{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.1,
  "eval_steps": 500,
  "global_step": 10000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1e-05,
      "grad_norm": 1.085229352812366,
      "learning_rate": 3e-06,
      "loss": 10.849,
      "step": 1
    },
    {
      "epoch": 2e-05,
      "grad_norm": 1.0764689186661929,
      "learning_rate": 6e-06,
      "loss": 10.8489,
      "step": 2
    },
    {
      "epoch": 3e-05,
      "grad_norm": 1.0926036068515363,
      "learning_rate": 9e-06,
      "loss": 10.8486,
      "step": 3
    },
    {
      "epoch": 4e-05,
      "grad_norm": 1.0859011783792423,
      "learning_rate": 1.2e-05,
      "loss": 10.848,
      "step": 4
    },
    {
      "epoch": 5e-05,
      "grad_norm": 1.0906873388641662,
      "learning_rate": 1.5e-05,
      "loss": 10.8453,
      "step": 5
    },
    {
      "epoch": 6e-05,
      "grad_norm": 1.0895888734627917,
      "learning_rate": 1.8e-05,
      "loss": 10.8447,
      "step": 6
    },
    {
      "epoch": 7e-05,
      "grad_norm": 1.0913329404483254,
      "learning_rate": 2.1000000000000002e-05,
      "loss": 10.8355,
      "step": 7
    },
    {
      "epoch": 8e-05,
      "grad_norm": 1.0766237663279077,
      "learning_rate": 2.4e-05,
      "loss": 10.8141,
      "step": 8
    },
    {
      "epoch": 9e-05,
      "grad_norm": 1.0617425338278697,
      "learning_rate": 2.7e-05,
      "loss": 10.8099,
      "step": 9
    },
    {
      "epoch": 0.0001,
      "grad_norm": 1.080271312928173,
      "learning_rate": 3e-05,
      "loss": 10.7983,
      "step": 10
    },
    {
      "epoch": 0.00011,
      "grad_norm": 1.0663119503184246,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 10.7844,
      "step": 11
    },
    {
      "epoch": 0.00012,
      "grad_norm": 1.0694505502575085,
      "learning_rate": 3.6e-05,
      "loss": 10.7739,
      "step": 12
    },
    {
      "epoch": 0.00013,
      "grad_norm": 1.0509206118753271,
      "learning_rate": 3.9e-05,
      "loss": 10.7554,
      "step": 13
    },
    {
      "epoch": 0.00014,
      "grad_norm": 1.0427932624919014,
      "learning_rate": 4.2000000000000004e-05,
      "loss": 10.7452,
      "step": 14
    },
    {
      "epoch": 0.00015,
      "grad_norm": 1.0323600599358198,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 10.736,
      "step": 15
    },
    {
      "epoch": 0.00016,
      "grad_norm": 1.0039727604034705,
      "learning_rate": 4.8e-05,
      "loss": 10.7191,
      "step": 16
    },
    {
      "epoch": 0.00017,
      "grad_norm": 0.9736033767894897,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 10.7049,
      "step": 17
    },
    {
      "epoch": 0.00018,
      "grad_norm": 0.9665971933165531,
      "learning_rate": 5.4e-05,
      "loss": 10.69,
      "step": 18
    },
    {
      "epoch": 0.00019,
      "grad_norm": 0.9428089164290921,
      "learning_rate": 5.7e-05,
      "loss": 10.6767,
      "step": 19
    },
    {
      "epoch": 0.0002,
      "grad_norm": 0.943192779381936,
      "learning_rate": 6e-05,
      "loss": 10.666,
      "step": 20
    },
    {
      "epoch": 0.00021,
      "grad_norm": 0.9246956411226879,
      "learning_rate": 6.3e-05,
      "loss": 10.6525,
      "step": 21
    },
    {
      "epoch": 0.00022,
      "grad_norm": 0.9206776841693647,
      "learning_rate": 6.599999999999999e-05,
      "loss": 10.6378,
      "step": 22
    },
    {
      "epoch": 0.00023,
      "grad_norm": 0.9152142474292139,
      "learning_rate": 6.9e-05,
      "loss": 10.626,
      "step": 23
    },
    {
      "epoch": 0.00024,
      "grad_norm": 0.9101286615991204,
      "learning_rate": 7.2e-05,
      "loss": 10.613,
      "step": 24
    },
    {
      "epoch": 0.00025,
      "grad_norm": 0.9100685942512898,
      "learning_rate": 7.500000000000001e-05,
      "loss": 10.5998,
      "step": 25
    },
    {
      "epoch": 0.00026,
      "grad_norm": 0.9093157110032968,
      "learning_rate": 7.8e-05,
      "loss": 10.5859,
      "step": 26
    },
    {
      "epoch": 0.00027,
      "grad_norm": 0.9036082412984342,
      "learning_rate": 8.1e-05,
      "loss": 10.5742,
      "step": 27
    },
    {
      "epoch": 0.00028,
      "grad_norm": 0.9033464449252919,
      "learning_rate": 8.400000000000001e-05,
      "loss": 10.5612,
      "step": 28
    },
    {
      "epoch": 0.00029,
      "grad_norm": 0.9067163029768021,
      "learning_rate": 8.7e-05,
      "loss": 10.5465,
      "step": 29
    },
    {
      "epoch": 0.0003,
      "grad_norm": 0.9152916501496612,
      "learning_rate": 8.999999999999999e-05,
      "loss": 10.5321,
      "step": 30
    },
    {
      "epoch": 0.00031,
      "grad_norm": 0.9122391999445313,
      "learning_rate": 9.3e-05,
      "loss": 10.5167,
      "step": 31
    },
    {
      "epoch": 0.00032,
      "grad_norm": 0.9090668181794543,
      "learning_rate": 9.6e-05,
      "loss": 10.5034,
      "step": 32
    },
    {
      "epoch": 0.00033,
      "grad_norm": 0.9095629132760054,
      "learning_rate": 9.900000000000001e-05,
      "loss": 10.4865,
      "step": 33
    },
    {
      "epoch": 0.00034,
      "grad_norm": 0.9107692784496799,
      "learning_rate": 0.00010200000000000001,
      "loss": 10.4706,
      "step": 34
    },
    {
      "epoch": 0.00035,
      "grad_norm": 0.9162740070354684,
      "learning_rate": 0.00010500000000000002,
      "loss": 10.4535,
      "step": 35
    },
    {
      "epoch": 0.00036,
      "grad_norm": 0.9114324016484531,
      "learning_rate": 0.000108,
      "loss": 10.4367,
      "step": 36
    },
    {
      "epoch": 0.00037,
      "grad_norm": 0.901331547883087,
      "learning_rate": 0.000111,
      "loss": 10.4196,
      "step": 37
    },
    {
      "epoch": 0.00038,
      "grad_norm": 0.9056352937834914,
      "learning_rate": 0.000114,
      "loss": 10.4003,
      "step": 38
    },
    {
      "epoch": 0.00039,
      "grad_norm": 0.9071824304759276,
      "learning_rate": 0.000117,
      "loss": 10.3815,
      "step": 39
    },
    {
      "epoch": 0.0004,
      "grad_norm": 0.911150856035614,
      "learning_rate": 0.00012,
      "loss": 10.3599,
      "step": 40
    },
    {
      "epoch": 0.00041,
      "grad_norm": 0.9001357892993758,
      "learning_rate": 0.000123,
      "loss": 10.3411,
      "step": 41
    },
    {
      "epoch": 0.00042,
      "grad_norm": 0.9057215347961051,
      "learning_rate": 0.000126,
      "loss": 10.3208,
      "step": 42
    },
    {
      "epoch": 0.00043,
      "grad_norm": 0.9101872556211666,
      "learning_rate": 0.000129,
      "loss": 10.2978,
      "step": 43
    },
    {
      "epoch": 0.00044,
      "grad_norm": 0.9097345366629672,
      "learning_rate": 0.00013199999999999998,
      "loss": 10.2768,
      "step": 44
    },
    {
      "epoch": 0.00045,
      "grad_norm": 0.9111670268341607,
      "learning_rate": 0.000135,
      "loss": 10.2539,
      "step": 45
    },
    {
      "epoch": 0.00046,
      "grad_norm": 0.9030473796809102,
      "learning_rate": 0.000138,
      "loss": 10.2322,
      "step": 46
    },
    {
      "epoch": 0.00047,
      "grad_norm": 0.9086055117133346,
      "learning_rate": 0.000141,
      "loss": 10.2064,
      "step": 47
    },
    {
      "epoch": 0.00048,
      "grad_norm": 0.9060414917101882,
      "learning_rate": 0.000144,
      "loss": 10.1837,
      "step": 48
    },
    {
      "epoch": 0.00049,
      "grad_norm": 0.9059366436676172,
      "learning_rate": 0.000147,
      "loss": 10.1599,
      "step": 49
    },
    {
      "epoch": 0.0005,
      "grad_norm": 0.9106165244124662,
      "learning_rate": 0.00015000000000000001,
      "loss": 10.1343,
      "step": 50
    },
    {
      "epoch": 0.00051,
      "grad_norm": 0.9110992341155927,
      "learning_rate": 0.000153,
      "loss": 10.1078,
      "step": 51
    },
    {
      "epoch": 0.00052,
      "grad_norm": 0.9077294523989683,
      "learning_rate": 0.000156,
      "loss": 10.0815,
      "step": 52
    },
    {
      "epoch": 0.00053,
      "grad_norm": 0.8995624264210066,
      "learning_rate": 0.000159,
      "loss": 10.0581,
      "step": 53
    },
    {
      "epoch": 0.00054,
      "grad_norm": 0.9187536344258231,
      "learning_rate": 0.000162,
      "loss": 10.026,
      "step": 54
    },
    {
      "epoch": 0.00055,
      "grad_norm": 0.9102867236672618,
      "learning_rate": 0.000165,
      "loss": 10.0019,
      "step": 55
    },
    {
      "epoch": 0.00056,
      "grad_norm": 0.9031170239719724,
      "learning_rate": 0.00016800000000000002,
      "loss": 9.9743,
      "step": 56
    },
    {
      "epoch": 0.00057,
      "grad_norm": 0.9090239107255728,
      "learning_rate": 0.000171,
      "loss": 9.9467,
      "step": 57
    },
    {
      "epoch": 0.00058,
      "grad_norm": 0.9070896749665766,
      "learning_rate": 0.000174,
      "loss": 9.9223,
      "step": 58
    },
    {
      "epoch": 0.00059,
      "grad_norm": 0.9116582783399498,
      "learning_rate": 0.000177,
      "loss": 9.8905,
      "step": 59
    },
    {
      "epoch": 0.0006,
      "grad_norm": 0.899071209460366,
      "learning_rate": 0.00017999999999999998,
      "loss": 9.8642,
      "step": 60
    },
    {
      "epoch": 0.00061,
      "grad_norm": 0.8984032767613607,
      "learning_rate": 0.000183,
      "loss": 9.8367,
      "step": 61
    },
    {
      "epoch": 0.00062,
      "grad_norm": 0.9043779152804675,
      "learning_rate": 0.000186,
      "loss": 9.8058,
      "step": 62
    },
    {
      "epoch": 0.00063,
      "grad_norm": 0.8969230453763369,
      "learning_rate": 0.000189,
      "loss": 9.7784,
      "step": 63
    },
    {
      "epoch": 0.00064,
      "grad_norm": 0.8929896633083918,
      "learning_rate": 0.000192,
      "loss": 9.7485,
      "step": 64
    },
    {
      "epoch": 0.00065,
      "grad_norm": 0.9016512171852502,
      "learning_rate": 0.00019500000000000002,
      "loss": 9.7165,
      "step": 65
    },
    {
      "epoch": 0.00066,
      "grad_norm": 0.8946267084059,
      "learning_rate": 0.00019800000000000002,
      "loss": 9.6927,
      "step": 66
    },
    {
      "epoch": 0.00067,
      "grad_norm": 0.8963872265737496,
      "learning_rate": 0.000201,
      "loss": 9.6633,
      "step": 67
    },
    {
      "epoch": 0.00068,
      "grad_norm": 0.9035635569767697,
      "learning_rate": 0.00020400000000000003,
      "loss": 9.6313,
      "step": 68
    },
    {
      "epoch": 0.00069,
      "grad_norm": 0.8884430485081615,
      "learning_rate": 0.00020700000000000002,
      "loss": 9.604,
      "step": 69
    },
    {
      "epoch": 0.0007,
      "grad_norm": 0.901710386427562,
      "learning_rate": 0.00021000000000000004,
      "loss": 9.5675,
      "step": 70
    },
    {
      "epoch": 0.00071,
      "grad_norm": 0.8913815692585527,
      "learning_rate": 0.00021299999999999997,
      "loss": 9.5403,
      "step": 71
    },
    {
      "epoch": 0.00072,
      "grad_norm": 0.8947982003050186,
      "learning_rate": 0.000216,
      "loss": 9.5138,
      "step": 72
    },
    {
      "epoch": 0.00073,
      "grad_norm": 0.8936046055705469,
      "learning_rate": 0.00021899999999999998,
      "loss": 9.4802,
      "step": 73
    },
    {
      "epoch": 0.00074,
      "grad_norm": 0.894403551075387,
      "learning_rate": 0.000222,
      "loss": 9.4539,
      "step": 74
    },
    {
      "epoch": 0.00075,
      "grad_norm": 0.8897139679621429,
      "learning_rate": 0.000225,
      "loss": 9.4187,
      "step": 75
    },
    {
      "epoch": 0.00076,
      "grad_norm": 0.89755782108852,
      "learning_rate": 0.000228,
      "loss": 9.3929,
      "step": 76
    },
    {
      "epoch": 0.00077,
      "grad_norm": 0.8917412779828411,
      "learning_rate": 0.000231,
      "loss": 9.3649,
      "step": 77
    },
    {
      "epoch": 0.00078,
      "grad_norm": 0.886930008239094,
      "learning_rate": 0.000234,
      "loss": 9.3351,
      "step": 78
    },
    {
      "epoch": 0.00079,
      "grad_norm": 0.8893115023720741,
      "learning_rate": 0.00023700000000000001,
      "loss": 9.2992,
      "step": 79
    },
    {
      "epoch": 0.0008,
      "grad_norm": 0.8917772276535281,
      "learning_rate": 0.00024,
      "loss": 9.2632,
      "step": 80
    },
    {
      "epoch": 0.00081,
      "grad_norm": 0.8971231461853929,
      "learning_rate": 0.00024300000000000002,
      "loss": 9.2316,
      "step": 81
    },
    {
      "epoch": 0.00082,
      "grad_norm": 0.9071155901294882,
      "learning_rate": 0.000246,
      "loss": 9.206,
      "step": 82
    },
    {
      "epoch": 0.00083,
      "grad_norm": 0.8963243651316662,
      "learning_rate": 0.00024900000000000004,
      "loss": 9.1708,
      "step": 83
    },
    {
      "epoch": 0.00084,
      "grad_norm": 0.9002519637093493,
      "learning_rate": 0.000252,
      "loss": 9.1395,
      "step": 84
    },
    {
      "epoch": 0.00085,
      "grad_norm": 0.8962870771351267,
      "learning_rate": 0.000255,
      "loss": 9.1159,
      "step": 85
    },
    {
      "epoch": 0.00086,
      "grad_norm": 0.8973951423301171,
      "learning_rate": 0.000258,
      "loss": 9.0799,
      "step": 86
    },
    {
      "epoch": 0.00087,
      "grad_norm": 0.8894399774297843,
      "learning_rate": 0.000261,
      "loss": 9.0571,
      "step": 87
    },
    {
      "epoch": 0.00088,
      "grad_norm": 0.8954974565265822,
      "learning_rate": 0.00026399999999999997,
      "loss": 9.0211,
      "step": 88
    },
    {
      "epoch": 0.00089,
      "grad_norm": 0.8903887799592864,
      "learning_rate": 0.000267,
      "loss": 8.9943,
      "step": 89
    },
    {
      "epoch": 0.0009,
      "grad_norm": 0.8896831425922581,
      "learning_rate": 0.00027,
      "loss": 8.9666,
      "step": 90
    },
    {
      "epoch": 0.00091,
      "grad_norm": 0.8841935030693385,
      "learning_rate": 0.000273,
      "loss": 8.9365,
      "step": 91
    },
    {
      "epoch": 0.00092,
      "grad_norm": 0.8862593694142118,
      "learning_rate": 0.000276,
      "loss": 8.9052,
      "step": 92
    },
    {
      "epoch": 0.00093,
      "grad_norm": 0.8881041222250594,
      "learning_rate": 0.000279,
      "loss": 8.8752,
      "step": 93
    },
    {
      "epoch": 0.00094,
      "grad_norm": 0.8868931156100198,
      "learning_rate": 0.000282,
      "loss": 8.8494,
      "step": 94
    },
    {
      "epoch": 0.00095,
      "grad_norm": 0.8826333164427848,
      "learning_rate": 0.000285,
      "loss": 8.8203,
      "step": 95
    },
    {
      "epoch": 0.00096,
      "grad_norm": 0.8910955494857569,
      "learning_rate": 0.000288,
      "loss": 8.7905,
      "step": 96
    },
    {
      "epoch": 0.00097,
      "grad_norm": 0.8809279978723125,
      "learning_rate": 0.000291,
      "loss": 8.7668,
      "step": 97
    },
    {
      "epoch": 0.00098,
      "grad_norm": 0.8829202734417477,
      "learning_rate": 0.000294,
      "loss": 8.73,
      "step": 98
    },
    {
      "epoch": 0.00099,
      "grad_norm": 0.8858154061322314,
      "learning_rate": 0.000297,
      "loss": 8.7072,
      "step": 99
    },
    {
      "epoch": 0.001,
      "grad_norm": 0.8861735275161908,
      "learning_rate": 0.00030000000000000003,
      "loss": 8.6806,
      "step": 100
    },
    {
      "epoch": 0.00101,
      "grad_norm": 0.8886468466177152,
      "learning_rate": 0.00030300000000000005,
      "loss": 8.6455,
      "step": 101
    },
    {
      "epoch": 0.00102,
      "grad_norm": 0.8755547614914917,
      "learning_rate": 0.000306,
      "loss": 8.625,
      "step": 102
    },
    {
      "epoch": 0.00103,
      "grad_norm": 0.879237266457317,
      "learning_rate": 0.000309,
      "loss": 8.6003,
      "step": 103
    },
    {
      "epoch": 0.00104,
      "grad_norm": 0.8697854479956653,
      "learning_rate": 0.000312,
      "loss": 8.5786,
      "step": 104
    },
    {
      "epoch": 0.00105,
      "grad_norm": 0.8733949043334459,
      "learning_rate": 0.000315,
      "loss": 8.552,
      "step": 105
    },
    {
      "epoch": 0.00106,
      "grad_norm": 0.8706148334292045,
      "learning_rate": 0.000318,
      "loss": 8.5224,
      "step": 106
    },
    {
      "epoch": 0.00107,
      "grad_norm": 0.8653683244965015,
      "learning_rate": 0.000321,
      "loss": 8.4981,
      "step": 107
    },
    {
      "epoch": 0.00108,
      "grad_norm": 0.8737239863866451,
      "learning_rate": 0.000324,
      "loss": 8.4698,
      "step": 108
    },
    {
      "epoch": 0.00109,
      "grad_norm": 0.8706512729216435,
      "learning_rate": 0.000327,
      "loss": 8.4501,
      "step": 109
    },
    {
      "epoch": 0.0011,
      "grad_norm": 0.8842127152624679,
      "learning_rate": 0.00033,
      "loss": 8.4274,
      "step": 110
    },
    {
      "epoch": 0.00111,
      "grad_norm": 0.9007754832304464,
      "learning_rate": 0.000333,
      "loss": 8.3985,
      "step": 111
    },
    {
      "epoch": 0.00112,
      "grad_norm": 0.9159855921530741,
      "learning_rate": 0.00033600000000000004,
      "loss": 8.3784,
      "step": 112
    },
    {
      "epoch": 0.00113,
      "grad_norm": 0.9063278036144603,
      "learning_rate": 0.000339,
      "loss": 8.3391,
      "step": 113
    },
    {
      "epoch": 0.00114,
      "grad_norm": 0.8437820836704115,
      "learning_rate": 0.000342,
      "loss": 8.3286,
      "step": 114
    },
    {
      "epoch": 0.00115,
      "grad_norm": 0.8612821674982505,
      "learning_rate": 0.00034500000000000004,
      "loss": 8.3022,
      "step": 115
    },
    {
      "epoch": 0.00116,
      "grad_norm": 0.8581012057508914,
      "learning_rate": 0.000348,
      "loss": 8.2786,
      "step": 116
    },
    {
      "epoch": 0.00117,
      "grad_norm": 0.834681162463853,
      "learning_rate": 0.000351,
      "loss": 8.2472,
      "step": 117
    },
    {
      "epoch": 0.00118,
      "grad_norm": 0.8526508674143746,
      "learning_rate": 0.000354,
      "loss": 8.2312,
      "step": 118
    },
    {
      "epoch": 0.00119,
      "grad_norm": 0.8471842679679056,
      "learning_rate": 0.000357,
      "loss": 8.2186,
      "step": 119
    },
    {
      "epoch": 0.0012,
      "grad_norm": 0.8238029079166322,
      "learning_rate": 0.00035999999999999997,
      "loss": 8.1917,
      "step": 120
    },
    {
      "epoch": 0.00121,
      "grad_norm": 0.832770075662114,
      "learning_rate": 0.000363,
      "loss": 8.1641,
      "step": 121
    },
    {
      "epoch": 0.00122,
      "grad_norm": 0.8176689288160716,
      "learning_rate": 0.000366,
      "loss": 8.1471,
      "step": 122
    },
    {
      "epoch": 0.00123,
      "grad_norm": 0.8121975389696077,
      "learning_rate": 0.000369,
      "loss": 8.1243,
      "step": 123
    },
    {
      "epoch": 0.00124,
      "grad_norm": 0.8287555396618358,
      "learning_rate": 0.000372,
      "loss": 8.1037,
      "step": 124
    },
    {
      "epoch": 0.00125,
      "grad_norm": 0.8387006564379252,
      "learning_rate": 0.000375,
      "loss": 8.0802,
      "step": 125
    },
    {
      "epoch": 0.00126,
      "grad_norm": 0.7985855201767323,
      "learning_rate": 0.000378,
      "loss": 8.0672,
      "step": 126
    },
    {
      "epoch": 0.00127,
      "grad_norm": 0.7830434817433392,
      "learning_rate": 0.000381,
      "loss": 8.0437,
      "step": 127
    },
    {
      "epoch": 0.00128,
      "grad_norm": 0.8466276209824322,
      "learning_rate": 0.000384,
      "loss": 8.0264,
      "step": 128
    },
    {
      "epoch": 0.00129,
      "grad_norm": 1.1259201196462498,
      "learning_rate": 0.00038700000000000003,
      "loss": 8.0199,
      "step": 129
    },
    {
      "epoch": 0.0013,
      "grad_norm": 1.356061683657538,
      "learning_rate": 0.00039000000000000005,
      "loss": 8.0054,
      "step": 130
    },
    {
      "epoch": 0.00131,
      "grad_norm": 0.781068809712636,
      "learning_rate": 0.000393,
      "loss": 7.9576,
      "step": 131
    },
    {
      "epoch": 0.00132,
      "grad_norm": 1.0367188815921278,
      "learning_rate": 0.00039600000000000003,
      "loss": 7.9591,
      "step": 132
    },
    {
      "epoch": 0.00133,
      "grad_norm": 0.7841067594214959,
      "learning_rate": 0.00039900000000000005,
      "loss": 7.9293,
      "step": 133
    },
    {
      "epoch": 0.00134,
      "grad_norm": 0.8594113418007541,
      "learning_rate": 0.000402,
      "loss": 7.9217,
      "step": 134
    },
    {
      "epoch": 0.00135,
      "grad_norm": 0.7468719464596628,
      "learning_rate": 0.00040500000000000003,
      "loss": 7.8883,
      "step": 135
    },
    {
      "epoch": 0.00136,
      "grad_norm": 0.8415432392576198,
      "learning_rate": 0.00040800000000000005,
      "loss": 7.8771,
      "step": 136
    },
    {
      "epoch": 0.00137,
      "grad_norm": 0.7386877288273068,
      "learning_rate": 0.000411,
      "loss": 7.8542,
      "step": 137
    },
    {
      "epoch": 0.00138,
      "grad_norm": 0.7450176106214967,
      "learning_rate": 0.00041400000000000003,
      "loss": 7.8322,
      "step": 138
    },
    {
      "epoch": 0.00139,
      "grad_norm": 0.7424698225823185,
      "learning_rate": 0.00041700000000000005,
      "loss": 7.8197,
      "step": 139
    },
    {
      "epoch": 0.0014,
      "grad_norm": 0.7268306219948636,
      "learning_rate": 0.00042000000000000007,
      "loss": 7.8018,
      "step": 140
    },
    {
      "epoch": 0.00141,
      "grad_norm": 0.7115037341291065,
      "learning_rate": 0.000423,
      "loss": 7.7905,
      "step": 141
    },
    {
      "epoch": 0.00142,
      "grad_norm": 0.6725409058271569,
      "learning_rate": 0.00042599999999999995,
      "loss": 7.7772,
      "step": 142
    },
    {
      "epoch": 0.00143,
      "grad_norm": 0.6830809074405504,
      "learning_rate": 0.00042899999999999997,
      "loss": 7.7496,
      "step": 143
    },
    {
      "epoch": 0.00144,
      "grad_norm": 0.6791849969278475,
      "learning_rate": 0.000432,
      "loss": 7.7318,
      "step": 144
    },
    {
      "epoch": 0.00145,
      "grad_norm": 0.6462817248800249,
      "learning_rate": 0.000435,
      "loss": 7.7251,
      "step": 145
    },
    {
      "epoch": 0.00146,
      "grad_norm": 0.6695930969912641,
      "learning_rate": 0.00043799999999999997,
      "loss": 7.6893,
      "step": 146
    },
    {
      "epoch": 0.00147,
      "grad_norm": 0.7012896651032599,
      "learning_rate": 0.000441,
      "loss": 7.6817,
      "step": 147
    },
    {
      "epoch": 0.00148,
      "grad_norm": 0.7755580877429182,
      "learning_rate": 0.000444,
      "loss": 7.6698,
      "step": 148
    },
    {
      "epoch": 0.00149,
      "grad_norm": 1.088947674236225,
      "learning_rate": 0.00044699999999999997,
      "loss": 7.6713,
      "step": 149
    },
    {
      "epoch": 0.0015,
      "grad_norm": 0.9396541473912592,
      "learning_rate": 0.00045,
      "loss": 7.6436,
      "step": 150
    },
    {
      "epoch": 0.00151,
      "grad_norm": 0.640217037835256,
      "learning_rate": 0.000453,
      "loss": 7.6133,
      "step": 151
    },
    {
      "epoch": 0.00152,
      "grad_norm": 0.9098496631236208,
      "learning_rate": 0.000456,
      "loss": 7.6149,
      "step": 152
    },
    {
      "epoch": 0.00153,
      "grad_norm": 0.6175759444520236,
      "learning_rate": 0.000459,
      "loss": 7.5962,
      "step": 153
    },
    {
      "epoch": 0.00154,
      "grad_norm": 0.6884616968083866,
      "learning_rate": 0.000462,
      "loss": 7.5772,
      "step": 154
    },
    {
      "epoch": 0.00155,
      "grad_norm": 0.6360938051960316,
      "learning_rate": 0.000465,
      "loss": 7.5598,
      "step": 155
    },
    {
      "epoch": 0.00156,
      "grad_norm": 0.6937352973699618,
      "learning_rate": 0.000468,
      "loss": 7.5366,
      "step": 156
    },
    {
      "epoch": 0.00157,
      "grad_norm": 0.6334556877551312,
      "learning_rate": 0.000471,
      "loss": 7.53,
      "step": 157
    },
    {
      "epoch": 0.00158,
      "grad_norm": 0.9707170671679272,
      "learning_rate": 0.00047400000000000003,
      "loss": 7.5153,
      "step": 158
    },
    {
      "epoch": 0.00159,
      "grad_norm": 1.0240750882994218,
      "learning_rate": 0.000477,
      "loss": 7.5076,
      "step": 159
    },
    {
      "epoch": 0.0016,
      "grad_norm": 0.9322894026235434,
      "learning_rate": 0.00048,
      "loss": 7.4836,
      "step": 160
    },
    {
      "epoch": 0.00161,
      "grad_norm": 0.5279037898518898,
      "learning_rate": 0.00048300000000000003,
      "loss": 7.4527,
      "step": 161
    },
    {
      "epoch": 0.00162,
      "grad_norm": 0.687356662308957,
      "learning_rate": 0.00048600000000000005,
      "loss": 7.4601,
      "step": 162
    },
    {
      "epoch": 0.00163,
      "grad_norm": 0.5623951705594973,
      "learning_rate": 0.0004890000000000001,
      "loss": 7.4388,
      "step": 163
    },
    {
      "epoch": 0.00164,
      "grad_norm": 0.5581337114560441,
      "learning_rate": 0.000492,
      "loss": 7.4399,
      "step": 164
    },
    {
      "epoch": 0.00165,
      "grad_norm": 0.5516159301488641,
      "learning_rate": 0.000495,
      "loss": 7.4126,
      "step": 165
    },
    {
      "epoch": 0.00166,
      "grad_norm": 0.5242244578051735,
      "learning_rate": 0.0004980000000000001,
      "loss": 7.3876,
      "step": 166
    },
    {
      "epoch": 0.00167,
      "grad_norm": 0.45997959649003123,
      "learning_rate": 0.000501,
      "loss": 7.3779,
      "step": 167
    },
    {
      "epoch": 0.00168,
      "grad_norm": 0.5436289820614866,
      "learning_rate": 0.000504,
      "loss": 7.3569,
      "step": 168
    },
    {
      "epoch": 0.00169,
      "grad_norm": 0.4983067598465849,
      "learning_rate": 0.0005070000000000001,
      "loss": 7.3495,
      "step": 169
    },
    {
      "epoch": 0.0017,
      "grad_norm": 0.4402852485923817,
      "learning_rate": 0.00051,
      "loss": 7.3316,
      "step": 170
    },
    {
      "epoch": 0.00171,
      "grad_norm": 0.5221521396945228,
      "learning_rate": 0.000513,
      "loss": 7.3138,
      "step": 171
    },
    {
      "epoch": 0.00172,
      "grad_norm": 0.45544347662440743,
      "learning_rate": 0.000516,
      "loss": 7.3129,
      "step": 172
    },
    {
      "epoch": 0.00173,
      "grad_norm": 0.4745602833877857,
      "learning_rate": 0.0005189999999999999,
      "loss": 7.2961,
      "step": 173
    },
    {
      "epoch": 0.00174,
      "grad_norm": 0.5121580158942174,
      "learning_rate": 0.000522,
      "loss": 7.311,
      "step": 174
    },
    {
      "epoch": 0.00175,
      "grad_norm": 0.680505499537256,
      "learning_rate": 0.000525,
      "loss": 7.2769,
      "step": 175
    },
    {
      "epoch": 0.00176,
      "grad_norm": 0.7210959926983863,
      "learning_rate": 0.0005279999999999999,
      "loss": 7.2549,
      "step": 176
    },
    {
      "epoch": 0.00177,
      "grad_norm": 0.7510526045152774,
      "learning_rate": 0.000531,
      "loss": 7.2572,
      "step": 177
    },
    {
      "epoch": 0.00178,
      "grad_norm": 0.7343359791017195,
      "learning_rate": 0.000534,
      "loss": 7.2506,
      "step": 178
    },
    {
      "epoch": 0.00179,
      "grad_norm": 0.6185909230661502,
      "learning_rate": 0.000537,
      "loss": 7.2371,
      "step": 179
    },
    {
      "epoch": 0.0018,
      "grad_norm": 0.42887581745789505,
      "learning_rate": 0.00054,
      "loss": 7.2041,
      "step": 180
    },
    {
      "epoch": 0.00181,
      "grad_norm": 0.5359194273155663,
      "learning_rate": 0.000543,
      "loss": 7.1961,
      "step": 181
    },
    {
      "epoch": 0.00182,
      "grad_norm": 0.4369454993609359,
      "learning_rate": 0.000546,
      "loss": 7.188,
      "step": 182
    },
    {
      "epoch": 0.00183,
      "grad_norm": 0.41001210068633426,
      "learning_rate": 0.000549,
      "loss": 7.1769,
      "step": 183
    },
    {
      "epoch": 0.00184,
      "grad_norm": 0.4396906853721559,
      "learning_rate": 0.000552,
      "loss": 7.1646,
      "step": 184
    },
    {
      "epoch": 0.00185,
      "grad_norm": 0.4311329351383538,
      "learning_rate": 0.000555,
      "loss": 7.1403,
      "step": 185
    },
    {
      "epoch": 0.00186,
      "grad_norm": 0.44013221474943204,
      "learning_rate": 0.000558,
      "loss": 7.1495,
      "step": 186
    },
    {
      "epoch": 0.00187,
      "grad_norm": 0.42719641993731927,
      "learning_rate": 0.000561,
      "loss": 7.1244,
      "step": 187
    },
    {
      "epoch": 0.00188,
      "grad_norm": 0.39192004433941763,
      "learning_rate": 0.000564,
      "loss": 7.1022,
      "step": 188
    },
    {
      "epoch": 0.00189,
      "grad_norm": 0.5474017259570552,
      "learning_rate": 0.000567,
      "loss": 7.1002,
      "step": 189
    },
    {
      "epoch": 0.0019,
      "grad_norm": 0.5320208949420774,
      "learning_rate": 0.00057,
      "loss": 7.0824,
      "step": 190
    },
    {
      "epoch": 0.00191,
      "grad_norm": 0.7661733453917681,
      "learning_rate": 0.000573,
      "loss": 7.0764,
      "step": 191
    },
    {
      "epoch": 0.00192,
      "grad_norm": 0.9858385620454592,
      "learning_rate": 0.000576,
      "loss": 7.0775,
      "step": 192
    },
    {
      "epoch": 0.00193,
      "grad_norm": 0.9591489873604585,
      "learning_rate": 0.000579,
      "loss": 7.0725,
      "step": 193
    },
    {
      "epoch": 0.00194,
      "grad_norm": 0.8774582305545432,
      "learning_rate": 0.000582,
      "loss": 7.0715,
      "step": 194
    },
    {
      "epoch": 0.00195,
      "grad_norm": 0.9015156284206135,
      "learning_rate": 0.000585,
      "loss": 7.0487,
      "step": 195
    },
    {
      "epoch": 0.00196,
      "grad_norm": 0.8000421121121074,
      "learning_rate": 0.000588,
      "loss": 7.0339,
      "step": 196
    },
    {
      "epoch": 0.00197,
      "grad_norm": 0.8468645119231468,
      "learning_rate": 0.000591,
      "loss": 7.0329,
      "step": 197
    },
    {
      "epoch": 0.00198,
      "grad_norm": 0.6924161058762034,
      "learning_rate": 0.000594,
      "loss": 7.0197,
      "step": 198
    },
    {
      "epoch": 0.00199,
      "grad_norm": 0.5671884633245193,
      "learning_rate": 0.0005970000000000001,
      "loss": 6.9992,
      "step": 199
    },
    {
      "epoch": 0.002,
      "grad_norm": 0.5557515599131739,
      "learning_rate": 0.0006000000000000001,
      "loss": 6.9982,
      "step": 200
    },
    {
      "epoch": 0.00201,
      "grad_norm": 0.4619645912916116,
      "learning_rate": 0.000603,
      "loss": 6.9694,
      "step": 201
    },
    {
      "epoch": 0.00202,
      "grad_norm": 0.4929222182059183,
      "learning_rate": 0.0006060000000000001,
      "loss": 6.9601,
      "step": 202
    },
    {
      "epoch": 0.00203,
      "grad_norm": 0.37702087689736097,
      "learning_rate": 0.0006090000000000001,
      "loss": 6.9599,
      "step": 203
    },
    {
      "epoch": 0.00204,
      "grad_norm": 0.45518892473332057,
      "learning_rate": 0.000612,
      "loss": 6.9476,
      "step": 204
    },
    {
      "epoch": 0.00205,
      "grad_norm": 0.384809834745988,
      "learning_rate": 0.000615,
      "loss": 6.9322,
      "step": 205
    },
    {
      "epoch": 0.00206,
      "grad_norm": 0.38976835520829006,
      "learning_rate": 0.000618,
      "loss": 6.9255,
      "step": 206
    },
    {
      "epoch": 0.00207,
      "grad_norm": 0.4375749283114229,
      "learning_rate": 0.000621,
      "loss": 6.9166,
      "step": 207
    },
    {
      "epoch": 0.00208,
      "grad_norm": 0.6317927736115376,
      "learning_rate": 0.000624,
      "loss": 6.9157,
      "step": 208
    },
    {
      "epoch": 0.00209,
      "grad_norm": 0.7772617938427908,
      "learning_rate": 0.000627,
      "loss": 6.9007,
      "step": 209
    },
    {
      "epoch": 0.0021,
      "grad_norm": 1.2298132092555871,
      "learning_rate": 0.00063,
      "loss": 6.9015,
      "step": 210
    },
    {
      "epoch": 0.00211,
      "grad_norm": 0.7435502579532725,
      "learning_rate": 0.000633,
      "loss": 6.8883,
      "step": 211
    },
    {
      "epoch": 0.00212,
      "grad_norm": 0.3794792840193541,
      "learning_rate": 0.000636,
      "loss": 6.8693,
      "step": 212
    },
    {
      "epoch": 0.00213,
      "grad_norm": 0.688114698891817,
      "learning_rate": 0.000639,
      "loss": 6.8589,
      "step": 213
    },
    {
      "epoch": 0.00214,
      "grad_norm": 0.49868168831557785,
      "learning_rate": 0.000642,
      "loss": 6.8504,
      "step": 214
    },
    {
      "epoch": 0.00215,
      "grad_norm": 0.6292678113208914,
      "learning_rate": 0.000645,
      "loss": 6.8506,
      "step": 215
    },
    {
      "epoch": 0.00216,
      "grad_norm": 0.3673676300147008,
      "learning_rate": 0.000648,
      "loss": 6.8383,
      "step": 216
    },
    {
      "epoch": 0.00217,
      "grad_norm": 0.4511278951279821,
      "learning_rate": 0.000651,
      "loss": 6.8131,
      "step": 217
    },
    {
      "epoch": 0.00218,
      "grad_norm": 0.3783399316849399,
      "learning_rate": 0.000654,
      "loss": 6.8146,
      "step": 218
    },
    {
      "epoch": 0.00219,
      "grad_norm": 0.3651267362185199,
      "learning_rate": 0.000657,
      "loss": 6.7867,
      "step": 219
    },
    {
      "epoch": 0.0022,
      "grad_norm": 0.4708567379601776,
      "learning_rate": 0.00066,
      "loss": 6.8074,
      "step": 220
    },
    {
      "epoch": 0.00221,
      "grad_norm": 0.5218809374007617,
      "learning_rate": 0.0006630000000000001,
      "loss": 6.7837,
      "step": 221
    },
    {
      "epoch": 0.00222,
      "grad_norm": 0.7518789100021657,
      "learning_rate": 0.000666,
      "loss": 6.7774,
      "step": 222
    },
    {
      "epoch": 0.00223,
      "grad_norm": 0.9964198530393009,
      "learning_rate": 0.000669,
      "loss": 6.7824,
      "step": 223
    },
    {
      "epoch": 0.00224,
      "grad_norm": 0.807522133159797,
      "learning_rate": 0.0006720000000000001,
      "loss": 6.7748,
      "step": 224
    },
    {
      "epoch": 0.00225,
      "grad_norm": 0.6803785939854445,
      "learning_rate": 0.000675,
      "loss": 6.7563,
      "step": 225
    },
    {
      "epoch": 0.00226,
      "grad_norm": 1.0762485305388094,
      "learning_rate": 0.000678,
      "loss": 6.7681,
      "step": 226
    },
    {
      "epoch": 0.00227,
      "grad_norm": 0.9129777590140102,
      "learning_rate": 0.0006810000000000001,
      "loss": 6.734,
      "step": 227
    },
    {
      "epoch": 0.00228,
      "grad_norm": 1.299087365220233,
      "learning_rate": 0.000684,
      "loss": 6.7517,
      "step": 228
    },
    {
      "epoch": 0.00229,
      "grad_norm": 0.7392729796935101,
      "learning_rate": 0.000687,
      "loss": 6.7331,
      "step": 229
    },
    {
      "epoch": 0.0023,
      "grad_norm": 0.6158971034672628,
      "learning_rate": 0.0006900000000000001,
      "loss": 6.7089,
      "step": 230
    },
    {
      "epoch": 0.00231,
      "grad_norm": 0.7016695647497186,
      "learning_rate": 0.000693,
      "loss": 6.6956,
      "step": 231
    },
    {
      "epoch": 0.00232,
      "grad_norm": 0.6434185638703606,
      "learning_rate": 0.000696,
      "loss": 6.7082,
      "step": 232
    },
    {
      "epoch": 0.00233,
      "grad_norm": 0.6044879418578446,
      "learning_rate": 0.0006990000000000001,
      "loss": 6.6921,
      "step": 233
    },
    {
      "epoch": 0.00234,
      "grad_norm": 0.4480061456613071,
      "learning_rate": 0.000702,
      "loss": 6.679,
      "step": 234
    },
    {
      "epoch": 0.00235,
      "grad_norm": 0.4492075259819563,
      "learning_rate": 0.000705,
      "loss": 6.6661,
      "step": 235
    },
    {
      "epoch": 0.00236,
      "grad_norm": 0.3889271878969786,
      "learning_rate": 0.000708,
      "loss": 6.6458,
      "step": 236
    },
    {
      "epoch": 0.00237,
      "grad_norm": 0.41607485698419117,
      "learning_rate": 0.0007109999999999999,
      "loss": 6.6575,
      "step": 237
    },
    {
      "epoch": 0.00238,
      "grad_norm": 0.3515958046168432,
      "learning_rate": 0.000714,
      "loss": 6.6432,
      "step": 238
    },
    {
      "epoch": 0.00239,
      "grad_norm": 0.4049983983930021,
      "learning_rate": 0.000717,
      "loss": 6.6274,
      "step": 239
    },
    {
      "epoch": 0.0024,
      "grad_norm": 0.3247748587680522,
      "learning_rate": 0.0007199999999999999,
      "loss": 6.6414,
      "step": 240
    },
    {
      "epoch": 0.00241,
      "grad_norm": 0.36262136655648425,
      "learning_rate": 0.000723,
      "loss": 6.6267,
      "step": 241
    },
    {
      "epoch": 0.00242,
      "grad_norm": 0.3238608381517686,
      "learning_rate": 0.000726,
      "loss": 6.598,
      "step": 242
    },
    {
      "epoch": 0.00243,
      "grad_norm": 0.31681258130419926,
      "learning_rate": 0.000729,
      "loss": 6.5877,
      "step": 243
    },
    {
      "epoch": 0.00244,
      "grad_norm": 0.3616370322868285,
      "learning_rate": 0.000732,
      "loss": 6.5911,
      "step": 244
    },
    {
      "epoch": 0.00245,
      "grad_norm": 0.4634076592576489,
      "learning_rate": 0.000735,
      "loss": 6.5815,
      "step": 245
    },
    {
      "epoch": 0.00246,
      "grad_norm": 0.7486687333799987,
      "learning_rate": 0.000738,
      "loss": 6.5693,
      "step": 246
    },
    {
      "epoch": 0.00247,
      "grad_norm": 1.662590620725261,
      "learning_rate": 0.000741,
      "loss": 6.5942,
      "step": 247
    },
    {
      "epoch": 0.00248,
      "grad_norm": 0.7962452736060305,
      "learning_rate": 0.000744,
      "loss": 6.5799,
      "step": 248
    },
    {
      "epoch": 0.00249,
      "grad_norm": 1.091558045553116,
      "learning_rate": 0.000747,
      "loss": 6.5681,
      "step": 249
    },
    {
      "epoch": 0.0025,
      "grad_norm": 1.1566118279124307,
      "learning_rate": 0.00075,
      "loss": 6.5821,
      "step": 250
    },
    {
      "epoch": 0.00251,
      "grad_norm": 0.7784764288643516,
      "learning_rate": 0.000753,
      "loss": 6.5311,
      "step": 251
    },
    {
      "epoch": 0.00252,
      "grad_norm": 0.8327004278614876,
      "learning_rate": 0.000756,
      "loss": 6.5476,
      "step": 252
    },
    {
      "epoch": 0.00253,
      "grad_norm": 0.8184368732684698,
      "learning_rate": 0.000759,
      "loss": 6.5327,
      "step": 253
    },
    {
      "epoch": 0.00254,
      "grad_norm": 1.1486609061260633,
      "learning_rate": 0.000762,
      "loss": 6.541,
      "step": 254
    },
    {
      "epoch": 0.00255,
      "grad_norm": 0.8085196503687528,
      "learning_rate": 0.0007650000000000001,
      "loss": 6.533,
      "step": 255
    },
    {
      "epoch": 0.00256,
      "grad_norm": 0.6227788989369496,
      "learning_rate": 0.000768,
      "loss": 6.5039,
      "step": 256
    },
    {
      "epoch": 0.00257,
      "grad_norm": 0.7880922816831363,
      "learning_rate": 0.000771,
      "loss": 6.5163,
      "step": 257
    },
    {
      "epoch": 0.00258,
      "grad_norm": 0.8888953354311911,
      "learning_rate": 0.0007740000000000001,
      "loss": 6.497,
      "step": 258
    },
    {
      "epoch": 0.00259,
      "grad_norm": 0.6556759438280303,
      "learning_rate": 0.000777,
      "loss": 6.491,
      "step": 259
    },
    {
      "epoch": 0.0026,
      "grad_norm": 0.5556045743581063,
      "learning_rate": 0.0007800000000000001,
      "loss": 6.4736,
      "step": 260
    },
    {
      "epoch": 0.00261,
      "grad_norm": 0.5757853625150707,
      "learning_rate": 0.0007830000000000001,
      "loss": 6.4835,
      "step": 261
    },
    {
      "epoch": 0.00262,
      "grad_norm": 0.4702991355160636,
      "learning_rate": 0.000786,
      "loss": 6.4605,
      "step": 262
    },
    {
      "epoch": 0.00263,
      "grad_norm": 0.5236270398277312,
      "learning_rate": 0.0007890000000000001,
      "loss": 6.4668,
      "step": 263
    },
    {
      "epoch": 0.00264,
      "grad_norm": 0.57918798311576,
      "learning_rate": 0.0007920000000000001,
      "loss": 6.4494,
      "step": 264
    },
    {
      "epoch": 0.00265,
      "grad_norm": 0.48386144230733374,
      "learning_rate": 0.000795,
      "loss": 6.4394,
      "step": 265
    },
    {
      "epoch": 0.00266,
      "grad_norm": 0.44170660781498655,
      "learning_rate": 0.0007980000000000001,
      "loss": 6.4372,
      "step": 266
    },
    {
      "epoch": 0.00267,
      "grad_norm": 0.35002993788441544,
      "learning_rate": 0.0008010000000000001,
      "loss": 6.4273,
      "step": 267
    },
    {
      "epoch": 0.00268,
      "grad_norm": 0.49450621569184094,
      "learning_rate": 0.000804,
      "loss": 6.426,
      "step": 268
    },
    {
      "epoch": 0.00269,
      "grad_norm": 0.5917094821390496,
      "learning_rate": 0.0008070000000000001,
      "loss": 6.4104,
      "step": 269
    },
    {
      "epoch": 0.0027,
      "grad_norm": 0.8610573273382283,
      "learning_rate": 0.0008100000000000001,
      "loss": 6.4165,
      "step": 270
    },
    {
      "epoch": 0.00271,
      "grad_norm": 0.9687752227819599,
      "learning_rate": 0.000813,
      "loss": 6.4342,
      "step": 271
    },
    {
      "epoch": 0.00272,
      "grad_norm": 0.8252297813091672,
      "learning_rate": 0.0008160000000000001,
      "loss": 6.4109,
      "step": 272
    },
    {
      "epoch": 0.00273,
      "grad_norm": 1.0509340869742279,
      "learning_rate": 0.0008190000000000001,
      "loss": 6.4091,
      "step": 273
    },
    {
      "epoch": 0.00274,
      "grad_norm": 1.2492890299831212,
      "learning_rate": 0.000822,
      "loss": 6.3984,
      "step": 274
    },
    {
      "epoch": 0.00275,
      "grad_norm": 0.7449833128353471,
      "learning_rate": 0.0008250000000000001,
      "loss": 6.3937,
      "step": 275
    },
    {
      "epoch": 0.00276,
      "grad_norm": 0.650217944416595,
      "learning_rate": 0.0008280000000000001,
      "loss": 6.383,
      "step": 276
    },
    {
      "epoch": 0.00277,
      "grad_norm": 0.6897831829588578,
      "learning_rate": 0.0008310000000000001,
      "loss": 6.3791,
      "step": 277
    },
    {
      "epoch": 0.00278,
      "grad_norm": 0.6845751052836879,
      "learning_rate": 0.0008340000000000001,
      "loss": 6.3625,
      "step": 278
    },
    {
      "epoch": 0.00279,
      "grad_norm": 0.6895680336476154,
      "learning_rate": 0.0008370000000000001,
      "loss": 6.3607,
      "step": 279
    },
    {
      "epoch": 0.0028,
      "grad_norm": 0.7962805776833526,
      "learning_rate": 0.0008400000000000001,
      "loss": 6.3497,
      "step": 280
    },
    {
      "epoch": 0.00281,
      "grad_norm": 0.7281805890288613,
      "learning_rate": 0.0008430000000000001,
      "loss": 6.3358,
      "step": 281
    },
    {
      "epoch": 0.00282,
      "grad_norm": 0.6077351466654652,
      "learning_rate": 0.000846,
      "loss": 6.3343,
      "step": 282
    },
    {
      "epoch": 0.00283,
      "grad_norm": 0.7021452115418912,
      "learning_rate": 0.0008489999999999999,
      "loss": 6.3268,
      "step": 283
    },
    {
      "epoch": 0.00284,
      "grad_norm": 0.8067589645945545,
      "learning_rate": 0.0008519999999999999,
      "loss": 6.3217,
      "step": 284
    },
    {
      "epoch": 0.00285,
      "grad_norm": 0.7238661685202905,
      "learning_rate": 0.000855,
      "loss": 6.3229,
      "step": 285
    },
    {
      "epoch": 0.00286,
      "grad_norm": 0.6566110557704579,
      "learning_rate": 0.0008579999999999999,
      "loss": 6.3002,
      "step": 286
    },
    {
      "epoch": 0.00287,
      "grad_norm": 0.5448339024770075,
      "learning_rate": 0.000861,
      "loss": 6.3016,
      "step": 287
    },
    {
      "epoch": 0.00288,
      "grad_norm": 0.4516947530472597,
      "learning_rate": 0.000864,
      "loss": 6.279,
      "step": 288
    },
    {
      "epoch": 0.00289,
      "grad_norm": 0.5415480649305122,
      "learning_rate": 0.0008669999999999999,
      "loss": 6.2929,
      "step": 289
    },
    {
      "epoch": 0.0029,
      "grad_norm": 0.5812113854990302,
      "learning_rate": 0.00087,
      "loss": 6.2743,
      "step": 290
    },
    {
      "epoch": 0.00291,
      "grad_norm": 0.5620937871349196,
      "learning_rate": 0.000873,
      "loss": 6.2679,
      "step": 291
    },
    {
      "epoch": 0.00292,
      "grad_norm": 0.43362286284353735,
      "learning_rate": 0.0008759999999999999,
      "loss": 6.2618,
      "step": 292
    },
    {
      "epoch": 0.00293,
      "grad_norm": 0.563953479113573,
      "learning_rate": 0.000879,
      "loss": 6.2491,
      "step": 293
    },
    {
      "epoch": 0.00294,
      "grad_norm": 0.5784107789505917,
      "learning_rate": 0.000882,
      "loss": 6.2636,
      "step": 294
    },
    {
      "epoch": 0.00295,
      "grad_norm": 0.5410292555611181,
      "learning_rate": 0.0008849999999999999,
      "loss": 6.2293,
      "step": 295
    },
    {
      "epoch": 0.00296,
      "grad_norm": 0.5413778994680785,
      "learning_rate": 0.000888,
      "loss": 6.2428,
      "step": 296
    },
    {
      "epoch": 0.00297,
      "grad_norm": 0.7587824592105622,
      "learning_rate": 0.000891,
      "loss": 6.2186,
      "step": 297
    },
    {
      "epoch": 0.00298,
      "grad_norm": 1.2390223277612324,
      "learning_rate": 0.0008939999999999999,
      "loss": 6.2454,
      "step": 298
    },
    {
      "epoch": 0.00299,
      "grad_norm": 1.0051162436627608,
      "learning_rate": 0.000897,
      "loss": 6.2195,
      "step": 299
    },
    {
      "epoch": 0.003,
      "grad_norm": 0.9301924220055711,
      "learning_rate": 0.0009,
      "loss": 6.2223,
      "step": 300
    },
    {
      "epoch": 0.00301,
      "grad_norm": 0.9092669241988458,
      "learning_rate": 0.0009029999999999999,
      "loss": 6.2133,
      "step": 301
    },
    {
      "epoch": 0.00302,
      "grad_norm": 1.0280476917286576,
      "learning_rate": 0.000906,
      "loss": 6.2045,
      "step": 302
    },
    {
      "epoch": 0.00303,
      "grad_norm": 0.955686909103224,
      "learning_rate": 0.000909,
      "loss": 6.2159,
      "step": 303
    },
    {
      "epoch": 0.00304,
      "grad_norm": 0.924153033838841,
      "learning_rate": 0.000912,
      "loss": 6.2068,
      "step": 304
    },
    {
      "epoch": 0.00305,
      "grad_norm": 1.0129569357421315,
      "learning_rate": 0.000915,
      "loss": 6.2162,
      "step": 305
    },
    {
      "epoch": 0.00306,
      "grad_norm": 0.908196615580651,
      "learning_rate": 0.000918,
      "loss": 6.1982,
      "step": 306
    },
    {
      "epoch": 0.00307,
      "grad_norm": 0.9412115799039344,
      "learning_rate": 0.000921,
      "loss": 6.1922,
      "step": 307
    },
    {
      "epoch": 0.00308,
      "grad_norm": 1.0133210251014897,
      "learning_rate": 0.000924,
      "loss": 6.1801,
      "step": 308
    },
    {
      "epoch": 0.00309,
      "grad_norm": 0.7885110954474147,
      "learning_rate": 0.000927,
      "loss": 6.1916,
      "step": 309
    },
    {
      "epoch": 0.0031,
      "grad_norm": 0.8502930513548597,
      "learning_rate": 0.00093,
      "loss": 6.1767,
      "step": 310
    },
    {
      "epoch": 0.00311,
      "grad_norm": 0.8830498260903274,
      "learning_rate": 0.000933,
      "loss": 6.1706,
      "step": 311
    },
    {
      "epoch": 0.00312,
      "grad_norm": 0.7731682808518218,
      "learning_rate": 0.000936,
      "loss": 6.1483,
      "step": 312
    },
    {
      "epoch": 0.00313,
      "grad_norm": 0.5034693382195244,
      "learning_rate": 0.0009390000000000001,
      "loss": 6.1541,
      "step": 313
    },
    {
      "epoch": 0.00314,
      "grad_norm": 0.5944206771214186,
      "learning_rate": 0.000942,
      "loss": 6.1615,
      "step": 314
    },
    {
      "epoch": 0.00315,
      "grad_norm": 0.6198949865082991,
      "learning_rate": 0.000945,
      "loss": 6.1382,
      "step": 315
    },
    {
      "epoch": 0.00316,
      "grad_norm": 0.6322301787116348,
      "learning_rate": 0.0009480000000000001,
      "loss": 6.112,
      "step": 316
    },
    {
      "epoch": 0.00317,
      "grad_norm": 0.6034799354522895,
      "learning_rate": 0.000951,
      "loss": 6.1294,
      "step": 317
    },
    {
      "epoch": 0.00318,
      "grad_norm": 0.594812791501401,
      "learning_rate": 0.000954,
      "loss": 6.1243,
      "step": 318
    },
    {
      "epoch": 0.00319,
      "grad_norm": 0.5121880287121718,
      "learning_rate": 0.0009570000000000001,
      "loss": 6.1104,
      "step": 319
    },
    {
      "epoch": 0.0032,
      "grad_norm": 0.4383869814323231,
      "learning_rate": 0.00096,
      "loss": 6.1078,
      "step": 320
    },
    {
      "epoch": 0.00321,
      "grad_norm": 0.46253996577569373,
      "learning_rate": 0.000963,
      "loss": 6.0957,
      "step": 321
    },
    {
      "epoch": 0.00322,
      "grad_norm": 0.5330974668531825,
      "learning_rate": 0.0009660000000000001,
      "loss": 6.0852,
      "step": 322
    },
    {
      "epoch": 0.00323,
      "grad_norm": 0.4269664457046623,
      "learning_rate": 0.000969,
      "loss": 6.0805,
      "step": 323
    },
    {
      "epoch": 0.00324,
      "grad_norm": 0.4120363869720864,
      "learning_rate": 0.0009720000000000001,
      "loss": 6.0833,
      "step": 324
    },
    {
      "epoch": 0.00325,
      "grad_norm": 0.43852524771556284,
      "learning_rate": 0.0009750000000000001,
      "loss": 6.0494,
      "step": 325
    },
    {
      "epoch": 0.00326,
      "grad_norm": 0.5390182954459198,
      "learning_rate": 0.0009780000000000001,
      "loss": 6.0675,
      "step": 326
    },
    {
      "epoch": 0.00327,
      "grad_norm": 0.7655676469366879,
      "learning_rate": 0.000981,
      "loss": 6.0729,
      "step": 327
    },
    {
      "epoch": 0.00328,
      "grad_norm": 1.3095809996152918,
      "learning_rate": 0.000984,
      "loss": 6.0661,
      "step": 328
    },
    {
      "epoch": 0.00329,
      "grad_norm": 0.9398963728555475,
      "learning_rate": 0.000987,
      "loss": 6.0593,
      "step": 329
    },
    {
      "epoch": 0.0033,
      "grad_norm": 0.9687365266328093,
      "learning_rate": 0.00099,
      "loss": 6.0595,
      "step": 330
    },
    {
      "epoch": 0.00331,
      "grad_norm": 1.0572725295338647,
      "learning_rate": 0.0009930000000000002,
      "loss": 6.0709,
      "step": 331
    },
    {
      "epoch": 0.00332,
      "grad_norm": 0.7804253243852415,
      "learning_rate": 0.0009960000000000001,
      "loss": 6.0449,
      "step": 332
    },
    {
      "epoch": 0.00333,
      "grad_norm": 0.9477444786898163,
      "learning_rate": 0.000999,
      "loss": 6.0352,
      "step": 333
    },
    {
      "epoch": 0.00334,
      "grad_norm": 1.119636024931192,
      "learning_rate": 0.001002,
      "loss": 6.0509,
      "step": 334
    },
    {
      "epoch": 0.00335,
      "grad_norm": 1.0398096390896505,
      "learning_rate": 0.001005,
      "loss": 6.0495,
      "step": 335
    },
    {
      "epoch": 0.00336,
      "grad_norm": 0.8946811171098662,
      "learning_rate": 0.001008,
      "loss": 6.0348,
      "step": 336
    },
    {
      "epoch": 0.00337,
      "grad_norm": 1.0240126692017184,
      "learning_rate": 0.0010110000000000002,
      "loss": 6.0215,
      "step": 337
    },
    {
      "epoch": 0.00338,
      "grad_norm": 1.0245564805630587,
      "learning_rate": 0.0010140000000000001,
      "loss": 6.0414,
      "step": 338
    },
    {
      "epoch": 0.00339,
      "grad_norm": 0.9606863659347135,
      "learning_rate": 0.0010170000000000001,
      "loss": 6.0439,
      "step": 339
    },
    {
      "epoch": 0.0034,
      "grad_norm": 0.7549695201896788,
      "learning_rate": 0.00102,
      "loss": 6.0067,
      "step": 340
    },
    {
      "epoch": 0.00341,
      "grad_norm": 0.7981881225838676,
      "learning_rate": 0.001023,
      "loss": 5.9935,
      "step": 341
    },
    {
      "epoch": 0.00342,
      "grad_norm": 0.6476806075689499,
      "learning_rate": 0.001026,
      "loss": 5.9919,
      "step": 342
    },
    {
      "epoch": 0.00343,
      "grad_norm": 0.6180530813345404,
      "learning_rate": 0.0010290000000000002,
      "loss": 5.9943,
      "step": 343
    },
    {
      "epoch": 0.00344,
      "grad_norm": 0.5904235085071222,
      "learning_rate": 0.001032,
      "loss": 5.9824,
      "step": 344
    },
    {
      "epoch": 0.00345,
      "grad_norm": 0.5662246103658737,
      "learning_rate": 0.001035,
      "loss": 5.9736,
      "step": 345
    },
    {
      "epoch": 0.00346,
      "grad_norm": 0.5815647759401746,
      "learning_rate": 0.0010379999999999999,
      "loss": 5.9787,
      "step": 346
    },
    {
      "epoch": 0.00347,
      "grad_norm": 0.5922585456492798,
      "learning_rate": 0.001041,
      "loss": 5.9505,
      "step": 347
    },
    {
      "epoch": 0.00348,
      "grad_norm": 0.546770754957902,
      "learning_rate": 0.001044,
      "loss": 5.9566,
      "step": 348
    },
    {
      "epoch": 0.00349,
      "grad_norm": 0.5259270224752429,
      "learning_rate": 0.001047,
      "loss": 5.9388,
      "step": 349
    },
    {
      "epoch": 0.0035,
      "grad_norm": 0.5639453872626701,
      "learning_rate": 0.00105,
      "loss": 5.96,
      "step": 350
    },
    {
      "epoch": 0.00351,
      "grad_norm": 0.562716119705,
      "learning_rate": 0.001053,
      "loss": 5.9575,
      "step": 351
    },
    {
      "epoch": 0.00352,
      "grad_norm": 0.6276157353012132,
      "learning_rate": 0.0010559999999999999,
      "loss": 5.9357,
      "step": 352
    },
    {
      "epoch": 0.00353,
      "grad_norm": 0.6899105072430818,
      "learning_rate": 0.001059,
      "loss": 5.939,
      "step": 353
    },
    {
      "epoch": 0.00354,
      "grad_norm": 0.7414855002893123,
      "learning_rate": 0.001062,
      "loss": 5.9142,
      "step": 354
    },
    {
      "epoch": 0.00355,
      "grad_norm": 0.8275245559939801,
      "learning_rate": 0.001065,
      "loss": 5.932,
      "step": 355
    },
    {
      "epoch": 0.00356,
      "grad_norm": 0.8334699202442165,
      "learning_rate": 0.001068,
      "loss": 5.925,
      "step": 356
    },
    {
      "epoch": 0.00357,
      "grad_norm": 0.733463396370079,
      "learning_rate": 0.001071,
      "loss": 5.9147,
      "step": 357
    },
    {
      "epoch": 0.00358,
      "grad_norm": 0.5898600387972718,
      "learning_rate": 0.001074,
      "loss": 5.9022,
      "step": 358
    },
    {
      "epoch": 0.00359,
      "grad_norm": 0.53807120274405,
      "learning_rate": 0.001077,
      "loss": 5.8933,
      "step": 359
    },
    {
      "epoch": 0.0036,
      "grad_norm": 0.5868526923580266,
      "learning_rate": 0.00108,
      "loss": 5.8984,
      "step": 360
    },
    {
      "epoch": 0.00361,
      "grad_norm": 0.6561371025439784,
      "learning_rate": 0.001083,
      "loss": 5.8927,
      "step": 361
    },
    {
      "epoch": 0.00362,
      "grad_norm": 1.003392418604553,
      "learning_rate": 0.001086,
      "loss": 5.8918,
      "step": 362
    },
    {
      "epoch": 0.00363,
      "grad_norm": 1.2318504963643775,
      "learning_rate": 0.001089,
      "loss": 5.899,
      "step": 363
    },
    {
      "epoch": 0.00364,
      "grad_norm": 0.7073595215993886,
      "learning_rate": 0.001092,
      "loss": 5.8956,
      "step": 364
    },
    {
      "epoch": 0.00365,
      "grad_norm": 0.6834408409212124,
      "learning_rate": 0.001095,
      "loss": 5.8924,
      "step": 365
    },
    {
      "epoch": 0.00366,
      "grad_norm": 0.808409762735137,
      "learning_rate": 0.001098,
      "loss": 5.8732,
      "step": 366
    },
    {
      "epoch": 0.00367,
      "grad_norm": 1.150363353399194,
      "learning_rate": 0.001101,
      "loss": 5.8871,
      "step": 367
    },
    {
      "epoch": 0.00368,
      "grad_norm": 1.5759048240116487,
      "learning_rate": 0.001104,
      "loss": 5.9113,
      "step": 368
    },
    {
      "epoch": 0.00369,
      "grad_norm": 0.6706600035773841,
      "learning_rate": 0.001107,
      "loss": 5.8826,
      "step": 369
    },
    {
      "epoch": 0.0037,
      "grad_norm": 1.0165372867603926,
      "learning_rate": 0.00111,
      "loss": 5.8773,
      "step": 370
    },
    {
      "epoch": 0.00371,
      "grad_norm": 1.199271508025637,
      "learning_rate": 0.001113,
      "loss": 5.8595,
      "step": 371
    },
    {
      "epoch": 0.00372,
      "grad_norm": 0.8890339964058547,
      "learning_rate": 0.001116,
      "loss": 5.8693,
      "step": 372
    },
    {
      "epoch": 0.00373,
      "grad_norm": 1.0853395614111072,
      "learning_rate": 0.001119,
      "loss": 5.8674,
      "step": 373
    },
    {
      "epoch": 0.00374,
      "grad_norm": 0.9530307319001724,
      "learning_rate": 0.001122,
      "loss": 5.8703,
      "step": 374
    },
    {
      "epoch": 0.00375,
      "grad_norm": 0.934750801302772,
      "learning_rate": 0.0011250000000000001,
      "loss": 5.8719,
      "step": 375
    },
    {
      "epoch": 0.00376,
      "grad_norm": 0.7986123477309184,
      "learning_rate": 0.001128,
      "loss": 5.863,
      "step": 376
    },
    {
      "epoch": 0.00377,
      "grad_norm": 1.034328210427756,
      "learning_rate": 0.001131,
      "loss": 5.8644,
      "step": 377
    },
    {
      "epoch": 0.00378,
      "grad_norm": 1.1394107219722458,
      "learning_rate": 0.001134,
      "loss": 5.8702,
      "step": 378
    },
    {
      "epoch": 0.00379,
      "grad_norm": 0.6778105205682995,
      "learning_rate": 0.001137,
      "loss": 5.8386,
      "step": 379
    },
    {
      "epoch": 0.0038,
      "grad_norm": 0.7735792074169382,
      "learning_rate": 0.00114,
      "loss": 5.8619,
      "step": 380
    },
    {
      "epoch": 0.00381,
      "grad_norm": 0.882590402307869,
      "learning_rate": 0.0011430000000000001,
      "loss": 5.83,
      "step": 381
    },
    {
      "epoch": 0.00382,
      "grad_norm": 0.8860425089624719,
      "learning_rate": 0.001146,
      "loss": 5.8197,
      "step": 382
    },
    {
      "epoch": 0.00383,
      "grad_norm": 1.0396684860073488,
      "learning_rate": 0.001149,
      "loss": 5.836,
      "step": 383
    },
    {
      "epoch": 0.00384,
      "grad_norm": 0.9683520339794217,
      "learning_rate": 0.001152,
      "loss": 5.8386,
      "step": 384
    },
    {
      "epoch": 0.00385,
      "grad_norm": 0.8210443052214895,
      "learning_rate": 0.001155,
      "loss": 5.8215,
      "step": 385
    },
    {
      "epoch": 0.00386,
      "grad_norm": 1.030920073327714,
      "learning_rate": 0.001158,
      "loss": 5.8336,
      "step": 386
    },
    {
      "epoch": 0.00387,
      "grad_norm": 0.8032114385497527,
      "learning_rate": 0.0011610000000000001,
      "loss": 5.8025,
      "step": 387
    },
    {
      "epoch": 0.00388,
      "grad_norm": 0.6803620347459473,
      "learning_rate": 0.001164,
      "loss": 5.7978,
      "step": 388
    },
    {
      "epoch": 0.00389,
      "grad_norm": 0.6752304208768743,
      "learning_rate": 0.001167,
      "loss": 5.8027,
      "step": 389
    },
    {
      "epoch": 0.0039,
      "grad_norm": 0.6054825081153106,
      "learning_rate": 0.00117,
      "loss": 5.7828,
      "step": 390
    },
    {
      "epoch": 0.00391,
      "grad_norm": 0.5156470046541872,
      "learning_rate": 0.001173,
      "loss": 5.7863,
      "step": 391
    },
    {
      "epoch": 0.00392,
      "grad_norm": 0.49609211852516366,
      "learning_rate": 0.001176,
      "loss": 5.7945,
      "step": 392
    },
    {
      "epoch": 0.00393,
      "grad_norm": 0.42817727304572534,
      "learning_rate": 0.0011790000000000001,
      "loss": 5.7664,
      "step": 393
    },
    {
      "epoch": 0.00394,
      "grad_norm": 0.43894767278563757,
      "learning_rate": 0.001182,
      "loss": 5.7539,
      "step": 394
    },
    {
      "epoch": 0.00395,
      "grad_norm": 0.41850660912289844,
      "learning_rate": 0.001185,
      "loss": 5.7366,
      "step": 395
    },
    {
      "epoch": 0.00396,
      "grad_norm": 0.41745423249833347,
      "learning_rate": 0.001188,
      "loss": 5.7516,
      "step": 396
    },
    {
      "epoch": 0.00397,
      "grad_norm": 0.40474070631964676,
      "learning_rate": 0.001191,
      "loss": 5.7433,
      "step": 397
    },
    {
      "epoch": 0.00398,
      "grad_norm": 0.3677722018443306,
      "learning_rate": 0.0011940000000000002,
      "loss": 5.7479,
      "step": 398
    },
    {
      "epoch": 0.00399,
      "grad_norm": 0.3480206547108819,
      "learning_rate": 0.0011970000000000001,
      "loss": 5.7478,
      "step": 399
    },
    {
      "epoch": 0.004,
      "grad_norm": 0.3434828622202681,
      "learning_rate": 0.0012000000000000001,
      "loss": 5.7345,
      "step": 400
    },
    {
      "epoch": 0.00401,
      "grad_norm": 0.34918136204349326,
      "learning_rate": 0.001203,
      "loss": 5.7155,
      "step": 401
    },
    {
      "epoch": 0.00402,
      "grad_norm": 0.30554980038341767,
      "learning_rate": 0.001206,
      "loss": 5.7215,
      "step": 402
    },
    {
      "epoch": 0.00403,
      "grad_norm": 0.38840665384838735,
      "learning_rate": 0.001209,
      "loss": 5.7184,
      "step": 403
    },
    {
      "epoch": 0.00404,
      "grad_norm": 0.5409513056663879,
      "learning_rate": 0.0012120000000000002,
      "loss": 5.714,
      "step": 404
    },
    {
      "epoch": 0.00405,
      "grad_norm": 0.845020924848713,
      "learning_rate": 0.0012150000000000002,
      "loss": 5.7181,
      "step": 405
    },
    {
      "epoch": 0.00406,
      "grad_norm": 1.1911410915070972,
      "learning_rate": 0.0012180000000000001,
      "loss": 5.7205,
      "step": 406
    },
    {
      "epoch": 0.00407,
      "grad_norm": 0.6577283880630926,
      "learning_rate": 0.0012209999999999999,
      "loss": 5.6994,
      "step": 407
    },
    {
      "epoch": 0.00408,
      "grad_norm": 0.7475745975098248,
      "learning_rate": 0.001224,
      "loss": 5.7213,
      "step": 408
    },
    {
      "epoch": 0.00409,
      "grad_norm": 0.9872391126413178,
      "learning_rate": 0.001227,
      "loss": 5.7126,
      "step": 409
    },
    {
      "epoch": 0.0041,
      "grad_norm": 1.1359034721668335,
      "learning_rate": 0.00123,
      "loss": 5.7088,
      "step": 410
    },
    {
      "epoch": 0.00411,
      "grad_norm": 1.3596329145222696,
      "learning_rate": 0.001233,
      "loss": 5.7402,
      "step": 411
    },
    {
      "epoch": 0.00412,
      "grad_norm": 0.7538358474928969,
      "learning_rate": 0.001236,
      "loss": 5.7066,
      "step": 412
    },
    {
      "epoch": 0.00413,
      "grad_norm": 0.9465320539051596,
      "learning_rate": 0.0012389999999999999,
      "loss": 5.7197,
      "step": 413
    },
    {
      "epoch": 0.00414,
      "grad_norm": 0.9262933655624658,
      "learning_rate": 0.001242,
      "loss": 5.6978,
      "step": 414
    },
    {
      "epoch": 0.00415,
      "grad_norm": 1.1564175286146172,
      "learning_rate": 0.001245,
      "loss": 5.7105,
      "step": 415
    },
    {
      "epoch": 0.00416,
      "grad_norm": 1.1001247072345506,
      "learning_rate": 0.001248,
      "loss": 5.6929,
      "step": 416
    },
    {
      "epoch": 0.00417,
      "grad_norm": 1.0416153435685582,
      "learning_rate": 0.001251,
      "loss": 5.7199,
      "step": 417
    },
    {
      "epoch": 0.00418,
      "grad_norm": 1.0281555694116995,
      "learning_rate": 0.001254,
      "loss": 5.6999,
      "step": 418
    },
    {
      "epoch": 0.00419,
      "grad_norm": 1.1154617103247704,
      "learning_rate": 0.0012569999999999999,
      "loss": 5.6876,
      "step": 419
    },
    {
      "epoch": 0.0042,
      "grad_norm": 0.9999912825556322,
      "learning_rate": 0.00126,
      "loss": 5.7045,
      "step": 420
    },
    {
      "epoch": 0.00421,
      "grad_norm": 0.9729638313238949,
      "learning_rate": 0.001263,
      "loss": 5.6933,
      "step": 421
    },
    {
      "epoch": 0.00422,
      "grad_norm": 1.0297208241186608,
      "learning_rate": 0.001266,
      "loss": 5.6894,
      "step": 422
    },
    {
      "epoch": 0.00423,
      "grad_norm": 0.8381062456476874,
      "learning_rate": 0.001269,
      "loss": 5.6811,
      "step": 423
    },
    {
      "epoch": 0.00424,
      "grad_norm": 0.73825928337582,
      "learning_rate": 0.001272,
      "loss": 5.6773,
      "step": 424
    },
    {
      "epoch": 0.00425,
      "grad_norm": 0.8151693610653118,
      "learning_rate": 0.001275,
      "loss": 5.6815,
      "step": 425
    },
    {
      "epoch": 0.00426,
      "grad_norm": 0.9877723869544494,
      "learning_rate": 0.001278,
      "loss": 5.6881,
      "step": 426
    },
    {
      "epoch": 0.00427,
      "grad_norm": 0.8875063218226354,
      "learning_rate": 0.001281,
      "loss": 5.6676,
      "step": 427
    },
    {
      "epoch": 0.00428,
      "grad_norm": 0.8761493100474917,
      "learning_rate": 0.001284,
      "loss": 5.6437,
      "step": 428
    },
    {
      "epoch": 0.00429,
      "grad_norm": 1.109439755296158,
      "learning_rate": 0.001287,
      "loss": 5.668,
      "step": 429
    },
    {
      "epoch": 0.0043,
      "grad_norm": 0.8726361797071612,
      "learning_rate": 0.00129,
      "loss": 5.6611,
      "step": 430
    },
    {
      "epoch": 0.00431,
      "grad_norm": 0.5601358924900582,
      "learning_rate": 0.001293,
      "loss": 5.6342,
      "step": 431
    },
    {
      "epoch": 0.00432,
      "grad_norm": 0.5940988397687699,
      "learning_rate": 0.001296,
      "loss": 5.6448,
      "step": 432
    },
    {
      "epoch": 0.00433,
      "grad_norm": 0.480538337754196,
      "learning_rate": 0.001299,
      "loss": 5.6336,
      "step": 433
    },
    {
      "epoch": 0.00434,
      "grad_norm": 0.5069699406038057,
      "learning_rate": 0.001302,
      "loss": 5.6331,
      "step": 434
    },
    {
      "epoch": 0.00435,
      "grad_norm": 0.417220569666801,
      "learning_rate": 0.001305,
      "loss": 5.6227,
      "step": 435
    },
    {
      "epoch": 0.00436,
      "grad_norm": 0.46597565539240443,
      "learning_rate": 0.001308,
      "loss": 5.6039,
      "step": 436
    },
    {
      "epoch": 0.00437,
      "grad_norm": 0.37605303006482044,
      "learning_rate": 0.001311,
      "loss": 5.6161,
      "step": 437
    },
    {
      "epoch": 0.00438,
      "grad_norm": 0.3526339213940271,
      "learning_rate": 0.001314,
      "loss": 5.5977,
      "step": 438
    },
    {
      "epoch": 0.00439,
      "grad_norm": 0.3618369277094543,
      "learning_rate": 0.001317,
      "loss": 5.6246,
      "step": 439
    },
    {
      "epoch": 0.0044,
      "grad_norm": 0.3223158135938896,
      "learning_rate": 0.00132,
      "loss": 5.5938,
      "step": 440
    },
    {
      "epoch": 0.00441,
      "grad_norm": 0.3386640445759432,
      "learning_rate": 0.001323,
      "loss": 5.5905,
      "step": 441
    },
    {
      "epoch": 0.00442,
      "grad_norm": 0.3397360216396013,
      "learning_rate": 0.0013260000000000001,
      "loss": 5.5838,
      "step": 442
    },
    {
      "epoch": 0.00443,
      "grad_norm": 0.34532354892574607,
      "learning_rate": 0.001329,
      "loss": 5.5832,
      "step": 443
    },
    {
      "epoch": 0.00444,
      "grad_norm": 0.37928556611065656,
      "learning_rate": 0.001332,
      "loss": 5.5853,
      "step": 444
    },
    {
      "epoch": 0.00445,
      "grad_norm": 0.44947894711961484,
      "learning_rate": 0.001335,
      "loss": 5.5723,
      "step": 445
    },
    {
      "epoch": 0.00446,
      "grad_norm": 0.5239308309237933,
      "learning_rate": 0.001338,
      "loss": 5.5751,
      "step": 446
    },
    {
      "epoch": 0.00447,
      "grad_norm": 0.60889528761374,
      "learning_rate": 0.001341,
      "loss": 5.5777,
      "step": 447
    },
    {
      "epoch": 0.00448,
      "grad_norm": 0.6150556040535831,
      "learning_rate": 0.0013440000000000001,
      "loss": 5.561,
      "step": 448
    },
    {
      "epoch": 0.00449,
      "grad_norm": 0.5444850551876294,
      "learning_rate": 0.001347,
      "loss": 5.5623,
      "step": 449
    },
    {
      "epoch": 0.0045,
      "grad_norm": 0.6163688303714219,
      "learning_rate": 0.00135,
      "loss": 5.5617,
      "step": 450
    },
    {
      "epoch": 0.00451,
      "grad_norm": 0.7972728836658292,
      "learning_rate": 0.001353,
      "loss": 5.5614,
      "step": 451
    },
    {
      "epoch": 0.00452,
      "grad_norm": 0.7711764764543457,
      "learning_rate": 0.001356,
      "loss": 5.5454,
      "step": 452
    },
    {
      "epoch": 0.00453,
      "grad_norm": 0.7702356986189732,
      "learning_rate": 0.001359,
      "loss": 5.5379,
      "step": 453
    },
    {
      "epoch": 0.00454,
      "grad_norm": 1.0838160417982272,
      "learning_rate": 0.0013620000000000001,
      "loss": 5.5686,
      "step": 454
    },
    {
      "epoch": 0.00455,
      "grad_norm": 1.152655585803101,
      "learning_rate": 0.0013650000000000001,
      "loss": 5.5757,
      "step": 455
    },
    {
      "epoch": 0.00456,
      "grad_norm": 1.2776355160689266,
      "learning_rate": 0.001368,
      "loss": 5.5831,
      "step": 456
    },
    {
      "epoch": 0.00457,
      "grad_norm": 1.091292995937963,
      "learning_rate": 0.001371,
      "loss": 5.5727,
      "step": 457
    },
    {
      "epoch": 0.00458,
      "grad_norm": 0.8963405103823251,
      "learning_rate": 0.001374,
      "loss": 5.5735,
      "step": 458
    },
    {
      "epoch": 0.00459,
      "grad_norm": 1.0168648046101516,
      "learning_rate": 0.0013770000000000002,
      "loss": 5.5669,
      "step": 459
    },
    {
      "epoch": 0.0046,
      "grad_norm": 1.296131601782866,
      "learning_rate": 0.0013800000000000002,
      "loss": 5.5665,
      "step": 460
    },
    {
      "epoch": 0.00461,
      "grad_norm": 0.9634420565591739,
      "learning_rate": 0.0013830000000000001,
      "loss": 5.556,
      "step": 461
    },
    {
      "epoch": 0.00462,
      "grad_norm": 0.9383281224355017,
      "learning_rate": 0.001386,
      "loss": 5.5634,
      "step": 462
    },
    {
      "epoch": 0.00463,
      "grad_norm": 1.2569585597421309,
      "learning_rate": 0.001389,
      "loss": 5.5542,
      "step": 463
    },
    {
      "epoch": 0.00464,
      "grad_norm": 0.9874595595654581,
      "learning_rate": 0.001392,
      "loss": 5.5689,
      "step": 464
    },
    {
      "epoch": 0.00465,
      "grad_norm": 1.085595749506429,
      "learning_rate": 0.0013950000000000002,
      "loss": 5.5385,
      "step": 465
    },
    {
      "epoch": 0.00466,
      "grad_norm": 1.0673943770446899,
      "learning_rate": 0.0013980000000000002,
      "loss": 5.5603,
      "step": 466
    },
    {
      "epoch": 0.00467,
      "grad_norm": 0.8139501043376736,
      "learning_rate": 0.0014010000000000001,
      "loss": 5.5432,
      "step": 467
    },
    {
      "epoch": 0.00468,
      "grad_norm": 0.7494382793960519,
      "learning_rate": 0.001404,
      "loss": 5.5245,
      "step": 468
    },
    {
      "epoch": 0.00469,
      "grad_norm": 0.7634992086588068,
      "learning_rate": 0.001407,
      "loss": 5.5282,
      "step": 469
    },
    {
      "epoch": 0.0047,
      "grad_norm": 0.8018093758476836,
      "learning_rate": 0.00141,
      "loss": 5.5404,
      "step": 470
    },
    {
      "epoch": 0.00471,
      "grad_norm": 0.7418690809708749,
      "learning_rate": 0.001413,
      "loss": 5.5115,
      "step": 471
    },
    {
      "epoch": 0.00472,
      "grad_norm": 0.7355325431039438,
      "learning_rate": 0.001416,
      "loss": 5.5216,
      "step": 472
    },
    {
      "epoch": 0.00473,
      "grad_norm": 0.709026539269664,
      "learning_rate": 0.001419,
      "loss": 5.5305,
      "step": 473
    },
    {
      "epoch": 0.00474,
      "grad_norm": 0.5742329758009745,
      "learning_rate": 0.0014219999999999999,
      "loss": 5.5064,
      "step": 474
    },
    {
      "epoch": 0.00475,
      "grad_norm": 0.5859758403725885,
      "learning_rate": 0.001425,
      "loss": 5.4971,
      "step": 475
    },
    {
      "epoch": 0.00476,
      "grad_norm": 0.6365922795308678,
      "learning_rate": 0.001428,
      "loss": 5.5308,
      "step": 476
    },
    {
      "epoch": 0.00477,
      "grad_norm": 0.6539516343537074,
      "learning_rate": 0.001431,
      "loss": 5.4846,
      "step": 477
    },
    {
      "epoch": 0.00478,
      "grad_norm": 0.6446859909585969,
      "learning_rate": 0.001434,
      "loss": 5.5074,
      "step": 478
    },
    {
      "epoch": 0.00479,
      "grad_norm": 0.7791938725908187,
      "learning_rate": 0.001437,
      "loss": 5.4897,
      "step": 479
    },
    {
      "epoch": 0.0048,
      "grad_norm": 0.7608695568360718,
      "learning_rate": 0.0014399999999999999,
      "loss": 5.4855,
      "step": 480
    },
    {
      "epoch": 0.00481,
      "grad_norm": 0.5435552532069989,
      "learning_rate": 0.001443,
      "loss": 5.4813,
      "step": 481
    },
    {
      "epoch": 0.00482,
      "grad_norm": 0.4961185149512517,
      "learning_rate": 0.001446,
      "loss": 5.4538,
      "step": 482
    },
    {
      "epoch": 0.00483,
      "grad_norm": 0.5120902463904886,
      "learning_rate": 0.001449,
      "loss": 5.4636,
      "step": 483
    },
    {
      "epoch": 0.00484,
      "grad_norm": 0.418197369903841,
      "learning_rate": 0.001452,
      "loss": 5.464,
      "step": 484
    },
    {
      "epoch": 0.00485,
      "grad_norm": 0.36311406822078424,
      "learning_rate": 0.001455,
      "loss": 5.4671,
      "step": 485
    },
    {
      "epoch": 0.00486,
      "grad_norm": 0.3913777576995821,
      "learning_rate": 0.001458,
      "loss": 5.4393,
      "step": 486
    },
    {
      "epoch": 0.00487,
      "grad_norm": 0.36874474197662527,
      "learning_rate": 0.001461,
      "loss": 5.4517,
      "step": 487
    },
    {
      "epoch": 0.00488,
      "grad_norm": 0.41593519822402414,
      "learning_rate": 0.001464,
      "loss": 5.43,
      "step": 488
    },
    {
      "epoch": 0.00489,
      "grad_norm": 0.4051383375955623,
      "learning_rate": 0.001467,
      "loss": 5.4204,
      "step": 489
    },
    {
      "epoch": 0.0049,
      "grad_norm": 0.4948319932753325,
      "learning_rate": 0.00147,
      "loss": 5.434,
      "step": 490
    },
    {
      "epoch": 0.00491,
      "grad_norm": 0.6887805434617323,
      "learning_rate": 0.001473,
      "loss": 5.4357,
      "step": 491
    },
    {
      "epoch": 0.00492,
      "grad_norm": 0.912047432270828,
      "learning_rate": 0.001476,
      "loss": 5.4432,
      "step": 492
    },
    {
      "epoch": 0.00493,
      "grad_norm": 0.8950009277905591,
      "learning_rate": 0.001479,
      "loss": 5.4415,
      "step": 493
    },
    {
      "epoch": 0.00494,
      "grad_norm": 0.807683799932231,
      "learning_rate": 0.001482,
      "loss": 5.4427,
      "step": 494
    },
    {
      "epoch": 0.00495,
      "grad_norm": 0.976722169908224,
      "learning_rate": 0.001485,
      "loss": 5.4601,
      "step": 495
    },
    {
      "epoch": 0.00496,
      "grad_norm": 0.7224780529872387,
      "learning_rate": 0.001488,
      "loss": 5.4314,
      "step": 496
    },
    {
      "epoch": 0.00497,
      "grad_norm": 0.7139613919522917,
      "learning_rate": 0.001491,
      "loss": 5.4172,
      "step": 497
    },
    {
      "epoch": 0.00498,
      "grad_norm": 0.7556637282468179,
      "learning_rate": 0.001494,
      "loss": 5.4443,
      "step": 498
    },
    {
      "epoch": 0.00499,
      "grad_norm": 0.8519321120342865,
      "learning_rate": 0.001497,
      "loss": 5.4223,
      "step": 499
    },
    {
      "epoch": 0.005,
      "grad_norm": 0.9868888032280079,
      "learning_rate": 0.0015,
      "loss": 5.4308,
      "step": 500
    },
    {
      "epoch": 0.00501,
      "grad_norm": 1.2028724714017198,
      "learning_rate": 0.001503,
      "loss": 5.4458,
      "step": 501
    },
    {
      "epoch": 0.00502,
      "grad_norm": 0.9548534640519003,
      "learning_rate": 0.001506,
      "loss": 5.4405,
      "step": 502
    },
    {
      "epoch": 0.00503,
      "grad_norm": 0.9423651398338494,
      "learning_rate": 0.0015090000000000001,
      "loss": 5.4484,
      "step": 503
    },
    {
      "epoch": 0.00504,
      "grad_norm": 0.9672711526274779,
      "learning_rate": 0.001512,
      "loss": 5.4328,
      "step": 504
    },
    {
      "epoch": 0.00505,
      "grad_norm": 0.9474889600476256,
      "learning_rate": 0.001515,
      "loss": 5.4283,
      "step": 505
    },
    {
      "epoch": 0.00506,
      "grad_norm": 1.3039451405080307,
      "learning_rate": 0.001518,
      "loss": 5.4432,
      "step": 506
    },
    {
      "epoch": 0.00507,
      "grad_norm": 1.2269782118632737,
      "learning_rate": 0.001521,
      "loss": 5.4452,
      "step": 507
    },
    {
      "epoch": 0.00508,
      "grad_norm": 0.8597426683067237,
      "learning_rate": 0.001524,
      "loss": 5.4268,
      "step": 508
    },
    {
      "epoch": 0.00509,
      "grad_norm": 0.9388657090373522,
      "learning_rate": 0.0015270000000000001,
      "loss": 5.4178,
      "step": 509
    },
    {
      "epoch": 0.0051,
      "grad_norm": 0.9594427813189665,
      "learning_rate": 0.0015300000000000001,
      "loss": 5.4356,
      "step": 510
    },
    {
      "epoch": 0.00511,
      "grad_norm": 1.04563577032056,
      "learning_rate": 0.001533,
      "loss": 5.4212,
      "step": 511
    },
    {
      "epoch": 0.00512,
      "grad_norm": 0.733703407645156,
      "learning_rate": 0.001536,
      "loss": 5.4003,
      "step": 512
    },
    {
      "epoch": 0.00513,
      "grad_norm": 0.8415210942026606,
      "learning_rate": 0.001539,
      "loss": 5.423,
      "step": 513
    },
    {
      "epoch": 0.00514,
      "grad_norm": 0.8791751992621939,
      "learning_rate": 0.001542,
      "loss": 5.4064,
      "step": 514
    },
    {
      "epoch": 0.00515,
      "grad_norm": 0.8161499995578689,
      "learning_rate": 0.0015450000000000001,
      "loss": 5.4094,
      "step": 515
    },
    {
      "epoch": 0.00516,
      "grad_norm": 0.878383695319614,
      "learning_rate": 0.0015480000000000001,
      "loss": 5.4087,
      "step": 516
    },
    {
      "epoch": 0.00517,
      "grad_norm": 0.9768725869756134,
      "learning_rate": 0.001551,
      "loss": 5.4055,
      "step": 517
    },
    {
      "epoch": 0.00518,
      "grad_norm": 0.8865017723772849,
      "learning_rate": 0.001554,
      "loss": 5.3907,
      "step": 518
    },
    {
      "epoch": 0.00519,
      "grad_norm": 0.8308797688973832,
      "learning_rate": 0.001557,
      "loss": 5.3905,
      "step": 519
    },
    {
      "epoch": 0.0052,
      "grad_norm": 0.6978413162257922,
      "learning_rate": 0.0015600000000000002,
      "loss": 5.3938,
      "step": 520
    },
    {
      "epoch": 0.00521,
      "grad_norm": 0.6562689530690187,
      "learning_rate": 0.0015630000000000002,
      "loss": 5.3676,
      "step": 521
    },
    {
      "epoch": 0.00522,
      "grad_norm": 0.5577523148431155,
      "learning_rate": 0.0015660000000000001,
      "loss": 5.3673,
      "step": 522
    },
    {
      "epoch": 0.00523,
      "grad_norm": 0.5298728018270966,
      "learning_rate": 0.001569,
      "loss": 5.3784,
      "step": 523
    },
    {
      "epoch": 0.00524,
      "grad_norm": 0.44216623389663734,
      "learning_rate": 0.001572,
      "loss": 5.3811,
      "step": 524
    },
    {
      "epoch": 0.00525,
      "grad_norm": 0.3702182111689363,
      "learning_rate": 0.001575,
      "loss": 5.369,
      "step": 525
    },
    {
      "epoch": 0.00526,
      "grad_norm": 0.40289552198632295,
      "learning_rate": 0.0015780000000000002,
      "loss": 5.3444,
      "step": 526
    },
    {
      "epoch": 0.00527,
      "grad_norm": 0.38284579814689895,
      "learning_rate": 0.0015810000000000002,
      "loss": 5.3551,
      "step": 527
    },
    {
      "epoch": 0.00528,
      "grad_norm": 0.33950473031510653,
      "learning_rate": 0.0015840000000000001,
      "loss": 5.3339,
      "step": 528
    },
    {
      "epoch": 0.00529,
      "grad_norm": 0.3777758983585419,
      "learning_rate": 0.001587,
      "loss": 5.343,
      "step": 529
    },
    {
      "epoch": 0.0053,
      "grad_norm": 0.4257451161382566,
      "learning_rate": 0.00159,
      "loss": 5.3483,
      "step": 530
    },
    {
      "epoch": 0.00531,
      "grad_norm": 0.5472045550610978,
      "learning_rate": 0.001593,
      "loss": 5.3387,
      "step": 531
    },
    {
      "epoch": 0.00532,
      "grad_norm": 0.696856419507981,
      "learning_rate": 0.0015960000000000002,
      "loss": 5.3455,
      "step": 532
    },
    {
      "epoch": 0.00533,
      "grad_norm": 0.7975941430607876,
      "learning_rate": 0.0015990000000000002,
      "loss": 5.3447,
      "step": 533
    },
    {
      "epoch": 0.00534,
      "grad_norm": 0.6576622452461942,
      "learning_rate": 0.0016020000000000001,
      "loss": 5.3339,
      "step": 534
    },
    {
      "epoch": 0.00535,
      "grad_norm": 0.5879483820814444,
      "learning_rate": 0.001605,
      "loss": 5.3347,
      "step": 535
    },
    {
      "epoch": 0.00536,
      "grad_norm": 0.8212636469840171,
      "learning_rate": 0.001608,
      "loss": 5.3454,
      "step": 536
    },
    {
      "epoch": 0.00537,
      "grad_norm": 0.7740232031924225,
      "learning_rate": 0.0016110000000000002,
      "loss": 5.3418,
      "step": 537
    },
    {
      "epoch": 0.00538,
      "grad_norm": 0.683098129060214,
      "learning_rate": 0.0016140000000000002,
      "loss": 5.3143,
      "step": 538
    },
    {
      "epoch": 0.00539,
      "grad_norm": 0.9170551975741953,
      "learning_rate": 0.0016170000000000002,
      "loss": 5.3245,
      "step": 539
    },
    {
      "epoch": 0.0054,
      "grad_norm": 0.765080131514484,
      "learning_rate": 0.0016200000000000001,
      "loss": 5.3262,
      "step": 540
    },
    {
      "epoch": 0.00541,
      "grad_norm": 0.658457026305436,
      "learning_rate": 0.001623,
      "loss": 5.3137,
      "step": 541
    },
    {
      "epoch": 0.00542,
      "grad_norm": 0.601942869875084,
      "learning_rate": 0.001626,
      "loss": 5.3315,
      "step": 542
    },
    {
      "epoch": 0.00543,
      "grad_norm": 0.6751097730454854,
      "learning_rate": 0.0016290000000000002,
      "loss": 5.2998,
      "step": 543
    },
    {
      "epoch": 0.00544,
      "grad_norm": 0.6943293389301006,
      "learning_rate": 0.0016320000000000002,
      "loss": 5.3191,
      "step": 544
    },
    {
      "epoch": 0.00545,
      "grad_norm": 0.757591621302123,
      "learning_rate": 0.0016350000000000002,
      "loss": 5.3293,
      "step": 545
    },
    {
      "epoch": 0.00546,
      "grad_norm": 0.825685624372282,
      "learning_rate": 0.0016380000000000001,
      "loss": 5.3061,
      "step": 546
    },
    {
      "epoch": 0.00547,
      "grad_norm": 0.8411824339962438,
      "learning_rate": 0.001641,
      "loss": 5.3051,
      "step": 547
    },
    {
      "epoch": 0.00548,
      "grad_norm": 0.8225125547688507,
      "learning_rate": 0.001644,
      "loss": 5.303,
      "step": 548
    },
    {
      "epoch": 0.00549,
      "grad_norm": 0.7692066087665821,
      "learning_rate": 0.0016470000000000002,
      "loss": 5.3016,
      "step": 549
    },
    {
      "epoch": 0.0055,
      "grad_norm": 0.7541680263658305,
      "learning_rate": 0.0016500000000000002,
      "loss": 5.3099,
      "step": 550
    },
    {
      "epoch": 0.00551,
      "grad_norm": 0.8632990593818363,
      "learning_rate": 0.0016530000000000002,
      "loss": 5.3123,
      "step": 551
    },
    {
      "epoch": 0.00552,
      "grad_norm": 0.9083583396116485,
      "learning_rate": 0.0016560000000000001,
      "loss": 5.3063,
      "step": 552
    },
    {
      "epoch": 0.00553,
      "grad_norm": 0.9125681148017897,
      "learning_rate": 0.001659,
      "loss": 5.304,
      "step": 553
    },
    {
      "epoch": 0.00554,
      "grad_norm": 0.9360850923631763,
      "learning_rate": 0.0016620000000000003,
      "loss": 5.3021,
      "step": 554
    },
    {
      "epoch": 0.00555,
      "grad_norm": 0.9519012966124738,
      "learning_rate": 0.0016650000000000002,
      "loss": 5.3154,
      "step": 555
    },
    {
      "epoch": 0.00556,
      "grad_norm": 0.9297335237671114,
      "learning_rate": 0.0016680000000000002,
      "loss": 5.2962,
      "step": 556
    },
    {
      "epoch": 0.00557,
      "grad_norm": 1.1187109821510643,
      "learning_rate": 0.0016710000000000002,
      "loss": 5.2935,
      "step": 557
    },
    {
      "epoch": 0.00558,
      "grad_norm": 0.9868599839032521,
      "learning_rate": 0.0016740000000000001,
      "loss": 5.3219,
      "step": 558
    },
    {
      "epoch": 0.00559,
      "grad_norm": 0.8786610310513777,
      "learning_rate": 0.001677,
      "loss": 5.2938,
      "step": 559
    },
    {
      "epoch": 0.0056,
      "grad_norm": 0.9118163270219521,
      "learning_rate": 0.0016800000000000003,
      "loss": 5.2767,
      "step": 560
    },
    {
      "epoch": 0.00561,
      "grad_norm": 0.8649386991384733,
      "learning_rate": 0.0016830000000000003,
      "loss": 5.3004,
      "step": 561
    },
    {
      "epoch": 0.00562,
      "grad_norm": 0.5982058914442256,
      "learning_rate": 0.0016860000000000002,
      "loss": 5.279,
      "step": 562
    },
    {
      "epoch": 0.00563,
      "grad_norm": 0.5830709316445766,
      "learning_rate": 0.001689,
      "loss": 5.2821,
      "step": 563
    },
    {
      "epoch": 0.00564,
      "grad_norm": 0.5564672250442253,
      "learning_rate": 0.001692,
      "loss": 5.2582,
      "step": 564
    },
    {
      "epoch": 0.00565,
      "grad_norm": 0.5903966881939692,
      "learning_rate": 0.001695,
      "loss": 5.2619,
      "step": 565
    },
    {
      "epoch": 0.00566,
      "grad_norm": 0.5777761648359326,
      "learning_rate": 0.0016979999999999999,
      "loss": 5.2552,
      "step": 566
    },
    {
      "epoch": 0.00567,
      "grad_norm": 0.6069030603134064,
      "learning_rate": 0.0017009999999999998,
      "loss": 5.2491,
      "step": 567
    },
    {
      "epoch": 0.00568,
      "grad_norm": 0.5693132087018719,
      "learning_rate": 0.0017039999999999998,
      "loss": 5.2604,
      "step": 568
    },
    {
      "epoch": 0.00569,
      "grad_norm": 0.47662406778838745,
      "learning_rate": 0.001707,
      "loss": 5.2359,
      "step": 569
    },
    {
      "epoch": 0.0057,
      "grad_norm": 0.49231021705037487,
      "learning_rate": 0.00171,
      "loss": 5.253,
      "step": 570
    },
    {
      "epoch": 0.00571,
      "grad_norm": 0.4167352661720621,
      "learning_rate": 0.001713,
      "loss": 5.2491,
      "step": 571
    },
    {
      "epoch": 0.00572,
      "grad_norm": 0.39011854138215074,
      "learning_rate": 0.0017159999999999999,
      "loss": 5.2258,
      "step": 572
    },
    {
      "epoch": 0.00573,
      "grad_norm": 0.45971389455263184,
      "learning_rate": 0.0017189999999999998,
      "loss": 5.2313,
      "step": 573
    },
    {
      "epoch": 0.00574,
      "grad_norm": 0.4643694238461635,
      "learning_rate": 0.001722,
      "loss": 5.2399,
      "step": 574
    },
    {
      "epoch": 0.00575,
      "grad_norm": 0.4439840100405838,
      "learning_rate": 0.001725,
      "loss": 5.2168,
      "step": 575
    },
    {
      "epoch": 0.00576,
      "grad_norm": 0.4794795740566699,
      "learning_rate": 0.001728,
      "loss": 5.2109,
      "step": 576
    },
    {
      "epoch": 0.00577,
      "grad_norm": 0.5445912068882687,
      "learning_rate": 0.001731,
      "loss": 5.2354,
      "step": 577
    },
    {
      "epoch": 0.00578,
      "grad_norm": 0.5881619532039025,
      "learning_rate": 0.0017339999999999999,
      "loss": 5.2156,
      "step": 578
    },
    {
      "epoch": 0.00579,
      "grad_norm": 0.6688964155004951,
      "learning_rate": 0.0017369999999999998,
      "loss": 5.2064,
      "step": 579
    },
    {
      "epoch": 0.0058,
      "grad_norm": 0.6446644513052245,
      "learning_rate": 0.00174,
      "loss": 5.2223,
      "step": 580
    },
    {
      "epoch": 0.00581,
      "grad_norm": 0.7385299343409043,
      "learning_rate": 0.001743,
      "loss": 5.2222,
      "step": 581
    },
    {
      "epoch": 0.00582,
      "grad_norm": 1.0486135149054512,
      "learning_rate": 0.001746,
      "loss": 5.2114,
      "step": 582
    },
    {
      "epoch": 0.00583,
      "grad_norm": 0.9184260585056472,
      "learning_rate": 0.001749,
      "loss": 5.2374,
      "step": 583
    },
    {
      "epoch": 0.00584,
      "grad_norm": 0.7724931135788974,
      "learning_rate": 0.0017519999999999999,
      "loss": 5.2235,
      "step": 584
    },
    {
      "epoch": 0.00585,
      "grad_norm": 0.903942587317279,
      "learning_rate": 0.0017549999999999998,
      "loss": 5.2218,
      "step": 585
    },
    {
      "epoch": 0.00586,
      "grad_norm": 0.8496888678331875,
      "learning_rate": 0.001758,
      "loss": 5.2272,
      "step": 586
    },
    {
      "epoch": 0.00587,
      "grad_norm": 0.8580070219006531,
      "learning_rate": 0.001761,
      "loss": 5.2094,
      "step": 587
    },
    {
      "epoch": 0.00588,
      "grad_norm": 1.0100984711915582,
      "learning_rate": 0.001764,
      "loss": 5.233,
      "step": 588
    },
    {
      "epoch": 0.00589,
      "grad_norm": 0.9311738464832717,
      "learning_rate": 0.001767,
      "loss": 5.2289,
      "step": 589
    },
    {
      "epoch": 0.0059,
      "grad_norm": 0.9363758859588256,
      "learning_rate": 0.0017699999999999999,
      "loss": 5.2155,
      "step": 590
    },
    {
      "epoch": 0.00591,
      "grad_norm": 0.9695025163620248,
      "learning_rate": 0.001773,
      "loss": 5.2165,
      "step": 591
    },
    {
      "epoch": 0.00592,
      "grad_norm": 0.9805216711841677,
      "learning_rate": 0.001776,
      "loss": 5.2417,
      "step": 592
    },
    {
      "epoch": 0.00593,
      "grad_norm": 0.9827241162755458,
      "learning_rate": 0.001779,
      "loss": 5.2392,
      "step": 593
    },
    {
      "epoch": 0.00594,
      "grad_norm": 1.1445433178010636,
      "learning_rate": 0.001782,
      "loss": 5.2522,
      "step": 594
    },
    {
      "epoch": 0.00595,
      "grad_norm": 0.950699335886284,
      "learning_rate": 0.001785,
      "loss": 5.2351,
      "step": 595
    },
    {
      "epoch": 0.00596,
      "grad_norm": 0.8373647963925929,
      "learning_rate": 0.0017879999999999999,
      "loss": 5.2269,
      "step": 596
    },
    {
      "epoch": 0.00597,
      "grad_norm": 0.9799263768822638,
      "learning_rate": 0.001791,
      "loss": 5.2118,
      "step": 597
    },
    {
      "epoch": 0.00598,
      "grad_norm": 0.9413379648187816,
      "learning_rate": 0.001794,
      "loss": 5.2215,
      "step": 598
    },
    {
      "epoch": 0.00599,
      "grad_norm": 0.8392983563516706,
      "learning_rate": 0.001797,
      "loss": 5.2191,
      "step": 599
    },
    {
      "epoch": 0.006,
      "grad_norm": 0.8414286288934597,
      "learning_rate": 0.0018,
      "loss": 5.2134,
      "step": 600
    },
    {
      "epoch": 0.00601,
      "grad_norm": 0.8993201270126511,
      "learning_rate": 0.001803,
      "loss": 5.2089,
      "step": 601
    },
    {
      "epoch": 0.00602,
      "grad_norm": 0.8678008919316418,
      "learning_rate": 0.0018059999999999999,
      "loss": 5.1957,
      "step": 602
    },
    {
      "epoch": 0.00603,
      "grad_norm": 0.6967382011990112,
      "learning_rate": 0.001809,
      "loss": 5.2122,
      "step": 603
    },
    {
      "epoch": 0.00604,
      "grad_norm": 0.5312401482691863,
      "learning_rate": 0.001812,
      "loss": 5.1788,
      "step": 604
    },
    {
      "epoch": 0.00605,
      "grad_norm": 0.5795052729398412,
      "learning_rate": 0.001815,
      "loss": 5.1915,
      "step": 605
    },
    {
      "epoch": 0.00606,
      "grad_norm": 0.6345039250131549,
      "learning_rate": 0.001818,
      "loss": 5.1894,
      "step": 606
    },
    {
      "epoch": 0.00607,
      "grad_norm": 0.6716049737606092,
      "learning_rate": 0.001821,
      "loss": 5.1916,
      "step": 607
    },
    {
      "epoch": 0.00608,
      "grad_norm": 0.631863512626499,
      "learning_rate": 0.001824,
      "loss": 5.1762,
      "step": 608
    },
    {
      "epoch": 0.00609,
      "grad_norm": 0.5345713642167306,
      "learning_rate": 0.001827,
      "loss": 5.177,
      "step": 609
    },
    {
      "epoch": 0.0061,
      "grad_norm": 0.455626202663485,
      "learning_rate": 0.00183,
      "loss": 5.1434,
      "step": 610
    },
    {
      "epoch": 0.00611,
      "grad_norm": 0.46197362206927406,
      "learning_rate": 0.001833,
      "loss": 5.1489,
      "step": 611
    },
    {
      "epoch": 0.00612,
      "grad_norm": 0.3943653846553885,
      "learning_rate": 0.001836,
      "loss": 5.1488,
      "step": 612
    },
    {
      "epoch": 0.00613,
      "grad_norm": 0.4053887102586243,
      "learning_rate": 0.001839,
      "loss": 5.1409,
      "step": 613
    },
    {
      "epoch": 0.00614,
      "grad_norm": 0.44363743610311057,
      "learning_rate": 0.001842,
      "loss": 5.1611,
      "step": 614
    },
    {
      "epoch": 0.00615,
      "grad_norm": 0.40025757691173514,
      "learning_rate": 0.001845,
      "loss": 5.145,
      "step": 615
    },
    {
      "epoch": 0.00616,
      "grad_norm": 0.4067321452673224,
      "learning_rate": 0.001848,
      "loss": 5.1299,
      "step": 616
    },
    {
      "epoch": 0.00617,
      "grad_norm": 0.42403873536126996,
      "learning_rate": 0.001851,
      "loss": 5.1397,
      "step": 617
    },
    {
      "epoch": 0.00618,
      "grad_norm": 0.44192084147381183,
      "learning_rate": 0.001854,
      "loss": 5.1265,
      "step": 618
    },
    {
      "epoch": 0.00619,
      "grad_norm": 0.5104655865853382,
      "learning_rate": 0.001857,
      "loss": 5.1299,
      "step": 619
    },
    {
      "epoch": 0.0062,
      "grad_norm": 0.6226145292933671,
      "learning_rate": 0.00186,
      "loss": 5.118,
      "step": 620
    },
    {
      "epoch": 0.00621,
      "grad_norm": 0.745237667035521,
      "learning_rate": 0.001863,
      "loss": 5.1237,
      "step": 621
    },
    {
      "epoch": 0.00622,
      "grad_norm": 0.7986865810762169,
      "learning_rate": 0.001866,
      "loss": 5.1267,
      "step": 622
    },
    {
      "epoch": 0.00623,
      "grad_norm": 0.7723208155309241,
      "learning_rate": 0.001869,
      "loss": 5.1084,
      "step": 623
    },
    {
      "epoch": 0.00624,
      "grad_norm": 0.8171878766946273,
      "learning_rate": 0.001872,
      "loss": 5.1246,
      "step": 624
    },
    {
      "epoch": 0.00625,
      "grad_norm": 0.7607424856276187,
      "learning_rate": 0.001875,
      "loss": 5.1359,
      "step": 625
    },
    {
      "epoch": 0.00626,
      "grad_norm": 0.7005729458198662,
      "learning_rate": 0.0018780000000000001,
      "loss": 5.1104,
      "step": 626
    },
    {
      "epoch": 0.00627,
      "grad_norm": 0.6735260479679158,
      "learning_rate": 0.001881,
      "loss": 5.1015,
      "step": 627
    },
    {
      "epoch": 0.00628,
      "grad_norm": 0.780486998163919,
      "learning_rate": 0.001884,
      "loss": 5.1349,
      "step": 628
    },
    {
      "epoch": 0.00629,
      "grad_norm": 0.8746830311438225,
      "learning_rate": 0.001887,
      "loss": 5.1097,
      "step": 629
    },
    {
      "epoch": 0.0063,
      "grad_norm": 0.9536711950620466,
      "learning_rate": 0.00189,
      "loss": 5.1137,
      "step": 630
    },
    {
      "epoch": 0.00631,
      "grad_norm": 0.8628106812141149,
      "learning_rate": 0.0018930000000000002,
      "loss": 5.1035,
      "step": 631
    },
    {
      "epoch": 0.00632,
      "grad_norm": 0.9320890333045916,
      "learning_rate": 0.0018960000000000001,
      "loss": 5.1281,
      "step": 632
    },
    {
      "epoch": 0.00633,
      "grad_norm": 1.0593968691082751,
      "learning_rate": 0.001899,
      "loss": 5.1321,
      "step": 633
    },
    {
      "epoch": 0.00634,
      "grad_norm": 0.943443350801409,
      "learning_rate": 0.001902,
      "loss": 5.1182,
      "step": 634
    },
    {
      "epoch": 0.00635,
      "grad_norm": 0.9628484504873114,
      "learning_rate": 0.001905,
      "loss": 5.1089,
      "step": 635
    },
    {
      "epoch": 0.00636,
      "grad_norm": 1.1043383962750646,
      "learning_rate": 0.001908,
      "loss": 5.1289,
      "step": 636
    },
    {
      "epoch": 0.00637,
      "grad_norm": 0.8987493866500654,
      "learning_rate": 0.0019110000000000002,
      "loss": 5.1351,
      "step": 637
    },
    {
      "epoch": 0.00638,
      "grad_norm": 0.9251804377428581,
      "learning_rate": 0.0019140000000000001,
      "loss": 5.1288,
      "step": 638
    },
    {
      "epoch": 0.00639,
      "grad_norm": 0.854418425044198,
      "learning_rate": 0.001917,
      "loss": 5.0998,
      "step": 639
    },
    {
      "epoch": 0.0064,
      "grad_norm": 0.9324816679284724,
      "learning_rate": 0.00192,
      "loss": 5.1038,
      "step": 640
    },
    {
      "epoch": 0.00641,
      "grad_norm": 0.9892889234371413,
      "learning_rate": 0.001923,
      "loss": 5.1163,
      "step": 641
    },
    {
      "epoch": 0.00642,
      "grad_norm": 1.0346602459121752,
      "learning_rate": 0.001926,
      "loss": 5.106,
      "step": 642
    },
    {
      "epoch": 0.00643,
      "grad_norm": 0.8661994645957561,
      "learning_rate": 0.0019290000000000002,
      "loss": 5.117,
      "step": 643
    },
    {
      "epoch": 0.00644,
      "grad_norm": 0.8724056100423225,
      "learning_rate": 0.0019320000000000001,
      "loss": 5.0889,
      "step": 644
    },
    {
      "epoch": 0.00645,
      "grad_norm": 0.8584186184200229,
      "learning_rate": 0.001935,
      "loss": 5.1004,
      "step": 645
    },
    {
      "epoch": 0.00646,
      "grad_norm": 0.7360558672224548,
      "learning_rate": 0.001938,
      "loss": 5.0955,
      "step": 646
    },
    {
      "epoch": 0.00647,
      "grad_norm": 0.7977702647925389,
      "learning_rate": 0.001941,
      "loss": 5.1058,
      "step": 647
    },
    {
      "epoch": 0.00648,
      "grad_norm": 0.7872116543506851,
      "learning_rate": 0.0019440000000000002,
      "loss": 5.0908,
      "step": 648
    },
    {
      "epoch": 0.00649,
      "grad_norm": 0.7104658813349117,
      "learning_rate": 0.0019470000000000002,
      "loss": 5.0718,
      "step": 649
    },
    {
      "epoch": 0.0065,
      "grad_norm": 0.7453763255239747,
      "learning_rate": 0.0019500000000000001,
      "loss": 5.0953,
      "step": 650
    },
    {
      "epoch": 0.00651,
      "grad_norm": 0.7781624388594444,
      "learning_rate": 0.001953,
      "loss": 5.0758,
      "step": 651
    },
    {
      "epoch": 0.00652,
      "grad_norm": 0.7616046275009601,
      "learning_rate": 0.0019560000000000003,
      "loss": 5.0661,
      "step": 652
    },
    {
      "epoch": 0.00653,
      "grad_norm": 0.5945469625366651,
      "learning_rate": 0.0019590000000000002,
      "loss": 5.0539,
      "step": 653
    },
    {
      "epoch": 0.00654,
      "grad_norm": 0.6024408595794577,
      "learning_rate": 0.001962,
      "loss": 5.0374,
      "step": 654
    },
    {
      "epoch": 0.00655,
      "grad_norm": 0.5905307565923603,
      "learning_rate": 0.001965,
      "loss": 5.048,
      "step": 655
    },
    {
      "epoch": 0.00656,
      "grad_norm": 0.5236322372626927,
      "learning_rate": 0.001968,
      "loss": 5.04,
      "step": 656
    },
    {
      "epoch": 0.00657,
      "grad_norm": 0.5283416618835216,
      "learning_rate": 0.001971,
      "loss": 5.0223,
      "step": 657
    },
    {
      "epoch": 0.00658,
      "grad_norm": 0.5563146586062104,
      "learning_rate": 0.001974,
      "loss": 5.0415,
      "step": 658
    },
    {
      "epoch": 0.00659,
      "grad_norm": 0.6297873363395704,
      "learning_rate": 0.001977,
      "loss": 5.0241,
      "step": 659
    },
    {
      "epoch": 0.0066,
      "grad_norm": 0.5780538180580159,
      "learning_rate": 0.00198,
      "loss": 5.0197,
      "step": 660
    },
    {
      "epoch": 0.00661,
      "grad_norm": 0.5505266007864265,
      "learning_rate": 0.001983,
      "loss": 4.9938,
      "step": 661
    },
    {
      "epoch": 0.00662,
      "grad_norm": 0.6077995116545319,
      "learning_rate": 0.0019860000000000004,
      "loss": 4.9946,
      "step": 662
    },
    {
      "epoch": 0.00663,
      "grad_norm": 0.693544741809416,
      "learning_rate": 0.0019890000000000003,
      "loss": 5.01,
      "step": 663
    },
    {
      "epoch": 0.00664,
      "grad_norm": 0.801492715102265,
      "learning_rate": 0.0019920000000000003,
      "loss": 5.0017,
      "step": 664
    },
    {
      "epoch": 0.00665,
      "grad_norm": 0.8803866066519176,
      "learning_rate": 0.0019950000000000002,
      "loss": 4.9845,
      "step": 665
    },
    {
      "epoch": 0.00666,
      "grad_norm": 0.9133314823416234,
      "learning_rate": 0.001998,
      "loss": 5.0048,
      "step": 666
    },
    {
      "epoch": 0.00667,
      "grad_norm": 0.9006055773427947,
      "learning_rate": 0.002001,
      "loss": 4.9967,
      "step": 667
    },
    {
      "epoch": 0.00668,
      "grad_norm": 0.8268536663020751,
      "learning_rate": 0.002004,
      "loss": 5.0164,
      "step": 668
    },
    {
      "epoch": 0.00669,
      "grad_norm": 1.1034218430158187,
      "learning_rate": 0.002007,
      "loss": 5.0358,
      "step": 669
    },
    {
      "epoch": 0.0067,
      "grad_norm": 1.0710112933622913,
      "learning_rate": 0.00201,
      "loss": 5.0286,
      "step": 670
    },
    {
      "epoch": 0.00671,
      "grad_norm": 1.0384170566197124,
      "learning_rate": 0.002013,
      "loss": 5.0085,
      "step": 671
    },
    {
      "epoch": 0.00672,
      "grad_norm": 0.9247071512184438,
      "learning_rate": 0.002016,
      "loss": 4.9973,
      "step": 672
    },
    {
      "epoch": 0.00673,
      "grad_norm": 0.9905174678816935,
      "learning_rate": 0.002019,
      "loss": 5.0151,
      "step": 673
    },
    {
      "epoch": 0.00674,
      "grad_norm": 0.8930692065183647,
      "learning_rate": 0.0020220000000000004,
      "loss": 5.0014,
      "step": 674
    },
    {
      "epoch": 0.00675,
      "grad_norm": 1.0067908289229996,
      "learning_rate": 0.0020250000000000003,
      "loss": 5.0182,
      "step": 675
    },
    {
      "epoch": 0.00676,
      "grad_norm": 0.972209366764115,
      "learning_rate": 0.0020280000000000003,
      "loss": 5.0082,
      "step": 676
    },
    {
      "epoch": 0.00677,
      "grad_norm": 1.0440090424594235,
      "learning_rate": 0.0020310000000000003,
      "loss": 5.0109,
      "step": 677
    },
    {
      "epoch": 0.00678,
      "grad_norm": 1.1292207401563255,
      "learning_rate": 0.0020340000000000002,
      "loss": 5.0278,
      "step": 678
    },
    {
      "epoch": 0.00679,
      "grad_norm": 0.9872089241398577,
      "learning_rate": 0.002037,
      "loss": 5.0054,
      "step": 679
    },
    {
      "epoch": 0.0068,
      "grad_norm": 0.8748535789650627,
      "learning_rate": 0.00204,
      "loss": 5.0004,
      "step": 680
    },
    {
      "epoch": 0.00681,
      "grad_norm": 1.0039141852392888,
      "learning_rate": 0.002043,
      "loss": 5.0018,
      "step": 681
    },
    {
      "epoch": 0.00682,
      "grad_norm": 1.0414085118011525,
      "learning_rate": 0.002046,
      "loss": 5.0026,
      "step": 682
    },
    {
      "epoch": 0.00683,
      "grad_norm": 0.8083819733286619,
      "learning_rate": 0.002049,
      "loss": 4.9746,
      "step": 683
    },
    {
      "epoch": 0.00684,
      "grad_norm": 0.618682103661302,
      "learning_rate": 0.002052,
      "loss": 4.9803,
      "step": 684
    },
    {
      "epoch": 0.00685,
      "grad_norm": 0.5775508276061283,
      "learning_rate": 0.0020550000000000004,
      "loss": 4.9792,
      "step": 685
    },
    {
      "epoch": 0.00686,
      "grad_norm": 0.5359292821422027,
      "learning_rate": 0.0020580000000000004,
      "loss": 4.9508,
      "step": 686
    },
    {
      "epoch": 0.00687,
      "grad_norm": 0.5439874489161504,
      "learning_rate": 0.0020610000000000003,
      "loss": 4.9456,
      "step": 687
    },
    {
      "epoch": 0.00688,
      "grad_norm": 0.6521058673701751,
      "learning_rate": 0.002064,
      "loss": 4.9472,
      "step": 688
    },
    {
      "epoch": 0.00689,
      "grad_norm": 0.7201992210148584,
      "learning_rate": 0.002067,
      "loss": 4.9514,
      "step": 689
    },
    {
      "epoch": 0.0069,
      "grad_norm": 0.6329359289093167,
      "learning_rate": 0.00207,
      "loss": 4.9393,
      "step": 690
    },
    {
      "epoch": 0.00691,
      "grad_norm": 0.5452341441548267,
      "learning_rate": 0.0020729999999999998,
      "loss": 4.9065,
      "step": 691
    },
    {
      "epoch": 0.00692,
      "grad_norm": 0.524421270821715,
      "learning_rate": 0.0020759999999999997,
      "loss": 4.9264,
      "step": 692
    },
    {
      "epoch": 0.00693,
      "grad_norm": 0.5237198583423548,
      "learning_rate": 0.0020789999999999997,
      "loss": 4.9275,
      "step": 693
    },
    {
      "epoch": 0.00694,
      "grad_norm": 0.5943284166970615,
      "learning_rate": 0.002082,
      "loss": 4.9065,
      "step": 694
    },
    {
      "epoch": 0.00695,
      "grad_norm": 0.6331066584123409,
      "learning_rate": 0.002085,
      "loss": 4.9096,
      "step": 695
    },
    {
      "epoch": 0.00696,
      "grad_norm": 0.6552141599381052,
      "learning_rate": 0.002088,
      "loss": 4.9103,
      "step": 696
    },
    {
      "epoch": 0.00697,
      "grad_norm": 0.7459258980378775,
      "learning_rate": 0.002091,
      "loss": 4.9188,
      "step": 697
    },
    {
      "epoch": 0.00698,
      "grad_norm": 0.902413176094811,
      "learning_rate": 0.002094,
      "loss": 4.9109,
      "step": 698
    },
    {
      "epoch": 0.00699,
      "grad_norm": 0.9488326760888935,
      "learning_rate": 0.002097,
      "loss": 4.9116,
      "step": 699
    },
    {
      "epoch": 0.007,
      "grad_norm": 0.9189763967394823,
      "learning_rate": 0.0021,
      "loss": 4.929,
      "step": 700
    },
    {
      "epoch": 0.00701,
      "grad_norm": 0.8924436046521577,
      "learning_rate": 0.002103,
      "loss": 4.9083,
      "step": 701
    },
    {
      "epoch": 0.00702,
      "grad_norm": 0.8863614629240012,
      "learning_rate": 0.002106,
      "loss": 4.9092,
      "step": 702
    },
    {
      "epoch": 0.00703,
      "grad_norm": 0.7940709541538681,
      "learning_rate": 0.0021089999999999998,
      "loss": 4.891,
      "step": 703
    },
    {
      "epoch": 0.00704,
      "grad_norm": 0.7939787736751149,
      "learning_rate": 0.0021119999999999997,
      "loss": 4.8964,
      "step": 704
    },
    {
      "epoch": 0.00705,
      "grad_norm": 1.0444656030359551,
      "learning_rate": 0.002115,
      "loss": 4.9151,
      "step": 705
    },
    {
      "epoch": 0.00706,
      "grad_norm": 1.1431311909042268,
      "learning_rate": 0.002118,
      "loss": 4.9234,
      "step": 706
    },
    {
      "epoch": 0.00707,
      "grad_norm": 0.8384635334186645,
      "learning_rate": 0.002121,
      "loss": 4.8902,
      "step": 707
    },
    {
      "epoch": 0.00708,
      "grad_norm": 0.9594405079672866,
      "learning_rate": 0.002124,
      "loss": 4.8745,
      "step": 708
    },
    {
      "epoch": 0.00709,
      "grad_norm": 0.8900382869322284,
      "learning_rate": 0.002127,
      "loss": 4.9161,
      "step": 709
    },
    {
      "epoch": 0.0071,
      "grad_norm": 0.8868570415283396,
      "learning_rate": 0.00213,
      "loss": 4.9065,
      "step": 710
    },
    {
      "epoch": 0.00711,
      "grad_norm": 0.8610490944817158,
      "learning_rate": 0.002133,
      "loss": 4.8804,
      "step": 711
    },
    {
      "epoch": 0.00712,
      "grad_norm": 0.8727164938852855,
      "learning_rate": 0.002136,
      "loss": 4.9046,
      "step": 712
    },
    {
      "epoch": 0.00713,
      "grad_norm": 0.8857525308493206,
      "learning_rate": 0.002139,
      "loss": 4.9135,
      "step": 713
    },
    {
      "epoch": 0.00714,
      "grad_norm": 0.9495661806955594,
      "learning_rate": 0.002142,
      "loss": 4.918,
      "step": 714
    },
    {
      "epoch": 0.00715,
      "grad_norm": 1.0263024097609161,
      "learning_rate": 0.0021449999999999998,
      "loss": 4.8857,
      "step": 715
    },
    {
      "epoch": 0.00716,
      "grad_norm": 0.8876358680026493,
      "learning_rate": 0.002148,
      "loss": 4.8749,
      "step": 716
    },
    {
      "epoch": 0.00717,
      "grad_norm": 0.8225498605776377,
      "learning_rate": 0.002151,
      "loss": 4.8925,
      "step": 717
    },
    {
      "epoch": 0.00718,
      "grad_norm": 0.628552485265691,
      "learning_rate": 0.002154,
      "loss": 4.8659,
      "step": 718
    },
    {
      "epoch": 0.00719,
      "grad_norm": 0.6584104654465238,
      "learning_rate": 0.002157,
      "loss": 4.8747,
      "step": 719
    },
    {
      "epoch": 0.0072,
      "grad_norm": 0.6698592474865601,
      "learning_rate": 0.00216,
      "loss": 4.8635,
      "step": 720
    },
    {
      "epoch": 0.00721,
      "grad_norm": 0.6673590176314685,
      "learning_rate": 0.002163,
      "loss": 4.8639,
      "step": 721
    },
    {
      "epoch": 0.00722,
      "grad_norm": 0.6674098225397388,
      "learning_rate": 0.002166,
      "loss": 4.8386,
      "step": 722
    },
    {
      "epoch": 0.00723,
      "grad_norm": 0.6090726175552883,
      "learning_rate": 0.002169,
      "loss": 4.8464,
      "step": 723
    },
    {
      "epoch": 0.00724,
      "grad_norm": 0.6325507361418539,
      "learning_rate": 0.002172,
      "loss": 4.8403,
      "step": 724
    },
    {
      "epoch": 0.00725,
      "grad_norm": 0.6927587431932604,
      "learning_rate": 0.002175,
      "loss": 4.8341,
      "step": 725
    },
    {
      "epoch": 0.00726,
      "grad_norm": 0.7422551683158218,
      "learning_rate": 0.002178,
      "loss": 4.8448,
      "step": 726
    },
    {
      "epoch": 0.00727,
      "grad_norm": 0.7946686392459241,
      "learning_rate": 0.0021809999999999998,
      "loss": 4.8264,
      "step": 727
    },
    {
      "epoch": 0.00728,
      "grad_norm": 0.651194780867581,
      "learning_rate": 0.002184,
      "loss": 4.8373,
      "step": 728
    },
    {
      "epoch": 0.00729,
      "grad_norm": 0.5507866158426874,
      "learning_rate": 0.002187,
      "loss": 4.8279,
      "step": 729
    },
    {
      "epoch": 0.0073,
      "grad_norm": 0.5770531279665235,
      "learning_rate": 0.00219,
      "loss": 4.8256,
      "step": 730
    },
    {
      "epoch": 0.00731,
      "grad_norm": 0.6604772562967653,
      "learning_rate": 0.002193,
      "loss": 4.8198,
      "step": 731
    },
    {
      "epoch": 0.00732,
      "grad_norm": 0.7902754963422924,
      "learning_rate": 0.002196,
      "loss": 4.832,
      "step": 732
    },
    {
      "epoch": 0.00733,
      "grad_norm": 0.8530754775219535,
      "learning_rate": 0.002199,
      "loss": 4.814,
      "step": 733
    },
    {
      "epoch": 0.00734,
      "grad_norm": 0.808835939559151,
      "learning_rate": 0.002202,
      "loss": 4.8365,
      "step": 734
    },
    {
      "epoch": 0.00735,
      "grad_norm": 0.7793455630355729,
      "learning_rate": 0.002205,
      "loss": 4.8484,
      "step": 735
    },
    {
      "epoch": 0.00736,
      "grad_norm": 0.8899384636665534,
      "learning_rate": 0.002208,
      "loss": 4.8322,
      "step": 736
    },
    {
      "epoch": 0.00737,
      "grad_norm": 1.0166382867407526,
      "learning_rate": 0.002211,
      "loss": 4.8145,
      "step": 737
    },
    {
      "epoch": 0.00738,
      "grad_norm": 0.9857126416807526,
      "learning_rate": 0.002214,
      "loss": 4.8213,
      "step": 738
    },
    {
      "epoch": 0.00739,
      "grad_norm": 0.8982016923721464,
      "learning_rate": 0.0022170000000000002,
      "loss": 4.809,
      "step": 739
    },
    {
      "epoch": 0.0074,
      "grad_norm": 0.8517808259905928,
      "learning_rate": 0.00222,
      "loss": 4.8138,
      "step": 740
    },
    {
      "epoch": 0.00741,
      "grad_norm": 0.6571636698582773,
      "learning_rate": 0.002223,
      "loss": 4.778,
      "step": 741
    },
    {
      "epoch": 0.00742,
      "grad_norm": 0.5983530041008951,
      "learning_rate": 0.002226,
      "loss": 4.8043,
      "step": 742
    },
    {
      "epoch": 0.00743,
      "grad_norm": 0.613767022264535,
      "learning_rate": 0.002229,
      "loss": 4.772,
      "step": 743
    },
    {
      "epoch": 0.00744,
      "grad_norm": 0.5746332772613801,
      "learning_rate": 0.002232,
      "loss": 4.7755,
      "step": 744
    },
    {
      "epoch": 0.00745,
      "grad_norm": 0.5491916195482714,
      "learning_rate": 0.002235,
      "loss": 4.7792,
      "step": 745
    },
    {
      "epoch": 0.00746,
      "grad_norm": 0.6224704559098753,
      "learning_rate": 0.002238,
      "loss": 4.7785,
      "step": 746
    },
    {
      "epoch": 0.00747,
      "grad_norm": 0.721423929849636,
      "learning_rate": 0.002241,
      "loss": 4.766,
      "step": 747
    },
    {
      "epoch": 0.00748,
      "grad_norm": 0.8211263973312402,
      "learning_rate": 0.002244,
      "loss": 4.7825,
      "step": 748
    },
    {
      "epoch": 0.00749,
      "grad_norm": 0.9630311220772746,
      "learning_rate": 0.002247,
      "loss": 4.7687,
      "step": 749
    },
    {
      "epoch": 0.0075,
      "grad_norm": 1.0671208539128567,
      "learning_rate": 0.0022500000000000003,
      "loss": 4.8201,
      "step": 750
    },
    {
      "epoch": 0.00751,
      "grad_norm": 1.0285377005887373,
      "learning_rate": 0.0022530000000000002,
      "loss": 4.8053,
      "step": 751
    },
    {
      "epoch": 0.00752,
      "grad_norm": 0.9937387135055332,
      "learning_rate": 0.002256,
      "loss": 4.7635,
      "step": 752
    },
    {
      "epoch": 0.00753,
      "grad_norm": 0.8939099354397223,
      "learning_rate": 0.002259,
      "loss": 4.8048,
      "step": 753
    },
    {
      "epoch": 0.00754,
      "grad_norm": 0.958657384547811,
      "learning_rate": 0.002262,
      "loss": 4.816,
      "step": 754
    },
    {
      "epoch": 0.00755,
      "grad_norm": 0.8579165829387244,
      "learning_rate": 0.002265,
      "loss": 4.7809,
      "step": 755
    },
    {
      "epoch": 0.00756,
      "grad_norm": 0.8147493051985796,
      "learning_rate": 0.002268,
      "loss": 4.7687,
      "step": 756
    },
    {
      "epoch": 0.00757,
      "grad_norm": 0.9899674342411924,
      "learning_rate": 0.002271,
      "loss": 4.7845,
      "step": 757
    },
    {
      "epoch": 0.00758,
      "grad_norm": 1.2327208130150207,
      "learning_rate": 0.002274,
      "loss": 4.7967,
      "step": 758
    },
    {
      "epoch": 0.00759,
      "grad_norm": 0.8760473410677282,
      "learning_rate": 0.002277,
      "loss": 4.7951,
      "step": 759
    },
    {
      "epoch": 0.0076,
      "grad_norm": 0.9553393765090987,
      "learning_rate": 0.00228,
      "loss": 4.7842,
      "step": 760
    },
    {
      "epoch": 0.00761,
      "grad_norm": 1.088855928225056,
      "learning_rate": 0.002283,
      "loss": 4.769,
      "step": 761
    },
    {
      "epoch": 0.00762,
      "grad_norm": 0.9818480683263884,
      "learning_rate": 0.0022860000000000003,
      "loss": 4.7512,
      "step": 762
    },
    {
      "epoch": 0.00763,
      "grad_norm": 0.9731540924632093,
      "learning_rate": 0.0022890000000000002,
      "loss": 4.7931,
      "step": 763
    },
    {
      "epoch": 0.00764,
      "grad_norm": 1.0508884929557651,
      "learning_rate": 0.002292,
      "loss": 4.8167,
      "step": 764
    },
    {
      "epoch": 0.00765,
      "grad_norm": 1.0020863769727308,
      "learning_rate": 0.002295,
      "loss": 4.7984,
      "step": 765
    },
    {
      "epoch": 0.00766,
      "grad_norm": 1.1527463652354557,
      "learning_rate": 0.002298,
      "loss": 4.8085,
      "step": 766
    },
    {
      "epoch": 0.00767,
      "grad_norm": 0.9657952239159258,
      "learning_rate": 0.002301,
      "loss": 4.7959,
      "step": 767
    },
    {
      "epoch": 0.00768,
      "grad_norm": 1.0234917976922082,
      "learning_rate": 0.002304,
      "loss": 4.8012,
      "step": 768
    },
    {
      "epoch": 0.00769,
      "grad_norm": 0.9850893067060651,
      "learning_rate": 0.002307,
      "loss": 4.8144,
      "step": 769
    },
    {
      "epoch": 0.0077,
      "grad_norm": 0.9062134932024389,
      "learning_rate": 0.00231,
      "loss": 4.7653,
      "step": 770
    },
    {
      "epoch": 0.00771,
      "grad_norm": 0.8476285286232204,
      "learning_rate": 0.002313,
      "loss": 4.7979,
      "step": 771
    },
    {
      "epoch": 0.00772,
      "grad_norm": 0.9122213123018311,
      "learning_rate": 0.002316,
      "loss": 4.7851,
      "step": 772
    },
    {
      "epoch": 0.00773,
      "grad_norm": 1.0718910624781612,
      "learning_rate": 0.0023190000000000003,
      "loss": 4.8052,
      "step": 773
    },
    {
      "epoch": 0.00774,
      "grad_norm": 0.7792131883523417,
      "learning_rate": 0.0023220000000000003,
      "loss": 4.7945,
      "step": 774
    },
    {
      "epoch": 0.00775,
      "grad_norm": 0.7995411986928386,
      "learning_rate": 0.0023250000000000002,
      "loss": 4.7914,
      "step": 775
    },
    {
      "epoch": 0.00776,
      "grad_norm": 0.7054590225014301,
      "learning_rate": 0.002328,
      "loss": 4.7883,
      "step": 776
    },
    {
      "epoch": 0.00777,
      "grad_norm": 0.6505869359405926,
      "learning_rate": 0.002331,
      "loss": 4.7585,
      "step": 777
    },
    {
      "epoch": 0.00778,
      "grad_norm": 0.6484695284206986,
      "learning_rate": 0.002334,
      "loss": 4.7652,
      "step": 778
    },
    {
      "epoch": 0.00779,
      "grad_norm": 0.6047799586124271,
      "learning_rate": 0.002337,
      "loss": 4.7239,
      "step": 779
    },
    {
      "epoch": 0.0078,
      "grad_norm": 0.5436502526586032,
      "learning_rate": 0.00234,
      "loss": 4.7364,
      "step": 780
    },
    {
      "epoch": 0.00781,
      "grad_norm": 0.5682167623371829,
      "learning_rate": 0.002343,
      "loss": 4.7355,
      "step": 781
    },
    {
      "epoch": 0.00782,
      "grad_norm": 0.628910610744215,
      "learning_rate": 0.002346,
      "loss": 4.7356,
      "step": 782
    },
    {
      "epoch": 0.00783,
      "grad_norm": 0.5845457532196663,
      "learning_rate": 0.002349,
      "loss": 4.7086,
      "step": 783
    },
    {
      "epoch": 0.00784,
      "grad_norm": 0.6345965495249546,
      "learning_rate": 0.002352,
      "loss": 4.723,
      "step": 784
    },
    {
      "epoch": 0.00785,
      "grad_norm": 0.4959017064429773,
      "learning_rate": 0.0023550000000000003,
      "loss": 4.7138,
      "step": 785
    },
    {
      "epoch": 0.00786,
      "grad_norm": 0.4358915945164792,
      "learning_rate": 0.0023580000000000003,
      "loss": 4.7131,
      "step": 786
    },
    {
      "epoch": 0.00787,
      "grad_norm": 0.4259891850743534,
      "learning_rate": 0.0023610000000000003,
      "loss": 4.7113,
      "step": 787
    },
    {
      "epoch": 0.00788,
      "grad_norm": 0.4529949912379573,
      "learning_rate": 0.002364,
      "loss": 4.7077,
      "step": 788
    },
    {
      "epoch": 0.00789,
      "grad_norm": 0.4944753699230628,
      "learning_rate": 0.002367,
      "loss": 4.7334,
      "step": 789
    },
    {
      "epoch": 0.0079,
      "grad_norm": 0.6295031827770177,
      "learning_rate": 0.00237,
      "loss": 4.7144,
      "step": 790
    },
    {
      "epoch": 0.00791,
      "grad_norm": 0.8474014146600959,
      "learning_rate": 0.002373,
      "loss": 4.6777,
      "step": 791
    },
    {
      "epoch": 0.00792,
      "grad_norm": 0.8040402155973354,
      "learning_rate": 0.002376,
      "loss": 4.7267,
      "step": 792
    },
    {
      "epoch": 0.00793,
      "grad_norm": 0.5568568781614048,
      "learning_rate": 0.002379,
      "loss": 4.7006,
      "step": 793
    },
    {
      "epoch": 0.00794,
      "grad_norm": 0.8220014797505664,
      "learning_rate": 0.002382,
      "loss": 4.7246,
      "step": 794
    },
    {
      "epoch": 0.00795,
      "grad_norm": 0.6551332917875898,
      "learning_rate": 0.002385,
      "loss": 4.6938,
      "step": 795
    },
    {
      "epoch": 0.00796,
      "grad_norm": 0.5666814801389223,
      "learning_rate": 0.0023880000000000004,
      "loss": 4.6826,
      "step": 796
    },
    {
      "epoch": 0.00797,
      "grad_norm": 0.654430341218369,
      "learning_rate": 0.0023910000000000003,
      "loss": 4.7248,
      "step": 797
    },
    {
      "epoch": 0.00798,
      "grad_norm": 0.5511512463730408,
      "learning_rate": 0.0023940000000000003,
      "loss": 4.6864,
      "step": 798
    },
    {
      "epoch": 0.00799,
      "grad_norm": 0.5084174359945534,
      "learning_rate": 0.0023970000000000003,
      "loss": 4.7161,
      "step": 799
    },
    {
      "epoch": 0.008,
      "grad_norm": 0.41968398472595075,
      "learning_rate": 0.0024000000000000002,
      "loss": 4.6652,
      "step": 800
    },
    {
      "epoch": 0.00801,
      "grad_norm": 0.4546309911468048,
      "learning_rate": 0.002403,
      "loss": 4.6967,
      "step": 801
    },
    {
      "epoch": 0.00802,
      "grad_norm": 0.430018994115786,
      "learning_rate": 0.002406,
      "loss": 4.6827,
      "step": 802
    },
    {
      "epoch": 0.00803,
      "grad_norm": 0.44614830716967085,
      "learning_rate": 0.002409,
      "loss": 4.6617,
      "step": 803
    },
    {
      "epoch": 0.00804,
      "grad_norm": 0.44722400844593674,
      "learning_rate": 0.002412,
      "loss": 4.6598,
      "step": 804
    },
    {
      "epoch": 0.00805,
      "grad_norm": 0.5179127215582825,
      "learning_rate": 0.002415,
      "loss": 4.6599,
      "step": 805
    },
    {
      "epoch": 0.00806,
      "grad_norm": 0.5610832008078775,
      "learning_rate": 0.002418,
      "loss": 4.677,
      "step": 806
    },
    {
      "epoch": 0.00807,
      "grad_norm": 0.5167453223410896,
      "learning_rate": 0.0024210000000000004,
      "loss": 4.6671,
      "step": 807
    },
    {
      "epoch": 0.00808,
      "grad_norm": 0.46468933196331563,
      "learning_rate": 0.0024240000000000004,
      "loss": 4.6511,
      "step": 808
    },
    {
      "epoch": 0.00809,
      "grad_norm": 0.5221883532574668,
      "learning_rate": 0.0024270000000000003,
      "loss": 4.6468,
      "step": 809
    },
    {
      "epoch": 0.0081,
      "grad_norm": 0.4992566900849729,
      "learning_rate": 0.0024300000000000003,
      "loss": 4.6744,
      "step": 810
    },
    {
      "epoch": 0.00811,
      "grad_norm": 0.4854147467055134,
      "learning_rate": 0.0024330000000000003,
      "loss": 4.646,
      "step": 811
    },
    {
      "epoch": 0.00812,
      "grad_norm": 0.650970729431075,
      "learning_rate": 0.0024360000000000002,
      "loss": 4.6307,
      "step": 812
    },
    {
      "epoch": 0.00813,
      "grad_norm": 0.8160691589494683,
      "learning_rate": 0.0024389999999999998,
      "loss": 4.6711,
      "step": 813
    },
    {
      "epoch": 0.00814,
      "grad_norm": 0.9918101747931352,
      "learning_rate": 0.0024419999999999997,
      "loss": 4.6946,
      "step": 814
    },
    {
      "epoch": 0.00815,
      "grad_norm": 1.247963175893729,
      "learning_rate": 0.0024449999999999997,
      "loss": 4.7226,
      "step": 815
    },
    {
      "epoch": 0.00816,
      "grad_norm": 0.8376200515557375,
      "learning_rate": 0.002448,
      "loss": 4.6777,
      "step": 816
    },
    {
      "epoch": 0.00817,
      "grad_norm": 0.9161032619759178,
      "learning_rate": 0.002451,
      "loss": 4.6939,
      "step": 817
    },
    {
      "epoch": 0.00818,
      "grad_norm": 1.0914649908014256,
      "learning_rate": 0.002454,
      "loss": 4.6886,
      "step": 818
    },
    {
      "epoch": 0.00819,
      "grad_norm": 0.9806171410774952,
      "learning_rate": 0.002457,
      "loss": 4.712,
      "step": 819
    },
    {
      "epoch": 0.0082,
      "grad_norm": 0.992236077471004,
      "learning_rate": 0.00246,
      "loss": 4.6918,
      "step": 820
    },
    {
      "epoch": 0.00821,
      "grad_norm": 1.0594557870263281,
      "learning_rate": 0.002463,
      "loss": 4.6759,
      "step": 821
    },
    {
      "epoch": 0.00822,
      "grad_norm": 1.0346800919438124,
      "learning_rate": 0.002466,
      "loss": 4.6853,
      "step": 822
    },
    {
      "epoch": 0.00823,
      "grad_norm": 0.9573573191186882,
      "learning_rate": 0.002469,
      "loss": 4.6833,
      "step": 823
    },
    {
      "epoch": 0.00824,
      "grad_norm": 1.1123514933123841,
      "learning_rate": 0.002472,
      "loss": 4.714,
      "step": 824
    },
    {
      "epoch": 0.00825,
      "grad_norm": 0.8463845700248506,
      "learning_rate": 0.0024749999999999998,
      "loss": 4.7191,
      "step": 825
    },
    {
      "epoch": 0.00826,
      "grad_norm": 0.8444785606085857,
      "learning_rate": 0.0024779999999999997,
      "loss": 4.672,
      "step": 826
    },
    {
      "epoch": 0.00827,
      "grad_norm": 0.9726341870117121,
      "learning_rate": 0.002481,
      "loss": 4.7078,
      "step": 827
    },
    {
      "epoch": 0.00828,
      "grad_norm": 0.9106448417621353,
      "learning_rate": 0.002484,
      "loss": 4.7003,
      "step": 828
    },
    {
      "epoch": 0.00829,
      "grad_norm": 0.7565680418878746,
      "learning_rate": 0.002487,
      "loss": 4.6856,
      "step": 829
    },
    {
      "epoch": 0.0083,
      "grad_norm": 0.8537774465977133,
      "learning_rate": 0.00249,
      "loss": 4.7017,
      "step": 830
    },
    {
      "epoch": 0.00831,
      "grad_norm": 0.9023323948099834,
      "learning_rate": 0.002493,
      "loss": 4.6871,
      "step": 831
    },
    {
      "epoch": 0.00832,
      "grad_norm": 0.8524529451127855,
      "learning_rate": 0.002496,
      "loss": 4.6815,
      "step": 832
    },
    {
      "epoch": 0.00833,
      "grad_norm": 0.9428655185832147,
      "learning_rate": 0.002499,
      "loss": 4.6808,
      "step": 833
    },
    {
      "epoch": 0.00834,
      "grad_norm": 0.9597220185428569,
      "learning_rate": 0.002502,
      "loss": 4.6879,
      "step": 834
    },
    {
      "epoch": 0.00835,
      "grad_norm": 0.7735101632354252,
      "learning_rate": 0.002505,
      "loss": 4.6724,
      "step": 835
    },
    {
      "epoch": 0.00836,
      "grad_norm": 0.9597202731139803,
      "learning_rate": 0.002508,
      "loss": 4.7061,
      "step": 836
    },
    {
      "epoch": 0.00837,
      "grad_norm": 0.9520863539431935,
      "learning_rate": 0.0025109999999999998,
      "loss": 4.6636,
      "step": 837
    },
    {
      "epoch": 0.00838,
      "grad_norm": 0.7800128524395746,
      "learning_rate": 0.0025139999999999997,
      "loss": 4.6721,
      "step": 838
    },
    {
      "epoch": 0.00839,
      "grad_norm": 0.8122589832425033,
      "learning_rate": 0.002517,
      "loss": 4.675,
      "step": 839
    },
    {
      "epoch": 0.0084,
      "grad_norm": 0.8183344402395425,
      "learning_rate": 0.00252,
      "loss": 4.6669,
      "step": 840
    },
    {
      "epoch": 0.00841,
      "grad_norm": 0.6918735110390536,
      "learning_rate": 0.002523,
      "loss": 4.6489,
      "step": 841
    },
    {
      "epoch": 0.00842,
      "grad_norm": 0.6201385747244391,
      "learning_rate": 0.002526,
      "loss": 4.6423,
      "step": 842
    },
    {
      "epoch": 0.00843,
      "grad_norm": 0.606127970479136,
      "learning_rate": 0.002529,
      "loss": 4.6465,
      "step": 843
    },
    {
      "epoch": 0.00844,
      "grad_norm": 0.5515773209874846,
      "learning_rate": 0.002532,
      "loss": 4.6607,
      "step": 844
    },
    {
      "epoch": 0.00845,
      "grad_norm": 0.6203742299859808,
      "learning_rate": 0.002535,
      "loss": 4.6293,
      "step": 845
    },
    {
      "epoch": 0.00846,
      "grad_norm": 0.5875832865020281,
      "learning_rate": 0.002538,
      "loss": 4.6474,
      "step": 846
    },
    {
      "epoch": 0.00847,
      "grad_norm": 0.5703256353430879,
      "learning_rate": 0.002541,
      "loss": 4.6282,
      "step": 847
    },
    {
      "epoch": 0.00848,
      "grad_norm": 0.602830367643936,
      "learning_rate": 0.002544,
      "loss": 4.6269,
      "step": 848
    },
    {
      "epoch": 0.00849,
      "grad_norm": 0.6741507039909044,
      "learning_rate": 0.002547,
      "loss": 4.6233,
      "step": 849
    },
    {
      "epoch": 0.0085,
      "grad_norm": 0.6288739006540759,
      "learning_rate": 0.00255,
      "loss": 4.6341,
      "step": 850
    },
    {
      "epoch": 0.00851,
      "grad_norm": 0.5820099008678455,
      "learning_rate": 0.002553,
      "loss": 4.644,
      "step": 851
    },
    {
      "epoch": 0.00852,
      "grad_norm": 0.586123912558797,
      "learning_rate": 0.002556,
      "loss": 4.6367,
      "step": 852
    },
    {
      "epoch": 0.00853,
      "grad_norm": 0.5127813487098001,
      "learning_rate": 0.002559,
      "loss": 4.6085,
      "step": 853
    },
    {
      "epoch": 0.00854,
      "grad_norm": 0.4730499644759358,
      "learning_rate": 0.002562,
      "loss": 4.6029,
      "step": 854
    },
    {
      "epoch": 0.00855,
      "grad_norm": 0.44708869980986227,
      "learning_rate": 0.002565,
      "loss": 4.5799,
      "step": 855
    },
    {
      "epoch": 0.00856,
      "grad_norm": 0.466044480858233,
      "learning_rate": 0.002568,
      "loss": 4.6142,
      "step": 856
    },
    {
      "epoch": 0.00857,
      "grad_norm": 0.5382201915945353,
      "learning_rate": 0.002571,
      "loss": 4.6036,
      "step": 857
    },
    {
      "epoch": 0.00858,
      "grad_norm": 0.6780662034295477,
      "learning_rate": 0.002574,
      "loss": 4.609,
      "step": 858
    },
    {
      "epoch": 0.00859,
      "grad_norm": 0.9086610382483981,
      "learning_rate": 0.002577,
      "loss": 4.6039,
      "step": 859
    },
    {
      "epoch": 0.0086,
      "grad_norm": 0.8563688949272525,
      "learning_rate": 0.00258,
      "loss": 4.6531,
      "step": 860
    },
    {
      "epoch": 0.00861,
      "grad_norm": 0.5965670098126366,
      "learning_rate": 0.0025830000000000002,
      "loss": 4.6461,
      "step": 861
    },
    {
      "epoch": 0.00862,
      "grad_norm": 0.7975052365958228,
      "learning_rate": 0.002586,
      "loss": 4.611,
      "step": 862
    },
    {
      "epoch": 0.00863,
      "grad_norm": 0.650099032572018,
      "learning_rate": 0.002589,
      "loss": 4.6284,
      "step": 863
    },
    {
      "epoch": 0.00864,
      "grad_norm": 0.6277114763068243,
      "learning_rate": 0.002592,
      "loss": 4.5809,
      "step": 864
    },
    {
      "epoch": 0.00865,
      "grad_norm": 0.7499269309750987,
      "learning_rate": 0.002595,
      "loss": 4.6024,
      "step": 865
    },
    {
      "epoch": 0.00866,
      "grad_norm": 0.7085973518403954,
      "learning_rate": 0.002598,
      "loss": 4.6233,
      "step": 866
    },
    {
      "epoch": 0.00867,
      "grad_norm": 0.605538925445329,
      "learning_rate": 0.002601,
      "loss": 4.6101,
      "step": 867
    },
    {
      "epoch": 0.00868,
      "grad_norm": 0.48200776305054654,
      "learning_rate": 0.002604,
      "loss": 4.586,
      "step": 868
    },
    {
      "epoch": 0.00869,
      "grad_norm": 0.5266950965425763,
      "learning_rate": 0.002607,
      "loss": 4.5966,
      "step": 869
    },
    {
      "epoch": 0.0087,
      "grad_norm": 0.48953699231607295,
      "learning_rate": 0.00261,
      "loss": 4.5832,
      "step": 870
    },
    {
      "epoch": 0.00871,
      "grad_norm": 0.5478274928438833,
      "learning_rate": 0.002613,
      "loss": 4.5992,
      "step": 871
    },
    {
      "epoch": 0.00872,
      "grad_norm": 0.6263670684952429,
      "learning_rate": 0.002616,
      "loss": 4.5864,
      "step": 872
    },
    {
      "epoch": 0.00873,
      "grad_norm": 0.6672951132542,
      "learning_rate": 0.0026190000000000002,
      "loss": 4.5977,
      "step": 873
    },
    {
      "epoch": 0.00874,
      "grad_norm": 0.677096167715366,
      "learning_rate": 0.002622,
      "loss": 4.617,
      "step": 874
    },
    {
      "epoch": 0.00875,
      "grad_norm": 0.6959913524482387,
      "learning_rate": 0.002625,
      "loss": 4.5696,
      "step": 875
    },
    {
      "epoch": 0.00876,
      "grad_norm": 0.6682762743495083,
      "learning_rate": 0.002628,
      "loss": 4.5958,
      "step": 876
    },
    {
      "epoch": 0.00877,
      "grad_norm": 0.6375266502117092,
      "learning_rate": 0.002631,
      "loss": 4.612,
      "step": 877
    },
    {
      "epoch": 0.00878,
      "grad_norm": 0.7079418482290942,
      "learning_rate": 0.002634,
      "loss": 4.5486,
      "step": 878
    },
    {
      "epoch": 0.00879,
      "grad_norm": 0.6282689223941402,
      "learning_rate": 0.002637,
      "loss": 4.578,
      "step": 879
    },
    {
      "epoch": 0.0088,
      "grad_norm": 0.5460943252882049,
      "learning_rate": 0.00264,
      "loss": 4.5852,
      "step": 880
    },
    {
      "epoch": 0.00881,
      "grad_norm": 0.5723972494402886,
      "learning_rate": 0.002643,
      "loss": 4.5869,
      "step": 881
    },
    {
      "epoch": 0.00882,
      "grad_norm": 0.647818443655113,
      "learning_rate": 0.002646,
      "loss": 4.588,
      "step": 882
    },
    {
      "epoch": 0.00883,
      "grad_norm": 0.8827068805337381,
      "learning_rate": 0.002649,
      "loss": 4.5935,
      "step": 883
    },
    {
      "epoch": 0.00884,
      "grad_norm": 1.3000201706023533,
      "learning_rate": 0.0026520000000000003,
      "loss": 4.6052,
      "step": 884
    },
    {
      "epoch": 0.00885,
      "grad_norm": 0.7527768384442359,
      "learning_rate": 0.0026550000000000002,
      "loss": 4.5797,
      "step": 885
    },
    {
      "epoch": 0.00886,
      "grad_norm": 0.7313595200920677,
      "learning_rate": 0.002658,
      "loss": 4.6019,
      "step": 886
    },
    {
      "epoch": 0.00887,
      "grad_norm": 0.553327654847044,
      "learning_rate": 0.002661,
      "loss": 4.5828,
      "step": 887
    },
    {
      "epoch": 0.00888,
      "grad_norm": 0.6064219625843388,
      "learning_rate": 0.002664,
      "loss": 4.5894,
      "step": 888
    },
    {
      "epoch": 0.00889,
      "grad_norm": 0.6392357596846293,
      "learning_rate": 0.002667,
      "loss": 4.5422,
      "step": 889
    },
    {
      "epoch": 0.0089,
      "grad_norm": 0.6860123914477424,
      "learning_rate": 0.00267,
      "loss": 4.5989,
      "step": 890
    },
    {
      "epoch": 0.00891,
      "grad_norm": 0.7088960904364014,
      "learning_rate": 0.002673,
      "loss": 4.5822,
      "step": 891
    },
    {
      "epoch": 0.00892,
      "grad_norm": 0.7157207147763361,
      "learning_rate": 0.002676,
      "loss": 4.5934,
      "step": 892
    },
    {
      "epoch": 0.00893,
      "grad_norm": 0.7412527752908875,
      "learning_rate": 0.002679,
      "loss": 4.5709,
      "step": 893
    },
    {
      "epoch": 0.00894,
      "grad_norm": 0.8084836835989728,
      "learning_rate": 0.002682,
      "loss": 4.5639,
      "step": 894
    },
    {
      "epoch": 0.00895,
      "grad_norm": 0.9923307111818513,
      "learning_rate": 0.0026850000000000003,
      "loss": 4.5864,
      "step": 895
    },
    {
      "epoch": 0.00896,
      "grad_norm": 1.2171682577354312,
      "learning_rate": 0.0026880000000000003,
      "loss": 4.6057,
      "step": 896
    },
    {
      "epoch": 0.00897,
      "grad_norm": 0.797478427208377,
      "learning_rate": 0.0026910000000000002,
      "loss": 4.5989,
      "step": 897
    },
    {
      "epoch": 0.00898,
      "grad_norm": 0.7928728804117916,
      "learning_rate": 0.002694,
      "loss": 4.594,
      "step": 898
    },
    {
      "epoch": 0.00899,
      "grad_norm": 0.8357403035452178,
      "learning_rate": 0.002697,
      "loss": 4.5983,
      "step": 899
    },
    {
      "epoch": 0.009,
      "grad_norm": 0.8448290091163538,
      "learning_rate": 0.0027,
      "loss": 4.6292,
      "step": 900
    },
    {
      "epoch": 0.00901,
      "grad_norm": 0.9488092229670547,
      "learning_rate": 0.002703,
      "loss": 4.5868,
      "step": 901
    },
    {
      "epoch": 0.00902,
      "grad_norm": 0.9434404658743749,
      "learning_rate": 0.002706,
      "loss": 4.5999,
      "step": 902
    },
    {
      "epoch": 0.00903,
      "grad_norm": 1.0122099567822476,
      "learning_rate": 0.002709,
      "loss": 4.6102,
      "step": 903
    },
    {
      "epoch": 0.00904,
      "grad_norm": 0.9358691681287052,
      "learning_rate": 0.002712,
      "loss": 4.5848,
      "step": 904
    },
    {
      "epoch": 0.00905,
      "grad_norm": 0.8321510442485943,
      "learning_rate": 0.002715,
      "loss": 4.5984,
      "step": 905
    },
    {
      "epoch": 0.00906,
      "grad_norm": 0.8914473393947665,
      "learning_rate": 0.002718,
      "loss": 4.6112,
      "step": 906
    },
    {
      "epoch": 0.00907,
      "grad_norm": 0.9883982303638487,
      "learning_rate": 0.0027210000000000003,
      "loss": 4.6386,
      "step": 907
    },
    {
      "epoch": 0.00908,
      "grad_norm": 0.86073203349026,
      "learning_rate": 0.0027240000000000003,
      "loss": 4.6116,
      "step": 908
    },
    {
      "epoch": 0.00909,
      "grad_norm": 0.7773747412069614,
      "learning_rate": 0.0027270000000000003,
      "loss": 4.6163,
      "step": 909
    },
    {
      "epoch": 0.0091,
      "grad_norm": 0.7370585718531062,
      "learning_rate": 0.0027300000000000002,
      "loss": 4.6234,
      "step": 910
    },
    {
      "epoch": 0.00911,
      "grad_norm": 0.6906269071273593,
      "learning_rate": 0.002733,
      "loss": 4.5785,
      "step": 911
    },
    {
      "epoch": 0.00912,
      "grad_norm": 0.6578032292778252,
      "learning_rate": 0.002736,
      "loss": 4.5778,
      "step": 912
    },
    {
      "epoch": 0.00913,
      "grad_norm": 0.6528626059582382,
      "learning_rate": 0.002739,
      "loss": 4.5704,
      "step": 913
    },
    {
      "epoch": 0.00914,
      "grad_norm": 0.599731896856576,
      "learning_rate": 0.002742,
      "loss": 4.595,
      "step": 914
    },
    {
      "epoch": 0.00915,
      "grad_norm": 0.5922054086035364,
      "learning_rate": 0.002745,
      "loss": 4.5555,
      "step": 915
    },
    {
      "epoch": 0.00916,
      "grad_norm": 0.528646140228931,
      "learning_rate": 0.002748,
      "loss": 4.5304,
      "step": 916
    },
    {
      "epoch": 0.00917,
      "grad_norm": 0.5305158198561161,
      "learning_rate": 0.002751,
      "loss": 4.5419,
      "step": 917
    },
    {
      "epoch": 0.00918,
      "grad_norm": 0.4736382884122071,
      "learning_rate": 0.0027540000000000004,
      "loss": 4.5569,
      "step": 918
    },
    {
      "epoch": 0.00919,
      "grad_norm": 0.45838817911808083,
      "learning_rate": 0.0027570000000000003,
      "loss": 4.5357,
      "step": 919
    },
    {
      "epoch": 0.0092,
      "grad_norm": 0.4361472184478695,
      "learning_rate": 0.0027600000000000003,
      "loss": 4.5493,
      "step": 920
    },
    {
      "epoch": 0.00921,
      "grad_norm": 0.43931126757575867,
      "learning_rate": 0.0027630000000000003,
      "loss": 4.5336,
      "step": 921
    },
    {
      "epoch": 0.00922,
      "grad_norm": 0.46920285200840567,
      "learning_rate": 0.0027660000000000002,
      "loss": 4.5412,
      "step": 922
    },
    {
      "epoch": 0.00923,
      "grad_norm": 0.5996209745093354,
      "learning_rate": 0.002769,
      "loss": 4.5218,
      "step": 923
    },
    {
      "epoch": 0.00924,
      "grad_norm": 0.702528530852281,
      "learning_rate": 0.002772,
      "loss": 4.5564,
      "step": 924
    },
    {
      "epoch": 0.00925,
      "grad_norm": 0.6651779892024297,
      "learning_rate": 0.002775,
      "loss": 4.5288,
      "step": 925
    },
    {
      "epoch": 0.00926,
      "grad_norm": 0.5723855300218565,
      "learning_rate": 0.002778,
      "loss": 4.5426,
      "step": 926
    },
    {
      "epoch": 0.00927,
      "grad_norm": 0.6080148945846373,
      "learning_rate": 0.002781,
      "loss": 4.5043,
      "step": 927
    },
    {
      "epoch": 0.00928,
      "grad_norm": 0.6957035289314476,
      "learning_rate": 0.002784,
      "loss": 4.5502,
      "step": 928
    },
    {
      "epoch": 0.00929,
      "grad_norm": 0.7366634335117053,
      "learning_rate": 0.0027870000000000004,
      "loss": 4.5403,
      "step": 929
    },
    {
      "epoch": 0.0093,
      "grad_norm": 0.7796102692352119,
      "learning_rate": 0.0027900000000000004,
      "loss": 4.5265,
      "step": 930
    },
    {
      "epoch": 0.00931,
      "grad_norm": 0.7116909602884058,
      "learning_rate": 0.0027930000000000003,
      "loss": 4.5333,
      "step": 931
    },
    {
      "epoch": 0.00932,
      "grad_norm": 0.8319854024158545,
      "learning_rate": 0.0027960000000000003,
      "loss": 4.5481,
      "step": 932
    },
    {
      "epoch": 0.00933,
      "grad_norm": 0.8001418279766108,
      "learning_rate": 0.0027990000000000003,
      "loss": 4.5286,
      "step": 933
    },
    {
      "epoch": 0.00934,
      "grad_norm": 0.6519619232143173,
      "learning_rate": 0.0028020000000000002,
      "loss": 4.5422,
      "step": 934
    },
    {
      "epoch": 0.00935,
      "grad_norm": 0.7580737482550882,
      "learning_rate": 0.002805,
      "loss": 4.5559,
      "step": 935
    },
    {
      "epoch": 0.00936,
      "grad_norm": 0.8499038627491867,
      "learning_rate": 0.002808,
      "loss": 4.5679,
      "step": 936
    },
    {
      "epoch": 0.00937,
      "grad_norm": 0.7597302495348821,
      "learning_rate": 0.002811,
      "loss": 4.5708,
      "step": 937
    },
    {
      "epoch": 0.00938,
      "grad_norm": 0.9878821641788273,
      "learning_rate": 0.002814,
      "loss": 4.5609,
      "step": 938
    },
    {
      "epoch": 0.00939,
      "grad_norm": 0.9691729918808772,
      "learning_rate": 0.002817,
      "loss": 4.563,
      "step": 939
    },
    {
      "epoch": 0.0094,
      "grad_norm": 0.8937843559478598,
      "learning_rate": 0.00282,
      "loss": 4.56,
      "step": 940
    },
    {
      "epoch": 0.00941,
      "grad_norm": 0.9477839045288606,
      "learning_rate": 0.002823,
      "loss": 4.5409,
      "step": 941
    },
    {
      "epoch": 0.00942,
      "grad_norm": 1.088560613057821,
      "learning_rate": 0.002826,
      "loss": 4.5819,
      "step": 942
    },
    {
      "epoch": 0.00943,
      "grad_norm": 0.8020128186220904,
      "learning_rate": 0.002829,
      "loss": 4.556,
      "step": 943
    },
    {
      "epoch": 0.00944,
      "grad_norm": 0.7970499406732843,
      "learning_rate": 0.002832,
      "loss": 4.5652,
      "step": 944
    },
    {
      "epoch": 0.00945,
      "grad_norm": 0.760430287307007,
      "learning_rate": 0.002835,
      "loss": 4.567,
      "step": 945
    },
    {
      "epoch": 0.00946,
      "grad_norm": 0.8410168172764453,
      "learning_rate": 0.002838,
      "loss": 4.5808,
      "step": 946
    },
    {
      "epoch": 0.00947,
      "grad_norm": 0.8502364092306604,
      "learning_rate": 0.0028409999999999998,
      "loss": 4.5581,
      "step": 947
    },
    {
      "epoch": 0.00948,
      "grad_norm": 0.7534324730199542,
      "learning_rate": 0.0028439999999999997,
      "loss": 4.533,
      "step": 948
    },
    {
      "epoch": 0.00949,
      "grad_norm": 0.8075715283027973,
      "learning_rate": 0.002847,
      "loss": 4.5789,
      "step": 949
    },
    {
      "epoch": 0.0095,
      "grad_norm": 0.8790685187514339,
      "learning_rate": 0.00285,
      "loss": 4.5764,
      "step": 950
    },
    {
      "epoch": 0.00951,
      "grad_norm": 0.8527621336415785,
      "learning_rate": 0.002853,
      "loss": 4.552,
      "step": 951
    },
    {
      "epoch": 0.00952,
      "grad_norm": 0.793648162843131,
      "learning_rate": 0.002856,
      "loss": 4.5771,
      "step": 952
    },
    {
      "epoch": 0.00953,
      "grad_norm": 0.7100823051409002,
      "learning_rate": 0.002859,
      "loss": 4.5151,
      "step": 953
    },
    {
      "epoch": 0.00954,
      "grad_norm": 0.776086010454581,
      "learning_rate": 0.002862,
      "loss": 4.5748,
      "step": 954
    },
    {
      "epoch": 0.00955,
      "grad_norm": 0.7357834745016256,
      "learning_rate": 0.002865,
      "loss": 4.5651,
      "step": 955
    },
    {
      "epoch": 0.00956,
      "grad_norm": 0.6871788604084053,
      "learning_rate": 0.002868,
      "loss": 4.5378,
      "step": 956
    },
    {
      "epoch": 0.00957,
      "grad_norm": 0.6293704920093642,
      "learning_rate": 0.002871,
      "loss": 4.5585,
      "step": 957
    },
    {
      "epoch": 0.00958,
      "grad_norm": 0.6933721545151298,
      "learning_rate": 0.002874,
      "loss": 4.5402,
      "step": 958
    },
    {
      "epoch": 0.00959,
      "grad_norm": 0.6216290945191316,
      "learning_rate": 0.002877,
      "loss": 4.5294,
      "step": 959
    },
    {
      "epoch": 0.0096,
      "grad_norm": 0.44090482568449035,
      "learning_rate": 0.0028799999999999997,
      "loss": 4.5205,
      "step": 960
    },
    {
      "epoch": 0.00961,
      "grad_norm": 0.5026549244936088,
      "learning_rate": 0.002883,
      "loss": 4.4973,
      "step": 961
    },
    {
      "epoch": 0.00962,
      "grad_norm": 0.46550744372429714,
      "learning_rate": 0.002886,
      "loss": 4.5199,
      "step": 962
    },
    {
      "epoch": 0.00963,
      "grad_norm": 0.4817462883709995,
      "learning_rate": 0.002889,
      "loss": 4.5204,
      "step": 963
    },
    {
      "epoch": 0.00964,
      "grad_norm": 0.5021989893794454,
      "learning_rate": 0.002892,
      "loss": 4.5105,
      "step": 964
    },
    {
      "epoch": 0.00965,
      "grad_norm": 0.6331237702649058,
      "learning_rate": 0.002895,
      "loss": 4.4888,
      "step": 965
    },
    {
      "epoch": 0.00966,
      "grad_norm": 0.7186463225121739,
      "learning_rate": 0.002898,
      "loss": 4.5122,
      "step": 966
    },
    {
      "epoch": 0.00967,
      "grad_norm": 0.835541291398658,
      "learning_rate": 0.002901,
      "loss": 4.5497,
      "step": 967
    },
    {
      "epoch": 0.00968,
      "grad_norm": 0.7770950591910699,
      "learning_rate": 0.002904,
      "loss": 4.5246,
      "step": 968
    },
    {
      "epoch": 0.00969,
      "grad_norm": 0.6421972738290654,
      "learning_rate": 0.002907,
      "loss": 4.5465,
      "step": 969
    },
    {
      "epoch": 0.0097,
      "grad_norm": 0.6170493579190435,
      "learning_rate": 0.00291,
      "loss": 4.5014,
      "step": 970
    },
    {
      "epoch": 0.00971,
      "grad_norm": 0.671893763395282,
      "learning_rate": 0.002913,
      "loss": 4.5134,
      "step": 971
    },
    {
      "epoch": 0.00972,
      "grad_norm": 0.5384349268117217,
      "learning_rate": 0.002916,
      "loss": 4.51,
      "step": 972
    },
    {
      "epoch": 0.00973,
      "grad_norm": 0.6570052261370841,
      "learning_rate": 0.002919,
      "loss": 4.5075,
      "step": 973
    },
    {
      "epoch": 0.00974,
      "grad_norm": 0.6469437996214488,
      "learning_rate": 0.002922,
      "loss": 4.5042,
      "step": 974
    },
    {
      "epoch": 0.00975,
      "grad_norm": 0.5139434995269291,
      "learning_rate": 0.002925,
      "loss": 4.5141,
      "step": 975
    },
    {
      "epoch": 0.00976,
      "grad_norm": 0.540350404123188,
      "learning_rate": 0.002928,
      "loss": 4.4984,
      "step": 976
    },
    {
      "epoch": 0.00977,
      "grad_norm": 0.5640158884340003,
      "learning_rate": 0.002931,
      "loss": 4.5359,
      "step": 977
    },
    {
      "epoch": 0.00978,
      "grad_norm": 0.5471232322596488,
      "learning_rate": 0.002934,
      "loss": 4.5069,
      "step": 978
    },
    {
      "epoch": 0.00979,
      "grad_norm": 0.6391692840311302,
      "learning_rate": 0.002937,
      "loss": 4.4972,
      "step": 979
    },
    {
      "epoch": 0.0098,
      "grad_norm": 0.7065424241899814,
      "learning_rate": 0.00294,
      "loss": 4.4963,
      "step": 980
    },
    {
      "epoch": 0.00981,
      "grad_norm": 0.6903724510426201,
      "learning_rate": 0.002943,
      "loss": 4.5078,
      "step": 981
    },
    {
      "epoch": 0.00982,
      "grad_norm": 0.653302049548968,
      "learning_rate": 0.002946,
      "loss": 4.4991,
      "step": 982
    },
    {
      "epoch": 0.00983,
      "grad_norm": 0.7786836590229197,
      "learning_rate": 0.0029490000000000002,
      "loss": 4.5061,
      "step": 983
    },
    {
      "epoch": 0.00984,
      "grad_norm": 0.8944946184941699,
      "learning_rate": 0.002952,
      "loss": 4.5043,
      "step": 984
    },
    {
      "epoch": 0.00985,
      "grad_norm": 0.9434362337974719,
      "learning_rate": 0.002955,
      "loss": 4.5335,
      "step": 985
    },
    {
      "epoch": 0.00986,
      "grad_norm": 0.9312010498282314,
      "learning_rate": 0.002958,
      "loss": 4.4908,
      "step": 986
    },
    {
      "epoch": 0.00987,
      "grad_norm": 0.7600059379317153,
      "learning_rate": 0.002961,
      "loss": 4.5038,
      "step": 987
    },
    {
      "epoch": 0.00988,
      "grad_norm": 0.6949983731085813,
      "learning_rate": 0.002964,
      "loss": 4.533,
      "step": 988
    },
    {
      "epoch": 0.00989,
      "grad_norm": 0.6958207764457225,
      "learning_rate": 0.002967,
      "loss": 4.4744,
      "step": 989
    },
    {
      "epoch": 0.0099,
      "grad_norm": 0.8314159874496235,
      "learning_rate": 0.00297,
      "loss": 4.5218,
      "step": 990
    },
    {
      "epoch": 0.00991,
      "grad_norm": 0.90109982550625,
      "learning_rate": 0.002973,
      "loss": 4.5224,
      "step": 991
    },
    {
      "epoch": 0.00992,
      "grad_norm": 0.951208720196979,
      "learning_rate": 0.002976,
      "loss": 4.5281,
      "step": 992
    },
    {
      "epoch": 0.00993,
      "grad_norm": 1.034654706401041,
      "learning_rate": 0.002979,
      "loss": 4.539,
      "step": 993
    },
    {
      "epoch": 0.00994,
      "grad_norm": 1.0078120294806783,
      "learning_rate": 0.002982,
      "loss": 4.5459,
      "step": 994
    },
    {
      "epoch": 0.00995,
      "grad_norm": 0.9905056664043064,
      "learning_rate": 0.0029850000000000002,
      "loss": 4.5461,
      "step": 995
    },
    {
      "epoch": 0.00996,
      "grad_norm": 1.1603429750232241,
      "learning_rate": 0.002988,
      "loss": 4.5605,
      "step": 996
    },
    {
      "epoch": 0.00997,
      "grad_norm": 0.8595677567729485,
      "learning_rate": 0.002991,
      "loss": 4.5366,
      "step": 997
    },
    {
      "epoch": 0.00998,
      "grad_norm": 0.9672785170741734,
      "learning_rate": 0.002994,
      "loss": 4.5467,
      "step": 998
    },
    {
      "epoch": 0.00999,
      "grad_norm": 0.8855476122040005,
      "learning_rate": 0.002997,
      "loss": 4.5306,
      "step": 999
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7479993027173912,
      "learning_rate": 0.003,
      "loss": 4.5436,
      "step": 1000
    },
    {
      "epoch": 0.01001,
      "grad_norm": 0.7121555496393158,
      "learning_rate": 0.003,
      "loss": 4.5401,
      "step": 1001
    },
    {
      "epoch": 0.01002,
      "grad_norm": 0.72505396659877,
      "learning_rate": 0.003,
      "loss": 4.5099,
      "step": 1002
    },
    {
      "epoch": 0.01003,
      "grad_norm": 0.7482409624375888,
      "learning_rate": 0.003,
      "loss": 4.5346,
      "step": 1003
    },
    {
      "epoch": 0.01004,
      "grad_norm": 0.5922239695560899,
      "learning_rate": 0.003,
      "loss": 4.513,
      "step": 1004
    },
    {
      "epoch": 0.01005,
      "grad_norm": 0.7163386651042714,
      "learning_rate": 0.003,
      "loss": 4.4897,
      "step": 1005
    },
    {
      "epoch": 0.01006,
      "grad_norm": 0.7664341729228217,
      "learning_rate": 0.003,
      "loss": 4.4943,
      "step": 1006
    },
    {
      "epoch": 0.01007,
      "grad_norm": 0.6325498197242884,
      "learning_rate": 0.003,
      "loss": 4.5316,
      "step": 1007
    },
    {
      "epoch": 0.01008,
      "grad_norm": 0.5334181570982156,
      "learning_rate": 0.003,
      "loss": 4.5325,
      "step": 1008
    },
    {
      "epoch": 0.01009,
      "grad_norm": 0.4756687788935663,
      "learning_rate": 0.003,
      "loss": 4.4998,
      "step": 1009
    },
    {
      "epoch": 0.0101,
      "grad_norm": 0.44081077367096977,
      "learning_rate": 0.003,
      "loss": 4.5025,
      "step": 1010
    },
    {
      "epoch": 0.01011,
      "grad_norm": 0.4706704982402646,
      "learning_rate": 0.003,
      "loss": 4.5082,
      "step": 1011
    },
    {
      "epoch": 0.01012,
      "grad_norm": 0.47223639219621605,
      "learning_rate": 0.003,
      "loss": 4.492,
      "step": 1012
    },
    {
      "epoch": 0.01013,
      "grad_norm": 0.4611333961201937,
      "learning_rate": 0.003,
      "loss": 4.5135,
      "step": 1013
    },
    {
      "epoch": 0.01014,
      "grad_norm": 0.36636268932008986,
      "learning_rate": 0.003,
      "loss": 4.4528,
      "step": 1014
    },
    {
      "epoch": 0.01015,
      "grad_norm": 0.4484711002206511,
      "learning_rate": 0.003,
      "loss": 4.4948,
      "step": 1015
    },
    {
      "epoch": 0.01016,
      "grad_norm": 0.553209882568818,
      "learning_rate": 0.003,
      "loss": 4.5082,
      "step": 1016
    },
    {
      "epoch": 0.01017,
      "grad_norm": 0.8898002239967567,
      "learning_rate": 0.003,
      "loss": 4.4579,
      "step": 1017
    },
    {
      "epoch": 0.01018,
      "grad_norm": 0.9938726896514534,
      "learning_rate": 0.003,
      "loss": 4.5557,
      "step": 1018
    },
    {
      "epoch": 0.01019,
      "grad_norm": 0.7008138530645361,
      "learning_rate": 0.003,
      "loss": 4.5003,
      "step": 1019
    },
    {
      "epoch": 0.0102,
      "grad_norm": 1.006685411589756,
      "learning_rate": 0.003,
      "loss": 4.5244,
      "step": 1020
    },
    {
      "epoch": 0.01021,
      "grad_norm": 0.6295405897908188,
      "learning_rate": 0.003,
      "loss": 4.4915,
      "step": 1021
    },
    {
      "epoch": 0.01022,
      "grad_norm": 0.6613412501940705,
      "learning_rate": 0.003,
      "loss": 4.5063,
      "step": 1022
    },
    {
      "epoch": 0.01023,
      "grad_norm": 0.6284055579476905,
      "learning_rate": 0.003,
      "loss": 4.4914,
      "step": 1023
    },
    {
      "epoch": 0.01024,
      "grad_norm": 0.6735486877543648,
      "learning_rate": 0.003,
      "loss": 4.4573,
      "step": 1024
    },
    {
      "epoch": 0.01025,
      "grad_norm": 0.6253815453908939,
      "learning_rate": 0.003,
      "loss": 4.4832,
      "step": 1025
    },
    {
      "epoch": 0.01026,
      "grad_norm": 0.5932239048767913,
      "learning_rate": 0.003,
      "loss": 4.4735,
      "step": 1026
    },
    {
      "epoch": 0.01027,
      "grad_norm": 0.6318859057234201,
      "learning_rate": 0.003,
      "loss": 4.4952,
      "step": 1027
    },
    {
      "epoch": 0.01028,
      "grad_norm": 0.6588163663127304,
      "learning_rate": 0.003,
      "loss": 4.4763,
      "step": 1028
    },
    {
      "epoch": 0.01029,
      "grad_norm": 0.55721560335989,
      "learning_rate": 0.003,
      "loss": 4.4524,
      "step": 1029
    },
    {
      "epoch": 0.0103,
      "grad_norm": 0.5468199872259095,
      "learning_rate": 0.003,
      "loss": 4.4809,
      "step": 1030
    },
    {
      "epoch": 0.01031,
      "grad_norm": 0.5601179932704955,
      "learning_rate": 0.003,
      "loss": 4.4934,
      "step": 1031
    },
    {
      "epoch": 0.01032,
      "grad_norm": 0.5514443283118416,
      "learning_rate": 0.003,
      "loss": 4.464,
      "step": 1032
    },
    {
      "epoch": 0.01033,
      "grad_norm": 0.49777794102419176,
      "learning_rate": 0.003,
      "loss": 4.496,
      "step": 1033
    },
    {
      "epoch": 0.01034,
      "grad_norm": 0.5163173300909295,
      "learning_rate": 0.003,
      "loss": 4.4625,
      "step": 1034
    },
    {
      "epoch": 0.01035,
      "grad_norm": 0.4798196088286999,
      "learning_rate": 0.003,
      "loss": 4.4573,
      "step": 1035
    },
    {
      "epoch": 0.01036,
      "grad_norm": 0.3958499097440651,
      "learning_rate": 0.003,
      "loss": 4.4459,
      "step": 1036
    },
    {
      "epoch": 0.01037,
      "grad_norm": 0.3643003447603396,
      "learning_rate": 0.003,
      "loss": 4.4698,
      "step": 1037
    },
    {
      "epoch": 0.01038,
      "grad_norm": 0.40989678679265595,
      "learning_rate": 0.003,
      "loss": 4.4667,
      "step": 1038
    },
    {
      "epoch": 0.01039,
      "grad_norm": 0.3544910886173235,
      "learning_rate": 0.003,
      "loss": 4.4556,
      "step": 1039
    },
    {
      "epoch": 0.0104,
      "grad_norm": 0.3719879786284374,
      "learning_rate": 0.003,
      "loss": 4.4667,
      "step": 1040
    },
    {
      "epoch": 0.01041,
      "grad_norm": 0.3770068025965155,
      "learning_rate": 0.003,
      "loss": 4.4469,
      "step": 1041
    },
    {
      "epoch": 0.01042,
      "grad_norm": 0.4469030721524434,
      "learning_rate": 0.003,
      "loss": 4.4752,
      "step": 1042
    },
    {
      "epoch": 0.01043,
      "grad_norm": 0.7272268008417163,
      "learning_rate": 0.003,
      "loss": 4.4274,
      "step": 1043
    },
    {
      "epoch": 0.01044,
      "grad_norm": 0.8992907359513324,
      "learning_rate": 0.003,
      "loss": 4.4949,
      "step": 1044
    },
    {
      "epoch": 0.01045,
      "grad_norm": 0.8302310255488462,
      "learning_rate": 0.003,
      "loss": 4.4953,
      "step": 1045
    },
    {
      "epoch": 0.01046,
      "grad_norm": 0.9975061123002056,
      "learning_rate": 0.003,
      "loss": 4.4816,
      "step": 1046
    },
    {
      "epoch": 0.01047,
      "grad_norm": 0.8067658442263902,
      "learning_rate": 0.003,
      "loss": 4.4954,
      "step": 1047
    },
    {
      "epoch": 0.01048,
      "grad_norm": 0.7572637033913173,
      "learning_rate": 0.003,
      "loss": 4.4741,
      "step": 1048
    },
    {
      "epoch": 0.01049,
      "grad_norm": 0.8994748129621607,
      "learning_rate": 0.003,
      "loss": 4.4647,
      "step": 1049
    },
    {
      "epoch": 0.0105,
      "grad_norm": 0.7962039985766673,
      "learning_rate": 0.003,
      "loss": 4.5006,
      "step": 1050
    },
    {
      "epoch": 0.01051,
      "grad_norm": 0.8207643682537104,
      "learning_rate": 0.003,
      "loss": 4.4811,
      "step": 1051
    },
    {
      "epoch": 0.01052,
      "grad_norm": 0.7458558375660994,
      "learning_rate": 0.003,
      "loss": 4.4881,
      "step": 1052
    },
    {
      "epoch": 0.01053,
      "grad_norm": 0.6759075645027314,
      "learning_rate": 0.003,
      "loss": 4.4946,
      "step": 1053
    },
    {
      "epoch": 0.01054,
      "grad_norm": 0.6790074754239308,
      "learning_rate": 0.003,
      "loss": 4.4852,
      "step": 1054
    },
    {
      "epoch": 0.01055,
      "grad_norm": 0.6920845821894167,
      "learning_rate": 0.003,
      "loss": 4.4644,
      "step": 1055
    },
    {
      "epoch": 0.01056,
      "grad_norm": 0.5219544124128492,
      "learning_rate": 0.003,
      "loss": 4.4569,
      "step": 1056
    },
    {
      "epoch": 0.01057,
      "grad_norm": 0.6021767240952963,
      "learning_rate": 0.003,
      "loss": 4.4651,
      "step": 1057
    },
    {
      "epoch": 0.01058,
      "grad_norm": 0.5219741420101096,
      "learning_rate": 0.003,
      "loss": 4.4552,
      "step": 1058
    },
    {
      "epoch": 0.01059,
      "grad_norm": 0.509217286199775,
      "learning_rate": 0.003,
      "loss": 4.4557,
      "step": 1059
    },
    {
      "epoch": 0.0106,
      "grad_norm": 0.6119681984085901,
      "learning_rate": 0.003,
      "loss": 4.4506,
      "step": 1060
    },
    {
      "epoch": 0.01061,
      "grad_norm": 0.6018802462486371,
      "learning_rate": 0.003,
      "loss": 4.4451,
      "step": 1061
    },
    {
      "epoch": 0.01062,
      "grad_norm": 0.6072220138777784,
      "learning_rate": 0.003,
      "loss": 4.4398,
      "step": 1062
    },
    {
      "epoch": 0.01063,
      "grad_norm": 0.5793152524763621,
      "learning_rate": 0.003,
      "loss": 4.4372,
      "step": 1063
    },
    {
      "epoch": 0.01064,
      "grad_norm": 0.740973950389461,
      "learning_rate": 0.003,
      "loss": 4.4576,
      "step": 1064
    },
    {
      "epoch": 0.01065,
      "grad_norm": 0.8998946238241571,
      "learning_rate": 0.003,
      "loss": 4.4756,
      "step": 1065
    },
    {
      "epoch": 0.01066,
      "grad_norm": 0.889057289027363,
      "learning_rate": 0.003,
      "loss": 4.4727,
      "step": 1066
    },
    {
      "epoch": 0.01067,
      "grad_norm": 0.9442468447502006,
      "learning_rate": 0.003,
      "loss": 4.4902,
      "step": 1067
    },
    {
      "epoch": 0.01068,
      "grad_norm": 1.018364097076677,
      "learning_rate": 0.003,
      "loss": 4.4857,
      "step": 1068
    },
    {
      "epoch": 0.01069,
      "grad_norm": 0.9799434684826753,
      "learning_rate": 0.003,
      "loss": 4.4924,
      "step": 1069
    },
    {
      "epoch": 0.0107,
      "grad_norm": 0.8992499083099482,
      "learning_rate": 0.003,
      "loss": 4.5446,
      "step": 1070
    },
    {
      "epoch": 0.01071,
      "grad_norm": 0.8735440073637682,
      "learning_rate": 0.003,
      "loss": 4.4651,
      "step": 1071
    },
    {
      "epoch": 0.01072,
      "grad_norm": 0.8443267804596191,
      "learning_rate": 0.003,
      "loss": 4.5031,
      "step": 1072
    },
    {
      "epoch": 0.01073,
      "grad_norm": 0.7866639405001601,
      "learning_rate": 0.003,
      "loss": 4.4993,
      "step": 1073
    },
    {
      "epoch": 0.01074,
      "grad_norm": 0.7412592488158714,
      "learning_rate": 0.003,
      "loss": 4.4612,
      "step": 1074
    },
    {
      "epoch": 0.01075,
      "grad_norm": 0.7055777296919327,
      "learning_rate": 0.003,
      "loss": 4.4949,
      "step": 1075
    },
    {
      "epoch": 0.01076,
      "grad_norm": 0.6602103525167045,
      "learning_rate": 0.003,
      "loss": 4.4903,
      "step": 1076
    },
    {
      "epoch": 0.01077,
      "grad_norm": 0.6221585201435259,
      "learning_rate": 0.003,
      "loss": 4.4791,
      "step": 1077
    },
    {
      "epoch": 0.01078,
      "grad_norm": 0.6312740380808073,
      "learning_rate": 0.003,
      "loss": 4.4373,
      "step": 1078
    },
    {
      "epoch": 0.01079,
      "grad_norm": 0.5465751942660528,
      "learning_rate": 0.003,
      "loss": 4.4641,
      "step": 1079
    },
    {
      "epoch": 0.0108,
      "grad_norm": 0.5040258102852178,
      "learning_rate": 0.003,
      "loss": 4.4392,
      "step": 1080
    },
    {
      "epoch": 0.01081,
      "grad_norm": 0.4557353176092044,
      "learning_rate": 0.003,
      "loss": 4.4439,
      "step": 1081
    },
    {
      "epoch": 0.01082,
      "grad_norm": 0.43657211030445375,
      "learning_rate": 0.003,
      "loss": 4.4598,
      "step": 1082
    },
    {
      "epoch": 0.01083,
      "grad_norm": 0.37468227521199704,
      "learning_rate": 0.003,
      "loss": 4.4419,
      "step": 1083
    },
    {
      "epoch": 0.01084,
      "grad_norm": 0.3702365616744669,
      "learning_rate": 0.003,
      "loss": 4.4465,
      "step": 1084
    },
    {
      "epoch": 0.01085,
      "grad_norm": 0.3248887668649726,
      "learning_rate": 0.003,
      "loss": 4.4323,
      "step": 1085
    },
    {
      "epoch": 0.01086,
      "grad_norm": 0.28646275942866467,
      "learning_rate": 0.003,
      "loss": 4.4059,
      "step": 1086
    },
    {
      "epoch": 0.01087,
      "grad_norm": 0.3109295903469056,
      "learning_rate": 0.003,
      "loss": 4.4215,
      "step": 1087
    },
    {
      "epoch": 0.01088,
      "grad_norm": 0.31305777007935864,
      "learning_rate": 0.003,
      "loss": 4.454,
      "step": 1088
    },
    {
      "epoch": 0.01089,
      "grad_norm": 0.3202194647994524,
      "learning_rate": 0.003,
      "loss": 4.423,
      "step": 1089
    },
    {
      "epoch": 0.0109,
      "grad_norm": 0.3546167101166988,
      "learning_rate": 0.003,
      "loss": 4.4466,
      "step": 1090
    },
    {
      "epoch": 0.01091,
      "grad_norm": 0.40979643182380626,
      "learning_rate": 0.003,
      "loss": 4.4162,
      "step": 1091
    },
    {
      "epoch": 0.01092,
      "grad_norm": 0.44852044367674554,
      "learning_rate": 0.003,
      "loss": 4.4131,
      "step": 1092
    },
    {
      "epoch": 0.01093,
      "grad_norm": 0.5395428853759895,
      "learning_rate": 0.003,
      "loss": 4.4351,
      "step": 1093
    },
    {
      "epoch": 0.01094,
      "grad_norm": 0.7419472173607777,
      "learning_rate": 0.003,
      "loss": 4.4344,
      "step": 1094
    },
    {
      "epoch": 0.01095,
      "grad_norm": 1.3268112301670232,
      "learning_rate": 0.003,
      "loss": 4.4875,
      "step": 1095
    },
    {
      "epoch": 0.01096,
      "grad_norm": 0.8107867356113418,
      "learning_rate": 0.003,
      "loss": 4.4576,
      "step": 1096
    },
    {
      "epoch": 0.01097,
      "grad_norm": 0.6635063888056452,
      "learning_rate": 0.003,
      "loss": 4.4388,
      "step": 1097
    },
    {
      "epoch": 0.01098,
      "grad_norm": 0.9662363707419442,
      "learning_rate": 0.003,
      "loss": 4.474,
      "step": 1098
    },
    {
      "epoch": 0.01099,
      "grad_norm": 0.7696101380141932,
      "learning_rate": 0.003,
      "loss": 4.4405,
      "step": 1099
    },
    {
      "epoch": 0.011,
      "grad_norm": 0.7450236895824817,
      "learning_rate": 0.003,
      "loss": 4.4451,
      "step": 1100
    },
    {
      "epoch": 0.01101,
      "grad_norm": 0.7254968126029079,
      "learning_rate": 0.003,
      "loss": 4.4513,
      "step": 1101
    },
    {
      "epoch": 0.01102,
      "grad_norm": 0.7537628081039499,
      "learning_rate": 0.003,
      "loss": 4.4588,
      "step": 1102
    },
    {
      "epoch": 0.01103,
      "grad_norm": 0.7123173358553468,
      "learning_rate": 0.003,
      "loss": 4.4551,
      "step": 1103
    },
    {
      "epoch": 0.01104,
      "grad_norm": 0.6033148725235706,
      "learning_rate": 0.003,
      "loss": 4.4384,
      "step": 1104
    },
    {
      "epoch": 0.01105,
      "grad_norm": 0.7223359575565769,
      "learning_rate": 0.003,
      "loss": 4.4643,
      "step": 1105
    },
    {
      "epoch": 0.01106,
      "grad_norm": 0.6749769688136343,
      "learning_rate": 0.003,
      "loss": 4.4355,
      "step": 1106
    },
    {
      "epoch": 0.01107,
      "grad_norm": 0.6519921729558845,
      "learning_rate": 0.003,
      "loss": 4.443,
      "step": 1107
    },
    {
      "epoch": 0.01108,
      "grad_norm": 0.6442390747026363,
      "learning_rate": 0.003,
      "loss": 4.4155,
      "step": 1108
    },
    {
      "epoch": 0.01109,
      "grad_norm": 0.5210761074063659,
      "learning_rate": 0.003,
      "loss": 4.4258,
      "step": 1109
    },
    {
      "epoch": 0.0111,
      "grad_norm": 0.49112961811162964,
      "learning_rate": 0.003,
      "loss": 4.4439,
      "step": 1110
    },
    {
      "epoch": 0.01111,
      "grad_norm": 0.5154689325771941,
      "learning_rate": 0.003,
      "loss": 4.4228,
      "step": 1111
    },
    {
      "epoch": 0.01112,
      "grad_norm": 0.5147734027357755,
      "learning_rate": 0.003,
      "loss": 4.4383,
      "step": 1112
    },
    {
      "epoch": 0.01113,
      "grad_norm": 0.4905886669705696,
      "learning_rate": 0.003,
      "loss": 4.4046,
      "step": 1113
    },
    {
      "epoch": 0.01114,
      "grad_norm": 0.47555859430135783,
      "learning_rate": 0.003,
      "loss": 4.4166,
      "step": 1114
    },
    {
      "epoch": 0.01115,
      "grad_norm": 0.42768329796719207,
      "learning_rate": 0.003,
      "loss": 4.4369,
      "step": 1115
    },
    {
      "epoch": 0.01116,
      "grad_norm": 0.4281709726428523,
      "learning_rate": 0.003,
      "loss": 4.4286,
      "step": 1116
    },
    {
      "epoch": 0.01117,
      "grad_norm": 0.4124788023594352,
      "learning_rate": 0.003,
      "loss": 4.414,
      "step": 1117
    },
    {
      "epoch": 0.01118,
      "grad_norm": 0.4531459371875401,
      "learning_rate": 0.003,
      "loss": 4.415,
      "step": 1118
    },
    {
      "epoch": 0.01119,
      "grad_norm": 0.5115410769007107,
      "learning_rate": 0.003,
      "loss": 4.4247,
      "step": 1119
    },
    {
      "epoch": 0.0112,
      "grad_norm": 0.6598099600325713,
      "learning_rate": 0.003,
      "loss": 4.4115,
      "step": 1120
    },
    {
      "epoch": 0.01121,
      "grad_norm": 0.7298362819040065,
      "learning_rate": 0.003,
      "loss": 4.4356,
      "step": 1121
    },
    {
      "epoch": 0.01122,
      "grad_norm": 0.6383918217981295,
      "learning_rate": 0.003,
      "loss": 4.4021,
      "step": 1122
    },
    {
      "epoch": 0.01123,
      "grad_norm": 0.7157274101955465,
      "learning_rate": 0.003,
      "loss": 4.3942,
      "step": 1123
    },
    {
      "epoch": 0.01124,
      "grad_norm": 0.6763302773515354,
      "learning_rate": 0.003,
      "loss": 4.4444,
      "step": 1124
    },
    {
      "epoch": 0.01125,
      "grad_norm": 0.5890253028490829,
      "learning_rate": 0.003,
      "loss": 4.4062,
      "step": 1125
    },
    {
      "epoch": 0.01126,
      "grad_norm": 0.6745452077137704,
      "learning_rate": 0.003,
      "loss": 4.4204,
      "step": 1126
    },
    {
      "epoch": 0.01127,
      "grad_norm": 0.6212265837956056,
      "learning_rate": 0.003,
      "loss": 4.446,
      "step": 1127
    },
    {
      "epoch": 0.01128,
      "grad_norm": 0.6579375902442356,
      "learning_rate": 0.003,
      "loss": 4.4285,
      "step": 1128
    },
    {
      "epoch": 0.01129,
      "grad_norm": 0.7038241088934594,
      "learning_rate": 0.003,
      "loss": 4.4197,
      "step": 1129
    },
    {
      "epoch": 0.0113,
      "grad_norm": 0.8537912734504166,
      "learning_rate": 0.003,
      "loss": 4.4099,
      "step": 1130
    },
    {
      "epoch": 0.01131,
      "grad_norm": 0.6676066708341881,
      "learning_rate": 0.003,
      "loss": 4.4184,
      "step": 1131
    },
    {
      "epoch": 0.01132,
      "grad_norm": 0.6262944206433371,
      "learning_rate": 0.003,
      "loss": 4.4418,
      "step": 1132
    },
    {
      "epoch": 0.01133,
      "grad_norm": 0.7096964978520276,
      "learning_rate": 0.003,
      "loss": 4.4077,
      "step": 1133
    },
    {
      "epoch": 0.01134,
      "grad_norm": 0.7063046600590702,
      "learning_rate": 0.003,
      "loss": 4.4219,
      "step": 1134
    },
    {
      "epoch": 0.01135,
      "grad_norm": 0.7389408975678705,
      "learning_rate": 0.003,
      "loss": 4.4511,
      "step": 1135
    },
    {
      "epoch": 0.01136,
      "grad_norm": 0.902932943298548,
      "learning_rate": 0.003,
      "loss": 4.4334,
      "step": 1136
    },
    {
      "epoch": 0.01137,
      "grad_norm": 0.8104098050855381,
      "learning_rate": 0.003,
      "loss": 4.4615,
      "step": 1137
    },
    {
      "epoch": 0.01138,
      "grad_norm": 0.6304382373186352,
      "learning_rate": 0.003,
      "loss": 4.4061,
      "step": 1138
    },
    {
      "epoch": 0.01139,
      "grad_norm": 0.6573985975094263,
      "learning_rate": 0.003,
      "loss": 4.4174,
      "step": 1139
    },
    {
      "epoch": 0.0114,
      "grad_norm": 0.6243548176788066,
      "learning_rate": 0.003,
      "loss": 4.4015,
      "step": 1140
    },
    {
      "epoch": 0.01141,
      "grad_norm": 0.6941761654842915,
      "learning_rate": 0.003,
      "loss": 4.4278,
      "step": 1141
    },
    {
      "epoch": 0.01142,
      "grad_norm": 0.6422841400280502,
      "learning_rate": 0.003,
      "loss": 4.4042,
      "step": 1142
    },
    {
      "epoch": 0.01143,
      "grad_norm": 0.6274623913648985,
      "learning_rate": 0.003,
      "loss": 4.4193,
      "step": 1143
    },
    {
      "epoch": 0.01144,
      "grad_norm": 0.6449672978265372,
      "learning_rate": 0.003,
      "loss": 4.3861,
      "step": 1144
    },
    {
      "epoch": 0.01145,
      "grad_norm": 0.8094555842673681,
      "learning_rate": 0.003,
      "loss": 4.4312,
      "step": 1145
    },
    {
      "epoch": 0.01146,
      "grad_norm": 0.8440842425424007,
      "learning_rate": 0.003,
      "loss": 4.4247,
      "step": 1146
    },
    {
      "epoch": 0.01147,
      "grad_norm": 0.8250289231813209,
      "learning_rate": 0.003,
      "loss": 4.4121,
      "step": 1147
    },
    {
      "epoch": 0.01148,
      "grad_norm": 0.9842608879774175,
      "learning_rate": 0.003,
      "loss": 4.4248,
      "step": 1148
    },
    {
      "epoch": 0.01149,
      "grad_norm": 1.140890319957312,
      "learning_rate": 0.003,
      "loss": 4.4515,
      "step": 1149
    },
    {
      "epoch": 0.0115,
      "grad_norm": 1.0049560913231448,
      "learning_rate": 0.003,
      "loss": 4.454,
      "step": 1150
    },
    {
      "epoch": 0.01151,
      "grad_norm": 0.9241096807179571,
      "learning_rate": 0.003,
      "loss": 4.4438,
      "step": 1151
    },
    {
      "epoch": 0.01152,
      "grad_norm": 0.8197935874101315,
      "learning_rate": 0.003,
      "loss": 4.4496,
      "step": 1152
    },
    {
      "epoch": 0.01153,
      "grad_norm": 0.8964418195355868,
      "learning_rate": 0.003,
      "loss": 4.4768,
      "step": 1153
    },
    {
      "epoch": 0.01154,
      "grad_norm": 0.7052475574958695,
      "learning_rate": 0.003,
      "loss": 4.4381,
      "step": 1154
    },
    {
      "epoch": 0.01155,
      "grad_norm": 0.6994161040159336,
      "learning_rate": 0.003,
      "loss": 4.425,
      "step": 1155
    },
    {
      "epoch": 0.01156,
      "grad_norm": 0.5866991021029688,
      "learning_rate": 0.003,
      "loss": 4.4267,
      "step": 1156
    },
    {
      "epoch": 0.01157,
      "grad_norm": 0.5890541796764595,
      "learning_rate": 0.003,
      "loss": 4.4503,
      "step": 1157
    },
    {
      "epoch": 0.01158,
      "grad_norm": 0.5558417103751193,
      "learning_rate": 0.003,
      "loss": 4.4146,
      "step": 1158
    },
    {
      "epoch": 0.01159,
      "grad_norm": 0.5494688252527891,
      "learning_rate": 0.003,
      "loss": 4.41,
      "step": 1159
    },
    {
      "epoch": 0.0116,
      "grad_norm": 0.5721755744732883,
      "learning_rate": 0.003,
      "loss": 4.4228,
      "step": 1160
    },
    {
      "epoch": 0.01161,
      "grad_norm": 0.6717218566295324,
      "learning_rate": 0.003,
      "loss": 4.4019,
      "step": 1161
    },
    {
      "epoch": 0.01162,
      "grad_norm": 0.7268973886687413,
      "learning_rate": 0.003,
      "loss": 4.4138,
      "step": 1162
    },
    {
      "epoch": 0.01163,
      "grad_norm": 0.5842283691456034,
      "learning_rate": 0.003,
      "loss": 4.4284,
      "step": 1163
    },
    {
      "epoch": 0.01164,
      "grad_norm": 0.4855856513804374,
      "learning_rate": 0.003,
      "loss": 4.4134,
      "step": 1164
    },
    {
      "epoch": 0.01165,
      "grad_norm": 0.4188604676147223,
      "learning_rate": 0.003,
      "loss": 4.4035,
      "step": 1165
    },
    {
      "epoch": 0.01166,
      "grad_norm": 0.3540156009620338,
      "learning_rate": 0.003,
      "loss": 4.4019,
      "step": 1166
    },
    {
      "epoch": 0.01167,
      "grad_norm": 0.34211903868248933,
      "learning_rate": 0.003,
      "loss": 4.3663,
      "step": 1167
    },
    {
      "epoch": 0.01168,
      "grad_norm": 0.3891962127459521,
      "learning_rate": 0.003,
      "loss": 4.374,
      "step": 1168
    },
    {
      "epoch": 0.01169,
      "grad_norm": 0.43062143986497,
      "learning_rate": 0.003,
      "loss": 4.4147,
      "step": 1169
    },
    {
      "epoch": 0.0117,
      "grad_norm": 0.5082945055708723,
      "learning_rate": 0.003,
      "loss": 4.404,
      "step": 1170
    },
    {
      "epoch": 0.01171,
      "grad_norm": 0.6470102757597544,
      "learning_rate": 0.003,
      "loss": 4.3885,
      "step": 1171
    },
    {
      "epoch": 0.01172,
      "grad_norm": 0.7362820078056241,
      "learning_rate": 0.003,
      "loss": 4.3869,
      "step": 1172
    },
    {
      "epoch": 0.01173,
      "grad_norm": 0.693269388291317,
      "learning_rate": 0.003,
      "loss": 4.4171,
      "step": 1173
    },
    {
      "epoch": 0.01174,
      "grad_norm": 0.5646769911878592,
      "learning_rate": 0.003,
      "loss": 4.3891,
      "step": 1174
    },
    {
      "epoch": 0.01175,
      "grad_norm": 0.5675605646992437,
      "learning_rate": 0.003,
      "loss": 4.3751,
      "step": 1175
    },
    {
      "epoch": 0.01176,
      "grad_norm": 0.5874888204753396,
      "learning_rate": 0.003,
      "loss": 4.4094,
      "step": 1176
    },
    {
      "epoch": 0.01177,
      "grad_norm": 0.4402531057426568,
      "learning_rate": 0.003,
      "loss": 4.3711,
      "step": 1177
    },
    {
      "epoch": 0.01178,
      "grad_norm": 0.542910388266956,
      "learning_rate": 0.003,
      "loss": 4.3959,
      "step": 1178
    },
    {
      "epoch": 0.01179,
      "grad_norm": 0.6077990883816204,
      "learning_rate": 0.003,
      "loss": 4.3751,
      "step": 1179
    },
    {
      "epoch": 0.0118,
      "grad_norm": 0.5935696360755632,
      "learning_rate": 0.003,
      "loss": 4.4393,
      "step": 1180
    },
    {
      "epoch": 0.01181,
      "grad_norm": 0.5077497618150401,
      "learning_rate": 0.003,
      "loss": 4.4227,
      "step": 1181
    },
    {
      "epoch": 0.01182,
      "grad_norm": 0.4525800009003507,
      "learning_rate": 0.003,
      "loss": 4.3791,
      "step": 1182
    },
    {
      "epoch": 0.01183,
      "grad_norm": 0.465730915302176,
      "learning_rate": 0.003,
      "loss": 4.3741,
      "step": 1183
    },
    {
      "epoch": 0.01184,
      "grad_norm": 0.5691375544130988,
      "learning_rate": 0.003,
      "loss": 4.4172,
      "step": 1184
    },
    {
      "epoch": 0.01185,
      "grad_norm": 0.6220094109188331,
      "learning_rate": 0.003,
      "loss": 4.3896,
      "step": 1185
    },
    {
      "epoch": 0.01186,
      "grad_norm": 0.7180162404965751,
      "learning_rate": 0.003,
      "loss": 4.417,
      "step": 1186
    },
    {
      "epoch": 0.01187,
      "grad_norm": 0.7918251317294935,
      "learning_rate": 0.003,
      "loss": 4.389,
      "step": 1187
    },
    {
      "epoch": 0.01188,
      "grad_norm": 0.8521642061482538,
      "learning_rate": 0.003,
      "loss": 4.4081,
      "step": 1188
    },
    {
      "epoch": 0.01189,
      "grad_norm": 0.884743262111852,
      "learning_rate": 0.003,
      "loss": 4.4173,
      "step": 1189
    },
    {
      "epoch": 0.0119,
      "grad_norm": 0.809315268231435,
      "learning_rate": 0.003,
      "loss": 4.3961,
      "step": 1190
    },
    {
      "epoch": 0.01191,
      "grad_norm": 0.8426026711506704,
      "learning_rate": 0.003,
      "loss": 4.3805,
      "step": 1191
    },
    {
      "epoch": 0.01192,
      "grad_norm": 0.8039017391990436,
      "learning_rate": 0.003,
      "loss": 4.4134,
      "step": 1192
    },
    {
      "epoch": 0.01193,
      "grad_norm": 1.0091767993853729,
      "learning_rate": 0.003,
      "loss": 4.4278,
      "step": 1193
    },
    {
      "epoch": 0.01194,
      "grad_norm": 1.021446612913599,
      "learning_rate": 0.003,
      "loss": 4.4458,
      "step": 1194
    },
    {
      "epoch": 0.01195,
      "grad_norm": 0.8197186317539255,
      "learning_rate": 0.003,
      "loss": 4.4136,
      "step": 1195
    },
    {
      "epoch": 0.01196,
      "grad_norm": 1.0663075325190898,
      "learning_rate": 0.003,
      "loss": 4.4346,
      "step": 1196
    },
    {
      "epoch": 0.01197,
      "grad_norm": 1.1163749964317249,
      "learning_rate": 0.003,
      "loss": 4.42,
      "step": 1197
    },
    {
      "epoch": 0.01198,
      "grad_norm": 0.7911818503418244,
      "learning_rate": 0.003,
      "loss": 4.4756,
      "step": 1198
    },
    {
      "epoch": 0.01199,
      "grad_norm": 0.7094954502714929,
      "learning_rate": 0.003,
      "loss": 4.4429,
      "step": 1199
    },
    {
      "epoch": 0.012,
      "grad_norm": 0.6371349856097592,
      "learning_rate": 0.003,
      "loss": 4.4311,
      "step": 1200
    },
    {
      "epoch": 0.01201,
      "grad_norm": 0.6670059038710348,
      "learning_rate": 0.003,
      "loss": 4.4586,
      "step": 1201
    },
    {
      "epoch": 0.01202,
      "grad_norm": 0.7057344621036731,
      "learning_rate": 0.003,
      "loss": 4.4253,
      "step": 1202
    },
    {
      "epoch": 0.01203,
      "grad_norm": 0.6813638781778574,
      "learning_rate": 0.003,
      "loss": 4.4314,
      "step": 1203
    },
    {
      "epoch": 0.01204,
      "grad_norm": 0.6487927748495244,
      "learning_rate": 0.003,
      "loss": 4.4252,
      "step": 1204
    },
    {
      "epoch": 0.01205,
      "grad_norm": 0.5702088035511471,
      "learning_rate": 0.003,
      "loss": 4.4134,
      "step": 1205
    },
    {
      "epoch": 0.01206,
      "grad_norm": 0.5185412999807271,
      "learning_rate": 0.003,
      "loss": 4.4527,
      "step": 1206
    },
    {
      "epoch": 0.01207,
      "grad_norm": 0.5529206635637681,
      "learning_rate": 0.003,
      "loss": 4.3748,
      "step": 1207
    },
    {
      "epoch": 0.01208,
      "grad_norm": 0.6027594396835911,
      "learning_rate": 0.003,
      "loss": 4.4355,
      "step": 1208
    },
    {
      "epoch": 0.01209,
      "grad_norm": 0.6294683615242583,
      "learning_rate": 0.003,
      "loss": 4.4139,
      "step": 1209
    },
    {
      "epoch": 0.0121,
      "grad_norm": 0.5634425369332556,
      "learning_rate": 0.003,
      "loss": 4.4142,
      "step": 1210
    },
    {
      "epoch": 0.01211,
      "grad_norm": 0.489983773731107,
      "learning_rate": 0.003,
      "loss": 4.3953,
      "step": 1211
    },
    {
      "epoch": 0.01212,
      "grad_norm": 0.48590408635878257,
      "learning_rate": 0.003,
      "loss": 4.4163,
      "step": 1212
    },
    {
      "epoch": 0.01213,
      "grad_norm": 0.4654031475329597,
      "learning_rate": 0.003,
      "loss": 4.3921,
      "step": 1213
    },
    {
      "epoch": 0.01214,
      "grad_norm": 0.4519325610344299,
      "learning_rate": 0.003,
      "loss": 4.3755,
      "step": 1214
    },
    {
      "epoch": 0.01215,
      "grad_norm": 0.4986453835908023,
      "learning_rate": 0.003,
      "loss": 4.3833,
      "step": 1215
    },
    {
      "epoch": 0.01216,
      "grad_norm": 0.6836671205807361,
      "learning_rate": 0.003,
      "loss": 4.3816,
      "step": 1216
    },
    {
      "epoch": 0.01217,
      "grad_norm": 0.8634560994181636,
      "learning_rate": 0.003,
      "loss": 4.4035,
      "step": 1217
    },
    {
      "epoch": 0.01218,
      "grad_norm": 0.7430166358012542,
      "learning_rate": 0.003,
      "loss": 4.3918,
      "step": 1218
    },
    {
      "epoch": 0.01219,
      "grad_norm": 0.5693399114743081,
      "learning_rate": 0.003,
      "loss": 4.3834,
      "step": 1219
    },
    {
      "epoch": 0.0122,
      "grad_norm": 0.6203041570130704,
      "learning_rate": 0.003,
      "loss": 4.3803,
      "step": 1220
    },
    {
      "epoch": 0.01221,
      "grad_norm": 0.6221216866599081,
      "learning_rate": 0.003,
      "loss": 4.3932,
      "step": 1221
    },
    {
      "epoch": 0.01222,
      "grad_norm": 0.5678017035835031,
      "learning_rate": 0.003,
      "loss": 4.3744,
      "step": 1222
    },
    {
      "epoch": 0.01223,
      "grad_norm": 0.4544260730538049,
      "learning_rate": 0.003,
      "loss": 4.3669,
      "step": 1223
    },
    {
      "epoch": 0.01224,
      "grad_norm": 0.4498657047747612,
      "learning_rate": 0.003,
      "loss": 4.3528,
      "step": 1224
    },
    {
      "epoch": 0.01225,
      "grad_norm": 0.44954702929076595,
      "learning_rate": 0.003,
      "loss": 4.3751,
      "step": 1225
    },
    {
      "epoch": 0.01226,
      "grad_norm": 0.45922987677597427,
      "learning_rate": 0.003,
      "loss": 4.3678,
      "step": 1226
    },
    {
      "epoch": 0.01227,
      "grad_norm": 0.444736072897216,
      "learning_rate": 0.003,
      "loss": 4.3955,
      "step": 1227
    },
    {
      "epoch": 0.01228,
      "grad_norm": 0.39964234776344315,
      "learning_rate": 0.003,
      "loss": 4.3749,
      "step": 1228
    },
    {
      "epoch": 0.01229,
      "grad_norm": 0.4304709469798317,
      "learning_rate": 0.003,
      "loss": 4.3636,
      "step": 1229
    },
    {
      "epoch": 0.0123,
      "grad_norm": 0.45096371519587236,
      "learning_rate": 0.003,
      "loss": 4.3946,
      "step": 1230
    },
    {
      "epoch": 0.01231,
      "grad_norm": 0.45260016324469293,
      "learning_rate": 0.003,
      "loss": 4.3662,
      "step": 1231
    },
    {
      "epoch": 0.01232,
      "grad_norm": 0.5056071407476121,
      "learning_rate": 0.003,
      "loss": 4.3416,
      "step": 1232
    },
    {
      "epoch": 0.01233,
      "grad_norm": 0.6137961234604853,
      "learning_rate": 0.003,
      "loss": 4.3562,
      "step": 1233
    },
    {
      "epoch": 0.01234,
      "grad_norm": 0.7580924108092045,
      "learning_rate": 0.003,
      "loss": 4.3771,
      "step": 1234
    },
    {
      "epoch": 0.01235,
      "grad_norm": 0.7682739117958468,
      "learning_rate": 0.003,
      "loss": 4.377,
      "step": 1235
    },
    {
      "epoch": 0.01236,
      "grad_norm": 0.7726095308103765,
      "learning_rate": 0.003,
      "loss": 4.379,
      "step": 1236
    },
    {
      "epoch": 0.01237,
      "grad_norm": 0.6515520396776402,
      "learning_rate": 0.003,
      "loss": 4.3868,
      "step": 1237
    },
    {
      "epoch": 0.01238,
      "grad_norm": 0.5911401797128396,
      "learning_rate": 0.003,
      "loss": 4.3811,
      "step": 1238
    },
    {
      "epoch": 0.01239,
      "grad_norm": 0.6868743136698521,
      "learning_rate": 0.003,
      "loss": 4.3689,
      "step": 1239
    },
    {
      "epoch": 0.0124,
      "grad_norm": 0.7235342138686123,
      "learning_rate": 0.003,
      "loss": 4.3578,
      "step": 1240
    },
    {
      "epoch": 0.01241,
      "grad_norm": 0.7086323715527123,
      "learning_rate": 0.003,
      "loss": 4.3714,
      "step": 1241
    },
    {
      "epoch": 0.01242,
      "grad_norm": 0.7273279516261576,
      "learning_rate": 0.003,
      "loss": 4.3772,
      "step": 1242
    },
    {
      "epoch": 0.01243,
      "grad_norm": 0.7241329186524991,
      "learning_rate": 0.003,
      "loss": 4.358,
      "step": 1243
    },
    {
      "epoch": 0.01244,
      "grad_norm": 0.659589685877625,
      "learning_rate": 0.003,
      "loss": 4.3697,
      "step": 1244
    },
    {
      "epoch": 0.01245,
      "grad_norm": 0.5679488489092839,
      "learning_rate": 0.003,
      "loss": 4.3588,
      "step": 1245
    },
    {
      "epoch": 0.01246,
      "grad_norm": 0.5843246373122641,
      "learning_rate": 0.003,
      "loss": 4.3356,
      "step": 1246
    },
    {
      "epoch": 0.01247,
      "grad_norm": 0.5386207185407424,
      "learning_rate": 0.003,
      "loss": 4.3884,
      "step": 1247
    },
    {
      "epoch": 0.01248,
      "grad_norm": 0.4947062329390487,
      "learning_rate": 0.003,
      "loss": 4.3718,
      "step": 1248
    },
    {
      "epoch": 0.01249,
      "grad_norm": 0.47539230004779465,
      "learning_rate": 0.003,
      "loss": 4.3597,
      "step": 1249
    },
    {
      "epoch": 0.0125,
      "grad_norm": 0.4876155686681362,
      "learning_rate": 0.003,
      "loss": 4.3353,
      "step": 1250
    },
    {
      "epoch": 0.01251,
      "grad_norm": 0.587984128950104,
      "learning_rate": 0.003,
      "loss": 4.3725,
      "step": 1251
    },
    {
      "epoch": 0.01252,
      "grad_norm": 0.74609447185642,
      "learning_rate": 0.003,
      "loss": 4.3585,
      "step": 1252
    },
    {
      "epoch": 0.01253,
      "grad_norm": 0.8297104750282515,
      "learning_rate": 0.003,
      "loss": 4.3426,
      "step": 1253
    },
    {
      "epoch": 0.01254,
      "grad_norm": 0.6714602053035584,
      "learning_rate": 0.003,
      "loss": 4.3879,
      "step": 1254
    },
    {
      "epoch": 0.01255,
      "grad_norm": 0.5575284697102026,
      "learning_rate": 0.003,
      "loss": 4.3328,
      "step": 1255
    },
    {
      "epoch": 0.01256,
      "grad_norm": 0.6531801875316168,
      "learning_rate": 0.003,
      "loss": 4.3617,
      "step": 1256
    },
    {
      "epoch": 0.01257,
      "grad_norm": 0.6725833064810157,
      "learning_rate": 0.003,
      "loss": 4.3523,
      "step": 1257
    },
    {
      "epoch": 0.01258,
      "grad_norm": 0.5415379584452896,
      "learning_rate": 0.003,
      "loss": 4.3533,
      "step": 1258
    },
    {
      "epoch": 0.01259,
      "grad_norm": 0.6189096611706483,
      "learning_rate": 0.003,
      "loss": 4.3565,
      "step": 1259
    },
    {
      "epoch": 0.0126,
      "grad_norm": 0.7719709473721972,
      "learning_rate": 0.003,
      "loss": 4.3943,
      "step": 1260
    },
    {
      "epoch": 0.01261,
      "grad_norm": 0.7141309260632757,
      "learning_rate": 0.003,
      "loss": 4.3828,
      "step": 1261
    },
    {
      "epoch": 0.01262,
      "grad_norm": 0.82201580215985,
      "learning_rate": 0.003,
      "loss": 4.3803,
      "step": 1262
    },
    {
      "epoch": 0.01263,
      "grad_norm": 0.7648394097891381,
      "learning_rate": 0.003,
      "loss": 4.3764,
      "step": 1263
    },
    {
      "epoch": 0.01264,
      "grad_norm": 0.6534936022081808,
      "learning_rate": 0.003,
      "loss": 4.3615,
      "step": 1264
    },
    {
      "epoch": 0.01265,
      "grad_norm": 0.8083064566897021,
      "learning_rate": 0.003,
      "loss": 4.3795,
      "step": 1265
    },
    {
      "epoch": 0.01266,
      "grad_norm": 0.7625627444580204,
      "learning_rate": 0.003,
      "loss": 4.3817,
      "step": 1266
    },
    {
      "epoch": 0.01267,
      "grad_norm": 0.7144669763786441,
      "learning_rate": 0.003,
      "loss": 4.3961,
      "step": 1267
    },
    {
      "epoch": 0.01268,
      "grad_norm": 0.6176763363932604,
      "learning_rate": 0.003,
      "loss": 4.3815,
      "step": 1268
    },
    {
      "epoch": 0.01269,
      "grad_norm": 0.5844219741679575,
      "learning_rate": 0.003,
      "loss": 4.3442,
      "step": 1269
    },
    {
      "epoch": 0.0127,
      "grad_norm": 0.5570920646541462,
      "learning_rate": 0.003,
      "loss": 4.3764,
      "step": 1270
    },
    {
      "epoch": 0.01271,
      "grad_norm": 0.665679672510672,
      "learning_rate": 0.003,
      "loss": 4.3628,
      "step": 1271
    },
    {
      "epoch": 0.01272,
      "grad_norm": 0.7700987517359916,
      "learning_rate": 0.003,
      "loss": 4.3638,
      "step": 1272
    },
    {
      "epoch": 0.01273,
      "grad_norm": 0.7473118276250652,
      "learning_rate": 0.003,
      "loss": 4.3932,
      "step": 1273
    },
    {
      "epoch": 0.01274,
      "grad_norm": 0.6491851354934152,
      "learning_rate": 0.003,
      "loss": 4.3789,
      "step": 1274
    },
    {
      "epoch": 0.01275,
      "grad_norm": 0.6744635149292406,
      "learning_rate": 0.003,
      "loss": 4.3494,
      "step": 1275
    },
    {
      "epoch": 0.01276,
      "grad_norm": 0.652401206080157,
      "learning_rate": 0.003,
      "loss": 4.3572,
      "step": 1276
    },
    {
      "epoch": 0.01277,
      "grad_norm": 0.8226392169345714,
      "learning_rate": 0.003,
      "loss": 4.3706,
      "step": 1277
    },
    {
      "epoch": 0.01278,
      "grad_norm": 0.8325255271008436,
      "learning_rate": 0.003,
      "loss": 4.3906,
      "step": 1278
    },
    {
      "epoch": 0.01279,
      "grad_norm": 0.7029576010205191,
      "learning_rate": 0.003,
      "loss": 4.3612,
      "step": 1279
    },
    {
      "epoch": 0.0128,
      "grad_norm": 0.6056600803694466,
      "learning_rate": 0.003,
      "loss": 4.3548,
      "step": 1280
    },
    {
      "epoch": 0.01281,
      "grad_norm": 0.6068824045406035,
      "learning_rate": 0.003,
      "loss": 4.3433,
      "step": 1281
    },
    {
      "epoch": 0.01282,
      "grad_norm": 0.5449477323585487,
      "learning_rate": 0.003,
      "loss": 4.3801,
      "step": 1282
    },
    {
      "epoch": 0.01283,
      "grad_norm": 0.5086376593278678,
      "learning_rate": 0.003,
      "loss": 4.3465,
      "step": 1283
    },
    {
      "epoch": 0.01284,
      "grad_norm": 0.49507526795006285,
      "learning_rate": 0.003,
      "loss": 4.3484,
      "step": 1284
    },
    {
      "epoch": 0.01285,
      "grad_norm": 0.4794063886412805,
      "learning_rate": 0.003,
      "loss": 4.3499,
      "step": 1285
    },
    {
      "epoch": 0.01286,
      "grad_norm": 0.4314216967223394,
      "learning_rate": 0.003,
      "loss": 4.37,
      "step": 1286
    },
    {
      "epoch": 0.01287,
      "grad_norm": 0.43751633400555473,
      "learning_rate": 0.003,
      "loss": 4.3355,
      "step": 1287
    },
    {
      "epoch": 0.01288,
      "grad_norm": 0.4402964980769675,
      "learning_rate": 0.003,
      "loss": 4.3477,
      "step": 1288
    },
    {
      "epoch": 0.01289,
      "grad_norm": 0.4330370123851316,
      "learning_rate": 0.003,
      "loss": 4.34,
      "step": 1289
    },
    {
      "epoch": 0.0129,
      "grad_norm": 0.4577360505134826,
      "learning_rate": 0.003,
      "loss": 4.3613,
      "step": 1290
    },
    {
      "epoch": 0.01291,
      "grad_norm": 0.4908761222350223,
      "learning_rate": 0.003,
      "loss": 4.3434,
      "step": 1291
    },
    {
      "epoch": 0.01292,
      "grad_norm": 0.5197962656056466,
      "learning_rate": 0.003,
      "loss": 4.3769,
      "step": 1292
    },
    {
      "epoch": 0.01293,
      "grad_norm": 0.5249407732060195,
      "learning_rate": 0.003,
      "loss": 4.3636,
      "step": 1293
    },
    {
      "epoch": 0.01294,
      "grad_norm": 0.611698986416238,
      "learning_rate": 0.003,
      "loss": 4.3337,
      "step": 1294
    },
    {
      "epoch": 0.01295,
      "grad_norm": 0.7156402524759599,
      "learning_rate": 0.003,
      "loss": 4.369,
      "step": 1295
    },
    {
      "epoch": 0.01296,
      "grad_norm": 0.8400871327514794,
      "learning_rate": 0.003,
      "loss": 4.3429,
      "step": 1296
    },
    {
      "epoch": 0.01297,
      "grad_norm": 0.9437097734689512,
      "learning_rate": 0.003,
      "loss": 4.3707,
      "step": 1297
    },
    {
      "epoch": 0.01298,
      "grad_norm": 0.9584374674542437,
      "learning_rate": 0.003,
      "loss": 4.3505,
      "step": 1298
    },
    {
      "epoch": 0.01299,
      "grad_norm": 0.8671660761915297,
      "learning_rate": 0.003,
      "loss": 4.3625,
      "step": 1299
    },
    {
      "epoch": 0.013,
      "grad_norm": 0.7502528073695865,
      "learning_rate": 0.003,
      "loss": 4.3495,
      "step": 1300
    },
    {
      "epoch": 0.01301,
      "grad_norm": 0.7063515582852367,
      "learning_rate": 0.003,
      "loss": 4.3619,
      "step": 1301
    },
    {
      "epoch": 0.01302,
      "grad_norm": 0.6149436229664508,
      "learning_rate": 0.003,
      "loss": 4.4034,
      "step": 1302
    },
    {
      "epoch": 0.01303,
      "grad_norm": 0.6398205689943912,
      "learning_rate": 0.003,
      "loss": 4.3511,
      "step": 1303
    },
    {
      "epoch": 0.01304,
      "grad_norm": 0.5125654671319883,
      "learning_rate": 0.003,
      "loss": 4.3743,
      "step": 1304
    },
    {
      "epoch": 0.01305,
      "grad_norm": 0.49151327704885306,
      "learning_rate": 0.003,
      "loss": 4.357,
      "step": 1305
    },
    {
      "epoch": 0.01306,
      "grad_norm": 0.449035259984806,
      "learning_rate": 0.003,
      "loss": 4.3496,
      "step": 1306
    },
    {
      "epoch": 0.01307,
      "grad_norm": 0.3786564732477656,
      "learning_rate": 0.003,
      "loss": 4.365,
      "step": 1307
    },
    {
      "epoch": 0.01308,
      "grad_norm": 0.40747339154306583,
      "learning_rate": 0.003,
      "loss": 4.3407,
      "step": 1308
    },
    {
      "epoch": 0.01309,
      "grad_norm": 0.3477777170477395,
      "learning_rate": 0.003,
      "loss": 4.3771,
      "step": 1309
    },
    {
      "epoch": 0.0131,
      "grad_norm": 0.33985557007437,
      "learning_rate": 0.003,
      "loss": 4.3221,
      "step": 1310
    },
    {
      "epoch": 0.01311,
      "grad_norm": 0.3781744690142208,
      "learning_rate": 0.003,
      "loss": 4.3621,
      "step": 1311
    },
    {
      "epoch": 0.01312,
      "grad_norm": 0.390430095045798,
      "learning_rate": 0.003,
      "loss": 4.3295,
      "step": 1312
    },
    {
      "epoch": 0.01313,
      "grad_norm": 0.4411339748954226,
      "learning_rate": 0.003,
      "loss": 4.3219,
      "step": 1313
    },
    {
      "epoch": 0.01314,
      "grad_norm": 0.4398621851217481,
      "learning_rate": 0.003,
      "loss": 4.3182,
      "step": 1314
    },
    {
      "epoch": 0.01315,
      "grad_norm": 0.4543520117536255,
      "learning_rate": 0.003,
      "loss": 4.3454,
      "step": 1315
    },
    {
      "epoch": 0.01316,
      "grad_norm": 0.4647119001484631,
      "learning_rate": 0.003,
      "loss": 4.3315,
      "step": 1316
    },
    {
      "epoch": 0.01317,
      "grad_norm": 0.6634726948283797,
      "learning_rate": 0.003,
      "loss": 4.3607,
      "step": 1317
    },
    {
      "epoch": 0.01318,
      "grad_norm": 1.0496539085022096,
      "learning_rate": 0.003,
      "loss": 4.3666,
      "step": 1318
    },
    {
      "epoch": 0.01319,
      "grad_norm": 1.3165114729988252,
      "learning_rate": 0.003,
      "loss": 4.392,
      "step": 1319
    },
    {
      "epoch": 0.0132,
      "grad_norm": 0.7650097283623507,
      "learning_rate": 0.003,
      "loss": 4.3723,
      "step": 1320
    },
    {
      "epoch": 0.01321,
      "grad_norm": 0.7692901736546863,
      "learning_rate": 0.003,
      "loss": 4.3563,
      "step": 1321
    },
    {
      "epoch": 0.01322,
      "grad_norm": 0.8430579356733313,
      "learning_rate": 0.003,
      "loss": 4.3811,
      "step": 1322
    },
    {
      "epoch": 0.01323,
      "grad_norm": 0.7787983772827455,
      "learning_rate": 0.003,
      "loss": 4.3749,
      "step": 1323
    },
    {
      "epoch": 0.01324,
      "grad_norm": 0.7585640630062677,
      "learning_rate": 0.003,
      "loss": 4.3595,
      "step": 1324
    },
    {
      "epoch": 0.01325,
      "grad_norm": 0.662871102926864,
      "learning_rate": 0.003,
      "loss": 4.3604,
      "step": 1325
    },
    {
      "epoch": 0.01326,
      "grad_norm": 0.6552542644296041,
      "learning_rate": 0.003,
      "loss": 4.3471,
      "step": 1326
    },
    {
      "epoch": 0.01327,
      "grad_norm": 0.6597442221568894,
      "learning_rate": 0.003,
      "loss": 4.3636,
      "step": 1327
    },
    {
      "epoch": 0.01328,
      "grad_norm": 0.6991774516177431,
      "learning_rate": 0.003,
      "loss": 4.3572,
      "step": 1328
    },
    {
      "epoch": 0.01329,
      "grad_norm": 0.7004568547673504,
      "learning_rate": 0.003,
      "loss": 4.323,
      "step": 1329
    },
    {
      "epoch": 0.0133,
      "grad_norm": 0.6451796491929548,
      "learning_rate": 0.003,
      "loss": 4.3619,
      "step": 1330
    },
    {
      "epoch": 0.01331,
      "grad_norm": 0.5603586741834496,
      "learning_rate": 0.003,
      "loss": 4.3461,
      "step": 1331
    },
    {
      "epoch": 0.01332,
      "grad_norm": 0.6118296137650349,
      "learning_rate": 0.003,
      "loss": 4.3674,
      "step": 1332
    },
    {
      "epoch": 0.01333,
      "grad_norm": 0.6556746462798643,
      "learning_rate": 0.003,
      "loss": 4.3385,
      "step": 1333
    },
    {
      "epoch": 0.01334,
      "grad_norm": 0.6569171248898434,
      "learning_rate": 0.003,
      "loss": 4.3292,
      "step": 1334
    },
    {
      "epoch": 0.01335,
      "grad_norm": 0.7308490257197903,
      "learning_rate": 0.003,
      "loss": 4.355,
      "step": 1335
    },
    {
      "epoch": 0.01336,
      "grad_norm": 0.7694263760249205,
      "learning_rate": 0.003,
      "loss": 4.3562,
      "step": 1336
    },
    {
      "epoch": 0.01337,
      "grad_norm": 0.829484027358136,
      "learning_rate": 0.003,
      "loss": 4.3739,
      "step": 1337
    },
    {
      "epoch": 0.01338,
      "grad_norm": 0.7394720058707402,
      "learning_rate": 0.003,
      "loss": 4.3375,
      "step": 1338
    },
    {
      "epoch": 0.01339,
      "grad_norm": 0.7663877331658486,
      "learning_rate": 0.003,
      "loss": 4.3614,
      "step": 1339
    },
    {
      "epoch": 0.0134,
      "grad_norm": 0.6814802672525545,
      "learning_rate": 0.003,
      "loss": 4.3431,
      "step": 1340
    },
    {
      "epoch": 0.01341,
      "grad_norm": 0.5804799698607723,
      "learning_rate": 0.003,
      "loss": 4.3566,
      "step": 1341
    },
    {
      "epoch": 0.01342,
      "grad_norm": 0.5489998904660267,
      "learning_rate": 0.003,
      "loss": 4.3242,
      "step": 1342
    },
    {
      "epoch": 0.01343,
      "grad_norm": 0.5528070757921824,
      "learning_rate": 0.003,
      "loss": 4.3308,
      "step": 1343
    },
    {
      "epoch": 0.01344,
      "grad_norm": 0.5249427916401641,
      "learning_rate": 0.003,
      "loss": 4.3286,
      "step": 1344
    },
    {
      "epoch": 0.01345,
      "grad_norm": 0.5796492875823627,
      "learning_rate": 0.003,
      "loss": 4.3698,
      "step": 1345
    },
    {
      "epoch": 0.01346,
      "grad_norm": 0.7791779206443923,
      "learning_rate": 0.003,
      "loss": 4.349,
      "step": 1346
    },
    {
      "epoch": 0.01347,
      "grad_norm": 0.9787862373774305,
      "learning_rate": 0.003,
      "loss": 4.3601,
      "step": 1347
    },
    {
      "epoch": 0.01348,
      "grad_norm": 1.124691751963475,
      "learning_rate": 0.003,
      "loss": 4.3695,
      "step": 1348
    },
    {
      "epoch": 0.01349,
      "grad_norm": 0.7239534112442599,
      "learning_rate": 0.003,
      "loss": 4.3703,
      "step": 1349
    },
    {
      "epoch": 0.0135,
      "grad_norm": 0.7058824966707771,
      "learning_rate": 0.003,
      "loss": 4.3665,
      "step": 1350
    },
    {
      "epoch": 0.01351,
      "grad_norm": 0.7269767228946674,
      "learning_rate": 0.003,
      "loss": 4.3913,
      "step": 1351
    },
    {
      "epoch": 0.01352,
      "grad_norm": 0.629762174967035,
      "learning_rate": 0.003,
      "loss": 4.3652,
      "step": 1352
    },
    {
      "epoch": 0.01353,
      "grad_norm": 0.46709847900974316,
      "learning_rate": 0.003,
      "loss": 4.3426,
      "step": 1353
    },
    {
      "epoch": 0.01354,
      "grad_norm": 0.4348570974243878,
      "learning_rate": 0.003,
      "loss": 4.33,
      "step": 1354
    },
    {
      "epoch": 0.01355,
      "grad_norm": 0.4058626528377844,
      "learning_rate": 0.003,
      "loss": 4.3397,
      "step": 1355
    },
    {
      "epoch": 0.01356,
      "grad_norm": 0.4077904728688183,
      "learning_rate": 0.003,
      "loss": 4.3329,
      "step": 1356
    },
    {
      "epoch": 0.01357,
      "grad_norm": 0.3881492908104237,
      "learning_rate": 0.003,
      "loss": 4.3196,
      "step": 1357
    },
    {
      "epoch": 0.01358,
      "grad_norm": 0.37883341535998916,
      "learning_rate": 0.003,
      "loss": 4.3135,
      "step": 1358
    },
    {
      "epoch": 0.01359,
      "grad_norm": 0.3673079033329598,
      "learning_rate": 0.003,
      "loss": 4.3515,
      "step": 1359
    },
    {
      "epoch": 0.0136,
      "grad_norm": 0.37976733712994504,
      "learning_rate": 0.003,
      "loss": 4.3177,
      "step": 1360
    },
    {
      "epoch": 0.01361,
      "grad_norm": 0.4118348378412375,
      "learning_rate": 0.003,
      "loss": 4.3313,
      "step": 1361
    },
    {
      "epoch": 0.01362,
      "grad_norm": 0.4407593683397167,
      "learning_rate": 0.003,
      "loss": 4.345,
      "step": 1362
    },
    {
      "epoch": 0.01363,
      "grad_norm": 0.5058671106167254,
      "learning_rate": 0.003,
      "loss": 4.3464,
      "step": 1363
    },
    {
      "epoch": 0.01364,
      "grad_norm": 0.6807238620694973,
      "learning_rate": 0.003,
      "loss": 4.3446,
      "step": 1364
    },
    {
      "epoch": 0.01365,
      "grad_norm": 0.8032111226209311,
      "learning_rate": 0.003,
      "loss": 4.3702,
      "step": 1365
    },
    {
      "epoch": 0.01366,
      "grad_norm": 0.7976336994683142,
      "learning_rate": 0.003,
      "loss": 4.3233,
      "step": 1366
    },
    {
      "epoch": 0.01367,
      "grad_norm": 0.7881501972797743,
      "learning_rate": 0.003,
      "loss": 4.3766,
      "step": 1367
    },
    {
      "epoch": 0.01368,
      "grad_norm": 0.6437176795201012,
      "learning_rate": 0.003,
      "loss": 4.346,
      "step": 1368
    },
    {
      "epoch": 0.01369,
      "grad_norm": 0.7544973412872852,
      "learning_rate": 0.003,
      "loss": 4.3393,
      "step": 1369
    },
    {
      "epoch": 0.0137,
      "grad_norm": 0.9164166166999201,
      "learning_rate": 0.003,
      "loss": 4.3794,
      "step": 1370
    },
    {
      "epoch": 0.01371,
      "grad_norm": 0.9303263432444765,
      "learning_rate": 0.003,
      "loss": 4.3692,
      "step": 1371
    },
    {
      "epoch": 0.01372,
      "grad_norm": 0.7939151912920627,
      "learning_rate": 0.003,
      "loss": 4.3176,
      "step": 1372
    },
    {
      "epoch": 0.01373,
      "grad_norm": 0.8088750323088586,
      "learning_rate": 0.003,
      "loss": 4.3626,
      "step": 1373
    },
    {
      "epoch": 0.01374,
      "grad_norm": 0.6721224672262793,
      "learning_rate": 0.003,
      "loss": 4.3585,
      "step": 1374
    },
    {
      "epoch": 0.01375,
      "grad_norm": 0.6771097427365759,
      "learning_rate": 0.003,
      "loss": 4.3249,
      "step": 1375
    },
    {
      "epoch": 0.01376,
      "grad_norm": 0.6762752470113543,
      "learning_rate": 0.003,
      "loss": 4.347,
      "step": 1376
    },
    {
      "epoch": 0.01377,
      "grad_norm": 0.6604690846076395,
      "learning_rate": 0.003,
      "loss": 4.3466,
      "step": 1377
    },
    {
      "epoch": 0.01378,
      "grad_norm": 0.8027631418948692,
      "learning_rate": 0.003,
      "loss": 4.3812,
      "step": 1378
    },
    {
      "epoch": 0.01379,
      "grad_norm": 0.8058991494927565,
      "learning_rate": 0.003,
      "loss": 4.3543,
      "step": 1379
    },
    {
      "epoch": 0.0138,
      "grad_norm": 0.6522516144796072,
      "learning_rate": 0.003,
      "loss": 4.3459,
      "step": 1380
    },
    {
      "epoch": 0.01381,
      "grad_norm": 0.6569912385058241,
      "learning_rate": 0.003,
      "loss": 4.3524,
      "step": 1381
    },
    {
      "epoch": 0.01382,
      "grad_norm": 0.5996275495095534,
      "learning_rate": 0.003,
      "loss": 4.3471,
      "step": 1382
    },
    {
      "epoch": 0.01383,
      "grad_norm": 0.7099037071072233,
      "learning_rate": 0.003,
      "loss": 4.3442,
      "step": 1383
    },
    {
      "epoch": 0.01384,
      "grad_norm": 0.9143332820449893,
      "learning_rate": 0.003,
      "loss": 4.3854,
      "step": 1384
    },
    {
      "epoch": 0.01385,
      "grad_norm": 1.0346171765908472,
      "learning_rate": 0.003,
      "loss": 4.3815,
      "step": 1385
    },
    {
      "epoch": 0.01386,
      "grad_norm": 0.8985740978326894,
      "learning_rate": 0.003,
      "loss": 4.3558,
      "step": 1386
    },
    {
      "epoch": 0.01387,
      "grad_norm": 0.6958974730086266,
      "learning_rate": 0.003,
      "loss": 4.335,
      "step": 1387
    },
    {
      "epoch": 0.01388,
      "grad_norm": 0.7497022428211844,
      "learning_rate": 0.003,
      "loss": 4.3735,
      "step": 1388
    },
    {
      "epoch": 0.01389,
      "grad_norm": 0.7334888360507559,
      "learning_rate": 0.003,
      "loss": 4.3541,
      "step": 1389
    },
    {
      "epoch": 0.0139,
      "grad_norm": 0.6634716599193375,
      "learning_rate": 0.003,
      "loss": 4.3533,
      "step": 1390
    },
    {
      "epoch": 0.01391,
      "grad_norm": 0.5884420459817105,
      "learning_rate": 0.003,
      "loss": 4.3483,
      "step": 1391
    },
    {
      "epoch": 0.01392,
      "grad_norm": 0.536202889592441,
      "learning_rate": 0.003,
      "loss": 4.3599,
      "step": 1392
    },
    {
      "epoch": 0.01393,
      "grad_norm": 0.5041805750347179,
      "learning_rate": 0.003,
      "loss": 4.327,
      "step": 1393
    },
    {
      "epoch": 0.01394,
      "grad_norm": 0.46187226586984687,
      "learning_rate": 0.003,
      "loss": 4.3118,
      "step": 1394
    },
    {
      "epoch": 0.01395,
      "grad_norm": 0.4642311356997612,
      "learning_rate": 0.003,
      "loss": 4.343,
      "step": 1395
    },
    {
      "epoch": 0.01396,
      "grad_norm": 0.49219234547480073,
      "learning_rate": 0.003,
      "loss": 4.3333,
      "step": 1396
    },
    {
      "epoch": 0.01397,
      "grad_norm": 0.535253115362672,
      "learning_rate": 0.003,
      "loss": 4.3471,
      "step": 1397
    },
    {
      "epoch": 0.01398,
      "grad_norm": 0.6052714387106423,
      "learning_rate": 0.003,
      "loss": 4.3267,
      "step": 1398
    },
    {
      "epoch": 0.01399,
      "grad_norm": 0.6277477924110956,
      "learning_rate": 0.003,
      "loss": 4.3504,
      "step": 1399
    },
    {
      "epoch": 0.014,
      "grad_norm": 0.6846592906181149,
      "learning_rate": 0.003,
      "loss": 4.3103,
      "step": 1400
    },
    {
      "epoch": 0.01401,
      "grad_norm": 0.8200984983857136,
      "learning_rate": 0.003,
      "loss": 4.3387,
      "step": 1401
    },
    {
      "epoch": 0.01402,
      "grad_norm": 0.8742108124137586,
      "learning_rate": 0.003,
      "loss": 4.3557,
      "step": 1402
    },
    {
      "epoch": 0.01403,
      "grad_norm": 0.7096844776193044,
      "learning_rate": 0.003,
      "loss": 4.3233,
      "step": 1403
    },
    {
      "epoch": 0.01404,
      "grad_norm": 0.5388679123577896,
      "learning_rate": 0.003,
      "loss": 4.3401,
      "step": 1404
    },
    {
      "epoch": 0.01405,
      "grad_norm": 0.4910540359683292,
      "learning_rate": 0.003,
      "loss": 4.3559,
      "step": 1405
    },
    {
      "epoch": 0.01406,
      "grad_norm": 0.5213447941957257,
      "learning_rate": 0.003,
      "loss": 4.3366,
      "step": 1406
    },
    {
      "epoch": 0.01407,
      "grad_norm": 0.555926299222398,
      "learning_rate": 0.003,
      "loss": 4.3521,
      "step": 1407
    },
    {
      "epoch": 0.01408,
      "grad_norm": 0.692327807092523,
      "learning_rate": 0.003,
      "loss": 4.3137,
      "step": 1408
    },
    {
      "epoch": 0.01409,
      "grad_norm": 0.8697281888439876,
      "learning_rate": 0.003,
      "loss": 4.3588,
      "step": 1409
    },
    {
      "epoch": 0.0141,
      "grad_norm": 0.9566955718451174,
      "learning_rate": 0.003,
      "loss": 4.3446,
      "step": 1410
    },
    {
      "epoch": 0.01411,
      "grad_norm": 0.7215244847418736,
      "learning_rate": 0.003,
      "loss": 4.3131,
      "step": 1411
    },
    {
      "epoch": 0.01412,
      "grad_norm": 0.6315141391735041,
      "learning_rate": 0.003,
      "loss": 4.3535,
      "step": 1412
    },
    {
      "epoch": 0.01413,
      "grad_norm": 0.6841381857718712,
      "learning_rate": 0.003,
      "loss": 4.342,
      "step": 1413
    },
    {
      "epoch": 0.01414,
      "grad_norm": 0.5798611042820683,
      "learning_rate": 0.003,
      "loss": 4.3235,
      "step": 1414
    },
    {
      "epoch": 0.01415,
      "grad_norm": 0.547398429036414,
      "learning_rate": 0.003,
      "loss": 4.3157,
      "step": 1415
    },
    {
      "epoch": 0.01416,
      "grad_norm": 0.532688568072748,
      "learning_rate": 0.003,
      "loss": 4.3388,
      "step": 1416
    },
    {
      "epoch": 0.01417,
      "grad_norm": 0.5474366342351658,
      "learning_rate": 0.003,
      "loss": 4.3317,
      "step": 1417
    },
    {
      "epoch": 0.01418,
      "grad_norm": 0.5499547169322332,
      "learning_rate": 0.003,
      "loss": 4.3477,
      "step": 1418
    },
    {
      "epoch": 0.01419,
      "grad_norm": 0.58527195779541,
      "learning_rate": 0.003,
      "loss": 4.3403,
      "step": 1419
    },
    {
      "epoch": 0.0142,
      "grad_norm": 0.4513568933907034,
      "learning_rate": 0.003,
      "loss": 4.3401,
      "step": 1420
    },
    {
      "epoch": 0.01421,
      "grad_norm": 0.433334289258551,
      "learning_rate": 0.003,
      "loss": 4.3072,
      "step": 1421
    },
    {
      "epoch": 0.01422,
      "grad_norm": 0.39636752625840327,
      "learning_rate": 0.003,
      "loss": 4.3095,
      "step": 1422
    },
    {
      "epoch": 0.01423,
      "grad_norm": 0.38308187149456013,
      "learning_rate": 0.003,
      "loss": 4.3141,
      "step": 1423
    },
    {
      "epoch": 0.01424,
      "grad_norm": 0.3793175762281477,
      "learning_rate": 0.003,
      "loss": 4.3274,
      "step": 1424
    },
    {
      "epoch": 0.01425,
      "grad_norm": 0.4426704955835019,
      "learning_rate": 0.003,
      "loss": 4.334,
      "step": 1425
    },
    {
      "epoch": 0.01426,
      "grad_norm": 0.5006373442027768,
      "learning_rate": 0.003,
      "loss": 4.3106,
      "step": 1426
    },
    {
      "epoch": 0.01427,
      "grad_norm": 0.654459732986379,
      "learning_rate": 0.003,
      "loss": 4.3121,
      "step": 1427
    },
    {
      "epoch": 0.01428,
      "grad_norm": 0.8067469794918197,
      "learning_rate": 0.003,
      "loss": 4.3172,
      "step": 1428
    },
    {
      "epoch": 0.01429,
      "grad_norm": 0.8729241550485854,
      "learning_rate": 0.003,
      "loss": 4.3306,
      "step": 1429
    },
    {
      "epoch": 0.0143,
      "grad_norm": 0.7278485959649278,
      "learning_rate": 0.003,
      "loss": 4.3432,
      "step": 1430
    },
    {
      "epoch": 0.01431,
      "grad_norm": 0.7004237859708793,
      "learning_rate": 0.003,
      "loss": 4.3301,
      "step": 1431
    },
    {
      "epoch": 0.01432,
      "grad_norm": 0.8171598559675952,
      "learning_rate": 0.003,
      "loss": 4.3267,
      "step": 1432
    },
    {
      "epoch": 0.01433,
      "grad_norm": 0.7196947939958487,
      "learning_rate": 0.003,
      "loss": 4.323,
      "step": 1433
    },
    {
      "epoch": 0.01434,
      "grad_norm": 0.6846858190137799,
      "learning_rate": 0.003,
      "loss": 4.3048,
      "step": 1434
    },
    {
      "epoch": 0.01435,
      "grad_norm": 0.8455490886566032,
      "learning_rate": 0.003,
      "loss": 4.3242,
      "step": 1435
    },
    {
      "epoch": 0.01436,
      "grad_norm": 0.83940902394985,
      "learning_rate": 0.003,
      "loss": 4.3453,
      "step": 1436
    },
    {
      "epoch": 0.01437,
      "grad_norm": 0.7328785180307785,
      "learning_rate": 0.003,
      "loss": 4.3547,
      "step": 1437
    },
    {
      "epoch": 0.01438,
      "grad_norm": 0.7369569320140708,
      "learning_rate": 0.003,
      "loss": 4.349,
      "step": 1438
    },
    {
      "epoch": 0.01439,
      "grad_norm": 0.7139149757788127,
      "learning_rate": 0.003,
      "loss": 4.3129,
      "step": 1439
    },
    {
      "epoch": 0.0144,
      "grad_norm": 0.7299011882512658,
      "learning_rate": 0.003,
      "loss": 4.3275,
      "step": 1440
    },
    {
      "epoch": 0.01441,
      "grad_norm": 0.670750314419178,
      "learning_rate": 0.003,
      "loss": 4.3309,
      "step": 1441
    },
    {
      "epoch": 0.01442,
      "grad_norm": 0.7691195459017623,
      "learning_rate": 0.003,
      "loss": 4.3219,
      "step": 1442
    },
    {
      "epoch": 0.01443,
      "grad_norm": 0.7471914299871487,
      "learning_rate": 0.003,
      "loss": 4.3829,
      "step": 1443
    },
    {
      "epoch": 0.01444,
      "grad_norm": 0.7410199276775529,
      "learning_rate": 0.003,
      "loss": 4.3445,
      "step": 1444
    },
    {
      "epoch": 0.01445,
      "grad_norm": 0.6515002318773166,
      "learning_rate": 0.003,
      "loss": 4.3041,
      "step": 1445
    },
    {
      "epoch": 0.01446,
      "grad_norm": 0.6781090959776701,
      "learning_rate": 0.003,
      "loss": 4.3207,
      "step": 1446
    },
    {
      "epoch": 0.01447,
      "grad_norm": 0.6643065072159584,
      "learning_rate": 0.003,
      "loss": 4.3293,
      "step": 1447
    },
    {
      "epoch": 0.01448,
      "grad_norm": 0.6532713033801402,
      "learning_rate": 0.003,
      "loss": 4.3375,
      "step": 1448
    },
    {
      "epoch": 0.01449,
      "grad_norm": 0.6344501403492658,
      "learning_rate": 0.003,
      "loss": 4.3302,
      "step": 1449
    },
    {
      "epoch": 0.0145,
      "grad_norm": 0.5404283744793761,
      "learning_rate": 0.003,
      "loss": 4.3401,
      "step": 1450
    },
    {
      "epoch": 0.01451,
      "grad_norm": 0.5381895515345636,
      "learning_rate": 0.003,
      "loss": 4.3181,
      "step": 1451
    },
    {
      "epoch": 0.01452,
      "grad_norm": 0.5085114651371503,
      "learning_rate": 0.003,
      "loss": 4.3176,
      "step": 1452
    },
    {
      "epoch": 0.01453,
      "grad_norm": 0.5297212710147206,
      "learning_rate": 0.003,
      "loss": 4.3302,
      "step": 1453
    },
    {
      "epoch": 0.01454,
      "grad_norm": 0.6082304279549751,
      "learning_rate": 0.003,
      "loss": 4.299,
      "step": 1454
    },
    {
      "epoch": 0.01455,
      "grad_norm": 0.799761378996593,
      "learning_rate": 0.003,
      "loss": 4.3379,
      "step": 1455
    },
    {
      "epoch": 0.01456,
      "grad_norm": 0.8200199061255905,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1456
    },
    {
      "epoch": 0.01457,
      "grad_norm": 0.6729328173983412,
      "learning_rate": 0.003,
      "loss": 4.3054,
      "step": 1457
    },
    {
      "epoch": 0.01458,
      "grad_norm": 0.5761787071333488,
      "learning_rate": 0.003,
      "loss": 4.2955,
      "step": 1458
    },
    {
      "epoch": 0.01459,
      "grad_norm": 0.6468019060138231,
      "learning_rate": 0.003,
      "loss": 4.3289,
      "step": 1459
    },
    {
      "epoch": 0.0146,
      "grad_norm": 0.6012651007343053,
      "learning_rate": 0.003,
      "loss": 4.3283,
      "step": 1460
    },
    {
      "epoch": 0.01461,
      "grad_norm": 0.4832305210154182,
      "learning_rate": 0.003,
      "loss": 4.3037,
      "step": 1461
    },
    {
      "epoch": 0.01462,
      "grad_norm": 0.4933402571331549,
      "learning_rate": 0.003,
      "loss": 4.3088,
      "step": 1462
    },
    {
      "epoch": 0.01463,
      "grad_norm": 0.4774020866387011,
      "learning_rate": 0.003,
      "loss": 4.3072,
      "step": 1463
    },
    {
      "epoch": 0.01464,
      "grad_norm": 0.4604171826341042,
      "learning_rate": 0.003,
      "loss": 4.2821,
      "step": 1464
    },
    {
      "epoch": 0.01465,
      "grad_norm": 0.4294089797917102,
      "learning_rate": 0.003,
      "loss": 4.3245,
      "step": 1465
    },
    {
      "epoch": 0.01466,
      "grad_norm": 0.41929661999999984,
      "learning_rate": 0.003,
      "loss": 4.2919,
      "step": 1466
    },
    {
      "epoch": 0.01467,
      "grad_norm": 0.4391785760355703,
      "learning_rate": 0.003,
      "loss": 4.3283,
      "step": 1467
    },
    {
      "epoch": 0.01468,
      "grad_norm": 0.4538846759969807,
      "learning_rate": 0.003,
      "loss": 4.3037,
      "step": 1468
    },
    {
      "epoch": 0.01469,
      "grad_norm": 0.5119886030394242,
      "learning_rate": 0.003,
      "loss": 4.3053,
      "step": 1469
    },
    {
      "epoch": 0.0147,
      "grad_norm": 0.716327286811815,
      "learning_rate": 0.003,
      "loss": 4.3041,
      "step": 1470
    },
    {
      "epoch": 0.01471,
      "grad_norm": 0.9150052503540761,
      "learning_rate": 0.003,
      "loss": 4.3274,
      "step": 1471
    },
    {
      "epoch": 0.01472,
      "grad_norm": 0.7809983181295278,
      "learning_rate": 0.003,
      "loss": 4.3143,
      "step": 1472
    },
    {
      "epoch": 0.01473,
      "grad_norm": 0.8086105929105535,
      "learning_rate": 0.003,
      "loss": 4.3289,
      "step": 1473
    },
    {
      "epoch": 0.01474,
      "grad_norm": 0.8656063557310713,
      "learning_rate": 0.003,
      "loss": 4.3258,
      "step": 1474
    },
    {
      "epoch": 0.01475,
      "grad_norm": 0.8637306499434225,
      "learning_rate": 0.003,
      "loss": 4.3398,
      "step": 1475
    },
    {
      "epoch": 0.01476,
      "grad_norm": 0.6743755524690542,
      "learning_rate": 0.003,
      "loss": 4.3227,
      "step": 1476
    },
    {
      "epoch": 0.01477,
      "grad_norm": 0.6414161546100536,
      "learning_rate": 0.003,
      "loss": 4.3256,
      "step": 1477
    },
    {
      "epoch": 0.01478,
      "grad_norm": 0.7067557912668699,
      "learning_rate": 0.003,
      "loss": 4.3113,
      "step": 1478
    },
    {
      "epoch": 0.01479,
      "grad_norm": 0.7427717638657159,
      "learning_rate": 0.003,
      "loss": 4.3499,
      "step": 1479
    },
    {
      "epoch": 0.0148,
      "grad_norm": 0.7972486147312002,
      "learning_rate": 0.003,
      "loss": 4.3212,
      "step": 1480
    },
    {
      "epoch": 0.01481,
      "grad_norm": 0.8238319142257903,
      "learning_rate": 0.003,
      "loss": 4.3261,
      "step": 1481
    },
    {
      "epoch": 0.01482,
      "grad_norm": 0.7010334962030825,
      "learning_rate": 0.003,
      "loss": 4.3089,
      "step": 1482
    },
    {
      "epoch": 0.01483,
      "grad_norm": 0.6287754589698554,
      "learning_rate": 0.003,
      "loss": 4.303,
      "step": 1483
    },
    {
      "epoch": 0.01484,
      "grad_norm": 0.5953574922658559,
      "learning_rate": 0.003,
      "loss": 4.3534,
      "step": 1484
    },
    {
      "epoch": 0.01485,
      "grad_norm": 0.6859671225358986,
      "learning_rate": 0.003,
      "loss": 4.3172,
      "step": 1485
    },
    {
      "epoch": 0.01486,
      "grad_norm": 0.7393843625917039,
      "learning_rate": 0.003,
      "loss": 4.3181,
      "step": 1486
    },
    {
      "epoch": 0.01487,
      "grad_norm": 0.6948333501030851,
      "learning_rate": 0.003,
      "loss": 4.3225,
      "step": 1487
    },
    {
      "epoch": 0.01488,
      "grad_norm": 0.5806619385051136,
      "learning_rate": 0.003,
      "loss": 4.3259,
      "step": 1488
    },
    {
      "epoch": 0.01489,
      "grad_norm": 0.6342989775695094,
      "learning_rate": 0.003,
      "loss": 4.3221,
      "step": 1489
    },
    {
      "epoch": 0.0149,
      "grad_norm": 0.6704129451979153,
      "learning_rate": 0.003,
      "loss": 4.3147,
      "step": 1490
    },
    {
      "epoch": 0.01491,
      "grad_norm": 0.7017206728292057,
      "learning_rate": 0.003,
      "loss": 4.3305,
      "step": 1491
    },
    {
      "epoch": 0.01492,
      "grad_norm": 0.6813188063148864,
      "learning_rate": 0.003,
      "loss": 4.2961,
      "step": 1492
    },
    {
      "epoch": 0.01493,
      "grad_norm": 0.8525559699799811,
      "learning_rate": 0.003,
      "loss": 4.3324,
      "step": 1493
    },
    {
      "epoch": 0.01494,
      "grad_norm": 0.8687834663259932,
      "learning_rate": 0.003,
      "loss": 4.3213,
      "step": 1494
    },
    {
      "epoch": 0.01495,
      "grad_norm": 0.7274136348295218,
      "learning_rate": 0.003,
      "loss": 4.3243,
      "step": 1495
    },
    {
      "epoch": 0.01496,
      "grad_norm": 0.5434712910501038,
      "learning_rate": 0.003,
      "loss": 4.2923,
      "step": 1496
    },
    {
      "epoch": 0.01497,
      "grad_norm": 0.6148760530691155,
      "learning_rate": 0.003,
      "loss": 4.2701,
      "step": 1497
    },
    {
      "epoch": 0.01498,
      "grad_norm": 0.5665585769325068,
      "learning_rate": 0.003,
      "loss": 4.3224,
      "step": 1498
    },
    {
      "epoch": 0.01499,
      "grad_norm": 0.543044486095261,
      "learning_rate": 0.003,
      "loss": 4.3408,
      "step": 1499
    },
    {
      "epoch": 0.015,
      "grad_norm": 0.5868245294183819,
      "learning_rate": 0.003,
      "loss": 4.2952,
      "step": 1500
    },
    {
      "epoch": 0.01501,
      "grad_norm": 0.6488345831013783,
      "learning_rate": 0.003,
      "loss": 4.2943,
      "step": 1501
    },
    {
      "epoch": 0.01502,
      "grad_norm": 0.5560720951051904,
      "learning_rate": 0.003,
      "loss": 4.3322,
      "step": 1502
    },
    {
      "epoch": 0.01503,
      "grad_norm": 0.6170651368212409,
      "learning_rate": 0.003,
      "loss": 4.3142,
      "step": 1503
    },
    {
      "epoch": 0.01504,
      "grad_norm": 0.7253173696408437,
      "learning_rate": 0.003,
      "loss": 4.3199,
      "step": 1504
    },
    {
      "epoch": 0.01505,
      "grad_norm": 0.7573152828389973,
      "learning_rate": 0.003,
      "loss": 4.3067,
      "step": 1505
    },
    {
      "epoch": 0.01506,
      "grad_norm": 0.7119425101662294,
      "learning_rate": 0.003,
      "loss": 4.3233,
      "step": 1506
    },
    {
      "epoch": 0.01507,
      "grad_norm": 0.7408719494100054,
      "learning_rate": 0.003,
      "loss": 4.3394,
      "step": 1507
    },
    {
      "epoch": 0.01508,
      "grad_norm": 0.6023053184100382,
      "learning_rate": 0.003,
      "loss": 4.3108,
      "step": 1508
    },
    {
      "epoch": 0.01509,
      "grad_norm": 0.5054062616818538,
      "learning_rate": 0.003,
      "loss": 4.3257,
      "step": 1509
    },
    {
      "epoch": 0.0151,
      "grad_norm": 0.4440427822667444,
      "learning_rate": 0.003,
      "loss": 4.2952,
      "step": 1510
    },
    {
      "epoch": 0.01511,
      "grad_norm": 0.503043960784533,
      "learning_rate": 0.003,
      "loss": 4.2838,
      "step": 1511
    },
    {
      "epoch": 0.01512,
      "grad_norm": 0.4478910644887885,
      "learning_rate": 0.003,
      "loss": 4.2937,
      "step": 1512
    },
    {
      "epoch": 0.01513,
      "grad_norm": 0.46313433916934627,
      "learning_rate": 0.003,
      "loss": 4.2989,
      "step": 1513
    },
    {
      "epoch": 0.01514,
      "grad_norm": 0.5237062692888148,
      "learning_rate": 0.003,
      "loss": 4.3066,
      "step": 1514
    },
    {
      "epoch": 0.01515,
      "grad_norm": 0.6686824710181497,
      "learning_rate": 0.003,
      "loss": 4.3056,
      "step": 1515
    },
    {
      "epoch": 0.01516,
      "grad_norm": 0.8162150436928566,
      "learning_rate": 0.003,
      "loss": 4.3198,
      "step": 1516
    },
    {
      "epoch": 0.01517,
      "grad_norm": 0.8250386268904906,
      "learning_rate": 0.003,
      "loss": 4.274,
      "step": 1517
    },
    {
      "epoch": 0.01518,
      "grad_norm": 0.8551846044396628,
      "learning_rate": 0.003,
      "loss": 4.3102,
      "step": 1518
    },
    {
      "epoch": 0.01519,
      "grad_norm": 0.7778994522735176,
      "learning_rate": 0.003,
      "loss": 4.3471,
      "step": 1519
    },
    {
      "epoch": 0.0152,
      "grad_norm": 0.6407333772962789,
      "learning_rate": 0.003,
      "loss": 4.2901,
      "step": 1520
    },
    {
      "epoch": 0.01521,
      "grad_norm": 0.590099483112598,
      "learning_rate": 0.003,
      "loss": 4.3053,
      "step": 1521
    },
    {
      "epoch": 0.01522,
      "grad_norm": 0.6220605799864671,
      "learning_rate": 0.003,
      "loss": 4.3014,
      "step": 1522
    },
    {
      "epoch": 0.01523,
      "grad_norm": 0.6378640000910417,
      "learning_rate": 0.003,
      "loss": 4.2957,
      "step": 1523
    },
    {
      "epoch": 0.01524,
      "grad_norm": 0.6800153761279158,
      "learning_rate": 0.003,
      "loss": 4.3332,
      "step": 1524
    },
    {
      "epoch": 0.01525,
      "grad_norm": 0.6552309843606878,
      "learning_rate": 0.003,
      "loss": 4.3306,
      "step": 1525
    },
    {
      "epoch": 0.01526,
      "grad_norm": 0.5832236683536917,
      "learning_rate": 0.003,
      "loss": 4.3154,
      "step": 1526
    },
    {
      "epoch": 0.01527,
      "grad_norm": 0.5923964968536319,
      "learning_rate": 0.003,
      "loss": 4.2868,
      "step": 1527
    },
    {
      "epoch": 0.01528,
      "grad_norm": 0.6438689697714796,
      "learning_rate": 0.003,
      "loss": 4.3067,
      "step": 1528
    },
    {
      "epoch": 0.01529,
      "grad_norm": 0.7863523458278316,
      "learning_rate": 0.003,
      "loss": 4.3071,
      "step": 1529
    },
    {
      "epoch": 0.0153,
      "grad_norm": 0.8716684688082496,
      "learning_rate": 0.003,
      "loss": 4.3236,
      "step": 1530
    },
    {
      "epoch": 0.01531,
      "grad_norm": 0.7632243352103186,
      "learning_rate": 0.003,
      "loss": 4.314,
      "step": 1531
    },
    {
      "epoch": 0.01532,
      "grad_norm": 0.5830764209700208,
      "learning_rate": 0.003,
      "loss": 4.2863,
      "step": 1532
    },
    {
      "epoch": 0.01533,
      "grad_norm": 0.6925578088506626,
      "learning_rate": 0.003,
      "loss": 4.3193,
      "step": 1533
    },
    {
      "epoch": 0.01534,
      "grad_norm": 0.6884106279584329,
      "learning_rate": 0.003,
      "loss": 4.3131,
      "step": 1534
    },
    {
      "epoch": 0.01535,
      "grad_norm": 0.612086877528724,
      "learning_rate": 0.003,
      "loss": 4.3216,
      "step": 1535
    },
    {
      "epoch": 0.01536,
      "grad_norm": 0.6277711721176384,
      "learning_rate": 0.003,
      "loss": 4.3039,
      "step": 1536
    },
    {
      "epoch": 0.01537,
      "grad_norm": 0.7930101311629847,
      "learning_rate": 0.003,
      "loss": 4.3078,
      "step": 1537
    },
    {
      "epoch": 0.01538,
      "grad_norm": 0.8987557100700295,
      "learning_rate": 0.003,
      "loss": 4.2971,
      "step": 1538
    },
    {
      "epoch": 0.01539,
      "grad_norm": 0.9391352557198052,
      "learning_rate": 0.003,
      "loss": 4.334,
      "step": 1539
    },
    {
      "epoch": 0.0154,
      "grad_norm": 0.9440902063787158,
      "learning_rate": 0.003,
      "loss": 4.3487,
      "step": 1540
    },
    {
      "epoch": 0.01541,
      "grad_norm": 0.9127476278073733,
      "learning_rate": 0.003,
      "loss": 4.328,
      "step": 1541
    },
    {
      "epoch": 0.01542,
      "grad_norm": 0.7753546344208377,
      "learning_rate": 0.003,
      "loss": 4.3528,
      "step": 1542
    },
    {
      "epoch": 0.01543,
      "grad_norm": 0.7317390533493494,
      "learning_rate": 0.003,
      "loss": 4.3318,
      "step": 1543
    },
    {
      "epoch": 0.01544,
      "grad_norm": 0.5946634053593371,
      "learning_rate": 0.003,
      "loss": 4.3162,
      "step": 1544
    },
    {
      "epoch": 0.01545,
      "grad_norm": 0.49808121797063876,
      "learning_rate": 0.003,
      "loss": 4.375,
      "step": 1545
    },
    {
      "epoch": 0.01546,
      "grad_norm": 0.5189277397467036,
      "learning_rate": 0.003,
      "loss": 4.3435,
      "step": 1546
    },
    {
      "epoch": 0.01547,
      "grad_norm": 0.5075409907094708,
      "learning_rate": 0.003,
      "loss": 4.3431,
      "step": 1547
    },
    {
      "epoch": 0.01548,
      "grad_norm": 0.4269782583856954,
      "learning_rate": 0.003,
      "loss": 4.3016,
      "step": 1548
    },
    {
      "epoch": 0.01549,
      "grad_norm": 0.44259739876123444,
      "learning_rate": 0.003,
      "loss": 4.2898,
      "step": 1549
    },
    {
      "epoch": 0.0155,
      "grad_norm": 0.39476104723697186,
      "learning_rate": 0.003,
      "loss": 4.3072,
      "step": 1550
    },
    {
      "epoch": 0.01551,
      "grad_norm": 0.39719568498427116,
      "learning_rate": 0.003,
      "loss": 4.2836,
      "step": 1551
    },
    {
      "epoch": 0.01552,
      "grad_norm": 0.5112856286097273,
      "learning_rate": 0.003,
      "loss": 4.3059,
      "step": 1552
    },
    {
      "epoch": 0.01553,
      "grad_norm": 0.6994822828905559,
      "learning_rate": 0.003,
      "loss": 4.3366,
      "step": 1553
    },
    {
      "epoch": 0.01554,
      "grad_norm": 0.9918562900778292,
      "learning_rate": 0.003,
      "loss": 4.3337,
      "step": 1554
    },
    {
      "epoch": 0.01555,
      "grad_norm": 0.8597749628867734,
      "learning_rate": 0.003,
      "loss": 4.3154,
      "step": 1555
    },
    {
      "epoch": 0.01556,
      "grad_norm": 0.4859845337872918,
      "learning_rate": 0.003,
      "loss": 4.3299,
      "step": 1556
    },
    {
      "epoch": 0.01557,
      "grad_norm": 0.8105266239108289,
      "learning_rate": 0.003,
      "loss": 4.317,
      "step": 1557
    },
    {
      "epoch": 0.01558,
      "grad_norm": 0.6797261746777549,
      "learning_rate": 0.003,
      "loss": 4.3032,
      "step": 1558
    },
    {
      "epoch": 0.01559,
      "grad_norm": 0.5540991472031954,
      "learning_rate": 0.003,
      "loss": 4.3349,
      "step": 1559
    },
    {
      "epoch": 0.0156,
      "grad_norm": 0.5222478848499021,
      "learning_rate": 0.003,
      "loss": 4.3207,
      "step": 1560
    },
    {
      "epoch": 0.01561,
      "grad_norm": 0.4547690370956997,
      "learning_rate": 0.003,
      "loss": 4.3006,
      "step": 1561
    },
    {
      "epoch": 0.01562,
      "grad_norm": 0.3960981969234329,
      "learning_rate": 0.003,
      "loss": 4.2916,
      "step": 1562
    },
    {
      "epoch": 0.01563,
      "grad_norm": 0.4365008440994641,
      "learning_rate": 0.003,
      "loss": 4.2998,
      "step": 1563
    },
    {
      "epoch": 0.01564,
      "grad_norm": 0.4153235041333595,
      "learning_rate": 0.003,
      "loss": 4.3237,
      "step": 1564
    },
    {
      "epoch": 0.01565,
      "grad_norm": 0.4172430370227481,
      "learning_rate": 0.003,
      "loss": 4.2984,
      "step": 1565
    },
    {
      "epoch": 0.01566,
      "grad_norm": 0.4030631404816366,
      "learning_rate": 0.003,
      "loss": 4.3153,
      "step": 1566
    },
    {
      "epoch": 0.01567,
      "grad_norm": 0.4006086575381243,
      "learning_rate": 0.003,
      "loss": 4.3082,
      "step": 1567
    },
    {
      "epoch": 0.01568,
      "grad_norm": 0.4692595605101829,
      "learning_rate": 0.003,
      "loss": 4.2921,
      "step": 1568
    },
    {
      "epoch": 0.01569,
      "grad_norm": 0.5156092105425303,
      "learning_rate": 0.003,
      "loss": 4.2975,
      "step": 1569
    },
    {
      "epoch": 0.0157,
      "grad_norm": 0.5574730520335043,
      "learning_rate": 0.003,
      "loss": 4.3166,
      "step": 1570
    },
    {
      "epoch": 0.01571,
      "grad_norm": 0.7147942584489586,
      "learning_rate": 0.003,
      "loss": 4.306,
      "step": 1571
    },
    {
      "epoch": 0.01572,
      "grad_norm": 0.8176596681794123,
      "learning_rate": 0.003,
      "loss": 4.3202,
      "step": 1572
    },
    {
      "epoch": 0.01573,
      "grad_norm": 0.8859299700656846,
      "learning_rate": 0.003,
      "loss": 4.311,
      "step": 1573
    },
    {
      "epoch": 0.01574,
      "grad_norm": 1.0167563301248608,
      "learning_rate": 0.003,
      "loss": 4.3365,
      "step": 1574
    },
    {
      "epoch": 0.01575,
      "grad_norm": 0.9276915846212054,
      "learning_rate": 0.003,
      "loss": 4.3012,
      "step": 1575
    },
    {
      "epoch": 0.01576,
      "grad_norm": 0.7519027259067711,
      "learning_rate": 0.003,
      "loss": 4.317,
      "step": 1576
    },
    {
      "epoch": 0.01577,
      "grad_norm": 0.5990656751852772,
      "learning_rate": 0.003,
      "loss": 4.2743,
      "step": 1577
    },
    {
      "epoch": 0.01578,
      "grad_norm": 0.652141379980981,
      "learning_rate": 0.003,
      "loss": 4.3047,
      "step": 1578
    },
    {
      "epoch": 0.01579,
      "grad_norm": 0.6130941880481341,
      "learning_rate": 0.003,
      "loss": 4.3001,
      "step": 1579
    },
    {
      "epoch": 0.0158,
      "grad_norm": 0.6566738181349618,
      "learning_rate": 0.003,
      "loss": 4.3101,
      "step": 1580
    },
    {
      "epoch": 0.01581,
      "grad_norm": 0.5602300919728743,
      "learning_rate": 0.003,
      "loss": 4.3154,
      "step": 1581
    },
    {
      "epoch": 0.01582,
      "grad_norm": 0.6614515024819504,
      "learning_rate": 0.003,
      "loss": 4.3018,
      "step": 1582
    },
    {
      "epoch": 0.01583,
      "grad_norm": 0.6685116238668469,
      "learning_rate": 0.003,
      "loss": 4.2884,
      "step": 1583
    },
    {
      "epoch": 0.01584,
      "grad_norm": 0.7663983884039552,
      "learning_rate": 0.003,
      "loss": 4.3227,
      "step": 1584
    },
    {
      "epoch": 0.01585,
      "grad_norm": 0.8471607724061854,
      "learning_rate": 0.003,
      "loss": 4.3183,
      "step": 1585
    },
    {
      "epoch": 0.01586,
      "grad_norm": 0.8029205674223586,
      "learning_rate": 0.003,
      "loss": 4.2999,
      "step": 1586
    },
    {
      "epoch": 0.01587,
      "grad_norm": 0.7335558361580696,
      "learning_rate": 0.003,
      "loss": 4.3092,
      "step": 1587
    },
    {
      "epoch": 0.01588,
      "grad_norm": 0.6576426322376321,
      "learning_rate": 0.003,
      "loss": 4.3341,
      "step": 1588
    },
    {
      "epoch": 0.01589,
      "grad_norm": 0.7314604361721292,
      "learning_rate": 0.003,
      "loss": 4.328,
      "step": 1589
    },
    {
      "epoch": 0.0159,
      "grad_norm": 0.7484813941023133,
      "learning_rate": 0.003,
      "loss": 4.323,
      "step": 1590
    },
    {
      "epoch": 0.01591,
      "grad_norm": 0.6836450786052791,
      "learning_rate": 0.003,
      "loss": 4.3264,
      "step": 1591
    },
    {
      "epoch": 0.01592,
      "grad_norm": 0.6522043667699231,
      "learning_rate": 0.003,
      "loss": 4.3127,
      "step": 1592
    },
    {
      "epoch": 0.01593,
      "grad_norm": 0.6311012555499182,
      "learning_rate": 0.003,
      "loss": 4.3174,
      "step": 1593
    },
    {
      "epoch": 0.01594,
      "grad_norm": 0.5510623120073839,
      "learning_rate": 0.003,
      "loss": 4.3009,
      "step": 1594
    },
    {
      "epoch": 0.01595,
      "grad_norm": 0.48793948155833994,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1595
    },
    {
      "epoch": 0.01596,
      "grad_norm": 0.42717068818596066,
      "learning_rate": 0.003,
      "loss": 4.3142,
      "step": 1596
    },
    {
      "epoch": 0.01597,
      "grad_norm": 0.43677376148637953,
      "learning_rate": 0.003,
      "loss": 4.2773,
      "step": 1597
    },
    {
      "epoch": 0.01598,
      "grad_norm": 0.47981748858846524,
      "learning_rate": 0.003,
      "loss": 4.2706,
      "step": 1598
    },
    {
      "epoch": 0.01599,
      "grad_norm": 0.5068183957178387,
      "learning_rate": 0.003,
      "loss": 4.2944,
      "step": 1599
    },
    {
      "epoch": 0.016,
      "grad_norm": 0.570411039839509,
      "learning_rate": 0.003,
      "loss": 4.261,
      "step": 1600
    },
    {
      "epoch": 0.01601,
      "grad_norm": 0.7428825863420937,
      "learning_rate": 0.003,
      "loss": 4.3164,
      "step": 1601
    },
    {
      "epoch": 0.01602,
      "grad_norm": 1.0316614845943484,
      "learning_rate": 0.003,
      "loss": 4.3056,
      "step": 1602
    },
    {
      "epoch": 0.01603,
      "grad_norm": 0.9265092099146023,
      "learning_rate": 0.003,
      "loss": 4.3273,
      "step": 1603
    },
    {
      "epoch": 0.01604,
      "grad_norm": 0.6714138671323193,
      "learning_rate": 0.003,
      "loss": 4.3099,
      "step": 1604
    },
    {
      "epoch": 0.01605,
      "grad_norm": 0.7449186454711473,
      "learning_rate": 0.003,
      "loss": 4.2937,
      "step": 1605
    },
    {
      "epoch": 0.01606,
      "grad_norm": 0.6551508425933098,
      "learning_rate": 0.003,
      "loss": 4.29,
      "step": 1606
    },
    {
      "epoch": 0.01607,
      "grad_norm": 0.6368386814210635,
      "learning_rate": 0.003,
      "loss": 4.3196,
      "step": 1607
    },
    {
      "epoch": 0.01608,
      "grad_norm": 0.6007545339051799,
      "learning_rate": 0.003,
      "loss": 4.287,
      "step": 1608
    },
    {
      "epoch": 0.01609,
      "grad_norm": 0.5685481620694516,
      "learning_rate": 0.003,
      "loss": 4.2973,
      "step": 1609
    },
    {
      "epoch": 0.0161,
      "grad_norm": 0.5084184586189073,
      "learning_rate": 0.003,
      "loss": 4.2592,
      "step": 1610
    },
    {
      "epoch": 0.01611,
      "grad_norm": 0.4840833495127669,
      "learning_rate": 0.003,
      "loss": 4.2802,
      "step": 1611
    },
    {
      "epoch": 0.01612,
      "grad_norm": 0.48942028074279065,
      "learning_rate": 0.003,
      "loss": 4.2702,
      "step": 1612
    },
    {
      "epoch": 0.01613,
      "grad_norm": 0.5011285237133881,
      "learning_rate": 0.003,
      "loss": 4.315,
      "step": 1613
    },
    {
      "epoch": 0.01614,
      "grad_norm": 0.43995362094592044,
      "learning_rate": 0.003,
      "loss": 4.3364,
      "step": 1614
    },
    {
      "epoch": 0.01615,
      "grad_norm": 0.42434808753845177,
      "learning_rate": 0.003,
      "loss": 4.3023,
      "step": 1615
    },
    {
      "epoch": 0.01616,
      "grad_norm": 0.40673555464346545,
      "learning_rate": 0.003,
      "loss": 4.2826,
      "step": 1616
    },
    {
      "epoch": 0.01617,
      "grad_norm": 0.3578337742407294,
      "learning_rate": 0.003,
      "loss": 4.2706,
      "step": 1617
    },
    {
      "epoch": 0.01618,
      "grad_norm": 0.41081917922167766,
      "learning_rate": 0.003,
      "loss": 4.2753,
      "step": 1618
    },
    {
      "epoch": 0.01619,
      "grad_norm": 0.47886760935645833,
      "learning_rate": 0.003,
      "loss": 4.2659,
      "step": 1619
    },
    {
      "epoch": 0.0162,
      "grad_norm": 0.6222321239813,
      "learning_rate": 0.003,
      "loss": 4.2767,
      "step": 1620
    },
    {
      "epoch": 0.01621,
      "grad_norm": 0.6903491585345826,
      "learning_rate": 0.003,
      "loss": 4.2786,
      "step": 1621
    },
    {
      "epoch": 0.01622,
      "grad_norm": 0.7573622136296652,
      "learning_rate": 0.003,
      "loss": 4.3118,
      "step": 1622
    },
    {
      "epoch": 0.01623,
      "grad_norm": 0.9256995453444036,
      "learning_rate": 0.003,
      "loss": 4.3026,
      "step": 1623
    },
    {
      "epoch": 0.01624,
      "grad_norm": 0.9558305805696413,
      "learning_rate": 0.003,
      "loss": 4.3216,
      "step": 1624
    },
    {
      "epoch": 0.01625,
      "grad_norm": 0.8114385487348762,
      "learning_rate": 0.003,
      "loss": 4.2898,
      "step": 1625
    },
    {
      "epoch": 0.01626,
      "grad_norm": 0.8322826717160557,
      "learning_rate": 0.003,
      "loss": 4.3177,
      "step": 1626
    },
    {
      "epoch": 0.01627,
      "grad_norm": 0.7727713367753223,
      "learning_rate": 0.003,
      "loss": 4.2816,
      "step": 1627
    },
    {
      "epoch": 0.01628,
      "grad_norm": 0.7109224896248458,
      "learning_rate": 0.003,
      "loss": 4.2929,
      "step": 1628
    },
    {
      "epoch": 0.01629,
      "grad_norm": 0.560897124973334,
      "learning_rate": 0.003,
      "loss": 4.2858,
      "step": 1629
    },
    {
      "epoch": 0.0163,
      "grad_norm": 0.6779380557060044,
      "learning_rate": 0.003,
      "loss": 4.2826,
      "step": 1630
    },
    {
      "epoch": 0.01631,
      "grad_norm": 0.6886638086492523,
      "learning_rate": 0.003,
      "loss": 4.3011,
      "step": 1631
    },
    {
      "epoch": 0.01632,
      "grad_norm": 0.6589349584439033,
      "learning_rate": 0.003,
      "loss": 4.3124,
      "step": 1632
    },
    {
      "epoch": 0.01633,
      "grad_norm": 0.7155276302347654,
      "learning_rate": 0.003,
      "loss": 4.2893,
      "step": 1633
    },
    {
      "epoch": 0.01634,
      "grad_norm": 0.7783909364325312,
      "learning_rate": 0.003,
      "loss": 4.3094,
      "step": 1634
    },
    {
      "epoch": 0.01635,
      "grad_norm": 0.8288192042951168,
      "learning_rate": 0.003,
      "loss": 4.2952,
      "step": 1635
    },
    {
      "epoch": 0.01636,
      "grad_norm": 0.8309682765627266,
      "learning_rate": 0.003,
      "loss": 4.3157,
      "step": 1636
    },
    {
      "epoch": 0.01637,
      "grad_norm": 0.7885672766922769,
      "learning_rate": 0.003,
      "loss": 4.2985,
      "step": 1637
    },
    {
      "epoch": 0.01638,
      "grad_norm": 0.6820787883112478,
      "learning_rate": 0.003,
      "loss": 4.3026,
      "step": 1638
    },
    {
      "epoch": 0.01639,
      "grad_norm": 0.5912531628997083,
      "learning_rate": 0.003,
      "loss": 4.3221,
      "step": 1639
    },
    {
      "epoch": 0.0164,
      "grad_norm": 0.7081315099543188,
      "learning_rate": 0.003,
      "loss": 4.3062,
      "step": 1640
    },
    {
      "epoch": 0.01641,
      "grad_norm": 0.7390837101488571,
      "learning_rate": 0.003,
      "loss": 4.279,
      "step": 1641
    },
    {
      "epoch": 0.01642,
      "grad_norm": 0.8019499352702314,
      "learning_rate": 0.003,
      "loss": 4.3244,
      "step": 1642
    },
    {
      "epoch": 0.01643,
      "grad_norm": 0.7922382978586373,
      "learning_rate": 0.003,
      "loss": 4.3126,
      "step": 1643
    },
    {
      "epoch": 0.01644,
      "grad_norm": 0.7260441686104165,
      "learning_rate": 0.003,
      "loss": 4.3282,
      "step": 1644
    },
    {
      "epoch": 0.01645,
      "grad_norm": 0.5765166219346322,
      "learning_rate": 0.003,
      "loss": 4.2686,
      "step": 1645
    },
    {
      "epoch": 0.01646,
      "grad_norm": 0.4846869130781467,
      "learning_rate": 0.003,
      "loss": 4.298,
      "step": 1646
    },
    {
      "epoch": 0.01647,
      "grad_norm": 0.4324516045425632,
      "learning_rate": 0.003,
      "loss": 4.294,
      "step": 1647
    },
    {
      "epoch": 0.01648,
      "grad_norm": 0.45801960619516824,
      "learning_rate": 0.003,
      "loss": 4.2725,
      "step": 1648
    },
    {
      "epoch": 0.01649,
      "grad_norm": 0.4519054523297515,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1649
    },
    {
      "epoch": 0.0165,
      "grad_norm": 0.5030069407293299,
      "learning_rate": 0.003,
      "loss": 4.2846,
      "step": 1650
    },
    {
      "epoch": 0.01651,
      "grad_norm": 0.7338143580585036,
      "learning_rate": 0.003,
      "loss": 4.2988,
      "step": 1651
    },
    {
      "epoch": 0.01652,
      "grad_norm": 0.9179472565177694,
      "learning_rate": 0.003,
      "loss": 4.2688,
      "step": 1652
    },
    {
      "epoch": 0.01653,
      "grad_norm": 0.7974402781604697,
      "learning_rate": 0.003,
      "loss": 4.307,
      "step": 1653
    },
    {
      "epoch": 0.01654,
      "grad_norm": 0.8094858670947136,
      "learning_rate": 0.003,
      "loss": 4.2972,
      "step": 1654
    },
    {
      "epoch": 0.01655,
      "grad_norm": 0.8204525750124729,
      "learning_rate": 0.003,
      "loss": 4.313,
      "step": 1655
    },
    {
      "epoch": 0.01656,
      "grad_norm": 0.6293302149986666,
      "learning_rate": 0.003,
      "loss": 4.271,
      "step": 1656
    },
    {
      "epoch": 0.01657,
      "grad_norm": 0.5706278835378605,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 1657
    },
    {
      "epoch": 0.01658,
      "grad_norm": 0.6110254736121277,
      "learning_rate": 0.003,
      "loss": 4.2809,
      "step": 1658
    },
    {
      "epoch": 0.01659,
      "grad_norm": 0.4984603935459828,
      "learning_rate": 0.003,
      "loss": 4.2824,
      "step": 1659
    },
    {
      "epoch": 0.0166,
      "grad_norm": 0.46804303085612786,
      "learning_rate": 0.003,
      "loss": 4.2649,
      "step": 1660
    },
    {
      "epoch": 0.01661,
      "grad_norm": 0.48879681023532395,
      "learning_rate": 0.003,
      "loss": 4.2671,
      "step": 1661
    },
    {
      "epoch": 0.01662,
      "grad_norm": 0.5571765680768687,
      "learning_rate": 0.003,
      "loss": 4.2719,
      "step": 1662
    },
    {
      "epoch": 0.01663,
      "grad_norm": 0.532882366154707,
      "learning_rate": 0.003,
      "loss": 4.294,
      "step": 1663
    },
    {
      "epoch": 0.01664,
      "grad_norm": 0.5088338427561463,
      "learning_rate": 0.003,
      "loss": 4.3094,
      "step": 1664
    },
    {
      "epoch": 0.01665,
      "grad_norm": 0.4963301596331138,
      "learning_rate": 0.003,
      "loss": 4.2712,
      "step": 1665
    },
    {
      "epoch": 0.01666,
      "grad_norm": 0.5472162068898109,
      "learning_rate": 0.003,
      "loss": 4.2732,
      "step": 1666
    },
    {
      "epoch": 0.01667,
      "grad_norm": 0.6863444215859392,
      "learning_rate": 0.003,
      "loss": 4.2951,
      "step": 1667
    },
    {
      "epoch": 0.01668,
      "grad_norm": 0.7945747889757943,
      "learning_rate": 0.003,
      "loss": 4.3019,
      "step": 1668
    },
    {
      "epoch": 0.01669,
      "grad_norm": 0.7029756676891371,
      "learning_rate": 0.003,
      "loss": 4.2868,
      "step": 1669
    },
    {
      "epoch": 0.0167,
      "grad_norm": 0.564779057565769,
      "learning_rate": 0.003,
      "loss": 4.2701,
      "step": 1670
    },
    {
      "epoch": 0.01671,
      "grad_norm": 0.5946383928046679,
      "learning_rate": 0.003,
      "loss": 4.2602,
      "step": 1671
    },
    {
      "epoch": 0.01672,
      "grad_norm": 0.7328231067257597,
      "learning_rate": 0.003,
      "loss": 4.2779,
      "step": 1672
    },
    {
      "epoch": 0.01673,
      "grad_norm": 0.7730808965686631,
      "learning_rate": 0.003,
      "loss": 4.2759,
      "step": 1673
    },
    {
      "epoch": 0.01674,
      "grad_norm": 0.8250670120229329,
      "learning_rate": 0.003,
      "loss": 4.3013,
      "step": 1674
    },
    {
      "epoch": 0.01675,
      "grad_norm": 0.7145691985141881,
      "learning_rate": 0.003,
      "loss": 4.2806,
      "step": 1675
    },
    {
      "epoch": 0.01676,
      "grad_norm": 0.6268084931910631,
      "learning_rate": 0.003,
      "loss": 4.2872,
      "step": 1676
    },
    {
      "epoch": 0.01677,
      "grad_norm": 0.6170941899124056,
      "learning_rate": 0.003,
      "loss": 4.2772,
      "step": 1677
    },
    {
      "epoch": 0.01678,
      "grad_norm": 0.5951034196601164,
      "learning_rate": 0.003,
      "loss": 4.2879,
      "step": 1678
    },
    {
      "epoch": 0.01679,
      "grad_norm": 0.5254502211849061,
      "learning_rate": 0.003,
      "loss": 4.2679,
      "step": 1679
    },
    {
      "epoch": 0.0168,
      "grad_norm": 0.6096842133031645,
      "learning_rate": 0.003,
      "loss": 4.2892,
      "step": 1680
    },
    {
      "epoch": 0.01681,
      "grad_norm": 0.6218177000062469,
      "learning_rate": 0.003,
      "loss": 4.2892,
      "step": 1681
    },
    {
      "epoch": 0.01682,
      "grad_norm": 0.5832829705689185,
      "learning_rate": 0.003,
      "loss": 4.2817,
      "step": 1682
    },
    {
      "epoch": 0.01683,
      "grad_norm": 0.5910768318994952,
      "learning_rate": 0.003,
      "loss": 4.2959,
      "step": 1683
    },
    {
      "epoch": 0.01684,
      "grad_norm": 0.6050344929193974,
      "learning_rate": 0.003,
      "loss": 4.2707,
      "step": 1684
    },
    {
      "epoch": 0.01685,
      "grad_norm": 0.6382986956510986,
      "learning_rate": 0.003,
      "loss": 4.2898,
      "step": 1685
    },
    {
      "epoch": 0.01686,
      "grad_norm": 0.8077022270811443,
      "learning_rate": 0.003,
      "loss": 4.2942,
      "step": 1686
    },
    {
      "epoch": 0.01687,
      "grad_norm": 0.8169553039156104,
      "learning_rate": 0.003,
      "loss": 4.2925,
      "step": 1687
    },
    {
      "epoch": 0.01688,
      "grad_norm": 0.8368087154638648,
      "learning_rate": 0.003,
      "loss": 4.2843,
      "step": 1688
    },
    {
      "epoch": 0.01689,
      "grad_norm": 0.869445348031783,
      "learning_rate": 0.003,
      "loss": 4.3002,
      "step": 1689
    },
    {
      "epoch": 0.0169,
      "grad_norm": 0.9062293741017381,
      "learning_rate": 0.003,
      "loss": 4.2906,
      "step": 1690
    },
    {
      "epoch": 0.01691,
      "grad_norm": 0.8487759964259872,
      "learning_rate": 0.003,
      "loss": 4.3075,
      "step": 1691
    },
    {
      "epoch": 0.01692,
      "grad_norm": 0.8483641489848167,
      "learning_rate": 0.003,
      "loss": 4.3088,
      "step": 1692
    },
    {
      "epoch": 0.01693,
      "grad_norm": 0.8279189520542338,
      "learning_rate": 0.003,
      "loss": 4.3369,
      "step": 1693
    },
    {
      "epoch": 0.01694,
      "grad_norm": 0.8089873138093994,
      "learning_rate": 0.003,
      "loss": 4.2887,
      "step": 1694
    },
    {
      "epoch": 0.01695,
      "grad_norm": 0.9455670720071184,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1695
    },
    {
      "epoch": 0.01696,
      "grad_norm": 1.064016692244757,
      "learning_rate": 0.003,
      "loss": 4.3315,
      "step": 1696
    },
    {
      "epoch": 0.01697,
      "grad_norm": 0.9264587036694925,
      "learning_rate": 0.003,
      "loss": 4.302,
      "step": 1697
    },
    {
      "epoch": 0.01698,
      "grad_norm": 0.6997503897611733,
      "learning_rate": 0.003,
      "loss": 4.289,
      "step": 1698
    },
    {
      "epoch": 0.01699,
      "grad_norm": 0.6394457308265625,
      "learning_rate": 0.003,
      "loss": 4.2969,
      "step": 1699
    },
    {
      "epoch": 0.017,
      "grad_norm": 0.5891910529153762,
      "learning_rate": 0.003,
      "loss": 4.3094,
      "step": 1700
    },
    {
      "epoch": 0.01701,
      "grad_norm": 0.6435550929707158,
      "learning_rate": 0.003,
      "loss": 4.2939,
      "step": 1701
    },
    {
      "epoch": 0.01702,
      "grad_norm": 0.6810306253220882,
      "learning_rate": 0.003,
      "loss": 4.3163,
      "step": 1702
    },
    {
      "epoch": 0.01703,
      "grad_norm": 0.6849363850272787,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1703
    },
    {
      "epoch": 0.01704,
      "grad_norm": 0.6708561417744424,
      "learning_rate": 0.003,
      "loss": 4.2869,
      "step": 1704
    },
    {
      "epoch": 0.01705,
      "grad_norm": 0.5968699215981799,
      "learning_rate": 0.003,
      "loss": 4.2879,
      "step": 1705
    },
    {
      "epoch": 0.01706,
      "grad_norm": 0.505702133399309,
      "learning_rate": 0.003,
      "loss": 4.2792,
      "step": 1706
    },
    {
      "epoch": 0.01707,
      "grad_norm": 0.5302615087387699,
      "learning_rate": 0.003,
      "loss": 4.3011,
      "step": 1707
    },
    {
      "epoch": 0.01708,
      "grad_norm": 0.5210003345198112,
      "learning_rate": 0.003,
      "loss": 4.2762,
      "step": 1708
    },
    {
      "epoch": 0.01709,
      "grad_norm": 0.46584995061994,
      "learning_rate": 0.003,
      "loss": 4.2953,
      "step": 1709
    },
    {
      "epoch": 0.0171,
      "grad_norm": 0.42345394468243425,
      "learning_rate": 0.003,
      "loss": 4.2414,
      "step": 1710
    },
    {
      "epoch": 0.01711,
      "grad_norm": 0.43754438467909607,
      "learning_rate": 0.003,
      "loss": 4.2717,
      "step": 1711
    },
    {
      "epoch": 0.01712,
      "grad_norm": 0.41267034274673536,
      "learning_rate": 0.003,
      "loss": 4.2801,
      "step": 1712
    },
    {
      "epoch": 0.01713,
      "grad_norm": 0.3568885732041511,
      "learning_rate": 0.003,
      "loss": 4.2523,
      "step": 1713
    },
    {
      "epoch": 0.01714,
      "grad_norm": 0.3457637645495226,
      "learning_rate": 0.003,
      "loss": 4.2876,
      "step": 1714
    },
    {
      "epoch": 0.01715,
      "grad_norm": 0.3722023563902025,
      "learning_rate": 0.003,
      "loss": 4.2958,
      "step": 1715
    },
    {
      "epoch": 0.01716,
      "grad_norm": 0.3390050605620623,
      "learning_rate": 0.003,
      "loss": 4.2697,
      "step": 1716
    },
    {
      "epoch": 0.01717,
      "grad_norm": 0.4093688383515523,
      "learning_rate": 0.003,
      "loss": 4.3028,
      "step": 1717
    },
    {
      "epoch": 0.01718,
      "grad_norm": 0.5445666194239754,
      "learning_rate": 0.003,
      "loss": 4.2735,
      "step": 1718
    },
    {
      "epoch": 0.01719,
      "grad_norm": 0.8818554274614839,
      "learning_rate": 0.003,
      "loss": 4.3037,
      "step": 1719
    },
    {
      "epoch": 0.0172,
      "grad_norm": 1.123998698583038,
      "learning_rate": 0.003,
      "loss": 4.2879,
      "step": 1720
    },
    {
      "epoch": 0.01721,
      "grad_norm": 0.6518188394707584,
      "learning_rate": 0.003,
      "loss": 4.2815,
      "step": 1721
    },
    {
      "epoch": 0.01722,
      "grad_norm": 0.7473961890284897,
      "learning_rate": 0.003,
      "loss": 4.281,
      "step": 1722
    },
    {
      "epoch": 0.01723,
      "grad_norm": 0.7241082541176558,
      "learning_rate": 0.003,
      "loss": 4.2942,
      "step": 1723
    },
    {
      "epoch": 0.01724,
      "grad_norm": 0.5406243852034243,
      "learning_rate": 0.003,
      "loss": 4.2839,
      "step": 1724
    },
    {
      "epoch": 0.01725,
      "grad_norm": 0.7661047717663526,
      "learning_rate": 0.003,
      "loss": 4.2758,
      "step": 1725
    },
    {
      "epoch": 0.01726,
      "grad_norm": 0.6519733535814134,
      "learning_rate": 0.003,
      "loss": 4.2678,
      "step": 1726
    },
    {
      "epoch": 0.01727,
      "grad_norm": 0.5095447102684948,
      "learning_rate": 0.003,
      "loss": 4.2681,
      "step": 1727
    },
    {
      "epoch": 0.01728,
      "grad_norm": 0.5716779713985857,
      "learning_rate": 0.003,
      "loss": 4.2703,
      "step": 1728
    },
    {
      "epoch": 0.01729,
      "grad_norm": 0.5506488662519434,
      "learning_rate": 0.003,
      "loss": 4.2777,
      "step": 1729
    },
    {
      "epoch": 0.0173,
      "grad_norm": 0.4864830783539529,
      "learning_rate": 0.003,
      "loss": 4.2498,
      "step": 1730
    },
    {
      "epoch": 0.01731,
      "grad_norm": 0.49289174915320866,
      "learning_rate": 0.003,
      "loss": 4.2691,
      "step": 1731
    },
    {
      "epoch": 0.01732,
      "grad_norm": 0.5258408981010323,
      "learning_rate": 0.003,
      "loss": 4.2782,
      "step": 1732
    },
    {
      "epoch": 0.01733,
      "grad_norm": 0.5689089752248919,
      "learning_rate": 0.003,
      "loss": 4.2578,
      "step": 1733
    },
    {
      "epoch": 0.01734,
      "grad_norm": 0.5515499021599949,
      "learning_rate": 0.003,
      "loss": 4.2704,
      "step": 1734
    },
    {
      "epoch": 0.01735,
      "grad_norm": 0.5101191368780025,
      "learning_rate": 0.003,
      "loss": 4.2394,
      "step": 1735
    },
    {
      "epoch": 0.01736,
      "grad_norm": 0.5538259418110674,
      "learning_rate": 0.003,
      "loss": 4.2405,
      "step": 1736
    },
    {
      "epoch": 0.01737,
      "grad_norm": 0.6368154932756357,
      "learning_rate": 0.003,
      "loss": 4.273,
      "step": 1737
    },
    {
      "epoch": 0.01738,
      "grad_norm": 0.6302527957263243,
      "learning_rate": 0.003,
      "loss": 4.2633,
      "step": 1738
    },
    {
      "epoch": 0.01739,
      "grad_norm": 0.5659864529447296,
      "learning_rate": 0.003,
      "loss": 4.2757,
      "step": 1739
    },
    {
      "epoch": 0.0174,
      "grad_norm": 0.6034857544005698,
      "learning_rate": 0.003,
      "loss": 4.278,
      "step": 1740
    },
    {
      "epoch": 0.01741,
      "grad_norm": 0.7023281656557057,
      "learning_rate": 0.003,
      "loss": 4.2883,
      "step": 1741
    },
    {
      "epoch": 0.01742,
      "grad_norm": 0.7127714842932543,
      "learning_rate": 0.003,
      "loss": 4.2766,
      "step": 1742
    },
    {
      "epoch": 0.01743,
      "grad_norm": 0.7178940660267726,
      "learning_rate": 0.003,
      "loss": 4.2889,
      "step": 1743
    },
    {
      "epoch": 0.01744,
      "grad_norm": 0.770440209657255,
      "learning_rate": 0.003,
      "loss": 4.2773,
      "step": 1744
    },
    {
      "epoch": 0.01745,
      "grad_norm": 0.7362175602563142,
      "learning_rate": 0.003,
      "loss": 4.2484,
      "step": 1745
    },
    {
      "epoch": 0.01746,
      "grad_norm": 0.7588375597390518,
      "learning_rate": 0.003,
      "loss": 4.2719,
      "step": 1746
    },
    {
      "epoch": 0.01747,
      "grad_norm": 0.6839956521205998,
      "learning_rate": 0.003,
      "loss": 4.267,
      "step": 1747
    },
    {
      "epoch": 0.01748,
      "grad_norm": 0.6088836411369196,
      "learning_rate": 0.003,
      "loss": 4.2555,
      "step": 1748
    },
    {
      "epoch": 0.01749,
      "grad_norm": 0.548971540706334,
      "learning_rate": 0.003,
      "loss": 4.2516,
      "step": 1749
    },
    {
      "epoch": 0.0175,
      "grad_norm": 0.5440778124239496,
      "learning_rate": 0.003,
      "loss": 4.2665,
      "step": 1750
    },
    {
      "epoch": 0.01751,
      "grad_norm": 0.5918941393242653,
      "learning_rate": 0.003,
      "loss": 4.2593,
      "step": 1751
    },
    {
      "epoch": 0.01752,
      "grad_norm": 0.6455620934648258,
      "learning_rate": 0.003,
      "loss": 4.3014,
      "step": 1752
    },
    {
      "epoch": 0.01753,
      "grad_norm": 0.6781800073296501,
      "learning_rate": 0.003,
      "loss": 4.2535,
      "step": 1753
    },
    {
      "epoch": 0.01754,
      "grad_norm": 0.6882980076107725,
      "learning_rate": 0.003,
      "loss": 4.2853,
      "step": 1754
    },
    {
      "epoch": 0.01755,
      "grad_norm": 0.7961632146248975,
      "learning_rate": 0.003,
      "loss": 4.233,
      "step": 1755
    },
    {
      "epoch": 0.01756,
      "grad_norm": 0.8494780429534167,
      "learning_rate": 0.003,
      "loss": 4.314,
      "step": 1756
    },
    {
      "epoch": 0.01757,
      "grad_norm": 0.9294824362906122,
      "learning_rate": 0.003,
      "loss": 4.2987,
      "step": 1757
    },
    {
      "epoch": 0.01758,
      "grad_norm": 0.852175947314296,
      "learning_rate": 0.003,
      "loss": 4.3126,
      "step": 1758
    },
    {
      "epoch": 0.01759,
      "grad_norm": 0.7347749740307682,
      "learning_rate": 0.003,
      "loss": 4.2719,
      "step": 1759
    },
    {
      "epoch": 0.0176,
      "grad_norm": 0.832142747460363,
      "learning_rate": 0.003,
      "loss": 4.2994,
      "step": 1760
    },
    {
      "epoch": 0.01761,
      "grad_norm": 0.8686282951443787,
      "learning_rate": 0.003,
      "loss": 4.2965,
      "step": 1761
    },
    {
      "epoch": 0.01762,
      "grad_norm": 0.9408046252281581,
      "learning_rate": 0.003,
      "loss": 4.3172,
      "step": 1762
    },
    {
      "epoch": 0.01763,
      "grad_norm": 0.8594428065536988,
      "learning_rate": 0.003,
      "loss": 4.3092,
      "step": 1763
    },
    {
      "epoch": 0.01764,
      "grad_norm": 0.7749341487974934,
      "learning_rate": 0.003,
      "loss": 4.2754,
      "step": 1764
    },
    {
      "epoch": 0.01765,
      "grad_norm": 0.7897730382355919,
      "learning_rate": 0.003,
      "loss": 4.3076,
      "step": 1765
    },
    {
      "epoch": 0.01766,
      "grad_norm": 0.7174878978412247,
      "learning_rate": 0.003,
      "loss": 4.317,
      "step": 1766
    },
    {
      "epoch": 0.01767,
      "grad_norm": 0.7171956437495152,
      "learning_rate": 0.003,
      "loss": 4.2757,
      "step": 1767
    },
    {
      "epoch": 0.01768,
      "grad_norm": 0.8077487259776668,
      "learning_rate": 0.003,
      "loss": 4.2866,
      "step": 1768
    },
    {
      "epoch": 0.01769,
      "grad_norm": 0.7100605985326003,
      "learning_rate": 0.003,
      "loss": 4.2811,
      "step": 1769
    },
    {
      "epoch": 0.0177,
      "grad_norm": 0.578964917508603,
      "learning_rate": 0.003,
      "loss": 4.2635,
      "step": 1770
    },
    {
      "epoch": 0.01771,
      "grad_norm": 0.6086271828330341,
      "learning_rate": 0.003,
      "loss": 4.2929,
      "step": 1771
    },
    {
      "epoch": 0.01772,
      "grad_norm": 0.497546957954685,
      "learning_rate": 0.003,
      "loss": 4.262,
      "step": 1772
    },
    {
      "epoch": 0.01773,
      "grad_norm": 0.490787842661778,
      "learning_rate": 0.003,
      "loss": 4.2539,
      "step": 1773
    },
    {
      "epoch": 0.01774,
      "grad_norm": 0.40528808868462435,
      "learning_rate": 0.003,
      "loss": 4.272,
      "step": 1774
    },
    {
      "epoch": 0.01775,
      "grad_norm": 0.3996997018074339,
      "learning_rate": 0.003,
      "loss": 4.2574,
      "step": 1775
    },
    {
      "epoch": 0.01776,
      "grad_norm": 0.40093115740510726,
      "learning_rate": 0.003,
      "loss": 4.2491,
      "step": 1776
    },
    {
      "epoch": 0.01777,
      "grad_norm": 0.4577551703461772,
      "learning_rate": 0.003,
      "loss": 4.27,
      "step": 1777
    },
    {
      "epoch": 0.01778,
      "grad_norm": 0.4894775985846586,
      "learning_rate": 0.003,
      "loss": 4.262,
      "step": 1778
    },
    {
      "epoch": 0.01779,
      "grad_norm": 0.619715296261353,
      "learning_rate": 0.003,
      "loss": 4.2821,
      "step": 1779
    },
    {
      "epoch": 0.0178,
      "grad_norm": 0.8092821046637155,
      "learning_rate": 0.003,
      "loss": 4.2576,
      "step": 1780
    },
    {
      "epoch": 0.01781,
      "grad_norm": 1.007368746628607,
      "learning_rate": 0.003,
      "loss": 4.2833,
      "step": 1781
    },
    {
      "epoch": 0.01782,
      "grad_norm": 1.0605028910459124,
      "learning_rate": 0.003,
      "loss": 4.314,
      "step": 1782
    },
    {
      "epoch": 0.01783,
      "grad_norm": 0.8361916169047299,
      "learning_rate": 0.003,
      "loss": 4.2912,
      "step": 1783
    },
    {
      "epoch": 0.01784,
      "grad_norm": 0.9950086095977337,
      "learning_rate": 0.003,
      "loss": 4.2963,
      "step": 1784
    },
    {
      "epoch": 0.01785,
      "grad_norm": 1.0228877834258285,
      "learning_rate": 0.003,
      "loss": 4.2673,
      "step": 1785
    },
    {
      "epoch": 0.01786,
      "grad_norm": 0.8766088407747356,
      "learning_rate": 0.003,
      "loss": 4.2899,
      "step": 1786
    },
    {
      "epoch": 0.01787,
      "grad_norm": 0.9346024274483619,
      "learning_rate": 0.003,
      "loss": 4.2991,
      "step": 1787
    },
    {
      "epoch": 0.01788,
      "grad_norm": 0.8655394124739175,
      "learning_rate": 0.003,
      "loss": 4.3097,
      "step": 1788
    },
    {
      "epoch": 0.01789,
      "grad_norm": 0.7320722132470949,
      "learning_rate": 0.003,
      "loss": 4.2742,
      "step": 1789
    },
    {
      "epoch": 0.0179,
      "grad_norm": 0.6043898276387406,
      "learning_rate": 0.003,
      "loss": 4.288,
      "step": 1790
    },
    {
      "epoch": 0.01791,
      "grad_norm": 0.551731430114501,
      "learning_rate": 0.003,
      "loss": 4.2984,
      "step": 1791
    },
    {
      "epoch": 0.01792,
      "grad_norm": 0.4116411374695503,
      "learning_rate": 0.003,
      "loss": 4.2944,
      "step": 1792
    },
    {
      "epoch": 0.01793,
      "grad_norm": 0.4368353784765224,
      "learning_rate": 0.003,
      "loss": 4.2574,
      "step": 1793
    },
    {
      "epoch": 0.01794,
      "grad_norm": 0.4134683249639472,
      "learning_rate": 0.003,
      "loss": 4.2696,
      "step": 1794
    },
    {
      "epoch": 0.01795,
      "grad_norm": 0.3898886643379656,
      "learning_rate": 0.003,
      "loss": 4.2661,
      "step": 1795
    },
    {
      "epoch": 0.01796,
      "grad_norm": 0.3690232703108766,
      "learning_rate": 0.003,
      "loss": 4.2797,
      "step": 1796
    },
    {
      "epoch": 0.01797,
      "grad_norm": 0.3352292075992397,
      "learning_rate": 0.003,
      "loss": 4.2421,
      "step": 1797
    },
    {
      "epoch": 0.01798,
      "grad_norm": 0.3577790623133374,
      "learning_rate": 0.003,
      "loss": 4.2547,
      "step": 1798
    },
    {
      "epoch": 0.01799,
      "grad_norm": 0.40125449243674877,
      "learning_rate": 0.003,
      "loss": 4.269,
      "step": 1799
    },
    {
      "epoch": 0.018,
      "grad_norm": 0.4677938601833102,
      "learning_rate": 0.003,
      "loss": 4.26,
      "step": 1800
    },
    {
      "epoch": 0.01801,
      "grad_norm": 0.5610552208097703,
      "learning_rate": 0.003,
      "loss": 4.2554,
      "step": 1801
    },
    {
      "epoch": 0.01802,
      "grad_norm": 0.5864236478232465,
      "learning_rate": 0.003,
      "loss": 4.2736,
      "step": 1802
    },
    {
      "epoch": 0.01803,
      "grad_norm": 0.540429672022444,
      "learning_rate": 0.003,
      "loss": 4.2776,
      "step": 1803
    },
    {
      "epoch": 0.01804,
      "grad_norm": 0.60459180760597,
      "learning_rate": 0.003,
      "loss": 4.2503,
      "step": 1804
    },
    {
      "epoch": 0.01805,
      "grad_norm": 0.6945199909530504,
      "learning_rate": 0.003,
      "loss": 4.2727,
      "step": 1805
    },
    {
      "epoch": 0.01806,
      "grad_norm": 0.6888751318415884,
      "learning_rate": 0.003,
      "loss": 4.2338,
      "step": 1806
    },
    {
      "epoch": 0.01807,
      "grad_norm": 0.652641222784165,
      "learning_rate": 0.003,
      "loss": 4.2674,
      "step": 1807
    },
    {
      "epoch": 0.01808,
      "grad_norm": 0.7163920878946057,
      "learning_rate": 0.003,
      "loss": 4.2613,
      "step": 1808
    },
    {
      "epoch": 0.01809,
      "grad_norm": 0.7231527059402414,
      "learning_rate": 0.003,
      "loss": 4.284,
      "step": 1809
    },
    {
      "epoch": 0.0181,
      "grad_norm": 0.7979758654578836,
      "learning_rate": 0.003,
      "loss": 4.2839,
      "step": 1810
    },
    {
      "epoch": 0.01811,
      "grad_norm": 0.7630566976684018,
      "learning_rate": 0.003,
      "loss": 4.2692,
      "step": 1811
    },
    {
      "epoch": 0.01812,
      "grad_norm": 0.68601985737994,
      "learning_rate": 0.003,
      "loss": 4.2545,
      "step": 1812
    },
    {
      "epoch": 0.01813,
      "grad_norm": 0.601926611394952,
      "learning_rate": 0.003,
      "loss": 4.2949,
      "step": 1813
    },
    {
      "epoch": 0.01814,
      "grad_norm": 0.66852657571521,
      "learning_rate": 0.003,
      "loss": 4.2647,
      "step": 1814
    },
    {
      "epoch": 0.01815,
      "grad_norm": 0.6050905896281803,
      "learning_rate": 0.003,
      "loss": 4.2474,
      "step": 1815
    },
    {
      "epoch": 0.01816,
      "grad_norm": 0.531953848915138,
      "learning_rate": 0.003,
      "loss": 4.2731,
      "step": 1816
    },
    {
      "epoch": 0.01817,
      "grad_norm": 0.6652005043351588,
      "learning_rate": 0.003,
      "loss": 4.2676,
      "step": 1817
    },
    {
      "epoch": 0.01818,
      "grad_norm": 0.6653492631522442,
      "learning_rate": 0.003,
      "loss": 4.2669,
      "step": 1818
    },
    {
      "epoch": 0.01819,
      "grad_norm": 0.6430139858317634,
      "learning_rate": 0.003,
      "loss": 4.2522,
      "step": 1819
    },
    {
      "epoch": 0.0182,
      "grad_norm": 0.7307512465670702,
      "learning_rate": 0.003,
      "loss": 4.2731,
      "step": 1820
    },
    {
      "epoch": 0.01821,
      "grad_norm": 0.7117919312146421,
      "learning_rate": 0.003,
      "loss": 4.2608,
      "step": 1821
    },
    {
      "epoch": 0.01822,
      "grad_norm": 0.6849992334976018,
      "learning_rate": 0.003,
      "loss": 4.2737,
      "step": 1822
    },
    {
      "epoch": 0.01823,
      "grad_norm": 0.6437329965546797,
      "learning_rate": 0.003,
      "loss": 4.2861,
      "step": 1823
    },
    {
      "epoch": 0.01824,
      "grad_norm": 0.6298055587951418,
      "learning_rate": 0.003,
      "loss": 4.2663,
      "step": 1824
    },
    {
      "epoch": 0.01825,
      "grad_norm": 0.6105471946808345,
      "learning_rate": 0.003,
      "loss": 4.2706,
      "step": 1825
    },
    {
      "epoch": 0.01826,
      "grad_norm": 0.5680736127986394,
      "learning_rate": 0.003,
      "loss": 4.2798,
      "step": 1826
    },
    {
      "epoch": 0.01827,
      "grad_norm": 0.6203032072106797,
      "learning_rate": 0.003,
      "loss": 4.2493,
      "step": 1827
    },
    {
      "epoch": 0.01828,
      "grad_norm": 0.5646483418870301,
      "learning_rate": 0.003,
      "loss": 4.2595,
      "step": 1828
    },
    {
      "epoch": 0.01829,
      "grad_norm": 0.43449345506164244,
      "learning_rate": 0.003,
      "loss": 4.2724,
      "step": 1829
    },
    {
      "epoch": 0.0183,
      "grad_norm": 0.4995474978675237,
      "learning_rate": 0.003,
      "loss": 4.268,
      "step": 1830
    },
    {
      "epoch": 0.01831,
      "grad_norm": 0.6731917230470822,
      "learning_rate": 0.003,
      "loss": 4.2625,
      "step": 1831
    },
    {
      "epoch": 0.01832,
      "grad_norm": 1.0735909763241713,
      "learning_rate": 0.003,
      "loss": 4.2749,
      "step": 1832
    },
    {
      "epoch": 0.01833,
      "grad_norm": 0.9454987640162065,
      "learning_rate": 0.003,
      "loss": 4.3126,
      "step": 1833
    },
    {
      "epoch": 0.01834,
      "grad_norm": 0.6279857981331142,
      "learning_rate": 0.003,
      "loss": 4.265,
      "step": 1834
    },
    {
      "epoch": 0.01835,
      "grad_norm": 0.7795333868843384,
      "learning_rate": 0.003,
      "loss": 4.2708,
      "step": 1835
    },
    {
      "epoch": 0.01836,
      "grad_norm": 0.7540135177559686,
      "learning_rate": 0.003,
      "loss": 4.269,
      "step": 1836
    },
    {
      "epoch": 0.01837,
      "grad_norm": 0.4873694552023737,
      "learning_rate": 0.003,
      "loss": 4.2594,
      "step": 1837
    },
    {
      "epoch": 0.01838,
      "grad_norm": 0.6475671331829288,
      "learning_rate": 0.003,
      "loss": 4.2679,
      "step": 1838
    },
    {
      "epoch": 0.01839,
      "grad_norm": 0.5413345448794652,
      "learning_rate": 0.003,
      "loss": 4.2849,
      "step": 1839
    },
    {
      "epoch": 0.0184,
      "grad_norm": 0.505876271981352,
      "learning_rate": 0.003,
      "loss": 4.2815,
      "step": 1840
    },
    {
      "epoch": 0.01841,
      "grad_norm": 0.5559518639320817,
      "learning_rate": 0.003,
      "loss": 4.2211,
      "step": 1841
    },
    {
      "epoch": 0.01842,
      "grad_norm": 0.565402658078461,
      "learning_rate": 0.003,
      "loss": 4.2563,
      "step": 1842
    },
    {
      "epoch": 0.01843,
      "grad_norm": 0.5176031889881334,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 1843
    },
    {
      "epoch": 0.01844,
      "grad_norm": 0.515476660965344,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 1844
    },
    {
      "epoch": 0.01845,
      "grad_norm": 0.537738729997812,
      "learning_rate": 0.003,
      "loss": 4.2366,
      "step": 1845
    },
    {
      "epoch": 0.01846,
      "grad_norm": 0.5825872331164712,
      "learning_rate": 0.003,
      "loss": 4.2629,
      "step": 1846
    },
    {
      "epoch": 0.01847,
      "grad_norm": 0.6172426924095343,
      "learning_rate": 0.003,
      "loss": 4.261,
      "step": 1847
    },
    {
      "epoch": 0.01848,
      "grad_norm": 0.5675520394460739,
      "learning_rate": 0.003,
      "loss": 4.2733,
      "step": 1848
    },
    {
      "epoch": 0.01849,
      "grad_norm": 0.5804129655862207,
      "learning_rate": 0.003,
      "loss": 4.2733,
      "step": 1849
    },
    {
      "epoch": 0.0185,
      "grad_norm": 0.5903509862738283,
      "learning_rate": 0.003,
      "loss": 4.2637,
      "step": 1850
    },
    {
      "epoch": 0.01851,
      "grad_norm": 0.5789415751735032,
      "learning_rate": 0.003,
      "loss": 4.2453,
      "step": 1851
    },
    {
      "epoch": 0.01852,
      "grad_norm": 0.6073114122094332,
      "learning_rate": 0.003,
      "loss": 4.249,
      "step": 1852
    },
    {
      "epoch": 0.01853,
      "grad_norm": 0.6172059367670675,
      "learning_rate": 0.003,
      "loss": 4.2514,
      "step": 1853
    },
    {
      "epoch": 0.01854,
      "grad_norm": 0.5952391168819953,
      "learning_rate": 0.003,
      "loss": 4.282,
      "step": 1854
    },
    {
      "epoch": 0.01855,
      "grad_norm": 0.599929499071102,
      "learning_rate": 0.003,
      "loss": 4.2645,
      "step": 1855
    },
    {
      "epoch": 0.01856,
      "grad_norm": 0.7854590983227943,
      "learning_rate": 0.003,
      "loss": 4.2582,
      "step": 1856
    },
    {
      "epoch": 0.01857,
      "grad_norm": 0.9479235992175757,
      "learning_rate": 0.003,
      "loss": 4.2521,
      "step": 1857
    },
    {
      "epoch": 0.01858,
      "grad_norm": 1.0547185212118042,
      "learning_rate": 0.003,
      "loss": 4.2782,
      "step": 1858
    },
    {
      "epoch": 0.01859,
      "grad_norm": 0.7498301751007965,
      "learning_rate": 0.003,
      "loss": 4.259,
      "step": 1859
    },
    {
      "epoch": 0.0186,
      "grad_norm": 0.6772314759733408,
      "learning_rate": 0.003,
      "loss": 4.238,
      "step": 1860
    },
    {
      "epoch": 0.01861,
      "grad_norm": 0.7554457171283483,
      "learning_rate": 0.003,
      "loss": 4.2858,
      "step": 1861
    },
    {
      "epoch": 0.01862,
      "grad_norm": 0.6601398192741809,
      "learning_rate": 0.003,
      "loss": 4.2482,
      "step": 1862
    },
    {
      "epoch": 0.01863,
      "grad_norm": 0.5632517970928785,
      "learning_rate": 0.003,
      "loss": 4.2548,
      "step": 1863
    },
    {
      "epoch": 0.01864,
      "grad_norm": 0.5518520472207163,
      "learning_rate": 0.003,
      "loss": 4.2848,
      "step": 1864
    },
    {
      "epoch": 0.01865,
      "grad_norm": 0.5191605026436766,
      "learning_rate": 0.003,
      "loss": 4.2571,
      "step": 1865
    },
    {
      "epoch": 0.01866,
      "grad_norm": 0.487977336215745,
      "learning_rate": 0.003,
      "loss": 4.2583,
      "step": 1866
    },
    {
      "epoch": 0.01867,
      "grad_norm": 0.38111750510593084,
      "learning_rate": 0.003,
      "loss": 4.2288,
      "step": 1867
    },
    {
      "epoch": 0.01868,
      "grad_norm": 0.36979378128361096,
      "learning_rate": 0.003,
      "loss": 4.2458,
      "step": 1868
    },
    {
      "epoch": 0.01869,
      "grad_norm": 0.4056719991155462,
      "learning_rate": 0.003,
      "loss": 4.259,
      "step": 1869
    },
    {
      "epoch": 0.0187,
      "grad_norm": 0.511550197815793,
      "learning_rate": 0.003,
      "loss": 4.2534,
      "step": 1870
    },
    {
      "epoch": 0.01871,
      "grad_norm": 0.6908775527772675,
      "learning_rate": 0.003,
      "loss": 4.2322,
      "step": 1871
    },
    {
      "epoch": 0.01872,
      "grad_norm": 0.8655201254185445,
      "learning_rate": 0.003,
      "loss": 4.2892,
      "step": 1872
    },
    {
      "epoch": 0.01873,
      "grad_norm": 0.903028038452321,
      "learning_rate": 0.003,
      "loss": 4.2527,
      "step": 1873
    },
    {
      "epoch": 0.01874,
      "grad_norm": 0.7655189256325052,
      "learning_rate": 0.003,
      "loss": 4.2534,
      "step": 1874
    },
    {
      "epoch": 0.01875,
      "grad_norm": 0.6845228657861865,
      "learning_rate": 0.003,
      "loss": 4.2697,
      "step": 1875
    },
    {
      "epoch": 0.01876,
      "grad_norm": 0.7967028641532641,
      "learning_rate": 0.003,
      "loss": 4.2796,
      "step": 1876
    },
    {
      "epoch": 0.01877,
      "grad_norm": 0.7276633971950639,
      "learning_rate": 0.003,
      "loss": 4.2792,
      "step": 1877
    },
    {
      "epoch": 0.01878,
      "grad_norm": 0.6745755254962531,
      "learning_rate": 0.003,
      "loss": 4.2647,
      "step": 1878
    },
    {
      "epoch": 0.01879,
      "grad_norm": 0.6794948462625433,
      "learning_rate": 0.003,
      "loss": 4.2421,
      "step": 1879
    },
    {
      "epoch": 0.0188,
      "grad_norm": 0.7496570961239313,
      "learning_rate": 0.003,
      "loss": 4.2581,
      "step": 1880
    },
    {
      "epoch": 0.01881,
      "grad_norm": 0.7216177452798419,
      "learning_rate": 0.003,
      "loss": 4.2527,
      "step": 1881
    },
    {
      "epoch": 0.01882,
      "grad_norm": 0.6411166720639763,
      "learning_rate": 0.003,
      "loss": 4.2764,
      "step": 1882
    },
    {
      "epoch": 0.01883,
      "grad_norm": 0.4976595430995637,
      "learning_rate": 0.003,
      "loss": 4.2463,
      "step": 1883
    },
    {
      "epoch": 0.01884,
      "grad_norm": 0.5806395175516162,
      "learning_rate": 0.003,
      "loss": 4.2463,
      "step": 1884
    },
    {
      "epoch": 0.01885,
      "grad_norm": 0.5932999040186268,
      "learning_rate": 0.003,
      "loss": 4.2267,
      "step": 1885
    },
    {
      "epoch": 0.01886,
      "grad_norm": 0.6243294916603833,
      "learning_rate": 0.003,
      "loss": 4.2637,
      "step": 1886
    },
    {
      "epoch": 0.01887,
      "grad_norm": 0.6594473434297762,
      "learning_rate": 0.003,
      "loss": 4.2583,
      "step": 1887
    },
    {
      "epoch": 0.01888,
      "grad_norm": 0.5833874852548948,
      "learning_rate": 0.003,
      "loss": 4.2402,
      "step": 1888
    },
    {
      "epoch": 0.01889,
      "grad_norm": 0.5554783687186964,
      "learning_rate": 0.003,
      "loss": 4.2547,
      "step": 1889
    },
    {
      "epoch": 0.0189,
      "grad_norm": 0.5267765258339018,
      "learning_rate": 0.003,
      "loss": 4.2329,
      "step": 1890
    },
    {
      "epoch": 0.01891,
      "grad_norm": 0.5533738473996538,
      "learning_rate": 0.003,
      "loss": 4.2819,
      "step": 1891
    },
    {
      "epoch": 0.01892,
      "grad_norm": 0.5233002886731548,
      "learning_rate": 0.003,
      "loss": 4.2568,
      "step": 1892
    },
    {
      "epoch": 0.01893,
      "grad_norm": 0.598656523235247,
      "learning_rate": 0.003,
      "loss": 4.2415,
      "step": 1893
    },
    {
      "epoch": 0.01894,
      "grad_norm": 0.6329032192123055,
      "learning_rate": 0.003,
      "loss": 4.2498,
      "step": 1894
    },
    {
      "epoch": 0.01895,
      "grad_norm": 0.7448049826558554,
      "learning_rate": 0.003,
      "loss": 4.2404,
      "step": 1895
    },
    {
      "epoch": 0.01896,
      "grad_norm": 0.7896808634621466,
      "learning_rate": 0.003,
      "loss": 4.2666,
      "step": 1896
    },
    {
      "epoch": 0.01897,
      "grad_norm": 0.8670691506268475,
      "learning_rate": 0.003,
      "loss": 4.267,
      "step": 1897
    },
    {
      "epoch": 0.01898,
      "grad_norm": 0.7564766735493696,
      "learning_rate": 0.003,
      "loss": 4.2371,
      "step": 1898
    },
    {
      "epoch": 0.01899,
      "grad_norm": 0.6014591021089313,
      "learning_rate": 0.003,
      "loss": 4.246,
      "step": 1899
    },
    {
      "epoch": 0.019,
      "grad_norm": 0.6487238549786433,
      "learning_rate": 0.003,
      "loss": 4.2427,
      "step": 1900
    },
    {
      "epoch": 0.01901,
      "grad_norm": 0.7335622998657767,
      "learning_rate": 0.003,
      "loss": 4.2939,
      "step": 1901
    },
    {
      "epoch": 0.01902,
      "grad_norm": 0.727489215463111,
      "learning_rate": 0.003,
      "loss": 4.2706,
      "step": 1902
    },
    {
      "epoch": 0.01903,
      "grad_norm": 0.6444511274857245,
      "learning_rate": 0.003,
      "loss": 4.2217,
      "step": 1903
    },
    {
      "epoch": 0.01904,
      "grad_norm": 0.5669107570524424,
      "learning_rate": 0.003,
      "loss": 4.2738,
      "step": 1904
    },
    {
      "epoch": 0.01905,
      "grad_norm": 0.47567496733321213,
      "learning_rate": 0.003,
      "loss": 4.2471,
      "step": 1905
    },
    {
      "epoch": 0.01906,
      "grad_norm": 0.568871434515019,
      "learning_rate": 0.003,
      "loss": 4.2665,
      "step": 1906
    },
    {
      "epoch": 0.01907,
      "grad_norm": 0.6437840413516579,
      "learning_rate": 0.003,
      "loss": 4.2605,
      "step": 1907
    },
    {
      "epoch": 0.01908,
      "grad_norm": 0.8102395542085408,
      "learning_rate": 0.003,
      "loss": 4.2806,
      "step": 1908
    },
    {
      "epoch": 0.01909,
      "grad_norm": 0.9019612301377007,
      "learning_rate": 0.003,
      "loss": 4.2522,
      "step": 1909
    },
    {
      "epoch": 0.0191,
      "grad_norm": 0.8060528165222425,
      "learning_rate": 0.003,
      "loss": 4.2682,
      "step": 1910
    },
    {
      "epoch": 0.01911,
      "grad_norm": 0.6514562643306322,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 1911
    },
    {
      "epoch": 0.01912,
      "grad_norm": 0.7512556786074466,
      "learning_rate": 0.003,
      "loss": 4.2481,
      "step": 1912
    },
    {
      "epoch": 0.01913,
      "grad_norm": 0.7206930768558686,
      "learning_rate": 0.003,
      "loss": 4.2688,
      "step": 1913
    },
    {
      "epoch": 0.01914,
      "grad_norm": 0.6802559754696879,
      "learning_rate": 0.003,
      "loss": 4.2815,
      "step": 1914
    },
    {
      "epoch": 0.01915,
      "grad_norm": 0.6329709283705148,
      "learning_rate": 0.003,
      "loss": 4.2674,
      "step": 1915
    },
    {
      "epoch": 0.01916,
      "grad_norm": 0.6539184714683666,
      "learning_rate": 0.003,
      "loss": 4.2462,
      "step": 1916
    },
    {
      "epoch": 0.01917,
      "grad_norm": 0.710899415708157,
      "learning_rate": 0.003,
      "loss": 4.2674,
      "step": 1917
    },
    {
      "epoch": 0.01918,
      "grad_norm": 0.7167946113405885,
      "learning_rate": 0.003,
      "loss": 4.2562,
      "step": 1918
    },
    {
      "epoch": 0.01919,
      "grad_norm": 0.7533071012807427,
      "learning_rate": 0.003,
      "loss": 4.2594,
      "step": 1919
    },
    {
      "epoch": 0.0192,
      "grad_norm": 0.6703396884694851,
      "learning_rate": 0.003,
      "loss": 4.2612,
      "step": 1920
    },
    {
      "epoch": 0.01921,
      "grad_norm": 0.5834644623189628,
      "learning_rate": 0.003,
      "loss": 4.2441,
      "step": 1921
    },
    {
      "epoch": 0.01922,
      "grad_norm": 0.6481277683352221,
      "learning_rate": 0.003,
      "loss": 4.2625,
      "step": 1922
    },
    {
      "epoch": 0.01923,
      "grad_norm": 0.6044219599617415,
      "learning_rate": 0.003,
      "loss": 4.2661,
      "step": 1923
    },
    {
      "epoch": 0.01924,
      "grad_norm": 0.5337323934986105,
      "learning_rate": 0.003,
      "loss": 4.2643,
      "step": 1924
    },
    {
      "epoch": 0.01925,
      "grad_norm": 0.5129044588459396,
      "learning_rate": 0.003,
      "loss": 4.251,
      "step": 1925
    },
    {
      "epoch": 0.01926,
      "grad_norm": 0.48201171188935904,
      "learning_rate": 0.003,
      "loss": 4.2563,
      "step": 1926
    },
    {
      "epoch": 0.01927,
      "grad_norm": 0.4295757083815179,
      "learning_rate": 0.003,
      "loss": 4.2456,
      "step": 1927
    },
    {
      "epoch": 0.01928,
      "grad_norm": 0.5221994128637466,
      "learning_rate": 0.003,
      "loss": 4.2241,
      "step": 1928
    },
    {
      "epoch": 0.01929,
      "grad_norm": 0.6151786259894082,
      "learning_rate": 0.003,
      "loss": 4.246,
      "step": 1929
    },
    {
      "epoch": 0.0193,
      "grad_norm": 0.717185888834206,
      "learning_rate": 0.003,
      "loss": 4.2399,
      "step": 1930
    },
    {
      "epoch": 0.01931,
      "grad_norm": 0.865419932280181,
      "learning_rate": 0.003,
      "loss": 4.2485,
      "step": 1931
    },
    {
      "epoch": 0.01932,
      "grad_norm": 1.015142018105296,
      "learning_rate": 0.003,
      "loss": 4.2679,
      "step": 1932
    },
    {
      "epoch": 0.01933,
      "grad_norm": 1.0909531598947217,
      "learning_rate": 0.003,
      "loss": 4.2582,
      "step": 1933
    },
    {
      "epoch": 0.01934,
      "grad_norm": 0.7995257446738059,
      "learning_rate": 0.003,
      "loss": 4.2554,
      "step": 1934
    },
    {
      "epoch": 0.01935,
      "grad_norm": 0.6434815333532291,
      "learning_rate": 0.003,
      "loss": 4.2669,
      "step": 1935
    },
    {
      "epoch": 0.01936,
      "grad_norm": 0.6569925784458591,
      "learning_rate": 0.003,
      "loss": 4.2489,
      "step": 1936
    },
    {
      "epoch": 0.01937,
      "grad_norm": 0.5996128457789688,
      "learning_rate": 0.003,
      "loss": 4.2577,
      "step": 1937
    },
    {
      "epoch": 0.01938,
      "grad_norm": 0.6329170311784792,
      "learning_rate": 0.003,
      "loss": 4.2615,
      "step": 1938
    },
    {
      "epoch": 0.01939,
      "grad_norm": 0.6354665323417178,
      "learning_rate": 0.003,
      "loss": 4.2808,
      "step": 1939
    },
    {
      "epoch": 0.0194,
      "grad_norm": 0.6008328437037136,
      "learning_rate": 0.003,
      "loss": 4.2581,
      "step": 1940
    },
    {
      "epoch": 0.01941,
      "grad_norm": 0.5343827591468221,
      "learning_rate": 0.003,
      "loss": 4.2378,
      "step": 1941
    },
    {
      "epoch": 0.01942,
      "grad_norm": 0.5512416106954112,
      "learning_rate": 0.003,
      "loss": 4.2611,
      "step": 1942
    },
    {
      "epoch": 0.01943,
      "grad_norm": 0.5866960320410021,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 1943
    },
    {
      "epoch": 0.01944,
      "grad_norm": 0.6240489288808161,
      "learning_rate": 0.003,
      "loss": 4.2751,
      "step": 1944
    },
    {
      "epoch": 0.01945,
      "grad_norm": 0.6243606216006328,
      "learning_rate": 0.003,
      "loss": 4.2437,
      "step": 1945
    },
    {
      "epoch": 0.01946,
      "grad_norm": 0.5880226151273782,
      "learning_rate": 0.003,
      "loss": 4.2379,
      "step": 1946
    },
    {
      "epoch": 0.01947,
      "grad_norm": 0.5934159043950513,
      "learning_rate": 0.003,
      "loss": 4.2471,
      "step": 1947
    },
    {
      "epoch": 0.01948,
      "grad_norm": 0.5378813319639308,
      "learning_rate": 0.003,
      "loss": 4.2659,
      "step": 1948
    },
    {
      "epoch": 0.01949,
      "grad_norm": 0.5777955568924656,
      "learning_rate": 0.003,
      "loss": 4.2595,
      "step": 1949
    },
    {
      "epoch": 0.0195,
      "grad_norm": 0.6588236649280669,
      "learning_rate": 0.003,
      "loss": 4.2493,
      "step": 1950
    },
    {
      "epoch": 0.01951,
      "grad_norm": 0.6662863362190039,
      "learning_rate": 0.003,
      "loss": 4.2457,
      "step": 1951
    },
    {
      "epoch": 0.01952,
      "grad_norm": 0.6956758803553106,
      "learning_rate": 0.003,
      "loss": 4.2563,
      "step": 1952
    },
    {
      "epoch": 0.01953,
      "grad_norm": 0.7309670323591311,
      "learning_rate": 0.003,
      "loss": 4.2663,
      "step": 1953
    },
    {
      "epoch": 0.01954,
      "grad_norm": 0.7102958436956044,
      "learning_rate": 0.003,
      "loss": 4.2656,
      "step": 1954
    },
    {
      "epoch": 0.01955,
      "grad_norm": 0.748927101392194,
      "learning_rate": 0.003,
      "loss": 4.2596,
      "step": 1955
    },
    {
      "epoch": 0.01956,
      "grad_norm": 0.6357959051502906,
      "learning_rate": 0.003,
      "loss": 4.2507,
      "step": 1956
    },
    {
      "epoch": 0.01957,
      "grad_norm": 0.5776415679220375,
      "learning_rate": 0.003,
      "loss": 4.2172,
      "step": 1957
    },
    {
      "epoch": 0.01958,
      "grad_norm": 0.5010308008271186,
      "learning_rate": 0.003,
      "loss": 4.2108,
      "step": 1958
    },
    {
      "epoch": 0.01959,
      "grad_norm": 0.500419502823399,
      "learning_rate": 0.003,
      "loss": 4.2342,
      "step": 1959
    },
    {
      "epoch": 0.0196,
      "grad_norm": 0.5017850552651191,
      "learning_rate": 0.003,
      "loss": 4.229,
      "step": 1960
    },
    {
      "epoch": 0.01961,
      "grad_norm": 0.49043553705552145,
      "learning_rate": 0.003,
      "loss": 4.2267,
      "step": 1961
    },
    {
      "epoch": 0.01962,
      "grad_norm": 0.49806058358446914,
      "learning_rate": 0.003,
      "loss": 4.2301,
      "step": 1962
    },
    {
      "epoch": 0.01963,
      "grad_norm": 0.5386284819858854,
      "learning_rate": 0.003,
      "loss": 4.2201,
      "step": 1963
    },
    {
      "epoch": 0.01964,
      "grad_norm": 0.6462166130085853,
      "learning_rate": 0.003,
      "loss": 4.2492,
      "step": 1964
    },
    {
      "epoch": 0.01965,
      "grad_norm": 0.8617584950578717,
      "learning_rate": 0.003,
      "loss": 4.2512,
      "step": 1965
    },
    {
      "epoch": 0.01966,
      "grad_norm": 1.064420589029421,
      "learning_rate": 0.003,
      "loss": 4.2624,
      "step": 1966
    },
    {
      "epoch": 0.01967,
      "grad_norm": 0.8064456905773513,
      "learning_rate": 0.003,
      "loss": 4.2378,
      "step": 1967
    },
    {
      "epoch": 0.01968,
      "grad_norm": 0.6120027249682956,
      "learning_rate": 0.003,
      "loss": 4.2516,
      "step": 1968
    },
    {
      "epoch": 0.01969,
      "grad_norm": 0.7508122971760963,
      "learning_rate": 0.003,
      "loss": 4.2526,
      "step": 1969
    },
    {
      "epoch": 0.0197,
      "grad_norm": 0.7405449826366749,
      "learning_rate": 0.003,
      "loss": 4.2514,
      "step": 1970
    },
    {
      "epoch": 0.01971,
      "grad_norm": 0.7056278637839583,
      "learning_rate": 0.003,
      "loss": 4.2377,
      "step": 1971
    },
    {
      "epoch": 0.01972,
      "grad_norm": 0.7105683738164339,
      "learning_rate": 0.003,
      "loss": 4.2407,
      "step": 1972
    },
    {
      "epoch": 0.01973,
      "grad_norm": 0.7508003521914575,
      "learning_rate": 0.003,
      "loss": 4.2264,
      "step": 1973
    },
    {
      "epoch": 0.01974,
      "grad_norm": 0.7061088132196116,
      "learning_rate": 0.003,
      "loss": 4.2641,
      "step": 1974
    },
    {
      "epoch": 0.01975,
      "grad_norm": 0.6140071978685933,
      "learning_rate": 0.003,
      "loss": 4.2539,
      "step": 1975
    },
    {
      "epoch": 0.01976,
      "grad_norm": 0.6214170582947839,
      "learning_rate": 0.003,
      "loss": 4.243,
      "step": 1976
    },
    {
      "epoch": 0.01977,
      "grad_norm": 0.6032091701282252,
      "learning_rate": 0.003,
      "loss": 4.2759,
      "step": 1977
    },
    {
      "epoch": 0.01978,
      "grad_norm": 0.6890579485446906,
      "learning_rate": 0.003,
      "loss": 4.2782,
      "step": 1978
    },
    {
      "epoch": 0.01979,
      "grad_norm": 0.7078287139411981,
      "learning_rate": 0.003,
      "loss": 4.269,
      "step": 1979
    },
    {
      "epoch": 0.0198,
      "grad_norm": 0.726915367409026,
      "learning_rate": 0.003,
      "loss": 4.2503,
      "step": 1980
    },
    {
      "epoch": 0.01981,
      "grad_norm": 0.7277727772925232,
      "learning_rate": 0.003,
      "loss": 4.2394,
      "step": 1981
    },
    {
      "epoch": 0.01982,
      "grad_norm": 0.6618110470758236,
      "learning_rate": 0.003,
      "loss": 4.262,
      "step": 1982
    },
    {
      "epoch": 0.01983,
      "grad_norm": 0.6282900232413579,
      "learning_rate": 0.003,
      "loss": 4.2497,
      "step": 1983
    },
    {
      "epoch": 0.01984,
      "grad_norm": 0.6651788332059134,
      "learning_rate": 0.003,
      "loss": 4.2635,
      "step": 1984
    },
    {
      "epoch": 0.01985,
      "grad_norm": 0.6195263920647736,
      "learning_rate": 0.003,
      "loss": 4.2484,
      "step": 1985
    },
    {
      "epoch": 0.01986,
      "grad_norm": 0.6562857171469267,
      "learning_rate": 0.003,
      "loss": 4.2792,
      "step": 1986
    },
    {
      "epoch": 0.01987,
      "grad_norm": 0.6274806569391715,
      "learning_rate": 0.003,
      "loss": 4.2331,
      "step": 1987
    },
    {
      "epoch": 0.01988,
      "grad_norm": 0.6627977107268054,
      "learning_rate": 0.003,
      "loss": 4.2466,
      "step": 1988
    },
    {
      "epoch": 0.01989,
      "grad_norm": 0.6286439849240877,
      "learning_rate": 0.003,
      "loss": 4.2344,
      "step": 1989
    },
    {
      "epoch": 0.0199,
      "grad_norm": 0.5786457654364954,
      "learning_rate": 0.003,
      "loss": 4.246,
      "step": 1990
    },
    {
      "epoch": 0.01991,
      "grad_norm": 0.5167049375276446,
      "learning_rate": 0.003,
      "loss": 4.2238,
      "step": 1991
    },
    {
      "epoch": 0.01992,
      "grad_norm": 0.50421095170271,
      "learning_rate": 0.003,
      "loss": 4.261,
      "step": 1992
    },
    {
      "epoch": 0.01993,
      "grad_norm": 0.4699320041008658,
      "learning_rate": 0.003,
      "loss": 4.2447,
      "step": 1993
    },
    {
      "epoch": 0.01994,
      "grad_norm": 0.5186369446769569,
      "learning_rate": 0.003,
      "loss": 4.2629,
      "step": 1994
    },
    {
      "epoch": 0.01995,
      "grad_norm": 0.5867398105885127,
      "learning_rate": 0.003,
      "loss": 4.2547,
      "step": 1995
    },
    {
      "epoch": 0.01996,
      "grad_norm": 0.8031924386202848,
      "learning_rate": 0.003,
      "loss": 4.2326,
      "step": 1996
    },
    {
      "epoch": 0.01997,
      "grad_norm": 0.9756949043315064,
      "learning_rate": 0.003,
      "loss": 4.2746,
      "step": 1997
    },
    {
      "epoch": 0.01998,
      "grad_norm": 0.8340147112264455,
      "learning_rate": 0.003,
      "loss": 4.2546,
      "step": 1998
    },
    {
      "epoch": 0.01999,
      "grad_norm": 0.6652469206537822,
      "learning_rate": 0.003,
      "loss": 4.2393,
      "step": 1999
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7695645348180246,
      "learning_rate": 0.003,
      "loss": 4.2318,
      "step": 2000
    },
    {
      "epoch": 0.02001,
      "grad_norm": 0.6865633733578148,
      "learning_rate": 0.003,
      "loss": 4.262,
      "step": 2001
    },
    {
      "epoch": 0.02002,
      "grad_norm": 0.7345949774088514,
      "learning_rate": 0.003,
      "loss": 4.2618,
      "step": 2002
    },
    {
      "epoch": 0.02003,
      "grad_norm": 0.5984396930779281,
      "learning_rate": 0.003,
      "loss": 4.2543,
      "step": 2003
    },
    {
      "epoch": 0.02004,
      "grad_norm": 0.5305961077422008,
      "learning_rate": 0.003,
      "loss": 4.2408,
      "step": 2004
    },
    {
      "epoch": 0.02005,
      "grad_norm": 0.48452673426344656,
      "learning_rate": 0.003,
      "loss": 4.2431,
      "step": 2005
    },
    {
      "epoch": 0.02006,
      "grad_norm": 0.5689487878863901,
      "learning_rate": 0.003,
      "loss": 4.2442,
      "step": 2006
    },
    {
      "epoch": 0.02007,
      "grad_norm": 0.5838792869042901,
      "learning_rate": 0.003,
      "loss": 4.2584,
      "step": 2007
    },
    {
      "epoch": 0.02008,
      "grad_norm": 0.6291452839633271,
      "learning_rate": 0.003,
      "loss": 4.2506,
      "step": 2008
    },
    {
      "epoch": 0.02009,
      "grad_norm": 0.6903355213664546,
      "learning_rate": 0.003,
      "loss": 4.2495,
      "step": 2009
    },
    {
      "epoch": 0.0201,
      "grad_norm": 0.6289287565263567,
      "learning_rate": 0.003,
      "loss": 4.2607,
      "step": 2010
    },
    {
      "epoch": 0.02011,
      "grad_norm": 0.5984818046752132,
      "learning_rate": 0.003,
      "loss": 4.2452,
      "step": 2011
    },
    {
      "epoch": 0.02012,
      "grad_norm": 0.6682118694640922,
      "learning_rate": 0.003,
      "loss": 4.2452,
      "step": 2012
    },
    {
      "epoch": 0.02013,
      "grad_norm": 0.7050046827190253,
      "learning_rate": 0.003,
      "loss": 4.2532,
      "step": 2013
    },
    {
      "epoch": 0.02014,
      "grad_norm": 0.6761323120848991,
      "learning_rate": 0.003,
      "loss": 4.2415,
      "step": 2014
    },
    {
      "epoch": 0.02015,
      "grad_norm": 0.6372907183843184,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 2015
    },
    {
      "epoch": 0.02016,
      "grad_norm": 0.5972921997645952,
      "learning_rate": 0.003,
      "loss": 4.2404,
      "step": 2016
    },
    {
      "epoch": 0.02017,
      "grad_norm": 0.49149348634303763,
      "learning_rate": 0.003,
      "loss": 4.2312,
      "step": 2017
    },
    {
      "epoch": 0.02018,
      "grad_norm": 0.6030678218331625,
      "learning_rate": 0.003,
      "loss": 4.2371,
      "step": 2018
    },
    {
      "epoch": 0.02019,
      "grad_norm": 0.6395667829535132,
      "learning_rate": 0.003,
      "loss": 4.2431,
      "step": 2019
    },
    {
      "epoch": 0.0202,
      "grad_norm": 0.6210152394675877,
      "learning_rate": 0.003,
      "loss": 4.2512,
      "step": 2020
    },
    {
      "epoch": 0.02021,
      "grad_norm": 0.6950283205655802,
      "learning_rate": 0.003,
      "loss": 4.268,
      "step": 2021
    },
    {
      "epoch": 0.02022,
      "grad_norm": 0.8101179933347831,
      "learning_rate": 0.003,
      "loss": 4.2328,
      "step": 2022
    },
    {
      "epoch": 0.02023,
      "grad_norm": 0.8379556185876997,
      "learning_rate": 0.003,
      "loss": 4.2128,
      "step": 2023
    },
    {
      "epoch": 0.02024,
      "grad_norm": 0.8247606718710992,
      "learning_rate": 0.003,
      "loss": 4.2373,
      "step": 2024
    },
    {
      "epoch": 0.02025,
      "grad_norm": 0.8736161526776162,
      "learning_rate": 0.003,
      "loss": 4.291,
      "step": 2025
    },
    {
      "epoch": 0.02026,
      "grad_norm": 0.8472496904448235,
      "learning_rate": 0.003,
      "loss": 4.2727,
      "step": 2026
    },
    {
      "epoch": 0.02027,
      "grad_norm": 0.8056170331271387,
      "learning_rate": 0.003,
      "loss": 4.271,
      "step": 2027
    },
    {
      "epoch": 0.02028,
      "grad_norm": 0.8498369070927115,
      "learning_rate": 0.003,
      "loss": 4.2569,
      "step": 2028
    },
    {
      "epoch": 0.02029,
      "grad_norm": 0.9047394863120564,
      "learning_rate": 0.003,
      "loss": 4.2658,
      "step": 2029
    },
    {
      "epoch": 0.0203,
      "grad_norm": 1.0823713408101607,
      "learning_rate": 0.003,
      "loss": 4.2684,
      "step": 2030
    },
    {
      "epoch": 0.02031,
      "grad_norm": 0.9309326946376139,
      "learning_rate": 0.003,
      "loss": 4.2732,
      "step": 2031
    },
    {
      "epoch": 0.02032,
      "grad_norm": 0.9146514324992777,
      "learning_rate": 0.003,
      "loss": 4.2447,
      "step": 2032
    },
    {
      "epoch": 0.02033,
      "grad_norm": 0.8591399303672403,
      "learning_rate": 0.003,
      "loss": 4.2848,
      "step": 2033
    },
    {
      "epoch": 0.02034,
      "grad_norm": 0.8501591519538507,
      "learning_rate": 0.003,
      "loss": 4.2786,
      "step": 2034
    },
    {
      "epoch": 0.02035,
      "grad_norm": 0.8000776065354223,
      "learning_rate": 0.003,
      "loss": 4.2613,
      "step": 2035
    },
    {
      "epoch": 0.02036,
      "grad_norm": 0.7789847842987989,
      "learning_rate": 0.003,
      "loss": 4.2629,
      "step": 2036
    },
    {
      "epoch": 0.02037,
      "grad_norm": 0.8301107777503408,
      "learning_rate": 0.003,
      "loss": 4.2828,
      "step": 2037
    },
    {
      "epoch": 0.02038,
      "grad_norm": 0.8246182480374772,
      "learning_rate": 0.003,
      "loss": 4.2649,
      "step": 2038
    },
    {
      "epoch": 0.02039,
      "grad_norm": 0.8433738197183226,
      "learning_rate": 0.003,
      "loss": 4.2745,
      "step": 2039
    },
    {
      "epoch": 0.0204,
      "grad_norm": 0.7922524759276299,
      "learning_rate": 0.003,
      "loss": 4.2766,
      "step": 2040
    },
    {
      "epoch": 0.02041,
      "grad_norm": 0.7214169410748629,
      "learning_rate": 0.003,
      "loss": 4.2381,
      "step": 2041
    },
    {
      "epoch": 0.02042,
      "grad_norm": 0.5980473954078982,
      "learning_rate": 0.003,
      "loss": 4.2311,
      "step": 2042
    },
    {
      "epoch": 0.02043,
      "grad_norm": 0.6544292652761489,
      "learning_rate": 0.003,
      "loss": 4.2765,
      "step": 2043
    },
    {
      "epoch": 0.02044,
      "grad_norm": 0.6815263020074647,
      "learning_rate": 0.003,
      "loss": 4.2533,
      "step": 2044
    },
    {
      "epoch": 0.02045,
      "grad_norm": 0.6716783756563022,
      "learning_rate": 0.003,
      "loss": 4.266,
      "step": 2045
    },
    {
      "epoch": 0.02046,
      "grad_norm": 0.5685869111952888,
      "learning_rate": 0.003,
      "loss": 4.2364,
      "step": 2046
    },
    {
      "epoch": 0.02047,
      "grad_norm": 0.539068698157252,
      "learning_rate": 0.003,
      "loss": 4.2693,
      "step": 2047
    },
    {
      "epoch": 0.02048,
      "grad_norm": 0.5462552506791943,
      "learning_rate": 0.003,
      "loss": 4.2521,
      "step": 2048
    },
    {
      "epoch": 0.02049,
      "grad_norm": 0.5078072105462605,
      "learning_rate": 0.003,
      "loss": 4.2715,
      "step": 2049
    },
    {
      "epoch": 0.0205,
      "grad_norm": 0.5740623442922046,
      "learning_rate": 0.003,
      "loss": 4.2205,
      "step": 2050
    },
    {
      "epoch": 0.02051,
      "grad_norm": 0.631335957620747,
      "learning_rate": 0.003,
      "loss": 4.2224,
      "step": 2051
    },
    {
      "epoch": 0.02052,
      "grad_norm": 0.6486110208729147,
      "learning_rate": 0.003,
      "loss": 4.2258,
      "step": 2052
    },
    {
      "epoch": 0.02053,
      "grad_norm": 0.5888200401710874,
      "learning_rate": 0.003,
      "loss": 4.2235,
      "step": 2053
    },
    {
      "epoch": 0.02054,
      "grad_norm": 0.5515462434036196,
      "learning_rate": 0.003,
      "loss": 4.2611,
      "step": 2054
    },
    {
      "epoch": 0.02055,
      "grad_norm": 0.4722653472823476,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2055
    },
    {
      "epoch": 0.02056,
      "grad_norm": 0.44396394612077134,
      "learning_rate": 0.003,
      "loss": 4.2149,
      "step": 2056
    },
    {
      "epoch": 0.02057,
      "grad_norm": 0.4548184991267254,
      "learning_rate": 0.003,
      "loss": 4.2514,
      "step": 2057
    },
    {
      "epoch": 0.02058,
      "grad_norm": 0.5196145685763705,
      "learning_rate": 0.003,
      "loss": 4.2286,
      "step": 2058
    },
    {
      "epoch": 0.02059,
      "grad_norm": 0.5150404447757911,
      "learning_rate": 0.003,
      "loss": 4.2089,
      "step": 2059
    },
    {
      "epoch": 0.0206,
      "grad_norm": 0.44383213075680966,
      "learning_rate": 0.003,
      "loss": 4.251,
      "step": 2060
    },
    {
      "epoch": 0.02061,
      "grad_norm": 0.37383563921339574,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2061
    },
    {
      "epoch": 0.02062,
      "grad_norm": 0.35393934819548667,
      "learning_rate": 0.003,
      "loss": 4.221,
      "step": 2062
    },
    {
      "epoch": 0.02063,
      "grad_norm": 0.3267679809833493,
      "learning_rate": 0.003,
      "loss": 4.2165,
      "step": 2063
    },
    {
      "epoch": 0.02064,
      "grad_norm": 0.34033928546810105,
      "learning_rate": 0.003,
      "loss": 4.2259,
      "step": 2064
    },
    {
      "epoch": 0.02065,
      "grad_norm": 0.35466259038416126,
      "learning_rate": 0.003,
      "loss": 4.2059,
      "step": 2065
    },
    {
      "epoch": 0.02066,
      "grad_norm": 0.3016586113305969,
      "learning_rate": 0.003,
      "loss": 4.2224,
      "step": 2066
    },
    {
      "epoch": 0.02067,
      "grad_norm": 0.3498296991074057,
      "learning_rate": 0.003,
      "loss": 4.22,
      "step": 2067
    },
    {
      "epoch": 0.02068,
      "grad_norm": 0.3676321022781276,
      "learning_rate": 0.003,
      "loss": 4.2017,
      "step": 2068
    },
    {
      "epoch": 0.02069,
      "grad_norm": 0.40962533052122363,
      "learning_rate": 0.003,
      "loss": 4.2244,
      "step": 2069
    },
    {
      "epoch": 0.0207,
      "grad_norm": 0.5995962395694695,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 2070
    },
    {
      "epoch": 0.02071,
      "grad_norm": 1.0281554874982604,
      "learning_rate": 0.003,
      "loss": 4.2499,
      "step": 2071
    },
    {
      "epoch": 0.02072,
      "grad_norm": 1.2174002009617175,
      "learning_rate": 0.003,
      "loss": 4.2817,
      "step": 2072
    },
    {
      "epoch": 0.02073,
      "grad_norm": 0.5766208529721838,
      "learning_rate": 0.003,
      "loss": 4.2339,
      "step": 2073
    },
    {
      "epoch": 0.02074,
      "grad_norm": 0.8482120839475051,
      "learning_rate": 0.003,
      "loss": 4.2516,
      "step": 2074
    },
    {
      "epoch": 0.02075,
      "grad_norm": 0.914249694296833,
      "learning_rate": 0.003,
      "loss": 4.2873,
      "step": 2075
    },
    {
      "epoch": 0.02076,
      "grad_norm": 1.0073872264521357,
      "learning_rate": 0.003,
      "loss": 4.2898,
      "step": 2076
    },
    {
      "epoch": 0.02077,
      "grad_norm": 0.9613925913617914,
      "learning_rate": 0.003,
      "loss": 4.2576,
      "step": 2077
    },
    {
      "epoch": 0.02078,
      "grad_norm": 1.0251025412787642,
      "learning_rate": 0.003,
      "loss": 4.2665,
      "step": 2078
    },
    {
      "epoch": 0.02079,
      "grad_norm": 0.8112389212571772,
      "learning_rate": 0.003,
      "loss": 4.243,
      "step": 2079
    },
    {
      "epoch": 0.0208,
      "grad_norm": 0.9069389526132752,
      "learning_rate": 0.003,
      "loss": 4.2693,
      "step": 2080
    },
    {
      "epoch": 0.02081,
      "grad_norm": 1.1627691288871134,
      "learning_rate": 0.003,
      "loss": 4.3161,
      "step": 2081
    },
    {
      "epoch": 0.02082,
      "grad_norm": 1.0285610491992647,
      "learning_rate": 0.003,
      "loss": 4.3126,
      "step": 2082
    },
    {
      "epoch": 0.02083,
      "grad_norm": 1.0850701155180529,
      "learning_rate": 0.003,
      "loss": 4.3091,
      "step": 2083
    },
    {
      "epoch": 0.02084,
      "grad_norm": 0.9461237384614024,
      "learning_rate": 0.003,
      "loss": 4.2987,
      "step": 2084
    },
    {
      "epoch": 0.02085,
      "grad_norm": 0.976394782581613,
      "learning_rate": 0.003,
      "loss": 4.284,
      "step": 2085
    },
    {
      "epoch": 0.02086,
      "grad_norm": 0.82502607345478,
      "learning_rate": 0.003,
      "loss": 4.2825,
      "step": 2086
    },
    {
      "epoch": 0.02087,
      "grad_norm": 0.6641978639980478,
      "learning_rate": 0.003,
      "loss": 4.2762,
      "step": 2087
    },
    {
      "epoch": 0.02088,
      "grad_norm": 0.5823960320780495,
      "learning_rate": 0.003,
      "loss": 4.3019,
      "step": 2088
    },
    {
      "epoch": 0.02089,
      "grad_norm": 0.6056623050213609,
      "learning_rate": 0.003,
      "loss": 4.2696,
      "step": 2089
    },
    {
      "epoch": 0.0209,
      "grad_norm": 0.6222015176775844,
      "learning_rate": 0.003,
      "loss": 4.2845,
      "step": 2090
    },
    {
      "epoch": 0.02091,
      "grad_norm": 0.8107138314533802,
      "learning_rate": 0.003,
      "loss": 4.275,
      "step": 2091
    },
    {
      "epoch": 0.02092,
      "grad_norm": 0.9481339301239347,
      "learning_rate": 0.003,
      "loss": 4.2799,
      "step": 2092
    },
    {
      "epoch": 0.02093,
      "grad_norm": 0.8740047196660798,
      "learning_rate": 0.003,
      "loss": 4.2456,
      "step": 2093
    },
    {
      "epoch": 0.02094,
      "grad_norm": 0.6305842273976254,
      "learning_rate": 0.003,
      "loss": 4.2436,
      "step": 2094
    },
    {
      "epoch": 0.02095,
      "grad_norm": 0.5189901063960811,
      "learning_rate": 0.003,
      "loss": 4.2773,
      "step": 2095
    },
    {
      "epoch": 0.02096,
      "grad_norm": 0.7024275821596554,
      "learning_rate": 0.003,
      "loss": 4.2741,
      "step": 2096
    },
    {
      "epoch": 0.02097,
      "grad_norm": 0.6837847162193671,
      "learning_rate": 0.003,
      "loss": 4.2518,
      "step": 2097
    },
    {
      "epoch": 0.02098,
      "grad_norm": 0.6707932407749881,
      "learning_rate": 0.003,
      "loss": 4.2705,
      "step": 2098
    },
    {
      "epoch": 0.02099,
      "grad_norm": 0.5110579033218214,
      "learning_rate": 0.003,
      "loss": 4.2724,
      "step": 2099
    },
    {
      "epoch": 0.021,
      "grad_norm": 0.4612182402663896,
      "learning_rate": 0.003,
      "loss": 4.2819,
      "step": 2100
    },
    {
      "epoch": 0.02101,
      "grad_norm": 0.45331179352856693,
      "learning_rate": 0.003,
      "loss": 4.2684,
      "step": 2101
    },
    {
      "epoch": 0.02102,
      "grad_norm": 0.42546683266841523,
      "learning_rate": 0.003,
      "loss": 4.271,
      "step": 2102
    },
    {
      "epoch": 0.02103,
      "grad_norm": 0.3547519868664197,
      "learning_rate": 0.003,
      "loss": 4.2273,
      "step": 2103
    },
    {
      "epoch": 0.02104,
      "grad_norm": 0.3900500431869311,
      "learning_rate": 0.003,
      "loss": 4.2306,
      "step": 2104
    },
    {
      "epoch": 0.02105,
      "grad_norm": 0.4122345047469997,
      "learning_rate": 0.003,
      "loss": 4.2329,
      "step": 2105
    },
    {
      "epoch": 0.02106,
      "grad_norm": 0.4185113160695593,
      "learning_rate": 0.003,
      "loss": 4.2247,
      "step": 2106
    },
    {
      "epoch": 0.02107,
      "grad_norm": 0.3925543469529075,
      "learning_rate": 0.003,
      "loss": 4.2367,
      "step": 2107
    },
    {
      "epoch": 0.02108,
      "grad_norm": 0.33327126218953085,
      "learning_rate": 0.003,
      "loss": 4.2153,
      "step": 2108
    },
    {
      "epoch": 0.02109,
      "grad_norm": 0.3153436836891661,
      "learning_rate": 0.003,
      "loss": 4.2035,
      "step": 2109
    },
    {
      "epoch": 0.0211,
      "grad_norm": 0.35294607433058894,
      "learning_rate": 0.003,
      "loss": 4.2193,
      "step": 2110
    },
    {
      "epoch": 0.02111,
      "grad_norm": 0.37843249197178286,
      "learning_rate": 0.003,
      "loss": 4.2278,
      "step": 2111
    },
    {
      "epoch": 0.02112,
      "grad_norm": 0.38829107411949254,
      "learning_rate": 0.003,
      "loss": 4.226,
      "step": 2112
    },
    {
      "epoch": 0.02113,
      "grad_norm": 0.36727891938554497,
      "learning_rate": 0.003,
      "loss": 4.2209,
      "step": 2113
    },
    {
      "epoch": 0.02114,
      "grad_norm": 0.3796589384288701,
      "learning_rate": 0.003,
      "loss": 4.2115,
      "step": 2114
    },
    {
      "epoch": 0.02115,
      "grad_norm": 0.49933195549505976,
      "learning_rate": 0.003,
      "loss": 4.2045,
      "step": 2115
    },
    {
      "epoch": 0.02116,
      "grad_norm": 0.7078427672787017,
      "learning_rate": 0.003,
      "loss": 4.2392,
      "step": 2116
    },
    {
      "epoch": 0.02117,
      "grad_norm": 0.9422169918121244,
      "learning_rate": 0.003,
      "loss": 4.2414,
      "step": 2117
    },
    {
      "epoch": 0.02118,
      "grad_norm": 0.8924259465708047,
      "learning_rate": 0.003,
      "loss": 4.2414,
      "step": 2118
    },
    {
      "epoch": 0.02119,
      "grad_norm": 0.5681560145798641,
      "learning_rate": 0.003,
      "loss": 4.2397,
      "step": 2119
    },
    {
      "epoch": 0.0212,
      "grad_norm": 0.5070503337124914,
      "learning_rate": 0.003,
      "loss": 4.213,
      "step": 2120
    },
    {
      "epoch": 0.02121,
      "grad_norm": 0.5980827300624205,
      "learning_rate": 0.003,
      "loss": 4.2118,
      "step": 2121
    },
    {
      "epoch": 0.02122,
      "grad_norm": 0.48859129709793475,
      "learning_rate": 0.003,
      "loss": 4.2167,
      "step": 2122
    },
    {
      "epoch": 0.02123,
      "grad_norm": 0.463982076261229,
      "learning_rate": 0.003,
      "loss": 4.1963,
      "step": 2123
    },
    {
      "epoch": 0.02124,
      "grad_norm": 0.44089509619617673,
      "learning_rate": 0.003,
      "loss": 4.2174,
      "step": 2124
    },
    {
      "epoch": 0.02125,
      "grad_norm": 0.48261858431331583,
      "learning_rate": 0.003,
      "loss": 4.2332,
      "step": 2125
    },
    {
      "epoch": 0.02126,
      "grad_norm": 0.5099262151069514,
      "learning_rate": 0.003,
      "loss": 4.2302,
      "step": 2126
    },
    {
      "epoch": 0.02127,
      "grad_norm": 0.5196898928719708,
      "learning_rate": 0.003,
      "loss": 4.2379,
      "step": 2127
    },
    {
      "epoch": 0.02128,
      "grad_norm": 0.5656439270875484,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2128
    },
    {
      "epoch": 0.02129,
      "grad_norm": 0.5221861829481255,
      "learning_rate": 0.003,
      "loss": 4.2297,
      "step": 2129
    },
    {
      "epoch": 0.0213,
      "grad_norm": 0.4521057397115596,
      "learning_rate": 0.003,
      "loss": 4.2027,
      "step": 2130
    },
    {
      "epoch": 0.02131,
      "grad_norm": 0.45364673608644235,
      "learning_rate": 0.003,
      "loss": 4.2268,
      "step": 2131
    },
    {
      "epoch": 0.02132,
      "grad_norm": 0.519443445123456,
      "learning_rate": 0.003,
      "loss": 4.2417,
      "step": 2132
    },
    {
      "epoch": 0.02133,
      "grad_norm": 0.7304407810627219,
      "learning_rate": 0.003,
      "loss": 4.2391,
      "step": 2133
    },
    {
      "epoch": 0.02134,
      "grad_norm": 0.9541768215835498,
      "learning_rate": 0.003,
      "loss": 4.2522,
      "step": 2134
    },
    {
      "epoch": 0.02135,
      "grad_norm": 1.0307643830076532,
      "learning_rate": 0.003,
      "loss": 4.2588,
      "step": 2135
    },
    {
      "epoch": 0.02136,
      "grad_norm": 0.997681208751024,
      "learning_rate": 0.003,
      "loss": 4.2363,
      "step": 2136
    },
    {
      "epoch": 0.02137,
      "grad_norm": 0.8370170695212052,
      "learning_rate": 0.003,
      "loss": 4.2696,
      "step": 2137
    },
    {
      "epoch": 0.02138,
      "grad_norm": 0.9176192838894591,
      "learning_rate": 0.003,
      "loss": 4.2858,
      "step": 2138
    },
    {
      "epoch": 0.02139,
      "grad_norm": 0.9058155406944488,
      "learning_rate": 0.003,
      "loss": 4.2833,
      "step": 2139
    },
    {
      "epoch": 0.0214,
      "grad_norm": 0.8679543052073141,
      "learning_rate": 0.003,
      "loss": 4.2707,
      "step": 2140
    },
    {
      "epoch": 0.02141,
      "grad_norm": 0.7817830154220525,
      "learning_rate": 0.003,
      "loss": 4.2527,
      "step": 2141
    },
    {
      "epoch": 0.02142,
      "grad_norm": 0.798437758428762,
      "learning_rate": 0.003,
      "loss": 4.2453,
      "step": 2142
    },
    {
      "epoch": 0.02143,
      "grad_norm": 0.7871322126205197,
      "learning_rate": 0.003,
      "loss": 4.2612,
      "step": 2143
    },
    {
      "epoch": 0.02144,
      "grad_norm": 0.8115888487824465,
      "learning_rate": 0.003,
      "loss": 4.2653,
      "step": 2144
    },
    {
      "epoch": 0.02145,
      "grad_norm": 0.7440847388519684,
      "learning_rate": 0.003,
      "loss": 4.2448,
      "step": 2145
    },
    {
      "epoch": 0.02146,
      "grad_norm": 0.6836186393980395,
      "learning_rate": 0.003,
      "loss": 4.2558,
      "step": 2146
    },
    {
      "epoch": 0.02147,
      "grad_norm": 0.6617245709810007,
      "learning_rate": 0.003,
      "loss": 4.2727,
      "step": 2147
    },
    {
      "epoch": 0.02148,
      "grad_norm": 0.7076496122050561,
      "learning_rate": 0.003,
      "loss": 4.2832,
      "step": 2148
    },
    {
      "epoch": 0.02149,
      "grad_norm": 0.729078859235357,
      "learning_rate": 0.003,
      "loss": 4.2694,
      "step": 2149
    },
    {
      "epoch": 0.0215,
      "grad_norm": 0.6732791642286783,
      "learning_rate": 0.003,
      "loss": 4.2532,
      "step": 2150
    },
    {
      "epoch": 0.02151,
      "grad_norm": 0.6647883181229363,
      "learning_rate": 0.003,
      "loss": 4.2571,
      "step": 2151
    },
    {
      "epoch": 0.02152,
      "grad_norm": 0.539629911002914,
      "learning_rate": 0.003,
      "loss": 4.2298,
      "step": 2152
    },
    {
      "epoch": 0.02153,
      "grad_norm": 0.4979893391256132,
      "learning_rate": 0.003,
      "loss": 4.2372,
      "step": 2153
    },
    {
      "epoch": 0.02154,
      "grad_norm": 0.43644785385536694,
      "learning_rate": 0.003,
      "loss": 4.2266,
      "step": 2154
    },
    {
      "epoch": 0.02155,
      "grad_norm": 0.3633150618094754,
      "learning_rate": 0.003,
      "loss": 4.2532,
      "step": 2155
    },
    {
      "epoch": 0.02156,
      "grad_norm": 0.39513834073913734,
      "learning_rate": 0.003,
      "loss": 4.239,
      "step": 2156
    },
    {
      "epoch": 0.02157,
      "grad_norm": 0.3860719279707547,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 2157
    },
    {
      "epoch": 0.02158,
      "grad_norm": 0.4137178291066844,
      "learning_rate": 0.003,
      "loss": 4.2241,
      "step": 2158
    },
    {
      "epoch": 0.02159,
      "grad_norm": 0.5163329954248042,
      "learning_rate": 0.003,
      "loss": 4.2392,
      "step": 2159
    },
    {
      "epoch": 0.0216,
      "grad_norm": 0.6337653160295451,
      "learning_rate": 0.003,
      "loss": 4.2443,
      "step": 2160
    },
    {
      "epoch": 0.02161,
      "grad_norm": 0.7844529335341662,
      "learning_rate": 0.003,
      "loss": 4.2371,
      "step": 2161
    },
    {
      "epoch": 0.02162,
      "grad_norm": 0.8327279956908368,
      "learning_rate": 0.003,
      "loss": 4.2454,
      "step": 2162
    },
    {
      "epoch": 0.02163,
      "grad_norm": 0.6324138237584291,
      "learning_rate": 0.003,
      "loss": 4.2546,
      "step": 2163
    },
    {
      "epoch": 0.02164,
      "grad_norm": 0.4272642114449097,
      "learning_rate": 0.003,
      "loss": 4.2393,
      "step": 2164
    },
    {
      "epoch": 0.02165,
      "grad_norm": 0.5755156335143842,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2165
    },
    {
      "epoch": 0.02166,
      "grad_norm": 0.5179351914703093,
      "learning_rate": 0.003,
      "loss": 4.2452,
      "step": 2166
    },
    {
      "epoch": 0.02167,
      "grad_norm": 0.47795702712318133,
      "learning_rate": 0.003,
      "loss": 4.2215,
      "step": 2167
    },
    {
      "epoch": 0.02168,
      "grad_norm": 0.39306911706804365,
      "learning_rate": 0.003,
      "loss": 4.2181,
      "step": 2168
    },
    {
      "epoch": 0.02169,
      "grad_norm": 0.43640956278445237,
      "learning_rate": 0.003,
      "loss": 4.2172,
      "step": 2169
    },
    {
      "epoch": 0.0217,
      "grad_norm": 0.44151372609355966,
      "learning_rate": 0.003,
      "loss": 4.2252,
      "step": 2170
    },
    {
      "epoch": 0.02171,
      "grad_norm": 0.4398868813684568,
      "learning_rate": 0.003,
      "loss": 4.238,
      "step": 2171
    },
    {
      "epoch": 0.02172,
      "grad_norm": 0.453127984813779,
      "learning_rate": 0.003,
      "loss": 4.2099,
      "step": 2172
    },
    {
      "epoch": 0.02173,
      "grad_norm": 0.4200137719224221,
      "learning_rate": 0.003,
      "loss": 4.2112,
      "step": 2173
    },
    {
      "epoch": 0.02174,
      "grad_norm": 0.40927864912838796,
      "learning_rate": 0.003,
      "loss": 4.2074,
      "step": 2174
    },
    {
      "epoch": 0.02175,
      "grad_norm": 0.37273652046629646,
      "learning_rate": 0.003,
      "loss": 4.2305,
      "step": 2175
    },
    {
      "epoch": 0.02176,
      "grad_norm": 0.3722266811459993,
      "learning_rate": 0.003,
      "loss": 4.2097,
      "step": 2176
    },
    {
      "epoch": 0.02177,
      "grad_norm": 0.43921830418385466,
      "learning_rate": 0.003,
      "loss": 4.2187,
      "step": 2177
    },
    {
      "epoch": 0.02178,
      "grad_norm": 0.5214363509579703,
      "learning_rate": 0.003,
      "loss": 4.1985,
      "step": 2178
    },
    {
      "epoch": 0.02179,
      "grad_norm": 0.8402711445332305,
      "learning_rate": 0.003,
      "loss": 4.2235,
      "step": 2179
    },
    {
      "epoch": 0.0218,
      "grad_norm": 1.247371610474746,
      "learning_rate": 0.003,
      "loss": 4.2635,
      "step": 2180
    },
    {
      "epoch": 0.02181,
      "grad_norm": 0.8116435611354782,
      "learning_rate": 0.003,
      "loss": 4.2484,
      "step": 2181
    },
    {
      "epoch": 0.02182,
      "grad_norm": 0.6837903180839037,
      "learning_rate": 0.003,
      "loss": 4.2355,
      "step": 2182
    },
    {
      "epoch": 0.02183,
      "grad_norm": 0.7005151877778275,
      "learning_rate": 0.003,
      "loss": 4.2476,
      "step": 2183
    },
    {
      "epoch": 0.02184,
      "grad_norm": 0.8236745566528523,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2184
    },
    {
      "epoch": 0.02185,
      "grad_norm": 0.91418771570573,
      "learning_rate": 0.003,
      "loss": 4.2327,
      "step": 2185
    },
    {
      "epoch": 0.02186,
      "grad_norm": 0.8323760045804593,
      "learning_rate": 0.003,
      "loss": 4.2482,
      "step": 2186
    },
    {
      "epoch": 0.02187,
      "grad_norm": 0.8176331489387878,
      "learning_rate": 0.003,
      "loss": 4.245,
      "step": 2187
    },
    {
      "epoch": 0.02188,
      "grad_norm": 0.8776011343747697,
      "learning_rate": 0.003,
      "loss": 4.2325,
      "step": 2188
    },
    {
      "epoch": 0.02189,
      "grad_norm": 0.8972484956693035,
      "learning_rate": 0.003,
      "loss": 4.2667,
      "step": 2189
    },
    {
      "epoch": 0.0219,
      "grad_norm": 0.8431917439790871,
      "learning_rate": 0.003,
      "loss": 4.2584,
      "step": 2190
    },
    {
      "epoch": 0.02191,
      "grad_norm": 0.7416940909972495,
      "learning_rate": 0.003,
      "loss": 4.2439,
      "step": 2191
    },
    {
      "epoch": 0.02192,
      "grad_norm": 0.6213980069901923,
      "learning_rate": 0.003,
      "loss": 4.2676,
      "step": 2192
    },
    {
      "epoch": 0.02193,
      "grad_norm": 0.6597970415430513,
      "learning_rate": 0.003,
      "loss": 4.2343,
      "step": 2193
    },
    {
      "epoch": 0.02194,
      "grad_norm": 0.7061865144419822,
      "learning_rate": 0.003,
      "loss": 4.2832,
      "step": 2194
    },
    {
      "epoch": 0.02195,
      "grad_norm": 0.7006382568298751,
      "learning_rate": 0.003,
      "loss": 4.251,
      "step": 2195
    },
    {
      "epoch": 0.02196,
      "grad_norm": 0.6722854605002367,
      "learning_rate": 0.003,
      "loss": 4.2565,
      "step": 2196
    },
    {
      "epoch": 0.02197,
      "grad_norm": 0.6383221093399062,
      "learning_rate": 0.003,
      "loss": 4.2472,
      "step": 2197
    },
    {
      "epoch": 0.02198,
      "grad_norm": 0.6598256711875262,
      "learning_rate": 0.003,
      "loss": 4.2361,
      "step": 2198
    },
    {
      "epoch": 0.02199,
      "grad_norm": 0.6310822502739497,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2199
    },
    {
      "epoch": 0.022,
      "grad_norm": 0.5807092706295028,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 2200
    },
    {
      "epoch": 0.02201,
      "grad_norm": 0.5164997329736423,
      "learning_rate": 0.003,
      "loss": 4.2429,
      "step": 2201
    },
    {
      "epoch": 0.02202,
      "grad_norm": 0.5150399093991901,
      "learning_rate": 0.003,
      "loss": 4.2262,
      "step": 2202
    },
    {
      "epoch": 0.02203,
      "grad_norm": 0.5301942341385608,
      "learning_rate": 0.003,
      "loss": 4.2546,
      "step": 2203
    },
    {
      "epoch": 0.02204,
      "grad_norm": 0.5633853956311704,
      "learning_rate": 0.003,
      "loss": 4.2484,
      "step": 2204
    },
    {
      "epoch": 0.02205,
      "grad_norm": 0.5924154634458846,
      "learning_rate": 0.003,
      "loss": 4.2516,
      "step": 2205
    },
    {
      "epoch": 0.02206,
      "grad_norm": 0.6443822475335407,
      "learning_rate": 0.003,
      "loss": 4.2165,
      "step": 2206
    },
    {
      "epoch": 0.02207,
      "grad_norm": 0.6602238711518633,
      "learning_rate": 0.003,
      "loss": 4.2293,
      "step": 2207
    },
    {
      "epoch": 0.02208,
      "grad_norm": 0.6586527270486826,
      "learning_rate": 0.003,
      "loss": 4.2124,
      "step": 2208
    },
    {
      "epoch": 0.02209,
      "grad_norm": 0.5991461642461526,
      "learning_rate": 0.003,
      "loss": 4.2276,
      "step": 2209
    },
    {
      "epoch": 0.0221,
      "grad_norm": 0.5268358370568649,
      "learning_rate": 0.003,
      "loss": 4.2071,
      "step": 2210
    },
    {
      "epoch": 0.02211,
      "grad_norm": 0.6149216818507237,
      "learning_rate": 0.003,
      "loss": 4.2195,
      "step": 2211
    },
    {
      "epoch": 0.02212,
      "grad_norm": 0.7463716016406938,
      "learning_rate": 0.003,
      "loss": 4.2226,
      "step": 2212
    },
    {
      "epoch": 0.02213,
      "grad_norm": 0.7963090983602238,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 2213
    },
    {
      "epoch": 0.02214,
      "grad_norm": 0.7746780288539415,
      "learning_rate": 0.003,
      "loss": 4.2283,
      "step": 2214
    },
    {
      "epoch": 0.02215,
      "grad_norm": 0.7305597710999601,
      "learning_rate": 0.003,
      "loss": 4.2338,
      "step": 2215
    },
    {
      "epoch": 0.02216,
      "grad_norm": 0.6802860521431111,
      "learning_rate": 0.003,
      "loss": 4.2097,
      "step": 2216
    },
    {
      "epoch": 0.02217,
      "grad_norm": 0.814645356813129,
      "learning_rate": 0.003,
      "loss": 4.2149,
      "step": 2217
    },
    {
      "epoch": 0.02218,
      "grad_norm": 0.8644777161103991,
      "learning_rate": 0.003,
      "loss": 4.2502,
      "step": 2218
    },
    {
      "epoch": 0.02219,
      "grad_norm": 0.8895266477572761,
      "learning_rate": 0.003,
      "loss": 4.2436,
      "step": 2219
    },
    {
      "epoch": 0.0222,
      "grad_norm": 0.8379487765766322,
      "learning_rate": 0.003,
      "loss": 4.241,
      "step": 2220
    },
    {
      "epoch": 0.02221,
      "grad_norm": 0.8832652718422841,
      "learning_rate": 0.003,
      "loss": 4.23,
      "step": 2221
    },
    {
      "epoch": 0.02222,
      "grad_norm": 0.7112453634526715,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2222
    },
    {
      "epoch": 0.02223,
      "grad_norm": 0.5578145775957193,
      "learning_rate": 0.003,
      "loss": 4.2345,
      "step": 2223
    },
    {
      "epoch": 0.02224,
      "grad_norm": 0.5281644921526645,
      "learning_rate": 0.003,
      "loss": 4.216,
      "step": 2224
    },
    {
      "epoch": 0.02225,
      "grad_norm": 0.499018726060841,
      "learning_rate": 0.003,
      "loss": 4.216,
      "step": 2225
    },
    {
      "epoch": 0.02226,
      "grad_norm": 0.5772234482094156,
      "learning_rate": 0.003,
      "loss": 4.2362,
      "step": 2226
    },
    {
      "epoch": 0.02227,
      "grad_norm": 0.6481086517118151,
      "learning_rate": 0.003,
      "loss": 4.2311,
      "step": 2227
    },
    {
      "epoch": 0.02228,
      "grad_norm": 0.669454673605761,
      "learning_rate": 0.003,
      "loss": 4.2187,
      "step": 2228
    },
    {
      "epoch": 0.02229,
      "grad_norm": 0.677093243217859,
      "learning_rate": 0.003,
      "loss": 4.1992,
      "step": 2229
    },
    {
      "epoch": 0.0223,
      "grad_norm": 0.6785810316864852,
      "learning_rate": 0.003,
      "loss": 4.2386,
      "step": 2230
    },
    {
      "epoch": 0.02231,
      "grad_norm": 0.697493374925287,
      "learning_rate": 0.003,
      "loss": 4.2447,
      "step": 2231
    },
    {
      "epoch": 0.02232,
      "grad_norm": 0.8916380613416046,
      "learning_rate": 0.003,
      "loss": 4.2575,
      "step": 2232
    },
    {
      "epoch": 0.02233,
      "grad_norm": 1.0887336160767505,
      "learning_rate": 0.003,
      "loss": 4.2371,
      "step": 2233
    },
    {
      "epoch": 0.02234,
      "grad_norm": 0.654547980354846,
      "learning_rate": 0.003,
      "loss": 4.2454,
      "step": 2234
    },
    {
      "epoch": 0.02235,
      "grad_norm": 0.5125878831412668,
      "learning_rate": 0.003,
      "loss": 4.2654,
      "step": 2235
    },
    {
      "epoch": 0.02236,
      "grad_norm": 0.7208588554620924,
      "learning_rate": 0.003,
      "loss": 4.2213,
      "step": 2236
    },
    {
      "epoch": 0.02237,
      "grad_norm": 0.7459329466933863,
      "learning_rate": 0.003,
      "loss": 4.2569,
      "step": 2237
    },
    {
      "epoch": 0.02238,
      "grad_norm": 0.7047944961318278,
      "learning_rate": 0.003,
      "loss": 4.2359,
      "step": 2238
    },
    {
      "epoch": 0.02239,
      "grad_norm": 0.5989118371963864,
      "learning_rate": 0.003,
      "loss": 4.2173,
      "step": 2239
    },
    {
      "epoch": 0.0224,
      "grad_norm": 0.4980068818057585,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 2240
    },
    {
      "epoch": 0.02241,
      "grad_norm": 0.4765997364648245,
      "learning_rate": 0.003,
      "loss": 4.2278,
      "step": 2241
    },
    {
      "epoch": 0.02242,
      "grad_norm": 0.46816719473424767,
      "learning_rate": 0.003,
      "loss": 4.2111,
      "step": 2242
    },
    {
      "epoch": 0.02243,
      "grad_norm": 0.4356611881103323,
      "learning_rate": 0.003,
      "loss": 4.2383,
      "step": 2243
    },
    {
      "epoch": 0.02244,
      "grad_norm": 0.5122363302058777,
      "learning_rate": 0.003,
      "loss": 4.2149,
      "step": 2244
    },
    {
      "epoch": 0.02245,
      "grad_norm": 0.48883381472091325,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 2245
    },
    {
      "epoch": 0.02246,
      "grad_norm": 0.46925512406079145,
      "learning_rate": 0.003,
      "loss": 4.2303,
      "step": 2246
    },
    {
      "epoch": 0.02247,
      "grad_norm": 0.4305805801856768,
      "learning_rate": 0.003,
      "loss": 4.2025,
      "step": 2247
    },
    {
      "epoch": 0.02248,
      "grad_norm": 0.49466299772622346,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2248
    },
    {
      "epoch": 0.02249,
      "grad_norm": 0.5526228295993102,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 2249
    },
    {
      "epoch": 0.0225,
      "grad_norm": 0.5428701465777451,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 2250
    },
    {
      "epoch": 0.02251,
      "grad_norm": 0.5748138491222456,
      "learning_rate": 0.003,
      "loss": 4.242,
      "step": 2251
    },
    {
      "epoch": 0.02252,
      "grad_norm": 0.5627410118560184,
      "learning_rate": 0.003,
      "loss": 4.2308,
      "step": 2252
    },
    {
      "epoch": 0.02253,
      "grad_norm": 0.5464235773202091,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2253
    },
    {
      "epoch": 0.02254,
      "grad_norm": 0.5460056774038026,
      "learning_rate": 0.003,
      "loss": 4.2195,
      "step": 2254
    },
    {
      "epoch": 0.02255,
      "grad_norm": 0.5905579639478153,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2255
    },
    {
      "epoch": 0.02256,
      "grad_norm": 0.5996630986155933,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 2256
    },
    {
      "epoch": 0.02257,
      "grad_norm": 0.5786721112589904,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 2257
    },
    {
      "epoch": 0.02258,
      "grad_norm": 0.6117786281389065,
      "learning_rate": 0.003,
      "loss": 4.2219,
      "step": 2258
    },
    {
      "epoch": 0.02259,
      "grad_norm": 0.6324462778570742,
      "learning_rate": 0.003,
      "loss": 4.1884,
      "step": 2259
    },
    {
      "epoch": 0.0226,
      "grad_norm": 0.5884786495082438,
      "learning_rate": 0.003,
      "loss": 4.2409,
      "step": 2260
    },
    {
      "epoch": 0.02261,
      "grad_norm": 0.57968484048356,
      "learning_rate": 0.003,
      "loss": 4.2168,
      "step": 2261
    },
    {
      "epoch": 0.02262,
      "grad_norm": 0.6074577050747278,
      "learning_rate": 0.003,
      "loss": 4.2183,
      "step": 2262
    },
    {
      "epoch": 0.02263,
      "grad_norm": 0.7175518494233851,
      "learning_rate": 0.003,
      "loss": 4.237,
      "step": 2263
    },
    {
      "epoch": 0.02264,
      "grad_norm": 0.7379078827688124,
      "learning_rate": 0.003,
      "loss": 4.2337,
      "step": 2264
    },
    {
      "epoch": 0.02265,
      "grad_norm": 0.8347809834095073,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 2265
    },
    {
      "epoch": 0.02266,
      "grad_norm": 0.9870112085605878,
      "learning_rate": 0.003,
      "loss": 4.2201,
      "step": 2266
    },
    {
      "epoch": 0.02267,
      "grad_norm": 1.1653841189257785,
      "learning_rate": 0.003,
      "loss": 4.2605,
      "step": 2267
    },
    {
      "epoch": 0.02268,
      "grad_norm": 0.7766778764734964,
      "learning_rate": 0.003,
      "loss": 4.2198,
      "step": 2268
    },
    {
      "epoch": 0.02269,
      "grad_norm": 0.7319600539673422,
      "learning_rate": 0.003,
      "loss": 4.2435,
      "step": 2269
    },
    {
      "epoch": 0.0227,
      "grad_norm": 0.8988269964618911,
      "learning_rate": 0.003,
      "loss": 4.2441,
      "step": 2270
    },
    {
      "epoch": 0.02271,
      "grad_norm": 1.0408382709602264,
      "learning_rate": 0.003,
      "loss": 4.2637,
      "step": 2271
    },
    {
      "epoch": 0.02272,
      "grad_norm": 1.0101614929632108,
      "learning_rate": 0.003,
      "loss": 4.2626,
      "step": 2272
    },
    {
      "epoch": 0.02273,
      "grad_norm": 0.8989352397704572,
      "learning_rate": 0.003,
      "loss": 4.2643,
      "step": 2273
    },
    {
      "epoch": 0.02274,
      "grad_norm": 0.68535867805162,
      "learning_rate": 0.003,
      "loss": 4.2412,
      "step": 2274
    },
    {
      "epoch": 0.02275,
      "grad_norm": 0.6683944403644445,
      "learning_rate": 0.003,
      "loss": 4.2325,
      "step": 2275
    },
    {
      "epoch": 0.02276,
      "grad_norm": 0.6494300645141144,
      "learning_rate": 0.003,
      "loss": 4.2263,
      "step": 2276
    },
    {
      "epoch": 0.02277,
      "grad_norm": 0.6973670182968851,
      "learning_rate": 0.003,
      "loss": 4.2382,
      "step": 2277
    },
    {
      "epoch": 0.02278,
      "grad_norm": 0.7902320444222205,
      "learning_rate": 0.003,
      "loss": 4.2463,
      "step": 2278
    },
    {
      "epoch": 0.02279,
      "grad_norm": 0.8146336114310585,
      "learning_rate": 0.003,
      "loss": 4.2598,
      "step": 2279
    },
    {
      "epoch": 0.0228,
      "grad_norm": 0.8796339021642547,
      "learning_rate": 0.003,
      "loss": 4.2507,
      "step": 2280
    },
    {
      "epoch": 0.02281,
      "grad_norm": 0.8184015106359132,
      "learning_rate": 0.003,
      "loss": 4.2484,
      "step": 2281
    },
    {
      "epoch": 0.02282,
      "grad_norm": 0.6892747359251186,
      "learning_rate": 0.003,
      "loss": 4.2524,
      "step": 2282
    },
    {
      "epoch": 0.02283,
      "grad_norm": 0.6565456722535347,
      "learning_rate": 0.003,
      "loss": 4.2659,
      "step": 2283
    },
    {
      "epoch": 0.02284,
      "grad_norm": 0.5726981092319336,
      "learning_rate": 0.003,
      "loss": 4.2571,
      "step": 2284
    },
    {
      "epoch": 0.02285,
      "grad_norm": 0.5302500755986935,
      "learning_rate": 0.003,
      "loss": 4.1961,
      "step": 2285
    },
    {
      "epoch": 0.02286,
      "grad_norm": 0.5726425860790162,
      "learning_rate": 0.003,
      "loss": 4.2575,
      "step": 2286
    },
    {
      "epoch": 0.02287,
      "grad_norm": 0.5919298156896359,
      "learning_rate": 0.003,
      "loss": 4.2367,
      "step": 2287
    },
    {
      "epoch": 0.02288,
      "grad_norm": 0.5989015507067659,
      "learning_rate": 0.003,
      "loss": 4.2599,
      "step": 2288
    },
    {
      "epoch": 0.02289,
      "grad_norm": 0.6577684390532245,
      "learning_rate": 0.003,
      "loss": 4.2586,
      "step": 2289
    },
    {
      "epoch": 0.0229,
      "grad_norm": 0.8096348684056334,
      "learning_rate": 0.003,
      "loss": 4.2424,
      "step": 2290
    },
    {
      "epoch": 0.02291,
      "grad_norm": 0.8395348686742803,
      "learning_rate": 0.003,
      "loss": 4.2207,
      "step": 2291
    },
    {
      "epoch": 0.02292,
      "grad_norm": 0.6824435433690341,
      "learning_rate": 0.003,
      "loss": 4.2133,
      "step": 2292
    },
    {
      "epoch": 0.02293,
      "grad_norm": 0.5182235922586542,
      "learning_rate": 0.003,
      "loss": 4.2116,
      "step": 2293
    },
    {
      "epoch": 0.02294,
      "grad_norm": 0.5893445423849215,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2294
    },
    {
      "epoch": 0.02295,
      "grad_norm": 0.5909258163581438,
      "learning_rate": 0.003,
      "loss": 4.2083,
      "step": 2295
    },
    {
      "epoch": 0.02296,
      "grad_norm": 0.49380609648697416,
      "learning_rate": 0.003,
      "loss": 4.2381,
      "step": 2296
    },
    {
      "epoch": 0.02297,
      "grad_norm": 0.45509453612267814,
      "learning_rate": 0.003,
      "loss": 4.2063,
      "step": 2297
    },
    {
      "epoch": 0.02298,
      "grad_norm": 0.39795367258729464,
      "learning_rate": 0.003,
      "loss": 4.2154,
      "step": 2298
    },
    {
      "epoch": 0.02299,
      "grad_norm": 0.41864078809034155,
      "learning_rate": 0.003,
      "loss": 4.2206,
      "step": 2299
    },
    {
      "epoch": 0.023,
      "grad_norm": 0.45486028161368913,
      "learning_rate": 0.003,
      "loss": 4.2385,
      "step": 2300
    },
    {
      "epoch": 0.02301,
      "grad_norm": 0.4383928009279485,
      "learning_rate": 0.003,
      "loss": 4.2022,
      "step": 2301
    },
    {
      "epoch": 0.02302,
      "grad_norm": 0.4499222406460722,
      "learning_rate": 0.003,
      "loss": 4.2089,
      "step": 2302
    },
    {
      "epoch": 0.02303,
      "grad_norm": 0.4500683500939326,
      "learning_rate": 0.003,
      "loss": 4.1978,
      "step": 2303
    },
    {
      "epoch": 0.02304,
      "grad_norm": 0.4684736001381324,
      "learning_rate": 0.003,
      "loss": 4.2294,
      "step": 2304
    },
    {
      "epoch": 0.02305,
      "grad_norm": 0.4868306442044598,
      "learning_rate": 0.003,
      "loss": 4.2051,
      "step": 2305
    },
    {
      "epoch": 0.02306,
      "grad_norm": 0.5334764519084825,
      "learning_rate": 0.003,
      "loss": 4.1822,
      "step": 2306
    },
    {
      "epoch": 0.02307,
      "grad_norm": 0.6492467780995748,
      "learning_rate": 0.003,
      "loss": 4.2166,
      "step": 2307
    },
    {
      "epoch": 0.02308,
      "grad_norm": 0.7095262909438931,
      "learning_rate": 0.003,
      "loss": 4.2125,
      "step": 2308
    },
    {
      "epoch": 0.02309,
      "grad_norm": 0.54863336292105,
      "learning_rate": 0.003,
      "loss": 4.2349,
      "step": 2309
    },
    {
      "epoch": 0.0231,
      "grad_norm": 0.5340829751385735,
      "learning_rate": 0.003,
      "loss": 4.2136,
      "step": 2310
    },
    {
      "epoch": 0.02311,
      "grad_norm": 0.6450529599308674,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2311
    },
    {
      "epoch": 0.02312,
      "grad_norm": 0.820530552930021,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 2312
    },
    {
      "epoch": 0.02313,
      "grad_norm": 0.8724275505071452,
      "learning_rate": 0.003,
      "loss": 4.2036,
      "step": 2313
    },
    {
      "epoch": 0.02314,
      "grad_norm": 0.7571797533207066,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 2314
    },
    {
      "epoch": 0.02315,
      "grad_norm": 0.6953439594266967,
      "learning_rate": 0.003,
      "loss": 4.221,
      "step": 2315
    },
    {
      "epoch": 0.02316,
      "grad_norm": 0.7680069713239076,
      "learning_rate": 0.003,
      "loss": 4.2379,
      "step": 2316
    },
    {
      "epoch": 0.02317,
      "grad_norm": 0.8161604949475788,
      "learning_rate": 0.003,
      "loss": 4.223,
      "step": 2317
    },
    {
      "epoch": 0.02318,
      "grad_norm": 0.7516223217475189,
      "learning_rate": 0.003,
      "loss": 4.2136,
      "step": 2318
    },
    {
      "epoch": 0.02319,
      "grad_norm": 0.7000350308630269,
      "learning_rate": 0.003,
      "loss": 4.2358,
      "step": 2319
    },
    {
      "epoch": 0.0232,
      "grad_norm": 0.7134745769645445,
      "learning_rate": 0.003,
      "loss": 4.2452,
      "step": 2320
    },
    {
      "epoch": 0.02321,
      "grad_norm": 0.771931890706088,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2321
    },
    {
      "epoch": 0.02322,
      "grad_norm": 0.9854722178210035,
      "learning_rate": 0.003,
      "loss": 4.2409,
      "step": 2322
    },
    {
      "epoch": 0.02323,
      "grad_norm": 0.9767405875240949,
      "learning_rate": 0.003,
      "loss": 4.2246,
      "step": 2323
    },
    {
      "epoch": 0.02324,
      "grad_norm": 0.8048742564770486,
      "learning_rate": 0.003,
      "loss": 4.2207,
      "step": 2324
    },
    {
      "epoch": 0.02325,
      "grad_norm": 0.7747178630515028,
      "learning_rate": 0.003,
      "loss": 4.242,
      "step": 2325
    },
    {
      "epoch": 0.02326,
      "grad_norm": 0.7969057026631506,
      "learning_rate": 0.003,
      "loss": 4.2598,
      "step": 2326
    },
    {
      "epoch": 0.02327,
      "grad_norm": 0.8966295433578796,
      "learning_rate": 0.003,
      "loss": 4.2563,
      "step": 2327
    },
    {
      "epoch": 0.02328,
      "grad_norm": 0.7936452621630464,
      "learning_rate": 0.003,
      "loss": 4.2524,
      "step": 2328
    },
    {
      "epoch": 0.02329,
      "grad_norm": 0.6876802756045969,
      "learning_rate": 0.003,
      "loss": 4.2386,
      "step": 2329
    },
    {
      "epoch": 0.0233,
      "grad_norm": 0.7923806233690935,
      "learning_rate": 0.003,
      "loss": 4.2331,
      "step": 2330
    },
    {
      "epoch": 0.02331,
      "grad_norm": 0.8629660125432402,
      "learning_rate": 0.003,
      "loss": 4.278,
      "step": 2331
    },
    {
      "epoch": 0.02332,
      "grad_norm": 0.8224989442560882,
      "learning_rate": 0.003,
      "loss": 4.2382,
      "step": 2332
    },
    {
      "epoch": 0.02333,
      "grad_norm": 0.7543416736322919,
      "learning_rate": 0.003,
      "loss": 4.2433,
      "step": 2333
    },
    {
      "epoch": 0.02334,
      "grad_norm": 0.6971129018109938,
      "learning_rate": 0.003,
      "loss": 4.209,
      "step": 2334
    },
    {
      "epoch": 0.02335,
      "grad_norm": 0.7296913197019357,
      "learning_rate": 0.003,
      "loss": 4.2519,
      "step": 2335
    },
    {
      "epoch": 0.02336,
      "grad_norm": 0.7163909054433683,
      "learning_rate": 0.003,
      "loss": 4.2383,
      "step": 2336
    },
    {
      "epoch": 0.02337,
      "grad_norm": 0.6354188850134015,
      "learning_rate": 0.003,
      "loss": 4.2384,
      "step": 2337
    },
    {
      "epoch": 0.02338,
      "grad_norm": 0.7380334344563654,
      "learning_rate": 0.003,
      "loss": 4.2287,
      "step": 2338
    },
    {
      "epoch": 0.02339,
      "grad_norm": 0.6731414857527438,
      "learning_rate": 0.003,
      "loss": 4.2221,
      "step": 2339
    },
    {
      "epoch": 0.0234,
      "grad_norm": 0.5402269466528953,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2340
    },
    {
      "epoch": 0.02341,
      "grad_norm": 0.442581637819893,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 2341
    },
    {
      "epoch": 0.02342,
      "grad_norm": 0.37485389657180546,
      "learning_rate": 0.003,
      "loss": 4.2138,
      "step": 2342
    },
    {
      "epoch": 0.02343,
      "grad_norm": 0.3841947544396341,
      "learning_rate": 0.003,
      "loss": 4.213,
      "step": 2343
    },
    {
      "epoch": 0.02344,
      "grad_norm": 0.37024132956414024,
      "learning_rate": 0.003,
      "loss": 4.2104,
      "step": 2344
    },
    {
      "epoch": 0.02345,
      "grad_norm": 0.3532869549675326,
      "learning_rate": 0.003,
      "loss": 4.2035,
      "step": 2345
    },
    {
      "epoch": 0.02346,
      "grad_norm": 0.3470711196570803,
      "learning_rate": 0.003,
      "loss": 4.1998,
      "step": 2346
    },
    {
      "epoch": 0.02347,
      "grad_norm": 0.4249636779242577,
      "learning_rate": 0.003,
      "loss": 4.2374,
      "step": 2347
    },
    {
      "epoch": 0.02348,
      "grad_norm": 0.5336097138298947,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 2348
    },
    {
      "epoch": 0.02349,
      "grad_norm": 0.7560598476745078,
      "learning_rate": 0.003,
      "loss": 4.2252,
      "step": 2349
    },
    {
      "epoch": 0.0235,
      "grad_norm": 0.9366290665402001,
      "learning_rate": 0.003,
      "loss": 4.2316,
      "step": 2350
    },
    {
      "epoch": 0.02351,
      "grad_norm": 0.8241230319323669,
      "learning_rate": 0.003,
      "loss": 4.2392,
      "step": 2351
    },
    {
      "epoch": 0.02352,
      "grad_norm": 0.4838312119944649,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 2352
    },
    {
      "epoch": 0.02353,
      "grad_norm": 0.5962852237534145,
      "learning_rate": 0.003,
      "loss": 4.2218,
      "step": 2353
    },
    {
      "epoch": 0.02354,
      "grad_norm": 0.6656896770625591,
      "learning_rate": 0.003,
      "loss": 4.2262,
      "step": 2354
    },
    {
      "epoch": 0.02355,
      "grad_norm": 0.5663365054891429,
      "learning_rate": 0.003,
      "loss": 4.2216,
      "step": 2355
    },
    {
      "epoch": 0.02356,
      "grad_norm": 0.4266142835453117,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 2356
    },
    {
      "epoch": 0.02357,
      "grad_norm": 0.4433400684484211,
      "learning_rate": 0.003,
      "loss": 4.1998,
      "step": 2357
    },
    {
      "epoch": 0.02358,
      "grad_norm": 0.5052976773226068,
      "learning_rate": 0.003,
      "loss": 4.2226,
      "step": 2358
    },
    {
      "epoch": 0.02359,
      "grad_norm": 0.45316233898993824,
      "learning_rate": 0.003,
      "loss": 4.1954,
      "step": 2359
    },
    {
      "epoch": 0.0236,
      "grad_norm": 0.4014671488900521,
      "learning_rate": 0.003,
      "loss": 4.2325,
      "step": 2360
    },
    {
      "epoch": 0.02361,
      "grad_norm": 0.36698656753000536,
      "learning_rate": 0.003,
      "loss": 4.212,
      "step": 2361
    },
    {
      "epoch": 0.02362,
      "grad_norm": 0.42314800926533436,
      "learning_rate": 0.003,
      "loss": 4.2053,
      "step": 2362
    },
    {
      "epoch": 0.02363,
      "grad_norm": 0.4217103245805847,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 2363
    },
    {
      "epoch": 0.02364,
      "grad_norm": 0.4210664479788169,
      "learning_rate": 0.003,
      "loss": 4.2013,
      "step": 2364
    },
    {
      "epoch": 0.02365,
      "grad_norm": 0.4933666084308532,
      "learning_rate": 0.003,
      "loss": 4.2208,
      "step": 2365
    },
    {
      "epoch": 0.02366,
      "grad_norm": 0.5419476181345154,
      "learning_rate": 0.003,
      "loss": 4.1974,
      "step": 2366
    },
    {
      "epoch": 0.02367,
      "grad_norm": 0.5315884871641553,
      "learning_rate": 0.003,
      "loss": 4.2086,
      "step": 2367
    },
    {
      "epoch": 0.02368,
      "grad_norm": 0.5513031505324636,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 2368
    },
    {
      "epoch": 0.02369,
      "grad_norm": 0.6237648160785263,
      "learning_rate": 0.003,
      "loss": 4.1978,
      "step": 2369
    },
    {
      "epoch": 0.0237,
      "grad_norm": 0.7305017234045693,
      "learning_rate": 0.003,
      "loss": 4.2311,
      "step": 2370
    },
    {
      "epoch": 0.02371,
      "grad_norm": 0.7544509646548179,
      "learning_rate": 0.003,
      "loss": 4.2181,
      "step": 2371
    },
    {
      "epoch": 0.02372,
      "grad_norm": 0.8927671954437898,
      "learning_rate": 0.003,
      "loss": 4.2125,
      "step": 2372
    },
    {
      "epoch": 0.02373,
      "grad_norm": 0.8832047313566855,
      "learning_rate": 0.003,
      "loss": 4.2061,
      "step": 2373
    },
    {
      "epoch": 0.02374,
      "grad_norm": 0.9298955170309446,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2374
    },
    {
      "epoch": 0.02375,
      "grad_norm": 0.8640645189121525,
      "learning_rate": 0.003,
      "loss": 4.2188,
      "step": 2375
    },
    {
      "epoch": 0.02376,
      "grad_norm": 0.8134531891441159,
      "learning_rate": 0.003,
      "loss": 4.2414,
      "step": 2376
    },
    {
      "epoch": 0.02377,
      "grad_norm": 0.8309736476293886,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2377
    },
    {
      "epoch": 0.02378,
      "grad_norm": 0.8951590746635169,
      "learning_rate": 0.003,
      "loss": 4.2274,
      "step": 2378
    },
    {
      "epoch": 0.02379,
      "grad_norm": 0.8950538453176826,
      "learning_rate": 0.003,
      "loss": 4.2181,
      "step": 2379
    },
    {
      "epoch": 0.0238,
      "grad_norm": 0.8724035600141309,
      "learning_rate": 0.003,
      "loss": 4.1913,
      "step": 2380
    },
    {
      "epoch": 0.02381,
      "grad_norm": 0.7988879477893653,
      "learning_rate": 0.003,
      "loss": 4.2027,
      "step": 2381
    },
    {
      "epoch": 0.02382,
      "grad_norm": 0.7723567017862097,
      "learning_rate": 0.003,
      "loss": 4.2578,
      "step": 2382
    },
    {
      "epoch": 0.02383,
      "grad_norm": 0.671665890468045,
      "learning_rate": 0.003,
      "loss": 4.2171,
      "step": 2383
    },
    {
      "epoch": 0.02384,
      "grad_norm": 0.6389101019379912,
      "learning_rate": 0.003,
      "loss": 4.2056,
      "step": 2384
    },
    {
      "epoch": 0.02385,
      "grad_norm": 0.580627279569301,
      "learning_rate": 0.003,
      "loss": 4.2294,
      "step": 2385
    },
    {
      "epoch": 0.02386,
      "grad_norm": 0.6112201616439135,
      "learning_rate": 0.003,
      "loss": 4.2052,
      "step": 2386
    },
    {
      "epoch": 0.02387,
      "grad_norm": 0.609528097348241,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 2387
    },
    {
      "epoch": 0.02388,
      "grad_norm": 0.6686638362844006,
      "learning_rate": 0.003,
      "loss": 4.2564,
      "step": 2388
    },
    {
      "epoch": 0.02389,
      "grad_norm": 0.740840757442375,
      "learning_rate": 0.003,
      "loss": 4.2206,
      "step": 2389
    },
    {
      "epoch": 0.0239,
      "grad_norm": 0.795817868194079,
      "learning_rate": 0.003,
      "loss": 4.2173,
      "step": 2390
    },
    {
      "epoch": 0.02391,
      "grad_norm": 0.6885254297129437,
      "learning_rate": 0.003,
      "loss": 4.2366,
      "step": 2391
    },
    {
      "epoch": 0.02392,
      "grad_norm": 0.57208190101736,
      "learning_rate": 0.003,
      "loss": 4.2339,
      "step": 2392
    },
    {
      "epoch": 0.02393,
      "grad_norm": 0.6766423034488698,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 2393
    },
    {
      "epoch": 0.02394,
      "grad_norm": 0.7708917314903707,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 2394
    },
    {
      "epoch": 0.02395,
      "grad_norm": 0.8419139355655553,
      "learning_rate": 0.003,
      "loss": 4.234,
      "step": 2395
    },
    {
      "epoch": 0.02396,
      "grad_norm": 0.7491203962086143,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 2396
    },
    {
      "epoch": 0.02397,
      "grad_norm": 0.6046499284946664,
      "learning_rate": 0.003,
      "loss": 4.2249,
      "step": 2397
    },
    {
      "epoch": 0.02398,
      "grad_norm": 0.654682549696845,
      "learning_rate": 0.003,
      "loss": 4.2309,
      "step": 2398
    },
    {
      "epoch": 0.02399,
      "grad_norm": 0.600390647492162,
      "learning_rate": 0.003,
      "loss": 4.209,
      "step": 2399
    },
    {
      "epoch": 0.024,
      "grad_norm": 0.5155181636394677,
      "learning_rate": 0.003,
      "loss": 4.2198,
      "step": 2400
    },
    {
      "epoch": 0.02401,
      "grad_norm": 0.48968723675185105,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2401
    },
    {
      "epoch": 0.02402,
      "grad_norm": 0.5172416496163427,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2402
    },
    {
      "epoch": 0.02403,
      "grad_norm": 0.5647625943348626,
      "learning_rate": 0.003,
      "loss": 4.1949,
      "step": 2403
    },
    {
      "epoch": 0.02404,
      "grad_norm": 0.5032714506791921,
      "learning_rate": 0.003,
      "loss": 4.233,
      "step": 2404
    },
    {
      "epoch": 0.02405,
      "grad_norm": 0.4675987670630769,
      "learning_rate": 0.003,
      "loss": 4.2334,
      "step": 2405
    },
    {
      "epoch": 0.02406,
      "grad_norm": 0.5178645102430114,
      "learning_rate": 0.003,
      "loss": 4.2192,
      "step": 2406
    },
    {
      "epoch": 0.02407,
      "grad_norm": 0.5559303502201767,
      "learning_rate": 0.003,
      "loss": 4.1976,
      "step": 2407
    },
    {
      "epoch": 0.02408,
      "grad_norm": 0.551443134128045,
      "learning_rate": 0.003,
      "loss": 4.2064,
      "step": 2408
    },
    {
      "epoch": 0.02409,
      "grad_norm": 0.5057926630793051,
      "learning_rate": 0.003,
      "loss": 4.2079,
      "step": 2409
    },
    {
      "epoch": 0.0241,
      "grad_norm": 0.6021866868934532,
      "learning_rate": 0.003,
      "loss": 4.2249,
      "step": 2410
    },
    {
      "epoch": 0.02411,
      "grad_norm": 0.7613042549642509,
      "learning_rate": 0.003,
      "loss": 4.1979,
      "step": 2411
    },
    {
      "epoch": 0.02412,
      "grad_norm": 0.7823591519195993,
      "learning_rate": 0.003,
      "loss": 4.2054,
      "step": 2412
    },
    {
      "epoch": 0.02413,
      "grad_norm": 0.6762814430739601,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 2413
    },
    {
      "epoch": 0.02414,
      "grad_norm": 0.6515291647943621,
      "learning_rate": 0.003,
      "loss": 4.2148,
      "step": 2414
    },
    {
      "epoch": 0.02415,
      "grad_norm": 0.5905595191694923,
      "learning_rate": 0.003,
      "loss": 4.2125,
      "step": 2415
    },
    {
      "epoch": 0.02416,
      "grad_norm": 0.5267459284750431,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 2416
    },
    {
      "epoch": 0.02417,
      "grad_norm": 0.6612447088265828,
      "learning_rate": 0.003,
      "loss": 4.2248,
      "step": 2417
    },
    {
      "epoch": 0.02418,
      "grad_norm": 0.7551414883340356,
      "learning_rate": 0.003,
      "loss": 4.2234,
      "step": 2418
    },
    {
      "epoch": 0.02419,
      "grad_norm": 0.8205697648408953,
      "learning_rate": 0.003,
      "loss": 4.2244,
      "step": 2419
    },
    {
      "epoch": 0.0242,
      "grad_norm": 0.8899430556748202,
      "learning_rate": 0.003,
      "loss": 4.2436,
      "step": 2420
    },
    {
      "epoch": 0.02421,
      "grad_norm": 0.7830186936394068,
      "learning_rate": 0.003,
      "loss": 4.2007,
      "step": 2421
    },
    {
      "epoch": 0.02422,
      "grad_norm": 0.707149933198692,
      "learning_rate": 0.003,
      "loss": 4.2528,
      "step": 2422
    },
    {
      "epoch": 0.02423,
      "grad_norm": 0.6635737571843986,
      "learning_rate": 0.003,
      "loss": 4.2424,
      "step": 2423
    },
    {
      "epoch": 0.02424,
      "grad_norm": 0.6518995770283964,
      "learning_rate": 0.003,
      "loss": 4.2374,
      "step": 2424
    },
    {
      "epoch": 0.02425,
      "grad_norm": 0.6795937334698865,
      "learning_rate": 0.003,
      "loss": 4.2233,
      "step": 2425
    },
    {
      "epoch": 0.02426,
      "grad_norm": 0.7581646116983769,
      "learning_rate": 0.003,
      "loss": 4.2315,
      "step": 2426
    },
    {
      "epoch": 0.02427,
      "grad_norm": 0.9154720690249456,
      "learning_rate": 0.003,
      "loss": 4.2647,
      "step": 2427
    },
    {
      "epoch": 0.02428,
      "grad_norm": 0.9816907611239863,
      "learning_rate": 0.003,
      "loss": 4.2433,
      "step": 2428
    },
    {
      "epoch": 0.02429,
      "grad_norm": 0.9289957548466957,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 2429
    },
    {
      "epoch": 0.0243,
      "grad_norm": 0.6787755121793266,
      "learning_rate": 0.003,
      "loss": 4.2464,
      "step": 2430
    },
    {
      "epoch": 0.02431,
      "grad_norm": 0.7835751457530344,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2431
    },
    {
      "epoch": 0.02432,
      "grad_norm": 0.7857412874952409,
      "learning_rate": 0.003,
      "loss": 4.2512,
      "step": 2432
    },
    {
      "epoch": 0.02433,
      "grad_norm": 0.7902925766285801,
      "learning_rate": 0.003,
      "loss": 4.263,
      "step": 2433
    },
    {
      "epoch": 0.02434,
      "grad_norm": 0.6888559592034768,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2434
    },
    {
      "epoch": 0.02435,
      "grad_norm": 0.7021573732138202,
      "learning_rate": 0.003,
      "loss": 4.2272,
      "step": 2435
    },
    {
      "epoch": 0.02436,
      "grad_norm": 0.7401895262850188,
      "learning_rate": 0.003,
      "loss": 4.235,
      "step": 2436
    },
    {
      "epoch": 0.02437,
      "grad_norm": 0.6162969571454292,
      "learning_rate": 0.003,
      "loss": 4.2159,
      "step": 2437
    },
    {
      "epoch": 0.02438,
      "grad_norm": 0.5086196755044915,
      "learning_rate": 0.003,
      "loss": 4.2268,
      "step": 2438
    },
    {
      "epoch": 0.02439,
      "grad_norm": 0.47970313870881265,
      "learning_rate": 0.003,
      "loss": 4.1957,
      "step": 2439
    },
    {
      "epoch": 0.0244,
      "grad_norm": 0.4288371919581958,
      "learning_rate": 0.003,
      "loss": 4.2421,
      "step": 2440
    },
    {
      "epoch": 0.02441,
      "grad_norm": 0.38882426885049926,
      "learning_rate": 0.003,
      "loss": 4.2052,
      "step": 2441
    },
    {
      "epoch": 0.02442,
      "grad_norm": 0.4241944568677858,
      "learning_rate": 0.003,
      "loss": 4.2386,
      "step": 2442
    },
    {
      "epoch": 0.02443,
      "grad_norm": 0.4820755060057781,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 2443
    },
    {
      "epoch": 0.02444,
      "grad_norm": 0.620820199570667,
      "learning_rate": 0.003,
      "loss": 4.2049,
      "step": 2444
    },
    {
      "epoch": 0.02445,
      "grad_norm": 0.8695985731954037,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2445
    },
    {
      "epoch": 0.02446,
      "grad_norm": 0.93656467702838,
      "learning_rate": 0.003,
      "loss": 4.2301,
      "step": 2446
    },
    {
      "epoch": 0.02447,
      "grad_norm": 0.7193216075423335,
      "learning_rate": 0.003,
      "loss": 4.2297,
      "step": 2447
    },
    {
      "epoch": 0.02448,
      "grad_norm": 0.77368716084308,
      "learning_rate": 0.003,
      "loss": 4.2355,
      "step": 2448
    },
    {
      "epoch": 0.02449,
      "grad_norm": 0.93482214454163,
      "learning_rate": 0.003,
      "loss": 4.2378,
      "step": 2449
    },
    {
      "epoch": 0.0245,
      "grad_norm": 0.7890224806153067,
      "learning_rate": 0.003,
      "loss": 4.2333,
      "step": 2450
    },
    {
      "epoch": 0.02451,
      "grad_norm": 0.6715615912076842,
      "learning_rate": 0.003,
      "loss": 4.2085,
      "step": 2451
    },
    {
      "epoch": 0.02452,
      "grad_norm": 0.6401183815147788,
      "learning_rate": 0.003,
      "loss": 4.2138,
      "step": 2452
    },
    {
      "epoch": 0.02453,
      "grad_norm": 0.5995798001531515,
      "learning_rate": 0.003,
      "loss": 4.1908,
      "step": 2453
    },
    {
      "epoch": 0.02454,
      "grad_norm": 0.5267826981306184,
      "learning_rate": 0.003,
      "loss": 4.1933,
      "step": 2454
    },
    {
      "epoch": 0.02455,
      "grad_norm": 0.4821819525696827,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 2455
    },
    {
      "epoch": 0.02456,
      "grad_norm": 0.5089117196246842,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 2456
    },
    {
      "epoch": 0.02457,
      "grad_norm": 0.5510451022091881,
      "learning_rate": 0.003,
      "loss": 4.2212,
      "step": 2457
    },
    {
      "epoch": 0.02458,
      "grad_norm": 0.48930227370319684,
      "learning_rate": 0.003,
      "loss": 4.1938,
      "step": 2458
    },
    {
      "epoch": 0.02459,
      "grad_norm": 0.4331220174779214,
      "learning_rate": 0.003,
      "loss": 4.2193,
      "step": 2459
    },
    {
      "epoch": 0.0246,
      "grad_norm": 0.41158812351788215,
      "learning_rate": 0.003,
      "loss": 4.2215,
      "step": 2460
    },
    {
      "epoch": 0.02461,
      "grad_norm": 0.39108806608008795,
      "learning_rate": 0.003,
      "loss": 4.1964,
      "step": 2461
    },
    {
      "epoch": 0.02462,
      "grad_norm": 0.3754314162918144,
      "learning_rate": 0.003,
      "loss": 4.2216,
      "step": 2462
    },
    {
      "epoch": 0.02463,
      "grad_norm": 0.4122698662760153,
      "learning_rate": 0.003,
      "loss": 4.2201,
      "step": 2463
    },
    {
      "epoch": 0.02464,
      "grad_norm": 0.4350495849540257,
      "learning_rate": 0.003,
      "loss": 4.1992,
      "step": 2464
    },
    {
      "epoch": 0.02465,
      "grad_norm": 0.49445669847439655,
      "learning_rate": 0.003,
      "loss": 4.2175,
      "step": 2465
    },
    {
      "epoch": 0.02466,
      "grad_norm": 0.5818409584444987,
      "learning_rate": 0.003,
      "loss": 4.1996,
      "step": 2466
    },
    {
      "epoch": 0.02467,
      "grad_norm": 0.6235372640193676,
      "learning_rate": 0.003,
      "loss": 4.1803,
      "step": 2467
    },
    {
      "epoch": 0.02468,
      "grad_norm": 0.5927407776473445,
      "learning_rate": 0.003,
      "loss": 4.223,
      "step": 2468
    },
    {
      "epoch": 0.02469,
      "grad_norm": 0.6223619407097589,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 2469
    },
    {
      "epoch": 0.0247,
      "grad_norm": 0.583979309128103,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2470
    },
    {
      "epoch": 0.02471,
      "grad_norm": 0.5491918199641306,
      "learning_rate": 0.003,
      "loss": 4.1836,
      "step": 2471
    },
    {
      "epoch": 0.02472,
      "grad_norm": 0.635376629969627,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 2472
    },
    {
      "epoch": 0.02473,
      "grad_norm": 0.9426912757457121,
      "learning_rate": 0.003,
      "loss": 4.1862,
      "step": 2473
    },
    {
      "epoch": 0.02474,
      "grad_norm": 1.1494984995182613,
      "learning_rate": 0.003,
      "loss": 4.2384,
      "step": 2474
    },
    {
      "epoch": 0.02475,
      "grad_norm": 0.6832524660748021,
      "learning_rate": 0.003,
      "loss": 4.1958,
      "step": 2475
    },
    {
      "epoch": 0.02476,
      "grad_norm": 0.6659288683302277,
      "learning_rate": 0.003,
      "loss": 4.2064,
      "step": 2476
    },
    {
      "epoch": 0.02477,
      "grad_norm": 0.7078551330691552,
      "learning_rate": 0.003,
      "loss": 4.229,
      "step": 2477
    },
    {
      "epoch": 0.02478,
      "grad_norm": 0.682879510763541,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2478
    },
    {
      "epoch": 0.02479,
      "grad_norm": 0.6453701850218885,
      "learning_rate": 0.003,
      "loss": 4.1922,
      "step": 2479
    },
    {
      "epoch": 0.0248,
      "grad_norm": 0.6214835257281583,
      "learning_rate": 0.003,
      "loss": 4.2323,
      "step": 2480
    },
    {
      "epoch": 0.02481,
      "grad_norm": 0.5814705370549468,
      "learning_rate": 0.003,
      "loss": 4.2104,
      "step": 2481
    },
    {
      "epoch": 0.02482,
      "grad_norm": 0.5717194029886628,
      "learning_rate": 0.003,
      "loss": 4.2206,
      "step": 2482
    },
    {
      "epoch": 0.02483,
      "grad_norm": 0.5507096505833119,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 2483
    },
    {
      "epoch": 0.02484,
      "grad_norm": 0.48818940634137703,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2484
    },
    {
      "epoch": 0.02485,
      "grad_norm": 0.5078768442645458,
      "learning_rate": 0.003,
      "loss": 4.2218,
      "step": 2485
    },
    {
      "epoch": 0.02486,
      "grad_norm": 0.6158842388358957,
      "learning_rate": 0.003,
      "loss": 4.2483,
      "step": 2486
    },
    {
      "epoch": 0.02487,
      "grad_norm": 0.7113530958465855,
      "learning_rate": 0.003,
      "loss": 4.2057,
      "step": 2487
    },
    {
      "epoch": 0.02488,
      "grad_norm": 0.8543927391668017,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 2488
    },
    {
      "epoch": 0.02489,
      "grad_norm": 0.9446890974964458,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2489
    },
    {
      "epoch": 0.0249,
      "grad_norm": 0.7778085412828454,
      "learning_rate": 0.003,
      "loss": 4.1871,
      "step": 2490
    },
    {
      "epoch": 0.02491,
      "grad_norm": 0.6481694251808287,
      "learning_rate": 0.003,
      "loss": 4.2111,
      "step": 2491
    },
    {
      "epoch": 0.02492,
      "grad_norm": 0.6221254984926949,
      "learning_rate": 0.003,
      "loss": 4.2219,
      "step": 2492
    },
    {
      "epoch": 0.02493,
      "grad_norm": 0.6659856187784612,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 2493
    },
    {
      "epoch": 0.02494,
      "grad_norm": 0.6928875164931092,
      "learning_rate": 0.003,
      "loss": 4.206,
      "step": 2494
    },
    {
      "epoch": 0.02495,
      "grad_norm": 0.6666259038056425,
      "learning_rate": 0.003,
      "loss": 4.2171,
      "step": 2495
    },
    {
      "epoch": 0.02496,
      "grad_norm": 0.6388189581951018,
      "learning_rate": 0.003,
      "loss": 4.2077,
      "step": 2496
    },
    {
      "epoch": 0.02497,
      "grad_norm": 0.6011544202060304,
      "learning_rate": 0.003,
      "loss": 4.2109,
      "step": 2497
    },
    {
      "epoch": 0.02498,
      "grad_norm": 0.6445306947945534,
      "learning_rate": 0.003,
      "loss": 4.1989,
      "step": 2498
    },
    {
      "epoch": 0.02499,
      "grad_norm": 0.6336412786289282,
      "learning_rate": 0.003,
      "loss": 4.2101,
      "step": 2499
    },
    {
      "epoch": 0.025,
      "grad_norm": 0.7751683320425504,
      "learning_rate": 0.003,
      "loss": 4.1983,
      "step": 2500
    },
    {
      "epoch": 0.02501,
      "grad_norm": 0.8226642072050464,
      "learning_rate": 0.003,
      "loss": 4.2348,
      "step": 2501
    },
    {
      "epoch": 0.02502,
      "grad_norm": 0.7525954653781657,
      "learning_rate": 0.003,
      "loss": 4.2151,
      "step": 2502
    },
    {
      "epoch": 0.02503,
      "grad_norm": 0.6443708527662554,
      "learning_rate": 0.003,
      "loss": 4.22,
      "step": 2503
    },
    {
      "epoch": 0.02504,
      "grad_norm": 0.7431879561442126,
      "learning_rate": 0.003,
      "loss": 4.2035,
      "step": 2504
    },
    {
      "epoch": 0.02505,
      "grad_norm": 0.8444858068379042,
      "learning_rate": 0.003,
      "loss": 4.206,
      "step": 2505
    },
    {
      "epoch": 0.02506,
      "grad_norm": 0.8357688056745337,
      "learning_rate": 0.003,
      "loss": 4.2242,
      "step": 2506
    },
    {
      "epoch": 0.02507,
      "grad_norm": 0.8015204150666452,
      "learning_rate": 0.003,
      "loss": 4.21,
      "step": 2507
    },
    {
      "epoch": 0.02508,
      "grad_norm": 0.7988221880335916,
      "learning_rate": 0.003,
      "loss": 4.2096,
      "step": 2508
    },
    {
      "epoch": 0.02509,
      "grad_norm": 0.8648254923532768,
      "learning_rate": 0.003,
      "loss": 4.2197,
      "step": 2509
    },
    {
      "epoch": 0.0251,
      "grad_norm": 0.9052649096369385,
      "learning_rate": 0.003,
      "loss": 4.2352,
      "step": 2510
    },
    {
      "epoch": 0.02511,
      "grad_norm": 0.8287055915677601,
      "learning_rate": 0.003,
      "loss": 4.243,
      "step": 2511
    },
    {
      "epoch": 0.02512,
      "grad_norm": 0.6322731338734815,
      "learning_rate": 0.003,
      "loss": 4.2191,
      "step": 2512
    },
    {
      "epoch": 0.02513,
      "grad_norm": 0.6027563249296022,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 2513
    },
    {
      "epoch": 0.02514,
      "grad_norm": 0.5639092637703635,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2514
    },
    {
      "epoch": 0.02515,
      "grad_norm": 0.5109276454863091,
      "learning_rate": 0.003,
      "loss": 4.1846,
      "step": 2515
    },
    {
      "epoch": 0.02516,
      "grad_norm": 0.49613462202561587,
      "learning_rate": 0.003,
      "loss": 4.2057,
      "step": 2516
    },
    {
      "epoch": 0.02517,
      "grad_norm": 0.4638933308319014,
      "learning_rate": 0.003,
      "loss": 4.2043,
      "step": 2517
    },
    {
      "epoch": 0.02518,
      "grad_norm": 0.47895707517955144,
      "learning_rate": 0.003,
      "loss": 4.2156,
      "step": 2518
    },
    {
      "epoch": 0.02519,
      "grad_norm": 0.540563379395588,
      "learning_rate": 0.003,
      "loss": 4.191,
      "step": 2519
    },
    {
      "epoch": 0.0252,
      "grad_norm": 0.593588962667406,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2520
    },
    {
      "epoch": 0.02521,
      "grad_norm": 0.5859708147834043,
      "learning_rate": 0.003,
      "loss": 4.2007,
      "step": 2521
    },
    {
      "epoch": 0.02522,
      "grad_norm": 0.6750984334319393,
      "learning_rate": 0.003,
      "loss": 4.1967,
      "step": 2522
    },
    {
      "epoch": 0.02523,
      "grad_norm": 0.8294919931793535,
      "learning_rate": 0.003,
      "loss": 4.2041,
      "step": 2523
    },
    {
      "epoch": 0.02524,
      "grad_norm": 0.8897442405025344,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2524
    },
    {
      "epoch": 0.02525,
      "grad_norm": 0.8626054901448138,
      "learning_rate": 0.003,
      "loss": 4.2361,
      "step": 2525
    },
    {
      "epoch": 0.02526,
      "grad_norm": 0.6609622839443053,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 2526
    },
    {
      "epoch": 0.02527,
      "grad_norm": 0.5788832246003424,
      "learning_rate": 0.003,
      "loss": 4.2082,
      "step": 2527
    },
    {
      "epoch": 0.02528,
      "grad_norm": 0.6673373775723666,
      "learning_rate": 0.003,
      "loss": 4.2334,
      "step": 2528
    },
    {
      "epoch": 0.02529,
      "grad_norm": 0.6949939363717212,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 2529
    },
    {
      "epoch": 0.0253,
      "grad_norm": 0.6926157434556843,
      "learning_rate": 0.003,
      "loss": 4.2244,
      "step": 2530
    },
    {
      "epoch": 0.02531,
      "grad_norm": 0.6799840691277568,
      "learning_rate": 0.003,
      "loss": 4.2264,
      "step": 2531
    },
    {
      "epoch": 0.02532,
      "grad_norm": 0.8293024556709889,
      "learning_rate": 0.003,
      "loss": 4.1997,
      "step": 2532
    },
    {
      "epoch": 0.02533,
      "grad_norm": 0.9626324272072302,
      "learning_rate": 0.003,
      "loss": 4.2444,
      "step": 2533
    },
    {
      "epoch": 0.02534,
      "grad_norm": 0.9642180813694794,
      "learning_rate": 0.003,
      "loss": 4.2316,
      "step": 2534
    },
    {
      "epoch": 0.02535,
      "grad_norm": 0.8934366837294853,
      "learning_rate": 0.003,
      "loss": 4.2235,
      "step": 2535
    },
    {
      "epoch": 0.02536,
      "grad_norm": 0.8077546891906852,
      "learning_rate": 0.003,
      "loss": 4.2184,
      "step": 2536
    },
    {
      "epoch": 0.02537,
      "grad_norm": 0.6532019172502265,
      "learning_rate": 0.003,
      "loss": 4.2093,
      "step": 2537
    },
    {
      "epoch": 0.02538,
      "grad_norm": 0.6905511076952489,
      "learning_rate": 0.003,
      "loss": 4.184,
      "step": 2538
    },
    {
      "epoch": 0.02539,
      "grad_norm": 0.7499705806639466,
      "learning_rate": 0.003,
      "loss": 4.2378,
      "step": 2539
    },
    {
      "epoch": 0.0254,
      "grad_norm": 0.6787160415395472,
      "learning_rate": 0.003,
      "loss": 4.2251,
      "step": 2540
    },
    {
      "epoch": 0.02541,
      "grad_norm": 0.6164160513405081,
      "learning_rate": 0.003,
      "loss": 4.2157,
      "step": 2541
    },
    {
      "epoch": 0.02542,
      "grad_norm": 0.6175874501025589,
      "learning_rate": 0.003,
      "loss": 4.1908,
      "step": 2542
    },
    {
      "epoch": 0.02543,
      "grad_norm": 0.5640476735952517,
      "learning_rate": 0.003,
      "loss": 4.2185,
      "step": 2543
    },
    {
      "epoch": 0.02544,
      "grad_norm": 0.6088525590071204,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 2544
    },
    {
      "epoch": 0.02545,
      "grad_norm": 0.5774041309556538,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2545
    },
    {
      "epoch": 0.02546,
      "grad_norm": 0.5080999994690095,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 2546
    },
    {
      "epoch": 0.02547,
      "grad_norm": 0.491841642905056,
      "learning_rate": 0.003,
      "loss": 4.2303,
      "step": 2547
    },
    {
      "epoch": 0.02548,
      "grad_norm": 0.5507165108887961,
      "learning_rate": 0.003,
      "loss": 4.2283,
      "step": 2548
    },
    {
      "epoch": 0.02549,
      "grad_norm": 0.6923243100094933,
      "learning_rate": 0.003,
      "loss": 4.2227,
      "step": 2549
    },
    {
      "epoch": 0.0255,
      "grad_norm": 0.7817347310990386,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 2550
    },
    {
      "epoch": 0.02551,
      "grad_norm": 0.7447240542500363,
      "learning_rate": 0.003,
      "loss": 4.2189,
      "step": 2551
    },
    {
      "epoch": 0.02552,
      "grad_norm": 0.6307913070627245,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 2552
    },
    {
      "epoch": 0.02553,
      "grad_norm": 0.6438248127745729,
      "learning_rate": 0.003,
      "loss": 4.2512,
      "step": 2553
    },
    {
      "epoch": 0.02554,
      "grad_norm": 0.5917787379710305,
      "learning_rate": 0.003,
      "loss": 4.1983,
      "step": 2554
    },
    {
      "epoch": 0.02555,
      "grad_norm": 0.6080534723714437,
      "learning_rate": 0.003,
      "loss": 4.1896,
      "step": 2555
    },
    {
      "epoch": 0.02556,
      "grad_norm": 0.6279427379491616,
      "learning_rate": 0.003,
      "loss": 4.2002,
      "step": 2556
    },
    {
      "epoch": 0.02557,
      "grad_norm": 0.5766106602168388,
      "learning_rate": 0.003,
      "loss": 4.2029,
      "step": 2557
    },
    {
      "epoch": 0.02558,
      "grad_norm": 0.5614919439663087,
      "learning_rate": 0.003,
      "loss": 4.2176,
      "step": 2558
    },
    {
      "epoch": 0.02559,
      "grad_norm": 0.5870187050191585,
      "learning_rate": 0.003,
      "loss": 4.2144,
      "step": 2559
    },
    {
      "epoch": 0.0256,
      "grad_norm": 0.5866194992858084,
      "learning_rate": 0.003,
      "loss": 4.1883,
      "step": 2560
    },
    {
      "epoch": 0.02561,
      "grad_norm": 0.56223387782931,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 2561
    },
    {
      "epoch": 0.02562,
      "grad_norm": 0.4873971454432025,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 2562
    },
    {
      "epoch": 0.02563,
      "grad_norm": 0.6113475215270017,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 2563
    },
    {
      "epoch": 0.02564,
      "grad_norm": 0.7598159388308933,
      "learning_rate": 0.003,
      "loss": 4.2128,
      "step": 2564
    },
    {
      "epoch": 0.02565,
      "grad_norm": 0.9584592936312263,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 2565
    },
    {
      "epoch": 0.02566,
      "grad_norm": 1.059905791929432,
      "learning_rate": 0.003,
      "loss": 4.231,
      "step": 2566
    },
    {
      "epoch": 0.02567,
      "grad_norm": 0.6761577058111206,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 2567
    },
    {
      "epoch": 0.02568,
      "grad_norm": 0.582893785111441,
      "learning_rate": 0.003,
      "loss": 4.2104,
      "step": 2568
    },
    {
      "epoch": 0.02569,
      "grad_norm": 0.7274372672775176,
      "learning_rate": 0.003,
      "loss": 4.1941,
      "step": 2569
    },
    {
      "epoch": 0.0257,
      "grad_norm": 0.6382797975462534,
      "learning_rate": 0.003,
      "loss": 4.1988,
      "step": 2570
    },
    {
      "epoch": 0.02571,
      "grad_norm": 0.5842618410643182,
      "learning_rate": 0.003,
      "loss": 4.2168,
      "step": 2571
    },
    {
      "epoch": 0.02572,
      "grad_norm": 0.5733806386596177,
      "learning_rate": 0.003,
      "loss": 4.1929,
      "step": 2572
    },
    {
      "epoch": 0.02573,
      "grad_norm": 0.6098335904085562,
      "learning_rate": 0.003,
      "loss": 4.2012,
      "step": 2573
    },
    {
      "epoch": 0.02574,
      "grad_norm": 0.7180717115741516,
      "learning_rate": 0.003,
      "loss": 4.233,
      "step": 2574
    },
    {
      "epoch": 0.02575,
      "grad_norm": 0.7732503452131877,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 2575
    },
    {
      "epoch": 0.02576,
      "grad_norm": 0.8039085737347494,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 2576
    },
    {
      "epoch": 0.02577,
      "grad_norm": 0.7708873473006739,
      "learning_rate": 0.003,
      "loss": 4.183,
      "step": 2577
    },
    {
      "epoch": 0.02578,
      "grad_norm": 0.6416087124695665,
      "learning_rate": 0.003,
      "loss": 4.2007,
      "step": 2578
    },
    {
      "epoch": 0.02579,
      "grad_norm": 0.6401520636312507,
      "learning_rate": 0.003,
      "loss": 4.1503,
      "step": 2579
    },
    {
      "epoch": 0.0258,
      "grad_norm": 0.609541740323575,
      "learning_rate": 0.003,
      "loss": 4.2065,
      "step": 2580
    },
    {
      "epoch": 0.02581,
      "grad_norm": 0.6036763081432622,
      "learning_rate": 0.003,
      "loss": 4.2268,
      "step": 2581
    },
    {
      "epoch": 0.02582,
      "grad_norm": 0.5959601922482883,
      "learning_rate": 0.003,
      "loss": 4.2238,
      "step": 2582
    },
    {
      "epoch": 0.02583,
      "grad_norm": 0.6183189840573183,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2583
    },
    {
      "epoch": 0.02584,
      "grad_norm": 0.7087135376074408,
      "learning_rate": 0.003,
      "loss": 4.2138,
      "step": 2584
    },
    {
      "epoch": 0.02585,
      "grad_norm": 0.6812145006805832,
      "learning_rate": 0.003,
      "loss": 4.1952,
      "step": 2585
    },
    {
      "epoch": 0.02586,
      "grad_norm": 0.7034361159454041,
      "learning_rate": 0.003,
      "loss": 4.2173,
      "step": 2586
    },
    {
      "epoch": 0.02587,
      "grad_norm": 0.7458480631965212,
      "learning_rate": 0.003,
      "loss": 4.2079,
      "step": 2587
    },
    {
      "epoch": 0.02588,
      "grad_norm": 0.6631960688574444,
      "learning_rate": 0.003,
      "loss": 4.2037,
      "step": 2588
    },
    {
      "epoch": 0.02589,
      "grad_norm": 0.6894974904796883,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2589
    },
    {
      "epoch": 0.0259,
      "grad_norm": 0.6129883156712106,
      "learning_rate": 0.003,
      "loss": 4.2205,
      "step": 2590
    },
    {
      "epoch": 0.02591,
      "grad_norm": 0.52233908649411,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 2591
    },
    {
      "epoch": 0.02592,
      "grad_norm": 0.5621629848199292,
      "learning_rate": 0.003,
      "loss": 4.1958,
      "step": 2592
    },
    {
      "epoch": 0.02593,
      "grad_norm": 0.579018019134519,
      "learning_rate": 0.003,
      "loss": 4.1981,
      "step": 2593
    },
    {
      "epoch": 0.02594,
      "grad_norm": 0.7055147915095833,
      "learning_rate": 0.003,
      "loss": 4.2202,
      "step": 2594
    },
    {
      "epoch": 0.02595,
      "grad_norm": 0.9044844479304169,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 2595
    },
    {
      "epoch": 0.02596,
      "grad_norm": 0.9468656207805218,
      "learning_rate": 0.003,
      "loss": 4.2266,
      "step": 2596
    },
    {
      "epoch": 0.02597,
      "grad_norm": 0.6963050525881186,
      "learning_rate": 0.003,
      "loss": 4.1873,
      "step": 2597
    },
    {
      "epoch": 0.02598,
      "grad_norm": 0.6825938651999739,
      "learning_rate": 0.003,
      "loss": 4.2289,
      "step": 2598
    },
    {
      "epoch": 0.02599,
      "grad_norm": 0.7880468862794131,
      "learning_rate": 0.003,
      "loss": 4.225,
      "step": 2599
    },
    {
      "epoch": 0.026,
      "grad_norm": 0.8906440360951334,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 2600
    },
    {
      "epoch": 0.02601,
      "grad_norm": 0.8491652376595475,
      "learning_rate": 0.003,
      "loss": 4.2355,
      "step": 2601
    },
    {
      "epoch": 0.02602,
      "grad_norm": 0.7504592913224991,
      "learning_rate": 0.003,
      "loss": 4.2094,
      "step": 2602
    },
    {
      "epoch": 0.02603,
      "grad_norm": 0.6893393420157942,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 2603
    },
    {
      "epoch": 0.02604,
      "grad_norm": 0.8097986318712986,
      "learning_rate": 0.003,
      "loss": 4.1981,
      "step": 2604
    },
    {
      "epoch": 0.02605,
      "grad_norm": 0.9104130090907385,
      "learning_rate": 0.003,
      "loss": 4.2234,
      "step": 2605
    },
    {
      "epoch": 0.02606,
      "grad_norm": 1.0310289052047719,
      "learning_rate": 0.003,
      "loss": 4.2436,
      "step": 2606
    },
    {
      "epoch": 0.02607,
      "grad_norm": 0.9582413171497328,
      "learning_rate": 0.003,
      "loss": 4.2397,
      "step": 2607
    },
    {
      "epoch": 0.02608,
      "grad_norm": 0.8048575866472404,
      "learning_rate": 0.003,
      "loss": 4.2123,
      "step": 2608
    },
    {
      "epoch": 0.02609,
      "grad_norm": 0.8595866026929856,
      "learning_rate": 0.003,
      "loss": 4.2586,
      "step": 2609
    },
    {
      "epoch": 0.0261,
      "grad_norm": 0.823900371529962,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 2610
    },
    {
      "epoch": 0.02611,
      "grad_norm": 0.7141217984189516,
      "learning_rate": 0.003,
      "loss": 4.1945,
      "step": 2611
    },
    {
      "epoch": 0.02612,
      "grad_norm": 0.716568582474741,
      "learning_rate": 0.003,
      "loss": 4.2324,
      "step": 2612
    },
    {
      "epoch": 0.02613,
      "grad_norm": 0.6977866203993734,
      "learning_rate": 0.003,
      "loss": 4.2227,
      "step": 2613
    },
    {
      "epoch": 0.02614,
      "grad_norm": 0.6120459222412155,
      "learning_rate": 0.003,
      "loss": 4.1988,
      "step": 2614
    },
    {
      "epoch": 0.02615,
      "grad_norm": 0.48978414596737385,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 2615
    },
    {
      "epoch": 0.02616,
      "grad_norm": 0.4111484941092938,
      "learning_rate": 0.003,
      "loss": 4.193,
      "step": 2616
    },
    {
      "epoch": 0.02617,
      "grad_norm": 0.4314798871804994,
      "learning_rate": 0.003,
      "loss": 4.1808,
      "step": 2617
    },
    {
      "epoch": 0.02618,
      "grad_norm": 0.45257700858495165,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 2618
    },
    {
      "epoch": 0.02619,
      "grad_norm": 0.43820325303953606,
      "learning_rate": 0.003,
      "loss": 4.2202,
      "step": 2619
    },
    {
      "epoch": 0.0262,
      "grad_norm": 0.48135056397229203,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 2620
    },
    {
      "epoch": 0.02621,
      "grad_norm": 0.5541701604207973,
      "learning_rate": 0.003,
      "loss": 4.2081,
      "step": 2621
    },
    {
      "epoch": 0.02622,
      "grad_norm": 0.6217167828478687,
      "learning_rate": 0.003,
      "loss": 4.2095,
      "step": 2622
    },
    {
      "epoch": 0.02623,
      "grad_norm": 0.5922780087971841,
      "learning_rate": 0.003,
      "loss": 4.2099,
      "step": 2623
    },
    {
      "epoch": 0.02624,
      "grad_norm": 0.6090506282888395,
      "learning_rate": 0.003,
      "loss": 4.246,
      "step": 2624
    },
    {
      "epoch": 0.02625,
      "grad_norm": 0.5994929489144406,
      "learning_rate": 0.003,
      "loss": 4.2398,
      "step": 2625
    },
    {
      "epoch": 0.02626,
      "grad_norm": 0.6828660510269309,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 2626
    },
    {
      "epoch": 0.02627,
      "grad_norm": 0.7689467228473569,
      "learning_rate": 0.003,
      "loss": 4.1978,
      "step": 2627
    },
    {
      "epoch": 0.02628,
      "grad_norm": 0.772945921723307,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2628
    },
    {
      "epoch": 0.02629,
      "grad_norm": 0.7998878360923921,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 2629
    },
    {
      "epoch": 0.0263,
      "grad_norm": 0.8071005154576086,
      "learning_rate": 0.003,
      "loss": 4.2029,
      "step": 2630
    },
    {
      "epoch": 0.02631,
      "grad_norm": 0.7063928050043093,
      "learning_rate": 0.003,
      "loss": 4.2214,
      "step": 2631
    },
    {
      "epoch": 0.02632,
      "grad_norm": 0.5358237009716528,
      "learning_rate": 0.003,
      "loss": 4.1777,
      "step": 2632
    },
    {
      "epoch": 0.02633,
      "grad_norm": 0.5777990078306479,
      "learning_rate": 0.003,
      "loss": 4.1915,
      "step": 2633
    },
    {
      "epoch": 0.02634,
      "grad_norm": 0.5758302081275357,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 2634
    },
    {
      "epoch": 0.02635,
      "grad_norm": 0.5582382698428011,
      "learning_rate": 0.003,
      "loss": 4.2059,
      "step": 2635
    },
    {
      "epoch": 0.02636,
      "grad_norm": 0.5270921433631331,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 2636
    },
    {
      "epoch": 0.02637,
      "grad_norm": 0.4952392838330578,
      "learning_rate": 0.003,
      "loss": 4.2209,
      "step": 2637
    },
    {
      "epoch": 0.02638,
      "grad_norm": 0.42063663579214783,
      "learning_rate": 0.003,
      "loss": 4.1814,
      "step": 2638
    },
    {
      "epoch": 0.02639,
      "grad_norm": 0.38387035795548136,
      "learning_rate": 0.003,
      "loss": 4.1685,
      "step": 2639
    },
    {
      "epoch": 0.0264,
      "grad_norm": 0.39356773800785,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 2640
    },
    {
      "epoch": 0.02641,
      "grad_norm": 0.4027429798510006,
      "learning_rate": 0.003,
      "loss": 4.1789,
      "step": 2641
    },
    {
      "epoch": 0.02642,
      "grad_norm": 0.4954271688101395,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 2642
    },
    {
      "epoch": 0.02643,
      "grad_norm": 0.544589775918662,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 2643
    },
    {
      "epoch": 0.02644,
      "grad_norm": 0.5577274451115164,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 2644
    },
    {
      "epoch": 0.02645,
      "grad_norm": 0.7183426191932024,
      "learning_rate": 0.003,
      "loss": 4.2101,
      "step": 2645
    },
    {
      "epoch": 0.02646,
      "grad_norm": 0.9615412387915055,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 2646
    },
    {
      "epoch": 0.02647,
      "grad_norm": 1.0740028901393692,
      "learning_rate": 0.003,
      "loss": 4.2196,
      "step": 2647
    },
    {
      "epoch": 0.02648,
      "grad_norm": 0.7345014390620105,
      "learning_rate": 0.003,
      "loss": 4.2256,
      "step": 2648
    },
    {
      "epoch": 0.02649,
      "grad_norm": 0.6618619800661987,
      "learning_rate": 0.003,
      "loss": 4.2112,
      "step": 2649
    },
    {
      "epoch": 0.0265,
      "grad_norm": 0.7347893947046258,
      "learning_rate": 0.003,
      "loss": 4.1759,
      "step": 2650
    },
    {
      "epoch": 0.02651,
      "grad_norm": 0.6484062784740668,
      "learning_rate": 0.003,
      "loss": 4.1885,
      "step": 2651
    },
    {
      "epoch": 0.02652,
      "grad_norm": 0.7667553931843348,
      "learning_rate": 0.003,
      "loss": 4.2255,
      "step": 2652
    },
    {
      "epoch": 0.02653,
      "grad_norm": 0.7289832175113301,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 2653
    },
    {
      "epoch": 0.02654,
      "grad_norm": 0.7103736472197925,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2654
    },
    {
      "epoch": 0.02655,
      "grad_norm": 0.7580675808718691,
      "learning_rate": 0.003,
      "loss": 4.2053,
      "step": 2655
    },
    {
      "epoch": 0.02656,
      "grad_norm": 0.7969812553544864,
      "learning_rate": 0.003,
      "loss": 4.1802,
      "step": 2656
    },
    {
      "epoch": 0.02657,
      "grad_norm": 0.6347600401053554,
      "learning_rate": 0.003,
      "loss": 4.2057,
      "step": 2657
    },
    {
      "epoch": 0.02658,
      "grad_norm": 0.5896722976882536,
      "learning_rate": 0.003,
      "loss": 4.2116,
      "step": 2658
    },
    {
      "epoch": 0.02659,
      "grad_norm": 0.651082411912306,
      "learning_rate": 0.003,
      "loss": 4.2054,
      "step": 2659
    },
    {
      "epoch": 0.0266,
      "grad_norm": 0.6950508733673917,
      "learning_rate": 0.003,
      "loss": 4.1947,
      "step": 2660
    },
    {
      "epoch": 0.02661,
      "grad_norm": 0.7309821342606786,
      "learning_rate": 0.003,
      "loss": 4.2168,
      "step": 2661
    },
    {
      "epoch": 0.02662,
      "grad_norm": 0.8443517188841829,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 2662
    },
    {
      "epoch": 0.02663,
      "grad_norm": 0.9107871566375996,
      "learning_rate": 0.003,
      "loss": 4.1952,
      "step": 2663
    },
    {
      "epoch": 0.02664,
      "grad_norm": 0.8948636609916674,
      "learning_rate": 0.003,
      "loss": 4.2015,
      "step": 2664
    },
    {
      "epoch": 0.02665,
      "grad_norm": 0.7895030310092762,
      "learning_rate": 0.003,
      "loss": 4.2447,
      "step": 2665
    },
    {
      "epoch": 0.02666,
      "grad_norm": 0.7544169624146947,
      "learning_rate": 0.003,
      "loss": 4.2351,
      "step": 2666
    },
    {
      "epoch": 0.02667,
      "grad_norm": 0.7917512909838754,
      "learning_rate": 0.003,
      "loss": 4.2156,
      "step": 2667
    },
    {
      "epoch": 0.02668,
      "grad_norm": 0.8660435271921453,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 2668
    },
    {
      "epoch": 0.02669,
      "grad_norm": 0.8211806952379437,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 2669
    },
    {
      "epoch": 0.0267,
      "grad_norm": 0.7376659322382964,
      "learning_rate": 0.003,
      "loss": 4.2095,
      "step": 2670
    },
    {
      "epoch": 0.02671,
      "grad_norm": 0.7319080249702333,
      "learning_rate": 0.003,
      "loss": 4.2252,
      "step": 2671
    },
    {
      "epoch": 0.02672,
      "grad_norm": 0.674274482359015,
      "learning_rate": 0.003,
      "loss": 4.2161,
      "step": 2672
    },
    {
      "epoch": 0.02673,
      "grad_norm": 0.7262456950172295,
      "learning_rate": 0.003,
      "loss": 4.2152,
      "step": 2673
    },
    {
      "epoch": 0.02674,
      "grad_norm": 0.7478436146050179,
      "learning_rate": 0.003,
      "loss": 4.1824,
      "step": 2674
    },
    {
      "epoch": 0.02675,
      "grad_norm": 0.6974424835283173,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 2675
    },
    {
      "epoch": 0.02676,
      "grad_norm": 0.7773148175185325,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 2676
    },
    {
      "epoch": 0.02677,
      "grad_norm": 0.9696205190703804,
      "learning_rate": 0.003,
      "loss": 4.2371,
      "step": 2677
    },
    {
      "epoch": 0.02678,
      "grad_norm": 0.9527960429936483,
      "learning_rate": 0.003,
      "loss": 4.237,
      "step": 2678
    },
    {
      "epoch": 0.02679,
      "grad_norm": 0.686114258952289,
      "learning_rate": 0.003,
      "loss": 4.1997,
      "step": 2679
    },
    {
      "epoch": 0.0268,
      "grad_norm": 0.587978184638772,
      "learning_rate": 0.003,
      "loss": 4.2094,
      "step": 2680
    },
    {
      "epoch": 0.02681,
      "grad_norm": 0.5919245996634437,
      "learning_rate": 0.003,
      "loss": 4.2288,
      "step": 2681
    },
    {
      "epoch": 0.02682,
      "grad_norm": 0.5747684944993391,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2682
    },
    {
      "epoch": 0.02683,
      "grad_norm": 0.49692411048438434,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2683
    },
    {
      "epoch": 0.02684,
      "grad_norm": 0.42371991981886814,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2684
    },
    {
      "epoch": 0.02685,
      "grad_norm": 0.43270831648127167,
      "learning_rate": 0.003,
      "loss": 4.1987,
      "step": 2685
    },
    {
      "epoch": 0.02686,
      "grad_norm": 0.4373309691769235,
      "learning_rate": 0.003,
      "loss": 4.1811,
      "step": 2686
    },
    {
      "epoch": 0.02687,
      "grad_norm": 0.4372836741597173,
      "learning_rate": 0.003,
      "loss": 4.189,
      "step": 2687
    },
    {
      "epoch": 0.02688,
      "grad_norm": 0.49637578837338964,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 2688
    },
    {
      "epoch": 0.02689,
      "grad_norm": 0.42425016692512946,
      "learning_rate": 0.003,
      "loss": 4.197,
      "step": 2689
    },
    {
      "epoch": 0.0269,
      "grad_norm": 0.3687359570047172,
      "learning_rate": 0.003,
      "loss": 4.199,
      "step": 2690
    },
    {
      "epoch": 0.02691,
      "grad_norm": 0.422533215672721,
      "learning_rate": 0.003,
      "loss": 4.1917,
      "step": 2691
    },
    {
      "epoch": 0.02692,
      "grad_norm": 0.36876790177094304,
      "learning_rate": 0.003,
      "loss": 4.1944,
      "step": 2692
    },
    {
      "epoch": 0.02693,
      "grad_norm": 0.33344364158726264,
      "learning_rate": 0.003,
      "loss": 4.2054,
      "step": 2693
    },
    {
      "epoch": 0.02694,
      "grad_norm": 0.35193571774116406,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 2694
    },
    {
      "epoch": 0.02695,
      "grad_norm": 0.37618933218650874,
      "learning_rate": 0.003,
      "loss": 4.1828,
      "step": 2695
    },
    {
      "epoch": 0.02696,
      "grad_norm": 0.41130959986438415,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 2696
    },
    {
      "epoch": 0.02697,
      "grad_norm": 0.5181896124624418,
      "learning_rate": 0.003,
      "loss": 4.198,
      "step": 2697
    },
    {
      "epoch": 0.02698,
      "grad_norm": 0.8288766752985389,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 2698
    },
    {
      "epoch": 0.02699,
      "grad_norm": 1.1773125298816605,
      "learning_rate": 0.003,
      "loss": 4.23,
      "step": 2699
    },
    {
      "epoch": 0.027,
      "grad_norm": 0.7553245674542797,
      "learning_rate": 0.003,
      "loss": 4.1855,
      "step": 2700
    },
    {
      "epoch": 0.02701,
      "grad_norm": 0.8921172592412265,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 2701
    },
    {
      "epoch": 0.02702,
      "grad_norm": 1.0764387127470998,
      "learning_rate": 0.003,
      "loss": 4.2606,
      "step": 2702
    },
    {
      "epoch": 0.02703,
      "grad_norm": 0.9044993098249169,
      "learning_rate": 0.003,
      "loss": 4.2449,
      "step": 2703
    },
    {
      "epoch": 0.02704,
      "grad_norm": 0.8806690266182866,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2704
    },
    {
      "epoch": 0.02705,
      "grad_norm": 0.8651796636346343,
      "learning_rate": 0.003,
      "loss": 4.2521,
      "step": 2705
    },
    {
      "epoch": 0.02706,
      "grad_norm": 0.8193327778471594,
      "learning_rate": 0.003,
      "loss": 4.2075,
      "step": 2706
    },
    {
      "epoch": 0.02707,
      "grad_norm": 0.8383738553676314,
      "learning_rate": 0.003,
      "loss": 4.2115,
      "step": 2707
    },
    {
      "epoch": 0.02708,
      "grad_norm": 0.7927971133508214,
      "learning_rate": 0.003,
      "loss": 4.2392,
      "step": 2708
    },
    {
      "epoch": 0.02709,
      "grad_norm": 0.702011577126357,
      "learning_rate": 0.003,
      "loss": 4.2044,
      "step": 2709
    },
    {
      "epoch": 0.0271,
      "grad_norm": 0.7610112656844941,
      "learning_rate": 0.003,
      "loss": 4.2165,
      "step": 2710
    },
    {
      "epoch": 0.02711,
      "grad_norm": 0.7565309559289577,
      "learning_rate": 0.003,
      "loss": 4.1983,
      "step": 2711
    },
    {
      "epoch": 0.02712,
      "grad_norm": 0.7177786599726909,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 2712
    },
    {
      "epoch": 0.02713,
      "grad_norm": 0.6187976568863948,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 2713
    },
    {
      "epoch": 0.02714,
      "grad_norm": 0.5823945630189209,
      "learning_rate": 0.003,
      "loss": 4.1941,
      "step": 2714
    },
    {
      "epoch": 0.02715,
      "grad_norm": 0.6590346464153963,
      "learning_rate": 0.003,
      "loss": 4.1907,
      "step": 2715
    },
    {
      "epoch": 0.02716,
      "grad_norm": 0.8913115921228741,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 2716
    },
    {
      "epoch": 0.02717,
      "grad_norm": 0.9991228875759847,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 2717
    },
    {
      "epoch": 0.02718,
      "grad_norm": 0.9312424963750244,
      "learning_rate": 0.003,
      "loss": 4.2354,
      "step": 2718
    },
    {
      "epoch": 0.02719,
      "grad_norm": 0.6956685262676899,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 2719
    },
    {
      "epoch": 0.0272,
      "grad_norm": 0.6830262491518131,
      "learning_rate": 0.003,
      "loss": 4.192,
      "step": 2720
    },
    {
      "epoch": 0.02721,
      "grad_norm": 0.5790742761201393,
      "learning_rate": 0.003,
      "loss": 4.2022,
      "step": 2721
    },
    {
      "epoch": 0.02722,
      "grad_norm": 0.6475692617370225,
      "learning_rate": 0.003,
      "loss": 4.2016,
      "step": 2722
    },
    {
      "epoch": 0.02723,
      "grad_norm": 0.6175956588274222,
      "learning_rate": 0.003,
      "loss": 4.2065,
      "step": 2723
    },
    {
      "epoch": 0.02724,
      "grad_norm": 0.5457779888221875,
      "learning_rate": 0.003,
      "loss": 4.2041,
      "step": 2724
    },
    {
      "epoch": 0.02725,
      "grad_norm": 0.5026905482131955,
      "learning_rate": 0.003,
      "loss": 4.1969,
      "step": 2725
    },
    {
      "epoch": 0.02726,
      "grad_norm": 0.42536659188846526,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 2726
    },
    {
      "epoch": 0.02727,
      "grad_norm": 0.5003680404401011,
      "learning_rate": 0.003,
      "loss": 4.201,
      "step": 2727
    },
    {
      "epoch": 0.02728,
      "grad_norm": 0.48640002941271926,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 2728
    },
    {
      "epoch": 0.02729,
      "grad_norm": 0.41280683340266905,
      "learning_rate": 0.003,
      "loss": 4.2125,
      "step": 2729
    },
    {
      "epoch": 0.0273,
      "grad_norm": 0.43446915053933527,
      "learning_rate": 0.003,
      "loss": 4.1856,
      "step": 2730
    },
    {
      "epoch": 0.02731,
      "grad_norm": 0.51314133885114,
      "learning_rate": 0.003,
      "loss": 4.1847,
      "step": 2731
    },
    {
      "epoch": 0.02732,
      "grad_norm": 0.5875155983292144,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 2732
    },
    {
      "epoch": 0.02733,
      "grad_norm": 0.7435494959956005,
      "learning_rate": 0.003,
      "loss": 4.2133,
      "step": 2733
    },
    {
      "epoch": 0.02734,
      "grad_norm": 0.8318623947885522,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 2734
    },
    {
      "epoch": 0.02735,
      "grad_norm": 0.7270117460217965,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2735
    },
    {
      "epoch": 0.02736,
      "grad_norm": 0.5421807988773258,
      "learning_rate": 0.003,
      "loss": 4.1837,
      "step": 2736
    },
    {
      "epoch": 0.02737,
      "grad_norm": 0.5828799726606484,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 2737
    },
    {
      "epoch": 0.02738,
      "grad_norm": 0.6293605543493668,
      "learning_rate": 0.003,
      "loss": 4.2025,
      "step": 2738
    },
    {
      "epoch": 0.02739,
      "grad_norm": 0.5663805709619927,
      "learning_rate": 0.003,
      "loss": 4.1907,
      "step": 2739
    },
    {
      "epoch": 0.0274,
      "grad_norm": 0.5687899337220772,
      "learning_rate": 0.003,
      "loss": 4.1804,
      "step": 2740
    },
    {
      "epoch": 0.02741,
      "grad_norm": 0.6075448196131129,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 2741
    },
    {
      "epoch": 0.02742,
      "grad_norm": 0.550237016515108,
      "learning_rate": 0.003,
      "loss": 4.1804,
      "step": 2742
    },
    {
      "epoch": 0.02743,
      "grad_norm": 0.6055901207765589,
      "learning_rate": 0.003,
      "loss": 4.2013,
      "step": 2743
    },
    {
      "epoch": 0.02744,
      "grad_norm": 0.7300871101889295,
      "learning_rate": 0.003,
      "loss": 4.1884,
      "step": 2744
    },
    {
      "epoch": 0.02745,
      "grad_norm": 0.8259955139853259,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2745
    },
    {
      "epoch": 0.02746,
      "grad_norm": 0.83396545170588,
      "learning_rate": 0.003,
      "loss": 4.1856,
      "step": 2746
    },
    {
      "epoch": 0.02747,
      "grad_norm": 0.729485971791691,
      "learning_rate": 0.003,
      "loss": 4.2152,
      "step": 2747
    },
    {
      "epoch": 0.02748,
      "grad_norm": 0.695156151246541,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 2748
    },
    {
      "epoch": 0.02749,
      "grad_norm": 0.6271520188600127,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 2749
    },
    {
      "epoch": 0.0275,
      "grad_norm": 0.6580176171244524,
      "learning_rate": 0.003,
      "loss": 4.2067,
      "step": 2750
    },
    {
      "epoch": 0.02751,
      "grad_norm": 0.672675352925045,
      "learning_rate": 0.003,
      "loss": 4.2192,
      "step": 2751
    },
    {
      "epoch": 0.02752,
      "grad_norm": 0.6249956014069801,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 2752
    },
    {
      "epoch": 0.02753,
      "grad_norm": 0.6422052590772455,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 2753
    },
    {
      "epoch": 0.02754,
      "grad_norm": 0.6380940449715866,
      "learning_rate": 0.003,
      "loss": 4.1719,
      "step": 2754
    },
    {
      "epoch": 0.02755,
      "grad_norm": 0.5972581452735276,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 2755
    },
    {
      "epoch": 0.02756,
      "grad_norm": 0.6077948327385506,
      "learning_rate": 0.003,
      "loss": 4.1967,
      "step": 2756
    },
    {
      "epoch": 0.02757,
      "grad_norm": 0.7032781486647851,
      "learning_rate": 0.003,
      "loss": 4.2127,
      "step": 2757
    },
    {
      "epoch": 0.02758,
      "grad_norm": 0.793287683394734,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 2758
    },
    {
      "epoch": 0.02759,
      "grad_norm": 0.9418904888299408,
      "learning_rate": 0.003,
      "loss": 4.2024,
      "step": 2759
    },
    {
      "epoch": 0.0276,
      "grad_norm": 0.9351479329737886,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 2760
    },
    {
      "epoch": 0.02761,
      "grad_norm": 0.8139578363148168,
      "learning_rate": 0.003,
      "loss": 4.2282,
      "step": 2761
    },
    {
      "epoch": 0.02762,
      "grad_norm": 0.6757269439646322,
      "learning_rate": 0.003,
      "loss": 4.1918,
      "step": 2762
    },
    {
      "epoch": 0.02763,
      "grad_norm": 0.673672124363711,
      "learning_rate": 0.003,
      "loss": 4.2006,
      "step": 2763
    },
    {
      "epoch": 0.02764,
      "grad_norm": 0.8103484124048353,
      "learning_rate": 0.003,
      "loss": 4.199,
      "step": 2764
    },
    {
      "epoch": 0.02765,
      "grad_norm": 0.861735499978179,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2765
    },
    {
      "epoch": 0.02766,
      "grad_norm": 0.7923213801421585,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 2766
    },
    {
      "epoch": 0.02767,
      "grad_norm": 0.7631227515735746,
      "learning_rate": 0.003,
      "loss": 4.2302,
      "step": 2767
    },
    {
      "epoch": 0.02768,
      "grad_norm": 0.8386215859490249,
      "learning_rate": 0.003,
      "loss": 4.1676,
      "step": 2768
    },
    {
      "epoch": 0.02769,
      "grad_norm": 0.8716391723421844,
      "learning_rate": 0.003,
      "loss": 4.2312,
      "step": 2769
    },
    {
      "epoch": 0.0277,
      "grad_norm": 0.79734786858364,
      "learning_rate": 0.003,
      "loss": 4.1984,
      "step": 2770
    },
    {
      "epoch": 0.02771,
      "grad_norm": 0.7636865958720973,
      "learning_rate": 0.003,
      "loss": 4.187,
      "step": 2771
    },
    {
      "epoch": 0.02772,
      "grad_norm": 0.7711684731670819,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 2772
    },
    {
      "epoch": 0.02773,
      "grad_norm": 0.6855679524292124,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2773
    },
    {
      "epoch": 0.02774,
      "grad_norm": 0.6373232969577011,
      "learning_rate": 0.003,
      "loss": 4.2297,
      "step": 2774
    },
    {
      "epoch": 0.02775,
      "grad_norm": 0.5808981715198482,
      "learning_rate": 0.003,
      "loss": 4.1967,
      "step": 2775
    },
    {
      "epoch": 0.02776,
      "grad_norm": 0.551216477733259,
      "learning_rate": 0.003,
      "loss": 4.1912,
      "step": 2776
    },
    {
      "epoch": 0.02777,
      "grad_norm": 0.5502623559066758,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 2777
    },
    {
      "epoch": 0.02778,
      "grad_norm": 0.5471574038460382,
      "learning_rate": 0.003,
      "loss": 4.209,
      "step": 2778
    },
    {
      "epoch": 0.02779,
      "grad_norm": 0.5740086450201776,
      "learning_rate": 0.003,
      "loss": 4.1802,
      "step": 2779
    },
    {
      "epoch": 0.0278,
      "grad_norm": 0.6603722539886849,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 2780
    },
    {
      "epoch": 0.02781,
      "grad_norm": 0.7045319397684864,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 2781
    },
    {
      "epoch": 0.02782,
      "grad_norm": 0.7384112585585232,
      "learning_rate": 0.003,
      "loss": 4.189,
      "step": 2782
    },
    {
      "epoch": 0.02783,
      "grad_norm": 0.7302997601128179,
      "learning_rate": 0.003,
      "loss": 4.1951,
      "step": 2783
    },
    {
      "epoch": 0.02784,
      "grad_norm": 0.623059472233842,
      "learning_rate": 0.003,
      "loss": 4.1896,
      "step": 2784
    },
    {
      "epoch": 0.02785,
      "grad_norm": 0.6333396786278916,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2785
    },
    {
      "epoch": 0.02786,
      "grad_norm": 0.7323878741557076,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 2786
    },
    {
      "epoch": 0.02787,
      "grad_norm": 0.7244716928344251,
      "learning_rate": 0.003,
      "loss": 4.2035,
      "step": 2787
    },
    {
      "epoch": 0.02788,
      "grad_norm": 0.6542769748417147,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 2788
    },
    {
      "epoch": 0.02789,
      "grad_norm": 0.6574058601306716,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2789
    },
    {
      "epoch": 0.0279,
      "grad_norm": 0.7081427842141735,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 2790
    },
    {
      "epoch": 0.02791,
      "grad_norm": 0.6942117835439664,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 2791
    },
    {
      "epoch": 0.02792,
      "grad_norm": 0.6067557855062717,
      "learning_rate": 0.003,
      "loss": 4.1954,
      "step": 2792
    },
    {
      "epoch": 0.02793,
      "grad_norm": 0.5947324428169551,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 2793
    },
    {
      "epoch": 0.02794,
      "grad_norm": 0.5401502710295608,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 2794
    },
    {
      "epoch": 0.02795,
      "grad_norm": 0.5028686206702352,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2795
    },
    {
      "epoch": 0.02796,
      "grad_norm": 0.49627952868559183,
      "learning_rate": 0.003,
      "loss": 4.191,
      "step": 2796
    },
    {
      "epoch": 0.02797,
      "grad_norm": 0.4876291995281765,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 2797
    },
    {
      "epoch": 0.02798,
      "grad_norm": 0.5294169569466303,
      "learning_rate": 0.003,
      "loss": 4.1979,
      "step": 2798
    },
    {
      "epoch": 0.02799,
      "grad_norm": 0.5927669437619445,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 2799
    },
    {
      "epoch": 0.028,
      "grad_norm": 0.6682805931240281,
      "learning_rate": 0.003,
      "loss": 4.1988,
      "step": 2800
    },
    {
      "epoch": 0.02801,
      "grad_norm": 0.9068693661645921,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 2801
    },
    {
      "epoch": 0.02802,
      "grad_norm": 1.0254407756501984,
      "learning_rate": 0.003,
      "loss": 4.1982,
      "step": 2802
    },
    {
      "epoch": 0.02803,
      "grad_norm": 0.8128875626073906,
      "learning_rate": 0.003,
      "loss": 4.2114,
      "step": 2803
    },
    {
      "epoch": 0.02804,
      "grad_norm": 0.7353680218352796,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 2804
    },
    {
      "epoch": 0.02805,
      "grad_norm": 0.7856799021536293,
      "learning_rate": 0.003,
      "loss": 4.222,
      "step": 2805
    },
    {
      "epoch": 0.02806,
      "grad_norm": 0.7319188472007512,
      "learning_rate": 0.003,
      "loss": 4.1915,
      "step": 2806
    },
    {
      "epoch": 0.02807,
      "grad_norm": 0.6281145778193726,
      "learning_rate": 0.003,
      "loss": 4.1913,
      "step": 2807
    },
    {
      "epoch": 0.02808,
      "grad_norm": 0.6007996487222561,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 2808
    },
    {
      "epoch": 0.02809,
      "grad_norm": 0.6107598223212796,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2809
    },
    {
      "epoch": 0.0281,
      "grad_norm": 0.5658938842173376,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 2810
    },
    {
      "epoch": 0.02811,
      "grad_norm": 0.5188959758106707,
      "learning_rate": 0.003,
      "loss": 4.1753,
      "step": 2811
    },
    {
      "epoch": 0.02812,
      "grad_norm": 0.47017604842162736,
      "learning_rate": 0.003,
      "loss": 4.19,
      "step": 2812
    },
    {
      "epoch": 0.02813,
      "grad_norm": 0.43817425390848724,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 2813
    },
    {
      "epoch": 0.02814,
      "grad_norm": 0.49665633422440225,
      "learning_rate": 0.003,
      "loss": 4.1789,
      "step": 2814
    },
    {
      "epoch": 0.02815,
      "grad_norm": 0.5737919498303556,
      "learning_rate": 0.003,
      "loss": 4.2179,
      "step": 2815
    },
    {
      "epoch": 0.02816,
      "grad_norm": 0.5958113314856148,
      "learning_rate": 0.003,
      "loss": 4.1978,
      "step": 2816
    },
    {
      "epoch": 0.02817,
      "grad_norm": 0.7178958177645314,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 2817
    },
    {
      "epoch": 0.02818,
      "grad_norm": 0.856700432217493,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 2818
    },
    {
      "epoch": 0.02819,
      "grad_norm": 0.8381122457158061,
      "learning_rate": 0.003,
      "loss": 4.2052,
      "step": 2819
    },
    {
      "epoch": 0.0282,
      "grad_norm": 0.9145463072400436,
      "learning_rate": 0.003,
      "loss": 4.2085,
      "step": 2820
    },
    {
      "epoch": 0.02821,
      "grad_norm": 0.8180716884994599,
      "learning_rate": 0.003,
      "loss": 4.2204,
      "step": 2821
    },
    {
      "epoch": 0.02822,
      "grad_norm": 0.749853073972982,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2822
    },
    {
      "epoch": 0.02823,
      "grad_norm": 0.8042960763019706,
      "learning_rate": 0.003,
      "loss": 4.2228,
      "step": 2823
    },
    {
      "epoch": 0.02824,
      "grad_norm": 0.9373801354350708,
      "learning_rate": 0.003,
      "loss": 4.2156,
      "step": 2824
    },
    {
      "epoch": 0.02825,
      "grad_norm": 1.0578826545081004,
      "learning_rate": 0.003,
      "loss": 4.2331,
      "step": 2825
    },
    {
      "epoch": 0.02826,
      "grad_norm": 0.7996056170859099,
      "learning_rate": 0.003,
      "loss": 4.1963,
      "step": 2826
    },
    {
      "epoch": 0.02827,
      "grad_norm": 0.8051765401350005,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 2827
    },
    {
      "epoch": 0.02828,
      "grad_norm": 0.8027875497458861,
      "learning_rate": 0.003,
      "loss": 4.2252,
      "step": 2828
    },
    {
      "epoch": 0.02829,
      "grad_norm": 0.7733380033017664,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 2829
    },
    {
      "epoch": 0.0283,
      "grad_norm": 0.7838154488025357,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 2830
    },
    {
      "epoch": 0.02831,
      "grad_norm": 0.7961302139800379,
      "learning_rate": 0.003,
      "loss": 4.2125,
      "step": 2831
    },
    {
      "epoch": 0.02832,
      "grad_norm": 0.8562794848672607,
      "learning_rate": 0.003,
      "loss": 4.2283,
      "step": 2832
    },
    {
      "epoch": 0.02833,
      "grad_norm": 0.9219375196224617,
      "learning_rate": 0.003,
      "loss": 4.2033,
      "step": 2833
    },
    {
      "epoch": 0.02834,
      "grad_norm": 1.066449364140732,
      "learning_rate": 0.003,
      "loss": 4.2018,
      "step": 2834
    },
    {
      "epoch": 0.02835,
      "grad_norm": 0.9222685550708334,
      "learning_rate": 0.003,
      "loss": 4.2111,
      "step": 2835
    },
    {
      "epoch": 0.02836,
      "grad_norm": 0.7684380143787258,
      "learning_rate": 0.003,
      "loss": 4.2263,
      "step": 2836
    },
    {
      "epoch": 0.02837,
      "grad_norm": 0.6557155817543934,
      "learning_rate": 0.003,
      "loss": 4.1966,
      "step": 2837
    },
    {
      "epoch": 0.02838,
      "grad_norm": 0.635865016407609,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 2838
    },
    {
      "epoch": 0.02839,
      "grad_norm": 0.6046255360999631,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 2839
    },
    {
      "epoch": 0.0284,
      "grad_norm": 0.5605750331386489,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 2840
    },
    {
      "epoch": 0.02841,
      "grad_norm": 0.5616642948887489,
      "learning_rate": 0.003,
      "loss": 4.206,
      "step": 2841
    },
    {
      "epoch": 0.02842,
      "grad_norm": 0.57068171158362,
      "learning_rate": 0.003,
      "loss": 4.1872,
      "step": 2842
    },
    {
      "epoch": 0.02843,
      "grad_norm": 0.5405942011469851,
      "learning_rate": 0.003,
      "loss": 4.2077,
      "step": 2843
    },
    {
      "epoch": 0.02844,
      "grad_norm": 0.5611073031217337,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2844
    },
    {
      "epoch": 0.02845,
      "grad_norm": 0.6289027848917312,
      "learning_rate": 0.003,
      "loss": 4.1835,
      "step": 2845
    },
    {
      "epoch": 0.02846,
      "grad_norm": 0.6920775632747969,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 2846
    },
    {
      "epoch": 0.02847,
      "grad_norm": 0.767579826338002,
      "learning_rate": 0.003,
      "loss": 4.2364,
      "step": 2847
    },
    {
      "epoch": 0.02848,
      "grad_norm": 0.7249701206375546,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 2848
    },
    {
      "epoch": 0.02849,
      "grad_norm": 0.516455337675884,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 2849
    },
    {
      "epoch": 0.0285,
      "grad_norm": 0.5059786145152202,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 2850
    },
    {
      "epoch": 0.02851,
      "grad_norm": 0.5163885986759773,
      "learning_rate": 0.003,
      "loss": 4.2134,
      "step": 2851
    },
    {
      "epoch": 0.02852,
      "grad_norm": 0.49258101706112184,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 2852
    },
    {
      "epoch": 0.02853,
      "grad_norm": 0.500381435832321,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 2853
    },
    {
      "epoch": 0.02854,
      "grad_norm": 0.4710049420804533,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 2854
    },
    {
      "epoch": 0.02855,
      "grad_norm": 0.4796815944787949,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 2855
    },
    {
      "epoch": 0.02856,
      "grad_norm": 0.5101632070884339,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 2856
    },
    {
      "epoch": 0.02857,
      "grad_norm": 0.4913306093184678,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 2857
    },
    {
      "epoch": 0.02858,
      "grad_norm": 0.4711311880423705,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 2858
    },
    {
      "epoch": 0.02859,
      "grad_norm": 0.5082622329894141,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 2859
    },
    {
      "epoch": 0.0286,
      "grad_norm": 0.5184604450005492,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 2860
    },
    {
      "epoch": 0.02861,
      "grad_norm": 0.48267289618449316,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 2861
    },
    {
      "epoch": 0.02862,
      "grad_norm": 0.4761246257076003,
      "learning_rate": 0.003,
      "loss": 4.1721,
      "step": 2862
    },
    {
      "epoch": 0.02863,
      "grad_norm": 0.555990003004704,
      "learning_rate": 0.003,
      "loss": 4.1996,
      "step": 2863
    },
    {
      "epoch": 0.02864,
      "grad_norm": 0.6398184414941633,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 2864
    },
    {
      "epoch": 0.02865,
      "grad_norm": 0.7105267215046387,
      "learning_rate": 0.003,
      "loss": 4.2076,
      "step": 2865
    },
    {
      "epoch": 0.02866,
      "grad_norm": 0.7313418455236842,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2866
    },
    {
      "epoch": 0.02867,
      "grad_norm": 0.6641528598178345,
      "learning_rate": 0.003,
      "loss": 4.1753,
      "step": 2867
    },
    {
      "epoch": 0.02868,
      "grad_norm": 0.5930404639850378,
      "learning_rate": 0.003,
      "loss": 4.1816,
      "step": 2868
    },
    {
      "epoch": 0.02869,
      "grad_norm": 0.6719622014213373,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 2869
    },
    {
      "epoch": 0.0287,
      "grad_norm": 0.6550610400030609,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 2870
    },
    {
      "epoch": 0.02871,
      "grad_norm": 0.6686050845776436,
      "learning_rate": 0.003,
      "loss": 4.1824,
      "step": 2871
    },
    {
      "epoch": 0.02872,
      "grad_norm": 0.7301947656997149,
      "learning_rate": 0.003,
      "loss": 4.2172,
      "step": 2872
    },
    {
      "epoch": 0.02873,
      "grad_norm": 0.6807495486455808,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 2873
    },
    {
      "epoch": 0.02874,
      "grad_norm": 0.625144556477654,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 2874
    },
    {
      "epoch": 0.02875,
      "grad_norm": 0.6566904903408397,
      "learning_rate": 0.003,
      "loss": 4.2051,
      "step": 2875
    },
    {
      "epoch": 0.02876,
      "grad_norm": 0.759141268501963,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2876
    },
    {
      "epoch": 0.02877,
      "grad_norm": 0.8807634127058958,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 2877
    },
    {
      "epoch": 0.02878,
      "grad_norm": 1.0261973428455773,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 2878
    },
    {
      "epoch": 0.02879,
      "grad_norm": 0.888483399285398,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 2879
    },
    {
      "epoch": 0.0288,
      "grad_norm": 0.8435453856609076,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 2880
    },
    {
      "epoch": 0.02881,
      "grad_norm": 0.8865486451512147,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 2881
    },
    {
      "epoch": 0.02882,
      "grad_norm": 0.8928993496723746,
      "learning_rate": 0.003,
      "loss": 4.2242,
      "step": 2882
    },
    {
      "epoch": 0.02883,
      "grad_norm": 0.8226859467424321,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 2883
    },
    {
      "epoch": 0.02884,
      "grad_norm": 0.77429031635637,
      "learning_rate": 0.003,
      "loss": 4.2401,
      "step": 2884
    },
    {
      "epoch": 0.02885,
      "grad_norm": 0.8922281925752906,
      "learning_rate": 0.003,
      "loss": 4.1845,
      "step": 2885
    },
    {
      "epoch": 0.02886,
      "grad_norm": 0.9214758547731667,
      "learning_rate": 0.003,
      "loss": 4.2251,
      "step": 2886
    },
    {
      "epoch": 0.02887,
      "grad_norm": 0.7646945576422838,
      "learning_rate": 0.003,
      "loss": 4.205,
      "step": 2887
    },
    {
      "epoch": 0.02888,
      "grad_norm": 0.5765970376943947,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 2888
    },
    {
      "epoch": 0.02889,
      "grad_norm": 0.6149544262580506,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 2889
    },
    {
      "epoch": 0.0289,
      "grad_norm": 0.5856358962779576,
      "learning_rate": 0.003,
      "loss": 4.2099,
      "step": 2890
    },
    {
      "epoch": 0.02891,
      "grad_norm": 0.5989364430008874,
      "learning_rate": 0.003,
      "loss": 4.2274,
      "step": 2891
    },
    {
      "epoch": 0.02892,
      "grad_norm": 0.6055146901336932,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 2892
    },
    {
      "epoch": 0.02893,
      "grad_norm": 0.6302785437079876,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 2893
    },
    {
      "epoch": 0.02894,
      "grad_norm": 0.5540105467551689,
      "learning_rate": 0.003,
      "loss": 4.1878,
      "step": 2894
    },
    {
      "epoch": 0.02895,
      "grad_norm": 0.4023612128343426,
      "learning_rate": 0.003,
      "loss": 4.2266,
      "step": 2895
    },
    {
      "epoch": 0.02896,
      "grad_norm": 0.46314453746126943,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 2896
    },
    {
      "epoch": 0.02897,
      "grad_norm": 0.5689975115372741,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 2897
    },
    {
      "epoch": 0.02898,
      "grad_norm": 0.5759099877999215,
      "learning_rate": 0.003,
      "loss": 4.2076,
      "step": 2898
    },
    {
      "epoch": 0.02899,
      "grad_norm": 0.5841372181240183,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2899
    },
    {
      "epoch": 0.029,
      "grad_norm": 0.5742774291659906,
      "learning_rate": 0.003,
      "loss": 4.1791,
      "step": 2900
    },
    {
      "epoch": 0.02901,
      "grad_norm": 0.505503655503479,
      "learning_rate": 0.003,
      "loss": 4.1922,
      "step": 2901
    },
    {
      "epoch": 0.02902,
      "grad_norm": 0.47503490148304944,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 2902
    },
    {
      "epoch": 0.02903,
      "grad_norm": 0.5474247569951972,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 2903
    },
    {
      "epoch": 0.02904,
      "grad_norm": 0.6382457425375883,
      "learning_rate": 0.003,
      "loss": 4.1897,
      "step": 2904
    },
    {
      "epoch": 0.02905,
      "grad_norm": 0.707934254197877,
      "learning_rate": 0.003,
      "loss": 4.2073,
      "step": 2905
    },
    {
      "epoch": 0.02906,
      "grad_norm": 0.8403476091404251,
      "learning_rate": 0.003,
      "loss": 4.192,
      "step": 2906
    },
    {
      "epoch": 0.02907,
      "grad_norm": 0.8318613643365537,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 2907
    },
    {
      "epoch": 0.02908,
      "grad_norm": 0.7272031628626594,
      "learning_rate": 0.003,
      "loss": 4.1912,
      "step": 2908
    },
    {
      "epoch": 0.02909,
      "grad_norm": 0.669946483884344,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2909
    },
    {
      "epoch": 0.0291,
      "grad_norm": 0.6105283375602145,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 2910
    },
    {
      "epoch": 0.02911,
      "grad_norm": 0.6965941358145008,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2911
    },
    {
      "epoch": 0.02912,
      "grad_norm": 0.6975491884532862,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 2912
    },
    {
      "epoch": 0.02913,
      "grad_norm": 0.7857229509367818,
      "learning_rate": 0.003,
      "loss": 4.1992,
      "step": 2913
    },
    {
      "epoch": 0.02914,
      "grad_norm": 0.811281083177593,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 2914
    },
    {
      "epoch": 0.02915,
      "grad_norm": 0.6690033742013554,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 2915
    },
    {
      "epoch": 0.02916,
      "grad_norm": 0.5795750985593426,
      "learning_rate": 0.003,
      "loss": 4.1847,
      "step": 2916
    },
    {
      "epoch": 0.02917,
      "grad_norm": 0.6521162065161483,
      "learning_rate": 0.003,
      "loss": 4.2172,
      "step": 2917
    },
    {
      "epoch": 0.02918,
      "grad_norm": 0.6434845720798231,
      "learning_rate": 0.003,
      "loss": 4.1719,
      "step": 2918
    },
    {
      "epoch": 0.02919,
      "grad_norm": 0.6415437095317365,
      "learning_rate": 0.003,
      "loss": 4.2133,
      "step": 2919
    },
    {
      "epoch": 0.0292,
      "grad_norm": 0.6445863883052798,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 2920
    },
    {
      "epoch": 0.02921,
      "grad_norm": 0.5711168990862754,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 2921
    },
    {
      "epoch": 0.02922,
      "grad_norm": 0.6255743522411465,
      "learning_rate": 0.003,
      "loss": 4.1926,
      "step": 2922
    },
    {
      "epoch": 0.02923,
      "grad_norm": 0.7756468163112304,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 2923
    },
    {
      "epoch": 0.02924,
      "grad_norm": 0.8580521016504452,
      "learning_rate": 0.003,
      "loss": 4.2067,
      "step": 2924
    },
    {
      "epoch": 0.02925,
      "grad_norm": 0.8062805285298449,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2925
    },
    {
      "epoch": 0.02926,
      "grad_norm": 0.6123939436837456,
      "learning_rate": 0.003,
      "loss": 4.1817,
      "step": 2926
    },
    {
      "epoch": 0.02927,
      "grad_norm": 0.5880261599317353,
      "learning_rate": 0.003,
      "loss": 4.1862,
      "step": 2927
    },
    {
      "epoch": 0.02928,
      "grad_norm": 0.6052041112444663,
      "learning_rate": 0.003,
      "loss": 4.2074,
      "step": 2928
    },
    {
      "epoch": 0.02929,
      "grad_norm": 0.6235842112912531,
      "learning_rate": 0.003,
      "loss": 4.192,
      "step": 2929
    },
    {
      "epoch": 0.0293,
      "grad_norm": 0.660214912657757,
      "learning_rate": 0.003,
      "loss": 4.1869,
      "step": 2930
    },
    {
      "epoch": 0.02931,
      "grad_norm": 0.5466629516534421,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 2931
    },
    {
      "epoch": 0.02932,
      "grad_norm": 0.5338608341273119,
      "learning_rate": 0.003,
      "loss": 4.1686,
      "step": 2932
    },
    {
      "epoch": 0.02933,
      "grad_norm": 0.4969939045829935,
      "learning_rate": 0.003,
      "loss": 4.1951,
      "step": 2933
    },
    {
      "epoch": 0.02934,
      "grad_norm": 0.49477597477846896,
      "learning_rate": 0.003,
      "loss": 4.1818,
      "step": 2934
    },
    {
      "epoch": 0.02935,
      "grad_norm": 0.4464607167642475,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 2935
    },
    {
      "epoch": 0.02936,
      "grad_norm": 0.49350443190661963,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 2936
    },
    {
      "epoch": 0.02937,
      "grad_norm": 0.6198926034158319,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 2937
    },
    {
      "epoch": 0.02938,
      "grad_norm": 0.7272182520738907,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 2938
    },
    {
      "epoch": 0.02939,
      "grad_norm": 0.8486086179503688,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 2939
    },
    {
      "epoch": 0.0294,
      "grad_norm": 0.8446986466812662,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 2940
    },
    {
      "epoch": 0.02941,
      "grad_norm": 0.7723043017297002,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 2941
    },
    {
      "epoch": 0.02942,
      "grad_norm": 0.7697250754745237,
      "learning_rate": 0.003,
      "loss": 4.197,
      "step": 2942
    },
    {
      "epoch": 0.02943,
      "grad_norm": 0.741727859654241,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 2943
    },
    {
      "epoch": 0.02944,
      "grad_norm": 0.691917422319892,
      "learning_rate": 0.003,
      "loss": 4.2047,
      "step": 2944
    },
    {
      "epoch": 0.02945,
      "grad_norm": 0.6550566422880618,
      "learning_rate": 0.003,
      "loss": 4.1868,
      "step": 2945
    },
    {
      "epoch": 0.02946,
      "grad_norm": 0.7775286126916453,
      "learning_rate": 0.003,
      "loss": 4.1841,
      "step": 2946
    },
    {
      "epoch": 0.02947,
      "grad_norm": 0.989103701566936,
      "learning_rate": 0.003,
      "loss": 4.2271,
      "step": 2947
    },
    {
      "epoch": 0.02948,
      "grad_norm": 1.1031413486335138,
      "learning_rate": 0.003,
      "loss": 4.2196,
      "step": 2948
    },
    {
      "epoch": 0.02949,
      "grad_norm": 0.9071172160919259,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 2949
    },
    {
      "epoch": 0.0295,
      "grad_norm": 0.9640749695276918,
      "learning_rate": 0.003,
      "loss": 4.2161,
      "step": 2950
    },
    {
      "epoch": 0.02951,
      "grad_norm": 1.0552794520418902,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 2951
    },
    {
      "epoch": 0.02952,
      "grad_norm": 1.0245244401252407,
      "learning_rate": 0.003,
      "loss": 4.2154,
      "step": 2952
    },
    {
      "epoch": 0.02953,
      "grad_norm": 1.0404102150119146,
      "learning_rate": 0.003,
      "loss": 4.2329,
      "step": 2953
    },
    {
      "epoch": 0.02954,
      "grad_norm": 1.200181394447266,
      "learning_rate": 0.003,
      "loss": 4.2778,
      "step": 2954
    },
    {
      "epoch": 0.02955,
      "grad_norm": 0.8161704211969852,
      "learning_rate": 0.003,
      "loss": 4.2394,
      "step": 2955
    },
    {
      "epoch": 0.02956,
      "grad_norm": 0.8295432503492015,
      "learning_rate": 0.003,
      "loss": 4.2255,
      "step": 2956
    },
    {
      "epoch": 0.02957,
      "grad_norm": 0.8024603706062242,
      "learning_rate": 0.003,
      "loss": 4.2343,
      "step": 2957
    },
    {
      "epoch": 0.02958,
      "grad_norm": 0.9622635969093861,
      "learning_rate": 0.003,
      "loss": 4.2527,
      "step": 2958
    },
    {
      "epoch": 0.02959,
      "grad_norm": 1.144055914507281,
      "learning_rate": 0.003,
      "loss": 4.2509,
      "step": 2959
    },
    {
      "epoch": 0.0296,
      "grad_norm": 0.8230875124438284,
      "learning_rate": 0.003,
      "loss": 4.2477,
      "step": 2960
    },
    {
      "epoch": 0.02961,
      "grad_norm": 0.7761645102479322,
      "learning_rate": 0.003,
      "loss": 4.2177,
      "step": 2961
    },
    {
      "epoch": 0.02962,
      "grad_norm": 0.5798322996384551,
      "learning_rate": 0.003,
      "loss": 4.2146,
      "step": 2962
    },
    {
      "epoch": 0.02963,
      "grad_norm": 0.6082007449676078,
      "learning_rate": 0.003,
      "loss": 4.1885,
      "step": 2963
    },
    {
      "epoch": 0.02964,
      "grad_norm": 0.5952539583915469,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 2964
    },
    {
      "epoch": 0.02965,
      "grad_norm": 0.6067339945313683,
      "learning_rate": 0.003,
      "loss": 4.2192,
      "step": 2965
    },
    {
      "epoch": 0.02966,
      "grad_norm": 0.6300423766569044,
      "learning_rate": 0.003,
      "loss": 4.2275,
      "step": 2966
    },
    {
      "epoch": 0.02967,
      "grad_norm": 0.6210602868633931,
      "learning_rate": 0.003,
      "loss": 4.2121,
      "step": 2967
    },
    {
      "epoch": 0.02968,
      "grad_norm": 0.5447153342269296,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 2968
    },
    {
      "epoch": 0.02969,
      "grad_norm": 0.5492051288557623,
      "learning_rate": 0.003,
      "loss": 4.2026,
      "step": 2969
    },
    {
      "epoch": 0.0297,
      "grad_norm": 0.5667658969213387,
      "learning_rate": 0.003,
      "loss": 4.1857,
      "step": 2970
    },
    {
      "epoch": 0.02971,
      "grad_norm": 0.5386297699076645,
      "learning_rate": 0.003,
      "loss": 4.2025,
      "step": 2971
    },
    {
      "epoch": 0.02972,
      "grad_norm": 0.4946208196601275,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 2972
    },
    {
      "epoch": 0.02973,
      "grad_norm": 0.5841767565220707,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 2973
    },
    {
      "epoch": 0.02974,
      "grad_norm": 0.7076979032412727,
      "learning_rate": 0.003,
      "loss": 4.1953,
      "step": 2974
    },
    {
      "epoch": 0.02975,
      "grad_norm": 0.9835377294492083,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 2975
    },
    {
      "epoch": 0.02976,
      "grad_norm": 1.1071095377980806,
      "learning_rate": 0.003,
      "loss": 4.2143,
      "step": 2976
    },
    {
      "epoch": 0.02977,
      "grad_norm": 0.7259690969712574,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 2977
    },
    {
      "epoch": 0.02978,
      "grad_norm": 0.7734023943128602,
      "learning_rate": 0.003,
      "loss": 4.2094,
      "step": 2978
    },
    {
      "epoch": 0.02979,
      "grad_norm": 0.7278346414839765,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 2979
    },
    {
      "epoch": 0.0298,
      "grad_norm": 0.660497169411274,
      "learning_rate": 0.003,
      "loss": 4.1931,
      "step": 2980
    },
    {
      "epoch": 0.02981,
      "grad_norm": 0.6667577508097533,
      "learning_rate": 0.003,
      "loss": 4.2009,
      "step": 2981
    },
    {
      "epoch": 0.02982,
      "grad_norm": 0.6386554072904321,
      "learning_rate": 0.003,
      "loss": 4.1958,
      "step": 2982
    },
    {
      "epoch": 0.02983,
      "grad_norm": 0.5418480426752083,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 2983
    },
    {
      "epoch": 0.02984,
      "grad_norm": 0.5015711298760291,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 2984
    },
    {
      "epoch": 0.02985,
      "grad_norm": 0.47335120613727616,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2985
    },
    {
      "epoch": 0.02986,
      "grad_norm": 0.45169284672546406,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 2986
    },
    {
      "epoch": 0.02987,
      "grad_norm": 0.3872178111815347,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 2987
    },
    {
      "epoch": 0.02988,
      "grad_norm": 0.5156766146850885,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 2988
    },
    {
      "epoch": 0.02989,
      "grad_norm": 0.5734795816734266,
      "learning_rate": 0.003,
      "loss": 4.1903,
      "step": 2989
    },
    {
      "epoch": 0.0299,
      "grad_norm": 0.6108581797072585,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 2990
    },
    {
      "epoch": 0.02991,
      "grad_norm": 0.6465308189507616,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 2991
    },
    {
      "epoch": 0.02992,
      "grad_norm": 0.6024426991759066,
      "learning_rate": 0.003,
      "loss": 4.1963,
      "step": 2992
    },
    {
      "epoch": 0.02993,
      "grad_norm": 0.5450535416942648,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 2993
    },
    {
      "epoch": 0.02994,
      "grad_norm": 0.5271060232078253,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 2994
    },
    {
      "epoch": 0.02995,
      "grad_norm": 0.45562729812821884,
      "learning_rate": 0.003,
      "loss": 4.2083,
      "step": 2995
    },
    {
      "epoch": 0.02996,
      "grad_norm": 0.460767745641914,
      "learning_rate": 0.003,
      "loss": 4.1977,
      "step": 2996
    },
    {
      "epoch": 0.02997,
      "grad_norm": 0.49883686701327856,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 2997
    },
    {
      "epoch": 0.02998,
      "grad_norm": 0.5495557644619213,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 2998
    },
    {
      "epoch": 0.02999,
      "grad_norm": 0.6523752894410377,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 2999
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7129180858865645,
      "learning_rate": 0.003,
      "loss": 4.2061,
      "step": 3000
    },
    {
      "epoch": 0.03001,
      "grad_norm": 0.7274677458257305,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3001
    },
    {
      "epoch": 0.03002,
      "grad_norm": 0.6467948184705592,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 3002
    },
    {
      "epoch": 0.03003,
      "grad_norm": 0.6600036116283045,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 3003
    },
    {
      "epoch": 0.03004,
      "grad_norm": 0.7176122662961554,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 3004
    },
    {
      "epoch": 0.03005,
      "grad_norm": 0.6686551022549297,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 3005
    },
    {
      "epoch": 0.03006,
      "grad_norm": 0.7091833087612831,
      "learning_rate": 0.003,
      "loss": 4.1949,
      "step": 3006
    },
    {
      "epoch": 0.03007,
      "grad_norm": 0.6742803685976605,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 3007
    },
    {
      "epoch": 0.03008,
      "grad_norm": 0.6530504414551399,
      "learning_rate": 0.003,
      "loss": 4.1926,
      "step": 3008
    },
    {
      "epoch": 0.03009,
      "grad_norm": 0.7147134856842983,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 3009
    },
    {
      "epoch": 0.0301,
      "grad_norm": 0.803723268068343,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 3010
    },
    {
      "epoch": 0.03011,
      "grad_norm": 0.8381383897634889,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 3011
    },
    {
      "epoch": 0.03012,
      "grad_norm": 0.6693037897815421,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 3012
    },
    {
      "epoch": 0.03013,
      "grad_norm": 0.5695069096544523,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 3013
    },
    {
      "epoch": 0.03014,
      "grad_norm": 0.5914912051432484,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 3014
    },
    {
      "epoch": 0.03015,
      "grad_norm": 0.6342676904398465,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 3015
    },
    {
      "epoch": 0.03016,
      "grad_norm": 0.643666546817278,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 3016
    },
    {
      "epoch": 0.03017,
      "grad_norm": 0.6248622988385808,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 3017
    },
    {
      "epoch": 0.03018,
      "grad_norm": 0.5659743586845699,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3018
    },
    {
      "epoch": 0.03019,
      "grad_norm": 0.543953415462997,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 3019
    },
    {
      "epoch": 0.0302,
      "grad_norm": 0.4979549215791465,
      "learning_rate": 0.003,
      "loss": 4.171,
      "step": 3020
    },
    {
      "epoch": 0.03021,
      "grad_norm": 0.5988839209053217,
      "learning_rate": 0.003,
      "loss": 4.1986,
      "step": 3021
    },
    {
      "epoch": 0.03022,
      "grad_norm": 0.6917001050231241,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 3022
    },
    {
      "epoch": 0.03023,
      "grad_norm": 0.767246802866098,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 3023
    },
    {
      "epoch": 0.03024,
      "grad_norm": 0.7171193879296623,
      "learning_rate": 0.003,
      "loss": 4.2081,
      "step": 3024
    },
    {
      "epoch": 0.03025,
      "grad_norm": 0.8041231345280121,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 3025
    },
    {
      "epoch": 0.03026,
      "grad_norm": 0.8792691727950304,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 3026
    },
    {
      "epoch": 0.03027,
      "grad_norm": 1.0175395228935675,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 3027
    },
    {
      "epoch": 0.03028,
      "grad_norm": 0.923391249014174,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 3028
    },
    {
      "epoch": 0.03029,
      "grad_norm": 0.7571617694978044,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 3029
    },
    {
      "epoch": 0.0303,
      "grad_norm": 0.8910919036216922,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 3030
    },
    {
      "epoch": 0.03031,
      "grad_norm": 0.9641578814377423,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 3031
    },
    {
      "epoch": 0.03032,
      "grad_norm": 0.9409295179057755,
      "learning_rate": 0.003,
      "loss": 4.2025,
      "step": 3032
    },
    {
      "epoch": 0.03033,
      "grad_norm": 0.8317283746521951,
      "learning_rate": 0.003,
      "loss": 4.1969,
      "step": 3033
    },
    {
      "epoch": 0.03034,
      "grad_norm": 0.7313668328730979,
      "learning_rate": 0.003,
      "loss": 4.1976,
      "step": 3034
    },
    {
      "epoch": 0.03035,
      "grad_norm": 0.7132789456809936,
      "learning_rate": 0.003,
      "loss": 4.2248,
      "step": 3035
    },
    {
      "epoch": 0.03036,
      "grad_norm": 0.674714029441459,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 3036
    },
    {
      "epoch": 0.03037,
      "grad_norm": 0.7009049198448731,
      "learning_rate": 0.003,
      "loss": 4.1813,
      "step": 3037
    },
    {
      "epoch": 0.03038,
      "grad_norm": 0.6254153988955986,
      "learning_rate": 0.003,
      "loss": 4.1778,
      "step": 3038
    },
    {
      "epoch": 0.03039,
      "grad_norm": 0.6327013070531727,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 3039
    },
    {
      "epoch": 0.0304,
      "grad_norm": 0.6070726773833675,
      "learning_rate": 0.003,
      "loss": 4.1868,
      "step": 3040
    },
    {
      "epoch": 0.03041,
      "grad_norm": 0.5945854551887333,
      "learning_rate": 0.003,
      "loss": 4.1928,
      "step": 3041
    },
    {
      "epoch": 0.03042,
      "grad_norm": 0.5653702360336539,
      "learning_rate": 0.003,
      "loss": 4.2138,
      "step": 3042
    },
    {
      "epoch": 0.03043,
      "grad_norm": 0.48339928862065246,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 3043
    },
    {
      "epoch": 0.03044,
      "grad_norm": 0.4882765542102775,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 3044
    },
    {
      "epoch": 0.03045,
      "grad_norm": 0.49906211495716746,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 3045
    },
    {
      "epoch": 0.03046,
      "grad_norm": 0.6073778663358335,
      "learning_rate": 0.003,
      "loss": 4.1808,
      "step": 3046
    },
    {
      "epoch": 0.03047,
      "grad_norm": 0.767480218845333,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 3047
    },
    {
      "epoch": 0.03048,
      "grad_norm": 0.8141142563495186,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3048
    },
    {
      "epoch": 0.03049,
      "grad_norm": 0.882596084462819,
      "learning_rate": 0.003,
      "loss": 4.164,
      "step": 3049
    },
    {
      "epoch": 0.0305,
      "grad_norm": 0.872061418696622,
      "learning_rate": 0.003,
      "loss": 4.1847,
      "step": 3050
    },
    {
      "epoch": 0.03051,
      "grad_norm": 0.8096508372990807,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 3051
    },
    {
      "epoch": 0.03052,
      "grad_norm": 0.7731212144212846,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 3052
    },
    {
      "epoch": 0.03053,
      "grad_norm": 0.8121040926587938,
      "learning_rate": 0.003,
      "loss": 4.2108,
      "step": 3053
    },
    {
      "epoch": 0.03054,
      "grad_norm": 0.8031490148625648,
      "learning_rate": 0.003,
      "loss": 4.19,
      "step": 3054
    },
    {
      "epoch": 0.03055,
      "grad_norm": 0.7123837580650032,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 3055
    },
    {
      "epoch": 0.03056,
      "grad_norm": 0.6851652643314644,
      "learning_rate": 0.003,
      "loss": 4.2013,
      "step": 3056
    },
    {
      "epoch": 0.03057,
      "grad_norm": 0.7381209914846429,
      "learning_rate": 0.003,
      "loss": 4.205,
      "step": 3057
    },
    {
      "epoch": 0.03058,
      "grad_norm": 0.8318560657251862,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 3058
    },
    {
      "epoch": 0.03059,
      "grad_norm": 0.8791895081718305,
      "learning_rate": 0.003,
      "loss": 4.1917,
      "step": 3059
    },
    {
      "epoch": 0.0306,
      "grad_norm": 0.814947872922507,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 3060
    },
    {
      "epoch": 0.03061,
      "grad_norm": 0.6938901592564853,
      "learning_rate": 0.003,
      "loss": 4.1987,
      "step": 3061
    },
    {
      "epoch": 0.03062,
      "grad_norm": 0.7477036973476255,
      "learning_rate": 0.003,
      "loss": 4.225,
      "step": 3062
    },
    {
      "epoch": 0.03063,
      "grad_norm": 0.6602150387540897,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3063
    },
    {
      "epoch": 0.03064,
      "grad_norm": 0.588757114074522,
      "learning_rate": 0.003,
      "loss": 4.1973,
      "step": 3064
    },
    {
      "epoch": 0.03065,
      "grad_norm": 0.5129768984617096,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 3065
    },
    {
      "epoch": 0.03066,
      "grad_norm": 0.5610278353764739,
      "learning_rate": 0.003,
      "loss": 4.2055,
      "step": 3066
    },
    {
      "epoch": 0.03067,
      "grad_norm": 0.5604919558505147,
      "learning_rate": 0.003,
      "loss": 4.209,
      "step": 3067
    },
    {
      "epoch": 0.03068,
      "grad_norm": 0.6135705525861784,
      "learning_rate": 0.003,
      "loss": 4.1804,
      "step": 3068
    },
    {
      "epoch": 0.03069,
      "grad_norm": 0.6668599988517637,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 3069
    },
    {
      "epoch": 0.0307,
      "grad_norm": 0.7555455276952089,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 3070
    },
    {
      "epoch": 0.03071,
      "grad_norm": 0.6246930367493289,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3071
    },
    {
      "epoch": 0.03072,
      "grad_norm": 0.5918561396926248,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3072
    },
    {
      "epoch": 0.03073,
      "grad_norm": 0.543534573866796,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3073
    },
    {
      "epoch": 0.03074,
      "grad_norm": 0.5906611145957643,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3074
    },
    {
      "epoch": 0.03075,
      "grad_norm": 0.6400240036041446,
      "learning_rate": 0.003,
      "loss": 4.2013,
      "step": 3075
    },
    {
      "epoch": 0.03076,
      "grad_norm": 0.5704867962206898,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 3076
    },
    {
      "epoch": 0.03077,
      "grad_norm": 0.5207074704544816,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 3077
    },
    {
      "epoch": 0.03078,
      "grad_norm": 0.49735495203144253,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 3078
    },
    {
      "epoch": 0.03079,
      "grad_norm": 0.49371618460089295,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3079
    },
    {
      "epoch": 0.0308,
      "grad_norm": 0.5286026033951621,
      "learning_rate": 0.003,
      "loss": 4.1869,
      "step": 3080
    },
    {
      "epoch": 0.03081,
      "grad_norm": 0.597818150474852,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 3081
    },
    {
      "epoch": 0.03082,
      "grad_norm": 0.673672308591032,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 3082
    },
    {
      "epoch": 0.03083,
      "grad_norm": 0.742812911307247,
      "learning_rate": 0.003,
      "loss": 4.1694,
      "step": 3083
    },
    {
      "epoch": 0.03084,
      "grad_norm": 0.7219826879276637,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 3084
    },
    {
      "epoch": 0.03085,
      "grad_norm": 0.6340360256960204,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 3085
    },
    {
      "epoch": 0.03086,
      "grad_norm": 0.5692756028815859,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 3086
    },
    {
      "epoch": 0.03087,
      "grad_norm": 0.6389893757240679,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 3087
    },
    {
      "epoch": 0.03088,
      "grad_norm": 0.6896729441240512,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 3088
    },
    {
      "epoch": 0.03089,
      "grad_norm": 0.8131458592240662,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 3089
    },
    {
      "epoch": 0.0309,
      "grad_norm": 0.8512834574988407,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3090
    },
    {
      "epoch": 0.03091,
      "grad_norm": 0.8422191636508738,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 3091
    },
    {
      "epoch": 0.03092,
      "grad_norm": 0.7541794006345667,
      "learning_rate": 0.003,
      "loss": 4.1613,
      "step": 3092
    },
    {
      "epoch": 0.03093,
      "grad_norm": 0.6725158454989346,
      "learning_rate": 0.003,
      "loss": 4.1914,
      "step": 3093
    },
    {
      "epoch": 0.03094,
      "grad_norm": 0.7201284858059102,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3094
    },
    {
      "epoch": 0.03095,
      "grad_norm": 0.7621210210503538,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 3095
    },
    {
      "epoch": 0.03096,
      "grad_norm": 0.6578929196198599,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3096
    },
    {
      "epoch": 0.03097,
      "grad_norm": 0.695067390761494,
      "learning_rate": 0.003,
      "loss": 4.1922,
      "step": 3097
    },
    {
      "epoch": 0.03098,
      "grad_norm": 0.740033596710499,
      "learning_rate": 0.003,
      "loss": 4.1912,
      "step": 3098
    },
    {
      "epoch": 0.03099,
      "grad_norm": 0.7403836940802626,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 3099
    },
    {
      "epoch": 0.031,
      "grad_norm": 0.6979751976411313,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 3100
    },
    {
      "epoch": 0.03101,
      "grad_norm": 0.7670376808263826,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 3101
    },
    {
      "epoch": 0.03102,
      "grad_norm": 0.8040700995473709,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 3102
    },
    {
      "epoch": 0.03103,
      "grad_norm": 0.7605055478614147,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 3103
    },
    {
      "epoch": 0.03104,
      "grad_norm": 0.7299989579866449,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3104
    },
    {
      "epoch": 0.03105,
      "grad_norm": 0.8293979417721875,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 3105
    },
    {
      "epoch": 0.03106,
      "grad_norm": 0.7907107069059506,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3106
    },
    {
      "epoch": 0.03107,
      "grad_norm": 0.7368559192868307,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 3107
    },
    {
      "epoch": 0.03108,
      "grad_norm": 0.6702525929923716,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 3108
    },
    {
      "epoch": 0.03109,
      "grad_norm": 0.7199266096748919,
      "learning_rate": 0.003,
      "loss": 4.1721,
      "step": 3109
    },
    {
      "epoch": 0.0311,
      "grad_norm": 0.7526329246324333,
      "learning_rate": 0.003,
      "loss": 4.1926,
      "step": 3110
    },
    {
      "epoch": 0.03111,
      "grad_norm": 0.7398905284889996,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3111
    },
    {
      "epoch": 0.03112,
      "grad_norm": 0.8085890464082365,
      "learning_rate": 0.003,
      "loss": 4.1956,
      "step": 3112
    },
    {
      "epoch": 0.03113,
      "grad_norm": 0.8975031706869222,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3113
    },
    {
      "epoch": 0.03114,
      "grad_norm": 0.9143315414333046,
      "learning_rate": 0.003,
      "loss": 4.1926,
      "step": 3114
    },
    {
      "epoch": 0.03115,
      "grad_norm": 0.7984002320222364,
      "learning_rate": 0.003,
      "loss": 4.1932,
      "step": 3115
    },
    {
      "epoch": 0.03116,
      "grad_norm": 0.7251259112559764,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 3116
    },
    {
      "epoch": 0.03117,
      "grad_norm": 0.6765418867772506,
      "learning_rate": 0.003,
      "loss": 4.1803,
      "step": 3117
    },
    {
      "epoch": 0.03118,
      "grad_norm": 0.6564914593167945,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 3118
    },
    {
      "epoch": 0.03119,
      "grad_norm": 0.6469214981930856,
      "learning_rate": 0.003,
      "loss": 4.1951,
      "step": 3119
    },
    {
      "epoch": 0.0312,
      "grad_norm": 0.6004553747682227,
      "learning_rate": 0.003,
      "loss": 4.1991,
      "step": 3120
    },
    {
      "epoch": 0.03121,
      "grad_norm": 0.5447439216302157,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 3121
    },
    {
      "epoch": 0.03122,
      "grad_norm": 0.472628289298214,
      "learning_rate": 0.003,
      "loss": 4.1883,
      "step": 3122
    },
    {
      "epoch": 0.03123,
      "grad_norm": 0.4637134521023551,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3123
    },
    {
      "epoch": 0.03124,
      "grad_norm": 0.5031937627853148,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 3124
    },
    {
      "epoch": 0.03125,
      "grad_norm": 0.49290767185737955,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 3125
    },
    {
      "epoch": 0.03126,
      "grad_norm": 0.5046421633962421,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 3126
    },
    {
      "epoch": 0.03127,
      "grad_norm": 0.5295145475165267,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 3127
    },
    {
      "epoch": 0.03128,
      "grad_norm": 0.5768647951144659,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 3128
    },
    {
      "epoch": 0.03129,
      "grad_norm": 0.5526301834618427,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 3129
    },
    {
      "epoch": 0.0313,
      "grad_norm": 0.6096664293295351,
      "learning_rate": 0.003,
      "loss": 4.187,
      "step": 3130
    },
    {
      "epoch": 0.03131,
      "grad_norm": 0.6262440538187416,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 3131
    },
    {
      "epoch": 0.03132,
      "grad_norm": 0.6399356853738972,
      "learning_rate": 0.003,
      "loss": 4.1889,
      "step": 3132
    },
    {
      "epoch": 0.03133,
      "grad_norm": 0.6986536472969243,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3133
    },
    {
      "epoch": 0.03134,
      "grad_norm": 0.8533836264241755,
      "learning_rate": 0.003,
      "loss": 4.1841,
      "step": 3134
    },
    {
      "epoch": 0.03135,
      "grad_norm": 0.9857590467203682,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 3135
    },
    {
      "epoch": 0.03136,
      "grad_norm": 0.985916813895558,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 3136
    },
    {
      "epoch": 0.03137,
      "grad_norm": 0.7996019468603507,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3137
    },
    {
      "epoch": 0.03138,
      "grad_norm": 0.7409168937664911,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3138
    },
    {
      "epoch": 0.03139,
      "grad_norm": 0.6767699843822652,
      "learning_rate": 0.003,
      "loss": 4.1828,
      "step": 3139
    },
    {
      "epoch": 0.0314,
      "grad_norm": 0.7304920419704491,
      "learning_rate": 0.003,
      "loss": 4.1836,
      "step": 3140
    },
    {
      "epoch": 0.03141,
      "grad_norm": 0.8085325981037152,
      "learning_rate": 0.003,
      "loss": 4.2179,
      "step": 3141
    },
    {
      "epoch": 0.03142,
      "grad_norm": 0.7280300882309108,
      "learning_rate": 0.003,
      "loss": 4.1945,
      "step": 3142
    },
    {
      "epoch": 0.03143,
      "grad_norm": 0.723146974389316,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 3143
    },
    {
      "epoch": 0.03144,
      "grad_norm": 0.6065206122686737,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 3144
    },
    {
      "epoch": 0.03145,
      "grad_norm": 0.644934935501266,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3145
    },
    {
      "epoch": 0.03146,
      "grad_norm": 0.6592280348022013,
      "learning_rate": 0.003,
      "loss": 4.1869,
      "step": 3146
    },
    {
      "epoch": 0.03147,
      "grad_norm": 0.6530269899474325,
      "learning_rate": 0.003,
      "loss": 4.1855,
      "step": 3147
    },
    {
      "epoch": 0.03148,
      "grad_norm": 0.5745843025691852,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3148
    },
    {
      "epoch": 0.03149,
      "grad_norm": 0.6065668720311537,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 3149
    },
    {
      "epoch": 0.0315,
      "grad_norm": 0.6922007846823143,
      "learning_rate": 0.003,
      "loss": 4.1787,
      "step": 3150
    },
    {
      "epoch": 0.03151,
      "grad_norm": 0.7094328965622133,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 3151
    },
    {
      "epoch": 0.03152,
      "grad_norm": 0.6960957361399359,
      "learning_rate": 0.003,
      "loss": 4.1913,
      "step": 3152
    },
    {
      "epoch": 0.03153,
      "grad_norm": 0.7486079877329721,
      "learning_rate": 0.003,
      "loss": 4.1997,
      "step": 3153
    },
    {
      "epoch": 0.03154,
      "grad_norm": 0.7456103075414062,
      "learning_rate": 0.003,
      "loss": 4.1959,
      "step": 3154
    },
    {
      "epoch": 0.03155,
      "grad_norm": 0.8177011172560863,
      "learning_rate": 0.003,
      "loss": 4.1844,
      "step": 3155
    },
    {
      "epoch": 0.03156,
      "grad_norm": 0.8793941608963775,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3156
    },
    {
      "epoch": 0.03157,
      "grad_norm": 0.704857810319287,
      "learning_rate": 0.003,
      "loss": 4.1956,
      "step": 3157
    },
    {
      "epoch": 0.03158,
      "grad_norm": 0.6113967004695787,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 3158
    },
    {
      "epoch": 0.03159,
      "grad_norm": 0.66211174258146,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 3159
    },
    {
      "epoch": 0.0316,
      "grad_norm": 0.5523265028957371,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 3160
    },
    {
      "epoch": 0.03161,
      "grad_norm": 0.608325644395274,
      "learning_rate": 0.003,
      "loss": 4.1998,
      "step": 3161
    },
    {
      "epoch": 0.03162,
      "grad_norm": 0.6269366928420552,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 3162
    },
    {
      "epoch": 0.03163,
      "grad_norm": 0.6819002848904914,
      "learning_rate": 0.003,
      "loss": 4.188,
      "step": 3163
    },
    {
      "epoch": 0.03164,
      "grad_norm": 0.7746466933382178,
      "learning_rate": 0.003,
      "loss": 4.1862,
      "step": 3164
    },
    {
      "epoch": 0.03165,
      "grad_norm": 0.8342216139344243,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 3165
    },
    {
      "epoch": 0.03166,
      "grad_norm": 0.854621355905751,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 3166
    },
    {
      "epoch": 0.03167,
      "grad_norm": 0.9512159018248347,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 3167
    },
    {
      "epoch": 0.03168,
      "grad_norm": 0.9746218610383118,
      "learning_rate": 0.003,
      "loss": 4.2065,
      "step": 3168
    },
    {
      "epoch": 0.03169,
      "grad_norm": 0.9858598226166518,
      "learning_rate": 0.003,
      "loss": 4.224,
      "step": 3169
    },
    {
      "epoch": 0.0317,
      "grad_norm": 0.7498416277122048,
      "learning_rate": 0.003,
      "loss": 4.1691,
      "step": 3170
    },
    {
      "epoch": 0.03171,
      "grad_norm": 0.6044881164445568,
      "learning_rate": 0.003,
      "loss": 4.2016,
      "step": 3171
    },
    {
      "epoch": 0.03172,
      "grad_norm": 0.6815636176834087,
      "learning_rate": 0.003,
      "loss": 4.191,
      "step": 3172
    },
    {
      "epoch": 0.03173,
      "grad_norm": 0.6654525131928357,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 3173
    },
    {
      "epoch": 0.03174,
      "grad_norm": 0.6474897594742762,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3174
    },
    {
      "epoch": 0.03175,
      "grad_norm": 0.6418698031488217,
      "learning_rate": 0.003,
      "loss": 4.201,
      "step": 3175
    },
    {
      "epoch": 0.03176,
      "grad_norm": 0.6191051450233946,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 3176
    },
    {
      "epoch": 0.03177,
      "grad_norm": 0.5618627528537171,
      "learning_rate": 0.003,
      "loss": 4.1676,
      "step": 3177
    },
    {
      "epoch": 0.03178,
      "grad_norm": 0.574624051159771,
      "learning_rate": 0.003,
      "loss": 4.2133,
      "step": 3178
    },
    {
      "epoch": 0.03179,
      "grad_norm": 0.5170487183975983,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 3179
    },
    {
      "epoch": 0.0318,
      "grad_norm": 0.5443225922897362,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 3180
    },
    {
      "epoch": 0.03181,
      "grad_norm": 0.48857658279023347,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 3181
    },
    {
      "epoch": 0.03182,
      "grad_norm": 0.5003587458379335,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 3182
    },
    {
      "epoch": 0.03183,
      "grad_norm": 0.47427786201564304,
      "learning_rate": 0.003,
      "loss": 4.1791,
      "step": 3183
    },
    {
      "epoch": 0.03184,
      "grad_norm": 0.5056283615164667,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 3184
    },
    {
      "epoch": 0.03185,
      "grad_norm": 0.5400490284012653,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3185
    },
    {
      "epoch": 0.03186,
      "grad_norm": 0.6985528632077707,
      "learning_rate": 0.003,
      "loss": 4.1767,
      "step": 3186
    },
    {
      "epoch": 0.03187,
      "grad_norm": 0.8150902063241722,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 3187
    },
    {
      "epoch": 0.03188,
      "grad_norm": 0.9325360176643979,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 3188
    },
    {
      "epoch": 0.03189,
      "grad_norm": 0.8937715656838259,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 3189
    },
    {
      "epoch": 0.0319,
      "grad_norm": 0.744447678504765,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 3190
    },
    {
      "epoch": 0.03191,
      "grad_norm": 0.6776373550333729,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 3191
    },
    {
      "epoch": 0.03192,
      "grad_norm": 0.6440396042092965,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 3192
    },
    {
      "epoch": 0.03193,
      "grad_norm": 0.6355371339608338,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3193
    },
    {
      "epoch": 0.03194,
      "grad_norm": 0.6540570284226682,
      "learning_rate": 0.003,
      "loss": 4.1966,
      "step": 3194
    },
    {
      "epoch": 0.03195,
      "grad_norm": 0.5776431167458079,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 3195
    },
    {
      "epoch": 0.03196,
      "grad_norm": 0.5567486580718531,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 3196
    },
    {
      "epoch": 0.03197,
      "grad_norm": 0.6211522115218998,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 3197
    },
    {
      "epoch": 0.03198,
      "grad_norm": 0.6754236762792072,
      "learning_rate": 0.003,
      "loss": 4.1783,
      "step": 3198
    },
    {
      "epoch": 0.03199,
      "grad_norm": 0.6909028789552336,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 3199
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.6851130691205759,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3200
    },
    {
      "epoch": 0.03201,
      "grad_norm": 0.8236929229091875,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 3201
    },
    {
      "epoch": 0.03202,
      "grad_norm": 0.951709446477759,
      "learning_rate": 0.003,
      "loss": 4.1883,
      "step": 3202
    },
    {
      "epoch": 0.03203,
      "grad_norm": 0.9116506118668389,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 3203
    },
    {
      "epoch": 0.03204,
      "grad_norm": 0.737364092313657,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3204
    },
    {
      "epoch": 0.03205,
      "grad_norm": 0.5733762258500861,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3205
    },
    {
      "epoch": 0.03206,
      "grad_norm": 0.6520837940224375,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 3206
    },
    {
      "epoch": 0.03207,
      "grad_norm": 0.6713973023408082,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 3207
    },
    {
      "epoch": 0.03208,
      "grad_norm": 0.6638942407046395,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3208
    },
    {
      "epoch": 0.03209,
      "grad_norm": 0.6926100039707507,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 3209
    },
    {
      "epoch": 0.0321,
      "grad_norm": 0.7197536710037978,
      "learning_rate": 0.003,
      "loss": 4.1931,
      "step": 3210
    },
    {
      "epoch": 0.03211,
      "grad_norm": 0.7299656936668171,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 3211
    },
    {
      "epoch": 0.03212,
      "grad_norm": 0.6674134844073308,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 3212
    },
    {
      "epoch": 0.03213,
      "grad_norm": 0.602090480831783,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 3213
    },
    {
      "epoch": 0.03214,
      "grad_norm": 0.5632210286186823,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 3214
    },
    {
      "epoch": 0.03215,
      "grad_norm": 0.6424317594884178,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 3215
    },
    {
      "epoch": 0.03216,
      "grad_norm": 0.6946877458250511,
      "learning_rate": 0.003,
      "loss": 4.1642,
      "step": 3216
    },
    {
      "epoch": 0.03217,
      "grad_norm": 0.7162006314166186,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 3217
    },
    {
      "epoch": 0.03218,
      "grad_norm": 0.6688710116110083,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 3218
    },
    {
      "epoch": 0.03219,
      "grad_norm": 0.6849969993386218,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 3219
    },
    {
      "epoch": 0.0322,
      "grad_norm": 0.6568426784434412,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3220
    },
    {
      "epoch": 0.03221,
      "grad_norm": 0.7253483658822476,
      "learning_rate": 0.003,
      "loss": 4.1857,
      "step": 3221
    },
    {
      "epoch": 0.03222,
      "grad_norm": 0.793341013671458,
      "learning_rate": 0.003,
      "loss": 4.183,
      "step": 3222
    },
    {
      "epoch": 0.03223,
      "grad_norm": 0.824497677720901,
      "learning_rate": 0.003,
      "loss": 4.1862,
      "step": 3223
    },
    {
      "epoch": 0.03224,
      "grad_norm": 0.7552119583336456,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 3224
    },
    {
      "epoch": 0.03225,
      "grad_norm": 0.5734236212732641,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 3225
    },
    {
      "epoch": 0.03226,
      "grad_norm": 0.6220562562626234,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 3226
    },
    {
      "epoch": 0.03227,
      "grad_norm": 0.6820845746631659,
      "learning_rate": 0.003,
      "loss": 4.1783,
      "step": 3227
    },
    {
      "epoch": 0.03228,
      "grad_norm": 0.5933306652689198,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 3228
    },
    {
      "epoch": 0.03229,
      "grad_norm": 0.5119085736725962,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 3229
    },
    {
      "epoch": 0.0323,
      "grad_norm": 0.5310047730371823,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 3230
    },
    {
      "epoch": 0.03231,
      "grad_norm": 0.444458596160572,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3231
    },
    {
      "epoch": 0.03232,
      "grad_norm": 0.48863148670767037,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 3232
    },
    {
      "epoch": 0.03233,
      "grad_norm": 0.5890179829718989,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 3233
    },
    {
      "epoch": 0.03234,
      "grad_norm": 0.6507977897435386,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 3234
    },
    {
      "epoch": 0.03235,
      "grad_norm": 0.6735430538160202,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3235
    },
    {
      "epoch": 0.03236,
      "grad_norm": 0.6224436805872925,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3236
    },
    {
      "epoch": 0.03237,
      "grad_norm": 0.6153286251420005,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 3237
    },
    {
      "epoch": 0.03238,
      "grad_norm": 0.5653843948515308,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 3238
    },
    {
      "epoch": 0.03239,
      "grad_norm": 0.5579474828935078,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 3239
    },
    {
      "epoch": 0.0324,
      "grad_norm": 0.5628889716808217,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 3240
    },
    {
      "epoch": 0.03241,
      "grad_norm": 0.618198116761749,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3241
    },
    {
      "epoch": 0.03242,
      "grad_norm": 0.7766046798616605,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 3242
    },
    {
      "epoch": 0.03243,
      "grad_norm": 0.9785268652346966,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 3243
    },
    {
      "epoch": 0.03244,
      "grad_norm": 0.9580973048940927,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3244
    },
    {
      "epoch": 0.03245,
      "grad_norm": 0.8576334375089313,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 3245
    },
    {
      "epoch": 0.03246,
      "grad_norm": 0.8423468581019681,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 3246
    },
    {
      "epoch": 0.03247,
      "grad_norm": 0.9544112800192335,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 3247
    },
    {
      "epoch": 0.03248,
      "grad_norm": 1.108974194257081,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 3248
    },
    {
      "epoch": 0.03249,
      "grad_norm": 0.922677655020294,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 3249
    },
    {
      "epoch": 0.0325,
      "grad_norm": 0.87283028555209,
      "learning_rate": 0.003,
      "loss": 4.2167,
      "step": 3250
    },
    {
      "epoch": 0.03251,
      "grad_norm": 0.9163748774315567,
      "learning_rate": 0.003,
      "loss": 4.2201,
      "step": 3251
    },
    {
      "epoch": 0.03252,
      "grad_norm": 0.81456143100221,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 3252
    },
    {
      "epoch": 0.03253,
      "grad_norm": 0.7968908258913222,
      "learning_rate": 0.003,
      "loss": 4.1861,
      "step": 3253
    },
    {
      "epoch": 0.03254,
      "grad_norm": 0.7235050634300025,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 3254
    },
    {
      "epoch": 0.03255,
      "grad_norm": 0.6266440712092289,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 3255
    },
    {
      "epoch": 0.03256,
      "grad_norm": 0.6449674375686789,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3256
    },
    {
      "epoch": 0.03257,
      "grad_norm": 0.5622485206306796,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 3257
    },
    {
      "epoch": 0.03258,
      "grad_norm": 0.5825851281947257,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 3258
    },
    {
      "epoch": 0.03259,
      "grad_norm": 0.6444343388033511,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 3259
    },
    {
      "epoch": 0.0326,
      "grad_norm": 0.6989193151493163,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 3260
    },
    {
      "epoch": 0.03261,
      "grad_norm": 0.8302855362092207,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 3261
    },
    {
      "epoch": 0.03262,
      "grad_norm": 0.9498523114829631,
      "learning_rate": 0.003,
      "loss": 4.1887,
      "step": 3262
    },
    {
      "epoch": 0.03263,
      "grad_norm": 0.983071934935445,
      "learning_rate": 0.003,
      "loss": 4.2007,
      "step": 3263
    },
    {
      "epoch": 0.03264,
      "grad_norm": 0.8331337315585593,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 3264
    },
    {
      "epoch": 0.03265,
      "grad_norm": 0.7498656994034634,
      "learning_rate": 0.003,
      "loss": 4.2003,
      "step": 3265
    },
    {
      "epoch": 0.03266,
      "grad_norm": 0.7807040232610372,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 3266
    },
    {
      "epoch": 0.03267,
      "grad_norm": 0.8713471135209633,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3267
    },
    {
      "epoch": 0.03268,
      "grad_norm": 0.8995394972894192,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 3268
    },
    {
      "epoch": 0.03269,
      "grad_norm": 0.7799010442197561,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 3269
    },
    {
      "epoch": 0.0327,
      "grad_norm": 0.6333137664038297,
      "learning_rate": 0.003,
      "loss": 4.2075,
      "step": 3270
    },
    {
      "epoch": 0.03271,
      "grad_norm": 0.5431600237872565,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 3271
    },
    {
      "epoch": 0.03272,
      "grad_norm": 0.5431026489790501,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3272
    },
    {
      "epoch": 0.03273,
      "grad_norm": 0.45854609240948824,
      "learning_rate": 0.003,
      "loss": 4.1729,
      "step": 3273
    },
    {
      "epoch": 0.03274,
      "grad_norm": 0.44688508931158466,
      "learning_rate": 0.003,
      "loss": 4.1868,
      "step": 3274
    },
    {
      "epoch": 0.03275,
      "grad_norm": 0.45373894096093864,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3275
    },
    {
      "epoch": 0.03276,
      "grad_norm": 0.41006589936140064,
      "learning_rate": 0.003,
      "loss": 4.1613,
      "step": 3276
    },
    {
      "epoch": 0.03277,
      "grad_norm": 0.38524925786232783,
      "learning_rate": 0.003,
      "loss": 4.1604,
      "step": 3277
    },
    {
      "epoch": 0.03278,
      "grad_norm": 0.35950981826812556,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3278
    },
    {
      "epoch": 0.03279,
      "grad_norm": 0.38292545999031286,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 3279
    },
    {
      "epoch": 0.0328,
      "grad_norm": 0.4242164622322683,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 3280
    },
    {
      "epoch": 0.03281,
      "grad_norm": 0.488691641648266,
      "learning_rate": 0.003,
      "loss": 4.1888,
      "step": 3281
    },
    {
      "epoch": 0.03282,
      "grad_norm": 0.5739177249611346,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 3282
    },
    {
      "epoch": 0.03283,
      "grad_norm": 0.554048169102402,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 3283
    },
    {
      "epoch": 0.03284,
      "grad_norm": 0.4915028252102749,
      "learning_rate": 0.003,
      "loss": 4.1816,
      "step": 3284
    },
    {
      "epoch": 0.03285,
      "grad_norm": 0.5580359822634506,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 3285
    },
    {
      "epoch": 0.03286,
      "grad_norm": 0.6079781410505047,
      "learning_rate": 0.003,
      "loss": 4.1723,
      "step": 3286
    },
    {
      "epoch": 0.03287,
      "grad_norm": 0.5821835778978477,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 3287
    },
    {
      "epoch": 0.03288,
      "grad_norm": 0.5438338131339984,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3288
    },
    {
      "epoch": 0.03289,
      "grad_norm": 0.570468604964685,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 3289
    },
    {
      "epoch": 0.0329,
      "grad_norm": 0.6096925395272201,
      "learning_rate": 0.003,
      "loss": 4.1931,
      "step": 3290
    },
    {
      "epoch": 0.03291,
      "grad_norm": 0.7214109941856076,
      "learning_rate": 0.003,
      "loss": 4.1807,
      "step": 3291
    },
    {
      "epoch": 0.03292,
      "grad_norm": 0.9305508295977749,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 3292
    },
    {
      "epoch": 0.03293,
      "grad_norm": 0.9205488988397595,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 3293
    },
    {
      "epoch": 0.03294,
      "grad_norm": 0.8983261374723139,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 3294
    },
    {
      "epoch": 0.03295,
      "grad_norm": 0.9704372414132879,
      "learning_rate": 0.003,
      "loss": 4.1739,
      "step": 3295
    },
    {
      "epoch": 0.03296,
      "grad_norm": 1.4039738619774527,
      "learning_rate": 0.003,
      "loss": 4.1989,
      "step": 3296
    },
    {
      "epoch": 0.03297,
      "grad_norm": 0.8478482918673438,
      "learning_rate": 0.003,
      "loss": 4.188,
      "step": 3297
    },
    {
      "epoch": 0.03298,
      "grad_norm": 0.8625233369451641,
      "learning_rate": 0.003,
      "loss": 4.2085,
      "step": 3298
    },
    {
      "epoch": 0.03299,
      "grad_norm": 0.946434504765274,
      "learning_rate": 0.003,
      "loss": 4.201,
      "step": 3299
    },
    {
      "epoch": 0.033,
      "grad_norm": 0.9408833966323229,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 3300
    },
    {
      "epoch": 0.03301,
      "grad_norm": 1.0399735000004064,
      "learning_rate": 0.003,
      "loss": 4.2331,
      "step": 3301
    },
    {
      "epoch": 0.03302,
      "grad_norm": 1.060889182219052,
      "learning_rate": 0.003,
      "loss": 4.2111,
      "step": 3302
    },
    {
      "epoch": 0.03303,
      "grad_norm": 0.910480011898513,
      "learning_rate": 0.003,
      "loss": 4.2117,
      "step": 3303
    },
    {
      "epoch": 0.03304,
      "grad_norm": 0.9051761770071215,
      "learning_rate": 0.003,
      "loss": 4.2061,
      "step": 3304
    },
    {
      "epoch": 0.03305,
      "grad_norm": 0.8593496427676475,
      "learning_rate": 0.003,
      "loss": 4.2046,
      "step": 3305
    },
    {
      "epoch": 0.03306,
      "grad_norm": 0.788406787032699,
      "learning_rate": 0.003,
      "loss": 4.2208,
      "step": 3306
    },
    {
      "epoch": 0.03307,
      "grad_norm": 0.6264143830894844,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 3307
    },
    {
      "epoch": 0.03308,
      "grad_norm": 0.5647145644735359,
      "learning_rate": 0.003,
      "loss": 4.1897,
      "step": 3308
    },
    {
      "epoch": 0.03309,
      "grad_norm": 0.5464103710288922,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 3309
    },
    {
      "epoch": 0.0331,
      "grad_norm": 0.4958683830756014,
      "learning_rate": 0.003,
      "loss": 4.1767,
      "step": 3310
    },
    {
      "epoch": 0.03311,
      "grad_norm": 0.4522879146976332,
      "learning_rate": 0.003,
      "loss": 4.1941,
      "step": 3311
    },
    {
      "epoch": 0.03312,
      "grad_norm": 0.4414886343825775,
      "learning_rate": 0.003,
      "loss": 4.1614,
      "step": 3312
    },
    {
      "epoch": 0.03313,
      "grad_norm": 0.4255185332999733,
      "learning_rate": 0.003,
      "loss": 4.1887,
      "step": 3313
    },
    {
      "epoch": 0.03314,
      "grad_norm": 0.4381607158366914,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 3314
    },
    {
      "epoch": 0.03315,
      "grad_norm": 0.42262937541001916,
      "learning_rate": 0.003,
      "loss": 4.1816,
      "step": 3315
    },
    {
      "epoch": 0.03316,
      "grad_norm": 0.4970426554871505,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 3316
    },
    {
      "epoch": 0.03317,
      "grad_norm": 0.5627776265011807,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 3317
    },
    {
      "epoch": 0.03318,
      "grad_norm": 0.6893835679497516,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 3318
    },
    {
      "epoch": 0.03319,
      "grad_norm": 0.7468347673987352,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 3319
    },
    {
      "epoch": 0.0332,
      "grad_norm": 0.8688918354050132,
      "learning_rate": 0.003,
      "loss": 4.1819,
      "step": 3320
    },
    {
      "epoch": 0.03321,
      "grad_norm": 0.9441155117884158,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 3321
    },
    {
      "epoch": 0.03322,
      "grad_norm": 0.8347403031059345,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 3322
    },
    {
      "epoch": 0.03323,
      "grad_norm": 0.7804052129413228,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 3323
    },
    {
      "epoch": 0.03324,
      "grad_norm": 0.7276937121862084,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3324
    },
    {
      "epoch": 0.03325,
      "grad_norm": 0.7364714952754682,
      "learning_rate": 0.003,
      "loss": 4.1961,
      "step": 3325
    },
    {
      "epoch": 0.03326,
      "grad_norm": 0.6138475460467973,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 3326
    },
    {
      "epoch": 0.03327,
      "grad_norm": 0.5484242765422025,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3327
    },
    {
      "epoch": 0.03328,
      "grad_norm": 0.5021334597809965,
      "learning_rate": 0.003,
      "loss": 4.1949,
      "step": 3328
    },
    {
      "epoch": 0.03329,
      "grad_norm": 0.531052392624268,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 3329
    },
    {
      "epoch": 0.0333,
      "grad_norm": 0.5172636277516343,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 3330
    },
    {
      "epoch": 0.03331,
      "grad_norm": 0.5347276948931869,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 3331
    },
    {
      "epoch": 0.03332,
      "grad_norm": 0.5266769298617034,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 3332
    },
    {
      "epoch": 0.03333,
      "grad_norm": 0.5119624308137339,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 3333
    },
    {
      "epoch": 0.03334,
      "grad_norm": 0.5227592678561254,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 3334
    },
    {
      "epoch": 0.03335,
      "grad_norm": 0.5475154089666096,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 3335
    },
    {
      "epoch": 0.03336,
      "grad_norm": 0.6301047632403133,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 3336
    },
    {
      "epoch": 0.03337,
      "grad_norm": 0.8194265063356702,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 3337
    },
    {
      "epoch": 0.03338,
      "grad_norm": 1.1445574176004276,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 3338
    },
    {
      "epoch": 0.03339,
      "grad_norm": 0.8189097928363258,
      "learning_rate": 0.003,
      "loss": 4.1787,
      "step": 3339
    },
    {
      "epoch": 0.0334,
      "grad_norm": 0.660449401420429,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 3340
    },
    {
      "epoch": 0.03341,
      "grad_norm": 0.7754430989891157,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3341
    },
    {
      "epoch": 0.03342,
      "grad_norm": 0.8792442379704091,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 3342
    },
    {
      "epoch": 0.03343,
      "grad_norm": 0.8167670859904461,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 3343
    },
    {
      "epoch": 0.03344,
      "grad_norm": 0.7480631734050127,
      "learning_rate": 0.003,
      "loss": 4.1936,
      "step": 3344
    },
    {
      "epoch": 0.03345,
      "grad_norm": 0.6496750703892737,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 3345
    },
    {
      "epoch": 0.03346,
      "grad_norm": 0.6081719260800688,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 3346
    },
    {
      "epoch": 0.03347,
      "grad_norm": 0.6721063598583025,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3347
    },
    {
      "epoch": 0.03348,
      "grad_norm": 0.6283159116392718,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 3348
    },
    {
      "epoch": 0.03349,
      "grad_norm": 0.5798454168131916,
      "learning_rate": 0.003,
      "loss": 4.2006,
      "step": 3349
    },
    {
      "epoch": 0.0335,
      "grad_norm": 0.588557434396112,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 3350
    },
    {
      "epoch": 0.03351,
      "grad_norm": 0.572435380329449,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 3351
    },
    {
      "epoch": 0.03352,
      "grad_norm": 0.5946129998027531,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 3352
    },
    {
      "epoch": 0.03353,
      "grad_norm": 0.6650155372083263,
      "learning_rate": 0.003,
      "loss": 4.1783,
      "step": 3353
    },
    {
      "epoch": 0.03354,
      "grad_norm": 0.6819620362483191,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 3354
    },
    {
      "epoch": 0.03355,
      "grad_norm": 0.7239397479894265,
      "learning_rate": 0.003,
      "loss": 4.1701,
      "step": 3355
    },
    {
      "epoch": 0.03356,
      "grad_norm": 0.7696928848130231,
      "learning_rate": 0.003,
      "loss": 4.197,
      "step": 3356
    },
    {
      "epoch": 0.03357,
      "grad_norm": 0.6661695560767795,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 3357
    },
    {
      "epoch": 0.03358,
      "grad_norm": 0.6123082718814552,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 3358
    },
    {
      "epoch": 0.03359,
      "grad_norm": 0.770393140263048,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 3359
    },
    {
      "epoch": 0.0336,
      "grad_norm": 0.8303005904862942,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 3360
    },
    {
      "epoch": 0.03361,
      "grad_norm": 0.9034060736336073,
      "learning_rate": 0.003,
      "loss": 4.1953,
      "step": 3361
    },
    {
      "epoch": 0.03362,
      "grad_norm": 0.8488627855760285,
      "learning_rate": 0.003,
      "loss": 4.1936,
      "step": 3362
    },
    {
      "epoch": 0.03363,
      "grad_norm": 0.8500592898597722,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 3363
    },
    {
      "epoch": 0.03364,
      "grad_norm": 0.8053495290807956,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3364
    },
    {
      "epoch": 0.03365,
      "grad_norm": 0.6918565332600101,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 3365
    },
    {
      "epoch": 0.03366,
      "grad_norm": 0.6631827092506345,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 3366
    },
    {
      "epoch": 0.03367,
      "grad_norm": 0.5898333237323294,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 3367
    },
    {
      "epoch": 0.03368,
      "grad_norm": 0.5105692997315376,
      "learning_rate": 0.003,
      "loss": 4.1944,
      "step": 3368
    },
    {
      "epoch": 0.03369,
      "grad_norm": 0.4833179204197231,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 3369
    },
    {
      "epoch": 0.0337,
      "grad_norm": 0.4434936266579207,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 3370
    },
    {
      "epoch": 0.03371,
      "grad_norm": 0.43752326668000696,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 3371
    },
    {
      "epoch": 0.03372,
      "grad_norm": 0.5212167164579171,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 3372
    },
    {
      "epoch": 0.03373,
      "grad_norm": 0.5782660861068077,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 3373
    },
    {
      "epoch": 0.03374,
      "grad_norm": 0.6248054334937597,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3374
    },
    {
      "epoch": 0.03375,
      "grad_norm": 0.5656089592222998,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 3375
    },
    {
      "epoch": 0.03376,
      "grad_norm": 0.533250430480493,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 3376
    },
    {
      "epoch": 0.03377,
      "grad_norm": 0.6893576110724466,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3377
    },
    {
      "epoch": 0.03378,
      "grad_norm": 0.8724259510840375,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 3378
    },
    {
      "epoch": 0.03379,
      "grad_norm": 1.0969642859125113,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 3379
    },
    {
      "epoch": 0.0338,
      "grad_norm": 0.9222105012112505,
      "learning_rate": 0.003,
      "loss": 4.1929,
      "step": 3380
    },
    {
      "epoch": 0.03381,
      "grad_norm": 0.6636117483340874,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 3381
    },
    {
      "epoch": 0.03382,
      "grad_norm": 0.6448818843218406,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 3382
    },
    {
      "epoch": 0.03383,
      "grad_norm": 0.7558871316698539,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 3383
    },
    {
      "epoch": 0.03384,
      "grad_norm": 0.784895312607238,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3384
    },
    {
      "epoch": 0.03385,
      "grad_norm": 0.7068746569705847,
      "learning_rate": 0.003,
      "loss": 4.1685,
      "step": 3385
    },
    {
      "epoch": 0.03386,
      "grad_norm": 0.5763966935533144,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 3386
    },
    {
      "epoch": 0.03387,
      "grad_norm": 0.5637260656795919,
      "learning_rate": 0.003,
      "loss": 4.1642,
      "step": 3387
    },
    {
      "epoch": 0.03388,
      "grad_norm": 0.5764694412310111,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3388
    },
    {
      "epoch": 0.03389,
      "grad_norm": 0.5063857748030218,
      "learning_rate": 0.003,
      "loss": 4.1917,
      "step": 3389
    },
    {
      "epoch": 0.0339,
      "grad_norm": 0.48268849043201856,
      "learning_rate": 0.003,
      "loss": 4.1773,
      "step": 3390
    },
    {
      "epoch": 0.03391,
      "grad_norm": 0.46369864578553305,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 3391
    },
    {
      "epoch": 0.03392,
      "grad_norm": 0.44960224345626854,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 3392
    },
    {
      "epoch": 0.03393,
      "grad_norm": 0.4909883361811901,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 3393
    },
    {
      "epoch": 0.03394,
      "grad_norm": 0.6259272860691563,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 3394
    },
    {
      "epoch": 0.03395,
      "grad_norm": 0.7506927184483131,
      "learning_rate": 0.003,
      "loss": 4.1835,
      "step": 3395
    },
    {
      "epoch": 0.03396,
      "grad_norm": 0.9370743209078207,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 3396
    },
    {
      "epoch": 0.03397,
      "grad_norm": 1.1783116967146359,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3397
    },
    {
      "epoch": 0.03398,
      "grad_norm": 0.8089632724306486,
      "learning_rate": 0.003,
      "loss": 4.1941,
      "step": 3398
    },
    {
      "epoch": 0.03399,
      "grad_norm": 0.7596909526208954,
      "learning_rate": 0.003,
      "loss": 4.2163,
      "step": 3399
    },
    {
      "epoch": 0.034,
      "grad_norm": 0.8562517296822207,
      "learning_rate": 0.003,
      "loss": 4.2009,
      "step": 3400
    },
    {
      "epoch": 0.03401,
      "grad_norm": 1.0068688368658953,
      "learning_rate": 0.003,
      "loss": 4.188,
      "step": 3401
    },
    {
      "epoch": 0.03402,
      "grad_norm": 0.99343979847827,
      "learning_rate": 0.003,
      "loss": 4.2136,
      "step": 3402
    },
    {
      "epoch": 0.03403,
      "grad_norm": 1.1467610890884863,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 3403
    },
    {
      "epoch": 0.03404,
      "grad_norm": 1.0385468087744694,
      "learning_rate": 0.003,
      "loss": 4.2248,
      "step": 3404
    },
    {
      "epoch": 0.03405,
      "grad_norm": 0.9754588896396058,
      "learning_rate": 0.003,
      "loss": 4.2094,
      "step": 3405
    },
    {
      "epoch": 0.03406,
      "grad_norm": 1.0318386962627828,
      "learning_rate": 0.003,
      "loss": 4.193,
      "step": 3406
    },
    {
      "epoch": 0.03407,
      "grad_norm": 1.0407604591460764,
      "learning_rate": 0.003,
      "loss": 4.2114,
      "step": 3407
    },
    {
      "epoch": 0.03408,
      "grad_norm": 0.9991368525042554,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 3408
    },
    {
      "epoch": 0.03409,
      "grad_norm": 0.9201374434882152,
      "learning_rate": 0.003,
      "loss": 4.2017,
      "step": 3409
    },
    {
      "epoch": 0.0341,
      "grad_norm": 0.7897132061669699,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 3410
    },
    {
      "epoch": 0.03411,
      "grad_norm": 0.8111203199333381,
      "learning_rate": 0.003,
      "loss": 4.2024,
      "step": 3411
    },
    {
      "epoch": 0.03412,
      "grad_norm": 0.7752208914216027,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 3412
    },
    {
      "epoch": 0.03413,
      "grad_norm": 0.8018027400540946,
      "learning_rate": 0.003,
      "loss": 4.2281,
      "step": 3413
    },
    {
      "epoch": 0.03414,
      "grad_norm": 0.7248771456875713,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 3414
    },
    {
      "epoch": 0.03415,
      "grad_norm": 0.6901233244425752,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 3415
    },
    {
      "epoch": 0.03416,
      "grad_norm": 0.6847318566109114,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 3416
    },
    {
      "epoch": 0.03417,
      "grad_norm": 0.6799194082415111,
      "learning_rate": 0.003,
      "loss": 4.1883,
      "step": 3417
    },
    {
      "epoch": 0.03418,
      "grad_norm": 0.6775692861547594,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3418
    },
    {
      "epoch": 0.03419,
      "grad_norm": 0.6946056576603092,
      "learning_rate": 0.003,
      "loss": 4.1802,
      "step": 3419
    },
    {
      "epoch": 0.0342,
      "grad_norm": 0.7015136907830581,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 3420
    },
    {
      "epoch": 0.03421,
      "grad_norm": 0.8322618875343657,
      "learning_rate": 0.003,
      "loss": 4.184,
      "step": 3421
    },
    {
      "epoch": 0.03422,
      "grad_norm": 0.877410615683893,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 3422
    },
    {
      "epoch": 0.03423,
      "grad_norm": 0.7815901156339347,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3423
    },
    {
      "epoch": 0.03424,
      "grad_norm": 0.7461859372495792,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 3424
    },
    {
      "epoch": 0.03425,
      "grad_norm": 0.6612154502175626,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 3425
    },
    {
      "epoch": 0.03426,
      "grad_norm": 0.5350775915801592,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3426
    },
    {
      "epoch": 0.03427,
      "grad_norm": 0.5758126205992435,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 3427
    },
    {
      "epoch": 0.03428,
      "grad_norm": 0.57747553400605,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 3428
    },
    {
      "epoch": 0.03429,
      "grad_norm": 0.6495727211029925,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 3429
    },
    {
      "epoch": 0.0343,
      "grad_norm": 0.6970333948744754,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3430
    },
    {
      "epoch": 0.03431,
      "grad_norm": 0.7157526291407539,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 3431
    },
    {
      "epoch": 0.03432,
      "grad_norm": 0.6010910249067187,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 3432
    },
    {
      "epoch": 0.03433,
      "grad_norm": 0.42007320422820055,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 3433
    },
    {
      "epoch": 0.03434,
      "grad_norm": 0.47622504855038245,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 3434
    },
    {
      "epoch": 0.03435,
      "grad_norm": 0.5574984023909559,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 3435
    },
    {
      "epoch": 0.03436,
      "grad_norm": 0.6117380821653216,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 3436
    },
    {
      "epoch": 0.03437,
      "grad_norm": 0.5586731673929375,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 3437
    },
    {
      "epoch": 0.03438,
      "grad_norm": 0.41091865924021914,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 3438
    },
    {
      "epoch": 0.03439,
      "grad_norm": 0.3844506479299855,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 3439
    },
    {
      "epoch": 0.0344,
      "grad_norm": 0.47026248688155026,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 3440
    },
    {
      "epoch": 0.03441,
      "grad_norm": 0.49960141515204526,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 3441
    },
    {
      "epoch": 0.03442,
      "grad_norm": 0.48160504106231705,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 3442
    },
    {
      "epoch": 0.03443,
      "grad_norm": 0.4503151278381613,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 3443
    },
    {
      "epoch": 0.03444,
      "grad_norm": 0.4618551136735495,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 3444
    },
    {
      "epoch": 0.03445,
      "grad_norm": 0.5002612465674857,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 3445
    },
    {
      "epoch": 0.03446,
      "grad_norm": 0.49371947051027193,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 3446
    },
    {
      "epoch": 0.03447,
      "grad_norm": 0.48167624788075264,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 3447
    },
    {
      "epoch": 0.03448,
      "grad_norm": 0.48645338009833317,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 3448
    },
    {
      "epoch": 0.03449,
      "grad_norm": 0.42185936569719606,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 3449
    },
    {
      "epoch": 0.0345,
      "grad_norm": 0.39079332947086687,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 3450
    },
    {
      "epoch": 0.03451,
      "grad_norm": 0.42129819422128534,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 3451
    },
    {
      "epoch": 0.03452,
      "grad_norm": 0.5355871931302184,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3452
    },
    {
      "epoch": 0.03453,
      "grad_norm": 0.7320917006665237,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 3453
    },
    {
      "epoch": 0.03454,
      "grad_norm": 1.1401386064666243,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 3454
    },
    {
      "epoch": 0.03455,
      "grad_norm": 1.05569726012347,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 3455
    },
    {
      "epoch": 0.03456,
      "grad_norm": 0.7753833366744568,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 3456
    },
    {
      "epoch": 0.03457,
      "grad_norm": 0.8398952536178134,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 3457
    },
    {
      "epoch": 0.03458,
      "grad_norm": 0.9216186608796643,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 3458
    },
    {
      "epoch": 0.03459,
      "grad_norm": 0.8741651793984747,
      "learning_rate": 0.003,
      "loss": 4.1982,
      "step": 3459
    },
    {
      "epoch": 0.0346,
      "grad_norm": 0.7072954882447885,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 3460
    },
    {
      "epoch": 0.03461,
      "grad_norm": 0.6974866321122205,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3461
    },
    {
      "epoch": 0.03462,
      "grad_norm": 0.6828331664556785,
      "learning_rate": 0.003,
      "loss": 4.1676,
      "step": 3462
    },
    {
      "epoch": 0.03463,
      "grad_norm": 0.8097332341464493,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 3463
    },
    {
      "epoch": 0.03464,
      "grad_norm": 0.8057895102942476,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 3464
    },
    {
      "epoch": 0.03465,
      "grad_norm": 0.754598199709423,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 3465
    },
    {
      "epoch": 0.03466,
      "grad_norm": 0.6771478949940347,
      "learning_rate": 0.003,
      "loss": 4.1778,
      "step": 3466
    },
    {
      "epoch": 0.03467,
      "grad_norm": 0.6186375156162432,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 3467
    },
    {
      "epoch": 0.03468,
      "grad_norm": 0.6769160235929496,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 3468
    },
    {
      "epoch": 0.03469,
      "grad_norm": 0.6809204256661774,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 3469
    },
    {
      "epoch": 0.0347,
      "grad_norm": 0.7756647140071914,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 3470
    },
    {
      "epoch": 0.03471,
      "grad_norm": 0.8803633516066148,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 3471
    },
    {
      "epoch": 0.03472,
      "grad_norm": 0.8377841449731822,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3472
    },
    {
      "epoch": 0.03473,
      "grad_norm": 0.8583404258765982,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 3473
    },
    {
      "epoch": 0.03474,
      "grad_norm": 0.821362643848504,
      "learning_rate": 0.003,
      "loss": 4.2036,
      "step": 3474
    },
    {
      "epoch": 0.03475,
      "grad_norm": 0.7366543444137226,
      "learning_rate": 0.003,
      "loss": 4.171,
      "step": 3475
    },
    {
      "epoch": 0.03476,
      "grad_norm": 0.8990414035833436,
      "learning_rate": 0.003,
      "loss": 4.1854,
      "step": 3476
    },
    {
      "epoch": 0.03477,
      "grad_norm": 0.9251586453057641,
      "learning_rate": 0.003,
      "loss": 4.1836,
      "step": 3477
    },
    {
      "epoch": 0.03478,
      "grad_norm": 0.9081762714287535,
      "learning_rate": 0.003,
      "loss": 4.1933,
      "step": 3478
    },
    {
      "epoch": 0.03479,
      "grad_norm": 0.95014653125921,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 3479
    },
    {
      "epoch": 0.0348,
      "grad_norm": 0.8211598773714466,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 3480
    },
    {
      "epoch": 0.03481,
      "grad_norm": 0.6714433574061417,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3481
    },
    {
      "epoch": 0.03482,
      "grad_norm": 0.6624685031516976,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 3482
    },
    {
      "epoch": 0.03483,
      "grad_norm": 0.6364069563539299,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 3483
    },
    {
      "epoch": 0.03484,
      "grad_norm": 0.7252638284821149,
      "learning_rate": 0.003,
      "loss": 4.2109,
      "step": 3484
    },
    {
      "epoch": 0.03485,
      "grad_norm": 0.6599203755955035,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 3485
    },
    {
      "epoch": 0.03486,
      "grad_norm": 0.6351455036938864,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 3486
    },
    {
      "epoch": 0.03487,
      "grad_norm": 0.7047689274973644,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 3487
    },
    {
      "epoch": 0.03488,
      "grad_norm": 0.5998971286625794,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 3488
    },
    {
      "epoch": 0.03489,
      "grad_norm": 0.5928637335909946,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 3489
    },
    {
      "epoch": 0.0349,
      "grad_norm": 0.7224065582654237,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 3490
    },
    {
      "epoch": 0.03491,
      "grad_norm": 0.8948866147367468,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 3491
    },
    {
      "epoch": 0.03492,
      "grad_norm": 1.0698897809029222,
      "learning_rate": 0.003,
      "loss": 4.1887,
      "step": 3492
    },
    {
      "epoch": 0.03493,
      "grad_norm": 0.879044396872005,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 3493
    },
    {
      "epoch": 0.03494,
      "grad_norm": 0.7308396541826835,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 3494
    },
    {
      "epoch": 0.03495,
      "grad_norm": 0.7338448975678681,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 3495
    },
    {
      "epoch": 0.03496,
      "grad_norm": 0.8206218724499419,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 3496
    },
    {
      "epoch": 0.03497,
      "grad_norm": 0.8420218799600944,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 3497
    },
    {
      "epoch": 0.03498,
      "grad_norm": 0.7276352170340858,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 3498
    },
    {
      "epoch": 0.03499,
      "grad_norm": 0.6164613363129237,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 3499
    },
    {
      "epoch": 0.035,
      "grad_norm": 0.5533625720370539,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 3500
    },
    {
      "epoch": 0.03501,
      "grad_norm": 0.5440318172984431,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 3501
    },
    {
      "epoch": 0.03502,
      "grad_norm": 0.570449376318866,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 3502
    },
    {
      "epoch": 0.03503,
      "grad_norm": 0.5938098529331512,
      "learning_rate": 0.003,
      "loss": 4.1865,
      "step": 3503
    },
    {
      "epoch": 0.03504,
      "grad_norm": 0.5807097638534694,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 3504
    },
    {
      "epoch": 0.03505,
      "grad_norm": 0.5346263597287685,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 3505
    },
    {
      "epoch": 0.03506,
      "grad_norm": 0.5906950615820623,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3506
    },
    {
      "epoch": 0.03507,
      "grad_norm": 0.5854291049171028,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 3507
    },
    {
      "epoch": 0.03508,
      "grad_norm": 0.5919853377246803,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 3508
    },
    {
      "epoch": 0.03509,
      "grad_norm": 0.6611088078675262,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 3509
    },
    {
      "epoch": 0.0351,
      "grad_norm": 0.694428361139572,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 3510
    },
    {
      "epoch": 0.03511,
      "grad_norm": 0.5309620036695369,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 3511
    },
    {
      "epoch": 0.03512,
      "grad_norm": 0.4298187490647193,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 3512
    },
    {
      "epoch": 0.03513,
      "grad_norm": 0.4111172909761725,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 3513
    },
    {
      "epoch": 0.03514,
      "grad_norm": 0.45911964934305066,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 3514
    },
    {
      "epoch": 0.03515,
      "grad_norm": 0.5301625577665102,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 3515
    },
    {
      "epoch": 0.03516,
      "grad_norm": 0.6537735331219108,
      "learning_rate": 0.003,
      "loss": 4.1947,
      "step": 3516
    },
    {
      "epoch": 0.03517,
      "grad_norm": 0.8603631139968849,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 3517
    },
    {
      "epoch": 0.03518,
      "grad_norm": 0.9381334187285012,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 3518
    },
    {
      "epoch": 0.03519,
      "grad_norm": 0.7499517998636877,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 3519
    },
    {
      "epoch": 0.0352,
      "grad_norm": 0.6063052992661483,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 3520
    },
    {
      "epoch": 0.03521,
      "grad_norm": 0.6898167453059805,
      "learning_rate": 0.003,
      "loss": 4.1613,
      "step": 3521
    },
    {
      "epoch": 0.03522,
      "grad_norm": 0.7285294145828302,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 3522
    },
    {
      "epoch": 0.03523,
      "grad_norm": 0.786382262009744,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 3523
    },
    {
      "epoch": 0.03524,
      "grad_norm": 0.8485073208145252,
      "learning_rate": 0.003,
      "loss": 4.1811,
      "step": 3524
    },
    {
      "epoch": 0.03525,
      "grad_norm": 0.7884181271432541,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 3525
    },
    {
      "epoch": 0.03526,
      "grad_norm": 0.6895666981950603,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 3526
    },
    {
      "epoch": 0.03527,
      "grad_norm": 0.6609448619741584,
      "learning_rate": 0.003,
      "loss": 4.1841,
      "step": 3527
    },
    {
      "epoch": 0.03528,
      "grad_norm": 0.6552099898208482,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 3528
    },
    {
      "epoch": 0.03529,
      "grad_norm": 0.5657486402844111,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 3529
    },
    {
      "epoch": 0.0353,
      "grad_norm": 0.6271377691303136,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 3530
    },
    {
      "epoch": 0.03531,
      "grad_norm": 0.6215946372415877,
      "learning_rate": 0.003,
      "loss": 4.1691,
      "step": 3531
    },
    {
      "epoch": 0.03532,
      "grad_norm": 0.6387289961881946,
      "learning_rate": 0.003,
      "loss": 4.1723,
      "step": 3532
    },
    {
      "epoch": 0.03533,
      "grad_norm": 0.6872453476711173,
      "learning_rate": 0.003,
      "loss": 4.1733,
      "step": 3533
    },
    {
      "epoch": 0.03534,
      "grad_norm": 0.7783886281484835,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 3534
    },
    {
      "epoch": 0.03535,
      "grad_norm": 0.8725146995031193,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 3535
    },
    {
      "epoch": 0.03536,
      "grad_norm": 0.9246122970504058,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 3536
    },
    {
      "epoch": 0.03537,
      "grad_norm": 1.065477970890386,
      "learning_rate": 0.003,
      "loss": 4.1712,
      "step": 3537
    },
    {
      "epoch": 0.03538,
      "grad_norm": 0.9594169294264835,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3538
    },
    {
      "epoch": 0.03539,
      "grad_norm": 0.8330662665517189,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 3539
    },
    {
      "epoch": 0.0354,
      "grad_norm": 0.8526080452896662,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 3540
    },
    {
      "epoch": 0.03541,
      "grad_norm": 0.9085329983346553,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 3541
    },
    {
      "epoch": 0.03542,
      "grad_norm": 0.833985994066796,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3542
    },
    {
      "epoch": 0.03543,
      "grad_norm": 0.7626821688858162,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 3543
    },
    {
      "epoch": 0.03544,
      "grad_norm": 0.806738250706975,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3544
    },
    {
      "epoch": 0.03545,
      "grad_norm": 0.9242341338381844,
      "learning_rate": 0.003,
      "loss": 4.1613,
      "step": 3545
    },
    {
      "epoch": 0.03546,
      "grad_norm": 0.8293267592237652,
      "learning_rate": 0.003,
      "loss": 4.1759,
      "step": 3546
    },
    {
      "epoch": 0.03547,
      "grad_norm": 0.7605341067555178,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 3547
    },
    {
      "epoch": 0.03548,
      "grad_norm": 0.697933140723856,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 3548
    },
    {
      "epoch": 0.03549,
      "grad_norm": 0.5804050326776297,
      "learning_rate": 0.003,
      "loss": 4.1739,
      "step": 3549
    },
    {
      "epoch": 0.0355,
      "grad_norm": 0.603076289815789,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 3550
    },
    {
      "epoch": 0.03551,
      "grad_norm": 0.5646969274817457,
      "learning_rate": 0.003,
      "loss": 4.1847,
      "step": 3551
    },
    {
      "epoch": 0.03552,
      "grad_norm": 0.5353048682624879,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 3552
    },
    {
      "epoch": 0.03553,
      "grad_norm": 0.43948675430004575,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 3553
    },
    {
      "epoch": 0.03554,
      "grad_norm": 0.4404992722313449,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 3554
    },
    {
      "epoch": 0.03555,
      "grad_norm": 0.4408782434286909,
      "learning_rate": 0.003,
      "loss": 4.1708,
      "step": 3555
    },
    {
      "epoch": 0.03556,
      "grad_norm": 0.4089405885332322,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 3556
    },
    {
      "epoch": 0.03557,
      "grad_norm": 0.4886396097558162,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 3557
    },
    {
      "epoch": 0.03558,
      "grad_norm": 0.5683554684886454,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 3558
    },
    {
      "epoch": 0.03559,
      "grad_norm": 0.7367360745457323,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3559
    },
    {
      "epoch": 0.0356,
      "grad_norm": 0.8520355513833522,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3560
    },
    {
      "epoch": 0.03561,
      "grad_norm": 0.8538012995310599,
      "learning_rate": 0.003,
      "loss": 4.1983,
      "step": 3561
    },
    {
      "epoch": 0.03562,
      "grad_norm": 0.6793279389206137,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 3562
    },
    {
      "epoch": 0.03563,
      "grad_norm": 0.5212190292484681,
      "learning_rate": 0.003,
      "loss": 4.1876,
      "step": 3563
    },
    {
      "epoch": 0.03564,
      "grad_norm": 0.6272603340476319,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 3564
    },
    {
      "epoch": 0.03565,
      "grad_norm": 0.7385189206717337,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 3565
    },
    {
      "epoch": 0.03566,
      "grad_norm": 0.6717765632340529,
      "learning_rate": 0.003,
      "loss": 4.1622,
      "step": 3566
    },
    {
      "epoch": 0.03567,
      "grad_norm": 0.6137260235822769,
      "learning_rate": 0.003,
      "loss": 4.154,
      "step": 3567
    },
    {
      "epoch": 0.03568,
      "grad_norm": 0.5473859937693163,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 3568
    },
    {
      "epoch": 0.03569,
      "grad_norm": 0.5260257267109332,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 3569
    },
    {
      "epoch": 0.0357,
      "grad_norm": 0.5464547294802432,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 3570
    },
    {
      "epoch": 0.03571,
      "grad_norm": 0.5170616946027875,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 3571
    },
    {
      "epoch": 0.03572,
      "grad_norm": 0.5245302678289541,
      "learning_rate": 0.003,
      "loss": 4.1721,
      "step": 3572
    },
    {
      "epoch": 0.03573,
      "grad_norm": 0.5539674018003774,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 3573
    },
    {
      "epoch": 0.03574,
      "grad_norm": 0.5717457987893073,
      "learning_rate": 0.003,
      "loss": 4.1694,
      "step": 3574
    },
    {
      "epoch": 0.03575,
      "grad_norm": 0.6597241159959504,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 3575
    },
    {
      "epoch": 0.03576,
      "grad_norm": 0.7987769449923495,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 3576
    },
    {
      "epoch": 0.03577,
      "grad_norm": 0.9166920267526417,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 3577
    },
    {
      "epoch": 0.03578,
      "grad_norm": 0.8840919078143189,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 3578
    },
    {
      "epoch": 0.03579,
      "grad_norm": 0.8228224911003803,
      "learning_rate": 0.003,
      "loss": 4.1888,
      "step": 3579
    },
    {
      "epoch": 0.0358,
      "grad_norm": 0.9419353853796917,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 3580
    },
    {
      "epoch": 0.03581,
      "grad_norm": 0.8980601226690328,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 3581
    },
    {
      "epoch": 0.03582,
      "grad_norm": 0.8072733688617605,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 3582
    },
    {
      "epoch": 0.03583,
      "grad_norm": 0.8230842831411875,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 3583
    },
    {
      "epoch": 0.03584,
      "grad_norm": 0.6285337030399905,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 3584
    },
    {
      "epoch": 0.03585,
      "grad_norm": 0.6220326404370078,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 3585
    },
    {
      "epoch": 0.03586,
      "grad_norm": 0.6861320623389772,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3586
    },
    {
      "epoch": 0.03587,
      "grad_norm": 0.7015780277654053,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 3587
    },
    {
      "epoch": 0.03588,
      "grad_norm": 0.6838138540479132,
      "learning_rate": 0.003,
      "loss": 4.1653,
      "step": 3588
    },
    {
      "epoch": 0.03589,
      "grad_norm": 0.6735354715174068,
      "learning_rate": 0.003,
      "loss": 4.1806,
      "step": 3589
    },
    {
      "epoch": 0.0359,
      "grad_norm": 0.6330341634771458,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 3590
    },
    {
      "epoch": 0.03591,
      "grad_norm": 0.5910317698548738,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 3591
    },
    {
      "epoch": 0.03592,
      "grad_norm": 0.5618523931442762,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 3592
    },
    {
      "epoch": 0.03593,
      "grad_norm": 0.8151654550358379,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 3593
    },
    {
      "epoch": 0.03594,
      "grad_norm": 1.0948812647746626,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 3594
    },
    {
      "epoch": 0.03595,
      "grad_norm": 0.8469599816388995,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 3595
    },
    {
      "epoch": 0.03596,
      "grad_norm": 0.5860459737545368,
      "learning_rate": 0.003,
      "loss": 4.1904,
      "step": 3596
    },
    {
      "epoch": 0.03597,
      "grad_norm": 0.6216520906256218,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 3597
    },
    {
      "epoch": 0.03598,
      "grad_norm": 0.716220967428177,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3598
    },
    {
      "epoch": 0.03599,
      "grad_norm": 0.7599051671476444,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 3599
    },
    {
      "epoch": 0.036,
      "grad_norm": 0.6520846019848653,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 3600
    },
    {
      "epoch": 0.03601,
      "grad_norm": 0.5364426998203575,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 3601
    },
    {
      "epoch": 0.03602,
      "grad_norm": 0.5192638145820389,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 3602
    },
    {
      "epoch": 0.03603,
      "grad_norm": 0.48444783985893647,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 3603
    },
    {
      "epoch": 0.03604,
      "grad_norm": 0.44019612697059707,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 3604
    },
    {
      "epoch": 0.03605,
      "grad_norm": 0.46867930201266456,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3605
    },
    {
      "epoch": 0.03606,
      "grad_norm": 0.4990137003463985,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 3606
    },
    {
      "epoch": 0.03607,
      "grad_norm": 0.6032395506240065,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 3607
    },
    {
      "epoch": 0.03608,
      "grad_norm": 0.6248433459385002,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 3608
    },
    {
      "epoch": 0.03609,
      "grad_norm": 0.6134188004656983,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 3609
    },
    {
      "epoch": 0.0361,
      "grad_norm": 0.6290690472912259,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3610
    },
    {
      "epoch": 0.03611,
      "grad_norm": 0.72557440929179,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 3611
    },
    {
      "epoch": 0.03612,
      "grad_norm": 0.8893798610273862,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 3612
    },
    {
      "epoch": 0.03613,
      "grad_norm": 0.9668223551113271,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 3613
    },
    {
      "epoch": 0.03614,
      "grad_norm": 0.734407911853949,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 3614
    },
    {
      "epoch": 0.03615,
      "grad_norm": 0.6729100967664783,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 3615
    },
    {
      "epoch": 0.03616,
      "grad_norm": 0.8657549358453647,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3616
    },
    {
      "epoch": 0.03617,
      "grad_norm": 0.9260684291150802,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 3617
    },
    {
      "epoch": 0.03618,
      "grad_norm": 0.8570047461014029,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3618
    },
    {
      "epoch": 0.03619,
      "grad_norm": 0.866479505486592,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3619
    },
    {
      "epoch": 0.0362,
      "grad_norm": 0.8958378530573641,
      "learning_rate": 0.003,
      "loss": 4.1828,
      "step": 3620
    },
    {
      "epoch": 0.03621,
      "grad_norm": 0.8754496119000232,
      "learning_rate": 0.003,
      "loss": 4.1836,
      "step": 3621
    },
    {
      "epoch": 0.03622,
      "grad_norm": 0.7829921953128642,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 3622
    },
    {
      "epoch": 0.03623,
      "grad_norm": 0.7264263724470046,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 3623
    },
    {
      "epoch": 0.03624,
      "grad_norm": 0.7103073858138196,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 3624
    },
    {
      "epoch": 0.03625,
      "grad_norm": 0.7362483226450557,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3625
    },
    {
      "epoch": 0.03626,
      "grad_norm": 0.6700204258570956,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 3626
    },
    {
      "epoch": 0.03627,
      "grad_norm": 0.8160965768748923,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 3627
    },
    {
      "epoch": 0.03628,
      "grad_norm": 0.7835131802445204,
      "learning_rate": 0.003,
      "loss": 4.1739,
      "step": 3628
    },
    {
      "epoch": 0.03629,
      "grad_norm": 0.6985149126811981,
      "learning_rate": 0.003,
      "loss": 4.1895,
      "step": 3629
    },
    {
      "epoch": 0.0363,
      "grad_norm": 0.6039041743694569,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 3630
    },
    {
      "epoch": 0.03631,
      "grad_norm": 0.5592412769390824,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 3631
    },
    {
      "epoch": 0.03632,
      "grad_norm": 0.49011079854965545,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 3632
    },
    {
      "epoch": 0.03633,
      "grad_norm": 0.5382307200734877,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 3633
    },
    {
      "epoch": 0.03634,
      "grad_norm": 0.5367696632611142,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 3634
    },
    {
      "epoch": 0.03635,
      "grad_norm": 0.5142486579138349,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 3635
    },
    {
      "epoch": 0.03636,
      "grad_norm": 0.5652066346276485,
      "learning_rate": 0.003,
      "loss": 4.1613,
      "step": 3636
    },
    {
      "epoch": 0.03637,
      "grad_norm": 0.59668194695564,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 3637
    },
    {
      "epoch": 0.03638,
      "grad_norm": 0.6925042816123054,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 3638
    },
    {
      "epoch": 0.03639,
      "grad_norm": 0.8037659165365143,
      "learning_rate": 0.003,
      "loss": 4.1869,
      "step": 3639
    },
    {
      "epoch": 0.0364,
      "grad_norm": 0.8148685837655183,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 3640
    },
    {
      "epoch": 0.03641,
      "grad_norm": 0.7919945416520671,
      "learning_rate": 0.003,
      "loss": 4.1676,
      "step": 3641
    },
    {
      "epoch": 0.03642,
      "grad_norm": 0.8387878217923936,
      "learning_rate": 0.003,
      "loss": 4.1883,
      "step": 3642
    },
    {
      "epoch": 0.03643,
      "grad_norm": 0.7524748876313961,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 3643
    },
    {
      "epoch": 0.03644,
      "grad_norm": 0.7273454421844323,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 3644
    },
    {
      "epoch": 0.03645,
      "grad_norm": 0.7955485790365344,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 3645
    },
    {
      "epoch": 0.03646,
      "grad_norm": 0.7720894887238786,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 3646
    },
    {
      "epoch": 0.03647,
      "grad_norm": 0.7086978218067327,
      "learning_rate": 0.003,
      "loss": 4.1616,
      "step": 3647
    },
    {
      "epoch": 0.03648,
      "grad_norm": 0.6770260559623669,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 3648
    },
    {
      "epoch": 0.03649,
      "grad_norm": 0.5881027107624417,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 3649
    },
    {
      "epoch": 0.0365,
      "grad_norm": 0.6131807187241587,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 3650
    },
    {
      "epoch": 0.03651,
      "grad_norm": 0.594204825708316,
      "learning_rate": 0.003,
      "loss": 4.1729,
      "step": 3651
    },
    {
      "epoch": 0.03652,
      "grad_norm": 0.6399255321695406,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 3652
    },
    {
      "epoch": 0.03653,
      "grad_norm": 0.6119522794582246,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 3653
    },
    {
      "epoch": 0.03654,
      "grad_norm": 0.5566016631522781,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 3654
    },
    {
      "epoch": 0.03655,
      "grad_norm": 0.5193773149688491,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3655
    },
    {
      "epoch": 0.03656,
      "grad_norm": 0.6264013089429903,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 3656
    },
    {
      "epoch": 0.03657,
      "grad_norm": 0.6979744938653328,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 3657
    },
    {
      "epoch": 0.03658,
      "grad_norm": 0.8558728547971467,
      "learning_rate": 0.003,
      "loss": 4.1964,
      "step": 3658
    },
    {
      "epoch": 0.03659,
      "grad_norm": 0.8839768938776124,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 3659
    },
    {
      "epoch": 0.0366,
      "grad_norm": 0.738686352865498,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3660
    },
    {
      "epoch": 0.03661,
      "grad_norm": 0.7093891716169941,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 3661
    },
    {
      "epoch": 0.03662,
      "grad_norm": 0.7587081351482545,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 3662
    },
    {
      "epoch": 0.03663,
      "grad_norm": 0.7076929569924106,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 3663
    },
    {
      "epoch": 0.03664,
      "grad_norm": 0.7127846105810646,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 3664
    },
    {
      "epoch": 0.03665,
      "grad_norm": 0.7078459190405836,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3665
    },
    {
      "epoch": 0.03666,
      "grad_norm": 0.6805684369907511,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 3666
    },
    {
      "epoch": 0.03667,
      "grad_norm": 0.7150672278612442,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 3667
    },
    {
      "epoch": 0.03668,
      "grad_norm": 0.7699785342164407,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3668
    },
    {
      "epoch": 0.03669,
      "grad_norm": 0.7952877417083865,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 3669
    },
    {
      "epoch": 0.0367,
      "grad_norm": 0.7569891220037553,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 3670
    },
    {
      "epoch": 0.03671,
      "grad_norm": 0.7098731435624128,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 3671
    },
    {
      "epoch": 0.03672,
      "grad_norm": 0.6352485396363875,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 3672
    },
    {
      "epoch": 0.03673,
      "grad_norm": 0.6885305441098153,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3673
    },
    {
      "epoch": 0.03674,
      "grad_norm": 0.7059131456483599,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 3674
    },
    {
      "epoch": 0.03675,
      "grad_norm": 0.6038823828670427,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3675
    },
    {
      "epoch": 0.03676,
      "grad_norm": 0.57222994957781,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 3676
    },
    {
      "epoch": 0.03677,
      "grad_norm": 0.6802347986469087,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 3677
    },
    {
      "epoch": 0.03678,
      "grad_norm": 0.7740620961238671,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 3678
    },
    {
      "epoch": 0.03679,
      "grad_norm": 0.8635840591832115,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 3679
    },
    {
      "epoch": 0.0368,
      "grad_norm": 0.9019802875931285,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3680
    },
    {
      "epoch": 0.03681,
      "grad_norm": 0.9372442716730703,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3681
    },
    {
      "epoch": 0.03682,
      "grad_norm": 0.9892266156518691,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 3682
    },
    {
      "epoch": 0.03683,
      "grad_norm": 1.1018562984300873,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 3683
    },
    {
      "epoch": 0.03684,
      "grad_norm": 0.9775119392439015,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 3684
    },
    {
      "epoch": 0.03685,
      "grad_norm": 0.9024302815079079,
      "learning_rate": 0.003,
      "loss": 4.1723,
      "step": 3685
    },
    {
      "epoch": 0.03686,
      "grad_norm": 0.9419765899805256,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 3686
    },
    {
      "epoch": 0.03687,
      "grad_norm": 1.1488204202771457,
      "learning_rate": 0.003,
      "loss": 4.2226,
      "step": 3687
    },
    {
      "epoch": 0.03688,
      "grad_norm": 0.9052265907645448,
      "learning_rate": 0.003,
      "loss": 4.2006,
      "step": 3688
    },
    {
      "epoch": 0.03689,
      "grad_norm": 0.8161052846618367,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 3689
    },
    {
      "epoch": 0.0369,
      "grad_norm": 0.8630363350313656,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 3690
    },
    {
      "epoch": 0.03691,
      "grad_norm": 1.000347066652698,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 3691
    },
    {
      "epoch": 0.03692,
      "grad_norm": 1.0125314193391763,
      "learning_rate": 0.003,
      "loss": 4.2237,
      "step": 3692
    },
    {
      "epoch": 0.03693,
      "grad_norm": 0.7792710807207066,
      "learning_rate": 0.003,
      "loss": 4.187,
      "step": 3693
    },
    {
      "epoch": 0.03694,
      "grad_norm": 0.6567620211689106,
      "learning_rate": 0.003,
      "loss": 4.1772,
      "step": 3694
    },
    {
      "epoch": 0.03695,
      "grad_norm": 0.6069200650214434,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3695
    },
    {
      "epoch": 0.03696,
      "grad_norm": 0.6245228527312463,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 3696
    },
    {
      "epoch": 0.03697,
      "grad_norm": 0.5956476153449444,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 3697
    },
    {
      "epoch": 0.03698,
      "grad_norm": 0.6172146239896532,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 3698
    },
    {
      "epoch": 0.03699,
      "grad_norm": 0.6369440678819488,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 3699
    },
    {
      "epoch": 0.037,
      "grad_norm": 0.7397659340274994,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 3700
    },
    {
      "epoch": 0.03701,
      "grad_norm": 0.7400755144798878,
      "learning_rate": 0.003,
      "loss": 4.1708,
      "step": 3701
    },
    {
      "epoch": 0.03702,
      "grad_norm": 0.6011624195612609,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 3702
    },
    {
      "epoch": 0.03703,
      "grad_norm": 0.45688971739502526,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 3703
    },
    {
      "epoch": 0.03704,
      "grad_norm": 0.4625467744385196,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 3704
    },
    {
      "epoch": 0.03705,
      "grad_norm": 0.4601609886250813,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 3705
    },
    {
      "epoch": 0.03706,
      "grad_norm": 0.4966406347936221,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 3706
    },
    {
      "epoch": 0.03707,
      "grad_norm": 0.42477832568061835,
      "learning_rate": 0.003,
      "loss": 4.142,
      "step": 3707
    },
    {
      "epoch": 0.03708,
      "grad_norm": 0.40682526312634587,
      "learning_rate": 0.003,
      "loss": 4.1616,
      "step": 3708
    },
    {
      "epoch": 0.03709,
      "grad_norm": 0.4133271539515952,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 3709
    },
    {
      "epoch": 0.0371,
      "grad_norm": 0.43081117839287136,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 3710
    },
    {
      "epoch": 0.03711,
      "grad_norm": 0.46229609767440477,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 3711
    },
    {
      "epoch": 0.03712,
      "grad_norm": 0.4820292193047058,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 3712
    },
    {
      "epoch": 0.03713,
      "grad_norm": 0.5404513625982291,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 3713
    },
    {
      "epoch": 0.03714,
      "grad_norm": 0.6436389631284867,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 3714
    },
    {
      "epoch": 0.03715,
      "grad_norm": 0.8290141406987748,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 3715
    },
    {
      "epoch": 0.03716,
      "grad_norm": 0.9048610214356393,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 3716
    },
    {
      "epoch": 0.03717,
      "grad_norm": 0.7058124491704675,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3717
    },
    {
      "epoch": 0.03718,
      "grad_norm": 0.4918162555153287,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 3718
    },
    {
      "epoch": 0.03719,
      "grad_norm": 0.6622934600969125,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3719
    },
    {
      "epoch": 0.0372,
      "grad_norm": 0.7298879766213411,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3720
    },
    {
      "epoch": 0.03721,
      "grad_norm": 0.7368836467114779,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 3721
    },
    {
      "epoch": 0.03722,
      "grad_norm": 0.6530001576353514,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 3722
    },
    {
      "epoch": 0.03723,
      "grad_norm": 0.6828551463380831,
      "learning_rate": 0.003,
      "loss": 4.1778,
      "step": 3723
    },
    {
      "epoch": 0.03724,
      "grad_norm": 0.8098638547098356,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3724
    },
    {
      "epoch": 0.03725,
      "grad_norm": 0.787968577179379,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 3725
    },
    {
      "epoch": 0.03726,
      "grad_norm": 0.790721512863178,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 3726
    },
    {
      "epoch": 0.03727,
      "grad_norm": 0.7875255929753493,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 3727
    },
    {
      "epoch": 0.03728,
      "grad_norm": 0.7766724710074185,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 3728
    },
    {
      "epoch": 0.03729,
      "grad_norm": 0.7446811620432703,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 3729
    },
    {
      "epoch": 0.0373,
      "grad_norm": 0.7631969023181137,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 3730
    },
    {
      "epoch": 0.03731,
      "grad_norm": 0.8301810283690936,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 3731
    },
    {
      "epoch": 0.03732,
      "grad_norm": 0.795728492246486,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 3732
    },
    {
      "epoch": 0.03733,
      "grad_norm": 0.6502399531894205,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 3733
    },
    {
      "epoch": 0.03734,
      "grad_norm": 0.6112985055672381,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 3734
    },
    {
      "epoch": 0.03735,
      "grad_norm": 0.7852832594186502,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 3735
    },
    {
      "epoch": 0.03736,
      "grad_norm": 0.9467543719299969,
      "learning_rate": 0.003,
      "loss": 4.1902,
      "step": 3736
    },
    {
      "epoch": 0.03737,
      "grad_norm": 0.9659184343775771,
      "learning_rate": 0.003,
      "loss": 4.1847,
      "step": 3737
    },
    {
      "epoch": 0.03738,
      "grad_norm": 0.7573591347533927,
      "learning_rate": 0.003,
      "loss": 4.1903,
      "step": 3738
    },
    {
      "epoch": 0.03739,
      "grad_norm": 0.7992106244448375,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 3739
    },
    {
      "epoch": 0.0374,
      "grad_norm": 0.7486649819121083,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 3740
    },
    {
      "epoch": 0.03741,
      "grad_norm": 0.5976608138563914,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 3741
    },
    {
      "epoch": 0.03742,
      "grad_norm": 0.5995036064200935,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 3742
    },
    {
      "epoch": 0.03743,
      "grad_norm": 0.6901682901857126,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 3743
    },
    {
      "epoch": 0.03744,
      "grad_norm": 0.6080608979861319,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 3744
    },
    {
      "epoch": 0.03745,
      "grad_norm": 0.6576923197541351,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3745
    },
    {
      "epoch": 0.03746,
      "grad_norm": 0.6727287733203032,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3746
    },
    {
      "epoch": 0.03747,
      "grad_norm": 0.7572729409130596,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3747
    },
    {
      "epoch": 0.03748,
      "grad_norm": 0.7027720695663551,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 3748
    },
    {
      "epoch": 0.03749,
      "grad_norm": 0.7669711680997199,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 3749
    },
    {
      "epoch": 0.0375,
      "grad_norm": 0.7408458574924264,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 3750
    },
    {
      "epoch": 0.03751,
      "grad_norm": 0.6282480274827662,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 3751
    },
    {
      "epoch": 0.03752,
      "grad_norm": 0.6086039975418626,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 3752
    },
    {
      "epoch": 0.03753,
      "grad_norm": 0.6045335662772724,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 3753
    },
    {
      "epoch": 0.03754,
      "grad_norm": 0.6756444191450836,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3754
    },
    {
      "epoch": 0.03755,
      "grad_norm": 0.6814576187053439,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3755
    },
    {
      "epoch": 0.03756,
      "grad_norm": 0.6979671812601237,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 3756
    },
    {
      "epoch": 0.03757,
      "grad_norm": 0.7762754751971193,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 3757
    },
    {
      "epoch": 0.03758,
      "grad_norm": 0.9320275397343457,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 3758
    },
    {
      "epoch": 0.03759,
      "grad_norm": 1.029430729208351,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 3759
    },
    {
      "epoch": 0.0376,
      "grad_norm": 0.6884103372684509,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 3760
    },
    {
      "epoch": 0.03761,
      "grad_norm": 0.5726368511742514,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 3761
    },
    {
      "epoch": 0.03762,
      "grad_norm": 0.7431155489308362,
      "learning_rate": 0.003,
      "loss": 4.1857,
      "step": 3762
    },
    {
      "epoch": 0.03763,
      "grad_norm": 0.7334829914103421,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3763
    },
    {
      "epoch": 0.03764,
      "grad_norm": 0.7755402436276377,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3764
    },
    {
      "epoch": 0.03765,
      "grad_norm": 0.8071077893148694,
      "learning_rate": 0.003,
      "loss": 4.2095,
      "step": 3765
    },
    {
      "epoch": 0.03766,
      "grad_norm": 0.8062094613718822,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 3766
    },
    {
      "epoch": 0.03767,
      "grad_norm": 0.7450385148464626,
      "learning_rate": 0.003,
      "loss": 4.1846,
      "step": 3767
    },
    {
      "epoch": 0.03768,
      "grad_norm": 0.8501525775153304,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 3768
    },
    {
      "epoch": 0.03769,
      "grad_norm": 0.8490270851264778,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3769
    },
    {
      "epoch": 0.0377,
      "grad_norm": 0.6528754246262384,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 3770
    },
    {
      "epoch": 0.03771,
      "grad_norm": 0.514456285674655,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 3771
    },
    {
      "epoch": 0.03772,
      "grad_norm": 0.5778069788565984,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 3772
    },
    {
      "epoch": 0.03773,
      "grad_norm": 0.590491520248524,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3773
    },
    {
      "epoch": 0.03774,
      "grad_norm": 0.6156498757993644,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 3774
    },
    {
      "epoch": 0.03775,
      "grad_norm": 0.5373760491110208,
      "learning_rate": 0.003,
      "loss": 4.1714,
      "step": 3775
    },
    {
      "epoch": 0.03776,
      "grad_norm": 0.58872450634308,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 3776
    },
    {
      "epoch": 0.03777,
      "grad_norm": 0.57682380030575,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 3777
    },
    {
      "epoch": 0.03778,
      "grad_norm": 0.5253188081583163,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 3778
    },
    {
      "epoch": 0.03779,
      "grad_norm": 0.47209668216344913,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 3779
    },
    {
      "epoch": 0.0378,
      "grad_norm": 0.5332773200600602,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 3780
    },
    {
      "epoch": 0.03781,
      "grad_norm": 0.6825162371448047,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 3781
    },
    {
      "epoch": 0.03782,
      "grad_norm": 0.8418563836091014,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 3782
    },
    {
      "epoch": 0.03783,
      "grad_norm": 0.8512086413504435,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 3783
    },
    {
      "epoch": 0.03784,
      "grad_norm": 0.6671577794115535,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 3784
    },
    {
      "epoch": 0.03785,
      "grad_norm": 0.5730225846531266,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 3785
    },
    {
      "epoch": 0.03786,
      "grad_norm": 0.6939236048367627,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 3786
    },
    {
      "epoch": 0.03787,
      "grad_norm": 0.7368644792806699,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3787
    },
    {
      "epoch": 0.03788,
      "grad_norm": 0.6874505246320723,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 3788
    },
    {
      "epoch": 0.03789,
      "grad_norm": 0.6323049623986694,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 3789
    },
    {
      "epoch": 0.0379,
      "grad_norm": 0.6673306922274641,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 3790
    },
    {
      "epoch": 0.03791,
      "grad_norm": 0.6943632593992325,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 3791
    },
    {
      "epoch": 0.03792,
      "grad_norm": 0.6791072764782452,
      "learning_rate": 0.003,
      "loss": 4.1837,
      "step": 3792
    },
    {
      "epoch": 0.03793,
      "grad_norm": 0.6789029894480806,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 3793
    },
    {
      "epoch": 0.03794,
      "grad_norm": 0.7369905694175998,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 3794
    },
    {
      "epoch": 0.03795,
      "grad_norm": 0.7808489227373923,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 3795
    },
    {
      "epoch": 0.03796,
      "grad_norm": 0.8037760915310143,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 3796
    },
    {
      "epoch": 0.03797,
      "grad_norm": 0.7083927011043929,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 3797
    },
    {
      "epoch": 0.03798,
      "grad_norm": 0.706481627859496,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3798
    },
    {
      "epoch": 0.03799,
      "grad_norm": 0.6536139692203335,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 3799
    },
    {
      "epoch": 0.038,
      "grad_norm": 0.667914966075993,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 3800
    },
    {
      "epoch": 0.03801,
      "grad_norm": 0.6933040831212794,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 3801
    },
    {
      "epoch": 0.03802,
      "grad_norm": 0.6580821376084047,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3802
    },
    {
      "epoch": 0.03803,
      "grad_norm": 0.7779130392910976,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 3803
    },
    {
      "epoch": 0.03804,
      "grad_norm": 0.8777193399244893,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 3804
    },
    {
      "epoch": 0.03805,
      "grad_norm": 1.0020099764204873,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 3805
    },
    {
      "epoch": 0.03806,
      "grad_norm": 0.9634820397091458,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3806
    },
    {
      "epoch": 0.03807,
      "grad_norm": 0.7477715351876884,
      "learning_rate": 0.003,
      "loss": 4.161,
      "step": 3807
    },
    {
      "epoch": 0.03808,
      "grad_norm": 0.6392070927979916,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3808
    },
    {
      "epoch": 0.03809,
      "grad_norm": 0.6899923101840185,
      "learning_rate": 0.003,
      "loss": 4.1548,
      "step": 3809
    },
    {
      "epoch": 0.0381,
      "grad_norm": 0.6421320844146812,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 3810
    },
    {
      "epoch": 0.03811,
      "grad_norm": 0.5707513677972008,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 3811
    },
    {
      "epoch": 0.03812,
      "grad_norm": 0.6104281618032581,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3812
    },
    {
      "epoch": 0.03813,
      "grad_norm": 0.7409188794471754,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 3813
    },
    {
      "epoch": 0.03814,
      "grad_norm": 0.9399869172219258,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3814
    },
    {
      "epoch": 0.03815,
      "grad_norm": 1.0585281835423583,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 3815
    },
    {
      "epoch": 0.03816,
      "grad_norm": 0.9799276300500098,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 3816
    },
    {
      "epoch": 0.03817,
      "grad_norm": 0.9941910477954538,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 3817
    },
    {
      "epoch": 0.03818,
      "grad_norm": 0.9313720636167057,
      "learning_rate": 0.003,
      "loss": 4.2098,
      "step": 3818
    },
    {
      "epoch": 0.03819,
      "grad_norm": 0.8314238810075018,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 3819
    },
    {
      "epoch": 0.0382,
      "grad_norm": 0.7056825175881773,
      "learning_rate": 0.003,
      "loss": 4.2022,
      "step": 3820
    },
    {
      "epoch": 0.03821,
      "grad_norm": 0.6762157784042842,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 3821
    },
    {
      "epoch": 0.03822,
      "grad_norm": 0.6810353799131027,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 3822
    },
    {
      "epoch": 0.03823,
      "grad_norm": 0.693219699873982,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3823
    },
    {
      "epoch": 0.03824,
      "grad_norm": 0.7816987127966681,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 3824
    },
    {
      "epoch": 0.03825,
      "grad_norm": 0.8580364555019204,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 3825
    },
    {
      "epoch": 0.03826,
      "grad_norm": 0.737317401518091,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3826
    },
    {
      "epoch": 0.03827,
      "grad_norm": 0.6028910295160049,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3827
    },
    {
      "epoch": 0.03828,
      "grad_norm": 0.6588227313580575,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 3828
    },
    {
      "epoch": 0.03829,
      "grad_norm": 0.6720290422366224,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 3829
    },
    {
      "epoch": 0.0383,
      "grad_norm": 0.5935234244676734,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 3830
    },
    {
      "epoch": 0.03831,
      "grad_norm": 0.5027973321474926,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 3831
    },
    {
      "epoch": 0.03832,
      "grad_norm": 0.4793372599756564,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 3832
    },
    {
      "epoch": 0.03833,
      "grad_norm": 0.46250153234520147,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 3833
    },
    {
      "epoch": 0.03834,
      "grad_norm": 0.5005957325640198,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 3834
    },
    {
      "epoch": 0.03835,
      "grad_norm": 0.5638040713097137,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3835
    },
    {
      "epoch": 0.03836,
      "grad_norm": 0.7454000939697492,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 3836
    },
    {
      "epoch": 0.03837,
      "grad_norm": 0.84129158678745,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3837
    },
    {
      "epoch": 0.03838,
      "grad_norm": 0.8657748961672681,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3838
    },
    {
      "epoch": 0.03839,
      "grad_norm": 0.7593243937614058,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 3839
    },
    {
      "epoch": 0.0384,
      "grad_norm": 0.6950210484679211,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 3840
    },
    {
      "epoch": 0.03841,
      "grad_norm": 0.6626542758150601,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 3841
    },
    {
      "epoch": 0.03842,
      "grad_norm": 0.6047316460014022,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3842
    },
    {
      "epoch": 0.03843,
      "grad_norm": 0.662178693915001,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 3843
    },
    {
      "epoch": 0.03844,
      "grad_norm": 0.706679379810732,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 3844
    },
    {
      "epoch": 0.03845,
      "grad_norm": 0.7114665877564104,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3845
    },
    {
      "epoch": 0.03846,
      "grad_norm": 0.6429293496562273,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 3846
    },
    {
      "epoch": 0.03847,
      "grad_norm": 0.6208196779861823,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3847
    },
    {
      "epoch": 0.03848,
      "grad_norm": 0.5991718704769494,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 3848
    },
    {
      "epoch": 0.03849,
      "grad_norm": 0.6327221679341399,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 3849
    },
    {
      "epoch": 0.0385,
      "grad_norm": 0.6393009934280616,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 3850
    },
    {
      "epoch": 0.03851,
      "grad_norm": 0.7219030496541264,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 3851
    },
    {
      "epoch": 0.03852,
      "grad_norm": 0.741478675399483,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 3852
    },
    {
      "epoch": 0.03853,
      "grad_norm": 0.6467861453700433,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3853
    },
    {
      "epoch": 0.03854,
      "grad_norm": 0.698278301052957,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3854
    },
    {
      "epoch": 0.03855,
      "grad_norm": 0.7121845392242947,
      "learning_rate": 0.003,
      "loss": 4.1765,
      "step": 3855
    },
    {
      "epoch": 0.03856,
      "grad_norm": 0.6723670160555435,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 3856
    },
    {
      "epoch": 0.03857,
      "grad_norm": 0.6243401732081879,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 3857
    },
    {
      "epoch": 0.03858,
      "grad_norm": 0.5301294562723087,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 3858
    },
    {
      "epoch": 0.03859,
      "grad_norm": 0.50901761681548,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 3859
    },
    {
      "epoch": 0.0386,
      "grad_norm": 0.5147249879881411,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 3860
    },
    {
      "epoch": 0.03861,
      "grad_norm": 0.4939497869282373,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3861
    },
    {
      "epoch": 0.03862,
      "grad_norm": 0.5565978694443094,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 3862
    },
    {
      "epoch": 0.03863,
      "grad_norm": 0.6930312131073376,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 3863
    },
    {
      "epoch": 0.03864,
      "grad_norm": 0.7338788764609174,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 3864
    },
    {
      "epoch": 0.03865,
      "grad_norm": 0.716452591015328,
      "learning_rate": 0.003,
      "loss": 4.1614,
      "step": 3865
    },
    {
      "epoch": 0.03866,
      "grad_norm": 0.8035841485064186,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 3866
    },
    {
      "epoch": 0.03867,
      "grad_norm": 0.9790747131837212,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 3867
    },
    {
      "epoch": 0.03868,
      "grad_norm": 0.9557607039135098,
      "learning_rate": 0.003,
      "loss": 4.1685,
      "step": 3868
    },
    {
      "epoch": 0.03869,
      "grad_norm": 0.8269471619194556,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 3869
    },
    {
      "epoch": 0.0387,
      "grad_norm": 0.736530154455655,
      "learning_rate": 0.003,
      "loss": 4.1642,
      "step": 3870
    },
    {
      "epoch": 0.03871,
      "grad_norm": 0.5662896465927079,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 3871
    },
    {
      "epoch": 0.03872,
      "grad_norm": 0.7064773963599414,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 3872
    },
    {
      "epoch": 0.03873,
      "grad_norm": 0.7511440909675522,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 3873
    },
    {
      "epoch": 0.03874,
      "grad_norm": 0.8087487769850815,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 3874
    },
    {
      "epoch": 0.03875,
      "grad_norm": 0.840451180426337,
      "learning_rate": 0.003,
      "loss": 4.1553,
      "step": 3875
    },
    {
      "epoch": 0.03876,
      "grad_norm": 0.8711005630005544,
      "learning_rate": 0.003,
      "loss": 4.1888,
      "step": 3876
    },
    {
      "epoch": 0.03877,
      "grad_norm": 0.7489732856457578,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 3877
    },
    {
      "epoch": 0.03878,
      "grad_norm": 0.6479494485701219,
      "learning_rate": 0.003,
      "loss": 4.1753,
      "step": 3878
    },
    {
      "epoch": 0.03879,
      "grad_norm": 0.5974171547568563,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3879
    },
    {
      "epoch": 0.0388,
      "grad_norm": 0.6504260745071084,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 3880
    },
    {
      "epoch": 0.03881,
      "grad_norm": 0.6733083067264833,
      "learning_rate": 0.003,
      "loss": 4.1685,
      "step": 3881
    },
    {
      "epoch": 0.03882,
      "grad_norm": 0.660030960264416,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 3882
    },
    {
      "epoch": 0.03883,
      "grad_norm": 0.7215073977898953,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3883
    },
    {
      "epoch": 0.03884,
      "grad_norm": 0.6784371169912143,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 3884
    },
    {
      "epoch": 0.03885,
      "grad_norm": 0.6961417852606488,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 3885
    },
    {
      "epoch": 0.03886,
      "grad_norm": 0.6846762234353307,
      "learning_rate": 0.003,
      "loss": 4.1479,
      "step": 3886
    },
    {
      "epoch": 0.03887,
      "grad_norm": 0.6018705177852628,
      "learning_rate": 0.003,
      "loss": 4.1575,
      "step": 3887
    },
    {
      "epoch": 0.03888,
      "grad_norm": 0.7545299954550155,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 3888
    },
    {
      "epoch": 0.03889,
      "grad_norm": 0.9698514203515268,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 3889
    },
    {
      "epoch": 0.0389,
      "grad_norm": 1.1649616934398974,
      "learning_rate": 0.003,
      "loss": 4.1644,
      "step": 3890
    },
    {
      "epoch": 0.03891,
      "grad_norm": 0.7396396017755505,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 3891
    },
    {
      "epoch": 0.03892,
      "grad_norm": 0.7127971514741257,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 3892
    },
    {
      "epoch": 0.03893,
      "grad_norm": 0.8147903386909522,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 3893
    },
    {
      "epoch": 0.03894,
      "grad_norm": 0.8553199932435573,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 3894
    },
    {
      "epoch": 0.03895,
      "grad_norm": 0.8064783210870279,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 3895
    },
    {
      "epoch": 0.03896,
      "grad_norm": 0.7327396202925885,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 3896
    },
    {
      "epoch": 0.03897,
      "grad_norm": 0.6832165978250844,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 3897
    },
    {
      "epoch": 0.03898,
      "grad_norm": 0.7381820210282796,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3898
    },
    {
      "epoch": 0.03899,
      "grad_norm": 0.7120798145498066,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 3899
    },
    {
      "epoch": 0.039,
      "grad_norm": 0.706935890187933,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 3900
    },
    {
      "epoch": 0.03901,
      "grad_norm": 0.7172271819943744,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 3901
    },
    {
      "epoch": 0.03902,
      "grad_norm": 0.6436464145708447,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 3902
    },
    {
      "epoch": 0.03903,
      "grad_norm": 0.7120130287904197,
      "learning_rate": 0.003,
      "loss": 4.1839,
      "step": 3903
    },
    {
      "epoch": 0.03904,
      "grad_norm": 0.7363979573377551,
      "learning_rate": 0.003,
      "loss": 4.1613,
      "step": 3904
    },
    {
      "epoch": 0.03905,
      "grad_norm": 0.6989523475801707,
      "learning_rate": 0.003,
      "loss": 4.1777,
      "step": 3905
    },
    {
      "epoch": 0.03906,
      "grad_norm": 0.7664511025672687,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 3906
    },
    {
      "epoch": 0.03907,
      "grad_norm": 0.6973849941812954,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 3907
    },
    {
      "epoch": 0.03908,
      "grad_norm": 0.6968835116178974,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 3908
    },
    {
      "epoch": 0.03909,
      "grad_norm": 0.6214766339187717,
      "learning_rate": 0.003,
      "loss": 4.1479,
      "step": 3909
    },
    {
      "epoch": 0.0391,
      "grad_norm": 0.606928879147604,
      "learning_rate": 0.003,
      "loss": 4.1783,
      "step": 3910
    },
    {
      "epoch": 0.03911,
      "grad_norm": 0.6150917723982117,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3911
    },
    {
      "epoch": 0.03912,
      "grad_norm": 0.5315811614956929,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3912
    },
    {
      "epoch": 0.03913,
      "grad_norm": 0.49513392829810154,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 3913
    },
    {
      "epoch": 0.03914,
      "grad_norm": 0.5316802449083329,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3914
    },
    {
      "epoch": 0.03915,
      "grad_norm": 0.5092855956655066,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 3915
    },
    {
      "epoch": 0.03916,
      "grad_norm": 0.45924359376251017,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3916
    },
    {
      "epoch": 0.03917,
      "grad_norm": 0.45484869852430065,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 3917
    },
    {
      "epoch": 0.03918,
      "grad_norm": 0.4501493787308128,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 3918
    },
    {
      "epoch": 0.03919,
      "grad_norm": 0.5428853651303225,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 3919
    },
    {
      "epoch": 0.0392,
      "grad_norm": 0.7284545482558801,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 3920
    },
    {
      "epoch": 0.03921,
      "grad_norm": 0.9552348159276692,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 3921
    },
    {
      "epoch": 0.03922,
      "grad_norm": 1.132432474415542,
      "learning_rate": 0.003,
      "loss": 4.1914,
      "step": 3922
    },
    {
      "epoch": 0.03923,
      "grad_norm": 0.6932654094565733,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 3923
    },
    {
      "epoch": 0.03924,
      "grad_norm": 0.6532999077661216,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 3924
    },
    {
      "epoch": 0.03925,
      "grad_norm": 0.9135572051042363,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 3925
    },
    {
      "epoch": 0.03926,
      "grad_norm": 0.9923603322118285,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 3926
    },
    {
      "epoch": 0.03927,
      "grad_norm": 0.9789133497360833,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3927
    },
    {
      "epoch": 0.03928,
      "grad_norm": 0.8942775586441536,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 3928
    },
    {
      "epoch": 0.03929,
      "grad_norm": 1.0319586305334827,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 3929
    },
    {
      "epoch": 0.0393,
      "grad_norm": 1.0750467044406122,
      "learning_rate": 0.003,
      "loss": 4.1904,
      "step": 3930
    },
    {
      "epoch": 0.03931,
      "grad_norm": 0.9184595530968939,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 3931
    },
    {
      "epoch": 0.03932,
      "grad_norm": 0.7648894274462913,
      "learning_rate": 0.003,
      "loss": 4.1855,
      "step": 3932
    },
    {
      "epoch": 0.03933,
      "grad_norm": 0.724112317823659,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3933
    },
    {
      "epoch": 0.03934,
      "grad_norm": 0.6449540845441923,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 3934
    },
    {
      "epoch": 0.03935,
      "grad_norm": 0.7490619555451586,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3935
    },
    {
      "epoch": 0.03936,
      "grad_norm": 0.7177217614636706,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 3936
    },
    {
      "epoch": 0.03937,
      "grad_norm": 0.653893710648041,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 3937
    },
    {
      "epoch": 0.03938,
      "grad_norm": 0.689030208208419,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 3938
    },
    {
      "epoch": 0.03939,
      "grad_norm": 0.7098991679111472,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 3939
    },
    {
      "epoch": 0.0394,
      "grad_norm": 0.6955232506989616,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 3940
    },
    {
      "epoch": 0.03941,
      "grad_norm": 0.6598355292317952,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 3941
    },
    {
      "epoch": 0.03942,
      "grad_norm": 0.738032728669511,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 3942
    },
    {
      "epoch": 0.03943,
      "grad_norm": 0.8081766812397826,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 3943
    },
    {
      "epoch": 0.03944,
      "grad_norm": 0.8593029564204311,
      "learning_rate": 0.003,
      "loss": 4.1857,
      "step": 3944
    },
    {
      "epoch": 0.03945,
      "grad_norm": 0.8814910808518954,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 3945
    },
    {
      "epoch": 0.03946,
      "grad_norm": 0.8816608783626361,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3946
    },
    {
      "epoch": 0.03947,
      "grad_norm": 0.8064135711097247,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 3947
    },
    {
      "epoch": 0.03948,
      "grad_norm": 0.8804914835705606,
      "learning_rate": 0.003,
      "loss": 4.2055,
      "step": 3948
    },
    {
      "epoch": 0.03949,
      "grad_norm": 0.8976280856029432,
      "learning_rate": 0.003,
      "loss": 4.1479,
      "step": 3949
    },
    {
      "epoch": 0.0395,
      "grad_norm": 0.7915245308890937,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3950
    },
    {
      "epoch": 0.03951,
      "grad_norm": 0.7023597435152198,
      "learning_rate": 0.003,
      "loss": 4.2015,
      "step": 3951
    },
    {
      "epoch": 0.03952,
      "grad_norm": 0.5970926719547551,
      "learning_rate": 0.003,
      "loss": 4.1776,
      "step": 3952
    },
    {
      "epoch": 0.03953,
      "grad_norm": 0.5127193759525883,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 3953
    },
    {
      "epoch": 0.03954,
      "grad_norm": 0.5083765022250437,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 3954
    },
    {
      "epoch": 0.03955,
      "grad_norm": 0.5581234697293194,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3955
    },
    {
      "epoch": 0.03956,
      "grad_norm": 0.5875448440882535,
      "learning_rate": 0.003,
      "loss": 4.1566,
      "step": 3956
    },
    {
      "epoch": 0.03957,
      "grad_norm": 0.597994191226493,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 3957
    },
    {
      "epoch": 0.03958,
      "grad_norm": 0.5717890299511298,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 3958
    },
    {
      "epoch": 0.03959,
      "grad_norm": 0.5436034530009164,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 3959
    },
    {
      "epoch": 0.0396,
      "grad_norm": 0.49029266691650636,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 3960
    },
    {
      "epoch": 0.03961,
      "grad_norm": 0.517401344634264,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 3961
    },
    {
      "epoch": 0.03962,
      "grad_norm": 0.5330594146947688,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3962
    },
    {
      "epoch": 0.03963,
      "grad_norm": 0.5905667821025009,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 3963
    },
    {
      "epoch": 0.03964,
      "grad_norm": 0.558494601973704,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 3964
    },
    {
      "epoch": 0.03965,
      "grad_norm": 0.5647463812091732,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 3965
    },
    {
      "epoch": 0.03966,
      "grad_norm": 0.5672438671895076,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 3966
    },
    {
      "epoch": 0.03967,
      "grad_norm": 0.5990092369456635,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 3967
    },
    {
      "epoch": 0.03968,
      "grad_norm": 0.5306380158511645,
      "learning_rate": 0.003,
      "loss": 4.1653,
      "step": 3968
    },
    {
      "epoch": 0.03969,
      "grad_norm": 0.5335941140441545,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 3969
    },
    {
      "epoch": 0.0397,
      "grad_norm": 0.5101263285016783,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 3970
    },
    {
      "epoch": 0.03971,
      "grad_norm": 0.4861456325644423,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 3971
    },
    {
      "epoch": 0.03972,
      "grad_norm": 0.5082221340172262,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 3972
    },
    {
      "epoch": 0.03973,
      "grad_norm": 0.6064259836747726,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 3973
    },
    {
      "epoch": 0.03974,
      "grad_norm": 0.8810228923861501,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 3974
    },
    {
      "epoch": 0.03975,
      "grad_norm": 1.1594440991062704,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 3975
    },
    {
      "epoch": 0.03976,
      "grad_norm": 0.6242516132147851,
      "learning_rate": 0.003,
      "loss": 4.1795,
      "step": 3976
    },
    {
      "epoch": 0.03977,
      "grad_norm": 0.6310756758939645,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 3977
    },
    {
      "epoch": 0.03978,
      "grad_norm": 0.9800843159523053,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 3978
    },
    {
      "epoch": 0.03979,
      "grad_norm": 0.8894332385272683,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3979
    },
    {
      "epoch": 0.0398,
      "grad_norm": 0.5906456865918563,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 3980
    },
    {
      "epoch": 0.03981,
      "grad_norm": 0.6582973485295706,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 3981
    },
    {
      "epoch": 0.03982,
      "grad_norm": 0.7206583498121457,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 3982
    },
    {
      "epoch": 0.03983,
      "grad_norm": 0.8701732888100308,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 3983
    },
    {
      "epoch": 0.03984,
      "grad_norm": 0.8661641192026968,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 3984
    },
    {
      "epoch": 0.03985,
      "grad_norm": 0.9243378871553372,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 3985
    },
    {
      "epoch": 0.03986,
      "grad_norm": 0.9956369167150667,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 3986
    },
    {
      "epoch": 0.03987,
      "grad_norm": 0.8887499652438708,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 3987
    },
    {
      "epoch": 0.03988,
      "grad_norm": 0.7725996597052387,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3988
    },
    {
      "epoch": 0.03989,
      "grad_norm": 0.9540752093113514,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 3989
    },
    {
      "epoch": 0.0399,
      "grad_norm": 0.9866842358730207,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 3990
    },
    {
      "epoch": 0.03991,
      "grad_norm": 1.0032653576960733,
      "learning_rate": 0.003,
      "loss": 4.1979,
      "step": 3991
    },
    {
      "epoch": 0.03992,
      "grad_norm": 0.9799832060702593,
      "learning_rate": 0.003,
      "loss": 4.1809,
      "step": 3992
    },
    {
      "epoch": 0.03993,
      "grad_norm": 0.9143397893823952,
      "learning_rate": 0.003,
      "loss": 4.1772,
      "step": 3993
    },
    {
      "epoch": 0.03994,
      "grad_norm": 0.8867228330129578,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 3994
    },
    {
      "epoch": 0.03995,
      "grad_norm": 0.9777127443180489,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 3995
    },
    {
      "epoch": 0.03996,
      "grad_norm": 1.121916971358567,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 3996
    },
    {
      "epoch": 0.03997,
      "grad_norm": 1.0260835587044206,
      "learning_rate": 0.003,
      "loss": 4.21,
      "step": 3997
    },
    {
      "epoch": 0.03998,
      "grad_norm": 0.9733483073411772,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 3998
    },
    {
      "epoch": 0.03999,
      "grad_norm": 0.9253787235141333,
      "learning_rate": 0.003,
      "loss": 4.2098,
      "step": 3999
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.860383401913097,
      "learning_rate": 0.003,
      "loss": 4.2101,
      "step": 4000
    },
    {
      "epoch": 0.04001,
      "grad_norm": 0.8527166028367188,
      "learning_rate": 0.003,
      "loss": 4.2089,
      "step": 4001
    },
    {
      "epoch": 0.04002,
      "grad_norm": 0.7955822227821913,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 4002
    },
    {
      "epoch": 0.04003,
      "grad_norm": 0.8159059659987256,
      "learning_rate": 0.003,
      "loss": 4.1966,
      "step": 4003
    },
    {
      "epoch": 0.04004,
      "grad_norm": 0.8242608605578613,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 4004
    },
    {
      "epoch": 0.04005,
      "grad_norm": 0.7442104756929572,
      "learning_rate": 0.003,
      "loss": 4.2052,
      "step": 4005
    },
    {
      "epoch": 0.04006,
      "grad_norm": 0.6862974810950646,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4006
    },
    {
      "epoch": 0.04007,
      "grad_norm": 0.6713665895022044,
      "learning_rate": 0.003,
      "loss": 4.1851,
      "step": 4007
    },
    {
      "epoch": 0.04008,
      "grad_norm": 0.6515792038366235,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 4008
    },
    {
      "epoch": 0.04009,
      "grad_norm": 0.6670829454786272,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 4009
    },
    {
      "epoch": 0.0401,
      "grad_norm": 0.5890722285246915,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 4010
    },
    {
      "epoch": 0.04011,
      "grad_norm": 0.5154042218197626,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 4011
    },
    {
      "epoch": 0.04012,
      "grad_norm": 0.511904267386738,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 4012
    },
    {
      "epoch": 0.04013,
      "grad_norm": 0.4301130431551674,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 4013
    },
    {
      "epoch": 0.04014,
      "grad_norm": 0.4755384019802336,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 4014
    },
    {
      "epoch": 0.04015,
      "grad_norm": 0.4526642613482027,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 4015
    },
    {
      "epoch": 0.04016,
      "grad_norm": 0.5025871602100715,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 4016
    },
    {
      "epoch": 0.04017,
      "grad_norm": 0.613279269884007,
      "learning_rate": 0.003,
      "loss": 4.1847,
      "step": 4017
    },
    {
      "epoch": 0.04018,
      "grad_norm": 0.8281359691209577,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 4018
    },
    {
      "epoch": 0.04019,
      "grad_norm": 0.941035832708172,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 4019
    },
    {
      "epoch": 0.0402,
      "grad_norm": 0.8795192610150039,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 4020
    },
    {
      "epoch": 0.04021,
      "grad_norm": 0.6515100017542129,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 4021
    },
    {
      "epoch": 0.04022,
      "grad_norm": 0.5063361011739121,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 4022
    },
    {
      "epoch": 0.04023,
      "grad_norm": 0.6481572855287395,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 4023
    },
    {
      "epoch": 0.04024,
      "grad_norm": 0.7010903711503769,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 4024
    },
    {
      "epoch": 0.04025,
      "grad_norm": 0.7840447772680712,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 4025
    },
    {
      "epoch": 0.04026,
      "grad_norm": 0.7341068822639932,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 4026
    },
    {
      "epoch": 0.04027,
      "grad_norm": 0.6238472142782437,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 4027
    },
    {
      "epoch": 0.04028,
      "grad_norm": 0.4935095791712522,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 4028
    },
    {
      "epoch": 0.04029,
      "grad_norm": 0.4905704871714714,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 4029
    },
    {
      "epoch": 0.0403,
      "grad_norm": 0.5059882465087885,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 4030
    },
    {
      "epoch": 0.04031,
      "grad_norm": 0.49185506270358387,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4031
    },
    {
      "epoch": 0.04032,
      "grad_norm": 0.5505039809907691,
      "learning_rate": 0.003,
      "loss": 4.1625,
      "step": 4032
    },
    {
      "epoch": 0.04033,
      "grad_norm": 0.5349055922421693,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 4033
    },
    {
      "epoch": 0.04034,
      "grad_norm": 0.4887336563792077,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 4034
    },
    {
      "epoch": 0.04035,
      "grad_norm": 0.5026476582372887,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 4035
    },
    {
      "epoch": 0.04036,
      "grad_norm": 0.5741733182678186,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 4036
    },
    {
      "epoch": 0.04037,
      "grad_norm": 0.6338724243857007,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 4037
    },
    {
      "epoch": 0.04038,
      "grad_norm": 0.5866801909982847,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4038
    },
    {
      "epoch": 0.04039,
      "grad_norm": 0.4469263176133214,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 4039
    },
    {
      "epoch": 0.0404,
      "grad_norm": 0.5203352096732194,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 4040
    },
    {
      "epoch": 0.04041,
      "grad_norm": 0.5903209349905957,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 4041
    },
    {
      "epoch": 0.04042,
      "grad_norm": 0.7850173938632014,
      "learning_rate": 0.003,
      "loss": 4.1729,
      "step": 4042
    },
    {
      "epoch": 0.04043,
      "grad_norm": 1.0374432013836312,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 4043
    },
    {
      "epoch": 0.04044,
      "grad_norm": 1.0008411828594295,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 4044
    },
    {
      "epoch": 0.04045,
      "grad_norm": 0.721767328260429,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 4045
    },
    {
      "epoch": 0.04046,
      "grad_norm": 0.6027053735164074,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 4046
    },
    {
      "epoch": 0.04047,
      "grad_norm": 0.8146571762133998,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 4047
    },
    {
      "epoch": 0.04048,
      "grad_norm": 0.8947220522958645,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 4048
    },
    {
      "epoch": 0.04049,
      "grad_norm": 0.9307304863945178,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 4049
    },
    {
      "epoch": 0.0405,
      "grad_norm": 0.7045766049752963,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 4050
    },
    {
      "epoch": 0.04051,
      "grad_norm": 0.6846375832757631,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 4051
    },
    {
      "epoch": 0.04052,
      "grad_norm": 0.8615665313113385,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 4052
    },
    {
      "epoch": 0.04053,
      "grad_norm": 0.9373878770279569,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 4053
    },
    {
      "epoch": 0.04054,
      "grad_norm": 0.8371593982203549,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 4054
    },
    {
      "epoch": 0.04055,
      "grad_norm": 0.6967664819995452,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 4055
    },
    {
      "epoch": 0.04056,
      "grad_norm": 0.6741494833756686,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 4056
    },
    {
      "epoch": 0.04057,
      "grad_norm": 0.8030676782045835,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 4057
    },
    {
      "epoch": 0.04058,
      "grad_norm": 0.774339384731219,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 4058
    },
    {
      "epoch": 0.04059,
      "grad_norm": 0.8430830883517224,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 4059
    },
    {
      "epoch": 0.0406,
      "grad_norm": 0.6742657640683123,
      "learning_rate": 0.003,
      "loss": 4.1739,
      "step": 4060
    },
    {
      "epoch": 0.04061,
      "grad_norm": 0.6572926238853734,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 4061
    },
    {
      "epoch": 0.04062,
      "grad_norm": 0.5997992092489303,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 4062
    },
    {
      "epoch": 0.04063,
      "grad_norm": 0.5451542397429993,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 4063
    },
    {
      "epoch": 0.04064,
      "grad_norm": 0.5706051829517365,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 4064
    },
    {
      "epoch": 0.04065,
      "grad_norm": 0.6573749206271802,
      "learning_rate": 0.003,
      "loss": 4.1936,
      "step": 4065
    },
    {
      "epoch": 0.04066,
      "grad_norm": 0.6711377067040638,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4066
    },
    {
      "epoch": 0.04067,
      "grad_norm": 0.6151905993857413,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 4067
    },
    {
      "epoch": 0.04068,
      "grad_norm": 0.610139511076282,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 4068
    },
    {
      "epoch": 0.04069,
      "grad_norm": 0.6435562548550586,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 4069
    },
    {
      "epoch": 0.0407,
      "grad_norm": 0.5473010440748541,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4070
    },
    {
      "epoch": 0.04071,
      "grad_norm": 0.601566675296323,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 4071
    },
    {
      "epoch": 0.04072,
      "grad_norm": 0.6509579069490173,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 4072
    },
    {
      "epoch": 0.04073,
      "grad_norm": 0.5973245201602951,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 4073
    },
    {
      "epoch": 0.04074,
      "grad_norm": 0.607520676438575,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 4074
    },
    {
      "epoch": 0.04075,
      "grad_norm": 0.8288720279764606,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 4075
    },
    {
      "epoch": 0.04076,
      "grad_norm": 0.9232154672269041,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 4076
    },
    {
      "epoch": 0.04077,
      "grad_norm": 0.9604554082892389,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 4077
    },
    {
      "epoch": 0.04078,
      "grad_norm": 0.9176998486040583,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 4078
    },
    {
      "epoch": 0.04079,
      "grad_norm": 0.7781092418368329,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4079
    },
    {
      "epoch": 0.0408,
      "grad_norm": 0.7144390148261879,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 4080
    },
    {
      "epoch": 0.04081,
      "grad_norm": 0.7985914865590669,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 4081
    },
    {
      "epoch": 0.04082,
      "grad_norm": 0.8262338793037239,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 4082
    },
    {
      "epoch": 0.04083,
      "grad_norm": 0.7465376916541703,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4083
    },
    {
      "epoch": 0.04084,
      "grad_norm": 0.6728219238709465,
      "learning_rate": 0.003,
      "loss": 4.1811,
      "step": 4084
    },
    {
      "epoch": 0.04085,
      "grad_norm": 0.755821930925202,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 4085
    },
    {
      "epoch": 0.04086,
      "grad_norm": 0.637261349390268,
      "learning_rate": 0.003,
      "loss": 4.1791,
      "step": 4086
    },
    {
      "epoch": 0.04087,
      "grad_norm": 0.5906659081612129,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 4087
    },
    {
      "epoch": 0.04088,
      "grad_norm": 0.547995647785304,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 4088
    },
    {
      "epoch": 0.04089,
      "grad_norm": 0.5708520101184565,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4089
    },
    {
      "epoch": 0.0409,
      "grad_norm": 0.6099350559443698,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 4090
    },
    {
      "epoch": 0.04091,
      "grad_norm": 0.696520536628286,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 4091
    },
    {
      "epoch": 0.04092,
      "grad_norm": 0.8551793129611549,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 4092
    },
    {
      "epoch": 0.04093,
      "grad_norm": 1.0511160284337668,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4093
    },
    {
      "epoch": 0.04094,
      "grad_norm": 0.9303098870853981,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 4094
    },
    {
      "epoch": 0.04095,
      "grad_norm": 0.7394320217083529,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4095
    },
    {
      "epoch": 0.04096,
      "grad_norm": 0.7856690666821323,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 4096
    },
    {
      "epoch": 0.04097,
      "grad_norm": 0.7724952999564255,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4097
    },
    {
      "epoch": 0.04098,
      "grad_norm": 0.6418338461235239,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 4098
    },
    {
      "epoch": 0.04099,
      "grad_norm": 0.6141872194967258,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 4099
    },
    {
      "epoch": 0.041,
      "grad_norm": 0.6337061381438701,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 4100
    },
    {
      "epoch": 0.04101,
      "grad_norm": 0.7005605990414404,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 4101
    },
    {
      "epoch": 0.04102,
      "grad_norm": 0.7781518273799513,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 4102
    },
    {
      "epoch": 0.04103,
      "grad_norm": 0.7717092605525707,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 4103
    },
    {
      "epoch": 0.04104,
      "grad_norm": 0.7998045960483534,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 4104
    },
    {
      "epoch": 0.04105,
      "grad_norm": 0.7266397106851961,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 4105
    },
    {
      "epoch": 0.04106,
      "grad_norm": 0.7375729801536005,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 4106
    },
    {
      "epoch": 0.04107,
      "grad_norm": 0.8517602436934472,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 4107
    },
    {
      "epoch": 0.04108,
      "grad_norm": 0.7155354553416942,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4108
    },
    {
      "epoch": 0.04109,
      "grad_norm": 0.8472831379025649,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 4109
    },
    {
      "epoch": 0.0411,
      "grad_norm": 0.9994412956163683,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 4110
    },
    {
      "epoch": 0.04111,
      "grad_norm": 1.0865359108420907,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 4111
    },
    {
      "epoch": 0.04112,
      "grad_norm": 0.7944687116955357,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 4112
    },
    {
      "epoch": 0.04113,
      "grad_norm": 0.6695248184455779,
      "learning_rate": 0.003,
      "loss": 4.2009,
      "step": 4113
    },
    {
      "epoch": 0.04114,
      "grad_norm": 0.8304852025142981,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 4114
    },
    {
      "epoch": 0.04115,
      "grad_norm": 0.9696455727697199,
      "learning_rate": 0.003,
      "loss": 4.164,
      "step": 4115
    },
    {
      "epoch": 0.04116,
      "grad_norm": 1.0300022050612345,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 4116
    },
    {
      "epoch": 0.04117,
      "grad_norm": 0.7267792671268358,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 4117
    },
    {
      "epoch": 0.04118,
      "grad_norm": 0.6123016813019814,
      "learning_rate": 0.003,
      "loss": 4.1925,
      "step": 4118
    },
    {
      "epoch": 0.04119,
      "grad_norm": 0.6465840606735817,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 4119
    },
    {
      "epoch": 0.0412,
      "grad_norm": 0.7384604480296841,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 4120
    },
    {
      "epoch": 0.04121,
      "grad_norm": 0.6577587889661327,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 4121
    },
    {
      "epoch": 0.04122,
      "grad_norm": 0.5559068248334778,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4122
    },
    {
      "epoch": 0.04123,
      "grad_norm": 0.596362393082762,
      "learning_rate": 0.003,
      "loss": 4.1613,
      "step": 4123
    },
    {
      "epoch": 0.04124,
      "grad_norm": 0.5443723484819623,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 4124
    },
    {
      "epoch": 0.04125,
      "grad_norm": 0.5563286329797662,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 4125
    },
    {
      "epoch": 0.04126,
      "grad_norm": 0.523384653689282,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4126
    },
    {
      "epoch": 0.04127,
      "grad_norm": 0.4495176170519381,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4127
    },
    {
      "epoch": 0.04128,
      "grad_norm": 0.5072318783524735,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 4128
    },
    {
      "epoch": 0.04129,
      "grad_norm": 0.5052405740035331,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 4129
    },
    {
      "epoch": 0.0413,
      "grad_norm": 0.575809172371195,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4130
    },
    {
      "epoch": 0.04131,
      "grad_norm": 0.6143169354531615,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 4131
    },
    {
      "epoch": 0.04132,
      "grad_norm": 0.6223256630595382,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 4132
    },
    {
      "epoch": 0.04133,
      "grad_norm": 0.5894212387409717,
      "learning_rate": 0.003,
      "loss": 4.161,
      "step": 4133
    },
    {
      "epoch": 0.04134,
      "grad_norm": 0.5069566465455504,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4134
    },
    {
      "epoch": 0.04135,
      "grad_norm": 0.47827820618941536,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 4135
    },
    {
      "epoch": 0.04136,
      "grad_norm": 0.4653725022496872,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 4136
    },
    {
      "epoch": 0.04137,
      "grad_norm": 0.47773052230780755,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 4137
    },
    {
      "epoch": 0.04138,
      "grad_norm": 0.5380882786268274,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4138
    },
    {
      "epoch": 0.04139,
      "grad_norm": 0.6469827605874696,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 4139
    },
    {
      "epoch": 0.0414,
      "grad_norm": 0.6947344870174545,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 4140
    },
    {
      "epoch": 0.04141,
      "grad_norm": 0.748345923041137,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 4141
    },
    {
      "epoch": 0.04142,
      "grad_norm": 0.7319349091665185,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 4142
    },
    {
      "epoch": 0.04143,
      "grad_norm": 0.628358425706577,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 4143
    },
    {
      "epoch": 0.04144,
      "grad_norm": 0.6834841445843723,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 4144
    },
    {
      "epoch": 0.04145,
      "grad_norm": 0.7376058073772749,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 4145
    },
    {
      "epoch": 0.04146,
      "grad_norm": 0.7587407207350676,
      "learning_rate": 0.003,
      "loss": 4.1526,
      "step": 4146
    },
    {
      "epoch": 0.04147,
      "grad_norm": 0.7947181963356904,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 4147
    },
    {
      "epoch": 0.04148,
      "grad_norm": 0.7182381427435356,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 4148
    },
    {
      "epoch": 0.04149,
      "grad_norm": 0.7323007766800972,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 4149
    },
    {
      "epoch": 0.0415,
      "grad_norm": 0.857281738554918,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 4150
    },
    {
      "epoch": 0.04151,
      "grad_norm": 0.9753058360121778,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 4151
    },
    {
      "epoch": 0.04152,
      "grad_norm": 1.0237271601242721,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 4152
    },
    {
      "epoch": 0.04153,
      "grad_norm": 0.9306677581318379,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 4153
    },
    {
      "epoch": 0.04154,
      "grad_norm": 0.8382038016807482,
      "learning_rate": 0.003,
      "loss": 4.1686,
      "step": 4154
    },
    {
      "epoch": 0.04155,
      "grad_norm": 0.7006562175213897,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4155
    },
    {
      "epoch": 0.04156,
      "grad_norm": 0.6295351230747147,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 4156
    },
    {
      "epoch": 0.04157,
      "grad_norm": 0.7233308933121653,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 4157
    },
    {
      "epoch": 0.04158,
      "grad_norm": 0.7870559063957332,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 4158
    },
    {
      "epoch": 0.04159,
      "grad_norm": 0.9866760938792348,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 4159
    },
    {
      "epoch": 0.0416,
      "grad_norm": 1.0470738285698136,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 4160
    },
    {
      "epoch": 0.04161,
      "grad_norm": 0.9307304193810675,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 4161
    },
    {
      "epoch": 0.04162,
      "grad_norm": 0.8721482244521436,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 4162
    },
    {
      "epoch": 0.04163,
      "grad_norm": 0.8678721020792026,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 4163
    },
    {
      "epoch": 0.04164,
      "grad_norm": 0.8226440605014521,
      "learning_rate": 0.003,
      "loss": 4.1918,
      "step": 4164
    },
    {
      "epoch": 0.04165,
      "grad_norm": 0.825975020523602,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 4165
    },
    {
      "epoch": 0.04166,
      "grad_norm": 0.8820368404280395,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 4166
    },
    {
      "epoch": 0.04167,
      "grad_norm": 0.8857197236082065,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 4167
    },
    {
      "epoch": 0.04168,
      "grad_norm": 0.9094188409333465,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 4168
    },
    {
      "epoch": 0.04169,
      "grad_norm": 0.8608115262399164,
      "learning_rate": 0.003,
      "loss": 4.1926,
      "step": 4169
    },
    {
      "epoch": 0.0417,
      "grad_norm": 0.8339384653513617,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 4170
    },
    {
      "epoch": 0.04171,
      "grad_norm": 0.8053429340063298,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 4171
    },
    {
      "epoch": 0.04172,
      "grad_norm": 0.8778996219659019,
      "learning_rate": 0.003,
      "loss": 4.1968,
      "step": 4172
    },
    {
      "epoch": 0.04173,
      "grad_norm": 0.9642055884954371,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 4173
    },
    {
      "epoch": 0.04174,
      "grad_norm": 1.0380118480990967,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 4174
    },
    {
      "epoch": 0.04175,
      "grad_norm": 0.9536264797990276,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 4175
    },
    {
      "epoch": 0.04176,
      "grad_norm": 0.8997005452165456,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 4176
    },
    {
      "epoch": 0.04177,
      "grad_norm": 0.9315086547062951,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 4177
    },
    {
      "epoch": 0.04178,
      "grad_norm": 0.7486982272609372,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 4178
    },
    {
      "epoch": 0.04179,
      "grad_norm": 0.7322985384737938,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 4179
    },
    {
      "epoch": 0.0418,
      "grad_norm": 0.6634118485342885,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 4180
    },
    {
      "epoch": 0.04181,
      "grad_norm": 0.6257131964317331,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 4181
    },
    {
      "epoch": 0.04182,
      "grad_norm": 0.6155966650745514,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 4182
    },
    {
      "epoch": 0.04183,
      "grad_norm": 0.5888122942739422,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4183
    },
    {
      "epoch": 0.04184,
      "grad_norm": 0.48396597336103814,
      "learning_rate": 0.003,
      "loss": 4.1975,
      "step": 4184
    },
    {
      "epoch": 0.04185,
      "grad_norm": 0.46085090618271063,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 4185
    },
    {
      "epoch": 0.04186,
      "grad_norm": 0.40263981014725103,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 4186
    },
    {
      "epoch": 0.04187,
      "grad_norm": 0.4072854325073614,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 4187
    },
    {
      "epoch": 0.04188,
      "grad_norm": 0.3949031488342765,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 4188
    },
    {
      "epoch": 0.04189,
      "grad_norm": 0.4614400695553244,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4189
    },
    {
      "epoch": 0.0419,
      "grad_norm": 0.591341946932479,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 4190
    },
    {
      "epoch": 0.04191,
      "grad_norm": 0.6794599625720497,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 4191
    },
    {
      "epoch": 0.04192,
      "grad_norm": 0.8639843192836074,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 4192
    },
    {
      "epoch": 0.04193,
      "grad_norm": 0.9261826031541774,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 4193
    },
    {
      "epoch": 0.04194,
      "grad_norm": 0.6816042869205202,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 4194
    },
    {
      "epoch": 0.04195,
      "grad_norm": 0.4846458487192246,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 4195
    },
    {
      "epoch": 0.04196,
      "grad_norm": 0.6733457250112305,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 4196
    },
    {
      "epoch": 0.04197,
      "grad_norm": 0.8628147041804719,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 4197
    },
    {
      "epoch": 0.04198,
      "grad_norm": 0.8069270994807441,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 4198
    },
    {
      "epoch": 0.04199,
      "grad_norm": 0.5843819059317946,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 4199
    },
    {
      "epoch": 0.042,
      "grad_norm": 0.6524001973236165,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 4200
    },
    {
      "epoch": 0.04201,
      "grad_norm": 0.7455750389366074,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 4201
    },
    {
      "epoch": 0.04202,
      "grad_norm": 0.8073311014615949,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 4202
    },
    {
      "epoch": 0.04203,
      "grad_norm": 0.6797266710402026,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 4203
    },
    {
      "epoch": 0.04204,
      "grad_norm": 0.6818305150769544,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 4204
    },
    {
      "epoch": 0.04205,
      "grad_norm": 0.5915119548922819,
      "learning_rate": 0.003,
      "loss": 4.1526,
      "step": 4205
    },
    {
      "epoch": 0.04206,
      "grad_norm": 0.5385651548116344,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 4206
    },
    {
      "epoch": 0.04207,
      "grad_norm": 0.48635296552930585,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 4207
    },
    {
      "epoch": 0.04208,
      "grad_norm": 0.501480728550346,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4208
    },
    {
      "epoch": 0.04209,
      "grad_norm": 0.5039540680425979,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 4209
    },
    {
      "epoch": 0.0421,
      "grad_norm": 0.5190305179039434,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 4210
    },
    {
      "epoch": 0.04211,
      "grad_norm": 0.5620311787567622,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 4211
    },
    {
      "epoch": 0.04212,
      "grad_norm": 0.5635968224621314,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 4212
    },
    {
      "epoch": 0.04213,
      "grad_norm": 0.5728113060887386,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 4213
    },
    {
      "epoch": 0.04214,
      "grad_norm": 0.548258707730465,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 4214
    },
    {
      "epoch": 0.04215,
      "grad_norm": 0.599545067228449,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 4215
    },
    {
      "epoch": 0.04216,
      "grad_norm": 0.5748773014990753,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 4216
    },
    {
      "epoch": 0.04217,
      "grad_norm": 0.48220599086889576,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 4217
    },
    {
      "epoch": 0.04218,
      "grad_norm": 0.5634734523353527,
      "learning_rate": 0.003,
      "loss": 4.1642,
      "step": 4218
    },
    {
      "epoch": 0.04219,
      "grad_norm": 0.6074224940320497,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 4219
    },
    {
      "epoch": 0.0422,
      "grad_norm": 0.6971835957336069,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 4220
    },
    {
      "epoch": 0.04221,
      "grad_norm": 0.8744022212709924,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 4221
    },
    {
      "epoch": 0.04222,
      "grad_norm": 1.091644449624445,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 4222
    },
    {
      "epoch": 0.04223,
      "grad_norm": 0.9670759098939631,
      "learning_rate": 0.003,
      "loss": 4.1714,
      "step": 4223
    },
    {
      "epoch": 0.04224,
      "grad_norm": 1.0117671228316643,
      "learning_rate": 0.003,
      "loss": 4.1837,
      "step": 4224
    },
    {
      "epoch": 0.04225,
      "grad_norm": 0.9113076215312503,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 4225
    },
    {
      "epoch": 0.04226,
      "grad_norm": 0.8960930319481113,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4226
    },
    {
      "epoch": 0.04227,
      "grad_norm": 0.9167424550159433,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 4227
    },
    {
      "epoch": 0.04228,
      "grad_norm": 0.967478343035116,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 4228
    },
    {
      "epoch": 0.04229,
      "grad_norm": 0.8656039877374752,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 4229
    },
    {
      "epoch": 0.0423,
      "grad_norm": 0.8626263746774161,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 4230
    },
    {
      "epoch": 0.04231,
      "grad_norm": 0.7823486614059999,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 4231
    },
    {
      "epoch": 0.04232,
      "grad_norm": 0.6738192762506559,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 4232
    },
    {
      "epoch": 0.04233,
      "grad_norm": 0.6857804646466646,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 4233
    },
    {
      "epoch": 0.04234,
      "grad_norm": 0.7651833364067767,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 4234
    },
    {
      "epoch": 0.04235,
      "grad_norm": 0.8557695250207406,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 4235
    },
    {
      "epoch": 0.04236,
      "grad_norm": 0.8879388664685244,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4236
    },
    {
      "epoch": 0.04237,
      "grad_norm": 0.913885936433401,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 4237
    },
    {
      "epoch": 0.04238,
      "grad_norm": 0.8386396394298068,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 4238
    },
    {
      "epoch": 0.04239,
      "grad_norm": 0.8690235809658854,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 4239
    },
    {
      "epoch": 0.0424,
      "grad_norm": 0.8289113096147429,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 4240
    },
    {
      "epoch": 0.04241,
      "grad_norm": 0.7332475305569869,
      "learning_rate": 0.003,
      "loss": 4.1777,
      "step": 4241
    },
    {
      "epoch": 0.04242,
      "grad_norm": 0.6138420318092145,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 4242
    },
    {
      "epoch": 0.04243,
      "grad_norm": 0.6214280439461072,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 4243
    },
    {
      "epoch": 0.04244,
      "grad_norm": 0.618625833924493,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 4244
    },
    {
      "epoch": 0.04245,
      "grad_norm": 0.5938287908990628,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 4245
    },
    {
      "epoch": 0.04246,
      "grad_norm": 0.5233831349524802,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 4246
    },
    {
      "epoch": 0.04247,
      "grad_norm": 0.4617503229970775,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 4247
    },
    {
      "epoch": 0.04248,
      "grad_norm": 0.4580443757090383,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 4248
    },
    {
      "epoch": 0.04249,
      "grad_norm": 0.4036456069229838,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 4249
    },
    {
      "epoch": 0.0425,
      "grad_norm": 0.4380751494962389,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 4250
    },
    {
      "epoch": 0.04251,
      "grad_norm": 0.5168262583557149,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 4251
    },
    {
      "epoch": 0.04252,
      "grad_norm": 0.6808353915492127,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 4252
    },
    {
      "epoch": 0.04253,
      "grad_norm": 0.8432737064416014,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 4253
    },
    {
      "epoch": 0.04254,
      "grad_norm": 0.852211211018834,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 4254
    },
    {
      "epoch": 0.04255,
      "grad_norm": 0.6240445618140698,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 4255
    },
    {
      "epoch": 0.04256,
      "grad_norm": 0.4411537713654708,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 4256
    },
    {
      "epoch": 0.04257,
      "grad_norm": 0.5991668514983637,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 4257
    },
    {
      "epoch": 0.04258,
      "grad_norm": 0.7037912926105352,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 4258
    },
    {
      "epoch": 0.04259,
      "grad_norm": 0.6762366578254241,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 4259
    },
    {
      "epoch": 0.0426,
      "grad_norm": 0.6629398663628234,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4260
    },
    {
      "epoch": 0.04261,
      "grad_norm": 0.7122714093545782,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 4261
    },
    {
      "epoch": 0.04262,
      "grad_norm": 0.6594037833976022,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 4262
    },
    {
      "epoch": 0.04263,
      "grad_norm": 0.6354881188170681,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 4263
    },
    {
      "epoch": 0.04264,
      "grad_norm": 0.6654557887834439,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 4264
    },
    {
      "epoch": 0.04265,
      "grad_norm": 0.6938449312414295,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4265
    },
    {
      "epoch": 0.04266,
      "grad_norm": 0.7221897294325286,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4266
    },
    {
      "epoch": 0.04267,
      "grad_norm": 0.7436634345125996,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 4267
    },
    {
      "epoch": 0.04268,
      "grad_norm": 0.8458877078713634,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 4268
    },
    {
      "epoch": 0.04269,
      "grad_norm": 0.8904618229753511,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 4269
    },
    {
      "epoch": 0.0427,
      "grad_norm": 0.7112249953312418,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 4270
    },
    {
      "epoch": 0.04271,
      "grad_norm": 0.7216953907198649,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4271
    },
    {
      "epoch": 0.04272,
      "grad_norm": 0.7847449594647588,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 4272
    },
    {
      "epoch": 0.04273,
      "grad_norm": 1.004517275551493,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 4273
    },
    {
      "epoch": 0.04274,
      "grad_norm": 1.255471199279273,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 4274
    },
    {
      "epoch": 0.04275,
      "grad_norm": 0.6280388986370492,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 4275
    },
    {
      "epoch": 0.04276,
      "grad_norm": 0.6763844718810441,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 4276
    },
    {
      "epoch": 0.04277,
      "grad_norm": 0.9807377268843779,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 4277
    },
    {
      "epoch": 0.04278,
      "grad_norm": 1.0197483171024961,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 4278
    },
    {
      "epoch": 0.04279,
      "grad_norm": 0.9328879956310648,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4279
    },
    {
      "epoch": 0.0428,
      "grad_norm": 0.7779948668786022,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 4280
    },
    {
      "epoch": 0.04281,
      "grad_norm": 0.7431305171276716,
      "learning_rate": 0.003,
      "loss": 4.1851,
      "step": 4281
    },
    {
      "epoch": 0.04282,
      "grad_norm": 0.6283341114449361,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 4282
    },
    {
      "epoch": 0.04283,
      "grad_norm": 0.6394182134064753,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 4283
    },
    {
      "epoch": 0.04284,
      "grad_norm": 0.6758957287997666,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4284
    },
    {
      "epoch": 0.04285,
      "grad_norm": 0.6131823431058723,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 4285
    },
    {
      "epoch": 0.04286,
      "grad_norm": 0.6284313302797474,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 4286
    },
    {
      "epoch": 0.04287,
      "grad_norm": 0.6980874101445044,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 4287
    },
    {
      "epoch": 0.04288,
      "grad_norm": 0.696667601658298,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 4288
    },
    {
      "epoch": 0.04289,
      "grad_norm": 0.5891879790472746,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 4289
    },
    {
      "epoch": 0.0429,
      "grad_norm": 0.6679196724071268,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 4290
    },
    {
      "epoch": 0.04291,
      "grad_norm": 0.7132152030025544,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 4291
    },
    {
      "epoch": 0.04292,
      "grad_norm": 0.7831211413634629,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 4292
    },
    {
      "epoch": 0.04293,
      "grad_norm": 0.7420395140404202,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 4293
    },
    {
      "epoch": 0.04294,
      "grad_norm": 0.664819817807019,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 4294
    },
    {
      "epoch": 0.04295,
      "grad_norm": 0.6533174489648946,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 4295
    },
    {
      "epoch": 0.04296,
      "grad_norm": 0.6539485681842259,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 4296
    },
    {
      "epoch": 0.04297,
      "grad_norm": 0.5445244387293794,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 4297
    },
    {
      "epoch": 0.04298,
      "grad_norm": 0.509532772235553,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 4298
    },
    {
      "epoch": 0.04299,
      "grad_norm": 0.45809506613975526,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 4299
    },
    {
      "epoch": 0.043,
      "grad_norm": 0.4552028711707873,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 4300
    },
    {
      "epoch": 0.04301,
      "grad_norm": 0.5511852682855607,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 4301
    },
    {
      "epoch": 0.04302,
      "grad_norm": 0.7137895720159015,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 4302
    },
    {
      "epoch": 0.04303,
      "grad_norm": 0.968773809309932,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 4303
    },
    {
      "epoch": 0.04304,
      "grad_norm": 1.0008705380109228,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 4304
    },
    {
      "epoch": 0.04305,
      "grad_norm": 0.7364014697863502,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4305
    },
    {
      "epoch": 0.04306,
      "grad_norm": 0.8422192934651608,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 4306
    },
    {
      "epoch": 0.04307,
      "grad_norm": 0.8960004698755968,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 4307
    },
    {
      "epoch": 0.04308,
      "grad_norm": 0.8900888278943823,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 4308
    },
    {
      "epoch": 0.04309,
      "grad_norm": 0.7851447078463367,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 4309
    },
    {
      "epoch": 0.0431,
      "grad_norm": 0.8003446758283859,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 4310
    },
    {
      "epoch": 0.04311,
      "grad_norm": 0.9468061968202307,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 4311
    },
    {
      "epoch": 0.04312,
      "grad_norm": 0.8521671702866287,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 4312
    },
    {
      "epoch": 0.04313,
      "grad_norm": 0.7923961445552284,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 4313
    },
    {
      "epoch": 0.04314,
      "grad_norm": 0.739461542428857,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 4314
    },
    {
      "epoch": 0.04315,
      "grad_norm": 0.8509444047900186,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 4315
    },
    {
      "epoch": 0.04316,
      "grad_norm": 0.8601049555561022,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 4316
    },
    {
      "epoch": 0.04317,
      "grad_norm": 0.7623314110787873,
      "learning_rate": 0.003,
      "loss": 4.1871,
      "step": 4317
    },
    {
      "epoch": 0.04318,
      "grad_norm": 0.6587491558124051,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 4318
    },
    {
      "epoch": 0.04319,
      "grad_norm": 0.5849951298753531,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 4319
    },
    {
      "epoch": 0.0432,
      "grad_norm": 0.57401029054835,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 4320
    },
    {
      "epoch": 0.04321,
      "grad_norm": 0.6311486059735384,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 4321
    },
    {
      "epoch": 0.04322,
      "grad_norm": 0.6281833102961318,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 4322
    },
    {
      "epoch": 0.04323,
      "grad_norm": 0.7033374538455368,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4323
    },
    {
      "epoch": 0.04324,
      "grad_norm": 0.6437486136063232,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4324
    },
    {
      "epoch": 0.04325,
      "grad_norm": 0.6232679102675521,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 4325
    },
    {
      "epoch": 0.04326,
      "grad_norm": 0.6515618951702632,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 4326
    },
    {
      "epoch": 0.04327,
      "grad_norm": 0.6736681660679901,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 4327
    },
    {
      "epoch": 0.04328,
      "grad_norm": 0.719714363226295,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 4328
    },
    {
      "epoch": 0.04329,
      "grad_norm": 0.8360323655381567,
      "learning_rate": 0.003,
      "loss": 4.164,
      "step": 4329
    },
    {
      "epoch": 0.0433,
      "grad_norm": 0.8102704190603504,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4330
    },
    {
      "epoch": 0.04331,
      "grad_norm": 0.7636451395661018,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 4331
    },
    {
      "epoch": 0.04332,
      "grad_norm": 0.7789172902676315,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 4332
    },
    {
      "epoch": 0.04333,
      "grad_norm": 0.6881019378708992,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 4333
    },
    {
      "epoch": 0.04334,
      "grad_norm": 0.5741071284174022,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 4334
    },
    {
      "epoch": 0.04335,
      "grad_norm": 0.6016578373290047,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 4335
    },
    {
      "epoch": 0.04336,
      "grad_norm": 0.6309338575418205,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 4336
    },
    {
      "epoch": 0.04337,
      "grad_norm": 0.5725441440390032,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 4337
    },
    {
      "epoch": 0.04338,
      "grad_norm": 0.5493515413414659,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4338
    },
    {
      "epoch": 0.04339,
      "grad_norm": 0.5145808044001364,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 4339
    },
    {
      "epoch": 0.0434,
      "grad_norm": 0.5155260956336473,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 4340
    },
    {
      "epoch": 0.04341,
      "grad_norm": 0.49209801146871524,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4341
    },
    {
      "epoch": 0.04342,
      "grad_norm": 0.5776994696106369,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 4342
    },
    {
      "epoch": 0.04343,
      "grad_norm": 0.8551512196078731,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 4343
    },
    {
      "epoch": 0.04344,
      "grad_norm": 1.0783725488561569,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 4344
    },
    {
      "epoch": 0.04345,
      "grad_norm": 0.9655180256016967,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 4345
    },
    {
      "epoch": 0.04346,
      "grad_norm": 0.7582683532329971,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 4346
    },
    {
      "epoch": 0.04347,
      "grad_norm": 0.6941505792622187,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 4347
    },
    {
      "epoch": 0.04348,
      "grad_norm": 0.8405157089264003,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 4348
    },
    {
      "epoch": 0.04349,
      "grad_norm": 0.8523333890856811,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 4349
    },
    {
      "epoch": 0.0435,
      "grad_norm": 0.7396539502996479,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4350
    },
    {
      "epoch": 0.04351,
      "grad_norm": 0.6266986999030948,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4351
    },
    {
      "epoch": 0.04352,
      "grad_norm": 0.5282839844073566,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 4352
    },
    {
      "epoch": 0.04353,
      "grad_norm": 0.5005795044157166,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 4353
    },
    {
      "epoch": 0.04354,
      "grad_norm": 0.4851237319282445,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4354
    },
    {
      "epoch": 0.04355,
      "grad_norm": 0.5595788754843884,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 4355
    },
    {
      "epoch": 0.04356,
      "grad_norm": 0.5667796441106236,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 4356
    },
    {
      "epoch": 0.04357,
      "grad_norm": 0.5983610407473247,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 4357
    },
    {
      "epoch": 0.04358,
      "grad_norm": 0.6456602317217889,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 4358
    },
    {
      "epoch": 0.04359,
      "grad_norm": 0.6069596406446627,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 4359
    },
    {
      "epoch": 0.0436,
      "grad_norm": 0.5908374036987593,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 4360
    },
    {
      "epoch": 0.04361,
      "grad_norm": 0.6735288968429441,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 4361
    },
    {
      "epoch": 0.04362,
      "grad_norm": 0.7257053819990018,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 4362
    },
    {
      "epoch": 0.04363,
      "grad_norm": 0.6914693554965775,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 4363
    },
    {
      "epoch": 0.04364,
      "grad_norm": 0.7451580269204352,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4364
    },
    {
      "epoch": 0.04365,
      "grad_norm": 0.7858248696802116,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4365
    },
    {
      "epoch": 0.04366,
      "grad_norm": 0.663121974241136,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 4366
    },
    {
      "epoch": 0.04367,
      "grad_norm": 0.7175426458581545,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 4367
    },
    {
      "epoch": 0.04368,
      "grad_norm": 0.7351488980711208,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 4368
    },
    {
      "epoch": 0.04369,
      "grad_norm": 0.8536889689371039,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 4369
    },
    {
      "epoch": 0.0437,
      "grad_norm": 0.854111892194005,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 4370
    },
    {
      "epoch": 0.04371,
      "grad_norm": 1.046284154850743,
      "learning_rate": 0.003,
      "loss": 4.1575,
      "step": 4371
    },
    {
      "epoch": 0.04372,
      "grad_norm": 1.2262372791586333,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 4372
    },
    {
      "epoch": 0.04373,
      "grad_norm": 0.8968734983342949,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 4373
    },
    {
      "epoch": 0.04374,
      "grad_norm": 0.9753259228506409,
      "learning_rate": 0.003,
      "loss": 4.2155,
      "step": 4374
    },
    {
      "epoch": 0.04375,
      "grad_norm": 0.9920981237100678,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 4375
    },
    {
      "epoch": 0.04376,
      "grad_norm": 1.0294125384070931,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 4376
    },
    {
      "epoch": 0.04377,
      "grad_norm": 1.0068899366919466,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 4377
    },
    {
      "epoch": 0.04378,
      "grad_norm": 1.0341191509965633,
      "learning_rate": 0.003,
      "loss": 4.191,
      "step": 4378
    },
    {
      "epoch": 0.04379,
      "grad_norm": 0.9717889790676081,
      "learning_rate": 0.003,
      "loss": 4.2061,
      "step": 4379
    },
    {
      "epoch": 0.0438,
      "grad_norm": 1.0471882312294167,
      "learning_rate": 0.003,
      "loss": 4.1807,
      "step": 4380
    },
    {
      "epoch": 0.04381,
      "grad_norm": 0.8790592687859848,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 4381
    },
    {
      "epoch": 0.04382,
      "grad_norm": 0.8968294394751086,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 4382
    },
    {
      "epoch": 0.04383,
      "grad_norm": 0.8820678241101328,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 4383
    },
    {
      "epoch": 0.04384,
      "grad_norm": 1.0214298134194522,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 4384
    },
    {
      "epoch": 0.04385,
      "grad_norm": 0.900876137883741,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 4385
    },
    {
      "epoch": 0.04386,
      "grad_norm": 0.7692038806596491,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 4386
    },
    {
      "epoch": 0.04387,
      "grad_norm": 0.653702371460772,
      "learning_rate": 0.003,
      "loss": 4.1914,
      "step": 4387
    },
    {
      "epoch": 0.04388,
      "grad_norm": 0.6492229267746281,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4388
    },
    {
      "epoch": 0.04389,
      "grad_norm": 0.5786656683989158,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 4389
    },
    {
      "epoch": 0.0439,
      "grad_norm": 0.5012080835531559,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 4390
    },
    {
      "epoch": 0.04391,
      "grad_norm": 0.48775107179113014,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 4391
    },
    {
      "epoch": 0.04392,
      "grad_norm": 0.46374848591785856,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 4392
    },
    {
      "epoch": 0.04393,
      "grad_norm": 0.450356925230617,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 4393
    },
    {
      "epoch": 0.04394,
      "grad_norm": 0.4674972403264732,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 4394
    },
    {
      "epoch": 0.04395,
      "grad_norm": 0.5893115115334077,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 4395
    },
    {
      "epoch": 0.04396,
      "grad_norm": 0.7658769835507094,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 4396
    },
    {
      "epoch": 0.04397,
      "grad_norm": 0.9383339109696239,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 4397
    },
    {
      "epoch": 0.04398,
      "grad_norm": 0.8777679801246571,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 4398
    },
    {
      "epoch": 0.04399,
      "grad_norm": 0.7720983136183994,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4399
    },
    {
      "epoch": 0.044,
      "grad_norm": 0.7229759596282145,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4400
    },
    {
      "epoch": 0.04401,
      "grad_norm": 0.8501411177864325,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 4401
    },
    {
      "epoch": 0.04402,
      "grad_norm": 0.8319303801252986,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 4402
    },
    {
      "epoch": 0.04403,
      "grad_norm": 0.6805684530525856,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 4403
    },
    {
      "epoch": 0.04404,
      "grad_norm": 0.5579714343751891,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 4404
    },
    {
      "epoch": 0.04405,
      "grad_norm": 0.6272316247141385,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 4405
    },
    {
      "epoch": 0.04406,
      "grad_norm": 0.7189270603588521,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 4406
    },
    {
      "epoch": 0.04407,
      "grad_norm": 0.762459715751194,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 4407
    },
    {
      "epoch": 0.04408,
      "grad_norm": 0.7250097639288197,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 4408
    },
    {
      "epoch": 0.04409,
      "grad_norm": 0.7106703945924645,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4409
    },
    {
      "epoch": 0.0441,
      "grad_norm": 0.5937514209420578,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 4410
    },
    {
      "epoch": 0.04411,
      "grad_norm": 0.6265606257146901,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 4411
    },
    {
      "epoch": 0.04412,
      "grad_norm": 0.684509287657226,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 4412
    },
    {
      "epoch": 0.04413,
      "grad_norm": 0.7827609729511189,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 4413
    },
    {
      "epoch": 0.04414,
      "grad_norm": 0.8264979154885277,
      "learning_rate": 0.003,
      "loss": 4.1806,
      "step": 4414
    },
    {
      "epoch": 0.04415,
      "grad_norm": 0.7365639173006763,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 4415
    },
    {
      "epoch": 0.04416,
      "grad_norm": 0.5988319699730835,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 4416
    },
    {
      "epoch": 0.04417,
      "grad_norm": 0.6185388120705024,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 4417
    },
    {
      "epoch": 0.04418,
      "grad_norm": 0.7024277083040185,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 4418
    },
    {
      "epoch": 0.04419,
      "grad_norm": 0.6522881055559766,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 4419
    },
    {
      "epoch": 0.0442,
      "grad_norm": 0.6345542061505771,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 4420
    },
    {
      "epoch": 0.04421,
      "grad_norm": 0.6652846957230868,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 4421
    },
    {
      "epoch": 0.04422,
      "grad_norm": 0.7005501321538639,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 4422
    },
    {
      "epoch": 0.04423,
      "grad_norm": 0.7044884290461375,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 4423
    },
    {
      "epoch": 0.04424,
      "grad_norm": 0.6741401349701589,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 4424
    },
    {
      "epoch": 0.04425,
      "grad_norm": 0.6934002596368363,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 4425
    },
    {
      "epoch": 0.04426,
      "grad_norm": 0.6003050918470045,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 4426
    },
    {
      "epoch": 0.04427,
      "grad_norm": 0.6672810210726654,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 4427
    },
    {
      "epoch": 0.04428,
      "grad_norm": 0.6318153966455053,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 4428
    },
    {
      "epoch": 0.04429,
      "grad_norm": 0.623912292564508,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 4429
    },
    {
      "epoch": 0.0443,
      "grad_norm": 0.7110054552912546,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 4430
    },
    {
      "epoch": 0.04431,
      "grad_norm": 0.7246952184785321,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 4431
    },
    {
      "epoch": 0.04432,
      "grad_norm": 0.7298502791104271,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 4432
    },
    {
      "epoch": 0.04433,
      "grad_norm": 0.8250488623045142,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 4433
    },
    {
      "epoch": 0.04434,
      "grad_norm": 0.9624884861403035,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 4434
    },
    {
      "epoch": 0.04435,
      "grad_norm": 0.9889176292713632,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 4435
    },
    {
      "epoch": 0.04436,
      "grad_norm": 0.9593070985375517,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 4436
    },
    {
      "epoch": 0.04437,
      "grad_norm": 0.754019474989507,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 4437
    },
    {
      "epoch": 0.04438,
      "grad_norm": 0.6112092782747661,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 4438
    },
    {
      "epoch": 0.04439,
      "grad_norm": 0.8008555881950424,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 4439
    },
    {
      "epoch": 0.0444,
      "grad_norm": 0.8796518350119269,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 4440
    },
    {
      "epoch": 0.04441,
      "grad_norm": 0.881979649964416,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 4441
    },
    {
      "epoch": 0.04442,
      "grad_norm": 0.8372483887218065,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 4442
    },
    {
      "epoch": 0.04443,
      "grad_norm": 0.6759359878775629,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 4443
    },
    {
      "epoch": 0.04444,
      "grad_norm": 0.6554603066254844,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 4444
    },
    {
      "epoch": 0.04445,
      "grad_norm": 0.5717002727777339,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 4445
    },
    {
      "epoch": 0.04446,
      "grad_norm": 0.49898785075563756,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 4446
    },
    {
      "epoch": 0.04447,
      "grad_norm": 0.5124508174653755,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 4447
    },
    {
      "epoch": 0.04448,
      "grad_norm": 0.46026046835883233,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 4448
    },
    {
      "epoch": 0.04449,
      "grad_norm": 0.40191804438918904,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 4449
    },
    {
      "epoch": 0.0445,
      "grad_norm": 0.4056694001793259,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 4450
    },
    {
      "epoch": 0.04451,
      "grad_norm": 0.4140029705225631,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 4451
    },
    {
      "epoch": 0.04452,
      "grad_norm": 0.4961606990959985,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 4452
    },
    {
      "epoch": 0.04453,
      "grad_norm": 0.7225243712391776,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4453
    },
    {
      "epoch": 0.04454,
      "grad_norm": 0.9541612332110946,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 4454
    },
    {
      "epoch": 0.04455,
      "grad_norm": 1.004523212032084,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4455
    },
    {
      "epoch": 0.04456,
      "grad_norm": 0.7366078757936858,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 4456
    },
    {
      "epoch": 0.04457,
      "grad_norm": 0.7291092880614075,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 4457
    },
    {
      "epoch": 0.04458,
      "grad_norm": 0.8886507308195684,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 4458
    },
    {
      "epoch": 0.04459,
      "grad_norm": 1.023956211330971,
      "learning_rate": 0.003,
      "loss": 4.1783,
      "step": 4459
    },
    {
      "epoch": 0.0446,
      "grad_norm": 0.9783050212355968,
      "learning_rate": 0.003,
      "loss": 4.1765,
      "step": 4460
    },
    {
      "epoch": 0.04461,
      "grad_norm": 0.8910569073793239,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 4461
    },
    {
      "epoch": 0.04462,
      "grad_norm": 1.0114250150322548,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 4462
    },
    {
      "epoch": 0.04463,
      "grad_norm": 1.109635261633442,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 4463
    },
    {
      "epoch": 0.04464,
      "grad_norm": 0.9892179336121286,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 4464
    },
    {
      "epoch": 0.04465,
      "grad_norm": 0.9496628250922594,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 4465
    },
    {
      "epoch": 0.04466,
      "grad_norm": 0.9698557150392482,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 4466
    },
    {
      "epoch": 0.04467,
      "grad_norm": 0.9285468685707471,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 4467
    },
    {
      "epoch": 0.04468,
      "grad_norm": 0.9397451172133555,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4468
    },
    {
      "epoch": 0.04469,
      "grad_norm": 0.9565209269681922,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 4469
    },
    {
      "epoch": 0.0447,
      "grad_norm": 0.8769720683928874,
      "learning_rate": 0.003,
      "loss": 4.1844,
      "step": 4470
    },
    {
      "epoch": 0.04471,
      "grad_norm": 0.795338760730945,
      "learning_rate": 0.003,
      "loss": 4.2008,
      "step": 4471
    },
    {
      "epoch": 0.04472,
      "grad_norm": 0.7843283232673075,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 4472
    },
    {
      "epoch": 0.04473,
      "grad_norm": 0.8909484017306056,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 4473
    },
    {
      "epoch": 0.04474,
      "grad_norm": 0.7784392006819544,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 4474
    },
    {
      "epoch": 0.04475,
      "grad_norm": 0.7042700663520809,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 4475
    },
    {
      "epoch": 0.04476,
      "grad_norm": 0.6389974316195782,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 4476
    },
    {
      "epoch": 0.04477,
      "grad_norm": 0.6071760811715112,
      "learning_rate": 0.003,
      "loss": 4.1616,
      "step": 4477
    },
    {
      "epoch": 0.04478,
      "grad_norm": 0.5219307858429795,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 4478
    },
    {
      "epoch": 0.04479,
      "grad_norm": 0.52342819477722,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 4479
    },
    {
      "epoch": 0.0448,
      "grad_norm": 0.5268154790194748,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4480
    },
    {
      "epoch": 0.04481,
      "grad_norm": 0.5399282796391196,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 4481
    },
    {
      "epoch": 0.04482,
      "grad_norm": 0.6507902172065512,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 4482
    },
    {
      "epoch": 0.04483,
      "grad_norm": 0.867713110982925,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 4483
    },
    {
      "epoch": 0.04484,
      "grad_norm": 0.9801856447775387,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 4484
    },
    {
      "epoch": 0.04485,
      "grad_norm": 0.8358981195735171,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 4485
    },
    {
      "epoch": 0.04486,
      "grad_norm": 0.726373516850574,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 4486
    },
    {
      "epoch": 0.04487,
      "grad_norm": 0.7123593368535847,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4487
    },
    {
      "epoch": 0.04488,
      "grad_norm": 0.726522221164894,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 4488
    },
    {
      "epoch": 0.04489,
      "grad_norm": 0.7095321623068442,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 4489
    },
    {
      "epoch": 0.0449,
      "grad_norm": 0.6252082433565278,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 4490
    },
    {
      "epoch": 0.04491,
      "grad_norm": 0.6640250192773183,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 4491
    },
    {
      "epoch": 0.04492,
      "grad_norm": 0.6961055886888501,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 4492
    },
    {
      "epoch": 0.04493,
      "grad_norm": 0.8458020064467313,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4493
    },
    {
      "epoch": 0.04494,
      "grad_norm": 0.7474034453269607,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 4494
    },
    {
      "epoch": 0.04495,
      "grad_norm": 0.6451994546687867,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 4495
    },
    {
      "epoch": 0.04496,
      "grad_norm": 0.540011005371107,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 4496
    },
    {
      "epoch": 0.04497,
      "grad_norm": 0.4950562236330396,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 4497
    },
    {
      "epoch": 0.04498,
      "grad_norm": 0.4943734925550196,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 4498
    },
    {
      "epoch": 0.04499,
      "grad_norm": 0.47393717260525386,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 4499
    },
    {
      "epoch": 0.045,
      "grad_norm": 0.41885472582226696,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 4500
    },
    {
      "epoch": 0.04501,
      "grad_norm": 0.4834752022098693,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 4501
    },
    {
      "epoch": 0.04502,
      "grad_norm": 0.45694364065485266,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 4502
    },
    {
      "epoch": 0.04503,
      "grad_norm": 0.48865003766237863,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 4503
    },
    {
      "epoch": 0.04504,
      "grad_norm": 0.5438297922402094,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 4504
    },
    {
      "epoch": 0.04505,
      "grad_norm": 0.5871203728299376,
      "learning_rate": 0.003,
      "loss": 4.154,
      "step": 4505
    },
    {
      "epoch": 0.04506,
      "grad_norm": 0.5917294880351803,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 4506
    },
    {
      "epoch": 0.04507,
      "grad_norm": 0.7183532404676503,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 4507
    },
    {
      "epoch": 0.04508,
      "grad_norm": 0.8049591048914756,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 4508
    },
    {
      "epoch": 0.04509,
      "grad_norm": 0.9156955919586133,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 4509
    },
    {
      "epoch": 0.0451,
      "grad_norm": 0.8889056537954558,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4510
    },
    {
      "epoch": 0.04511,
      "grad_norm": 0.693800418390815,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4511
    },
    {
      "epoch": 0.04512,
      "grad_norm": 0.7135831824345485,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 4512
    },
    {
      "epoch": 0.04513,
      "grad_norm": 0.6688562196054199,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 4513
    },
    {
      "epoch": 0.04514,
      "grad_norm": 0.6423991582569938,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4514
    },
    {
      "epoch": 0.04515,
      "grad_norm": 0.7711378950211699,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4515
    },
    {
      "epoch": 0.04516,
      "grad_norm": 0.8939759348625197,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 4516
    },
    {
      "epoch": 0.04517,
      "grad_norm": 0.7606381004553531,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 4517
    },
    {
      "epoch": 0.04518,
      "grad_norm": 0.6218485906194273,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 4518
    },
    {
      "epoch": 0.04519,
      "grad_norm": 0.7964516800865484,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 4519
    },
    {
      "epoch": 0.0452,
      "grad_norm": 0.8973981028823347,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 4520
    },
    {
      "epoch": 0.04521,
      "grad_norm": 0.807736716714071,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4521
    },
    {
      "epoch": 0.04522,
      "grad_norm": 0.8117844738071615,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4522
    },
    {
      "epoch": 0.04523,
      "grad_norm": 0.7829719745475495,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 4523
    },
    {
      "epoch": 0.04524,
      "grad_norm": 0.7479885268590871,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 4524
    },
    {
      "epoch": 0.04525,
      "grad_norm": 0.7243373752912751,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 4525
    },
    {
      "epoch": 0.04526,
      "grad_norm": 0.7384870219800298,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 4526
    },
    {
      "epoch": 0.04527,
      "grad_norm": 0.8786326776352553,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 4527
    },
    {
      "epoch": 0.04528,
      "grad_norm": 0.9239987067801139,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 4528
    },
    {
      "epoch": 0.04529,
      "grad_norm": 0.8389910760544824,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 4529
    },
    {
      "epoch": 0.0453,
      "grad_norm": 0.748168700266448,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 4530
    },
    {
      "epoch": 0.04531,
      "grad_norm": 0.7202788367913475,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 4531
    },
    {
      "epoch": 0.04532,
      "grad_norm": 0.6571475637254836,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4532
    },
    {
      "epoch": 0.04533,
      "grad_norm": 0.7185379271005126,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 4533
    },
    {
      "epoch": 0.04534,
      "grad_norm": 0.6309360237693905,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 4534
    },
    {
      "epoch": 0.04535,
      "grad_norm": 0.6779837587183748,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 4535
    },
    {
      "epoch": 0.04536,
      "grad_norm": 0.6922814218735922,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 4536
    },
    {
      "epoch": 0.04537,
      "grad_norm": 0.6707461416618695,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 4537
    },
    {
      "epoch": 0.04538,
      "grad_norm": 0.6533648131571776,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4538
    },
    {
      "epoch": 0.04539,
      "grad_norm": 0.7226394008940196,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 4539
    },
    {
      "epoch": 0.0454,
      "grad_norm": 0.7809591559130085,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 4540
    },
    {
      "epoch": 0.04541,
      "grad_norm": 0.8598716052287259,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 4541
    },
    {
      "epoch": 0.04542,
      "grad_norm": 0.7887863556115788,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 4542
    },
    {
      "epoch": 0.04543,
      "grad_norm": 0.5682149072319508,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 4543
    },
    {
      "epoch": 0.04544,
      "grad_norm": 0.5505936637479015,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 4544
    },
    {
      "epoch": 0.04545,
      "grad_norm": 0.6999578034297054,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4545
    },
    {
      "epoch": 0.04546,
      "grad_norm": 0.6684727088847572,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 4546
    },
    {
      "epoch": 0.04547,
      "grad_norm": 0.6131696984235054,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 4547
    },
    {
      "epoch": 0.04548,
      "grad_norm": 0.6233401797280433,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 4548
    },
    {
      "epoch": 0.04549,
      "grad_norm": 0.7411565175033444,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 4549
    },
    {
      "epoch": 0.0455,
      "grad_norm": 0.7508874744904848,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 4550
    },
    {
      "epoch": 0.04551,
      "grad_norm": 0.7534956856948848,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 4551
    },
    {
      "epoch": 0.04552,
      "grad_norm": 0.689574501986947,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 4552
    },
    {
      "epoch": 0.04553,
      "grad_norm": 0.6536678327481642,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4553
    },
    {
      "epoch": 0.04554,
      "grad_norm": 0.6521633276844322,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 4554
    },
    {
      "epoch": 0.04555,
      "grad_norm": 0.7438153849355749,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 4555
    },
    {
      "epoch": 0.04556,
      "grad_norm": 0.813199349395972,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 4556
    },
    {
      "epoch": 0.04557,
      "grad_norm": 0.8670448776825898,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 4557
    },
    {
      "epoch": 0.04558,
      "grad_norm": 0.9173442370434104,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 4558
    },
    {
      "epoch": 0.04559,
      "grad_norm": 0.9293838780402448,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 4559
    },
    {
      "epoch": 0.0456,
      "grad_norm": 0.9721860045634062,
      "learning_rate": 0.003,
      "loss": 4.142,
      "step": 4560
    },
    {
      "epoch": 0.04561,
      "grad_norm": 0.9130374577828355,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 4561
    },
    {
      "epoch": 0.04562,
      "grad_norm": 0.8204531300346364,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 4562
    },
    {
      "epoch": 0.04563,
      "grad_norm": 0.941182990714371,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 4563
    },
    {
      "epoch": 0.04564,
      "grad_norm": 0.9950132484736411,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 4564
    },
    {
      "epoch": 0.04565,
      "grad_norm": 0.8713983224408589,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 4565
    },
    {
      "epoch": 0.04566,
      "grad_norm": 0.7742242108390238,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 4566
    },
    {
      "epoch": 0.04567,
      "grad_norm": 0.7732353202308767,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 4567
    },
    {
      "epoch": 0.04568,
      "grad_norm": 0.7949276600918046,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 4568
    },
    {
      "epoch": 0.04569,
      "grad_norm": 0.76216115917809,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4569
    },
    {
      "epoch": 0.0457,
      "grad_norm": 0.665803751777818,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 4570
    },
    {
      "epoch": 0.04571,
      "grad_norm": 0.61898217206735,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 4571
    },
    {
      "epoch": 0.04572,
      "grad_norm": 0.6630102965712841,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 4572
    },
    {
      "epoch": 0.04573,
      "grad_norm": 0.6545388007004492,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 4573
    },
    {
      "epoch": 0.04574,
      "grad_norm": 0.6434199338973368,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4574
    },
    {
      "epoch": 0.04575,
      "grad_norm": 0.6084613496329172,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4575
    },
    {
      "epoch": 0.04576,
      "grad_norm": 0.5926985992517696,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 4576
    },
    {
      "epoch": 0.04577,
      "grad_norm": 0.7068872849239232,
      "learning_rate": 0.003,
      "loss": 4.161,
      "step": 4577
    },
    {
      "epoch": 0.04578,
      "grad_norm": 0.7815668152529673,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 4578
    },
    {
      "epoch": 0.04579,
      "grad_norm": 0.705130762977042,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 4579
    },
    {
      "epoch": 0.0458,
      "grad_norm": 0.706838854182526,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 4580
    },
    {
      "epoch": 0.04581,
      "grad_norm": 0.7719932048413563,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 4581
    },
    {
      "epoch": 0.04582,
      "grad_norm": 0.8196871294195923,
      "learning_rate": 0.003,
      "loss": 4.1644,
      "step": 4582
    },
    {
      "epoch": 0.04583,
      "grad_norm": 0.9747741235878243,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 4583
    },
    {
      "epoch": 0.04584,
      "grad_norm": 1.0614405255220143,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 4584
    },
    {
      "epoch": 0.04585,
      "grad_norm": 0.8692609981568404,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 4585
    },
    {
      "epoch": 0.04586,
      "grad_norm": 0.8017360688038584,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 4586
    },
    {
      "epoch": 0.04587,
      "grad_norm": 0.737110938579538,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 4587
    },
    {
      "epoch": 0.04588,
      "grad_norm": 0.6638294236901521,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 4588
    },
    {
      "epoch": 0.04589,
      "grad_norm": 0.6411690157984203,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 4589
    },
    {
      "epoch": 0.0459,
      "grad_norm": 0.6088730568376585,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 4590
    },
    {
      "epoch": 0.04591,
      "grad_norm": 0.6362826036737852,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 4591
    },
    {
      "epoch": 0.04592,
      "grad_norm": 0.6389570140066912,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 4592
    },
    {
      "epoch": 0.04593,
      "grad_norm": 0.7097656474131322,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 4593
    },
    {
      "epoch": 0.04594,
      "grad_norm": 0.6175568409115185,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 4594
    },
    {
      "epoch": 0.04595,
      "grad_norm": 0.620699543326453,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 4595
    },
    {
      "epoch": 0.04596,
      "grad_norm": 0.5962342876114253,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 4596
    },
    {
      "epoch": 0.04597,
      "grad_norm": 0.6324593221701938,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 4597
    },
    {
      "epoch": 0.04598,
      "grad_norm": 0.6611397602971393,
      "learning_rate": 0.003,
      "loss": 4.1806,
      "step": 4598
    },
    {
      "epoch": 0.04599,
      "grad_norm": 0.6234361489034259,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 4599
    },
    {
      "epoch": 0.046,
      "grad_norm": 0.6075516815675354,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4600
    },
    {
      "epoch": 0.04601,
      "grad_norm": 0.6042104496947458,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 4601
    },
    {
      "epoch": 0.04602,
      "grad_norm": 0.6164999421939368,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 4602
    },
    {
      "epoch": 0.04603,
      "grad_norm": 0.6490308990358886,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4603
    },
    {
      "epoch": 0.04604,
      "grad_norm": 0.7661434840060077,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 4604
    },
    {
      "epoch": 0.04605,
      "grad_norm": 1.0569916498623362,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 4605
    },
    {
      "epoch": 0.04606,
      "grad_norm": 1.027715504643133,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 4606
    },
    {
      "epoch": 0.04607,
      "grad_norm": 0.8342746434973768,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 4607
    },
    {
      "epoch": 0.04608,
      "grad_norm": 0.7850330384795643,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 4608
    },
    {
      "epoch": 0.04609,
      "grad_norm": 0.7443196821550939,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 4609
    },
    {
      "epoch": 0.0461,
      "grad_norm": 0.7785237638327334,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 4610
    },
    {
      "epoch": 0.04611,
      "grad_norm": 0.8838166156331966,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 4611
    },
    {
      "epoch": 0.04612,
      "grad_norm": 0.8874166663110993,
      "learning_rate": 0.003,
      "loss": 4.1625,
      "step": 4612
    },
    {
      "epoch": 0.04613,
      "grad_norm": 0.8148264887171576,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 4613
    },
    {
      "epoch": 0.04614,
      "grad_norm": 0.7820861967247282,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 4614
    },
    {
      "epoch": 0.04615,
      "grad_norm": 0.7280484691047371,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4615
    },
    {
      "epoch": 0.04616,
      "grad_norm": 0.7490023925524317,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 4616
    },
    {
      "epoch": 0.04617,
      "grad_norm": 0.7518558085096564,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 4617
    },
    {
      "epoch": 0.04618,
      "grad_norm": 0.625986793466225,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4618
    },
    {
      "epoch": 0.04619,
      "grad_norm": 0.6472629361360867,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 4619
    },
    {
      "epoch": 0.0462,
      "grad_norm": 0.6557801977850327,
      "learning_rate": 0.003,
      "loss": 4.1625,
      "step": 4620
    },
    {
      "epoch": 0.04621,
      "grad_norm": 0.73229721178544,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 4621
    },
    {
      "epoch": 0.04622,
      "grad_norm": 0.9280201932819085,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 4622
    },
    {
      "epoch": 0.04623,
      "grad_norm": 1.042833156417488,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 4623
    },
    {
      "epoch": 0.04624,
      "grad_norm": 0.8256463072634586,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 4624
    },
    {
      "epoch": 0.04625,
      "grad_norm": 0.7149715088227584,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4625
    },
    {
      "epoch": 0.04626,
      "grad_norm": 0.6877823576969472,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 4626
    },
    {
      "epoch": 0.04627,
      "grad_norm": 0.6742469710012651,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 4627
    },
    {
      "epoch": 0.04628,
      "grad_norm": 0.62532013220358,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4628
    },
    {
      "epoch": 0.04629,
      "grad_norm": 0.5130605726615276,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 4629
    },
    {
      "epoch": 0.0463,
      "grad_norm": 0.5654280379865769,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 4630
    },
    {
      "epoch": 0.04631,
      "grad_norm": 0.6014781265202847,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 4631
    },
    {
      "epoch": 0.04632,
      "grad_norm": 0.6307983761305929,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 4632
    },
    {
      "epoch": 0.04633,
      "grad_norm": 0.6189113146917462,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 4633
    },
    {
      "epoch": 0.04634,
      "grad_norm": 0.6421436906797112,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 4634
    },
    {
      "epoch": 0.04635,
      "grad_norm": 0.5590583262211648,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 4635
    },
    {
      "epoch": 0.04636,
      "grad_norm": 0.503140076293387,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 4636
    },
    {
      "epoch": 0.04637,
      "grad_norm": 0.5299879986072056,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 4637
    },
    {
      "epoch": 0.04638,
      "grad_norm": 0.4214634549110216,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 4638
    },
    {
      "epoch": 0.04639,
      "grad_norm": 0.397393509216712,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 4639
    },
    {
      "epoch": 0.0464,
      "grad_norm": 0.43447214142466917,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4640
    },
    {
      "epoch": 0.04641,
      "grad_norm": 0.5062252032170508,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 4641
    },
    {
      "epoch": 0.04642,
      "grad_norm": 0.5711310545610656,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 4642
    },
    {
      "epoch": 0.04643,
      "grad_norm": 0.6248454986616232,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 4643
    },
    {
      "epoch": 0.04644,
      "grad_norm": 0.7210053036878883,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 4644
    },
    {
      "epoch": 0.04645,
      "grad_norm": 0.90249716060301,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 4645
    },
    {
      "epoch": 0.04646,
      "grad_norm": 1.2409050533897181,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 4646
    },
    {
      "epoch": 0.04647,
      "grad_norm": 0.8393901607525605,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 4647
    },
    {
      "epoch": 0.04648,
      "grad_norm": 0.7746500580810028,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 4648
    },
    {
      "epoch": 0.04649,
      "grad_norm": 0.8739807831426791,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 4649
    },
    {
      "epoch": 0.0465,
      "grad_norm": 0.9522286444793494,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 4650
    },
    {
      "epoch": 0.04651,
      "grad_norm": 1.1296690683804316,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 4651
    },
    {
      "epoch": 0.04652,
      "grad_norm": 0.9265569184990113,
      "learning_rate": 0.003,
      "loss": 4.1575,
      "step": 4652
    },
    {
      "epoch": 0.04653,
      "grad_norm": 0.900473448775536,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 4653
    },
    {
      "epoch": 0.04654,
      "grad_norm": 1.0602801619166438,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 4654
    },
    {
      "epoch": 0.04655,
      "grad_norm": 0.9413459563751835,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 4655
    },
    {
      "epoch": 0.04656,
      "grad_norm": 0.7433440151400471,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 4656
    },
    {
      "epoch": 0.04657,
      "grad_norm": 0.7855838722575367,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 4657
    },
    {
      "epoch": 0.04658,
      "grad_norm": 0.7026074720403149,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 4658
    },
    {
      "epoch": 0.04659,
      "grad_norm": 0.7054872170953301,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 4659
    },
    {
      "epoch": 0.0466,
      "grad_norm": 0.7253225489512056,
      "learning_rate": 0.003,
      "loss": 4.162,
      "step": 4660
    },
    {
      "epoch": 0.04661,
      "grad_norm": 0.6938113299882246,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 4661
    },
    {
      "epoch": 0.04662,
      "grad_norm": 0.6661030601372132,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 4662
    },
    {
      "epoch": 0.04663,
      "grad_norm": 0.7495766554410687,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 4663
    },
    {
      "epoch": 0.04664,
      "grad_norm": 0.7803399105366602,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 4664
    },
    {
      "epoch": 0.04665,
      "grad_norm": 0.7715228893902641,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 4665
    },
    {
      "epoch": 0.04666,
      "grad_norm": 0.8189022468897561,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 4666
    },
    {
      "epoch": 0.04667,
      "grad_norm": 0.9073148500434436,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 4667
    },
    {
      "epoch": 0.04668,
      "grad_norm": 0.8663369305932774,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 4668
    },
    {
      "epoch": 0.04669,
      "grad_norm": 0.7981760342295242,
      "learning_rate": 0.003,
      "loss": 4.1642,
      "step": 4669
    },
    {
      "epoch": 0.0467,
      "grad_norm": 0.7399100042776996,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 4670
    },
    {
      "epoch": 0.04671,
      "grad_norm": 0.79400390511152,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 4671
    },
    {
      "epoch": 0.04672,
      "grad_norm": 0.7886276399898601,
      "learning_rate": 0.003,
      "loss": 4.1526,
      "step": 4672
    },
    {
      "epoch": 0.04673,
      "grad_norm": 0.7587981117840258,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 4673
    },
    {
      "epoch": 0.04674,
      "grad_norm": 0.6833849503950475,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 4674
    },
    {
      "epoch": 0.04675,
      "grad_norm": 0.5521830886006064,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 4675
    },
    {
      "epoch": 0.04676,
      "grad_norm": 0.5551519809403406,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 4676
    },
    {
      "epoch": 0.04677,
      "grad_norm": 0.4956612676914781,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4677
    },
    {
      "epoch": 0.04678,
      "grad_norm": 0.5248385967176291,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 4678
    },
    {
      "epoch": 0.04679,
      "grad_norm": 0.5169108616408382,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 4679
    },
    {
      "epoch": 0.0468,
      "grad_norm": 0.7044973029382834,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 4680
    },
    {
      "epoch": 0.04681,
      "grad_norm": 0.845934209130146,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 4681
    },
    {
      "epoch": 0.04682,
      "grad_norm": 0.8277130051517226,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 4682
    },
    {
      "epoch": 0.04683,
      "grad_norm": 0.7576471069505033,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 4683
    },
    {
      "epoch": 0.04684,
      "grad_norm": 0.6931368300408345,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4684
    },
    {
      "epoch": 0.04685,
      "grad_norm": 0.6709891251557847,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 4685
    },
    {
      "epoch": 0.04686,
      "grad_norm": 0.5581781513115139,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 4686
    },
    {
      "epoch": 0.04687,
      "grad_norm": 0.6261887980349435,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 4687
    },
    {
      "epoch": 0.04688,
      "grad_norm": 0.7985962391701531,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 4688
    },
    {
      "epoch": 0.04689,
      "grad_norm": 1.0550803093277015,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 4689
    },
    {
      "epoch": 0.0469,
      "grad_norm": 0.9618900206251452,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 4690
    },
    {
      "epoch": 0.04691,
      "grad_norm": 0.6523829450195608,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 4691
    },
    {
      "epoch": 0.04692,
      "grad_norm": 0.6896621979107347,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 4692
    },
    {
      "epoch": 0.04693,
      "grad_norm": 0.9401038472008131,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 4693
    },
    {
      "epoch": 0.04694,
      "grad_norm": 0.9720475262714349,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 4694
    },
    {
      "epoch": 0.04695,
      "grad_norm": 0.9719251158591558,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 4695
    },
    {
      "epoch": 0.04696,
      "grad_norm": 0.8427853123728982,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 4696
    },
    {
      "epoch": 0.04697,
      "grad_norm": 0.8499034656864161,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 4697
    },
    {
      "epoch": 0.04698,
      "grad_norm": 0.7021228867021835,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 4698
    },
    {
      "epoch": 0.04699,
      "grad_norm": 0.7398294832914439,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 4699
    },
    {
      "epoch": 0.047,
      "grad_norm": 0.6942643738589905,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 4700
    },
    {
      "epoch": 0.04701,
      "grad_norm": 0.6683525277935326,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 4701
    },
    {
      "epoch": 0.04702,
      "grad_norm": 0.5837305852992254,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 4702
    },
    {
      "epoch": 0.04703,
      "grad_norm": 0.5004181371776402,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 4703
    },
    {
      "epoch": 0.04704,
      "grad_norm": 0.500112229497055,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 4704
    },
    {
      "epoch": 0.04705,
      "grad_norm": 0.46870504628905124,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 4705
    },
    {
      "epoch": 0.04706,
      "grad_norm": 0.5084124708467841,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 4706
    },
    {
      "epoch": 0.04707,
      "grad_norm": 0.5592830079070097,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 4707
    },
    {
      "epoch": 0.04708,
      "grad_norm": 0.6734818421768685,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 4708
    },
    {
      "epoch": 0.04709,
      "grad_norm": 0.7820705785181057,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 4709
    },
    {
      "epoch": 0.0471,
      "grad_norm": 0.8173042330246184,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 4710
    },
    {
      "epoch": 0.04711,
      "grad_norm": 0.7323489999342726,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 4711
    },
    {
      "epoch": 0.04712,
      "grad_norm": 0.6858752455645303,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 4712
    },
    {
      "epoch": 0.04713,
      "grad_norm": 0.5541474886267923,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 4713
    },
    {
      "epoch": 0.04714,
      "grad_norm": 0.5810454942772322,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 4714
    },
    {
      "epoch": 0.04715,
      "grad_norm": 0.6021618978261117,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 4715
    },
    {
      "epoch": 0.04716,
      "grad_norm": 0.6343272321037665,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 4716
    },
    {
      "epoch": 0.04717,
      "grad_norm": 0.7013052653860089,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4717
    },
    {
      "epoch": 0.04718,
      "grad_norm": 0.66349485254104,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 4718
    },
    {
      "epoch": 0.04719,
      "grad_norm": 0.8713918643231059,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 4719
    },
    {
      "epoch": 0.0472,
      "grad_norm": 1.0008190365871503,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 4720
    },
    {
      "epoch": 0.04721,
      "grad_norm": 0.9716425512049767,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 4721
    },
    {
      "epoch": 0.04722,
      "grad_norm": 1.1120781580345849,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 4722
    },
    {
      "epoch": 0.04723,
      "grad_norm": 0.8161047263055594,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 4723
    },
    {
      "epoch": 0.04724,
      "grad_norm": 0.8833523326572259,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 4724
    },
    {
      "epoch": 0.04725,
      "grad_norm": 0.8898592287073411,
      "learning_rate": 0.003,
      "loss": 4.1569,
      "step": 4725
    },
    {
      "epoch": 0.04726,
      "grad_norm": 0.9859327457193534,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4726
    },
    {
      "epoch": 0.04727,
      "grad_norm": 0.9165866060948171,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 4727
    },
    {
      "epoch": 0.04728,
      "grad_norm": 0.9068684412672875,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 4728
    },
    {
      "epoch": 0.04729,
      "grad_norm": 0.8318177657045663,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 4729
    },
    {
      "epoch": 0.0473,
      "grad_norm": 0.8027800490874897,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 4730
    },
    {
      "epoch": 0.04731,
      "grad_norm": 0.7978896274361092,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4731
    },
    {
      "epoch": 0.04732,
      "grad_norm": 0.8802600387002112,
      "learning_rate": 0.003,
      "loss": 4.164,
      "step": 4732
    },
    {
      "epoch": 0.04733,
      "grad_norm": 0.8234843622598483,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 4733
    },
    {
      "epoch": 0.04734,
      "grad_norm": 0.6599234947722757,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 4734
    },
    {
      "epoch": 0.04735,
      "grad_norm": 0.8208592352781166,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 4735
    },
    {
      "epoch": 0.04736,
      "grad_norm": 1.1729409563655009,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 4736
    },
    {
      "epoch": 0.04737,
      "grad_norm": 0.9659463338122776,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 4737
    },
    {
      "epoch": 0.04738,
      "grad_norm": 0.7651058866068673,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 4738
    },
    {
      "epoch": 0.04739,
      "grad_norm": 0.717869067195337,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 4739
    },
    {
      "epoch": 0.0474,
      "grad_norm": 0.6851887051422195,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 4740
    },
    {
      "epoch": 0.04741,
      "grad_norm": 0.6717713840804643,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 4741
    },
    {
      "epoch": 0.04742,
      "grad_norm": 0.6107738091433065,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 4742
    },
    {
      "epoch": 0.04743,
      "grad_norm": 0.5441883526059924,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 4743
    },
    {
      "epoch": 0.04744,
      "grad_norm": 0.5198168694209138,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 4744
    },
    {
      "epoch": 0.04745,
      "grad_norm": 0.43524115400010815,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 4745
    },
    {
      "epoch": 0.04746,
      "grad_norm": 0.48653308583746263,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 4746
    },
    {
      "epoch": 0.04747,
      "grad_norm": 0.5339120307603674,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 4747
    },
    {
      "epoch": 0.04748,
      "grad_norm": 0.5479071965461091,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 4748
    },
    {
      "epoch": 0.04749,
      "grad_norm": 0.5417055693516375,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 4749
    },
    {
      "epoch": 0.0475,
      "grad_norm": 0.5280951958997921,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 4750
    },
    {
      "epoch": 0.04751,
      "grad_norm": 0.5350329833617672,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 4751
    },
    {
      "epoch": 0.04752,
      "grad_norm": 0.57851596805048,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 4752
    },
    {
      "epoch": 0.04753,
      "grad_norm": 0.72902319198213,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 4753
    },
    {
      "epoch": 0.04754,
      "grad_norm": 0.871184049188565,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 4754
    },
    {
      "epoch": 0.04755,
      "grad_norm": 0.9103037208186765,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 4755
    },
    {
      "epoch": 0.04756,
      "grad_norm": 0.859550535967343,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 4756
    },
    {
      "epoch": 0.04757,
      "grad_norm": 0.7466205077193813,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4757
    },
    {
      "epoch": 0.04758,
      "grad_norm": 0.7507245421944043,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 4758
    },
    {
      "epoch": 0.04759,
      "grad_norm": 0.8294869880303999,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 4759
    },
    {
      "epoch": 0.0476,
      "grad_norm": 0.8251883621259142,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 4760
    },
    {
      "epoch": 0.04761,
      "grad_norm": 0.7765732752545424,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 4761
    },
    {
      "epoch": 0.04762,
      "grad_norm": 0.6280062378308506,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 4762
    },
    {
      "epoch": 0.04763,
      "grad_norm": 0.5686058703500096,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 4763
    },
    {
      "epoch": 0.04764,
      "grad_norm": 0.46584482494791357,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 4764
    },
    {
      "epoch": 0.04765,
      "grad_norm": 0.45316919798397814,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 4765
    },
    {
      "epoch": 0.04766,
      "grad_norm": 0.5554760863440187,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 4766
    },
    {
      "epoch": 0.04767,
      "grad_norm": 0.6418170833104156,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 4767
    },
    {
      "epoch": 0.04768,
      "grad_norm": 0.5645471402502725,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 4768
    },
    {
      "epoch": 0.04769,
      "grad_norm": 0.5927099803110258,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 4769
    },
    {
      "epoch": 0.0477,
      "grad_norm": 0.6595637919437056,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 4770
    },
    {
      "epoch": 0.04771,
      "grad_norm": 0.5520312176680346,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 4771
    },
    {
      "epoch": 0.04772,
      "grad_norm": 0.5616366446629109,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 4772
    },
    {
      "epoch": 0.04773,
      "grad_norm": 0.7069869424163182,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4773
    },
    {
      "epoch": 0.04774,
      "grad_norm": 0.8943049067297537,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4774
    },
    {
      "epoch": 0.04775,
      "grad_norm": 1.0379517887129035,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 4775
    },
    {
      "epoch": 0.04776,
      "grad_norm": 1.0601938367245354,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 4776
    },
    {
      "epoch": 0.04777,
      "grad_norm": 0.8123830997039955,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 4777
    },
    {
      "epoch": 0.04778,
      "grad_norm": 0.6466306116567022,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 4778
    },
    {
      "epoch": 0.04779,
      "grad_norm": 0.8059965327611258,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4779
    },
    {
      "epoch": 0.0478,
      "grad_norm": 1.005661350204286,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 4780
    },
    {
      "epoch": 0.04781,
      "grad_norm": 0.9783065445314736,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 4781
    },
    {
      "epoch": 0.04782,
      "grad_norm": 0.9777333193008619,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 4782
    },
    {
      "epoch": 0.04783,
      "grad_norm": 1.0138916385667516,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 4783
    },
    {
      "epoch": 0.04784,
      "grad_norm": 1.0210591580782744,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 4784
    },
    {
      "epoch": 0.04785,
      "grad_norm": 1.0867592349380752,
      "learning_rate": 0.003,
      "loss": 4.1783,
      "step": 4785
    },
    {
      "epoch": 0.04786,
      "grad_norm": 0.9270426683995183,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 4786
    },
    {
      "epoch": 0.04787,
      "grad_norm": 0.9581278828591302,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 4787
    },
    {
      "epoch": 0.04788,
      "grad_norm": 1.0463519085350252,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 4788
    },
    {
      "epoch": 0.04789,
      "grad_norm": 1.048457678252623,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 4789
    },
    {
      "epoch": 0.0479,
      "grad_norm": 1.0802030566717422,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 4790
    },
    {
      "epoch": 0.04791,
      "grad_norm": 0.9188967498168278,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 4791
    },
    {
      "epoch": 0.04792,
      "grad_norm": 1.0152593647469241,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 4792
    },
    {
      "epoch": 0.04793,
      "grad_norm": 1.018584486769021,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 4793
    },
    {
      "epoch": 0.04794,
      "grad_norm": 0.8402721654087,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 4794
    },
    {
      "epoch": 0.04795,
      "grad_norm": 0.7643727384922673,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 4795
    },
    {
      "epoch": 0.04796,
      "grad_norm": 0.813774811366339,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 4796
    },
    {
      "epoch": 0.04797,
      "grad_norm": 0.6494127001881433,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4797
    },
    {
      "epoch": 0.04798,
      "grad_norm": 0.664089910476502,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 4798
    },
    {
      "epoch": 0.04799,
      "grad_norm": 0.6031144588619443,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 4799
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.5598439245281802,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4800
    },
    {
      "epoch": 0.04801,
      "grad_norm": 0.6477000775289928,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 4801
    },
    {
      "epoch": 0.04802,
      "grad_norm": 0.706840356865147,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 4802
    },
    {
      "epoch": 0.04803,
      "grad_norm": 0.687443212408336,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 4803
    },
    {
      "epoch": 0.04804,
      "grad_norm": 0.6817235142027526,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 4804
    },
    {
      "epoch": 0.04805,
      "grad_norm": 0.7014528353258211,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 4805
    },
    {
      "epoch": 0.04806,
      "grad_norm": 0.6063087225759103,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 4806
    },
    {
      "epoch": 0.04807,
      "grad_norm": 0.4870524793552668,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 4807
    },
    {
      "epoch": 0.04808,
      "grad_norm": 0.4304737736314177,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 4808
    },
    {
      "epoch": 0.04809,
      "grad_norm": 0.444058010799397,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 4809
    },
    {
      "epoch": 0.0481,
      "grad_norm": 0.44159249278167406,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4810
    },
    {
      "epoch": 0.04811,
      "grad_norm": 0.46942470854293217,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 4811
    },
    {
      "epoch": 0.04812,
      "grad_norm": 0.4477933921171643,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 4812
    },
    {
      "epoch": 0.04813,
      "grad_norm": 0.43354928602746334,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 4813
    },
    {
      "epoch": 0.04814,
      "grad_norm": 0.45498153005997544,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 4814
    },
    {
      "epoch": 0.04815,
      "grad_norm": 0.48463241046460065,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 4815
    },
    {
      "epoch": 0.04816,
      "grad_norm": 0.5311818047677171,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 4816
    },
    {
      "epoch": 0.04817,
      "grad_norm": 0.631506579388757,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 4817
    },
    {
      "epoch": 0.04818,
      "grad_norm": 0.8170443890052518,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 4818
    },
    {
      "epoch": 0.04819,
      "grad_norm": 1.172132734592305,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 4819
    },
    {
      "epoch": 0.0482,
      "grad_norm": 0.838017513510012,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 4820
    },
    {
      "epoch": 0.04821,
      "grad_norm": 0.6975770749182754,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4821
    },
    {
      "epoch": 0.04822,
      "grad_norm": 0.7527718267712026,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 4822
    },
    {
      "epoch": 0.04823,
      "grad_norm": 0.7414326559856069,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 4823
    },
    {
      "epoch": 0.04824,
      "grad_norm": 0.858775362232865,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 4824
    },
    {
      "epoch": 0.04825,
      "grad_norm": 0.695583690590602,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 4825
    },
    {
      "epoch": 0.04826,
      "grad_norm": 0.5781760330827292,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 4826
    },
    {
      "epoch": 0.04827,
      "grad_norm": 0.5806936123154903,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 4827
    },
    {
      "epoch": 0.04828,
      "grad_norm": 0.7721046134763737,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 4828
    },
    {
      "epoch": 0.04829,
      "grad_norm": 0.7885233902013562,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 4829
    },
    {
      "epoch": 0.0483,
      "grad_norm": 0.7091409575187497,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 4830
    },
    {
      "epoch": 0.04831,
      "grad_norm": 0.7110846809520083,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 4831
    },
    {
      "epoch": 0.04832,
      "grad_norm": 0.680243507107158,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4832
    },
    {
      "epoch": 0.04833,
      "grad_norm": 0.7284136950991514,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 4833
    },
    {
      "epoch": 0.04834,
      "grad_norm": 0.7137894246322048,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 4834
    },
    {
      "epoch": 0.04835,
      "grad_norm": 0.636140630835025,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 4835
    },
    {
      "epoch": 0.04836,
      "grad_norm": 0.75652748414691,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 4836
    },
    {
      "epoch": 0.04837,
      "grad_norm": 0.785548659376736,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4837
    },
    {
      "epoch": 0.04838,
      "grad_norm": 0.8897930937393894,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 4838
    },
    {
      "epoch": 0.04839,
      "grad_norm": 0.9660346509912735,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 4839
    },
    {
      "epoch": 0.0484,
      "grad_norm": 0.9300911895678211,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4840
    },
    {
      "epoch": 0.04841,
      "grad_norm": 0.7818946985346545,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4841
    },
    {
      "epoch": 0.04842,
      "grad_norm": 0.7032209957018698,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 4842
    },
    {
      "epoch": 0.04843,
      "grad_norm": 0.6722771974917844,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 4843
    },
    {
      "epoch": 0.04844,
      "grad_norm": 0.6970990299983332,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4844
    },
    {
      "epoch": 0.04845,
      "grad_norm": 0.6367868288526063,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 4845
    },
    {
      "epoch": 0.04846,
      "grad_norm": 0.6829062431633811,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 4846
    },
    {
      "epoch": 0.04847,
      "grad_norm": 0.6460341296632274,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 4847
    },
    {
      "epoch": 0.04848,
      "grad_norm": 0.6713362647372938,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 4848
    },
    {
      "epoch": 0.04849,
      "grad_norm": 0.6714682036160667,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4849
    },
    {
      "epoch": 0.0485,
      "grad_norm": 0.7057113852906085,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 4850
    },
    {
      "epoch": 0.04851,
      "grad_norm": 0.6646770579142436,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 4851
    },
    {
      "epoch": 0.04852,
      "grad_norm": 0.6577931410879579,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 4852
    },
    {
      "epoch": 0.04853,
      "grad_norm": 0.6521737566217662,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4853
    },
    {
      "epoch": 0.04854,
      "grad_norm": 0.7638656389712092,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 4854
    },
    {
      "epoch": 0.04855,
      "grad_norm": 0.9457310244447173,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4855
    },
    {
      "epoch": 0.04856,
      "grad_norm": 1.1517585285172265,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 4856
    },
    {
      "epoch": 0.04857,
      "grad_norm": 0.9746651389714764,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 4857
    },
    {
      "epoch": 0.04858,
      "grad_norm": 0.8324446739497157,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 4858
    },
    {
      "epoch": 0.04859,
      "grad_norm": 0.7506712881500571,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 4859
    },
    {
      "epoch": 0.0486,
      "grad_norm": 0.7957902928578074,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 4860
    },
    {
      "epoch": 0.04861,
      "grad_norm": 1.0534393033014007,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 4861
    },
    {
      "epoch": 0.04862,
      "grad_norm": 1.072043293213455,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 4862
    },
    {
      "epoch": 0.04863,
      "grad_norm": 0.8542078395141963,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 4863
    },
    {
      "epoch": 0.04864,
      "grad_norm": 0.979149461429483,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 4864
    },
    {
      "epoch": 0.04865,
      "grad_norm": 0.9889727360280101,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 4865
    },
    {
      "epoch": 0.04866,
      "grad_norm": 0.87668738837822,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 4866
    },
    {
      "epoch": 0.04867,
      "grad_norm": 0.9480918411172308,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 4867
    },
    {
      "epoch": 0.04868,
      "grad_norm": 0.8900699953728738,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 4868
    },
    {
      "epoch": 0.04869,
      "grad_norm": 0.8410323236753658,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 4869
    },
    {
      "epoch": 0.0487,
      "grad_norm": 0.8052365609623746,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 4870
    },
    {
      "epoch": 0.04871,
      "grad_norm": 0.732052294025559,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 4871
    },
    {
      "epoch": 0.04872,
      "grad_norm": 0.6093794224074204,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 4872
    },
    {
      "epoch": 0.04873,
      "grad_norm": 0.5550033492784485,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 4873
    },
    {
      "epoch": 0.04874,
      "grad_norm": 0.6407687802372767,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 4874
    },
    {
      "epoch": 0.04875,
      "grad_norm": 0.6445083276518792,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 4875
    },
    {
      "epoch": 0.04876,
      "grad_norm": 0.598683403431915,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 4876
    },
    {
      "epoch": 0.04877,
      "grad_norm": 0.5879420193521339,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 4877
    },
    {
      "epoch": 0.04878,
      "grad_norm": 0.6015687919635526,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 4878
    },
    {
      "epoch": 0.04879,
      "grad_norm": 0.6406016359581286,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 4879
    },
    {
      "epoch": 0.0488,
      "grad_norm": 0.6520274839816634,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 4880
    },
    {
      "epoch": 0.04881,
      "grad_norm": 0.6746577474599795,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 4881
    },
    {
      "epoch": 0.04882,
      "grad_norm": 0.6988276300361237,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 4882
    },
    {
      "epoch": 0.04883,
      "grad_norm": 0.637601167920116,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 4883
    },
    {
      "epoch": 0.04884,
      "grad_norm": 0.6575805356479,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 4884
    },
    {
      "epoch": 0.04885,
      "grad_norm": 0.8560411189615229,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 4885
    },
    {
      "epoch": 0.04886,
      "grad_norm": 1.131590855439064,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 4886
    },
    {
      "epoch": 0.04887,
      "grad_norm": 0.8261960657958597,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 4887
    },
    {
      "epoch": 0.04888,
      "grad_norm": 0.5889636338082994,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4888
    },
    {
      "epoch": 0.04889,
      "grad_norm": 0.6875568675189105,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 4889
    },
    {
      "epoch": 0.0489,
      "grad_norm": 0.8178506331950943,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4890
    },
    {
      "epoch": 0.04891,
      "grad_norm": 0.8524321612267298,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 4891
    },
    {
      "epoch": 0.04892,
      "grad_norm": 0.6402952785912811,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 4892
    },
    {
      "epoch": 0.04893,
      "grad_norm": 0.49571623322484065,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 4893
    },
    {
      "epoch": 0.04894,
      "grad_norm": 0.6005303000219224,
      "learning_rate": 0.003,
      "loss": 4.1485,
      "step": 4894
    },
    {
      "epoch": 0.04895,
      "grad_norm": 0.7874188186491505,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 4895
    },
    {
      "epoch": 0.04896,
      "grad_norm": 0.8796160513302255,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 4896
    },
    {
      "epoch": 0.04897,
      "grad_norm": 0.7789626736982245,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 4897
    },
    {
      "epoch": 0.04898,
      "grad_norm": 0.7938347458338417,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 4898
    },
    {
      "epoch": 0.04899,
      "grad_norm": 0.6968501707866385,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 4899
    },
    {
      "epoch": 0.049,
      "grad_norm": 0.6737103261211691,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 4900
    },
    {
      "epoch": 0.04901,
      "grad_norm": 0.6928180762743227,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 4901
    },
    {
      "epoch": 0.04902,
      "grad_norm": 0.7387421279656656,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 4902
    },
    {
      "epoch": 0.04903,
      "grad_norm": 0.7365881444221591,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 4903
    },
    {
      "epoch": 0.04904,
      "grad_norm": 0.6809368406351045,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 4904
    },
    {
      "epoch": 0.04905,
      "grad_norm": 0.8596440024174532,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 4905
    },
    {
      "epoch": 0.04906,
      "grad_norm": 1.1779416676139556,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4906
    },
    {
      "epoch": 0.04907,
      "grad_norm": 0.9153726514069693,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 4907
    },
    {
      "epoch": 0.04908,
      "grad_norm": 0.6705432105024506,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4908
    },
    {
      "epoch": 0.04909,
      "grad_norm": 0.7177688849716874,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4909
    },
    {
      "epoch": 0.0491,
      "grad_norm": 0.8659923553333108,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 4910
    },
    {
      "epoch": 0.04911,
      "grad_norm": 1.0871210771334645,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 4911
    },
    {
      "epoch": 0.04912,
      "grad_norm": 0.825170311776902,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 4912
    },
    {
      "epoch": 0.04913,
      "grad_norm": 0.6669180449834547,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 4913
    },
    {
      "epoch": 0.04914,
      "grad_norm": 0.6589662640213191,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 4914
    },
    {
      "epoch": 0.04915,
      "grad_norm": 0.6453034668867473,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 4915
    },
    {
      "epoch": 0.04916,
      "grad_norm": 0.6379061303258432,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 4916
    },
    {
      "epoch": 0.04917,
      "grad_norm": 0.5924781741927655,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 4917
    },
    {
      "epoch": 0.04918,
      "grad_norm": 0.5392017954625788,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 4918
    },
    {
      "epoch": 0.04919,
      "grad_norm": 0.4926263926107212,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 4919
    },
    {
      "epoch": 0.0492,
      "grad_norm": 0.5657039568533931,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4920
    },
    {
      "epoch": 0.04921,
      "grad_norm": 0.6408113470012788,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 4921
    },
    {
      "epoch": 0.04922,
      "grad_norm": 0.7276187339344597,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 4922
    },
    {
      "epoch": 0.04923,
      "grad_norm": 0.7782906389614147,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 4923
    },
    {
      "epoch": 0.04924,
      "grad_norm": 0.8311620891758528,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 4924
    },
    {
      "epoch": 0.04925,
      "grad_norm": 0.9301177021878505,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 4925
    },
    {
      "epoch": 0.04926,
      "grad_norm": 0.8645596717575721,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 4926
    },
    {
      "epoch": 0.04927,
      "grad_norm": 0.7213043835497918,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 4927
    },
    {
      "epoch": 0.04928,
      "grad_norm": 0.6797037072765041,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 4928
    },
    {
      "epoch": 0.04929,
      "grad_norm": 0.5833040699084097,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 4929
    },
    {
      "epoch": 0.0493,
      "grad_norm": 0.7502027197799604,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 4930
    },
    {
      "epoch": 0.04931,
      "grad_norm": 0.7729294183641608,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 4931
    },
    {
      "epoch": 0.04932,
      "grad_norm": 0.7115365383624536,
      "learning_rate": 0.003,
      "loss": 4.1526,
      "step": 4932
    },
    {
      "epoch": 0.04933,
      "grad_norm": 0.619935413571119,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 4933
    },
    {
      "epoch": 0.04934,
      "grad_norm": 0.6328194752766774,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 4934
    },
    {
      "epoch": 0.04935,
      "grad_norm": 0.666067060827277,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 4935
    },
    {
      "epoch": 0.04936,
      "grad_norm": 0.7455790570370421,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 4936
    },
    {
      "epoch": 0.04937,
      "grad_norm": 0.8690467307626608,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 4937
    },
    {
      "epoch": 0.04938,
      "grad_norm": 0.9382478740758116,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4938
    },
    {
      "epoch": 0.04939,
      "grad_norm": 0.9269163238429858,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4939
    },
    {
      "epoch": 0.0494,
      "grad_norm": 0.7535488235404929,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4940
    },
    {
      "epoch": 0.04941,
      "grad_norm": 0.6231540125860376,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 4941
    },
    {
      "epoch": 0.04942,
      "grad_norm": 0.5994421619470709,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 4942
    },
    {
      "epoch": 0.04943,
      "grad_norm": 0.6316381391994007,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4943
    },
    {
      "epoch": 0.04944,
      "grad_norm": 0.6942848124244505,
      "learning_rate": 0.003,
      "loss": 4.1526,
      "step": 4944
    },
    {
      "epoch": 0.04945,
      "grad_norm": 0.6745431924540877,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 4945
    },
    {
      "epoch": 0.04946,
      "grad_norm": 0.7148234111999289,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 4946
    },
    {
      "epoch": 0.04947,
      "grad_norm": 0.8151644855471658,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 4947
    },
    {
      "epoch": 0.04948,
      "grad_norm": 0.901565456553654,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 4948
    },
    {
      "epoch": 0.04949,
      "grad_norm": 0.8388445391988596,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4949
    },
    {
      "epoch": 0.0495,
      "grad_norm": 0.769498516722048,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 4950
    },
    {
      "epoch": 0.04951,
      "grad_norm": 0.9271659871328424,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 4951
    },
    {
      "epoch": 0.04952,
      "grad_norm": 0.9430760383533365,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 4952
    },
    {
      "epoch": 0.04953,
      "grad_norm": 0.8839519050512392,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 4953
    },
    {
      "epoch": 0.04954,
      "grad_norm": 0.8691782058986859,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 4954
    },
    {
      "epoch": 0.04955,
      "grad_norm": 0.7804007935901659,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4955
    },
    {
      "epoch": 0.04956,
      "grad_norm": 0.7340937957959927,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 4956
    },
    {
      "epoch": 0.04957,
      "grad_norm": 0.7502559806635023,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 4957
    },
    {
      "epoch": 0.04958,
      "grad_norm": 0.7469357331628064,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 4958
    },
    {
      "epoch": 0.04959,
      "grad_norm": 0.7227020566857775,
      "learning_rate": 0.003,
      "loss": 4.1451,
      "step": 4959
    },
    {
      "epoch": 0.0496,
      "grad_norm": 0.6072369078325202,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 4960
    },
    {
      "epoch": 0.04961,
      "grad_norm": 0.676652458930749,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 4961
    },
    {
      "epoch": 0.04962,
      "grad_norm": 0.7166853924943506,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 4962
    },
    {
      "epoch": 0.04963,
      "grad_norm": 0.8282425384023169,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 4963
    },
    {
      "epoch": 0.04964,
      "grad_norm": 0.9085445932185267,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 4964
    },
    {
      "epoch": 0.04965,
      "grad_norm": 0.9373085689145864,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 4965
    },
    {
      "epoch": 0.04966,
      "grad_norm": 0.6879723306931469,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 4966
    },
    {
      "epoch": 0.04967,
      "grad_norm": 0.5346987676889187,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 4967
    },
    {
      "epoch": 0.04968,
      "grad_norm": 0.5973814985358372,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 4968
    },
    {
      "epoch": 0.04969,
      "grad_norm": 0.8137665544420138,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 4969
    },
    {
      "epoch": 0.0497,
      "grad_norm": 0.9213595657452577,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 4970
    },
    {
      "epoch": 0.04971,
      "grad_norm": 0.8081884217188025,
      "learning_rate": 0.003,
      "loss": 4.1548,
      "step": 4971
    },
    {
      "epoch": 0.04972,
      "grad_norm": 0.696216159144708,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 4972
    },
    {
      "epoch": 0.04973,
      "grad_norm": 0.8733113357334519,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 4973
    },
    {
      "epoch": 0.04974,
      "grad_norm": 1.0046232999118463,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 4974
    },
    {
      "epoch": 0.04975,
      "grad_norm": 0.7903522031468557,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4975
    },
    {
      "epoch": 0.04976,
      "grad_norm": 0.6802567984502358,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 4976
    },
    {
      "epoch": 0.04977,
      "grad_norm": 0.6880693634478566,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 4977
    },
    {
      "epoch": 0.04978,
      "grad_norm": 0.8224558060329078,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 4978
    },
    {
      "epoch": 0.04979,
      "grad_norm": 0.693860233816605,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 4979
    },
    {
      "epoch": 0.0498,
      "grad_norm": 0.5722513075301342,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 4980
    },
    {
      "epoch": 0.04981,
      "grad_norm": 0.5155541843719716,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 4981
    },
    {
      "epoch": 0.04982,
      "grad_norm": 0.5293279825762658,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 4982
    },
    {
      "epoch": 0.04983,
      "grad_norm": 0.5031810438056484,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 4983
    },
    {
      "epoch": 0.04984,
      "grad_norm": 0.50373719850947,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 4984
    },
    {
      "epoch": 0.04985,
      "grad_norm": 0.4503251156136766,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 4985
    },
    {
      "epoch": 0.04986,
      "grad_norm": 0.45313125147575284,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 4986
    },
    {
      "epoch": 0.04987,
      "grad_norm": 0.3893398619956913,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 4987
    },
    {
      "epoch": 0.04988,
      "grad_norm": 0.38877981714111726,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 4988
    },
    {
      "epoch": 0.04989,
      "grad_norm": 0.4389205190435005,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 4989
    },
    {
      "epoch": 0.0499,
      "grad_norm": 0.629712543859391,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 4990
    },
    {
      "epoch": 0.04991,
      "grad_norm": 0.8411643038205673,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 4991
    },
    {
      "epoch": 0.04992,
      "grad_norm": 1.0789241755274537,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 4992
    },
    {
      "epoch": 0.04993,
      "grad_norm": 0.894691990453197,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 4993
    },
    {
      "epoch": 0.04994,
      "grad_norm": 0.768409182457612,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 4994
    },
    {
      "epoch": 0.04995,
      "grad_norm": 0.7502833338777625,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 4995
    },
    {
      "epoch": 0.04996,
      "grad_norm": 0.7438036098652065,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 4996
    },
    {
      "epoch": 0.04997,
      "grad_norm": 0.7542393015064909,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 4997
    },
    {
      "epoch": 0.04998,
      "grad_norm": 0.7987106834768938,
      "learning_rate": 0.003,
      "loss": 4.1614,
      "step": 4998
    },
    {
      "epoch": 0.04999,
      "grad_norm": 0.7978624560845015,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 4999
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9161273312112043,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 5000
    },
    {
      "epoch": 0.05001,
      "grad_norm": 0.996412080112922,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 5001
    },
    {
      "epoch": 0.05002,
      "grad_norm": 0.9597914662378922,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 5002
    },
    {
      "epoch": 0.05003,
      "grad_norm": 0.8551476884140587,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 5003
    },
    {
      "epoch": 0.05004,
      "grad_norm": 0.8613602446247762,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 5004
    },
    {
      "epoch": 0.05005,
      "grad_norm": 0.9135522439159383,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 5005
    },
    {
      "epoch": 0.05006,
      "grad_norm": 1.0256214526238998,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5006
    },
    {
      "epoch": 0.05007,
      "grad_norm": 0.9189841175452944,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 5007
    },
    {
      "epoch": 0.05008,
      "grad_norm": 0.8069809344089516,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 5008
    },
    {
      "epoch": 0.05009,
      "grad_norm": 0.7363173254372745,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 5009
    },
    {
      "epoch": 0.0501,
      "grad_norm": 0.780930048247047,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 5010
    },
    {
      "epoch": 0.05011,
      "grad_norm": 0.8533571714235313,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 5011
    },
    {
      "epoch": 0.05012,
      "grad_norm": 0.7333946370690233,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 5012
    },
    {
      "epoch": 0.05013,
      "grad_norm": 0.7554255781670896,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 5013
    },
    {
      "epoch": 0.05014,
      "grad_norm": 0.8053934909287488,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 5014
    },
    {
      "epoch": 0.05015,
      "grad_norm": 0.6989364159311318,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 5015
    },
    {
      "epoch": 0.05016,
      "grad_norm": 0.6523590967522017,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 5016
    },
    {
      "epoch": 0.05017,
      "grad_norm": 0.6801442129276165,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 5017
    },
    {
      "epoch": 0.05018,
      "grad_norm": 0.7921166720657037,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 5018
    },
    {
      "epoch": 0.05019,
      "grad_norm": 0.9072386747671375,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 5019
    },
    {
      "epoch": 0.0502,
      "grad_norm": 1.01701838314669,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 5020
    },
    {
      "epoch": 0.05021,
      "grad_norm": 0.9399609124786235,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 5021
    },
    {
      "epoch": 0.05022,
      "grad_norm": 0.9060890151344648,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 5022
    },
    {
      "epoch": 0.05023,
      "grad_norm": 0.833401199404608,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 5023
    },
    {
      "epoch": 0.05024,
      "grad_norm": 0.6936980469006718,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 5024
    },
    {
      "epoch": 0.05025,
      "grad_norm": 0.6485227724397681,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 5025
    },
    {
      "epoch": 0.05026,
      "grad_norm": 0.6504679913105271,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 5026
    },
    {
      "epoch": 0.05027,
      "grad_norm": 0.6627173144511802,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 5027
    },
    {
      "epoch": 0.05028,
      "grad_norm": 0.6951447785935732,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 5028
    },
    {
      "epoch": 0.05029,
      "grad_norm": 0.6952175839048826,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 5029
    },
    {
      "epoch": 0.0503,
      "grad_norm": 0.5688496607028097,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 5030
    },
    {
      "epoch": 0.05031,
      "grad_norm": 0.6035161846572389,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 5031
    },
    {
      "epoch": 0.05032,
      "grad_norm": 0.670286035574246,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 5032
    },
    {
      "epoch": 0.05033,
      "grad_norm": 0.6467306224656293,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 5033
    },
    {
      "epoch": 0.05034,
      "grad_norm": 0.6063260055449672,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 5034
    },
    {
      "epoch": 0.05035,
      "grad_norm": 0.629592938048323,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 5035
    },
    {
      "epoch": 0.05036,
      "grad_norm": 0.6478567980439086,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 5036
    },
    {
      "epoch": 0.05037,
      "grad_norm": 0.7505964391613944,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 5037
    },
    {
      "epoch": 0.05038,
      "grad_norm": 0.8528209687755143,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 5038
    },
    {
      "epoch": 0.05039,
      "grad_norm": 1.0629743487440606,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 5039
    },
    {
      "epoch": 0.0504,
      "grad_norm": 1.1351742762146553,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 5040
    },
    {
      "epoch": 0.05041,
      "grad_norm": 0.7653122098370709,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 5041
    },
    {
      "epoch": 0.05042,
      "grad_norm": 0.7775733262103008,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 5042
    },
    {
      "epoch": 0.05043,
      "grad_norm": 0.7251140829300723,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 5043
    },
    {
      "epoch": 0.05044,
      "grad_norm": 0.7675590430218364,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 5044
    },
    {
      "epoch": 0.05045,
      "grad_norm": 0.796129916502261,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 5045
    },
    {
      "epoch": 0.05046,
      "grad_norm": 0.7730541689349488,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 5046
    },
    {
      "epoch": 0.05047,
      "grad_norm": 0.721813094114717,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 5047
    },
    {
      "epoch": 0.05048,
      "grad_norm": 0.6421472913380964,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 5048
    },
    {
      "epoch": 0.05049,
      "grad_norm": 0.6622380588723976,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 5049
    },
    {
      "epoch": 0.0505,
      "grad_norm": 0.5608629241012306,
      "learning_rate": 0.003,
      "loss": 4.183,
      "step": 5050
    },
    {
      "epoch": 0.05051,
      "grad_norm": 0.5745585056656611,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 5051
    },
    {
      "epoch": 0.05052,
      "grad_norm": 0.6379797741525732,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 5052
    },
    {
      "epoch": 0.05053,
      "grad_norm": 0.6833197049521939,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 5053
    },
    {
      "epoch": 0.05054,
      "grad_norm": 0.7634490766042622,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 5054
    },
    {
      "epoch": 0.05055,
      "grad_norm": 0.8837284078408866,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 5055
    },
    {
      "epoch": 0.05056,
      "grad_norm": 0.986372668426662,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 5056
    },
    {
      "epoch": 0.05057,
      "grad_norm": 1.1002958520848571,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 5057
    },
    {
      "epoch": 0.05058,
      "grad_norm": 0.8620407065983835,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 5058
    },
    {
      "epoch": 0.05059,
      "grad_norm": 0.7776023795769754,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 5059
    },
    {
      "epoch": 0.0506,
      "grad_norm": 0.8000191647790404,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 5060
    },
    {
      "epoch": 0.05061,
      "grad_norm": 0.8859446865423543,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 5061
    },
    {
      "epoch": 0.05062,
      "grad_norm": 0.8482279418144665,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 5062
    },
    {
      "epoch": 0.05063,
      "grad_norm": 0.8922032578947977,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 5063
    },
    {
      "epoch": 0.05064,
      "grad_norm": 0.8411761082071082,
      "learning_rate": 0.003,
      "loss": 4.1479,
      "step": 5064
    },
    {
      "epoch": 0.05065,
      "grad_norm": 0.8548177713084265,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 5065
    },
    {
      "epoch": 0.05066,
      "grad_norm": 0.8283313824637577,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 5066
    },
    {
      "epoch": 0.05067,
      "grad_norm": 0.8012249471762791,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 5067
    },
    {
      "epoch": 0.05068,
      "grad_norm": 0.7695096288578107,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 5068
    },
    {
      "epoch": 0.05069,
      "grad_norm": 0.9045793357271396,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 5069
    },
    {
      "epoch": 0.0507,
      "grad_norm": 0.864648963362809,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 5070
    },
    {
      "epoch": 0.05071,
      "grad_norm": 0.8168223538297239,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 5071
    },
    {
      "epoch": 0.05072,
      "grad_norm": 0.7496813210802163,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 5072
    },
    {
      "epoch": 0.05073,
      "grad_norm": 0.758480840663007,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 5073
    },
    {
      "epoch": 0.05074,
      "grad_norm": 0.696634586083982,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 5074
    },
    {
      "epoch": 0.05075,
      "grad_norm": 0.7650541290537565,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 5075
    },
    {
      "epoch": 0.05076,
      "grad_norm": 0.8193433029203343,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 5076
    },
    {
      "epoch": 0.05077,
      "grad_norm": 0.8494467870191629,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 5077
    },
    {
      "epoch": 0.05078,
      "grad_norm": 0.746324730026238,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 5078
    },
    {
      "epoch": 0.05079,
      "grad_norm": 0.7788241307468048,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 5079
    },
    {
      "epoch": 0.0508,
      "grad_norm": 0.6225546418109031,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 5080
    },
    {
      "epoch": 0.05081,
      "grad_norm": 0.5890146987434735,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 5081
    },
    {
      "epoch": 0.05082,
      "grad_norm": 0.7095924514939523,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 5082
    },
    {
      "epoch": 0.05083,
      "grad_norm": 0.8654893728046584,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 5083
    },
    {
      "epoch": 0.05084,
      "grad_norm": 0.9274741044534023,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 5084
    },
    {
      "epoch": 0.05085,
      "grad_norm": 0.863892065363985,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 5085
    },
    {
      "epoch": 0.05086,
      "grad_norm": 0.7754486240701118,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 5086
    },
    {
      "epoch": 0.05087,
      "grad_norm": 0.6467115070840203,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 5087
    },
    {
      "epoch": 0.05088,
      "grad_norm": 0.6210040639615062,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 5088
    },
    {
      "epoch": 0.05089,
      "grad_norm": 0.6024673934619533,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 5089
    },
    {
      "epoch": 0.0509,
      "grad_norm": 0.6180444267761035,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 5090
    },
    {
      "epoch": 0.05091,
      "grad_norm": 0.6761606632023955,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 5091
    },
    {
      "epoch": 0.05092,
      "grad_norm": 0.7124865882076785,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 5092
    },
    {
      "epoch": 0.05093,
      "grad_norm": 0.6974866206345641,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 5093
    },
    {
      "epoch": 0.05094,
      "grad_norm": 0.6044579073369937,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 5094
    },
    {
      "epoch": 0.05095,
      "grad_norm": 0.6081394607788756,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 5095
    },
    {
      "epoch": 0.05096,
      "grad_norm": 0.529589673235904,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 5096
    },
    {
      "epoch": 0.05097,
      "grad_norm": 0.5564475574602485,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 5097
    },
    {
      "epoch": 0.05098,
      "grad_norm": 0.6408190676261977,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 5098
    },
    {
      "epoch": 0.05099,
      "grad_norm": 0.6723429647324713,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 5099
    },
    {
      "epoch": 0.051,
      "grad_norm": 0.8300243600705887,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 5100
    },
    {
      "epoch": 0.05101,
      "grad_norm": 1.1968215120059336,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 5101
    },
    {
      "epoch": 0.05102,
      "grad_norm": 0.9163396896342757,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 5102
    },
    {
      "epoch": 0.05103,
      "grad_norm": 0.6393734271585462,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 5103
    },
    {
      "epoch": 0.05104,
      "grad_norm": 0.6360132483972007,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 5104
    },
    {
      "epoch": 0.05105,
      "grad_norm": 0.7716876792159864,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 5105
    },
    {
      "epoch": 0.05106,
      "grad_norm": 0.8361810564307343,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 5106
    },
    {
      "epoch": 0.05107,
      "grad_norm": 0.8838910237762364,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 5107
    },
    {
      "epoch": 0.05108,
      "grad_norm": 0.7935879173083066,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 5108
    },
    {
      "epoch": 0.05109,
      "grad_norm": 0.8192023142285442,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 5109
    },
    {
      "epoch": 0.0511,
      "grad_norm": 0.843755803433391,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5110
    },
    {
      "epoch": 0.05111,
      "grad_norm": 0.7470385931001929,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 5111
    },
    {
      "epoch": 0.05112,
      "grad_norm": 0.6896006523029916,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 5112
    },
    {
      "epoch": 0.05113,
      "grad_norm": 0.7291351113373098,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 5113
    },
    {
      "epoch": 0.05114,
      "grad_norm": 0.6627016065501992,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 5114
    },
    {
      "epoch": 0.05115,
      "grad_norm": 0.6079619066030912,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 5115
    },
    {
      "epoch": 0.05116,
      "grad_norm": 0.6260970910935404,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 5116
    },
    {
      "epoch": 0.05117,
      "grad_norm": 0.6494564538670361,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 5117
    },
    {
      "epoch": 0.05118,
      "grad_norm": 0.6314756962030549,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 5118
    },
    {
      "epoch": 0.05119,
      "grad_norm": 0.621412047113099,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 5119
    },
    {
      "epoch": 0.0512,
      "grad_norm": 0.6805129862001599,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 5120
    },
    {
      "epoch": 0.05121,
      "grad_norm": 0.6678436168391524,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 5121
    },
    {
      "epoch": 0.05122,
      "grad_norm": 0.6883310547366032,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 5122
    },
    {
      "epoch": 0.05123,
      "grad_norm": 0.6801681468514705,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 5123
    },
    {
      "epoch": 0.05124,
      "grad_norm": 0.6407786390586026,
      "learning_rate": 0.003,
      "loss": 4.154,
      "step": 5124
    },
    {
      "epoch": 0.05125,
      "grad_norm": 0.616883516629288,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 5125
    },
    {
      "epoch": 0.05126,
      "grad_norm": 0.7827204877464654,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 5126
    },
    {
      "epoch": 0.05127,
      "grad_norm": 0.8837164038749966,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 5127
    },
    {
      "epoch": 0.05128,
      "grad_norm": 0.8294720258543833,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 5128
    },
    {
      "epoch": 0.05129,
      "grad_norm": 0.8659985985101629,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 5129
    },
    {
      "epoch": 0.0513,
      "grad_norm": 0.9120789657636011,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 5130
    },
    {
      "epoch": 0.05131,
      "grad_norm": 1.071986933837498,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 5131
    },
    {
      "epoch": 0.05132,
      "grad_norm": 1.0458562389037795,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 5132
    },
    {
      "epoch": 0.05133,
      "grad_norm": 0.9786133915152717,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 5133
    },
    {
      "epoch": 0.05134,
      "grad_norm": 0.9697734611114633,
      "learning_rate": 0.003,
      "loss": 4.1714,
      "step": 5134
    },
    {
      "epoch": 0.05135,
      "grad_norm": 0.9363895927818596,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 5135
    },
    {
      "epoch": 0.05136,
      "grad_norm": 1.0170683776594798,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 5136
    },
    {
      "epoch": 0.05137,
      "grad_norm": 1.1145412465734708,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 5137
    },
    {
      "epoch": 0.05138,
      "grad_norm": 0.9941607682684069,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 5138
    },
    {
      "epoch": 0.05139,
      "grad_norm": 0.978207963075468,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 5139
    },
    {
      "epoch": 0.0514,
      "grad_norm": 1.0158981617955298,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 5140
    },
    {
      "epoch": 0.05141,
      "grad_norm": 0.8431378615881333,
      "learning_rate": 0.003,
      "loss": 4.2033,
      "step": 5141
    },
    {
      "epoch": 0.05142,
      "grad_norm": 0.9258386606688979,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 5142
    },
    {
      "epoch": 0.05143,
      "grad_norm": 0.8809664967212562,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 5143
    },
    {
      "epoch": 0.05144,
      "grad_norm": 0.808685397208462,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 5144
    },
    {
      "epoch": 0.05145,
      "grad_norm": 0.7632108507977365,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 5145
    },
    {
      "epoch": 0.05146,
      "grad_norm": 0.6541780509288703,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 5146
    },
    {
      "epoch": 0.05147,
      "grad_norm": 0.5225402009574097,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 5147
    },
    {
      "epoch": 0.05148,
      "grad_norm": 0.4857663422658965,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 5148
    },
    {
      "epoch": 0.05149,
      "grad_norm": 0.5849142561064581,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 5149
    },
    {
      "epoch": 0.0515,
      "grad_norm": 0.6123859001036767,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 5150
    },
    {
      "epoch": 0.05151,
      "grad_norm": 0.7026699600678006,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 5151
    },
    {
      "epoch": 0.05152,
      "grad_norm": 0.6504209042889696,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 5152
    },
    {
      "epoch": 0.05153,
      "grad_norm": 0.6252489466134844,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 5153
    },
    {
      "epoch": 0.05154,
      "grad_norm": 0.6371096455455558,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 5154
    },
    {
      "epoch": 0.05155,
      "grad_norm": 0.5613653424439898,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 5155
    },
    {
      "epoch": 0.05156,
      "grad_norm": 0.5387099035222069,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 5156
    },
    {
      "epoch": 0.05157,
      "grad_norm": 0.5796148701609515,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 5157
    },
    {
      "epoch": 0.05158,
      "grad_norm": 0.6280103262144628,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 5158
    },
    {
      "epoch": 0.05159,
      "grad_norm": 0.713561476333827,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 5159
    },
    {
      "epoch": 0.0516,
      "grad_norm": 0.7440717469043273,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 5160
    },
    {
      "epoch": 0.05161,
      "grad_norm": 0.693854318455779,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 5161
    },
    {
      "epoch": 0.05162,
      "grad_norm": 0.5713907895082873,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 5162
    },
    {
      "epoch": 0.05163,
      "grad_norm": 0.6389112592034603,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 5163
    },
    {
      "epoch": 0.05164,
      "grad_norm": 0.644621023044623,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 5164
    },
    {
      "epoch": 0.05165,
      "grad_norm": 0.6418172439331881,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 5165
    },
    {
      "epoch": 0.05166,
      "grad_norm": 0.6129087318474767,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 5166
    },
    {
      "epoch": 0.05167,
      "grad_norm": 0.6032353120193177,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 5167
    },
    {
      "epoch": 0.05168,
      "grad_norm": 0.5247219196682845,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 5168
    },
    {
      "epoch": 0.05169,
      "grad_norm": 0.5175916837961249,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 5169
    },
    {
      "epoch": 0.0517,
      "grad_norm": 0.6351381415951768,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 5170
    },
    {
      "epoch": 0.05171,
      "grad_norm": 0.9707906373263301,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 5171
    },
    {
      "epoch": 0.05172,
      "grad_norm": 1.3871012373265488,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 5172
    },
    {
      "epoch": 0.05173,
      "grad_norm": 0.6859878951673775,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 5173
    },
    {
      "epoch": 0.05174,
      "grad_norm": 0.688770612830069,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 5174
    },
    {
      "epoch": 0.05175,
      "grad_norm": 0.7348674861956664,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 5175
    },
    {
      "epoch": 0.05176,
      "grad_norm": 0.7293852449611181,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 5176
    },
    {
      "epoch": 0.05177,
      "grad_norm": 0.6564692961913474,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 5177
    },
    {
      "epoch": 0.05178,
      "grad_norm": 0.6743163086709401,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 5178
    },
    {
      "epoch": 0.05179,
      "grad_norm": 0.7124422218567592,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5179
    },
    {
      "epoch": 0.0518,
      "grad_norm": 0.7927407152562181,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 5180
    },
    {
      "epoch": 0.05181,
      "grad_norm": 0.7257533748695947,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 5181
    },
    {
      "epoch": 0.05182,
      "grad_norm": 0.6857538651077083,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 5182
    },
    {
      "epoch": 0.05183,
      "grad_norm": 0.6165169903841881,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 5183
    },
    {
      "epoch": 0.05184,
      "grad_norm": 0.5646217873608306,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 5184
    },
    {
      "epoch": 0.05185,
      "grad_norm": 0.6765417920434729,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 5185
    },
    {
      "epoch": 0.05186,
      "grad_norm": 0.763739343016591,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 5186
    },
    {
      "epoch": 0.05187,
      "grad_norm": 0.8133701961716747,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 5187
    },
    {
      "epoch": 0.05188,
      "grad_norm": 0.9136528986209898,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 5188
    },
    {
      "epoch": 0.05189,
      "grad_norm": 1.1309518830748815,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 5189
    },
    {
      "epoch": 0.0519,
      "grad_norm": 1.1084320754378056,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 5190
    },
    {
      "epoch": 0.05191,
      "grad_norm": 0.7822117289248245,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 5191
    },
    {
      "epoch": 0.05192,
      "grad_norm": 0.6054235823299604,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 5192
    },
    {
      "epoch": 0.05193,
      "grad_norm": 0.7272923940874335,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 5193
    },
    {
      "epoch": 0.05194,
      "grad_norm": 1.053286129123416,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 5194
    },
    {
      "epoch": 0.05195,
      "grad_norm": 1.0774389375766364,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 5195
    },
    {
      "epoch": 0.05196,
      "grad_norm": 0.7935135429431,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 5196
    },
    {
      "epoch": 0.05197,
      "grad_norm": 0.6765899030219741,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 5197
    },
    {
      "epoch": 0.05198,
      "grad_norm": 0.7288470154808702,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 5198
    },
    {
      "epoch": 0.05199,
      "grad_norm": 0.7762676094179997,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 5199
    },
    {
      "epoch": 0.052,
      "grad_norm": 0.8517147173377261,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 5200
    },
    {
      "epoch": 0.05201,
      "grad_norm": 0.7812345967785519,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 5201
    },
    {
      "epoch": 0.05202,
      "grad_norm": 0.7915734925208033,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 5202
    },
    {
      "epoch": 0.05203,
      "grad_norm": 0.779004263274718,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 5203
    },
    {
      "epoch": 0.05204,
      "grad_norm": 0.8440004396702558,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 5204
    },
    {
      "epoch": 0.05205,
      "grad_norm": 0.8044224640384809,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 5205
    },
    {
      "epoch": 0.05206,
      "grad_norm": 0.830818456670086,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 5206
    },
    {
      "epoch": 0.05207,
      "grad_norm": 0.840221891416657,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 5207
    },
    {
      "epoch": 0.05208,
      "grad_norm": 0.8228927798697641,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 5208
    },
    {
      "epoch": 0.05209,
      "grad_norm": 0.7538723463798628,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 5209
    },
    {
      "epoch": 0.0521,
      "grad_norm": 0.7269914660347427,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 5210
    },
    {
      "epoch": 0.05211,
      "grad_norm": 0.6396845788398585,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 5211
    },
    {
      "epoch": 0.05212,
      "grad_norm": 0.6578986704720922,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 5212
    },
    {
      "epoch": 0.05213,
      "grad_norm": 0.6665917425071297,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 5213
    },
    {
      "epoch": 0.05214,
      "grad_norm": 0.6696745228858743,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 5214
    },
    {
      "epoch": 0.05215,
      "grad_norm": 0.8360803674253388,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 5215
    },
    {
      "epoch": 0.05216,
      "grad_norm": 1.0234029588407536,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 5216
    },
    {
      "epoch": 0.05217,
      "grad_norm": 0.8990465132650213,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 5217
    },
    {
      "epoch": 0.05218,
      "grad_norm": 0.7789964318595968,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 5218
    },
    {
      "epoch": 0.05219,
      "grad_norm": 0.7298439479090568,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 5219
    },
    {
      "epoch": 0.0522,
      "grad_norm": 0.8002872146937169,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 5220
    },
    {
      "epoch": 0.05221,
      "grad_norm": 0.8077845914697496,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 5221
    },
    {
      "epoch": 0.05222,
      "grad_norm": 0.8937695724547757,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 5222
    },
    {
      "epoch": 0.05223,
      "grad_norm": 0.8660989097279149,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 5223
    },
    {
      "epoch": 0.05224,
      "grad_norm": 0.8551329970671594,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 5224
    },
    {
      "epoch": 0.05225,
      "grad_norm": 0.7915686333783567,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 5225
    },
    {
      "epoch": 0.05226,
      "grad_norm": 0.8295912864972166,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 5226
    },
    {
      "epoch": 0.05227,
      "grad_norm": 0.7362810272638042,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 5227
    },
    {
      "epoch": 0.05228,
      "grad_norm": 0.7001627229979629,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 5228
    },
    {
      "epoch": 0.05229,
      "grad_norm": 0.7602202740566725,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 5229
    },
    {
      "epoch": 0.0523,
      "grad_norm": 0.6912471723136686,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 5230
    },
    {
      "epoch": 0.05231,
      "grad_norm": 0.7018975883307987,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 5231
    },
    {
      "epoch": 0.05232,
      "grad_norm": 0.717854420616106,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 5232
    },
    {
      "epoch": 0.05233,
      "grad_norm": 0.7712695717441904,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 5233
    },
    {
      "epoch": 0.05234,
      "grad_norm": 0.8385680879325836,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 5234
    },
    {
      "epoch": 0.05235,
      "grad_norm": 0.9750232594843035,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 5235
    },
    {
      "epoch": 0.05236,
      "grad_norm": 0.8446359663530114,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 5236
    },
    {
      "epoch": 0.05237,
      "grad_norm": 0.9096810630918026,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 5237
    },
    {
      "epoch": 0.05238,
      "grad_norm": 0.9079420042262413,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 5238
    },
    {
      "epoch": 0.05239,
      "grad_norm": 0.8141480866664513,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 5239
    },
    {
      "epoch": 0.0524,
      "grad_norm": 0.7056408832787606,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5240
    },
    {
      "epoch": 0.05241,
      "grad_norm": 0.6654549467738141,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5241
    },
    {
      "epoch": 0.05242,
      "grad_norm": 0.7294744123105587,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 5242
    },
    {
      "epoch": 0.05243,
      "grad_norm": 0.7203093785089144,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 5243
    },
    {
      "epoch": 0.05244,
      "grad_norm": 0.6894002827090067,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 5244
    },
    {
      "epoch": 0.05245,
      "grad_norm": 0.6674737328058699,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 5245
    },
    {
      "epoch": 0.05246,
      "grad_norm": 0.6789552091924481,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5246
    },
    {
      "epoch": 0.05247,
      "grad_norm": 0.7027379397907318,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5247
    },
    {
      "epoch": 0.05248,
      "grad_norm": 0.7672915233226524,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 5248
    },
    {
      "epoch": 0.05249,
      "grad_norm": 0.8600082125105806,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 5249
    },
    {
      "epoch": 0.0525,
      "grad_norm": 0.8053204409517745,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 5250
    },
    {
      "epoch": 0.05251,
      "grad_norm": 0.7017613490500446,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 5251
    },
    {
      "epoch": 0.05252,
      "grad_norm": 0.6533051938670863,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 5252
    },
    {
      "epoch": 0.05253,
      "grad_norm": 0.6002345144376094,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 5253
    },
    {
      "epoch": 0.05254,
      "grad_norm": 0.5673372560372355,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 5254
    },
    {
      "epoch": 0.05255,
      "grad_norm": 0.5743257458749801,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 5255
    },
    {
      "epoch": 0.05256,
      "grad_norm": 0.5639251305320502,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 5256
    },
    {
      "epoch": 0.05257,
      "grad_norm": 0.5046284771633743,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 5257
    },
    {
      "epoch": 0.05258,
      "grad_norm": 0.474168541210226,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 5258
    },
    {
      "epoch": 0.05259,
      "grad_norm": 0.47547400835554277,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 5259
    },
    {
      "epoch": 0.0526,
      "grad_norm": 0.5164451968626312,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 5260
    },
    {
      "epoch": 0.05261,
      "grad_norm": 0.5250991374606924,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 5261
    },
    {
      "epoch": 0.05262,
      "grad_norm": 0.6851820393823259,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 5262
    },
    {
      "epoch": 0.05263,
      "grad_norm": 0.9977081433022779,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 5263
    },
    {
      "epoch": 0.05264,
      "grad_norm": 1.268054358654276,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 5264
    },
    {
      "epoch": 0.05265,
      "grad_norm": 0.5767306016131388,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 5265
    },
    {
      "epoch": 0.05266,
      "grad_norm": 0.7952501578678153,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 5266
    },
    {
      "epoch": 0.05267,
      "grad_norm": 1.0070088823226309,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 5267
    },
    {
      "epoch": 0.05268,
      "grad_norm": 0.9932744688310504,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 5268
    },
    {
      "epoch": 0.05269,
      "grad_norm": 1.0388601409327192,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 5269
    },
    {
      "epoch": 0.0527,
      "grad_norm": 0.956313275292693,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 5270
    },
    {
      "epoch": 0.05271,
      "grad_norm": 1.0292710271964594,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 5271
    },
    {
      "epoch": 0.05272,
      "grad_norm": 0.8930084447233932,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 5272
    },
    {
      "epoch": 0.05273,
      "grad_norm": 0.9405712293541173,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 5273
    },
    {
      "epoch": 0.05274,
      "grad_norm": 1.0695246435294683,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 5274
    },
    {
      "epoch": 0.05275,
      "grad_norm": 0.9058378894957746,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 5275
    },
    {
      "epoch": 0.05276,
      "grad_norm": 0.8493287470524594,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5276
    },
    {
      "epoch": 0.05277,
      "grad_norm": 0.9420450553310487,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 5277
    },
    {
      "epoch": 0.05278,
      "grad_norm": 1.1979114145894763,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 5278
    },
    {
      "epoch": 0.05279,
      "grad_norm": 0.9851879183907458,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 5279
    },
    {
      "epoch": 0.0528,
      "grad_norm": 0.9906333358429689,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 5280
    },
    {
      "epoch": 0.05281,
      "grad_norm": 1.0018246671795275,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 5281
    },
    {
      "epoch": 0.05282,
      "grad_norm": 0.9088563498086558,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 5282
    },
    {
      "epoch": 0.05283,
      "grad_norm": 0.8255004872484286,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 5283
    },
    {
      "epoch": 0.05284,
      "grad_norm": 0.7395222306618988,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 5284
    },
    {
      "epoch": 0.05285,
      "grad_norm": 0.7519757069394422,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 5285
    },
    {
      "epoch": 0.05286,
      "grad_norm": 0.8309929954214635,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 5286
    },
    {
      "epoch": 0.05287,
      "grad_norm": 0.8541420130148721,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 5287
    },
    {
      "epoch": 0.05288,
      "grad_norm": 0.9074648300179621,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 5288
    },
    {
      "epoch": 0.05289,
      "grad_norm": 0.934169321459613,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 5289
    },
    {
      "epoch": 0.0529,
      "grad_norm": 0.8614434047866291,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 5290
    },
    {
      "epoch": 0.05291,
      "grad_norm": 0.7851323632915874,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 5291
    },
    {
      "epoch": 0.05292,
      "grad_norm": 0.8846666231031659,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 5292
    },
    {
      "epoch": 0.05293,
      "grad_norm": 0.8272989171480324,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 5293
    },
    {
      "epoch": 0.05294,
      "grad_norm": 0.7925051182542472,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 5294
    },
    {
      "epoch": 0.05295,
      "grad_norm": 0.8458687864133015,
      "learning_rate": 0.003,
      "loss": 4.1485,
      "step": 5295
    },
    {
      "epoch": 0.05296,
      "grad_norm": 0.7329229801657224,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5296
    },
    {
      "epoch": 0.05297,
      "grad_norm": 0.590148048628958,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5297
    },
    {
      "epoch": 0.05298,
      "grad_norm": 0.5633959457343373,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 5298
    },
    {
      "epoch": 0.05299,
      "grad_norm": 0.6047041287435936,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 5299
    },
    {
      "epoch": 0.053,
      "grad_norm": 0.6026083589407333,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 5300
    },
    {
      "epoch": 0.05301,
      "grad_norm": 0.6614621931672765,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 5301
    },
    {
      "epoch": 0.05302,
      "grad_norm": 0.667460840133991,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 5302
    },
    {
      "epoch": 0.05303,
      "grad_norm": 0.7178043539507469,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 5303
    },
    {
      "epoch": 0.05304,
      "grad_norm": 0.7057622216619723,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 5304
    },
    {
      "epoch": 0.05305,
      "grad_norm": 0.6395996129884405,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 5305
    },
    {
      "epoch": 0.05306,
      "grad_norm": 0.6595661147403964,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 5306
    },
    {
      "epoch": 0.05307,
      "grad_norm": 0.5697470297170749,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 5307
    },
    {
      "epoch": 0.05308,
      "grad_norm": 0.6269529857264409,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 5308
    },
    {
      "epoch": 0.05309,
      "grad_norm": 0.7109602470701628,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 5309
    },
    {
      "epoch": 0.0531,
      "grad_norm": 0.8779710810213476,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 5310
    },
    {
      "epoch": 0.05311,
      "grad_norm": 0.9865994447177229,
      "learning_rate": 0.003,
      "loss": 4.161,
      "step": 5311
    },
    {
      "epoch": 0.05312,
      "grad_norm": 0.928167243969855,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 5312
    },
    {
      "epoch": 0.05313,
      "grad_norm": 0.65760273068672,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 5313
    },
    {
      "epoch": 0.05314,
      "grad_norm": 0.6095510092870687,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 5314
    },
    {
      "epoch": 0.05315,
      "grad_norm": 0.7424516970237663,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 5315
    },
    {
      "epoch": 0.05316,
      "grad_norm": 0.8440091626661397,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 5316
    },
    {
      "epoch": 0.05317,
      "grad_norm": 0.9378368997983185,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 5317
    },
    {
      "epoch": 0.05318,
      "grad_norm": 0.8048248745024449,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 5318
    },
    {
      "epoch": 0.05319,
      "grad_norm": 0.7236000931693785,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 5319
    },
    {
      "epoch": 0.0532,
      "grad_norm": 0.6461031092410674,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 5320
    },
    {
      "epoch": 0.05321,
      "grad_norm": 0.7679854561007805,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 5321
    },
    {
      "epoch": 0.05322,
      "grad_norm": 0.7986946965085385,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5322
    },
    {
      "epoch": 0.05323,
      "grad_norm": 0.7153375585884025,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 5323
    },
    {
      "epoch": 0.05324,
      "grad_norm": 0.749330813072559,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 5324
    },
    {
      "epoch": 0.05325,
      "grad_norm": 0.832758253039198,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 5325
    },
    {
      "epoch": 0.05326,
      "grad_norm": 0.7759697554212088,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 5326
    },
    {
      "epoch": 0.05327,
      "grad_norm": 0.7232266024198855,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 5327
    },
    {
      "epoch": 0.05328,
      "grad_norm": 0.7836949776849659,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 5328
    },
    {
      "epoch": 0.05329,
      "grad_norm": 0.7124265190034871,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 5329
    },
    {
      "epoch": 0.0533,
      "grad_norm": 0.7073201953779398,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 5330
    },
    {
      "epoch": 0.05331,
      "grad_norm": 0.6517329606524963,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 5331
    },
    {
      "epoch": 0.05332,
      "grad_norm": 0.7675020790673767,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 5332
    },
    {
      "epoch": 0.05333,
      "grad_norm": 0.8204477024627048,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 5333
    },
    {
      "epoch": 0.05334,
      "grad_norm": 0.7022074746197576,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 5334
    },
    {
      "epoch": 0.05335,
      "grad_norm": 0.6275459710277902,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 5335
    },
    {
      "epoch": 0.05336,
      "grad_norm": 0.6160252429147814,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 5336
    },
    {
      "epoch": 0.05337,
      "grad_norm": 0.6645784185963264,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 5337
    },
    {
      "epoch": 0.05338,
      "grad_norm": 0.755147688591901,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 5338
    },
    {
      "epoch": 0.05339,
      "grad_norm": 0.8974685697024821,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 5339
    },
    {
      "epoch": 0.0534,
      "grad_norm": 1.0247940652403054,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 5340
    },
    {
      "epoch": 0.05341,
      "grad_norm": 0.8398376000828047,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 5341
    },
    {
      "epoch": 0.05342,
      "grad_norm": 0.6251041860327448,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 5342
    },
    {
      "epoch": 0.05343,
      "grad_norm": 0.6653717662444403,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 5343
    },
    {
      "epoch": 0.05344,
      "grad_norm": 0.7626824156057184,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 5344
    },
    {
      "epoch": 0.05345,
      "grad_norm": 0.7412027124944542,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 5345
    },
    {
      "epoch": 0.05346,
      "grad_norm": 0.8204551363999621,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 5346
    },
    {
      "epoch": 0.05347,
      "grad_norm": 0.8691311780811011,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 5347
    },
    {
      "epoch": 0.05348,
      "grad_norm": 0.8297575781225915,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 5348
    },
    {
      "epoch": 0.05349,
      "grad_norm": 0.7800308176994237,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 5349
    },
    {
      "epoch": 0.0535,
      "grad_norm": 0.7428195062452232,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 5350
    },
    {
      "epoch": 0.05351,
      "grad_norm": 0.698577766705363,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 5351
    },
    {
      "epoch": 0.05352,
      "grad_norm": 0.7463247671330128,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 5352
    },
    {
      "epoch": 0.05353,
      "grad_norm": 0.8427380940568103,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 5353
    },
    {
      "epoch": 0.05354,
      "grad_norm": 0.8580345957059121,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 5354
    },
    {
      "epoch": 0.05355,
      "grad_norm": 0.7981948995131939,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 5355
    },
    {
      "epoch": 0.05356,
      "grad_norm": 0.9987056987093543,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 5356
    },
    {
      "epoch": 0.05357,
      "grad_norm": 1.0863630771998207,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 5357
    },
    {
      "epoch": 0.05358,
      "grad_norm": 0.8873987298102782,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 5358
    },
    {
      "epoch": 0.05359,
      "grad_norm": 0.8982951690299781,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 5359
    },
    {
      "epoch": 0.0536,
      "grad_norm": 0.7420851849216799,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 5360
    },
    {
      "epoch": 0.05361,
      "grad_norm": 0.6478690518338635,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 5361
    },
    {
      "epoch": 0.05362,
      "grad_norm": 0.7363423400186474,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 5362
    },
    {
      "epoch": 0.05363,
      "grad_norm": 0.8098849889259154,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 5363
    },
    {
      "epoch": 0.05364,
      "grad_norm": 0.8901440262592046,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 5364
    },
    {
      "epoch": 0.05365,
      "grad_norm": 1.0380282918537116,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 5365
    },
    {
      "epoch": 0.05366,
      "grad_norm": 1.0788058766201782,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 5366
    },
    {
      "epoch": 0.05367,
      "grad_norm": 0.7608583479176579,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 5367
    },
    {
      "epoch": 0.05368,
      "grad_norm": 0.5600149243746851,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 5368
    },
    {
      "epoch": 0.05369,
      "grad_norm": 0.5618528481117717,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 5369
    },
    {
      "epoch": 0.0537,
      "grad_norm": 0.6773018479702052,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 5370
    },
    {
      "epoch": 0.05371,
      "grad_norm": 0.8077481499415716,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 5371
    },
    {
      "epoch": 0.05372,
      "grad_norm": 0.9098981004391098,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 5372
    },
    {
      "epoch": 0.05373,
      "grad_norm": 0.8688252009595805,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 5373
    },
    {
      "epoch": 0.05374,
      "grad_norm": 0.7822281391029792,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 5374
    },
    {
      "epoch": 0.05375,
      "grad_norm": 0.6393823720543013,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 5375
    },
    {
      "epoch": 0.05376,
      "grad_norm": 0.6297681205873783,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 5376
    },
    {
      "epoch": 0.05377,
      "grad_norm": 0.5591158983975534,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 5377
    },
    {
      "epoch": 0.05378,
      "grad_norm": 0.5827833277734367,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 5378
    },
    {
      "epoch": 0.05379,
      "grad_norm": 0.5267920451244182,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 5379
    },
    {
      "epoch": 0.0538,
      "grad_norm": 0.5153469929794081,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 5380
    },
    {
      "epoch": 0.05381,
      "grad_norm": 0.541236643808538,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 5381
    },
    {
      "epoch": 0.05382,
      "grad_norm": 0.5524397786242985,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 5382
    },
    {
      "epoch": 0.05383,
      "grad_norm": 0.5171016444028097,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 5383
    },
    {
      "epoch": 0.05384,
      "grad_norm": 0.4322886075458978,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 5384
    },
    {
      "epoch": 0.05385,
      "grad_norm": 0.45234091719568803,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 5385
    },
    {
      "epoch": 0.05386,
      "grad_norm": 0.4669069348140255,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 5386
    },
    {
      "epoch": 0.05387,
      "grad_norm": 0.505314292670269,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 5387
    },
    {
      "epoch": 0.05388,
      "grad_norm": 0.6661928342962524,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 5388
    },
    {
      "epoch": 0.05389,
      "grad_norm": 1.0017958320911928,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 5389
    },
    {
      "epoch": 0.0539,
      "grad_norm": 1.2992679013311437,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 5390
    },
    {
      "epoch": 0.05391,
      "grad_norm": 0.5736546179541971,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 5391
    },
    {
      "epoch": 0.05392,
      "grad_norm": 0.7440996442059347,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 5392
    },
    {
      "epoch": 0.05393,
      "grad_norm": 0.9950521432118898,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 5393
    },
    {
      "epoch": 0.05394,
      "grad_norm": 0.7875014530755975,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 5394
    },
    {
      "epoch": 0.05395,
      "grad_norm": 0.7254081418838102,
      "learning_rate": 0.003,
      "loss": 4.1479,
      "step": 5395
    },
    {
      "epoch": 0.05396,
      "grad_norm": 0.8612136110888624,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 5396
    },
    {
      "epoch": 0.05397,
      "grad_norm": 0.8986673711381578,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 5397
    },
    {
      "epoch": 0.05398,
      "grad_norm": 0.778809786786069,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 5398
    },
    {
      "epoch": 0.05399,
      "grad_norm": 0.7995244546793957,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 5399
    },
    {
      "epoch": 0.054,
      "grad_norm": 0.7896689616881526,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 5400
    },
    {
      "epoch": 0.05401,
      "grad_norm": 0.7804724282282718,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 5401
    },
    {
      "epoch": 0.05402,
      "grad_norm": 0.7709619532218206,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 5402
    },
    {
      "epoch": 0.05403,
      "grad_norm": 0.7494119205086632,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 5403
    },
    {
      "epoch": 0.05404,
      "grad_norm": 0.7695850697537835,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 5404
    },
    {
      "epoch": 0.05405,
      "grad_norm": 0.6494919985184457,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 5405
    },
    {
      "epoch": 0.05406,
      "grad_norm": 0.7107256413996718,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 5406
    },
    {
      "epoch": 0.05407,
      "grad_norm": 0.8097891472467551,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 5407
    },
    {
      "epoch": 0.05408,
      "grad_norm": 0.9502406141177873,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 5408
    },
    {
      "epoch": 0.05409,
      "grad_norm": 0.9823219241510378,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 5409
    },
    {
      "epoch": 0.0541,
      "grad_norm": 0.878184005124274,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 5410
    },
    {
      "epoch": 0.05411,
      "grad_norm": 0.783228910710241,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 5411
    },
    {
      "epoch": 0.05412,
      "grad_norm": 0.7419337388906813,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 5412
    },
    {
      "epoch": 0.05413,
      "grad_norm": 0.8677947411304687,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 5413
    },
    {
      "epoch": 0.05414,
      "grad_norm": 0.754625111832618,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 5414
    },
    {
      "epoch": 0.05415,
      "grad_norm": 0.6166571551752229,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 5415
    },
    {
      "epoch": 0.05416,
      "grad_norm": 0.6149867683651459,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 5416
    },
    {
      "epoch": 0.05417,
      "grad_norm": 0.7707506569643702,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 5417
    },
    {
      "epoch": 0.05418,
      "grad_norm": 0.8827482178277591,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 5418
    },
    {
      "epoch": 0.05419,
      "grad_norm": 1.0696377805286104,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 5419
    },
    {
      "epoch": 0.0542,
      "grad_norm": 1.0762623846918855,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 5420
    },
    {
      "epoch": 0.05421,
      "grad_norm": 1.0370239550620723,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 5421
    },
    {
      "epoch": 0.05422,
      "grad_norm": 0.8836609588737956,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 5422
    },
    {
      "epoch": 0.05423,
      "grad_norm": 0.9202200994596317,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 5423
    },
    {
      "epoch": 0.05424,
      "grad_norm": 0.8512778459324802,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 5424
    },
    {
      "epoch": 0.05425,
      "grad_norm": 0.9221497425172506,
      "learning_rate": 0.003,
      "loss": 4.1457,
      "step": 5425
    },
    {
      "epoch": 0.05426,
      "grad_norm": 0.8858038824729005,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 5426
    },
    {
      "epoch": 0.05427,
      "grad_norm": 0.8461598509812891,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 5427
    },
    {
      "epoch": 0.05428,
      "grad_norm": 0.7838983529721615,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 5428
    },
    {
      "epoch": 0.05429,
      "grad_norm": 0.7381870936331452,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 5429
    },
    {
      "epoch": 0.0543,
      "grad_norm": 0.6675438001202542,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 5430
    },
    {
      "epoch": 0.05431,
      "grad_norm": 0.7468875444789291,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 5431
    },
    {
      "epoch": 0.05432,
      "grad_norm": 0.7685334970223217,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 5432
    },
    {
      "epoch": 0.05433,
      "grad_norm": 0.7841989897169739,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 5433
    },
    {
      "epoch": 0.05434,
      "grad_norm": 0.7643517687218986,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5434
    },
    {
      "epoch": 0.05435,
      "grad_norm": 0.7188445789131351,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 5435
    },
    {
      "epoch": 0.05436,
      "grad_norm": 0.6666336203457994,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 5436
    },
    {
      "epoch": 0.05437,
      "grad_norm": 0.5578867371709249,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 5437
    },
    {
      "epoch": 0.05438,
      "grad_norm": 0.6023052773273868,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 5438
    },
    {
      "epoch": 0.05439,
      "grad_norm": 0.686819083498365,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 5439
    },
    {
      "epoch": 0.0544,
      "grad_norm": 0.7877497063451424,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 5440
    },
    {
      "epoch": 0.05441,
      "grad_norm": 0.8224525297404207,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 5441
    },
    {
      "epoch": 0.05442,
      "grad_norm": 0.8963412844552312,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 5442
    },
    {
      "epoch": 0.05443,
      "grad_norm": 0.7843413799075878,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 5443
    },
    {
      "epoch": 0.05444,
      "grad_norm": 0.6677654509581011,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 5444
    },
    {
      "epoch": 0.05445,
      "grad_norm": 0.5372869992196785,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 5445
    },
    {
      "epoch": 0.05446,
      "grad_norm": 0.4853458301953657,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 5446
    },
    {
      "epoch": 0.05447,
      "grad_norm": 0.534235282926048,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 5447
    },
    {
      "epoch": 0.05448,
      "grad_norm": 0.6051573720194104,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 5448
    },
    {
      "epoch": 0.05449,
      "grad_norm": 0.7586494998052061,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 5449
    },
    {
      "epoch": 0.0545,
      "grad_norm": 0.8213010225949955,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 5450
    },
    {
      "epoch": 0.05451,
      "grad_norm": 0.7188380237062496,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 5451
    },
    {
      "epoch": 0.05452,
      "grad_norm": 0.5272820320310276,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 5452
    },
    {
      "epoch": 0.05453,
      "grad_norm": 0.4825898765859038,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 5453
    },
    {
      "epoch": 0.05454,
      "grad_norm": 0.5231707853046749,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 5454
    },
    {
      "epoch": 0.05455,
      "grad_norm": 0.6109918111895452,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 5455
    },
    {
      "epoch": 0.05456,
      "grad_norm": 0.717222959857415,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 5456
    },
    {
      "epoch": 0.05457,
      "grad_norm": 0.8551566509250617,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 5457
    },
    {
      "epoch": 0.05458,
      "grad_norm": 0.9372765867290779,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 5458
    },
    {
      "epoch": 0.05459,
      "grad_norm": 0.9320602167981404,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 5459
    },
    {
      "epoch": 0.0546,
      "grad_norm": 0.7627661268344466,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 5460
    },
    {
      "epoch": 0.05461,
      "grad_norm": 0.6315449710507808,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 5461
    },
    {
      "epoch": 0.05462,
      "grad_norm": 0.7123926231181666,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 5462
    },
    {
      "epoch": 0.05463,
      "grad_norm": 0.7549211607821605,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 5463
    },
    {
      "epoch": 0.05464,
      "grad_norm": 0.8370379354643269,
      "learning_rate": 0.003,
      "loss": 4.1503,
      "step": 5464
    },
    {
      "epoch": 0.05465,
      "grad_norm": 0.9542926928325179,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 5465
    },
    {
      "epoch": 0.05466,
      "grad_norm": 1.054115432669863,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 5466
    },
    {
      "epoch": 0.05467,
      "grad_norm": 0.8917210500728532,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 5467
    },
    {
      "epoch": 0.05468,
      "grad_norm": 0.8412209468493033,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 5468
    },
    {
      "epoch": 0.05469,
      "grad_norm": 0.7971647310022061,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 5469
    },
    {
      "epoch": 0.0547,
      "grad_norm": 0.8775383662555644,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 5470
    },
    {
      "epoch": 0.05471,
      "grad_norm": 0.9125028901592531,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 5471
    },
    {
      "epoch": 0.05472,
      "grad_norm": 0.7767839416762876,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 5472
    },
    {
      "epoch": 0.05473,
      "grad_norm": 0.7509158035092801,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 5473
    },
    {
      "epoch": 0.05474,
      "grad_norm": 0.8240995260680938,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 5474
    },
    {
      "epoch": 0.05475,
      "grad_norm": 0.9739217618215972,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 5475
    },
    {
      "epoch": 0.05476,
      "grad_norm": 1.0112035995282798,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 5476
    },
    {
      "epoch": 0.05477,
      "grad_norm": 0.8701723078990121,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 5477
    },
    {
      "epoch": 0.05478,
      "grad_norm": 0.745391995716312,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 5478
    },
    {
      "epoch": 0.05479,
      "grad_norm": 0.7047469678929648,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 5479
    },
    {
      "epoch": 0.0548,
      "grad_norm": 0.7568179897684124,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 5480
    },
    {
      "epoch": 0.05481,
      "grad_norm": 0.842291746689268,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 5481
    },
    {
      "epoch": 0.05482,
      "grad_norm": 0.8335768637729064,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 5482
    },
    {
      "epoch": 0.05483,
      "grad_norm": 0.8356448898440997,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 5483
    },
    {
      "epoch": 0.05484,
      "grad_norm": 0.8347694767449836,
      "learning_rate": 0.003,
      "loss": 4.1485,
      "step": 5484
    },
    {
      "epoch": 0.05485,
      "grad_norm": 0.7596611463998084,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 5485
    },
    {
      "epoch": 0.05486,
      "grad_norm": 0.7203868901609732,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 5486
    },
    {
      "epoch": 0.05487,
      "grad_norm": 0.7045034193470036,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 5487
    },
    {
      "epoch": 0.05488,
      "grad_norm": 0.7134164831679168,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 5488
    },
    {
      "epoch": 0.05489,
      "grad_norm": 0.7566204654677959,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 5489
    },
    {
      "epoch": 0.0549,
      "grad_norm": 0.753089787028579,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5490
    },
    {
      "epoch": 0.05491,
      "grad_norm": 0.7900961174130867,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 5491
    },
    {
      "epoch": 0.05492,
      "grad_norm": 0.7735481675330634,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 5492
    },
    {
      "epoch": 0.05493,
      "grad_norm": 0.7921324153077896,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 5493
    },
    {
      "epoch": 0.05494,
      "grad_norm": 0.7783756160879682,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 5494
    },
    {
      "epoch": 0.05495,
      "grad_norm": 0.8354402801680301,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 5495
    },
    {
      "epoch": 0.05496,
      "grad_norm": 0.6922018070920165,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 5496
    },
    {
      "epoch": 0.05497,
      "grad_norm": 0.6446395374277146,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 5497
    },
    {
      "epoch": 0.05498,
      "grad_norm": 0.6080103999530705,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 5498
    },
    {
      "epoch": 0.05499,
      "grad_norm": 0.6519375925877654,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 5499
    },
    {
      "epoch": 0.055,
      "grad_norm": 0.5807988789895889,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 5500
    },
    {
      "epoch": 0.05501,
      "grad_norm": 0.5252348957768659,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 5501
    },
    {
      "epoch": 0.05502,
      "grad_norm": 0.6442067499919607,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 5502
    },
    {
      "epoch": 0.05503,
      "grad_norm": 0.6540019754179668,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 5503
    },
    {
      "epoch": 0.05504,
      "grad_norm": 0.6146856306540659,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 5504
    },
    {
      "epoch": 0.05505,
      "grad_norm": 0.6120251915961363,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 5505
    },
    {
      "epoch": 0.05506,
      "grad_norm": 0.7454876837271709,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 5506
    },
    {
      "epoch": 0.05507,
      "grad_norm": 0.7761307027748946,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 5507
    },
    {
      "epoch": 0.05508,
      "grad_norm": 0.7800855127913302,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 5508
    },
    {
      "epoch": 0.05509,
      "grad_norm": 0.9996587539628538,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 5509
    },
    {
      "epoch": 0.0551,
      "grad_norm": 1.2203215437155306,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 5510
    },
    {
      "epoch": 0.05511,
      "grad_norm": 0.8793583971610104,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 5511
    },
    {
      "epoch": 0.05512,
      "grad_norm": 0.7370740348757334,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 5512
    },
    {
      "epoch": 0.05513,
      "grad_norm": 0.6510573891135065,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 5513
    },
    {
      "epoch": 0.05514,
      "grad_norm": 0.7249273397693745,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 5514
    },
    {
      "epoch": 0.05515,
      "grad_norm": 0.7885147514411773,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 5515
    },
    {
      "epoch": 0.05516,
      "grad_norm": 0.9148149024980577,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 5516
    },
    {
      "epoch": 0.05517,
      "grad_norm": 0.9500608738728219,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 5517
    },
    {
      "epoch": 0.05518,
      "grad_norm": 0.8711962258266909,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 5518
    },
    {
      "epoch": 0.05519,
      "grad_norm": 0.8912312013555587,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 5519
    },
    {
      "epoch": 0.0552,
      "grad_norm": 0.7356449479297673,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 5520
    },
    {
      "epoch": 0.05521,
      "grad_norm": 0.652079639982586,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 5521
    },
    {
      "epoch": 0.05522,
      "grad_norm": 0.7227034463026758,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 5522
    },
    {
      "epoch": 0.05523,
      "grad_norm": 0.6646962930116027,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 5523
    },
    {
      "epoch": 0.05524,
      "grad_norm": 0.6345664894540247,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5524
    },
    {
      "epoch": 0.05525,
      "grad_norm": 0.7628485020641987,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 5525
    },
    {
      "epoch": 0.05526,
      "grad_norm": 0.9521465562602499,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 5526
    },
    {
      "epoch": 0.05527,
      "grad_norm": 0.9654484523561925,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 5527
    },
    {
      "epoch": 0.05528,
      "grad_norm": 0.9467279300817267,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 5528
    },
    {
      "epoch": 0.05529,
      "grad_norm": 0.8441582406490136,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 5529
    },
    {
      "epoch": 0.0553,
      "grad_norm": 0.8409264258057348,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 5530
    },
    {
      "epoch": 0.05531,
      "grad_norm": 0.898368141131664,
      "learning_rate": 0.003,
      "loss": 4.1485,
      "step": 5531
    },
    {
      "epoch": 0.05532,
      "grad_norm": 0.9193518146488663,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 5532
    },
    {
      "epoch": 0.05533,
      "grad_norm": 0.9804221843926488,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 5533
    },
    {
      "epoch": 0.05534,
      "grad_norm": 1.0187659844050676,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 5534
    },
    {
      "epoch": 0.05535,
      "grad_norm": 0.9901354181855488,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 5535
    },
    {
      "epoch": 0.05536,
      "grad_norm": 0.8825476878279837,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 5536
    },
    {
      "epoch": 0.05537,
      "grad_norm": 0.7556391963162472,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 5537
    },
    {
      "epoch": 0.05538,
      "grad_norm": 0.6817528820545807,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 5538
    },
    {
      "epoch": 0.05539,
      "grad_norm": 0.7027437428476654,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 5539
    },
    {
      "epoch": 0.0554,
      "grad_norm": 0.7674376622806265,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 5540
    },
    {
      "epoch": 0.05541,
      "grad_norm": 0.9105163029193267,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 5541
    },
    {
      "epoch": 0.05542,
      "grad_norm": 0.8086875293524934,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 5542
    },
    {
      "epoch": 0.05543,
      "grad_norm": 0.7420293385687333,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 5543
    },
    {
      "epoch": 0.05544,
      "grad_norm": 0.8559082347379338,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 5544
    },
    {
      "epoch": 0.05545,
      "grad_norm": 0.8152608790796877,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 5545
    },
    {
      "epoch": 0.05546,
      "grad_norm": 0.6798845886566196,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 5546
    },
    {
      "epoch": 0.05547,
      "grad_norm": 0.69937059223718,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 5547
    },
    {
      "epoch": 0.05548,
      "grad_norm": 0.6313128061744391,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 5548
    },
    {
      "epoch": 0.05549,
      "grad_norm": 0.6803316037182424,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 5549
    },
    {
      "epoch": 0.0555,
      "grad_norm": 0.8849647916705993,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 5550
    },
    {
      "epoch": 0.05551,
      "grad_norm": 0.9946377104603599,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 5551
    },
    {
      "epoch": 0.05552,
      "grad_norm": 0.9054479922928002,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 5552
    },
    {
      "epoch": 0.05553,
      "grad_norm": 0.8058698994488415,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 5553
    },
    {
      "epoch": 0.05554,
      "grad_norm": 0.8113670227482532,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 5554
    },
    {
      "epoch": 0.05555,
      "grad_norm": 0.8623032272003974,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 5555
    },
    {
      "epoch": 0.05556,
      "grad_norm": 0.7815305295497466,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 5556
    },
    {
      "epoch": 0.05557,
      "grad_norm": 0.7613158957065711,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 5557
    },
    {
      "epoch": 0.05558,
      "grad_norm": 0.6848954615857884,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 5558
    },
    {
      "epoch": 0.05559,
      "grad_norm": 0.6558580775338294,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 5559
    },
    {
      "epoch": 0.0556,
      "grad_norm": 0.6949517459401227,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 5560
    },
    {
      "epoch": 0.05561,
      "grad_norm": 0.6020633152546068,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 5561
    },
    {
      "epoch": 0.05562,
      "grad_norm": 0.6130293848011331,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 5562
    },
    {
      "epoch": 0.05563,
      "grad_norm": 0.5305748367682133,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 5563
    },
    {
      "epoch": 0.05564,
      "grad_norm": 0.5360201658815928,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 5564
    },
    {
      "epoch": 0.05565,
      "grad_norm": 0.5782685165038048,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 5565
    },
    {
      "epoch": 0.05566,
      "grad_norm": 0.7133895920716543,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 5566
    },
    {
      "epoch": 0.05567,
      "grad_norm": 0.826072535494595,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 5567
    },
    {
      "epoch": 0.05568,
      "grad_norm": 0.9865934604781055,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 5568
    },
    {
      "epoch": 0.05569,
      "grad_norm": 1.1577665175633736,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 5569
    },
    {
      "epoch": 0.0557,
      "grad_norm": 0.836822155212909,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 5570
    },
    {
      "epoch": 0.05571,
      "grad_norm": 0.8107309429680243,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 5571
    },
    {
      "epoch": 0.05572,
      "grad_norm": 0.7471424734568298,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 5572
    },
    {
      "epoch": 0.05573,
      "grad_norm": 0.834234761336203,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 5573
    },
    {
      "epoch": 0.05574,
      "grad_norm": 0.9390828719578128,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 5574
    },
    {
      "epoch": 0.05575,
      "grad_norm": 1.0260723102103353,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 5575
    },
    {
      "epoch": 0.05576,
      "grad_norm": 0.8611281671804792,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 5576
    },
    {
      "epoch": 0.05577,
      "grad_norm": 0.7889811692592865,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 5577
    },
    {
      "epoch": 0.05578,
      "grad_norm": 0.8753881201720198,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 5578
    },
    {
      "epoch": 0.05579,
      "grad_norm": 0.8075261839099657,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 5579
    },
    {
      "epoch": 0.0558,
      "grad_norm": 0.6451347078049557,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 5580
    },
    {
      "epoch": 0.05581,
      "grad_norm": 0.5565812999963974,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 5581
    },
    {
      "epoch": 0.05582,
      "grad_norm": 0.5875801341192183,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 5582
    },
    {
      "epoch": 0.05583,
      "grad_norm": 0.6172149180481271,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 5583
    },
    {
      "epoch": 0.05584,
      "grad_norm": 0.6243564462276252,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 5584
    },
    {
      "epoch": 0.05585,
      "grad_norm": 0.7526662295082331,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 5585
    },
    {
      "epoch": 0.05586,
      "grad_norm": 0.926843716234651,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 5586
    },
    {
      "epoch": 0.05587,
      "grad_norm": 1.0068785999700622,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 5587
    },
    {
      "epoch": 0.05588,
      "grad_norm": 0.9183554908331668,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 5588
    },
    {
      "epoch": 0.05589,
      "grad_norm": 0.8399340851508159,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 5589
    },
    {
      "epoch": 0.0559,
      "grad_norm": 0.702522081897405,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 5590
    },
    {
      "epoch": 0.05591,
      "grad_norm": 0.8554841678862133,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 5591
    },
    {
      "epoch": 0.05592,
      "grad_norm": 1.0796189206376294,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 5592
    },
    {
      "epoch": 0.05593,
      "grad_norm": 0.8728376363454068,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 5593
    },
    {
      "epoch": 0.05594,
      "grad_norm": 0.6937202641640116,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 5594
    },
    {
      "epoch": 0.05595,
      "grad_norm": 0.66356619963666,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 5595
    },
    {
      "epoch": 0.05596,
      "grad_norm": 0.7927953508675363,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 5596
    },
    {
      "epoch": 0.05597,
      "grad_norm": 1.0914599310465052,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 5597
    },
    {
      "epoch": 0.05598,
      "grad_norm": 0.979078781931696,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 5598
    },
    {
      "epoch": 0.05599,
      "grad_norm": 0.8737639680425687,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 5599
    },
    {
      "epoch": 0.056,
      "grad_norm": 0.9488588659190235,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 5600
    },
    {
      "epoch": 0.05601,
      "grad_norm": 0.9787141618869671,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 5601
    },
    {
      "epoch": 0.05602,
      "grad_norm": 1.1091138120552808,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 5602
    },
    {
      "epoch": 0.05603,
      "grad_norm": 0.8587130498163756,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5603
    },
    {
      "epoch": 0.05604,
      "grad_norm": 0.7477443911515989,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 5604
    },
    {
      "epoch": 0.05605,
      "grad_norm": 0.6578770652147883,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 5605
    },
    {
      "epoch": 0.05606,
      "grad_norm": 0.6529008726545039,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 5606
    },
    {
      "epoch": 0.05607,
      "grad_norm": 0.6971218018514623,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 5607
    },
    {
      "epoch": 0.05608,
      "grad_norm": 0.8592836275322725,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 5608
    },
    {
      "epoch": 0.05609,
      "grad_norm": 1.0589082640672076,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 5609
    },
    {
      "epoch": 0.0561,
      "grad_norm": 0.8647971161448784,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 5610
    },
    {
      "epoch": 0.05611,
      "grad_norm": 0.7323461767341091,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 5611
    },
    {
      "epoch": 0.05612,
      "grad_norm": 0.7346877413349945,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 5612
    },
    {
      "epoch": 0.05613,
      "grad_norm": 0.7046408743706015,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 5613
    },
    {
      "epoch": 0.05614,
      "grad_norm": 0.7374788617059296,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 5614
    },
    {
      "epoch": 0.05615,
      "grad_norm": 0.8226983375167506,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 5615
    },
    {
      "epoch": 0.05616,
      "grad_norm": 0.8679780386223853,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 5616
    },
    {
      "epoch": 0.05617,
      "grad_norm": 0.9315265116837809,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5617
    },
    {
      "epoch": 0.05618,
      "grad_norm": 0.7604395735189772,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 5618
    },
    {
      "epoch": 0.05619,
      "grad_norm": 0.7485651636552902,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 5619
    },
    {
      "epoch": 0.0562,
      "grad_norm": 0.7181098485002947,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 5620
    },
    {
      "epoch": 0.05621,
      "grad_norm": 0.6644004914528993,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 5621
    },
    {
      "epoch": 0.05622,
      "grad_norm": 0.6212809614577611,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 5622
    },
    {
      "epoch": 0.05623,
      "grad_norm": 0.6362873457411731,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 5623
    },
    {
      "epoch": 0.05624,
      "grad_norm": 0.6095376423594582,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 5624
    },
    {
      "epoch": 0.05625,
      "grad_norm": 0.6012390884980424,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 5625
    },
    {
      "epoch": 0.05626,
      "grad_norm": 0.6373515773309039,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 5626
    },
    {
      "epoch": 0.05627,
      "grad_norm": 0.6997251309707232,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 5627
    },
    {
      "epoch": 0.05628,
      "grad_norm": 0.7740087467448155,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 5628
    },
    {
      "epoch": 0.05629,
      "grad_norm": 0.7635869528942756,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 5629
    },
    {
      "epoch": 0.0563,
      "grad_norm": 0.8230920756703966,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 5630
    },
    {
      "epoch": 0.05631,
      "grad_norm": 1.0656938602588153,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 5631
    },
    {
      "epoch": 0.05632,
      "grad_norm": 1.0708267403591085,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 5632
    },
    {
      "epoch": 0.05633,
      "grad_norm": 0.789050996667616,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 5633
    },
    {
      "epoch": 0.05634,
      "grad_norm": 0.703640516716815,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 5634
    },
    {
      "epoch": 0.05635,
      "grad_norm": 0.8170971379255785,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 5635
    },
    {
      "epoch": 0.05636,
      "grad_norm": 0.7643132742033529,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5636
    },
    {
      "epoch": 0.05637,
      "grad_norm": 0.7269418593201121,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 5637
    },
    {
      "epoch": 0.05638,
      "grad_norm": 0.6425160594147294,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 5638
    },
    {
      "epoch": 0.05639,
      "grad_norm": 0.59424152487318,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 5639
    },
    {
      "epoch": 0.0564,
      "grad_norm": 0.6307390916938302,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 5640
    },
    {
      "epoch": 0.05641,
      "grad_norm": 0.6398464382688712,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 5641
    },
    {
      "epoch": 0.05642,
      "grad_norm": 0.5607522531731973,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 5642
    },
    {
      "epoch": 0.05643,
      "grad_norm": 0.5848917477606379,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 5643
    },
    {
      "epoch": 0.05644,
      "grad_norm": 0.6555662778545387,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 5644
    },
    {
      "epoch": 0.05645,
      "grad_norm": 0.6055581079132784,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 5645
    },
    {
      "epoch": 0.05646,
      "grad_norm": 0.5727498420416256,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 5646
    },
    {
      "epoch": 0.05647,
      "grad_norm": 0.5143080586691178,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 5647
    },
    {
      "epoch": 0.05648,
      "grad_norm": 0.6091773925484103,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 5648
    },
    {
      "epoch": 0.05649,
      "grad_norm": 0.7510364666789875,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 5649
    },
    {
      "epoch": 0.0565,
      "grad_norm": 0.8427546995305852,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 5650
    },
    {
      "epoch": 0.05651,
      "grad_norm": 0.9454344139315918,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 5651
    },
    {
      "epoch": 0.05652,
      "grad_norm": 1.0512411398620864,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 5652
    },
    {
      "epoch": 0.05653,
      "grad_norm": 1.0651426178923196,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 5653
    },
    {
      "epoch": 0.05654,
      "grad_norm": 0.8316486999901311,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 5654
    },
    {
      "epoch": 0.05655,
      "grad_norm": 0.7989690345947827,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 5655
    },
    {
      "epoch": 0.05656,
      "grad_norm": 0.7220583330129677,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 5656
    },
    {
      "epoch": 0.05657,
      "grad_norm": 0.6534481422931128,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 5657
    },
    {
      "epoch": 0.05658,
      "grad_norm": 0.5840017265991315,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 5658
    },
    {
      "epoch": 0.05659,
      "grad_norm": 0.6445129008912936,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 5659
    },
    {
      "epoch": 0.0566,
      "grad_norm": 0.6350047659342725,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 5660
    },
    {
      "epoch": 0.05661,
      "grad_norm": 0.75443344313924,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 5661
    },
    {
      "epoch": 0.05662,
      "grad_norm": 0.8347625040490713,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 5662
    },
    {
      "epoch": 0.05663,
      "grad_norm": 0.7844268329201943,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 5663
    },
    {
      "epoch": 0.05664,
      "grad_norm": 0.6882757704839232,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 5664
    },
    {
      "epoch": 0.05665,
      "grad_norm": 0.7143954784573894,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 5665
    },
    {
      "epoch": 0.05666,
      "grad_norm": 0.7697196574672253,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 5666
    },
    {
      "epoch": 0.05667,
      "grad_norm": 0.6930833704435834,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 5667
    },
    {
      "epoch": 0.05668,
      "grad_norm": 0.7588668210696813,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 5668
    },
    {
      "epoch": 0.05669,
      "grad_norm": 0.7785704484779421,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 5669
    },
    {
      "epoch": 0.0567,
      "grad_norm": 0.7557718596536394,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 5670
    },
    {
      "epoch": 0.05671,
      "grad_norm": 0.7694361102868486,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 5671
    },
    {
      "epoch": 0.05672,
      "grad_norm": 0.8007083587054245,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 5672
    },
    {
      "epoch": 0.05673,
      "grad_norm": 0.8580257925305742,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 5673
    },
    {
      "epoch": 0.05674,
      "grad_norm": 0.9196710283055469,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 5674
    },
    {
      "epoch": 0.05675,
      "grad_norm": 0.9492581195457488,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 5675
    },
    {
      "epoch": 0.05676,
      "grad_norm": 1.0914070056598988,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 5676
    },
    {
      "epoch": 0.05677,
      "grad_norm": 0.8885367967761989,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 5677
    },
    {
      "epoch": 0.05678,
      "grad_norm": 0.7905086473346482,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 5678
    },
    {
      "epoch": 0.05679,
      "grad_norm": 0.6956868554948951,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 5679
    },
    {
      "epoch": 0.0568,
      "grad_norm": 0.7085015911445326,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 5680
    },
    {
      "epoch": 0.05681,
      "grad_norm": 0.7982290524952671,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 5681
    },
    {
      "epoch": 0.05682,
      "grad_norm": 0.8702967583238443,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 5682
    },
    {
      "epoch": 0.05683,
      "grad_norm": 0.8957753899417163,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 5683
    },
    {
      "epoch": 0.05684,
      "grad_norm": 0.7828031111834381,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 5684
    },
    {
      "epoch": 0.05685,
      "grad_norm": 0.7982392827676595,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 5685
    },
    {
      "epoch": 0.05686,
      "grad_norm": 0.7549245911981988,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 5686
    },
    {
      "epoch": 0.05687,
      "grad_norm": 0.7845211375597457,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 5687
    },
    {
      "epoch": 0.05688,
      "grad_norm": 0.8652123527501867,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 5688
    },
    {
      "epoch": 0.05689,
      "grad_norm": 1.1480876620771052,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 5689
    },
    {
      "epoch": 0.0569,
      "grad_norm": 1.0790768924826935,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 5690
    },
    {
      "epoch": 0.05691,
      "grad_norm": 1.015989841166836,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 5691
    },
    {
      "epoch": 0.05692,
      "grad_norm": 0.8594661410624701,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 5692
    },
    {
      "epoch": 0.05693,
      "grad_norm": 0.6762044853377914,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 5693
    },
    {
      "epoch": 0.05694,
      "grad_norm": 0.7821423521203583,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 5694
    },
    {
      "epoch": 0.05695,
      "grad_norm": 0.8134766301370081,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 5695
    },
    {
      "epoch": 0.05696,
      "grad_norm": 0.8476984030864564,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 5696
    },
    {
      "epoch": 0.05697,
      "grad_norm": 0.8589446601517351,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 5697
    },
    {
      "epoch": 0.05698,
      "grad_norm": 0.7927970537237816,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 5698
    },
    {
      "epoch": 0.05699,
      "grad_norm": 0.6663534787165875,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 5699
    },
    {
      "epoch": 0.057,
      "grad_norm": 0.6962826442411336,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 5700
    },
    {
      "epoch": 0.05701,
      "grad_norm": 0.748056579704057,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 5701
    },
    {
      "epoch": 0.05702,
      "grad_norm": 0.7276987471205317,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 5702
    },
    {
      "epoch": 0.05703,
      "grad_norm": 0.7077636085898174,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 5703
    },
    {
      "epoch": 0.05704,
      "grad_norm": 0.9037559825169571,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 5704
    },
    {
      "epoch": 0.05705,
      "grad_norm": 0.8075509768599368,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 5705
    },
    {
      "epoch": 0.05706,
      "grad_norm": 0.7674921460937386,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 5706
    },
    {
      "epoch": 0.05707,
      "grad_norm": 0.7539599082513933,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 5707
    },
    {
      "epoch": 0.05708,
      "grad_norm": 0.6799305657607112,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 5708
    },
    {
      "epoch": 0.05709,
      "grad_norm": 0.5479624493880318,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 5709
    },
    {
      "epoch": 0.0571,
      "grad_norm": 0.5121373459902853,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 5710
    },
    {
      "epoch": 0.05711,
      "grad_norm": 0.7280903597835207,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 5711
    },
    {
      "epoch": 0.05712,
      "grad_norm": 1.0090249543371728,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 5712
    },
    {
      "epoch": 0.05713,
      "grad_norm": 1.1666935879947706,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 5713
    },
    {
      "epoch": 0.05714,
      "grad_norm": 0.6574654065182656,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 5714
    },
    {
      "epoch": 0.05715,
      "grad_norm": 0.6534581937051788,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 5715
    },
    {
      "epoch": 0.05716,
      "grad_norm": 0.7637170840325201,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 5716
    },
    {
      "epoch": 0.05717,
      "grad_norm": 0.80055490534206,
      "learning_rate": 0.003,
      "loss": 4.1548,
      "step": 5717
    },
    {
      "epoch": 0.05718,
      "grad_norm": 0.8922767275182982,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 5718
    },
    {
      "epoch": 0.05719,
      "grad_norm": 0.8933371654984332,
      "learning_rate": 0.003,
      "loss": 4.142,
      "step": 5719
    },
    {
      "epoch": 0.0572,
      "grad_norm": 0.8966300715229761,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 5720
    },
    {
      "epoch": 0.05721,
      "grad_norm": 0.8505332145208457,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 5721
    },
    {
      "epoch": 0.05722,
      "grad_norm": 0.7349175001912129,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 5722
    },
    {
      "epoch": 0.05723,
      "grad_norm": 0.723220376175918,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 5723
    },
    {
      "epoch": 0.05724,
      "grad_norm": 0.680563858030686,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 5724
    },
    {
      "epoch": 0.05725,
      "grad_norm": 0.8088590173981038,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 5725
    },
    {
      "epoch": 0.05726,
      "grad_norm": 0.9418455002457423,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 5726
    },
    {
      "epoch": 0.05727,
      "grad_norm": 1.0373122053831039,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 5727
    },
    {
      "epoch": 0.05728,
      "grad_norm": 0.9012220097719827,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 5728
    },
    {
      "epoch": 0.05729,
      "grad_norm": 0.9082669562608017,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 5729
    },
    {
      "epoch": 0.0573,
      "grad_norm": 0.943356157606432,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 5730
    },
    {
      "epoch": 0.05731,
      "grad_norm": 0.8723864537449718,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 5731
    },
    {
      "epoch": 0.05732,
      "grad_norm": 0.8570219760726053,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 5732
    },
    {
      "epoch": 0.05733,
      "grad_norm": 0.8298818226350395,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 5733
    },
    {
      "epoch": 0.05734,
      "grad_norm": 0.8547854470967996,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 5734
    },
    {
      "epoch": 0.05735,
      "grad_norm": 0.8145784886674767,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 5735
    },
    {
      "epoch": 0.05736,
      "grad_norm": 0.6881823470117722,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 5736
    },
    {
      "epoch": 0.05737,
      "grad_norm": 0.5844418060230477,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 5737
    },
    {
      "epoch": 0.05738,
      "grad_norm": 0.5430304440202582,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 5738
    },
    {
      "epoch": 0.05739,
      "grad_norm": 0.5591474039370316,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 5739
    },
    {
      "epoch": 0.0574,
      "grad_norm": 0.5762447209115041,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 5740
    },
    {
      "epoch": 0.05741,
      "grad_norm": 0.5659495615021296,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 5741
    },
    {
      "epoch": 0.05742,
      "grad_norm": 0.5868859562225888,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 5742
    },
    {
      "epoch": 0.05743,
      "grad_norm": 0.6325935486380904,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 5743
    },
    {
      "epoch": 0.05744,
      "grad_norm": 0.7198789457341803,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 5744
    },
    {
      "epoch": 0.05745,
      "grad_norm": 0.8372703754668815,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 5745
    },
    {
      "epoch": 0.05746,
      "grad_norm": 1.013456290369712,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 5746
    },
    {
      "epoch": 0.05747,
      "grad_norm": 0.9926529931921089,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 5747
    },
    {
      "epoch": 0.05748,
      "grad_norm": 0.8673968423379839,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 5748
    },
    {
      "epoch": 0.05749,
      "grad_norm": 0.8265215218017055,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 5749
    },
    {
      "epoch": 0.0575,
      "grad_norm": 0.9141071050046009,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 5750
    },
    {
      "epoch": 0.05751,
      "grad_norm": 0.8867858497305796,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 5751
    },
    {
      "epoch": 0.05752,
      "grad_norm": 0.8160084253804927,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 5752
    },
    {
      "epoch": 0.05753,
      "grad_norm": 0.7570642462201534,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 5753
    },
    {
      "epoch": 0.05754,
      "grad_norm": 0.6931105461575188,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 5754
    },
    {
      "epoch": 0.05755,
      "grad_norm": 0.7892100308149763,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 5755
    },
    {
      "epoch": 0.05756,
      "grad_norm": 0.7797288548736725,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 5756
    },
    {
      "epoch": 0.05757,
      "grad_norm": 0.9143238759273115,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 5757
    },
    {
      "epoch": 0.05758,
      "grad_norm": 0.9568176868550978,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 5758
    },
    {
      "epoch": 0.05759,
      "grad_norm": 0.9850995699177724,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 5759
    },
    {
      "epoch": 0.0576,
      "grad_norm": 0.9114825113358783,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 5760
    },
    {
      "epoch": 0.05761,
      "grad_norm": 0.8393162691101879,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 5761
    },
    {
      "epoch": 0.05762,
      "grad_norm": 0.8185722596967431,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 5762
    },
    {
      "epoch": 0.05763,
      "grad_norm": 0.7941154884714461,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 5763
    },
    {
      "epoch": 0.05764,
      "grad_norm": 0.8167052006322443,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 5764
    },
    {
      "epoch": 0.05765,
      "grad_norm": 0.9075689951343894,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 5765
    },
    {
      "epoch": 0.05766,
      "grad_norm": 0.9379058663927884,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 5766
    },
    {
      "epoch": 0.05767,
      "grad_norm": 0.980509823532266,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 5767
    },
    {
      "epoch": 0.05768,
      "grad_norm": 1.0177076074132854,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 5768
    },
    {
      "epoch": 0.05769,
      "grad_norm": 1.0429502276363345,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 5769
    },
    {
      "epoch": 0.0577,
      "grad_norm": 1.1100087019496394,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 5770
    },
    {
      "epoch": 0.05771,
      "grad_norm": 0.7223949464608548,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 5771
    },
    {
      "epoch": 0.05772,
      "grad_norm": 0.5382829044206766,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 5772
    },
    {
      "epoch": 0.05773,
      "grad_norm": 0.6406979044556045,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 5773
    },
    {
      "epoch": 0.05774,
      "grad_norm": 0.8097734695290948,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 5774
    },
    {
      "epoch": 0.05775,
      "grad_norm": 1.0604370012894522,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 5775
    },
    {
      "epoch": 0.05776,
      "grad_norm": 1.0011226290122666,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 5776
    },
    {
      "epoch": 0.05777,
      "grad_norm": 0.799664737208272,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 5777
    },
    {
      "epoch": 0.05778,
      "grad_norm": 0.7400439760795068,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 5778
    },
    {
      "epoch": 0.05779,
      "grad_norm": 0.8139200513919773,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 5779
    },
    {
      "epoch": 0.0578,
      "grad_norm": 0.7887623553765343,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 5780
    },
    {
      "epoch": 0.05781,
      "grad_norm": 0.6516840422961846,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 5781
    },
    {
      "epoch": 0.05782,
      "grad_norm": 0.6059160139010061,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 5782
    },
    {
      "epoch": 0.05783,
      "grad_norm": 0.6322968665722302,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 5783
    },
    {
      "epoch": 0.05784,
      "grad_norm": 0.6350140210950438,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 5784
    },
    {
      "epoch": 0.05785,
      "grad_norm": 0.6503751753400261,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 5785
    },
    {
      "epoch": 0.05786,
      "grad_norm": 0.6778764586815985,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 5786
    },
    {
      "epoch": 0.05787,
      "grad_norm": 0.5988251572879624,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 5787
    },
    {
      "epoch": 0.05788,
      "grad_norm": 0.4761288844619093,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 5788
    },
    {
      "epoch": 0.05789,
      "grad_norm": 0.4810545422328699,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 5789
    },
    {
      "epoch": 0.0579,
      "grad_norm": 0.48719290090643896,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5790
    },
    {
      "epoch": 0.05791,
      "grad_norm": 0.6649633480208823,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 5791
    },
    {
      "epoch": 0.05792,
      "grad_norm": 0.836880563895898,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 5792
    },
    {
      "epoch": 0.05793,
      "grad_norm": 0.8850525856438148,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 5793
    },
    {
      "epoch": 0.05794,
      "grad_norm": 0.7567725522310005,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 5794
    },
    {
      "epoch": 0.05795,
      "grad_norm": 0.6015273097107093,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5795
    },
    {
      "epoch": 0.05796,
      "grad_norm": 0.6630617490292271,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 5796
    },
    {
      "epoch": 0.05797,
      "grad_norm": 0.6842122117299039,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 5797
    },
    {
      "epoch": 0.05798,
      "grad_norm": 0.6729171219216635,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 5798
    },
    {
      "epoch": 0.05799,
      "grad_norm": 0.6050205840034845,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 5799
    },
    {
      "epoch": 0.058,
      "grad_norm": 0.5873051359813098,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 5800
    },
    {
      "epoch": 0.05801,
      "grad_norm": 0.6839087744451585,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 5801
    },
    {
      "epoch": 0.05802,
      "grad_norm": 0.7630433000288757,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 5802
    },
    {
      "epoch": 0.05803,
      "grad_norm": 0.8857270500243197,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 5803
    },
    {
      "epoch": 0.05804,
      "grad_norm": 0.8401507038678849,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 5804
    },
    {
      "epoch": 0.05805,
      "grad_norm": 0.7475439434893316,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 5805
    },
    {
      "epoch": 0.05806,
      "grad_norm": 0.7978687405953363,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 5806
    },
    {
      "epoch": 0.05807,
      "grad_norm": 0.8733737077265966,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 5807
    },
    {
      "epoch": 0.05808,
      "grad_norm": 0.7983979140429381,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5808
    },
    {
      "epoch": 0.05809,
      "grad_norm": 0.8908751163199425,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 5809
    },
    {
      "epoch": 0.0581,
      "grad_norm": 0.9552572694017188,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 5810
    },
    {
      "epoch": 0.05811,
      "grad_norm": 0.9248914071476577,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 5811
    },
    {
      "epoch": 0.05812,
      "grad_norm": 0.9817321870382631,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 5812
    },
    {
      "epoch": 0.05813,
      "grad_norm": 1.0001250001988298,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 5813
    },
    {
      "epoch": 0.05814,
      "grad_norm": 0.8827566805540904,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 5814
    },
    {
      "epoch": 0.05815,
      "grad_norm": 0.8029521049860747,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 5815
    },
    {
      "epoch": 0.05816,
      "grad_norm": 0.70939056436584,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 5816
    },
    {
      "epoch": 0.05817,
      "grad_norm": 0.726146469102038,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 5817
    },
    {
      "epoch": 0.05818,
      "grad_norm": 0.6869846356437459,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 5818
    },
    {
      "epoch": 0.05819,
      "grad_norm": 0.7498344281131131,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 5819
    },
    {
      "epoch": 0.0582,
      "grad_norm": 0.7991793787271796,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 5820
    },
    {
      "epoch": 0.05821,
      "grad_norm": 0.8692851524335367,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 5821
    },
    {
      "epoch": 0.05822,
      "grad_norm": 0.9485327101586195,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 5822
    },
    {
      "epoch": 0.05823,
      "grad_norm": 0.8570830392648408,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 5823
    },
    {
      "epoch": 0.05824,
      "grad_norm": 0.8874388461019113,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 5824
    },
    {
      "epoch": 0.05825,
      "grad_norm": 0.7062702547735038,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 5825
    },
    {
      "epoch": 0.05826,
      "grad_norm": 0.6393477795637886,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 5826
    },
    {
      "epoch": 0.05827,
      "grad_norm": 0.8019064704761822,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 5827
    },
    {
      "epoch": 0.05828,
      "grad_norm": 0.935978602016839,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 5828
    },
    {
      "epoch": 0.05829,
      "grad_norm": 1.0118867950089439,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 5829
    },
    {
      "epoch": 0.0583,
      "grad_norm": 0.9012310576760173,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 5830
    },
    {
      "epoch": 0.05831,
      "grad_norm": 0.7921513486784892,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 5831
    },
    {
      "epoch": 0.05832,
      "grad_norm": 0.8500687381471804,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 5832
    },
    {
      "epoch": 0.05833,
      "grad_norm": 0.8901324912360562,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 5833
    },
    {
      "epoch": 0.05834,
      "grad_norm": 0.9308154213152057,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 5834
    },
    {
      "epoch": 0.05835,
      "grad_norm": 0.7729299081011445,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 5835
    },
    {
      "epoch": 0.05836,
      "grad_norm": 0.6500047045753582,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 5836
    },
    {
      "epoch": 0.05837,
      "grad_norm": 0.6365900961231694,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 5837
    },
    {
      "epoch": 0.05838,
      "grad_norm": 0.6835199033150741,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 5838
    },
    {
      "epoch": 0.05839,
      "grad_norm": 0.6793630131258913,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 5839
    },
    {
      "epoch": 0.0584,
      "grad_norm": 0.6942313892219114,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 5840
    },
    {
      "epoch": 0.05841,
      "grad_norm": 0.8257321604395392,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 5841
    },
    {
      "epoch": 0.05842,
      "grad_norm": 0.9932148908544386,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 5842
    },
    {
      "epoch": 0.05843,
      "grad_norm": 1.060226393564342,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 5843
    },
    {
      "epoch": 0.05844,
      "grad_norm": 0.7244723778858251,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 5844
    },
    {
      "epoch": 0.05845,
      "grad_norm": 0.5661816774133244,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 5845
    },
    {
      "epoch": 0.05846,
      "grad_norm": 0.6964991790304752,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 5846
    },
    {
      "epoch": 0.05847,
      "grad_norm": 0.8137963768049039,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5847
    },
    {
      "epoch": 0.05848,
      "grad_norm": 0.9321439032747159,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 5848
    },
    {
      "epoch": 0.05849,
      "grad_norm": 0.9690612770794925,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 5849
    },
    {
      "epoch": 0.0585,
      "grad_norm": 0.8070900144801142,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 5850
    },
    {
      "epoch": 0.05851,
      "grad_norm": 0.7978154280770444,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 5851
    },
    {
      "epoch": 0.05852,
      "grad_norm": 0.9108907816600593,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 5852
    },
    {
      "epoch": 0.05853,
      "grad_norm": 0.8478337317220017,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 5853
    },
    {
      "epoch": 0.05854,
      "grad_norm": 0.8702591663009297,
      "learning_rate": 0.003,
      "loss": 4.1814,
      "step": 5854
    },
    {
      "epoch": 0.05855,
      "grad_norm": 0.7888882543910517,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 5855
    },
    {
      "epoch": 0.05856,
      "grad_norm": 0.7469242966369579,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 5856
    },
    {
      "epoch": 0.05857,
      "grad_norm": 0.706624864851467,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 5857
    },
    {
      "epoch": 0.05858,
      "grad_norm": 0.7270168320884999,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 5858
    },
    {
      "epoch": 0.05859,
      "grad_norm": 0.7529935867004306,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 5859
    },
    {
      "epoch": 0.0586,
      "grad_norm": 0.6444427467245577,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 5860
    },
    {
      "epoch": 0.05861,
      "grad_norm": 0.5606393662770783,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 5861
    },
    {
      "epoch": 0.05862,
      "grad_norm": 0.5657377156785917,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 5862
    },
    {
      "epoch": 0.05863,
      "grad_norm": 0.5760395754129868,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 5863
    },
    {
      "epoch": 0.05864,
      "grad_norm": 0.6345646643457323,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 5864
    },
    {
      "epoch": 0.05865,
      "grad_norm": 0.6894044067934254,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 5865
    },
    {
      "epoch": 0.05866,
      "grad_norm": 0.783353091605477,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 5866
    },
    {
      "epoch": 0.05867,
      "grad_norm": 0.8917647666324704,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 5867
    },
    {
      "epoch": 0.05868,
      "grad_norm": 1.0201499997700771,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 5868
    },
    {
      "epoch": 0.05869,
      "grad_norm": 0.9506659421163607,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 5869
    },
    {
      "epoch": 0.0587,
      "grad_norm": 0.7664369973041759,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 5870
    },
    {
      "epoch": 0.05871,
      "grad_norm": 0.7087081834503488,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 5871
    },
    {
      "epoch": 0.05872,
      "grad_norm": 0.6953730352638628,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 5872
    },
    {
      "epoch": 0.05873,
      "grad_norm": 0.671656467983226,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 5873
    },
    {
      "epoch": 0.05874,
      "grad_norm": 0.6955207948522178,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 5874
    },
    {
      "epoch": 0.05875,
      "grad_norm": 0.6268815574500972,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 5875
    },
    {
      "epoch": 0.05876,
      "grad_norm": 0.6389310541639931,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 5876
    },
    {
      "epoch": 0.05877,
      "grad_norm": 0.6217227109072572,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5877
    },
    {
      "epoch": 0.05878,
      "grad_norm": 0.7692902630082031,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 5878
    },
    {
      "epoch": 0.05879,
      "grad_norm": 0.7440526157383079,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 5879
    },
    {
      "epoch": 0.0588,
      "grad_norm": 0.8420901226461514,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 5880
    },
    {
      "epoch": 0.05881,
      "grad_norm": 0.781449431655522,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 5881
    },
    {
      "epoch": 0.05882,
      "grad_norm": 0.7627663058529272,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 5882
    },
    {
      "epoch": 0.05883,
      "grad_norm": 0.8235508005883959,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 5883
    },
    {
      "epoch": 0.05884,
      "grad_norm": 0.9282232426526248,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 5884
    },
    {
      "epoch": 0.05885,
      "grad_norm": 0.8846373141454603,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 5885
    },
    {
      "epoch": 0.05886,
      "grad_norm": 1.1084709963523778,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 5886
    },
    {
      "epoch": 0.05887,
      "grad_norm": 0.9597419334563931,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 5887
    },
    {
      "epoch": 0.05888,
      "grad_norm": 1.1484270558519247,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 5888
    },
    {
      "epoch": 0.05889,
      "grad_norm": 0.7934646401717872,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 5889
    },
    {
      "epoch": 0.0589,
      "grad_norm": 0.7136730779704626,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 5890
    },
    {
      "epoch": 0.05891,
      "grad_norm": 0.725739807105299,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 5891
    },
    {
      "epoch": 0.05892,
      "grad_norm": 0.7256800823290586,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 5892
    },
    {
      "epoch": 0.05893,
      "grad_norm": 0.7692017239614891,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 5893
    },
    {
      "epoch": 0.05894,
      "grad_norm": 0.7387008350548316,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 5894
    },
    {
      "epoch": 0.05895,
      "grad_norm": 0.6999771063166865,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 5895
    },
    {
      "epoch": 0.05896,
      "grad_norm": 0.6586617304026913,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 5896
    },
    {
      "epoch": 0.05897,
      "grad_norm": 0.7731203514227012,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 5897
    },
    {
      "epoch": 0.05898,
      "grad_norm": 0.8305320510139119,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 5898
    },
    {
      "epoch": 0.05899,
      "grad_norm": 0.7742225479585695,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 5899
    },
    {
      "epoch": 0.059,
      "grad_norm": 0.9048055534854499,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 5900
    },
    {
      "epoch": 0.05901,
      "grad_norm": 0.9868685055861205,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 5901
    },
    {
      "epoch": 0.05902,
      "grad_norm": 1.0603964019326821,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 5902
    },
    {
      "epoch": 0.05903,
      "grad_norm": 1.0745850428190593,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 5903
    },
    {
      "epoch": 0.05904,
      "grad_norm": 0.8629637800138508,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 5904
    },
    {
      "epoch": 0.05905,
      "grad_norm": 0.746934964755465,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 5905
    },
    {
      "epoch": 0.05906,
      "grad_norm": 0.7381867732393038,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 5906
    },
    {
      "epoch": 0.05907,
      "grad_norm": 0.8572362096896168,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 5907
    },
    {
      "epoch": 0.05908,
      "grad_norm": 1.0322237476040848,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 5908
    },
    {
      "epoch": 0.05909,
      "grad_norm": 0.8902648134987068,
      "learning_rate": 0.003,
      "loss": 4.1457,
      "step": 5909
    },
    {
      "epoch": 0.0591,
      "grad_norm": 0.773959915980087,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 5910
    },
    {
      "epoch": 0.05911,
      "grad_norm": 0.8192300954731131,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 5911
    },
    {
      "epoch": 0.05912,
      "grad_norm": 0.8546496512225619,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 5912
    },
    {
      "epoch": 0.05913,
      "grad_norm": 0.7554616555250914,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 5913
    },
    {
      "epoch": 0.05914,
      "grad_norm": 0.7048163682769332,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 5914
    },
    {
      "epoch": 0.05915,
      "grad_norm": 0.7670702397330968,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 5915
    },
    {
      "epoch": 0.05916,
      "grad_norm": 0.7678316112541398,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 5916
    },
    {
      "epoch": 0.05917,
      "grad_norm": 0.6726151262347494,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 5917
    },
    {
      "epoch": 0.05918,
      "grad_norm": 0.6440357536146859,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 5918
    },
    {
      "epoch": 0.05919,
      "grad_norm": 0.6990263177794974,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 5919
    },
    {
      "epoch": 0.0592,
      "grad_norm": 0.7179007290375548,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 5920
    },
    {
      "epoch": 0.05921,
      "grad_norm": 0.7012296689409627,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 5921
    },
    {
      "epoch": 0.05922,
      "grad_norm": 0.7174027660375633,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 5922
    },
    {
      "epoch": 0.05923,
      "grad_norm": 0.7806176175934582,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 5923
    },
    {
      "epoch": 0.05924,
      "grad_norm": 0.8832244353079346,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 5924
    },
    {
      "epoch": 0.05925,
      "grad_norm": 0.8531781306643916,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 5925
    },
    {
      "epoch": 0.05926,
      "grad_norm": 0.6880234314584872,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 5926
    },
    {
      "epoch": 0.05927,
      "grad_norm": 0.5242997783745522,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 5927
    },
    {
      "epoch": 0.05928,
      "grad_norm": 0.7118263988410835,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 5928
    },
    {
      "epoch": 0.05929,
      "grad_norm": 0.9051922436752257,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 5929
    },
    {
      "epoch": 0.0593,
      "grad_norm": 0.9625875611083993,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 5930
    },
    {
      "epoch": 0.05931,
      "grad_norm": 0.8328206700541043,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 5931
    },
    {
      "epoch": 0.05932,
      "grad_norm": 0.6522428602922113,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 5932
    },
    {
      "epoch": 0.05933,
      "grad_norm": 0.8202258652975196,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 5933
    },
    {
      "epoch": 0.05934,
      "grad_norm": 0.945676463256145,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 5934
    },
    {
      "epoch": 0.05935,
      "grad_norm": 0.9491618675679084,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 5935
    },
    {
      "epoch": 0.05936,
      "grad_norm": 0.7346756788314402,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 5936
    },
    {
      "epoch": 0.05937,
      "grad_norm": 0.6849843625907307,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 5937
    },
    {
      "epoch": 0.05938,
      "grad_norm": 0.842493843061451,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 5938
    },
    {
      "epoch": 0.05939,
      "grad_norm": 0.9802131731998531,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 5939
    },
    {
      "epoch": 0.0594,
      "grad_norm": 1.0740955279537077,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 5940
    },
    {
      "epoch": 0.05941,
      "grad_norm": 0.80664432417099,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 5941
    },
    {
      "epoch": 0.05942,
      "grad_norm": 0.6673463401001458,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 5942
    },
    {
      "epoch": 0.05943,
      "grad_norm": 0.7136729446036693,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 5943
    },
    {
      "epoch": 0.05944,
      "grad_norm": 0.7748772994265909,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 5944
    },
    {
      "epoch": 0.05945,
      "grad_norm": 0.8319748600017105,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 5945
    },
    {
      "epoch": 0.05946,
      "grad_norm": 0.7912457127582416,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 5946
    },
    {
      "epoch": 0.05947,
      "grad_norm": 0.7794353450592082,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 5947
    },
    {
      "epoch": 0.05948,
      "grad_norm": 0.8118650335500223,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 5948
    },
    {
      "epoch": 0.05949,
      "grad_norm": 0.8550107923144452,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 5949
    },
    {
      "epoch": 0.0595,
      "grad_norm": 0.8915593726429895,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 5950
    },
    {
      "epoch": 0.05951,
      "grad_norm": 0.8945628866915003,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 5951
    },
    {
      "epoch": 0.05952,
      "grad_norm": 1.0089347086265024,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 5952
    },
    {
      "epoch": 0.05953,
      "grad_norm": 1.0649366317902482,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 5953
    },
    {
      "epoch": 0.05954,
      "grad_norm": 1.0265306010895543,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 5954
    },
    {
      "epoch": 0.05955,
      "grad_norm": 1.0053891091737848,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 5955
    },
    {
      "epoch": 0.05956,
      "grad_norm": 0.9418944733574852,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 5956
    },
    {
      "epoch": 0.05957,
      "grad_norm": 0.8074037334150802,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 5957
    },
    {
      "epoch": 0.05958,
      "grad_norm": 0.835853516513737,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 5958
    },
    {
      "epoch": 0.05959,
      "grad_norm": 0.9524075368140398,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 5959
    },
    {
      "epoch": 0.0596,
      "grad_norm": 0.9443691821790957,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 5960
    },
    {
      "epoch": 0.05961,
      "grad_norm": 0.8502001962303277,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 5961
    },
    {
      "epoch": 0.05962,
      "grad_norm": 0.7110403654732431,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 5962
    },
    {
      "epoch": 0.05963,
      "grad_norm": 0.6983877872125744,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 5963
    },
    {
      "epoch": 0.05964,
      "grad_norm": 0.6706508091828537,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 5964
    },
    {
      "epoch": 0.05965,
      "grad_norm": 0.5516979364941672,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 5965
    },
    {
      "epoch": 0.05966,
      "grad_norm": 0.5599182409877457,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 5966
    },
    {
      "epoch": 0.05967,
      "grad_norm": 0.5140228003501242,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 5967
    },
    {
      "epoch": 0.05968,
      "grad_norm": 0.5043516900053563,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 5968
    },
    {
      "epoch": 0.05969,
      "grad_norm": 0.4781424578723309,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 5969
    },
    {
      "epoch": 0.0597,
      "grad_norm": 0.49499059228808473,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 5970
    },
    {
      "epoch": 0.05971,
      "grad_norm": 0.5302688602898687,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 5971
    },
    {
      "epoch": 0.05972,
      "grad_norm": 0.5628921179090366,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 5972
    },
    {
      "epoch": 0.05973,
      "grad_norm": 0.6729538698391679,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 5973
    },
    {
      "epoch": 0.05974,
      "grad_norm": 0.8325991069119943,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 5974
    },
    {
      "epoch": 0.05975,
      "grad_norm": 1.0067237957827262,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 5975
    },
    {
      "epoch": 0.05976,
      "grad_norm": 1.2218569863697355,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 5976
    },
    {
      "epoch": 0.05977,
      "grad_norm": 0.6120282975916178,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 5977
    },
    {
      "epoch": 0.05978,
      "grad_norm": 0.634934658722155,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 5978
    },
    {
      "epoch": 0.05979,
      "grad_norm": 0.795528878435463,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 5979
    },
    {
      "epoch": 0.0598,
      "grad_norm": 0.7850331087825336,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 5980
    },
    {
      "epoch": 0.05981,
      "grad_norm": 0.7268016197563338,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 5981
    },
    {
      "epoch": 0.05982,
      "grad_norm": 0.6606189209405257,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 5982
    },
    {
      "epoch": 0.05983,
      "grad_norm": 0.7268589111503444,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 5983
    },
    {
      "epoch": 0.05984,
      "grad_norm": 0.7708148438016759,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 5984
    },
    {
      "epoch": 0.05985,
      "grad_norm": 0.7846506991954134,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 5985
    },
    {
      "epoch": 0.05986,
      "grad_norm": 0.7546183129518631,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 5986
    },
    {
      "epoch": 0.05987,
      "grad_norm": 0.8496680958314422,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 5987
    },
    {
      "epoch": 0.05988,
      "grad_norm": 0.9751701746085983,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 5988
    },
    {
      "epoch": 0.05989,
      "grad_norm": 1.0265361397551773,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 5989
    },
    {
      "epoch": 0.0599,
      "grad_norm": 1.032703400806406,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 5990
    },
    {
      "epoch": 0.05991,
      "grad_norm": 0.9301589844396891,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 5991
    },
    {
      "epoch": 0.05992,
      "grad_norm": 0.8529001190170848,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 5992
    },
    {
      "epoch": 0.05993,
      "grad_norm": 0.7232314574469758,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 5993
    },
    {
      "epoch": 0.05994,
      "grad_norm": 0.6055257270988292,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 5994
    },
    {
      "epoch": 0.05995,
      "grad_norm": 0.6461271491413968,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 5995
    },
    {
      "epoch": 0.05996,
      "grad_norm": 0.6721002555843104,
      "learning_rate": 0.003,
      "loss": 4.1485,
      "step": 5996
    },
    {
      "epoch": 0.05997,
      "grad_norm": 0.7240739565812643,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 5997
    },
    {
      "epoch": 0.05998,
      "grad_norm": 0.6950086390271106,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 5998
    },
    {
      "epoch": 0.05999,
      "grad_norm": 0.750185481402966,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 5999
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8158906083982681,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 6000
    },
    {
      "epoch": 0.06001,
      "grad_norm": 0.7786341857744299,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 6001
    },
    {
      "epoch": 0.06002,
      "grad_norm": 0.6362829691059452,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6002
    },
    {
      "epoch": 0.06003,
      "grad_norm": 0.6439798777608323,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 6003
    },
    {
      "epoch": 0.06004,
      "grad_norm": 0.7447283430547924,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 6004
    },
    {
      "epoch": 0.06005,
      "grad_norm": 0.8074281820106757,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 6005
    },
    {
      "epoch": 0.06006,
      "grad_norm": 0.8310801813091088,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 6006
    },
    {
      "epoch": 0.06007,
      "grad_norm": 0.739959036143615,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 6007
    },
    {
      "epoch": 0.06008,
      "grad_norm": 0.7512598056369182,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 6008
    },
    {
      "epoch": 0.06009,
      "grad_norm": 0.8453751856138448,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 6009
    },
    {
      "epoch": 0.0601,
      "grad_norm": 0.9834836783342701,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 6010
    },
    {
      "epoch": 0.06011,
      "grad_norm": 1.2294686050706798,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 6011
    },
    {
      "epoch": 0.06012,
      "grad_norm": 0.6713324153136399,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 6012
    },
    {
      "epoch": 0.06013,
      "grad_norm": 0.7264483540409177,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 6013
    },
    {
      "epoch": 0.06014,
      "grad_norm": 0.8963286802156893,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 6014
    },
    {
      "epoch": 0.06015,
      "grad_norm": 1.0516872025977433,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 6015
    },
    {
      "epoch": 0.06016,
      "grad_norm": 0.9867378972073549,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 6016
    },
    {
      "epoch": 0.06017,
      "grad_norm": 0.8189207241122054,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 6017
    },
    {
      "epoch": 0.06018,
      "grad_norm": 0.6248389222598227,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 6018
    },
    {
      "epoch": 0.06019,
      "grad_norm": 0.6382727015073211,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 6019
    },
    {
      "epoch": 0.0602,
      "grad_norm": 0.5942681644648503,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 6020
    },
    {
      "epoch": 0.06021,
      "grad_norm": 0.6307301681376707,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 6021
    },
    {
      "epoch": 0.06022,
      "grad_norm": 0.6750537233371731,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 6022
    },
    {
      "epoch": 0.06023,
      "grad_norm": 0.7128799339128652,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 6023
    },
    {
      "epoch": 0.06024,
      "grad_norm": 0.7520500789879113,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6024
    },
    {
      "epoch": 0.06025,
      "grad_norm": 0.7805119674993733,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 6025
    },
    {
      "epoch": 0.06026,
      "grad_norm": 0.7499035678190326,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 6026
    },
    {
      "epoch": 0.06027,
      "grad_norm": 0.7664750877261561,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 6027
    },
    {
      "epoch": 0.06028,
      "grad_norm": 0.7776094623285253,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 6028
    },
    {
      "epoch": 0.06029,
      "grad_norm": 0.8761047706417141,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 6029
    },
    {
      "epoch": 0.0603,
      "grad_norm": 0.9104716332204089,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 6030
    },
    {
      "epoch": 0.06031,
      "grad_norm": 0.8922902846451947,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 6031
    },
    {
      "epoch": 0.06032,
      "grad_norm": 0.9770316982513426,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 6032
    },
    {
      "epoch": 0.06033,
      "grad_norm": 0.9457538473651317,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 6033
    },
    {
      "epoch": 0.06034,
      "grad_norm": 0.9217776085904755,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 6034
    },
    {
      "epoch": 0.06035,
      "grad_norm": 0.9280484794744475,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 6035
    },
    {
      "epoch": 0.06036,
      "grad_norm": 1.0009116543043906,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 6036
    },
    {
      "epoch": 0.06037,
      "grad_norm": 1.100896351699259,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 6037
    },
    {
      "epoch": 0.06038,
      "grad_norm": 0.751886944643684,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 6038
    },
    {
      "epoch": 0.06039,
      "grad_norm": 0.6560741836996629,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 6039
    },
    {
      "epoch": 0.0604,
      "grad_norm": 0.6307042035546973,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 6040
    },
    {
      "epoch": 0.06041,
      "grad_norm": 0.6583077177213943,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 6041
    },
    {
      "epoch": 0.06042,
      "grad_norm": 0.5980033750895085,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 6042
    },
    {
      "epoch": 0.06043,
      "grad_norm": 0.6032908893943834,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 6043
    },
    {
      "epoch": 0.06044,
      "grad_norm": 0.6209590746215262,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 6044
    },
    {
      "epoch": 0.06045,
      "grad_norm": 0.6136074895659309,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 6045
    },
    {
      "epoch": 0.06046,
      "grad_norm": 0.6915063121187296,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6046
    },
    {
      "epoch": 0.06047,
      "grad_norm": 0.7812065657084274,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 6047
    },
    {
      "epoch": 0.06048,
      "grad_norm": 0.784591842396142,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 6048
    },
    {
      "epoch": 0.06049,
      "grad_norm": 0.856638395736723,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6049
    },
    {
      "epoch": 0.0605,
      "grad_norm": 1.1075444631976008,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 6050
    },
    {
      "epoch": 0.06051,
      "grad_norm": 1.0022518924058987,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 6051
    },
    {
      "epoch": 0.06052,
      "grad_norm": 0.9359706571597523,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 6052
    },
    {
      "epoch": 0.06053,
      "grad_norm": 0.8709162398571225,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 6053
    },
    {
      "epoch": 0.06054,
      "grad_norm": 0.9149227979278415,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 6054
    },
    {
      "epoch": 0.06055,
      "grad_norm": 1.0346925153764615,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 6055
    },
    {
      "epoch": 0.06056,
      "grad_norm": 0.9282475689671914,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 6056
    },
    {
      "epoch": 0.06057,
      "grad_norm": 0.9497057705026418,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 6057
    },
    {
      "epoch": 0.06058,
      "grad_norm": 1.003018583474644,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 6058
    },
    {
      "epoch": 0.06059,
      "grad_norm": 0.9574516546287778,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 6059
    },
    {
      "epoch": 0.0606,
      "grad_norm": 0.9267456613771629,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 6060
    },
    {
      "epoch": 0.06061,
      "grad_norm": 1.0318438116932864,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 6061
    },
    {
      "epoch": 0.06062,
      "grad_norm": 1.0400522917116977,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 6062
    },
    {
      "epoch": 0.06063,
      "grad_norm": 1.0497641878303774,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 6063
    },
    {
      "epoch": 0.06064,
      "grad_norm": 1.0662498014953086,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 6064
    },
    {
      "epoch": 0.06065,
      "grad_norm": 0.8198498372995232,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 6065
    },
    {
      "epoch": 0.06066,
      "grad_norm": 0.7109049478395263,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 6066
    },
    {
      "epoch": 0.06067,
      "grad_norm": 0.6804945226918797,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 6067
    },
    {
      "epoch": 0.06068,
      "grad_norm": 0.6372476248357833,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 6068
    },
    {
      "epoch": 0.06069,
      "grad_norm": 0.5605214633243844,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 6069
    },
    {
      "epoch": 0.0607,
      "grad_norm": 0.5452568213138966,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 6070
    },
    {
      "epoch": 0.06071,
      "grad_norm": 0.6180224930197478,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 6071
    },
    {
      "epoch": 0.06072,
      "grad_norm": 0.7813523705166946,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 6072
    },
    {
      "epoch": 0.06073,
      "grad_norm": 0.9736890155608084,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 6073
    },
    {
      "epoch": 0.06074,
      "grad_norm": 1.0327347386785892,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 6074
    },
    {
      "epoch": 0.06075,
      "grad_norm": 0.7472307019465172,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 6075
    },
    {
      "epoch": 0.06076,
      "grad_norm": 0.6670271053470518,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 6076
    },
    {
      "epoch": 0.06077,
      "grad_norm": 0.7254836202543244,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 6077
    },
    {
      "epoch": 0.06078,
      "grad_norm": 0.6208465711650167,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 6078
    },
    {
      "epoch": 0.06079,
      "grad_norm": 0.7113936555649242,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 6079
    },
    {
      "epoch": 0.0608,
      "grad_norm": 0.689258626457068,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 6080
    },
    {
      "epoch": 0.06081,
      "grad_norm": 0.6551455177926783,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 6081
    },
    {
      "epoch": 0.06082,
      "grad_norm": 0.6542633333810457,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 6082
    },
    {
      "epoch": 0.06083,
      "grad_norm": 0.6010838532362384,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 6083
    },
    {
      "epoch": 0.06084,
      "grad_norm": 0.5910563656180051,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 6084
    },
    {
      "epoch": 0.06085,
      "grad_norm": 0.6028680880478657,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 6085
    },
    {
      "epoch": 0.06086,
      "grad_norm": 0.5204888285801396,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 6086
    },
    {
      "epoch": 0.06087,
      "grad_norm": 0.5900462834410196,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 6087
    },
    {
      "epoch": 0.06088,
      "grad_norm": 0.6959450877948858,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 6088
    },
    {
      "epoch": 0.06089,
      "grad_norm": 0.6784516411313329,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 6089
    },
    {
      "epoch": 0.0609,
      "grad_norm": 0.7035758936099877,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 6090
    },
    {
      "epoch": 0.06091,
      "grad_norm": 0.8330650414758795,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 6091
    },
    {
      "epoch": 0.06092,
      "grad_norm": 0.8930098103439321,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 6092
    },
    {
      "epoch": 0.06093,
      "grad_norm": 1.0224308602932068,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 6093
    },
    {
      "epoch": 0.06094,
      "grad_norm": 0.9894448173843531,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 6094
    },
    {
      "epoch": 0.06095,
      "grad_norm": 0.8810812235713495,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 6095
    },
    {
      "epoch": 0.06096,
      "grad_norm": 0.6354430317367544,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 6096
    },
    {
      "epoch": 0.06097,
      "grad_norm": 0.6376701446414055,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 6097
    },
    {
      "epoch": 0.06098,
      "grad_norm": 0.8448719370273449,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 6098
    },
    {
      "epoch": 0.06099,
      "grad_norm": 0.969987418580439,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6099
    },
    {
      "epoch": 0.061,
      "grad_norm": 1.0255182095149356,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 6100
    },
    {
      "epoch": 0.06101,
      "grad_norm": 0.8139563380818391,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 6101
    },
    {
      "epoch": 0.06102,
      "grad_norm": 0.6862499005656783,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 6102
    },
    {
      "epoch": 0.06103,
      "grad_norm": 0.7284494857382159,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 6103
    },
    {
      "epoch": 0.06104,
      "grad_norm": 0.7385823683842133,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 6104
    },
    {
      "epoch": 0.06105,
      "grad_norm": 0.7289519050865993,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 6105
    },
    {
      "epoch": 0.06106,
      "grad_norm": 0.7418004159768921,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 6106
    },
    {
      "epoch": 0.06107,
      "grad_norm": 0.683000547263818,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 6107
    },
    {
      "epoch": 0.06108,
      "grad_norm": 0.5960849690912845,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 6108
    },
    {
      "epoch": 0.06109,
      "grad_norm": 0.6202810402417694,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 6109
    },
    {
      "epoch": 0.0611,
      "grad_norm": 0.7084749148334659,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 6110
    },
    {
      "epoch": 0.06111,
      "grad_norm": 0.8416354572669955,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6111
    },
    {
      "epoch": 0.06112,
      "grad_norm": 1.1477396706080005,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 6112
    },
    {
      "epoch": 0.06113,
      "grad_norm": 0.9298557660649742,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 6113
    },
    {
      "epoch": 0.06114,
      "grad_norm": 0.8426576582561913,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 6114
    },
    {
      "epoch": 0.06115,
      "grad_norm": 0.8345003178868361,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 6115
    },
    {
      "epoch": 0.06116,
      "grad_norm": 0.8891402025914743,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 6116
    },
    {
      "epoch": 0.06117,
      "grad_norm": 0.8733942465346064,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 6117
    },
    {
      "epoch": 0.06118,
      "grad_norm": 0.8300392838566886,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 6118
    },
    {
      "epoch": 0.06119,
      "grad_norm": 1.0718155087870083,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 6119
    },
    {
      "epoch": 0.0612,
      "grad_norm": 1.082038424508775,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 6120
    },
    {
      "epoch": 0.06121,
      "grad_norm": 1.0039524800797741,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 6121
    },
    {
      "epoch": 0.06122,
      "grad_norm": 0.858861812903392,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 6122
    },
    {
      "epoch": 0.06123,
      "grad_norm": 0.7517337557672539,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 6123
    },
    {
      "epoch": 0.06124,
      "grad_norm": 0.8054070621318578,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 6124
    },
    {
      "epoch": 0.06125,
      "grad_norm": 0.8798723198485504,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 6125
    },
    {
      "epoch": 0.06126,
      "grad_norm": 0.9746332514781364,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 6126
    },
    {
      "epoch": 0.06127,
      "grad_norm": 1.0519385433383126,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6127
    },
    {
      "epoch": 0.06128,
      "grad_norm": 0.7862904704912201,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 6128
    },
    {
      "epoch": 0.06129,
      "grad_norm": 0.8108064027519418,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 6129
    },
    {
      "epoch": 0.0613,
      "grad_norm": 0.7803383051940936,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 6130
    },
    {
      "epoch": 0.06131,
      "grad_norm": 0.807939857772501,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 6131
    },
    {
      "epoch": 0.06132,
      "grad_norm": 0.9270834066209759,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 6132
    },
    {
      "epoch": 0.06133,
      "grad_norm": 1.057590287106868,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 6133
    },
    {
      "epoch": 0.06134,
      "grad_norm": 0.9315718021788794,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 6134
    },
    {
      "epoch": 0.06135,
      "grad_norm": 0.7682639425500988,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 6135
    },
    {
      "epoch": 0.06136,
      "grad_norm": 0.6453509034899395,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 6136
    },
    {
      "epoch": 0.06137,
      "grad_norm": 0.5089285133051348,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 6137
    },
    {
      "epoch": 0.06138,
      "grad_norm": 0.5921446681322661,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 6138
    },
    {
      "epoch": 0.06139,
      "grad_norm": 0.6572428733185919,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 6139
    },
    {
      "epoch": 0.0614,
      "grad_norm": 0.7594320601513859,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 6140
    },
    {
      "epoch": 0.06141,
      "grad_norm": 0.7701960965013912,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 6141
    },
    {
      "epoch": 0.06142,
      "grad_norm": 0.708324006436222,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 6142
    },
    {
      "epoch": 0.06143,
      "grad_norm": 0.8072574094920275,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 6143
    },
    {
      "epoch": 0.06144,
      "grad_norm": 0.9407988342872765,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 6144
    },
    {
      "epoch": 0.06145,
      "grad_norm": 0.9922781126952929,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 6145
    },
    {
      "epoch": 0.06146,
      "grad_norm": 0.9137503629646905,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 6146
    },
    {
      "epoch": 0.06147,
      "grad_norm": 0.8208946959727162,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 6147
    },
    {
      "epoch": 0.06148,
      "grad_norm": 0.7535313704885915,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 6148
    },
    {
      "epoch": 0.06149,
      "grad_norm": 0.7754246480784573,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 6149
    },
    {
      "epoch": 0.0615,
      "grad_norm": 0.8086681823805969,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 6150
    },
    {
      "epoch": 0.06151,
      "grad_norm": 0.7520491914732798,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 6151
    },
    {
      "epoch": 0.06152,
      "grad_norm": 0.7385892490900827,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 6152
    },
    {
      "epoch": 0.06153,
      "grad_norm": 0.585306853249163,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 6153
    },
    {
      "epoch": 0.06154,
      "grad_norm": 0.6075200706981868,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 6154
    },
    {
      "epoch": 0.06155,
      "grad_norm": 0.7008926353906986,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 6155
    },
    {
      "epoch": 0.06156,
      "grad_norm": 0.7377701370473413,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 6156
    },
    {
      "epoch": 0.06157,
      "grad_norm": 0.6372062483340378,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 6157
    },
    {
      "epoch": 0.06158,
      "grad_norm": 0.5711395334899737,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 6158
    },
    {
      "epoch": 0.06159,
      "grad_norm": 0.6974553713335565,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 6159
    },
    {
      "epoch": 0.0616,
      "grad_norm": 0.7914523454155186,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 6160
    },
    {
      "epoch": 0.06161,
      "grad_norm": 0.9147836100156929,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 6161
    },
    {
      "epoch": 0.06162,
      "grad_norm": 1.0202822458279153,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 6162
    },
    {
      "epoch": 0.06163,
      "grad_norm": 0.8492609261522512,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 6163
    },
    {
      "epoch": 0.06164,
      "grad_norm": 0.6305588102799871,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 6164
    },
    {
      "epoch": 0.06165,
      "grad_norm": 0.70222905462584,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 6165
    },
    {
      "epoch": 0.06166,
      "grad_norm": 0.7330174697865238,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 6166
    },
    {
      "epoch": 0.06167,
      "grad_norm": 0.8120770794656365,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 6167
    },
    {
      "epoch": 0.06168,
      "grad_norm": 0.9461948541503907,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6168
    },
    {
      "epoch": 0.06169,
      "grad_norm": 0.8911742935865481,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 6169
    },
    {
      "epoch": 0.0617,
      "grad_norm": 0.7323433640480617,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 6170
    },
    {
      "epoch": 0.06171,
      "grad_norm": 0.6208020453162558,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 6171
    },
    {
      "epoch": 0.06172,
      "grad_norm": 0.7291939143005758,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 6172
    },
    {
      "epoch": 0.06173,
      "grad_norm": 0.7913577473386937,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 6173
    },
    {
      "epoch": 0.06174,
      "grad_norm": 0.9778156595493965,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 6174
    },
    {
      "epoch": 0.06175,
      "grad_norm": 1.1932141855839438,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 6175
    },
    {
      "epoch": 0.06176,
      "grad_norm": 0.7326715787881036,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 6176
    },
    {
      "epoch": 0.06177,
      "grad_norm": 0.7923716099483298,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 6177
    },
    {
      "epoch": 0.06178,
      "grad_norm": 0.9764360789021225,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 6178
    },
    {
      "epoch": 0.06179,
      "grad_norm": 1.0887682373744165,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 6179
    },
    {
      "epoch": 0.0618,
      "grad_norm": 0.8700014538602673,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 6180
    },
    {
      "epoch": 0.06181,
      "grad_norm": 0.8463904835451614,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 6181
    },
    {
      "epoch": 0.06182,
      "grad_norm": 0.8112163844886693,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 6182
    },
    {
      "epoch": 0.06183,
      "grad_norm": 0.8784890732030473,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 6183
    },
    {
      "epoch": 0.06184,
      "grad_norm": 0.8158970894757084,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 6184
    },
    {
      "epoch": 0.06185,
      "grad_norm": 0.8186525569563441,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 6185
    },
    {
      "epoch": 0.06186,
      "grad_norm": 0.8640122378248711,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 6186
    },
    {
      "epoch": 0.06187,
      "grad_norm": 0.9198602507085508,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 6187
    },
    {
      "epoch": 0.06188,
      "grad_norm": 0.9582358948075637,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 6188
    },
    {
      "epoch": 0.06189,
      "grad_norm": 0.9436060131627309,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 6189
    },
    {
      "epoch": 0.0619,
      "grad_norm": 0.8106004278533174,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 6190
    },
    {
      "epoch": 0.06191,
      "grad_norm": 0.7377926481554797,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 6191
    },
    {
      "epoch": 0.06192,
      "grad_norm": 0.7292550893414801,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 6192
    },
    {
      "epoch": 0.06193,
      "grad_norm": 0.796675035271722,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 6193
    },
    {
      "epoch": 0.06194,
      "grad_norm": 0.9834328051273128,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 6194
    },
    {
      "epoch": 0.06195,
      "grad_norm": 0.9674422632284655,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 6195
    },
    {
      "epoch": 0.06196,
      "grad_norm": 0.9127448537843487,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 6196
    },
    {
      "epoch": 0.06197,
      "grad_norm": 0.8018597016816854,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 6197
    },
    {
      "epoch": 0.06198,
      "grad_norm": 0.6492553167687579,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 6198
    },
    {
      "epoch": 0.06199,
      "grad_norm": 0.500075335938287,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 6199
    },
    {
      "epoch": 0.062,
      "grad_norm": 0.6408031097259872,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 6200
    },
    {
      "epoch": 0.06201,
      "grad_norm": 0.6882374713932283,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 6201
    },
    {
      "epoch": 0.06202,
      "grad_norm": 0.6650459048800811,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 6202
    },
    {
      "epoch": 0.06203,
      "grad_norm": 0.7279713021245527,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 6203
    },
    {
      "epoch": 0.06204,
      "grad_norm": 0.8176381462944218,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 6204
    },
    {
      "epoch": 0.06205,
      "grad_norm": 0.8635078911095838,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 6205
    },
    {
      "epoch": 0.06206,
      "grad_norm": 0.9051460085260756,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 6206
    },
    {
      "epoch": 0.06207,
      "grad_norm": 0.9683554461781614,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 6207
    },
    {
      "epoch": 0.06208,
      "grad_norm": 0.9491242929275403,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 6208
    },
    {
      "epoch": 0.06209,
      "grad_norm": 0.946544007384069,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 6209
    },
    {
      "epoch": 0.0621,
      "grad_norm": 0.7496490508161171,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 6210
    },
    {
      "epoch": 0.06211,
      "grad_norm": 0.7213205165847495,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6211
    },
    {
      "epoch": 0.06212,
      "grad_norm": 0.7495412802850021,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 6212
    },
    {
      "epoch": 0.06213,
      "grad_norm": 0.7517545861084816,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 6213
    },
    {
      "epoch": 0.06214,
      "grad_norm": 0.8206749672685049,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 6214
    },
    {
      "epoch": 0.06215,
      "grad_norm": 0.8335784823968934,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 6215
    },
    {
      "epoch": 0.06216,
      "grad_norm": 0.8600050966450496,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 6216
    },
    {
      "epoch": 0.06217,
      "grad_norm": 0.7434739467397622,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 6217
    },
    {
      "epoch": 0.06218,
      "grad_norm": 0.7048668827578362,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 6218
    },
    {
      "epoch": 0.06219,
      "grad_norm": 0.772659800309845,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 6219
    },
    {
      "epoch": 0.0622,
      "grad_norm": 0.8655188591453645,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 6220
    },
    {
      "epoch": 0.06221,
      "grad_norm": 0.855208686912549,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 6221
    },
    {
      "epoch": 0.06222,
      "grad_norm": 0.7753969968080353,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 6222
    },
    {
      "epoch": 0.06223,
      "grad_norm": 0.7347873178842909,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 6223
    },
    {
      "epoch": 0.06224,
      "grad_norm": 0.7453826365575011,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 6224
    },
    {
      "epoch": 0.06225,
      "grad_norm": 0.6725261598977015,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 6225
    },
    {
      "epoch": 0.06226,
      "grad_norm": 0.7149958935506678,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 6226
    },
    {
      "epoch": 0.06227,
      "grad_norm": 0.6062843965023491,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 6227
    },
    {
      "epoch": 0.06228,
      "grad_norm": 0.5956924427962597,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 6228
    },
    {
      "epoch": 0.06229,
      "grad_norm": 0.49453429887940603,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 6229
    },
    {
      "epoch": 0.0623,
      "grad_norm": 0.4966784851981883,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 6230
    },
    {
      "epoch": 0.06231,
      "grad_norm": 0.5841536691365076,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 6231
    },
    {
      "epoch": 0.06232,
      "grad_norm": 0.6363555647753291,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 6232
    },
    {
      "epoch": 0.06233,
      "grad_norm": 0.739340815218559,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 6233
    },
    {
      "epoch": 0.06234,
      "grad_norm": 0.983288242920816,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 6234
    },
    {
      "epoch": 0.06235,
      "grad_norm": 1.256808031451369,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 6235
    },
    {
      "epoch": 0.06236,
      "grad_norm": 0.879699004471445,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 6236
    },
    {
      "epoch": 0.06237,
      "grad_norm": 0.7521683358827187,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 6237
    },
    {
      "epoch": 0.06238,
      "grad_norm": 0.9148201380927862,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 6238
    },
    {
      "epoch": 0.06239,
      "grad_norm": 0.9607286756265107,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 6239
    },
    {
      "epoch": 0.0624,
      "grad_norm": 0.9539386861231751,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 6240
    },
    {
      "epoch": 0.06241,
      "grad_norm": 0.9355231782983046,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 6241
    },
    {
      "epoch": 0.06242,
      "grad_norm": 0.9288421332433013,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 6242
    },
    {
      "epoch": 0.06243,
      "grad_norm": 0.8502254135089574,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 6243
    },
    {
      "epoch": 0.06244,
      "grad_norm": 0.9732395834043053,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 6244
    },
    {
      "epoch": 0.06245,
      "grad_norm": 1.1071153201796413,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 6245
    },
    {
      "epoch": 0.06246,
      "grad_norm": 1.0940195095069707,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 6246
    },
    {
      "epoch": 0.06247,
      "grad_norm": 0.9412652481549822,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 6247
    },
    {
      "epoch": 0.06248,
      "grad_norm": 0.895195578619345,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 6248
    },
    {
      "epoch": 0.06249,
      "grad_norm": 0.9518706201636767,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 6249
    },
    {
      "epoch": 0.0625,
      "grad_norm": 0.9872959824439677,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 6250
    },
    {
      "epoch": 0.06251,
      "grad_norm": 0.8928606640381823,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6251
    },
    {
      "epoch": 0.06252,
      "grad_norm": 0.8613095945713102,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 6252
    },
    {
      "epoch": 0.06253,
      "grad_norm": 0.8745234996707003,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 6253
    },
    {
      "epoch": 0.06254,
      "grad_norm": 0.9945273839471781,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 6254
    },
    {
      "epoch": 0.06255,
      "grad_norm": 0.9317981123576337,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 6255
    },
    {
      "epoch": 0.06256,
      "grad_norm": 0.935398061718941,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 6256
    },
    {
      "epoch": 0.06257,
      "grad_norm": 0.9753404112298208,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 6257
    },
    {
      "epoch": 0.06258,
      "grad_norm": 0.7914387634721403,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 6258
    },
    {
      "epoch": 0.06259,
      "grad_norm": 0.8459303686065615,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 6259
    },
    {
      "epoch": 0.0626,
      "grad_norm": 0.8608798731117852,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 6260
    },
    {
      "epoch": 0.06261,
      "grad_norm": 0.9015697577574606,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 6261
    },
    {
      "epoch": 0.06262,
      "grad_norm": 0.8626927231908963,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 6262
    },
    {
      "epoch": 0.06263,
      "grad_norm": 0.8363061035136914,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 6263
    },
    {
      "epoch": 0.06264,
      "grad_norm": 0.870156868679017,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 6264
    },
    {
      "epoch": 0.06265,
      "grad_norm": 0.9555307920316731,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 6265
    },
    {
      "epoch": 0.06266,
      "grad_norm": 0.8998498765721897,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 6266
    },
    {
      "epoch": 0.06267,
      "grad_norm": 0.8579446764983243,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 6267
    },
    {
      "epoch": 0.06268,
      "grad_norm": 0.7491708731971836,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 6268
    },
    {
      "epoch": 0.06269,
      "grad_norm": 0.5769598306382601,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 6269
    },
    {
      "epoch": 0.0627,
      "grad_norm": 0.612130695073576,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 6270
    },
    {
      "epoch": 0.06271,
      "grad_norm": 0.5804514862016513,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 6271
    },
    {
      "epoch": 0.06272,
      "grad_norm": 0.6768271666465923,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 6272
    },
    {
      "epoch": 0.06273,
      "grad_norm": 0.725620628836043,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 6273
    },
    {
      "epoch": 0.06274,
      "grad_norm": 0.7124587885763234,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 6274
    },
    {
      "epoch": 0.06275,
      "grad_norm": 0.6549941998076714,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 6275
    },
    {
      "epoch": 0.06276,
      "grad_norm": 0.651422535989974,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 6276
    },
    {
      "epoch": 0.06277,
      "grad_norm": 0.7632957652022478,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 6277
    },
    {
      "epoch": 0.06278,
      "grad_norm": 0.8921311551644955,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 6278
    },
    {
      "epoch": 0.06279,
      "grad_norm": 0.9316054360840504,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 6279
    },
    {
      "epoch": 0.0628,
      "grad_norm": 0.9480087249404298,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 6280
    },
    {
      "epoch": 0.06281,
      "grad_norm": 0.9437555976288077,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 6281
    },
    {
      "epoch": 0.06282,
      "grad_norm": 0.8181620617485008,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6282
    },
    {
      "epoch": 0.06283,
      "grad_norm": 0.6530022259763903,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 6283
    },
    {
      "epoch": 0.06284,
      "grad_norm": 0.5914752081352821,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 6284
    },
    {
      "epoch": 0.06285,
      "grad_norm": 0.5833107165282243,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 6285
    },
    {
      "epoch": 0.06286,
      "grad_norm": 0.6373596862276327,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 6286
    },
    {
      "epoch": 0.06287,
      "grad_norm": 0.6834837784355613,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 6287
    },
    {
      "epoch": 0.06288,
      "grad_norm": 0.6535796667468495,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 6288
    },
    {
      "epoch": 0.06289,
      "grad_norm": 0.6174828688423645,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 6289
    },
    {
      "epoch": 0.0629,
      "grad_norm": 0.4959130253790628,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 6290
    },
    {
      "epoch": 0.06291,
      "grad_norm": 0.5595818315925254,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 6291
    },
    {
      "epoch": 0.06292,
      "grad_norm": 0.5248847935016591,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 6292
    },
    {
      "epoch": 0.06293,
      "grad_norm": 0.5063929729505692,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 6293
    },
    {
      "epoch": 0.06294,
      "grad_norm": 0.46785323488692016,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 6294
    },
    {
      "epoch": 0.06295,
      "grad_norm": 0.4202500873613126,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 6295
    },
    {
      "epoch": 0.06296,
      "grad_norm": 0.42143259177649955,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 6296
    },
    {
      "epoch": 0.06297,
      "grad_norm": 0.49833908617620376,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 6297
    },
    {
      "epoch": 0.06298,
      "grad_norm": 0.7251722641880546,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 6298
    },
    {
      "epoch": 0.06299,
      "grad_norm": 1.1796989504352162,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 6299
    },
    {
      "epoch": 0.063,
      "grad_norm": 1.0778085487061984,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 6300
    },
    {
      "epoch": 0.06301,
      "grad_norm": 0.890093583413934,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 6301
    },
    {
      "epoch": 0.06302,
      "grad_norm": 0.8849818441147115,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 6302
    },
    {
      "epoch": 0.06303,
      "grad_norm": 0.9016935377264436,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 6303
    },
    {
      "epoch": 0.06304,
      "grad_norm": 0.9975700771133877,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 6304
    },
    {
      "epoch": 0.06305,
      "grad_norm": 1.078399082415521,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 6305
    },
    {
      "epoch": 0.06306,
      "grad_norm": 0.8656165382853225,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 6306
    },
    {
      "epoch": 0.06307,
      "grad_norm": 0.9837508136152356,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 6307
    },
    {
      "epoch": 0.06308,
      "grad_norm": 1.0753663335239507,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 6308
    },
    {
      "epoch": 0.06309,
      "grad_norm": 0.9920769784954451,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 6309
    },
    {
      "epoch": 0.0631,
      "grad_norm": 0.8930854451381682,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 6310
    },
    {
      "epoch": 0.06311,
      "grad_norm": 0.9784115790794385,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 6311
    },
    {
      "epoch": 0.06312,
      "grad_norm": 0.9052068566792085,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 6312
    },
    {
      "epoch": 0.06313,
      "grad_norm": 0.8062060403102312,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 6313
    },
    {
      "epoch": 0.06314,
      "grad_norm": 0.8612542206730062,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 6314
    },
    {
      "epoch": 0.06315,
      "grad_norm": 1.2349281974671547,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 6315
    },
    {
      "epoch": 0.06316,
      "grad_norm": 0.9748496951834336,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 6316
    },
    {
      "epoch": 0.06317,
      "grad_norm": 0.8725933515560151,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 6317
    },
    {
      "epoch": 0.06318,
      "grad_norm": 0.7900742702838245,
      "learning_rate": 0.003,
      "loss": 4.1613,
      "step": 6318
    },
    {
      "epoch": 0.06319,
      "grad_norm": 0.8185448176929361,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 6319
    },
    {
      "epoch": 0.0632,
      "grad_norm": 0.7867763226056398,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 6320
    },
    {
      "epoch": 0.06321,
      "grad_norm": 0.7397415591963721,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 6321
    },
    {
      "epoch": 0.06322,
      "grad_norm": 0.7295018218458792,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 6322
    },
    {
      "epoch": 0.06323,
      "grad_norm": 0.8038126764870634,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 6323
    },
    {
      "epoch": 0.06324,
      "grad_norm": 0.8381105892893556,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 6324
    },
    {
      "epoch": 0.06325,
      "grad_norm": 0.7753841480841397,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 6325
    },
    {
      "epoch": 0.06326,
      "grad_norm": 0.7397370113159383,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 6326
    },
    {
      "epoch": 0.06327,
      "grad_norm": 0.8110748971387649,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6327
    },
    {
      "epoch": 0.06328,
      "grad_norm": 0.8165737037187061,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 6328
    },
    {
      "epoch": 0.06329,
      "grad_norm": 0.8459755592595166,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 6329
    },
    {
      "epoch": 0.0633,
      "grad_norm": 0.7837172627440523,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 6330
    },
    {
      "epoch": 0.06331,
      "grad_norm": 0.7382526500041345,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 6331
    },
    {
      "epoch": 0.06332,
      "grad_norm": 0.8446347024742388,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 6332
    },
    {
      "epoch": 0.06333,
      "grad_norm": 0.8755392261266487,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 6333
    },
    {
      "epoch": 0.06334,
      "grad_norm": 0.9464338840507761,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 6334
    },
    {
      "epoch": 0.06335,
      "grad_norm": 0.9871255748238474,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 6335
    },
    {
      "epoch": 0.06336,
      "grad_norm": 1.011604006822508,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 6336
    },
    {
      "epoch": 0.06337,
      "grad_norm": 1.054372126242909,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 6337
    },
    {
      "epoch": 0.06338,
      "grad_norm": 0.8149090879744815,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 6338
    },
    {
      "epoch": 0.06339,
      "grad_norm": 0.8045420062877509,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 6339
    },
    {
      "epoch": 0.0634,
      "grad_norm": 0.773692028277079,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 6340
    },
    {
      "epoch": 0.06341,
      "grad_norm": 0.834869748376275,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 6341
    },
    {
      "epoch": 0.06342,
      "grad_norm": 0.8239421334821241,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 6342
    },
    {
      "epoch": 0.06343,
      "grad_norm": 0.7317579621776237,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 6343
    },
    {
      "epoch": 0.06344,
      "grad_norm": 0.7009311944721036,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 6344
    },
    {
      "epoch": 0.06345,
      "grad_norm": 0.6736306168677829,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 6345
    },
    {
      "epoch": 0.06346,
      "grad_norm": 0.6489992606346455,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 6346
    },
    {
      "epoch": 0.06347,
      "grad_norm": 0.7318992775914919,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6347
    },
    {
      "epoch": 0.06348,
      "grad_norm": 0.8506294348624643,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 6348
    },
    {
      "epoch": 0.06349,
      "grad_norm": 0.8414498040161846,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 6349
    },
    {
      "epoch": 0.0635,
      "grad_norm": 0.7853415582358513,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 6350
    },
    {
      "epoch": 0.06351,
      "grad_norm": 0.8089386421880802,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 6351
    },
    {
      "epoch": 0.06352,
      "grad_norm": 0.7578096372347558,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 6352
    },
    {
      "epoch": 0.06353,
      "grad_norm": 0.7726928843900696,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 6353
    },
    {
      "epoch": 0.06354,
      "grad_norm": 0.8235268790134066,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 6354
    },
    {
      "epoch": 0.06355,
      "grad_norm": 0.8685123509492527,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 6355
    },
    {
      "epoch": 0.06356,
      "grad_norm": 0.9033483279727317,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 6356
    },
    {
      "epoch": 0.06357,
      "grad_norm": 0.8189510174670123,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 6357
    },
    {
      "epoch": 0.06358,
      "grad_norm": 0.7233022129389708,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 6358
    },
    {
      "epoch": 0.06359,
      "grad_norm": 0.591618308364435,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6359
    },
    {
      "epoch": 0.0636,
      "grad_norm": 0.6724366774358936,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 6360
    },
    {
      "epoch": 0.06361,
      "grad_norm": 0.7710988154121916,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 6361
    },
    {
      "epoch": 0.06362,
      "grad_norm": 0.8624550458659666,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 6362
    },
    {
      "epoch": 0.06363,
      "grad_norm": 0.9165629494523708,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 6363
    },
    {
      "epoch": 0.06364,
      "grad_norm": 0.7471123965901227,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 6364
    },
    {
      "epoch": 0.06365,
      "grad_norm": 0.674364493786534,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 6365
    },
    {
      "epoch": 0.06366,
      "grad_norm": 0.7335045602903685,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 6366
    },
    {
      "epoch": 0.06367,
      "grad_norm": 0.829927307349712,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 6367
    },
    {
      "epoch": 0.06368,
      "grad_norm": 0.9487139825002684,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 6368
    },
    {
      "epoch": 0.06369,
      "grad_norm": 0.911983799114357,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 6369
    },
    {
      "epoch": 0.0637,
      "grad_norm": 1.002706686712587,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 6370
    },
    {
      "epoch": 0.06371,
      "grad_norm": 1.0290249495706973,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 6371
    },
    {
      "epoch": 0.06372,
      "grad_norm": 0.9956241176141073,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 6372
    },
    {
      "epoch": 0.06373,
      "grad_norm": 0.9694674226068865,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 6373
    },
    {
      "epoch": 0.06374,
      "grad_norm": 0.8725376570060053,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 6374
    },
    {
      "epoch": 0.06375,
      "grad_norm": 0.8676467464492347,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 6375
    },
    {
      "epoch": 0.06376,
      "grad_norm": 0.8676286162915308,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 6376
    },
    {
      "epoch": 0.06377,
      "grad_norm": 0.9874746445128965,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 6377
    },
    {
      "epoch": 0.06378,
      "grad_norm": 1.0158658512452112,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 6378
    },
    {
      "epoch": 0.06379,
      "grad_norm": 1.0804519520556228,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 6379
    },
    {
      "epoch": 0.0638,
      "grad_norm": 1.0517174792581394,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 6380
    },
    {
      "epoch": 0.06381,
      "grad_norm": 0.9191882933740307,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 6381
    },
    {
      "epoch": 0.06382,
      "grad_norm": 1.112305209925744,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 6382
    },
    {
      "epoch": 0.06383,
      "grad_norm": 0.887564897445918,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 6383
    },
    {
      "epoch": 0.06384,
      "grad_norm": 0.7444199605679525,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 6384
    },
    {
      "epoch": 0.06385,
      "grad_norm": 0.6861155001402504,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 6385
    },
    {
      "epoch": 0.06386,
      "grad_norm": 0.6803077477027196,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 6386
    },
    {
      "epoch": 0.06387,
      "grad_norm": 0.6585413586302139,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6387
    },
    {
      "epoch": 0.06388,
      "grad_norm": 0.6526729101850073,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 6388
    },
    {
      "epoch": 0.06389,
      "grad_norm": 0.6961400897039125,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 6389
    },
    {
      "epoch": 0.0639,
      "grad_norm": 0.7971727357460624,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 6390
    },
    {
      "epoch": 0.06391,
      "grad_norm": 0.8100704061167459,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 6391
    },
    {
      "epoch": 0.06392,
      "grad_norm": 0.7878863222547036,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 6392
    },
    {
      "epoch": 0.06393,
      "grad_norm": 0.7832197554643082,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 6393
    },
    {
      "epoch": 0.06394,
      "grad_norm": 0.7676826793962613,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 6394
    },
    {
      "epoch": 0.06395,
      "grad_norm": 0.7865004179204782,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 6395
    },
    {
      "epoch": 0.06396,
      "grad_norm": 0.7481149498310921,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 6396
    },
    {
      "epoch": 0.06397,
      "grad_norm": 0.6578310984328573,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 6397
    },
    {
      "epoch": 0.06398,
      "grad_norm": 0.7814732896391083,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 6398
    },
    {
      "epoch": 0.06399,
      "grad_norm": 0.7910300530341595,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 6399
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.7384277513313977,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 6400
    },
    {
      "epoch": 0.06401,
      "grad_norm": 0.6044786328825004,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 6401
    },
    {
      "epoch": 0.06402,
      "grad_norm": 0.5635787730251838,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 6402
    },
    {
      "epoch": 0.06403,
      "grad_norm": 0.4804036029092371,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 6403
    },
    {
      "epoch": 0.06404,
      "grad_norm": 0.46851547499916696,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 6404
    },
    {
      "epoch": 0.06405,
      "grad_norm": 0.4343792187108702,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 6405
    },
    {
      "epoch": 0.06406,
      "grad_norm": 0.4387967998008624,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 6406
    },
    {
      "epoch": 0.06407,
      "grad_norm": 0.3763862974231301,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 6407
    },
    {
      "epoch": 0.06408,
      "grad_norm": 0.3618832104359289,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 6408
    },
    {
      "epoch": 0.06409,
      "grad_norm": 0.3640424177742778,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 6409
    },
    {
      "epoch": 0.0641,
      "grad_norm": 0.5071096861354427,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 6410
    },
    {
      "epoch": 0.06411,
      "grad_norm": 0.7056594230315663,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 6411
    },
    {
      "epoch": 0.06412,
      "grad_norm": 1.0514700079261672,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 6412
    },
    {
      "epoch": 0.06413,
      "grad_norm": 1.4637822933171232,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 6413
    },
    {
      "epoch": 0.06414,
      "grad_norm": 0.6368174491993011,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 6414
    },
    {
      "epoch": 0.06415,
      "grad_norm": 0.6864691858033185,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 6415
    },
    {
      "epoch": 0.06416,
      "grad_norm": 0.6967129148461769,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 6416
    },
    {
      "epoch": 0.06417,
      "grad_norm": 0.7877830132689768,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 6417
    },
    {
      "epoch": 0.06418,
      "grad_norm": 1.1025848418056756,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 6418
    },
    {
      "epoch": 0.06419,
      "grad_norm": 1.1303087225349653,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 6419
    },
    {
      "epoch": 0.0642,
      "grad_norm": 0.8179128267985711,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 6420
    },
    {
      "epoch": 0.06421,
      "grad_norm": 0.7061052538423624,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 6421
    },
    {
      "epoch": 0.06422,
      "grad_norm": 0.6712943095898026,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 6422
    },
    {
      "epoch": 0.06423,
      "grad_norm": 0.7842157557770016,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 6423
    },
    {
      "epoch": 0.06424,
      "grad_norm": 0.8391508667334445,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 6424
    },
    {
      "epoch": 0.06425,
      "grad_norm": 0.8032559980871449,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 6425
    },
    {
      "epoch": 0.06426,
      "grad_norm": 0.842568662503113,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 6426
    },
    {
      "epoch": 0.06427,
      "grad_norm": 0.9659133641186062,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 6427
    },
    {
      "epoch": 0.06428,
      "grad_norm": 0.9129797056099707,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 6428
    },
    {
      "epoch": 0.06429,
      "grad_norm": 0.8807646015484044,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 6429
    },
    {
      "epoch": 0.0643,
      "grad_norm": 0.9116849659316084,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 6430
    },
    {
      "epoch": 0.06431,
      "grad_norm": 0.9401397808995475,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 6431
    },
    {
      "epoch": 0.06432,
      "grad_norm": 1.094843463270304,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 6432
    },
    {
      "epoch": 0.06433,
      "grad_norm": 0.97330153788876,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 6433
    },
    {
      "epoch": 0.06434,
      "grad_norm": 0.8801080101022402,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 6434
    },
    {
      "epoch": 0.06435,
      "grad_norm": 0.7207019820857952,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 6435
    },
    {
      "epoch": 0.06436,
      "grad_norm": 0.7301567424764769,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 6436
    },
    {
      "epoch": 0.06437,
      "grad_norm": 0.7242287759939515,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 6437
    },
    {
      "epoch": 0.06438,
      "grad_norm": 0.6183084411636348,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 6438
    },
    {
      "epoch": 0.06439,
      "grad_norm": 0.6942308219572709,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 6439
    },
    {
      "epoch": 0.0644,
      "grad_norm": 0.6502050855985386,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 6440
    },
    {
      "epoch": 0.06441,
      "grad_norm": 0.6169845208363806,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 6441
    },
    {
      "epoch": 0.06442,
      "grad_norm": 0.6761282291621807,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 6442
    },
    {
      "epoch": 0.06443,
      "grad_norm": 0.7049850390444782,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 6443
    },
    {
      "epoch": 0.06444,
      "grad_norm": 0.7982502738327323,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 6444
    },
    {
      "epoch": 0.06445,
      "grad_norm": 0.8849927446065019,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 6445
    },
    {
      "epoch": 0.06446,
      "grad_norm": 0.9290562203347035,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 6446
    },
    {
      "epoch": 0.06447,
      "grad_norm": 0.9155039355826416,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 6447
    },
    {
      "epoch": 0.06448,
      "grad_norm": 0.915311592818687,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 6448
    },
    {
      "epoch": 0.06449,
      "grad_norm": 0.8915234617599778,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 6449
    },
    {
      "epoch": 0.0645,
      "grad_norm": 0.8516263875404035,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 6450
    },
    {
      "epoch": 0.06451,
      "grad_norm": 0.9911826973450533,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 6451
    },
    {
      "epoch": 0.06452,
      "grad_norm": 1.0818621297011708,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 6452
    },
    {
      "epoch": 0.06453,
      "grad_norm": 0.9321332755148247,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 6453
    },
    {
      "epoch": 0.06454,
      "grad_norm": 0.850408071613158,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 6454
    },
    {
      "epoch": 0.06455,
      "grad_norm": 0.8713116964180613,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 6455
    },
    {
      "epoch": 0.06456,
      "grad_norm": 0.7964782772027043,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 6456
    },
    {
      "epoch": 0.06457,
      "grad_norm": 0.8045164932744877,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 6457
    },
    {
      "epoch": 0.06458,
      "grad_norm": 0.8513566837706239,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 6458
    },
    {
      "epoch": 0.06459,
      "grad_norm": 0.8890380816261108,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 6459
    },
    {
      "epoch": 0.0646,
      "grad_norm": 0.9142267341528604,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 6460
    },
    {
      "epoch": 0.06461,
      "grad_norm": 1.1972918407769095,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 6461
    },
    {
      "epoch": 0.06462,
      "grad_norm": 1.0336978547847568,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 6462
    },
    {
      "epoch": 0.06463,
      "grad_norm": 1.0122741524242997,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 6463
    },
    {
      "epoch": 0.06464,
      "grad_norm": 1.0493974717346164,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 6464
    },
    {
      "epoch": 0.06465,
      "grad_norm": 0.9438301613214977,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 6465
    },
    {
      "epoch": 0.06466,
      "grad_norm": 0.8498735851960341,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 6466
    },
    {
      "epoch": 0.06467,
      "grad_norm": 0.8190041939424388,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 6467
    },
    {
      "epoch": 0.06468,
      "grad_norm": 0.8403763457939923,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 6468
    },
    {
      "epoch": 0.06469,
      "grad_norm": 0.7964142979953805,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 6469
    },
    {
      "epoch": 0.0647,
      "grad_norm": 0.7347258244702751,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 6470
    },
    {
      "epoch": 0.06471,
      "grad_norm": 0.6129309594714507,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 6471
    },
    {
      "epoch": 0.06472,
      "grad_norm": 0.6076460566383186,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 6472
    },
    {
      "epoch": 0.06473,
      "grad_norm": 0.5891815040105876,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 6473
    },
    {
      "epoch": 0.06474,
      "grad_norm": 0.5162733573289747,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 6474
    },
    {
      "epoch": 0.06475,
      "grad_norm": 0.5962513852003754,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 6475
    },
    {
      "epoch": 0.06476,
      "grad_norm": 0.7146539756300717,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 6476
    },
    {
      "epoch": 0.06477,
      "grad_norm": 0.8105712416755565,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 6477
    },
    {
      "epoch": 0.06478,
      "grad_norm": 0.9652008011579472,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 6478
    },
    {
      "epoch": 0.06479,
      "grad_norm": 1.1900520557936207,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 6479
    },
    {
      "epoch": 0.0648,
      "grad_norm": 0.7249331917094521,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 6480
    },
    {
      "epoch": 0.06481,
      "grad_norm": 0.6534504012207452,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 6481
    },
    {
      "epoch": 0.06482,
      "grad_norm": 0.7458945563014072,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 6482
    },
    {
      "epoch": 0.06483,
      "grad_norm": 0.909230494347854,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 6483
    },
    {
      "epoch": 0.06484,
      "grad_norm": 0.9786701503465991,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 6484
    },
    {
      "epoch": 0.06485,
      "grad_norm": 0.8056031175729698,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 6485
    },
    {
      "epoch": 0.06486,
      "grad_norm": 0.7315640607458516,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 6486
    },
    {
      "epoch": 0.06487,
      "grad_norm": 0.738123736009974,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 6487
    },
    {
      "epoch": 0.06488,
      "grad_norm": 0.7016088617855374,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 6488
    },
    {
      "epoch": 0.06489,
      "grad_norm": 0.7706221605393652,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 6489
    },
    {
      "epoch": 0.0649,
      "grad_norm": 0.8075185716569853,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 6490
    },
    {
      "epoch": 0.06491,
      "grad_norm": 0.7917671230247577,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 6491
    },
    {
      "epoch": 0.06492,
      "grad_norm": 0.851038360366963,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 6492
    },
    {
      "epoch": 0.06493,
      "grad_norm": 0.8479749542932367,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6493
    },
    {
      "epoch": 0.06494,
      "grad_norm": 0.7857228913822375,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 6494
    },
    {
      "epoch": 0.06495,
      "grad_norm": 0.7000090472611841,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 6495
    },
    {
      "epoch": 0.06496,
      "grad_norm": 0.6189238416133499,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 6496
    },
    {
      "epoch": 0.06497,
      "grad_norm": 0.5745909145349337,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 6497
    },
    {
      "epoch": 0.06498,
      "grad_norm": 0.5294157379174294,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 6498
    },
    {
      "epoch": 0.06499,
      "grad_norm": 0.5466886858120876,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 6499
    },
    {
      "epoch": 0.065,
      "grad_norm": 0.6539813137699607,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 6500
    },
    {
      "epoch": 0.06501,
      "grad_norm": 0.6647161266915989,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 6501
    },
    {
      "epoch": 0.06502,
      "grad_norm": 0.6443788622041227,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 6502
    },
    {
      "epoch": 0.06503,
      "grad_norm": 0.7571231315947718,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6503
    },
    {
      "epoch": 0.06504,
      "grad_norm": 0.9637682445146619,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 6504
    },
    {
      "epoch": 0.06505,
      "grad_norm": 1.3505877682488592,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 6505
    },
    {
      "epoch": 0.06506,
      "grad_norm": 0.7210710726778544,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 6506
    },
    {
      "epoch": 0.06507,
      "grad_norm": 0.8243540456063395,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 6507
    },
    {
      "epoch": 0.06508,
      "grad_norm": 0.9264270578051855,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 6508
    },
    {
      "epoch": 0.06509,
      "grad_norm": 0.8402439484381861,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 6509
    },
    {
      "epoch": 0.0651,
      "grad_norm": 0.8452797474799149,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 6510
    },
    {
      "epoch": 0.06511,
      "grad_norm": 0.8155771099505443,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 6511
    },
    {
      "epoch": 0.06512,
      "grad_norm": 0.8504514854470253,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 6512
    },
    {
      "epoch": 0.06513,
      "grad_norm": 0.7996056231914204,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 6513
    },
    {
      "epoch": 0.06514,
      "grad_norm": 0.8483202440944669,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 6514
    },
    {
      "epoch": 0.06515,
      "grad_norm": 0.729647231890123,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 6515
    },
    {
      "epoch": 0.06516,
      "grad_norm": 0.7668262347734042,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 6516
    },
    {
      "epoch": 0.06517,
      "grad_norm": 0.9668803624817275,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6517
    },
    {
      "epoch": 0.06518,
      "grad_norm": 0.9221337117494466,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 6518
    },
    {
      "epoch": 0.06519,
      "grad_norm": 0.8991712754647655,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 6519
    },
    {
      "epoch": 0.0652,
      "grad_norm": 0.9996762890374128,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 6520
    },
    {
      "epoch": 0.06521,
      "grad_norm": 0.9694147573902756,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 6521
    },
    {
      "epoch": 0.06522,
      "grad_norm": 1.1436487696458615,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 6522
    },
    {
      "epoch": 0.06523,
      "grad_norm": 1.1761311790032607,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 6523
    },
    {
      "epoch": 0.06524,
      "grad_norm": 0.9935715060905626,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 6524
    },
    {
      "epoch": 0.06525,
      "grad_norm": 1.0984134338267004,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 6525
    },
    {
      "epoch": 0.06526,
      "grad_norm": 0.8597157647493189,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 6526
    },
    {
      "epoch": 0.06527,
      "grad_norm": 0.9122953325479415,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 6527
    },
    {
      "epoch": 0.06528,
      "grad_norm": 0.9550047890046722,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 6528
    },
    {
      "epoch": 0.06529,
      "grad_norm": 0.9456502240272384,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 6529
    },
    {
      "epoch": 0.0653,
      "grad_norm": 0.8144448232250512,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 6530
    },
    {
      "epoch": 0.06531,
      "grad_norm": 0.7884345470232,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 6531
    },
    {
      "epoch": 0.06532,
      "grad_norm": 0.7546346553317457,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 6532
    },
    {
      "epoch": 0.06533,
      "grad_norm": 0.6337607386448925,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 6533
    },
    {
      "epoch": 0.06534,
      "grad_norm": 0.6571305549922228,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 6534
    },
    {
      "epoch": 0.06535,
      "grad_norm": 0.6727892441041884,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 6535
    },
    {
      "epoch": 0.06536,
      "grad_norm": 0.6796596200367593,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 6536
    },
    {
      "epoch": 0.06537,
      "grad_norm": 0.6850578598200898,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 6537
    },
    {
      "epoch": 0.06538,
      "grad_norm": 0.6903169559185911,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 6538
    },
    {
      "epoch": 0.06539,
      "grad_norm": 0.8869816313052468,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 6539
    },
    {
      "epoch": 0.0654,
      "grad_norm": 1.168073515327997,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 6540
    },
    {
      "epoch": 0.06541,
      "grad_norm": 0.7182356963906228,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 6541
    },
    {
      "epoch": 0.06542,
      "grad_norm": 0.6188811083162069,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 6542
    },
    {
      "epoch": 0.06543,
      "grad_norm": 0.8051980596249964,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 6543
    },
    {
      "epoch": 0.06544,
      "grad_norm": 0.9132061876425465,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 6544
    },
    {
      "epoch": 0.06545,
      "grad_norm": 0.9233000070732396,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 6545
    },
    {
      "epoch": 0.06546,
      "grad_norm": 0.7831417568832402,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 6546
    },
    {
      "epoch": 0.06547,
      "grad_norm": 0.6532883144447844,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 6547
    },
    {
      "epoch": 0.06548,
      "grad_norm": 0.6727801173157394,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 6548
    },
    {
      "epoch": 0.06549,
      "grad_norm": 0.6217127018777375,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 6549
    },
    {
      "epoch": 0.0655,
      "grad_norm": 0.7089420407652894,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 6550
    },
    {
      "epoch": 0.06551,
      "grad_norm": 0.8434472089359629,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 6551
    },
    {
      "epoch": 0.06552,
      "grad_norm": 0.957288546390429,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 6552
    },
    {
      "epoch": 0.06553,
      "grad_norm": 0.8538606588662652,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 6553
    },
    {
      "epoch": 0.06554,
      "grad_norm": 0.6821331682783609,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 6554
    },
    {
      "epoch": 0.06555,
      "grad_norm": 0.5970317370800081,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 6555
    },
    {
      "epoch": 0.06556,
      "grad_norm": 0.6339812838968247,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 6556
    },
    {
      "epoch": 0.06557,
      "grad_norm": 0.7300737435789836,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 6557
    },
    {
      "epoch": 0.06558,
      "grad_norm": 0.9340751362463283,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 6558
    },
    {
      "epoch": 0.06559,
      "grad_norm": 1.078857392761123,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 6559
    },
    {
      "epoch": 0.0656,
      "grad_norm": 0.7783300852395647,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 6560
    },
    {
      "epoch": 0.06561,
      "grad_norm": 0.6838855361080844,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 6561
    },
    {
      "epoch": 0.06562,
      "grad_norm": 0.8236313339666234,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 6562
    },
    {
      "epoch": 0.06563,
      "grad_norm": 0.9128215836077087,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 6563
    },
    {
      "epoch": 0.06564,
      "grad_norm": 1.0183713945436557,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 6564
    },
    {
      "epoch": 0.06565,
      "grad_norm": 0.9875484143170253,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 6565
    },
    {
      "epoch": 0.06566,
      "grad_norm": 0.8482203740641269,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 6566
    },
    {
      "epoch": 0.06567,
      "grad_norm": 0.8887069547124475,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 6567
    },
    {
      "epoch": 0.06568,
      "grad_norm": 1.0056485677045872,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 6568
    },
    {
      "epoch": 0.06569,
      "grad_norm": 0.9064862101322059,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 6569
    },
    {
      "epoch": 0.0657,
      "grad_norm": 0.7555124162684835,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 6570
    },
    {
      "epoch": 0.06571,
      "grad_norm": 0.7348370653563371,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 6571
    },
    {
      "epoch": 0.06572,
      "grad_norm": 0.8152888496093048,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 6572
    },
    {
      "epoch": 0.06573,
      "grad_norm": 0.7572678952633992,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 6573
    },
    {
      "epoch": 0.06574,
      "grad_norm": 0.7173617049609856,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 6574
    },
    {
      "epoch": 0.06575,
      "grad_norm": 0.7216202084765286,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 6575
    },
    {
      "epoch": 0.06576,
      "grad_norm": 0.7321185248183836,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 6576
    },
    {
      "epoch": 0.06577,
      "grad_norm": 0.812337799951959,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 6577
    },
    {
      "epoch": 0.06578,
      "grad_norm": 1.0097566236813522,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 6578
    },
    {
      "epoch": 0.06579,
      "grad_norm": 1.1289060393565276,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 6579
    },
    {
      "epoch": 0.0658,
      "grad_norm": 1.1236108631521293,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 6580
    },
    {
      "epoch": 0.06581,
      "grad_norm": 0.9095208566603907,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 6581
    },
    {
      "epoch": 0.06582,
      "grad_norm": 0.7078312720167307,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 6582
    },
    {
      "epoch": 0.06583,
      "grad_norm": 0.6250796377300359,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 6583
    },
    {
      "epoch": 0.06584,
      "grad_norm": 0.6182361788610296,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 6584
    },
    {
      "epoch": 0.06585,
      "grad_norm": 0.6199101947792578,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 6585
    },
    {
      "epoch": 0.06586,
      "grad_norm": 0.644608198538537,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 6586
    },
    {
      "epoch": 0.06587,
      "grad_norm": 0.7531896129898413,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 6587
    },
    {
      "epoch": 0.06588,
      "grad_norm": 0.9463136908207883,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 6588
    },
    {
      "epoch": 0.06589,
      "grad_norm": 1.0611698029062901,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 6589
    },
    {
      "epoch": 0.0659,
      "grad_norm": 0.9951512526105761,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 6590
    },
    {
      "epoch": 0.06591,
      "grad_norm": 0.9622203108833793,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 6591
    },
    {
      "epoch": 0.06592,
      "grad_norm": 0.7792553434807309,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 6592
    },
    {
      "epoch": 0.06593,
      "grad_norm": 0.7097135195972493,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 6593
    },
    {
      "epoch": 0.06594,
      "grad_norm": 0.7407425345106201,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 6594
    },
    {
      "epoch": 0.06595,
      "grad_norm": 0.7612553261817883,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 6595
    },
    {
      "epoch": 0.06596,
      "grad_norm": 0.7713906419670148,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 6596
    },
    {
      "epoch": 0.06597,
      "grad_norm": 0.95305210338701,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 6597
    },
    {
      "epoch": 0.06598,
      "grad_norm": 0.9912085601243025,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 6598
    },
    {
      "epoch": 0.06599,
      "grad_norm": 0.9563843538129239,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 6599
    },
    {
      "epoch": 0.066,
      "grad_norm": 0.9173651825854763,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 6600
    },
    {
      "epoch": 0.06601,
      "grad_norm": 0.8706761413967128,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 6601
    },
    {
      "epoch": 0.06602,
      "grad_norm": 0.8530638651302136,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 6602
    },
    {
      "epoch": 0.06603,
      "grad_norm": 0.7620877613851017,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6603
    },
    {
      "epoch": 0.06604,
      "grad_norm": 0.7157412127972154,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6604
    },
    {
      "epoch": 0.06605,
      "grad_norm": 0.7664275284896481,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 6605
    },
    {
      "epoch": 0.06606,
      "grad_norm": 0.8596918832087841,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 6606
    },
    {
      "epoch": 0.06607,
      "grad_norm": 0.9733248442463999,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 6607
    },
    {
      "epoch": 0.06608,
      "grad_norm": 1.261129457668947,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 6608
    },
    {
      "epoch": 0.06609,
      "grad_norm": 0.7875463153702758,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 6609
    },
    {
      "epoch": 0.0661,
      "grad_norm": 0.6418585115175501,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 6610
    },
    {
      "epoch": 0.06611,
      "grad_norm": 0.6118831582806695,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 6611
    },
    {
      "epoch": 0.06612,
      "grad_norm": 0.6131198639857838,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 6612
    },
    {
      "epoch": 0.06613,
      "grad_norm": 0.605891802900793,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 6613
    },
    {
      "epoch": 0.06614,
      "grad_norm": 0.7122334128472662,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 6614
    },
    {
      "epoch": 0.06615,
      "grad_norm": 0.986281825824359,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 6615
    },
    {
      "epoch": 0.06616,
      "grad_norm": 1.209509744232019,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 6616
    },
    {
      "epoch": 0.06617,
      "grad_norm": 0.589186192111392,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 6617
    },
    {
      "epoch": 0.06618,
      "grad_norm": 0.817010663602591,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 6618
    },
    {
      "epoch": 0.06619,
      "grad_norm": 1.1163167401139134,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 6619
    },
    {
      "epoch": 0.0662,
      "grad_norm": 0.861557189675986,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 6620
    },
    {
      "epoch": 0.06621,
      "grad_norm": 0.7734302388169766,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 6621
    },
    {
      "epoch": 0.06622,
      "grad_norm": 0.6681446519688905,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 6622
    },
    {
      "epoch": 0.06623,
      "grad_norm": 0.6086441119859902,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 6623
    },
    {
      "epoch": 0.06624,
      "grad_norm": 0.6161665012708084,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 6624
    },
    {
      "epoch": 0.06625,
      "grad_norm": 0.6016793043105603,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 6625
    },
    {
      "epoch": 0.06626,
      "grad_norm": 0.5731558676794132,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 6626
    },
    {
      "epoch": 0.06627,
      "grad_norm": 0.5748801713145845,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 6627
    },
    {
      "epoch": 0.06628,
      "grad_norm": 0.5724674668619956,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 6628
    },
    {
      "epoch": 0.06629,
      "grad_norm": 0.5329925465369826,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 6629
    },
    {
      "epoch": 0.0663,
      "grad_norm": 0.4945919230759341,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 6630
    },
    {
      "epoch": 0.06631,
      "grad_norm": 0.45658695378096065,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 6631
    },
    {
      "epoch": 0.06632,
      "grad_norm": 0.526478004374339,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 6632
    },
    {
      "epoch": 0.06633,
      "grad_norm": 0.6229089968891325,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 6633
    },
    {
      "epoch": 0.06634,
      "grad_norm": 0.7440020938095453,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 6634
    },
    {
      "epoch": 0.06635,
      "grad_norm": 1.0497670831766566,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 6635
    },
    {
      "epoch": 0.06636,
      "grad_norm": 1.14053551502134,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 6636
    },
    {
      "epoch": 0.06637,
      "grad_norm": 0.7502849737062908,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 6637
    },
    {
      "epoch": 0.06638,
      "grad_norm": 0.8150413391233561,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 6638
    },
    {
      "epoch": 0.06639,
      "grad_norm": 0.9898719461642121,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 6639
    },
    {
      "epoch": 0.0664,
      "grad_norm": 1.1511766317227263,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 6640
    },
    {
      "epoch": 0.06641,
      "grad_norm": 0.8614640565436265,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 6641
    },
    {
      "epoch": 0.06642,
      "grad_norm": 0.7977417905197698,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 6642
    },
    {
      "epoch": 0.06643,
      "grad_norm": 0.7780801354815946,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 6643
    },
    {
      "epoch": 0.06644,
      "grad_norm": 0.7755477432783495,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6644
    },
    {
      "epoch": 0.06645,
      "grad_norm": 0.9119617892094356,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 6645
    },
    {
      "epoch": 0.06646,
      "grad_norm": 0.878567897301363,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6646
    },
    {
      "epoch": 0.06647,
      "grad_norm": 0.8867140954494283,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 6647
    },
    {
      "epoch": 0.06648,
      "grad_norm": 0.8336313880890538,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 6648
    },
    {
      "epoch": 0.06649,
      "grad_norm": 0.8170814645370212,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 6649
    },
    {
      "epoch": 0.0665,
      "grad_norm": 0.7834859433585468,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 6650
    },
    {
      "epoch": 0.06651,
      "grad_norm": 0.8357692571639677,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 6651
    },
    {
      "epoch": 0.06652,
      "grad_norm": 0.9187801035749852,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 6652
    },
    {
      "epoch": 0.06653,
      "grad_norm": 0.969813469239403,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 6653
    },
    {
      "epoch": 0.06654,
      "grad_norm": 1.0521703710878825,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 6654
    },
    {
      "epoch": 0.06655,
      "grad_norm": 1.115077498474384,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 6655
    },
    {
      "epoch": 0.06656,
      "grad_norm": 0.9993667641705132,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 6656
    },
    {
      "epoch": 0.06657,
      "grad_norm": 1.1623793190867178,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 6657
    },
    {
      "epoch": 0.06658,
      "grad_norm": 0.8317171031791138,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 6658
    },
    {
      "epoch": 0.06659,
      "grad_norm": 0.7417282289820908,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 6659
    },
    {
      "epoch": 0.0666,
      "grad_norm": 0.7996438420969371,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 6660
    },
    {
      "epoch": 0.06661,
      "grad_norm": 0.8239468144667403,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 6661
    },
    {
      "epoch": 0.06662,
      "grad_norm": 0.8636078960543694,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 6662
    },
    {
      "epoch": 0.06663,
      "grad_norm": 0.8286265545981787,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 6663
    },
    {
      "epoch": 0.06664,
      "grad_norm": 0.7760590887675644,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 6664
    },
    {
      "epoch": 0.06665,
      "grad_norm": 0.6944243205061547,
      "learning_rate": 0.003,
      "loss": 4.1548,
      "step": 6665
    },
    {
      "epoch": 0.06666,
      "grad_norm": 0.7180657446089408,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 6666
    },
    {
      "epoch": 0.06667,
      "grad_norm": 0.8974376505212266,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 6667
    },
    {
      "epoch": 0.06668,
      "grad_norm": 1.0172055502989037,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 6668
    },
    {
      "epoch": 0.06669,
      "grad_norm": 1.1607157063808713,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 6669
    },
    {
      "epoch": 0.0667,
      "grad_norm": 0.6558890868526093,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 6670
    },
    {
      "epoch": 0.06671,
      "grad_norm": 0.5339830424514167,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 6671
    },
    {
      "epoch": 0.06672,
      "grad_norm": 0.6960933681548969,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6672
    },
    {
      "epoch": 0.06673,
      "grad_norm": 0.8613148280885923,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 6673
    },
    {
      "epoch": 0.06674,
      "grad_norm": 1.0944320576632818,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 6674
    },
    {
      "epoch": 0.06675,
      "grad_norm": 0.8624252033051901,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 6675
    },
    {
      "epoch": 0.06676,
      "grad_norm": 0.6153729593690301,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 6676
    },
    {
      "epoch": 0.06677,
      "grad_norm": 0.6138367390296136,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 6677
    },
    {
      "epoch": 0.06678,
      "grad_norm": 0.6766082571603963,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 6678
    },
    {
      "epoch": 0.06679,
      "grad_norm": 0.7509125610237447,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 6679
    },
    {
      "epoch": 0.0668,
      "grad_norm": 0.7800006560160819,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 6680
    },
    {
      "epoch": 0.06681,
      "grad_norm": 0.7276956103034433,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 6681
    },
    {
      "epoch": 0.06682,
      "grad_norm": 0.7677356222531484,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 6682
    },
    {
      "epoch": 0.06683,
      "grad_norm": 0.9103583128182188,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 6683
    },
    {
      "epoch": 0.06684,
      "grad_norm": 0.9184243344640678,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 6684
    },
    {
      "epoch": 0.06685,
      "grad_norm": 0.8953119859552124,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 6685
    },
    {
      "epoch": 0.06686,
      "grad_norm": 0.8649576016834076,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 6686
    },
    {
      "epoch": 0.06687,
      "grad_norm": 0.8737336563045488,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 6687
    },
    {
      "epoch": 0.06688,
      "grad_norm": 0.8001454108494693,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 6688
    },
    {
      "epoch": 0.06689,
      "grad_norm": 0.7007608354845569,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 6689
    },
    {
      "epoch": 0.0669,
      "grad_norm": 0.7346334248821653,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 6690
    },
    {
      "epoch": 0.06691,
      "grad_norm": 0.8242428244307782,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 6691
    },
    {
      "epoch": 0.06692,
      "grad_norm": 0.8785059768462102,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 6692
    },
    {
      "epoch": 0.06693,
      "grad_norm": 0.8419120527636385,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 6693
    },
    {
      "epoch": 0.06694,
      "grad_norm": 0.8364345690497847,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 6694
    },
    {
      "epoch": 0.06695,
      "grad_norm": 0.7282086608438694,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 6695
    },
    {
      "epoch": 0.06696,
      "grad_norm": 0.7156637313733002,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 6696
    },
    {
      "epoch": 0.06697,
      "grad_norm": 0.759716727374639,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6697
    },
    {
      "epoch": 0.06698,
      "grad_norm": 0.6814818122358287,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 6698
    },
    {
      "epoch": 0.06699,
      "grad_norm": 0.8881481031361592,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 6699
    },
    {
      "epoch": 0.067,
      "grad_norm": 0.916609985935385,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 6700
    },
    {
      "epoch": 0.06701,
      "grad_norm": 0.7647456152832255,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 6701
    },
    {
      "epoch": 0.06702,
      "grad_norm": 0.9878626082963966,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 6702
    },
    {
      "epoch": 0.06703,
      "grad_norm": 1.450715682692916,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 6703
    },
    {
      "epoch": 0.06704,
      "grad_norm": 0.8099658530003969,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 6704
    },
    {
      "epoch": 0.06705,
      "grad_norm": 0.8003246468539195,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 6705
    },
    {
      "epoch": 0.06706,
      "grad_norm": 0.9277530689587093,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 6706
    },
    {
      "epoch": 0.06707,
      "grad_norm": 1.081334232913671,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 6707
    },
    {
      "epoch": 0.06708,
      "grad_norm": 0.7824655188582949,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 6708
    },
    {
      "epoch": 0.06709,
      "grad_norm": 0.6810909588960559,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 6709
    },
    {
      "epoch": 0.0671,
      "grad_norm": 0.6175677027859562,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 6710
    },
    {
      "epoch": 0.06711,
      "grad_norm": 0.7248156251811534,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 6711
    },
    {
      "epoch": 0.06712,
      "grad_norm": 0.9236044790837477,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 6712
    },
    {
      "epoch": 0.06713,
      "grad_norm": 1.1656738764307082,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 6713
    },
    {
      "epoch": 0.06714,
      "grad_norm": 0.687130908864893,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 6714
    },
    {
      "epoch": 0.06715,
      "grad_norm": 0.6511308300062618,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 6715
    },
    {
      "epoch": 0.06716,
      "grad_norm": 0.9028221528538252,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 6716
    },
    {
      "epoch": 0.06717,
      "grad_norm": 0.9804042876410267,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 6717
    },
    {
      "epoch": 0.06718,
      "grad_norm": 1.0173300549817423,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 6718
    },
    {
      "epoch": 0.06719,
      "grad_norm": 0.9220403787562276,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 6719
    },
    {
      "epoch": 0.0672,
      "grad_norm": 0.7514745228001316,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 6720
    },
    {
      "epoch": 0.06721,
      "grad_norm": 0.8665408867149632,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 6721
    },
    {
      "epoch": 0.06722,
      "grad_norm": 0.9302053283399239,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 6722
    },
    {
      "epoch": 0.06723,
      "grad_norm": 0.9301656966667754,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 6723
    },
    {
      "epoch": 0.06724,
      "grad_norm": 0.8389657280038182,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 6724
    },
    {
      "epoch": 0.06725,
      "grad_norm": 0.7512173727545822,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 6725
    },
    {
      "epoch": 0.06726,
      "grad_norm": 0.7891627577920859,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 6726
    },
    {
      "epoch": 0.06727,
      "grad_norm": 0.9042190562513309,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6727
    },
    {
      "epoch": 0.06728,
      "grad_norm": 1.0541308359487462,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 6728
    },
    {
      "epoch": 0.06729,
      "grad_norm": 1.0843675552428607,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 6729
    },
    {
      "epoch": 0.0673,
      "grad_norm": 1.0045820679439086,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 6730
    },
    {
      "epoch": 0.06731,
      "grad_norm": 0.9002591648778789,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 6731
    },
    {
      "epoch": 0.06732,
      "grad_norm": 0.646912864765862,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 6732
    },
    {
      "epoch": 0.06733,
      "grad_norm": 0.7141084225358552,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 6733
    },
    {
      "epoch": 0.06734,
      "grad_norm": 0.7401896537351272,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 6734
    },
    {
      "epoch": 0.06735,
      "grad_norm": 0.716630242459992,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 6735
    },
    {
      "epoch": 0.06736,
      "grad_norm": 0.8346796605661073,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 6736
    },
    {
      "epoch": 0.06737,
      "grad_norm": 0.8558768913961102,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 6737
    },
    {
      "epoch": 0.06738,
      "grad_norm": 0.8608671123963944,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 6738
    },
    {
      "epoch": 0.06739,
      "grad_norm": 0.9596153104235727,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 6739
    },
    {
      "epoch": 0.0674,
      "grad_norm": 0.937939190033793,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 6740
    },
    {
      "epoch": 0.06741,
      "grad_norm": 0.7552811761560235,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 6741
    },
    {
      "epoch": 0.06742,
      "grad_norm": 0.650352878706116,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 6742
    },
    {
      "epoch": 0.06743,
      "grad_norm": 0.6887739825416035,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 6743
    },
    {
      "epoch": 0.06744,
      "grad_norm": 0.6459118608591984,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 6744
    },
    {
      "epoch": 0.06745,
      "grad_norm": 0.7004216000552352,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 6745
    },
    {
      "epoch": 0.06746,
      "grad_norm": 0.6099448791242965,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 6746
    },
    {
      "epoch": 0.06747,
      "grad_norm": 0.6988500937390398,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 6747
    },
    {
      "epoch": 0.06748,
      "grad_norm": 0.7156433755323733,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 6748
    },
    {
      "epoch": 0.06749,
      "grad_norm": 0.801421918564714,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 6749
    },
    {
      "epoch": 0.0675,
      "grad_norm": 0.9888166246308364,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 6750
    },
    {
      "epoch": 0.06751,
      "grad_norm": 0.9741131169646507,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 6751
    },
    {
      "epoch": 0.06752,
      "grad_norm": 0.8319653111077174,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 6752
    },
    {
      "epoch": 0.06753,
      "grad_norm": 0.7778459734702268,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 6753
    },
    {
      "epoch": 0.06754,
      "grad_norm": 0.6877593644712623,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 6754
    },
    {
      "epoch": 0.06755,
      "grad_norm": 0.5706976203238855,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 6755
    },
    {
      "epoch": 0.06756,
      "grad_norm": 0.6454631853072159,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 6756
    },
    {
      "epoch": 0.06757,
      "grad_norm": 0.7487310221163939,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 6757
    },
    {
      "epoch": 0.06758,
      "grad_norm": 0.9383683479121699,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 6758
    },
    {
      "epoch": 0.06759,
      "grad_norm": 1.2488321237213988,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 6759
    },
    {
      "epoch": 0.0676,
      "grad_norm": 0.8044219051814184,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 6760
    },
    {
      "epoch": 0.06761,
      "grad_norm": 0.6504031656659648,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 6761
    },
    {
      "epoch": 0.06762,
      "grad_norm": 0.5974848351395243,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 6762
    },
    {
      "epoch": 0.06763,
      "grad_norm": 0.7658766240767714,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 6763
    },
    {
      "epoch": 0.06764,
      "grad_norm": 1.0816158497707196,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 6764
    },
    {
      "epoch": 0.06765,
      "grad_norm": 1.0931948991927563,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 6765
    },
    {
      "epoch": 0.06766,
      "grad_norm": 0.7550131909735355,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 6766
    },
    {
      "epoch": 0.06767,
      "grad_norm": 0.6220201794198699,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 6767
    },
    {
      "epoch": 0.06768,
      "grad_norm": 0.7460575761496198,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 6768
    },
    {
      "epoch": 0.06769,
      "grad_norm": 0.7893758885707454,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 6769
    },
    {
      "epoch": 0.0677,
      "grad_norm": 0.7505537103644507,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 6770
    },
    {
      "epoch": 0.06771,
      "grad_norm": 0.7025624171913118,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 6771
    },
    {
      "epoch": 0.06772,
      "grad_norm": 0.7560527883448278,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 6772
    },
    {
      "epoch": 0.06773,
      "grad_norm": 0.7196524539900757,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 6773
    },
    {
      "epoch": 0.06774,
      "grad_norm": 0.7389219268859994,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 6774
    },
    {
      "epoch": 0.06775,
      "grad_norm": 0.6685763528477642,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 6775
    },
    {
      "epoch": 0.06776,
      "grad_norm": 0.5686448643650356,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 6776
    },
    {
      "epoch": 0.06777,
      "grad_norm": 0.5856469192051172,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 6777
    },
    {
      "epoch": 0.06778,
      "grad_norm": 0.5829647709549176,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 6778
    },
    {
      "epoch": 0.06779,
      "grad_norm": 0.58878250453052,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 6779
    },
    {
      "epoch": 0.0678,
      "grad_norm": 0.6705441528405734,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 6780
    },
    {
      "epoch": 0.06781,
      "grad_norm": 0.694256508354405,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 6781
    },
    {
      "epoch": 0.06782,
      "grad_norm": 0.7836721563560676,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 6782
    },
    {
      "epoch": 0.06783,
      "grad_norm": 0.9605929520599132,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 6783
    },
    {
      "epoch": 0.06784,
      "grad_norm": 0.9687369923954169,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 6784
    },
    {
      "epoch": 0.06785,
      "grad_norm": 0.8611250521831618,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6785
    },
    {
      "epoch": 0.06786,
      "grad_norm": 0.8150784557873674,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 6786
    },
    {
      "epoch": 0.06787,
      "grad_norm": 0.7294549203147801,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 6787
    },
    {
      "epoch": 0.06788,
      "grad_norm": 0.7384718901245526,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 6788
    },
    {
      "epoch": 0.06789,
      "grad_norm": 0.7373796160060775,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 6789
    },
    {
      "epoch": 0.0679,
      "grad_norm": 0.7304803645302286,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 6790
    },
    {
      "epoch": 0.06791,
      "grad_norm": 0.7119811531380004,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 6791
    },
    {
      "epoch": 0.06792,
      "grad_norm": 0.8288761752825445,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 6792
    },
    {
      "epoch": 0.06793,
      "grad_norm": 0.7074297261709009,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 6793
    },
    {
      "epoch": 0.06794,
      "grad_norm": 0.6917103888708278,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 6794
    },
    {
      "epoch": 0.06795,
      "grad_norm": 0.9056603235821777,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 6795
    },
    {
      "epoch": 0.06796,
      "grad_norm": 1.2619528323809053,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 6796
    },
    {
      "epoch": 0.06797,
      "grad_norm": 0.9779295543791401,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 6797
    },
    {
      "epoch": 0.06798,
      "grad_norm": 1.0466081925100106,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 6798
    },
    {
      "epoch": 0.06799,
      "grad_norm": 1.0487678592502534,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 6799
    },
    {
      "epoch": 0.068,
      "grad_norm": 0.9892002508567624,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 6800
    },
    {
      "epoch": 0.06801,
      "grad_norm": 0.9771195211817344,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 6801
    },
    {
      "epoch": 0.06802,
      "grad_norm": 0.9110534102852377,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 6802
    },
    {
      "epoch": 0.06803,
      "grad_norm": 0.9513266489274651,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 6803
    },
    {
      "epoch": 0.06804,
      "grad_norm": 1.1138225723663524,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 6804
    },
    {
      "epoch": 0.06805,
      "grad_norm": 1.0317572052095008,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 6805
    },
    {
      "epoch": 0.06806,
      "grad_norm": 0.797666875905461,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 6806
    },
    {
      "epoch": 0.06807,
      "grad_norm": 0.7541059465600338,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 6807
    },
    {
      "epoch": 0.06808,
      "grad_norm": 0.6793248525851846,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 6808
    },
    {
      "epoch": 0.06809,
      "grad_norm": 0.6873438348421407,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 6809
    },
    {
      "epoch": 0.0681,
      "grad_norm": 0.6049328221467462,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 6810
    },
    {
      "epoch": 0.06811,
      "grad_norm": 0.6187722825054933,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 6811
    },
    {
      "epoch": 0.06812,
      "grad_norm": 0.7495708505892162,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 6812
    },
    {
      "epoch": 0.06813,
      "grad_norm": 0.8952833843553192,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 6813
    },
    {
      "epoch": 0.06814,
      "grad_norm": 1.052298391151623,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 6814
    },
    {
      "epoch": 0.06815,
      "grad_norm": 0.8914337446957912,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 6815
    },
    {
      "epoch": 0.06816,
      "grad_norm": 0.9633140015019734,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 6816
    },
    {
      "epoch": 0.06817,
      "grad_norm": 0.8885850734774637,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 6817
    },
    {
      "epoch": 0.06818,
      "grad_norm": 0.7384023274306349,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 6818
    },
    {
      "epoch": 0.06819,
      "grad_norm": 0.7256819329745252,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 6819
    },
    {
      "epoch": 0.0682,
      "grad_norm": 0.8464960000732561,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 6820
    },
    {
      "epoch": 0.06821,
      "grad_norm": 0.959885711235584,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 6821
    },
    {
      "epoch": 0.06822,
      "grad_norm": 1.0407940708061996,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 6822
    },
    {
      "epoch": 0.06823,
      "grad_norm": 0.9542538781486196,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 6823
    },
    {
      "epoch": 0.06824,
      "grad_norm": 0.8548426620050126,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 6824
    },
    {
      "epoch": 0.06825,
      "grad_norm": 0.7562330738587587,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 6825
    },
    {
      "epoch": 0.06826,
      "grad_norm": 0.6682601342553707,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 6826
    },
    {
      "epoch": 0.06827,
      "grad_norm": 0.5882550844092198,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 6827
    },
    {
      "epoch": 0.06828,
      "grad_norm": 0.7012417787965474,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 6828
    },
    {
      "epoch": 0.06829,
      "grad_norm": 0.8327606673237082,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 6829
    },
    {
      "epoch": 0.0683,
      "grad_norm": 0.8528489072611283,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 6830
    },
    {
      "epoch": 0.06831,
      "grad_norm": 0.7377079020858234,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 6831
    },
    {
      "epoch": 0.06832,
      "grad_norm": 0.6999485550253548,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 6832
    },
    {
      "epoch": 0.06833,
      "grad_norm": 0.7712453382887129,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 6833
    },
    {
      "epoch": 0.06834,
      "grad_norm": 0.9741841802598422,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 6834
    },
    {
      "epoch": 0.06835,
      "grad_norm": 1.0151191229361403,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 6835
    },
    {
      "epoch": 0.06836,
      "grad_norm": 0.9928135661127082,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 6836
    },
    {
      "epoch": 0.06837,
      "grad_norm": 1.0176229330263218,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 6837
    },
    {
      "epoch": 0.06838,
      "grad_norm": 0.8062582072461082,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 6838
    },
    {
      "epoch": 0.06839,
      "grad_norm": 1.0001169867154327,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 6839
    },
    {
      "epoch": 0.0684,
      "grad_norm": 1.2793247468125468,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 6840
    },
    {
      "epoch": 0.06841,
      "grad_norm": 0.9352650749931306,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 6841
    },
    {
      "epoch": 0.06842,
      "grad_norm": 0.873691761023133,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 6842
    },
    {
      "epoch": 0.06843,
      "grad_norm": 0.943397035481235,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 6843
    },
    {
      "epoch": 0.06844,
      "grad_norm": 1.2419554124329863,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 6844
    },
    {
      "epoch": 0.06845,
      "grad_norm": 0.7627006179431822,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 6845
    },
    {
      "epoch": 0.06846,
      "grad_norm": 0.7761262500841255,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 6846
    },
    {
      "epoch": 0.06847,
      "grad_norm": 0.9754798916992613,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6847
    },
    {
      "epoch": 0.06848,
      "grad_norm": 1.2952128685404893,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 6848
    },
    {
      "epoch": 0.06849,
      "grad_norm": 0.8183956076455051,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 6849
    },
    {
      "epoch": 0.0685,
      "grad_norm": 0.7298588998388881,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 6850
    },
    {
      "epoch": 0.06851,
      "grad_norm": 0.7720044301801032,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 6851
    },
    {
      "epoch": 0.06852,
      "grad_norm": 0.9117508720265394,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 6852
    },
    {
      "epoch": 0.06853,
      "grad_norm": 1.0533798726154795,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 6853
    },
    {
      "epoch": 0.06854,
      "grad_norm": 1.027625478858702,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 6854
    },
    {
      "epoch": 0.06855,
      "grad_norm": 0.9135813747015725,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 6855
    },
    {
      "epoch": 0.06856,
      "grad_norm": 0.9132419414958513,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 6856
    },
    {
      "epoch": 0.06857,
      "grad_norm": 0.9212345733341637,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 6857
    },
    {
      "epoch": 0.06858,
      "grad_norm": 0.8783800930563433,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 6858
    },
    {
      "epoch": 0.06859,
      "grad_norm": 0.6985668013167255,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 6859
    },
    {
      "epoch": 0.0686,
      "grad_norm": 0.6082758464053263,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 6860
    },
    {
      "epoch": 0.06861,
      "grad_norm": 0.5184883931489526,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 6861
    },
    {
      "epoch": 0.06862,
      "grad_norm": 0.5465866791407239,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 6862
    },
    {
      "epoch": 0.06863,
      "grad_norm": 0.6294940342821269,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 6863
    },
    {
      "epoch": 0.06864,
      "grad_norm": 0.819027721570075,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 6864
    },
    {
      "epoch": 0.06865,
      "grad_norm": 0.9537722441632093,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 6865
    },
    {
      "epoch": 0.06866,
      "grad_norm": 0.9579871191173921,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 6866
    },
    {
      "epoch": 0.06867,
      "grad_norm": 0.7874535648082873,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 6867
    },
    {
      "epoch": 0.06868,
      "grad_norm": 0.7207326410237167,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 6868
    },
    {
      "epoch": 0.06869,
      "grad_norm": 0.8921760331303482,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 6869
    },
    {
      "epoch": 0.0687,
      "grad_norm": 1.0524247874039665,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 6870
    },
    {
      "epoch": 0.06871,
      "grad_norm": 0.9959937136708353,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 6871
    },
    {
      "epoch": 0.06872,
      "grad_norm": 0.7683343036479461,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 6872
    },
    {
      "epoch": 0.06873,
      "grad_norm": 0.6815142284894331,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 6873
    },
    {
      "epoch": 0.06874,
      "grad_norm": 0.681503918141428,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6874
    },
    {
      "epoch": 0.06875,
      "grad_norm": 0.671319808930849,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 6875
    },
    {
      "epoch": 0.06876,
      "grad_norm": 0.6432032608458409,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 6876
    },
    {
      "epoch": 0.06877,
      "grad_norm": 0.681246314917716,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 6877
    },
    {
      "epoch": 0.06878,
      "grad_norm": 0.6862800830361652,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 6878
    },
    {
      "epoch": 0.06879,
      "grad_norm": 0.6785172447705301,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 6879
    },
    {
      "epoch": 0.0688,
      "grad_norm": 0.6523592214155444,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 6880
    },
    {
      "epoch": 0.06881,
      "grad_norm": 0.7257213543926093,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 6881
    },
    {
      "epoch": 0.06882,
      "grad_norm": 0.8060377847691599,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 6882
    },
    {
      "epoch": 0.06883,
      "grad_norm": 0.792778596117736,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 6883
    },
    {
      "epoch": 0.06884,
      "grad_norm": 0.6894553250899708,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 6884
    },
    {
      "epoch": 0.06885,
      "grad_norm": 0.7168844072539149,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 6885
    },
    {
      "epoch": 0.06886,
      "grad_norm": 0.7480318972493727,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 6886
    },
    {
      "epoch": 0.06887,
      "grad_norm": 0.7617857801277111,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 6887
    },
    {
      "epoch": 0.06888,
      "grad_norm": 0.861679021778327,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 6888
    },
    {
      "epoch": 0.06889,
      "grad_norm": 0.8718407312295603,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 6889
    },
    {
      "epoch": 0.0689,
      "grad_norm": 1.0130605630972407,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 6890
    },
    {
      "epoch": 0.06891,
      "grad_norm": 1.3688291187731094,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 6891
    },
    {
      "epoch": 0.06892,
      "grad_norm": 0.7799161275909664,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 6892
    },
    {
      "epoch": 0.06893,
      "grad_norm": 0.7045235058505965,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 6893
    },
    {
      "epoch": 0.06894,
      "grad_norm": 0.6684728435369993,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 6894
    },
    {
      "epoch": 0.06895,
      "grad_norm": 0.6945126017991131,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 6895
    },
    {
      "epoch": 0.06896,
      "grad_norm": 0.7482855504159163,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 6896
    },
    {
      "epoch": 0.06897,
      "grad_norm": 0.8746143480093368,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 6897
    },
    {
      "epoch": 0.06898,
      "grad_norm": 0.8993489919013548,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 6898
    },
    {
      "epoch": 0.06899,
      "grad_norm": 1.0857843804381477,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 6899
    },
    {
      "epoch": 0.069,
      "grad_norm": 1.0648749747194601,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 6900
    },
    {
      "epoch": 0.06901,
      "grad_norm": 0.9006029334239385,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 6901
    },
    {
      "epoch": 0.06902,
      "grad_norm": 0.8416068012359819,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 6902
    },
    {
      "epoch": 0.06903,
      "grad_norm": 1.006337020622227,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 6903
    },
    {
      "epoch": 0.06904,
      "grad_norm": 1.045403521519028,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 6904
    },
    {
      "epoch": 0.06905,
      "grad_norm": 1.0035690205519932,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 6905
    },
    {
      "epoch": 0.06906,
      "grad_norm": 1.0866394477802588,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 6906
    },
    {
      "epoch": 0.06907,
      "grad_norm": 0.9795306254573748,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 6907
    },
    {
      "epoch": 0.06908,
      "grad_norm": 0.921664480593037,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 6908
    },
    {
      "epoch": 0.06909,
      "grad_norm": 0.7625577740017703,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 6909
    },
    {
      "epoch": 0.0691,
      "grad_norm": 0.7611464664895455,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 6910
    },
    {
      "epoch": 0.06911,
      "grad_norm": 0.7481854469654662,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 6911
    },
    {
      "epoch": 0.06912,
      "grad_norm": 0.9007946005791158,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 6912
    },
    {
      "epoch": 0.06913,
      "grad_norm": 1.0181585830069129,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 6913
    },
    {
      "epoch": 0.06914,
      "grad_norm": 1.1870130455108574,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 6914
    },
    {
      "epoch": 0.06915,
      "grad_norm": 0.9236319005490047,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 6915
    },
    {
      "epoch": 0.06916,
      "grad_norm": 0.8302058218315981,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 6916
    },
    {
      "epoch": 0.06917,
      "grad_norm": 0.8449547905827482,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 6917
    },
    {
      "epoch": 0.06918,
      "grad_norm": 1.0042057313137678,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 6918
    },
    {
      "epoch": 0.06919,
      "grad_norm": 1.0948578415550971,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 6919
    },
    {
      "epoch": 0.0692,
      "grad_norm": 0.8651976841261361,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 6920
    },
    {
      "epoch": 0.06921,
      "grad_norm": 0.843906688583314,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 6921
    },
    {
      "epoch": 0.06922,
      "grad_norm": 0.7494458953280497,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 6922
    },
    {
      "epoch": 0.06923,
      "grad_norm": 0.6493697075436607,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 6923
    },
    {
      "epoch": 0.06924,
      "grad_norm": 0.5960115740450397,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 6924
    },
    {
      "epoch": 0.06925,
      "grad_norm": 0.5439383509954068,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 6925
    },
    {
      "epoch": 0.06926,
      "grad_norm": 0.5566009950142305,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 6926
    },
    {
      "epoch": 0.06927,
      "grad_norm": 0.5929207701665649,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 6927
    },
    {
      "epoch": 0.06928,
      "grad_norm": 0.6959312150674015,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 6928
    },
    {
      "epoch": 0.06929,
      "grad_norm": 0.9859706292355523,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 6929
    },
    {
      "epoch": 0.0693,
      "grad_norm": 1.2296463232455344,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 6930
    },
    {
      "epoch": 0.06931,
      "grad_norm": 0.610382372425946,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 6931
    },
    {
      "epoch": 0.06932,
      "grad_norm": 0.6524546502395882,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 6932
    },
    {
      "epoch": 0.06933,
      "grad_norm": 0.7923321668003368,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 6933
    },
    {
      "epoch": 0.06934,
      "grad_norm": 0.824582431481253,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 6934
    },
    {
      "epoch": 0.06935,
      "grad_norm": 0.8727764667568022,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 6935
    },
    {
      "epoch": 0.06936,
      "grad_norm": 0.8326712177028426,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 6936
    },
    {
      "epoch": 0.06937,
      "grad_norm": 0.6666993557326405,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 6937
    },
    {
      "epoch": 0.06938,
      "grad_norm": 0.7059003515766912,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 6938
    },
    {
      "epoch": 0.06939,
      "grad_norm": 0.7910834610590093,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 6939
    },
    {
      "epoch": 0.0694,
      "grad_norm": 0.755416937286377,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 6940
    },
    {
      "epoch": 0.06941,
      "grad_norm": 0.6585307567475693,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 6941
    },
    {
      "epoch": 0.06942,
      "grad_norm": 0.686786611772453,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 6942
    },
    {
      "epoch": 0.06943,
      "grad_norm": 0.6748173855765259,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 6943
    },
    {
      "epoch": 0.06944,
      "grad_norm": 0.8309412479867045,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 6944
    },
    {
      "epoch": 0.06945,
      "grad_norm": 0.8694006981278489,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 6945
    },
    {
      "epoch": 0.06946,
      "grad_norm": 0.9735883766446164,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 6946
    },
    {
      "epoch": 0.06947,
      "grad_norm": 0.9569405363329243,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 6947
    },
    {
      "epoch": 0.06948,
      "grad_norm": 0.9911481171151727,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6948
    },
    {
      "epoch": 0.06949,
      "grad_norm": 1.1434234191553132,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 6949
    },
    {
      "epoch": 0.0695,
      "grad_norm": 0.8225443853899382,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 6950
    },
    {
      "epoch": 0.06951,
      "grad_norm": 0.7771139496183902,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 6951
    },
    {
      "epoch": 0.06952,
      "grad_norm": 0.8827808082345756,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 6952
    },
    {
      "epoch": 0.06953,
      "grad_norm": 1.0344317147347284,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 6953
    },
    {
      "epoch": 0.06954,
      "grad_norm": 0.9087885968248024,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 6954
    },
    {
      "epoch": 0.06955,
      "grad_norm": 0.8948872903159871,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 6955
    },
    {
      "epoch": 0.06956,
      "grad_norm": 0.8801466457771367,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 6956
    },
    {
      "epoch": 0.06957,
      "grad_norm": 0.7407954654118085,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 6957
    },
    {
      "epoch": 0.06958,
      "grad_norm": 0.6110627923040244,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 6958
    },
    {
      "epoch": 0.06959,
      "grad_norm": 0.640461355001315,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 6959
    },
    {
      "epoch": 0.0696,
      "grad_norm": 0.7327706567363643,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 6960
    },
    {
      "epoch": 0.06961,
      "grad_norm": 0.8343628226365843,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 6961
    },
    {
      "epoch": 0.06962,
      "grad_norm": 0.8816790067550263,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 6962
    },
    {
      "epoch": 0.06963,
      "grad_norm": 1.029137591999145,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 6963
    },
    {
      "epoch": 0.06964,
      "grad_norm": 1.083546288789342,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 6964
    },
    {
      "epoch": 0.06965,
      "grad_norm": 1.0972587017397695,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 6965
    },
    {
      "epoch": 0.06966,
      "grad_norm": 0.8505331999829698,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 6966
    },
    {
      "epoch": 0.06967,
      "grad_norm": 0.6926913786836009,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 6967
    },
    {
      "epoch": 0.06968,
      "grad_norm": 0.7470348630991248,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 6968
    },
    {
      "epoch": 0.06969,
      "grad_norm": 0.7260002360079123,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 6969
    },
    {
      "epoch": 0.0697,
      "grad_norm": 0.6766317452834478,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 6970
    },
    {
      "epoch": 0.06971,
      "grad_norm": 0.6911223172142554,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 6971
    },
    {
      "epoch": 0.06972,
      "grad_norm": 0.6699177971563128,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 6972
    },
    {
      "epoch": 0.06973,
      "grad_norm": 0.694687314086879,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 6973
    },
    {
      "epoch": 0.06974,
      "grad_norm": 0.9488840337040796,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 6974
    },
    {
      "epoch": 0.06975,
      "grad_norm": 1.2258549533377359,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 6975
    },
    {
      "epoch": 0.06976,
      "grad_norm": 0.7740323563047606,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 6976
    },
    {
      "epoch": 0.06977,
      "grad_norm": 0.575041977095538,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 6977
    },
    {
      "epoch": 0.06978,
      "grad_norm": 0.728998390696451,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 6978
    },
    {
      "epoch": 0.06979,
      "grad_norm": 1.013780311861753,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 6979
    },
    {
      "epoch": 0.0698,
      "grad_norm": 1.1763620482776194,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 6980
    },
    {
      "epoch": 0.06981,
      "grad_norm": 0.815900592117822,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 6981
    },
    {
      "epoch": 0.06982,
      "grad_norm": 0.7122151392371082,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 6982
    },
    {
      "epoch": 0.06983,
      "grad_norm": 0.6205206572999914,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 6983
    },
    {
      "epoch": 0.06984,
      "grad_norm": 0.6501948602208925,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 6984
    },
    {
      "epoch": 0.06985,
      "grad_norm": 0.6936542630188876,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 6985
    },
    {
      "epoch": 0.06986,
      "grad_norm": 0.7630402930183048,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 6986
    },
    {
      "epoch": 0.06987,
      "grad_norm": 0.8307718661807078,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 6987
    },
    {
      "epoch": 0.06988,
      "grad_norm": 0.8259567088608353,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 6988
    },
    {
      "epoch": 0.06989,
      "grad_norm": 0.8383919214928979,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 6989
    },
    {
      "epoch": 0.0699,
      "grad_norm": 0.8394764313790253,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 6990
    },
    {
      "epoch": 0.06991,
      "grad_norm": 0.7966606342946648,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 6991
    },
    {
      "epoch": 0.06992,
      "grad_norm": 0.6596406930588895,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 6992
    },
    {
      "epoch": 0.06993,
      "grad_norm": 0.6295320798498751,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 6993
    },
    {
      "epoch": 0.06994,
      "grad_norm": 0.6843612644051112,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 6994
    },
    {
      "epoch": 0.06995,
      "grad_norm": 0.7889219514350806,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 6995
    },
    {
      "epoch": 0.06996,
      "grad_norm": 0.8402572785911753,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 6996
    },
    {
      "epoch": 0.06997,
      "grad_norm": 0.7656897240238402,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 6997
    },
    {
      "epoch": 0.06998,
      "grad_norm": 0.8838339815381696,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 6998
    },
    {
      "epoch": 0.06999,
      "grad_norm": 1.0521884576052078,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 6999
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3428947555527946,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 7000
    },
    {
      "epoch": 0.07001,
      "grad_norm": 0.7773020396247668,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 7001
    },
    {
      "epoch": 0.07002,
      "grad_norm": 0.6360859962376562,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 7002
    },
    {
      "epoch": 0.07003,
      "grad_norm": 0.71049436819903,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 7003
    },
    {
      "epoch": 0.07004,
      "grad_norm": 0.7207589244507122,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 7004
    },
    {
      "epoch": 0.07005,
      "grad_norm": 0.8029274080334843,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 7005
    },
    {
      "epoch": 0.07006,
      "grad_norm": 0.9621397059134545,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 7006
    },
    {
      "epoch": 0.07007,
      "grad_norm": 1.0548395954148817,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 7007
    },
    {
      "epoch": 0.07008,
      "grad_norm": 0.9292949868098707,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 7008
    },
    {
      "epoch": 0.07009,
      "grad_norm": 1.1487800774682821,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 7009
    },
    {
      "epoch": 0.0701,
      "grad_norm": 1.0137599916747433,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 7010
    },
    {
      "epoch": 0.07011,
      "grad_norm": 0.9682428887494761,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 7011
    },
    {
      "epoch": 0.07012,
      "grad_norm": 0.8914447047092996,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 7012
    },
    {
      "epoch": 0.07013,
      "grad_norm": 0.9039463647228102,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 7013
    },
    {
      "epoch": 0.07014,
      "grad_norm": 1.134635997819684,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 7014
    },
    {
      "epoch": 0.07015,
      "grad_norm": 1.031076929207217,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 7015
    },
    {
      "epoch": 0.07016,
      "grad_norm": 1.0116310847444454,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7016
    },
    {
      "epoch": 0.07017,
      "grad_norm": 1.005782143604759,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 7017
    },
    {
      "epoch": 0.07018,
      "grad_norm": 0.9578767335403211,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 7018
    },
    {
      "epoch": 0.07019,
      "grad_norm": 0.8935381942583269,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 7019
    },
    {
      "epoch": 0.0702,
      "grad_norm": 0.744623485765837,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 7020
    },
    {
      "epoch": 0.07021,
      "grad_norm": 0.801305747893465,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 7021
    },
    {
      "epoch": 0.07022,
      "grad_norm": 0.7083760005593353,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 7022
    },
    {
      "epoch": 0.07023,
      "grad_norm": 0.621039433188525,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 7023
    },
    {
      "epoch": 0.07024,
      "grad_norm": 0.5806951914320432,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 7024
    },
    {
      "epoch": 0.07025,
      "grad_norm": 0.6152547374263034,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 7025
    },
    {
      "epoch": 0.07026,
      "grad_norm": 0.6694178964744939,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 7026
    },
    {
      "epoch": 0.07027,
      "grad_norm": 0.9841630681827739,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 7027
    },
    {
      "epoch": 0.07028,
      "grad_norm": 1.200254843619916,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 7028
    },
    {
      "epoch": 0.07029,
      "grad_norm": 0.7174951451462476,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 7029
    },
    {
      "epoch": 0.0703,
      "grad_norm": 0.6728970196461604,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 7030
    },
    {
      "epoch": 0.07031,
      "grad_norm": 0.7070157774496603,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 7031
    },
    {
      "epoch": 0.07032,
      "grad_norm": 0.7387898161770746,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 7032
    },
    {
      "epoch": 0.07033,
      "grad_norm": 0.8901816616568644,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 7033
    },
    {
      "epoch": 0.07034,
      "grad_norm": 1.0284878793130177,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 7034
    },
    {
      "epoch": 0.07035,
      "grad_norm": 1.058686540675132,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 7035
    },
    {
      "epoch": 0.07036,
      "grad_norm": 0.9359391304214265,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 7036
    },
    {
      "epoch": 0.07037,
      "grad_norm": 0.9481158673322754,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 7037
    },
    {
      "epoch": 0.07038,
      "grad_norm": 0.8334487757118696,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 7038
    },
    {
      "epoch": 0.07039,
      "grad_norm": 0.6326431662983385,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 7039
    },
    {
      "epoch": 0.0704,
      "grad_norm": 0.702514824071445,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 7040
    },
    {
      "epoch": 0.07041,
      "grad_norm": 0.7400349502050564,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 7041
    },
    {
      "epoch": 0.07042,
      "grad_norm": 0.7099648121514438,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 7042
    },
    {
      "epoch": 0.07043,
      "grad_norm": 0.6483290540107833,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 7043
    },
    {
      "epoch": 0.07044,
      "grad_norm": 0.6620145629117125,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 7044
    },
    {
      "epoch": 0.07045,
      "grad_norm": 0.5978749302357278,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 7045
    },
    {
      "epoch": 0.07046,
      "grad_norm": 0.5888111174272556,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 7046
    },
    {
      "epoch": 0.07047,
      "grad_norm": 0.5784250123506132,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 7047
    },
    {
      "epoch": 0.07048,
      "grad_norm": 0.6072433980268588,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 7048
    },
    {
      "epoch": 0.07049,
      "grad_norm": 0.7094471050898912,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 7049
    },
    {
      "epoch": 0.0705,
      "grad_norm": 0.953711232001365,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 7050
    },
    {
      "epoch": 0.07051,
      "grad_norm": 1.4359270563806132,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 7051
    },
    {
      "epoch": 0.07052,
      "grad_norm": 0.6731996114775755,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 7052
    },
    {
      "epoch": 0.07053,
      "grad_norm": 0.8452128279853137,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 7053
    },
    {
      "epoch": 0.07054,
      "grad_norm": 0.9361316507438332,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 7054
    },
    {
      "epoch": 0.07055,
      "grad_norm": 1.0791674499784152,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 7055
    },
    {
      "epoch": 0.07056,
      "grad_norm": 0.9840050657142604,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 7056
    },
    {
      "epoch": 0.07057,
      "grad_norm": 0.7477218913081056,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 7057
    },
    {
      "epoch": 0.07058,
      "grad_norm": 0.6446497473373758,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 7058
    },
    {
      "epoch": 0.07059,
      "grad_norm": 0.7865830553034692,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 7059
    },
    {
      "epoch": 0.0706,
      "grad_norm": 1.0013324953332357,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 7060
    },
    {
      "epoch": 0.07061,
      "grad_norm": 0.9809126840231154,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 7061
    },
    {
      "epoch": 0.07062,
      "grad_norm": 1.0431446716134858,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 7062
    },
    {
      "epoch": 0.07063,
      "grad_norm": 0.9216298810073756,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 7063
    },
    {
      "epoch": 0.07064,
      "grad_norm": 0.9309447457188175,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 7064
    },
    {
      "epoch": 0.07065,
      "grad_norm": 0.9478643723606038,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 7065
    },
    {
      "epoch": 0.07066,
      "grad_norm": 0.8659841976692164,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 7066
    },
    {
      "epoch": 0.07067,
      "grad_norm": 0.820962997351535,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 7067
    },
    {
      "epoch": 0.07068,
      "grad_norm": 0.8086119331768552,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 7068
    },
    {
      "epoch": 0.07069,
      "grad_norm": 0.7230672452463993,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 7069
    },
    {
      "epoch": 0.0707,
      "grad_norm": 0.6999038291572481,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 7070
    },
    {
      "epoch": 0.07071,
      "grad_norm": 0.646981573011097,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 7071
    },
    {
      "epoch": 0.07072,
      "grad_norm": 0.6484005206388396,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 7072
    },
    {
      "epoch": 0.07073,
      "grad_norm": 0.650657762847781,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 7073
    },
    {
      "epoch": 0.07074,
      "grad_norm": 0.6775125603608778,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 7074
    },
    {
      "epoch": 0.07075,
      "grad_norm": 0.6083872877550612,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 7075
    },
    {
      "epoch": 0.07076,
      "grad_norm": 0.6452208544105434,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 7076
    },
    {
      "epoch": 0.07077,
      "grad_norm": 0.6649673970366402,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 7077
    },
    {
      "epoch": 0.07078,
      "grad_norm": 0.6783088752414806,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 7078
    },
    {
      "epoch": 0.07079,
      "grad_norm": 0.7710592718558568,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7079
    },
    {
      "epoch": 0.0708,
      "grad_norm": 0.958668145233532,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 7080
    },
    {
      "epoch": 0.07081,
      "grad_norm": 1.2533157696452137,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 7081
    },
    {
      "epoch": 0.07082,
      "grad_norm": 0.7341633652839094,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 7082
    },
    {
      "epoch": 0.07083,
      "grad_norm": 0.7125445429868206,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 7083
    },
    {
      "epoch": 0.07084,
      "grad_norm": 0.8323202978555945,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 7084
    },
    {
      "epoch": 0.07085,
      "grad_norm": 0.8775075656235146,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 7085
    },
    {
      "epoch": 0.07086,
      "grad_norm": 0.8415506665597984,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 7086
    },
    {
      "epoch": 0.07087,
      "grad_norm": 0.8886268254263243,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 7087
    },
    {
      "epoch": 0.07088,
      "grad_norm": 1.1007254498444334,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 7088
    },
    {
      "epoch": 0.07089,
      "grad_norm": 1.077560245637077,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7089
    },
    {
      "epoch": 0.0709,
      "grad_norm": 1.0164885049934373,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 7090
    },
    {
      "epoch": 0.07091,
      "grad_norm": 0.9408820922271797,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 7091
    },
    {
      "epoch": 0.07092,
      "grad_norm": 0.9903430667983368,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 7092
    },
    {
      "epoch": 0.07093,
      "grad_norm": 1.0927513972609133,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 7093
    },
    {
      "epoch": 0.07094,
      "grad_norm": 0.9057260963699147,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 7094
    },
    {
      "epoch": 0.07095,
      "grad_norm": 0.961594432402731,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 7095
    },
    {
      "epoch": 0.07096,
      "grad_norm": 0.9337093837213043,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 7096
    },
    {
      "epoch": 0.07097,
      "grad_norm": 0.8978322204376042,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 7097
    },
    {
      "epoch": 0.07098,
      "grad_norm": 0.8250215892500493,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 7098
    },
    {
      "epoch": 0.07099,
      "grad_norm": 0.7161931173541564,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 7099
    },
    {
      "epoch": 0.071,
      "grad_norm": 0.7049995268154481,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 7100
    },
    {
      "epoch": 0.07101,
      "grad_norm": 0.7656370792822328,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 7101
    },
    {
      "epoch": 0.07102,
      "grad_norm": 0.7875547696719094,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 7102
    },
    {
      "epoch": 0.07103,
      "grad_norm": 0.8916531396092162,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 7103
    },
    {
      "epoch": 0.07104,
      "grad_norm": 1.0114205653924595,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 7104
    },
    {
      "epoch": 0.07105,
      "grad_norm": 1.1741497513128247,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7105
    },
    {
      "epoch": 0.07106,
      "grad_norm": 0.7625197875421197,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 7106
    },
    {
      "epoch": 0.07107,
      "grad_norm": 0.6257457801610031,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 7107
    },
    {
      "epoch": 0.07108,
      "grad_norm": 0.722291112408107,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 7108
    },
    {
      "epoch": 0.07109,
      "grad_norm": 0.8199337468043636,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 7109
    },
    {
      "epoch": 0.0711,
      "grad_norm": 0.9809541172196407,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 7110
    },
    {
      "epoch": 0.07111,
      "grad_norm": 1.3461787795853564,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 7111
    },
    {
      "epoch": 0.07112,
      "grad_norm": 0.8890390770134524,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 7112
    },
    {
      "epoch": 0.07113,
      "grad_norm": 0.795752170005651,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 7113
    },
    {
      "epoch": 0.07114,
      "grad_norm": 0.7764623741527109,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 7114
    },
    {
      "epoch": 0.07115,
      "grad_norm": 0.7626873101755763,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 7115
    },
    {
      "epoch": 0.07116,
      "grad_norm": 0.709900661004303,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 7116
    },
    {
      "epoch": 0.07117,
      "grad_norm": 0.7447218585864936,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 7117
    },
    {
      "epoch": 0.07118,
      "grad_norm": 0.8145999556594223,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 7118
    },
    {
      "epoch": 0.07119,
      "grad_norm": 0.969296790631849,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 7119
    },
    {
      "epoch": 0.0712,
      "grad_norm": 1.110298419738427,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 7120
    },
    {
      "epoch": 0.07121,
      "grad_norm": 0.7765027521208008,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 7121
    },
    {
      "epoch": 0.07122,
      "grad_norm": 0.6976330840832459,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7122
    },
    {
      "epoch": 0.07123,
      "grad_norm": 0.7282209002398681,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 7123
    },
    {
      "epoch": 0.07124,
      "grad_norm": 0.8034457748599896,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 7124
    },
    {
      "epoch": 0.07125,
      "grad_norm": 0.9445995479508306,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 7125
    },
    {
      "epoch": 0.07126,
      "grad_norm": 1.1268646555573658,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 7126
    },
    {
      "epoch": 0.07127,
      "grad_norm": 0.8675499183490605,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 7127
    },
    {
      "epoch": 0.07128,
      "grad_norm": 0.7495770001240798,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 7128
    },
    {
      "epoch": 0.07129,
      "grad_norm": 0.7927069334940464,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 7129
    },
    {
      "epoch": 0.0713,
      "grad_norm": 0.8376734219621176,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 7130
    },
    {
      "epoch": 0.07131,
      "grad_norm": 0.9019229212023059,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 7131
    },
    {
      "epoch": 0.07132,
      "grad_norm": 0.958492302536442,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 7132
    },
    {
      "epoch": 0.07133,
      "grad_norm": 0.9703595503667748,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 7133
    },
    {
      "epoch": 0.07134,
      "grad_norm": 1.0020602010967854,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7134
    },
    {
      "epoch": 0.07135,
      "grad_norm": 0.8801413035960939,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 7135
    },
    {
      "epoch": 0.07136,
      "grad_norm": 0.7990610722753618,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 7136
    },
    {
      "epoch": 0.07137,
      "grad_norm": 0.793252959668317,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 7137
    },
    {
      "epoch": 0.07138,
      "grad_norm": 0.9362895477284106,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 7138
    },
    {
      "epoch": 0.07139,
      "grad_norm": 1.0323509656605805,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 7139
    },
    {
      "epoch": 0.0714,
      "grad_norm": 0.9495162380496068,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 7140
    },
    {
      "epoch": 0.07141,
      "grad_norm": 0.9575955550409052,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 7141
    },
    {
      "epoch": 0.07142,
      "grad_norm": 0.8744238864306674,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 7142
    },
    {
      "epoch": 0.07143,
      "grad_norm": 0.8469591921258507,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 7143
    },
    {
      "epoch": 0.07144,
      "grad_norm": 0.8348336779926869,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 7144
    },
    {
      "epoch": 0.07145,
      "grad_norm": 1.20426435156208,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 7145
    },
    {
      "epoch": 0.07146,
      "grad_norm": 0.9331877874118307,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 7146
    },
    {
      "epoch": 0.07147,
      "grad_norm": 0.7956424973642607,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 7147
    },
    {
      "epoch": 0.07148,
      "grad_norm": 0.6333045591280563,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 7148
    },
    {
      "epoch": 0.07149,
      "grad_norm": 0.6566262136153223,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 7149
    },
    {
      "epoch": 0.0715,
      "grad_norm": 0.7318802572187021,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 7150
    },
    {
      "epoch": 0.07151,
      "grad_norm": 0.7829355969313044,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 7151
    },
    {
      "epoch": 0.07152,
      "grad_norm": 0.8583613260253945,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 7152
    },
    {
      "epoch": 0.07153,
      "grad_norm": 0.92982352203367,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 7153
    },
    {
      "epoch": 0.07154,
      "grad_norm": 0.9644437837333677,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 7154
    },
    {
      "epoch": 0.07155,
      "grad_norm": 0.978358778668515,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 7155
    },
    {
      "epoch": 0.07156,
      "grad_norm": 0.7507926446560859,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 7156
    },
    {
      "epoch": 0.07157,
      "grad_norm": 0.6945396009037457,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 7157
    },
    {
      "epoch": 0.07158,
      "grad_norm": 0.6202573792874552,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 7158
    },
    {
      "epoch": 0.07159,
      "grad_norm": 0.6406745858919619,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 7159
    },
    {
      "epoch": 0.0716,
      "grad_norm": 0.6759721738325275,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 7160
    },
    {
      "epoch": 0.07161,
      "grad_norm": 0.6773350977546571,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 7161
    },
    {
      "epoch": 0.07162,
      "grad_norm": 0.8134936196821616,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 7162
    },
    {
      "epoch": 0.07163,
      "grad_norm": 1.0238124509153523,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7163
    },
    {
      "epoch": 0.07164,
      "grad_norm": 1.0689438667971007,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 7164
    },
    {
      "epoch": 0.07165,
      "grad_norm": 0.7928160801351044,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 7165
    },
    {
      "epoch": 0.07166,
      "grad_norm": 0.7238005610299342,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 7166
    },
    {
      "epoch": 0.07167,
      "grad_norm": 0.7943353284460141,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 7167
    },
    {
      "epoch": 0.07168,
      "grad_norm": 0.9366138493792299,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 7168
    },
    {
      "epoch": 0.07169,
      "grad_norm": 0.9317917412995604,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 7169
    },
    {
      "epoch": 0.0717,
      "grad_norm": 0.9498234616222482,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 7170
    },
    {
      "epoch": 0.07171,
      "grad_norm": 1.0293570458877537,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 7171
    },
    {
      "epoch": 0.07172,
      "grad_norm": 0.9825020297184559,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 7172
    },
    {
      "epoch": 0.07173,
      "grad_norm": 0.8679097068915133,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 7173
    },
    {
      "epoch": 0.07174,
      "grad_norm": 0.8695176732805833,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 7174
    },
    {
      "epoch": 0.07175,
      "grad_norm": 0.9453745541920484,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 7175
    },
    {
      "epoch": 0.07176,
      "grad_norm": 0.9192625870625667,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 7176
    },
    {
      "epoch": 0.07177,
      "grad_norm": 0.9892316081377227,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 7177
    },
    {
      "epoch": 0.07178,
      "grad_norm": 1.1173320244326939,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 7178
    },
    {
      "epoch": 0.07179,
      "grad_norm": 0.9062788363818038,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 7179
    },
    {
      "epoch": 0.0718,
      "grad_norm": 0.8000050025573026,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 7180
    },
    {
      "epoch": 0.07181,
      "grad_norm": 0.873878962358314,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 7181
    },
    {
      "epoch": 0.07182,
      "grad_norm": 0.8940125268309896,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 7182
    },
    {
      "epoch": 0.07183,
      "grad_norm": 0.9553838594697793,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 7183
    },
    {
      "epoch": 0.07184,
      "grad_norm": 0.8409034686100991,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 7184
    },
    {
      "epoch": 0.07185,
      "grad_norm": 0.8965763690380733,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 7185
    },
    {
      "epoch": 0.07186,
      "grad_norm": 1.0738828307599333,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 7186
    },
    {
      "epoch": 0.07187,
      "grad_norm": 1.121879584063851,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 7187
    },
    {
      "epoch": 0.07188,
      "grad_norm": 0.7967451613183767,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 7188
    },
    {
      "epoch": 0.07189,
      "grad_norm": 0.8216455172743733,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 7189
    },
    {
      "epoch": 0.0719,
      "grad_norm": 0.8712547518094813,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 7190
    },
    {
      "epoch": 0.07191,
      "grad_norm": 0.7876295918843358,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 7191
    },
    {
      "epoch": 0.07192,
      "grad_norm": 0.8009874929822709,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 7192
    },
    {
      "epoch": 0.07193,
      "grad_norm": 0.8465208583922181,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 7193
    },
    {
      "epoch": 0.07194,
      "grad_norm": 0.919178707859649,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 7194
    },
    {
      "epoch": 0.07195,
      "grad_norm": 1.0483728768665097,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 7195
    },
    {
      "epoch": 0.07196,
      "grad_norm": 1.0529880709135624,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7196
    },
    {
      "epoch": 0.07197,
      "grad_norm": 0.8929728066328368,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 7197
    },
    {
      "epoch": 0.07198,
      "grad_norm": 0.8510797490053401,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 7198
    },
    {
      "epoch": 0.07199,
      "grad_norm": 0.7701251238667544,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 7199
    },
    {
      "epoch": 0.072,
      "grad_norm": 0.5992356375970112,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 7200
    },
    {
      "epoch": 0.07201,
      "grad_norm": 0.6127454964693743,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 7201
    },
    {
      "epoch": 0.07202,
      "grad_norm": 0.5750069845188995,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 7202
    },
    {
      "epoch": 0.07203,
      "grad_norm": 0.5644356454850563,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 7203
    },
    {
      "epoch": 0.07204,
      "grad_norm": 0.607948596229715,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 7204
    },
    {
      "epoch": 0.07205,
      "grad_norm": 0.575336251782507,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 7205
    },
    {
      "epoch": 0.07206,
      "grad_norm": 0.6206756982347824,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 7206
    },
    {
      "epoch": 0.07207,
      "grad_norm": 0.6744817053594698,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 7207
    },
    {
      "epoch": 0.07208,
      "grad_norm": 0.7220807677197786,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 7208
    },
    {
      "epoch": 0.07209,
      "grad_norm": 0.8377268449123019,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 7209
    },
    {
      "epoch": 0.0721,
      "grad_norm": 1.0721743343425194,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 7210
    },
    {
      "epoch": 0.07211,
      "grad_norm": 0.9661201701564104,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 7211
    },
    {
      "epoch": 0.07212,
      "grad_norm": 0.7690576613887181,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 7212
    },
    {
      "epoch": 0.07213,
      "grad_norm": 0.581038878715819,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 7213
    },
    {
      "epoch": 0.07214,
      "grad_norm": 0.6435911406628065,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 7214
    },
    {
      "epoch": 0.07215,
      "grad_norm": 0.8429785330520374,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 7215
    },
    {
      "epoch": 0.07216,
      "grad_norm": 1.0387688679965494,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 7216
    },
    {
      "epoch": 0.07217,
      "grad_norm": 1.0443475300961371,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 7217
    },
    {
      "epoch": 0.07218,
      "grad_norm": 0.8473614781155715,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 7218
    },
    {
      "epoch": 0.07219,
      "grad_norm": 0.8315491811072131,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 7219
    },
    {
      "epoch": 0.0722,
      "grad_norm": 1.0296189666178754,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 7220
    },
    {
      "epoch": 0.07221,
      "grad_norm": 1.058643563273535,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 7221
    },
    {
      "epoch": 0.07222,
      "grad_norm": 0.8979290937756685,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 7222
    },
    {
      "epoch": 0.07223,
      "grad_norm": 0.7842228557729956,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 7223
    },
    {
      "epoch": 0.07224,
      "grad_norm": 0.774784015172213,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 7224
    },
    {
      "epoch": 0.07225,
      "grad_norm": 0.77416880062085,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 7225
    },
    {
      "epoch": 0.07226,
      "grad_norm": 0.8594028278800052,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 7226
    },
    {
      "epoch": 0.07227,
      "grad_norm": 0.9339883501438095,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 7227
    },
    {
      "epoch": 0.07228,
      "grad_norm": 1.009238406341426,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 7228
    },
    {
      "epoch": 0.07229,
      "grad_norm": 0.9248237524022925,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 7229
    },
    {
      "epoch": 0.0723,
      "grad_norm": 0.7879348638628303,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 7230
    },
    {
      "epoch": 0.07231,
      "grad_norm": 0.8125590881523109,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 7231
    },
    {
      "epoch": 0.07232,
      "grad_norm": 0.7397244009053785,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7232
    },
    {
      "epoch": 0.07233,
      "grad_norm": 0.8631670605854277,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 7233
    },
    {
      "epoch": 0.07234,
      "grad_norm": 0.9590374718321888,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 7234
    },
    {
      "epoch": 0.07235,
      "grad_norm": 1.1821140388987355,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 7235
    },
    {
      "epoch": 0.07236,
      "grad_norm": 1.0912993445758663,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 7236
    },
    {
      "epoch": 0.07237,
      "grad_norm": 1.0186661817583151,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 7237
    },
    {
      "epoch": 0.07238,
      "grad_norm": 0.8827875731907374,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 7238
    },
    {
      "epoch": 0.07239,
      "grad_norm": 1.0101708637839817,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 7239
    },
    {
      "epoch": 0.0724,
      "grad_norm": 0.9436996667559744,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 7240
    },
    {
      "epoch": 0.07241,
      "grad_norm": 0.924594063482519,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 7241
    },
    {
      "epoch": 0.07242,
      "grad_norm": 1.0534829249112354,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 7242
    },
    {
      "epoch": 0.07243,
      "grad_norm": 0.9856885879894006,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 7243
    },
    {
      "epoch": 0.07244,
      "grad_norm": 0.8968354162645473,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 7244
    },
    {
      "epoch": 0.07245,
      "grad_norm": 0.7432909812609705,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 7245
    },
    {
      "epoch": 0.07246,
      "grad_norm": 0.7581401567213042,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 7246
    },
    {
      "epoch": 0.07247,
      "grad_norm": 0.7827779191397117,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 7247
    },
    {
      "epoch": 0.07248,
      "grad_norm": 0.8301715949958813,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 7248
    },
    {
      "epoch": 0.07249,
      "grad_norm": 0.8525741406819738,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 7249
    },
    {
      "epoch": 0.0725,
      "grad_norm": 0.8907335082596,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 7250
    },
    {
      "epoch": 0.07251,
      "grad_norm": 0.8802924249074248,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 7251
    },
    {
      "epoch": 0.07252,
      "grad_norm": 0.9494861985507634,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 7252
    },
    {
      "epoch": 0.07253,
      "grad_norm": 0.9816725505864071,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 7253
    },
    {
      "epoch": 0.07254,
      "grad_norm": 1.0443814177090267,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 7254
    },
    {
      "epoch": 0.07255,
      "grad_norm": 0.9747578800346165,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 7255
    },
    {
      "epoch": 0.07256,
      "grad_norm": 0.9623975370695055,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 7256
    },
    {
      "epoch": 0.07257,
      "grad_norm": 0.8844084755688933,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 7257
    },
    {
      "epoch": 0.07258,
      "grad_norm": 0.7873250461581538,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 7258
    },
    {
      "epoch": 0.07259,
      "grad_norm": 0.7048185689672526,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7259
    },
    {
      "epoch": 0.0726,
      "grad_norm": 0.7212279689840713,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 7260
    },
    {
      "epoch": 0.07261,
      "grad_norm": 0.6900512375900392,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 7261
    },
    {
      "epoch": 0.07262,
      "grad_norm": 0.7136129301757733,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 7262
    },
    {
      "epoch": 0.07263,
      "grad_norm": 0.7469633157532889,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 7263
    },
    {
      "epoch": 0.07264,
      "grad_norm": 0.8012211214960027,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 7264
    },
    {
      "epoch": 0.07265,
      "grad_norm": 0.7552911178426823,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 7265
    },
    {
      "epoch": 0.07266,
      "grad_norm": 0.7077347824210926,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 7266
    },
    {
      "epoch": 0.07267,
      "grad_norm": 0.6760989582415581,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 7267
    },
    {
      "epoch": 0.07268,
      "grad_norm": 0.7437332060079201,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 7268
    },
    {
      "epoch": 0.07269,
      "grad_norm": 0.9444827297437564,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 7269
    },
    {
      "epoch": 0.0727,
      "grad_norm": 1.1220594233552328,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 7270
    },
    {
      "epoch": 0.07271,
      "grad_norm": 0.8200379597299912,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 7271
    },
    {
      "epoch": 0.07272,
      "grad_norm": 0.7596490807886606,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 7272
    },
    {
      "epoch": 0.07273,
      "grad_norm": 0.7884580146314756,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 7273
    },
    {
      "epoch": 0.07274,
      "grad_norm": 0.7179271579858106,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 7274
    },
    {
      "epoch": 0.07275,
      "grad_norm": 0.6409693363218573,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 7275
    },
    {
      "epoch": 0.07276,
      "grad_norm": 0.561214096824011,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 7276
    },
    {
      "epoch": 0.07277,
      "grad_norm": 0.5256694600157773,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 7277
    },
    {
      "epoch": 0.07278,
      "grad_norm": 0.513949681102309,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 7278
    },
    {
      "epoch": 0.07279,
      "grad_norm": 0.49570529414604636,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 7279
    },
    {
      "epoch": 0.0728,
      "grad_norm": 0.6204492806369919,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 7280
    },
    {
      "epoch": 0.07281,
      "grad_norm": 0.9083869384535228,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 7281
    },
    {
      "epoch": 0.07282,
      "grad_norm": 1.2773527610872655,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 7282
    },
    {
      "epoch": 0.07283,
      "grad_norm": 0.7627102174193735,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 7283
    },
    {
      "epoch": 0.07284,
      "grad_norm": 0.6303760922439519,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 7284
    },
    {
      "epoch": 0.07285,
      "grad_norm": 0.703121523726995,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 7285
    },
    {
      "epoch": 0.07286,
      "grad_norm": 0.7516732999345842,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 7286
    },
    {
      "epoch": 0.07287,
      "grad_norm": 0.8402680521780728,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7287
    },
    {
      "epoch": 0.07288,
      "grad_norm": 0.8639792163628891,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 7288
    },
    {
      "epoch": 0.07289,
      "grad_norm": 0.9106770231204078,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 7289
    },
    {
      "epoch": 0.0729,
      "grad_norm": 0.8569039805553732,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 7290
    },
    {
      "epoch": 0.07291,
      "grad_norm": 0.8741502241373251,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 7291
    },
    {
      "epoch": 0.07292,
      "grad_norm": 0.9093181247050962,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 7292
    },
    {
      "epoch": 0.07293,
      "grad_norm": 1.0546113868958757,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 7293
    },
    {
      "epoch": 0.07294,
      "grad_norm": 1.0193664307418668,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 7294
    },
    {
      "epoch": 0.07295,
      "grad_norm": 1.0910195655187844,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 7295
    },
    {
      "epoch": 0.07296,
      "grad_norm": 0.8335499181973589,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 7296
    },
    {
      "epoch": 0.07297,
      "grad_norm": 0.6580676089467326,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 7297
    },
    {
      "epoch": 0.07298,
      "grad_norm": 0.5832643079122204,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 7298
    },
    {
      "epoch": 0.07299,
      "grad_norm": 0.6303194883819615,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 7299
    },
    {
      "epoch": 0.073,
      "grad_norm": 0.6469636223827426,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 7300
    },
    {
      "epoch": 0.07301,
      "grad_norm": 0.6597994798980442,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 7301
    },
    {
      "epoch": 0.07302,
      "grad_norm": 0.6982812827344924,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 7302
    },
    {
      "epoch": 0.07303,
      "grad_norm": 0.7027447626804699,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 7303
    },
    {
      "epoch": 0.07304,
      "grad_norm": 0.7898566798558829,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 7304
    },
    {
      "epoch": 0.07305,
      "grad_norm": 0.8619601714981096,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 7305
    },
    {
      "epoch": 0.07306,
      "grad_norm": 1.0874296421242524,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 7306
    },
    {
      "epoch": 0.07307,
      "grad_norm": 1.1513689835246426,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 7307
    },
    {
      "epoch": 0.07308,
      "grad_norm": 0.8638122639112095,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 7308
    },
    {
      "epoch": 0.07309,
      "grad_norm": 0.8255250333131889,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 7309
    },
    {
      "epoch": 0.0731,
      "grad_norm": 0.9337338907492057,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 7310
    },
    {
      "epoch": 0.07311,
      "grad_norm": 0.9258041196216139,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 7311
    },
    {
      "epoch": 0.07312,
      "grad_norm": 0.951937894172149,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 7312
    },
    {
      "epoch": 0.07313,
      "grad_norm": 1.0078253665907535,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 7313
    },
    {
      "epoch": 0.07314,
      "grad_norm": 0.901224372401486,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 7314
    },
    {
      "epoch": 0.07315,
      "grad_norm": 0.7944852254095126,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 7315
    },
    {
      "epoch": 0.07316,
      "grad_norm": 0.783421376743893,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 7316
    },
    {
      "epoch": 0.07317,
      "grad_norm": 0.8368936906337524,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 7317
    },
    {
      "epoch": 0.07318,
      "grad_norm": 1.033174150909009,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 7318
    },
    {
      "epoch": 0.07319,
      "grad_norm": 1.27108174493455,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 7319
    },
    {
      "epoch": 0.0732,
      "grad_norm": 0.7406462978297551,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 7320
    },
    {
      "epoch": 0.07321,
      "grad_norm": 0.7025965454631923,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 7321
    },
    {
      "epoch": 0.07322,
      "grad_norm": 0.8144605025587333,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7322
    },
    {
      "epoch": 0.07323,
      "grad_norm": 0.8053334048255597,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 7323
    },
    {
      "epoch": 0.07324,
      "grad_norm": 0.775372831647096,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 7324
    },
    {
      "epoch": 0.07325,
      "grad_norm": 0.8357997029570158,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 7325
    },
    {
      "epoch": 0.07326,
      "grad_norm": 0.8489307022162469,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 7326
    },
    {
      "epoch": 0.07327,
      "grad_norm": 0.8228446709367002,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 7327
    },
    {
      "epoch": 0.07328,
      "grad_norm": 0.7838796840554372,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 7328
    },
    {
      "epoch": 0.07329,
      "grad_norm": 0.7251399802311324,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 7329
    },
    {
      "epoch": 0.0733,
      "grad_norm": 0.766770759186389,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 7330
    },
    {
      "epoch": 0.07331,
      "grad_norm": 0.943369746260282,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 7331
    },
    {
      "epoch": 0.07332,
      "grad_norm": 1.106302603979319,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 7332
    },
    {
      "epoch": 0.07333,
      "grad_norm": 0.9429802103066166,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 7333
    },
    {
      "epoch": 0.07334,
      "grad_norm": 1.0320307959064938,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 7334
    },
    {
      "epoch": 0.07335,
      "grad_norm": 0.9938214659378108,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 7335
    },
    {
      "epoch": 0.07336,
      "grad_norm": 1.0914354610999535,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 7336
    },
    {
      "epoch": 0.07337,
      "grad_norm": 1.1269724240813392,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 7337
    },
    {
      "epoch": 0.07338,
      "grad_norm": 0.8570775179663218,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7338
    },
    {
      "epoch": 0.07339,
      "grad_norm": 0.8031931079538867,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 7339
    },
    {
      "epoch": 0.0734,
      "grad_norm": 0.8935576371259307,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 7340
    },
    {
      "epoch": 0.07341,
      "grad_norm": 0.9083095502174247,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 7341
    },
    {
      "epoch": 0.07342,
      "grad_norm": 0.9576486900412259,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 7342
    },
    {
      "epoch": 0.07343,
      "grad_norm": 0.9182147356345189,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 7343
    },
    {
      "epoch": 0.07344,
      "grad_norm": 0.8421322703250478,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 7344
    },
    {
      "epoch": 0.07345,
      "grad_norm": 0.7537817256778269,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 7345
    },
    {
      "epoch": 0.07346,
      "grad_norm": 0.7033688234808623,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 7346
    },
    {
      "epoch": 0.07347,
      "grad_norm": 0.7413990535788959,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 7347
    },
    {
      "epoch": 0.07348,
      "grad_norm": 0.7773942391883343,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 7348
    },
    {
      "epoch": 0.07349,
      "grad_norm": 0.9329173997319489,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 7349
    },
    {
      "epoch": 0.0735,
      "grad_norm": 1.093903209357276,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 7350
    },
    {
      "epoch": 0.07351,
      "grad_norm": 0.8109483676035741,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 7351
    },
    {
      "epoch": 0.07352,
      "grad_norm": 0.7939494918683481,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 7352
    },
    {
      "epoch": 0.07353,
      "grad_norm": 0.7869344084686208,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 7353
    },
    {
      "epoch": 0.07354,
      "grad_norm": 0.7788006156502201,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 7354
    },
    {
      "epoch": 0.07355,
      "grad_norm": 0.8552053349081179,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 7355
    },
    {
      "epoch": 0.07356,
      "grad_norm": 1.0518730489917372,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 7356
    },
    {
      "epoch": 0.07357,
      "grad_norm": 1.3412824434070465,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7357
    },
    {
      "epoch": 0.07358,
      "grad_norm": 0.9208430362293964,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 7358
    },
    {
      "epoch": 0.07359,
      "grad_norm": 0.8344769448824529,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 7359
    },
    {
      "epoch": 0.0736,
      "grad_norm": 0.6793410562113039,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 7360
    },
    {
      "epoch": 0.07361,
      "grad_norm": 0.6267991527813198,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 7361
    },
    {
      "epoch": 0.07362,
      "grad_norm": 0.690161715844307,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 7362
    },
    {
      "epoch": 0.07363,
      "grad_norm": 0.7869927998336663,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 7363
    },
    {
      "epoch": 0.07364,
      "grad_norm": 0.8027534998077103,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 7364
    },
    {
      "epoch": 0.07365,
      "grad_norm": 0.6955286691239817,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 7365
    },
    {
      "epoch": 0.07366,
      "grad_norm": 0.7372804832228109,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 7366
    },
    {
      "epoch": 0.07367,
      "grad_norm": 0.6394799490289566,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 7367
    },
    {
      "epoch": 0.07368,
      "grad_norm": 0.509023284762404,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 7368
    },
    {
      "epoch": 0.07369,
      "grad_norm": 0.5616436196591941,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 7369
    },
    {
      "epoch": 0.0737,
      "grad_norm": 0.6425475395508851,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 7370
    },
    {
      "epoch": 0.07371,
      "grad_norm": 0.7368588731675243,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 7371
    },
    {
      "epoch": 0.07372,
      "grad_norm": 0.9085719201340855,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 7372
    },
    {
      "epoch": 0.07373,
      "grad_norm": 1.2408545702032314,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 7373
    },
    {
      "epoch": 0.07374,
      "grad_norm": 0.7374902067189273,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 7374
    },
    {
      "epoch": 0.07375,
      "grad_norm": 0.6163993980166736,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 7375
    },
    {
      "epoch": 0.07376,
      "grad_norm": 0.6423268145367569,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 7376
    },
    {
      "epoch": 0.07377,
      "grad_norm": 0.7533167250559268,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 7377
    },
    {
      "epoch": 0.07378,
      "grad_norm": 0.8584299528911417,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 7378
    },
    {
      "epoch": 0.07379,
      "grad_norm": 1.0096294225047744,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 7379
    },
    {
      "epoch": 0.0738,
      "grad_norm": 1.0223154674326658,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 7380
    },
    {
      "epoch": 0.07381,
      "grad_norm": 0.8907126146182821,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 7381
    },
    {
      "epoch": 0.07382,
      "grad_norm": 0.7207053071921802,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 7382
    },
    {
      "epoch": 0.07383,
      "grad_norm": 0.9145179387624196,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 7383
    },
    {
      "epoch": 0.07384,
      "grad_norm": 1.0098905122982873,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 7384
    },
    {
      "epoch": 0.07385,
      "grad_norm": 1.198216983974359,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 7385
    },
    {
      "epoch": 0.07386,
      "grad_norm": 0.7055995485988885,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7386
    },
    {
      "epoch": 0.07387,
      "grad_norm": 0.6341184291168993,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 7387
    },
    {
      "epoch": 0.07388,
      "grad_norm": 0.6305102045443429,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7388
    },
    {
      "epoch": 0.07389,
      "grad_norm": 0.7729018243590356,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 7389
    },
    {
      "epoch": 0.0739,
      "grad_norm": 1.046784353921616,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 7390
    },
    {
      "epoch": 0.07391,
      "grad_norm": 1.003334204394043,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 7391
    },
    {
      "epoch": 0.07392,
      "grad_norm": 0.867653951162738,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 7392
    },
    {
      "epoch": 0.07393,
      "grad_norm": 0.7970634630521443,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 7393
    },
    {
      "epoch": 0.07394,
      "grad_norm": 0.9057360666683115,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 7394
    },
    {
      "epoch": 0.07395,
      "grad_norm": 0.9547098856968925,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 7395
    },
    {
      "epoch": 0.07396,
      "grad_norm": 0.9234933056161622,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7396
    },
    {
      "epoch": 0.07397,
      "grad_norm": 1.0368747780491556,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 7397
    },
    {
      "epoch": 0.07398,
      "grad_norm": 1.0403397267434613,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 7398
    },
    {
      "epoch": 0.07399,
      "grad_norm": 0.9544666764070975,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 7399
    },
    {
      "epoch": 0.074,
      "grad_norm": 0.8872295611618851,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 7400
    },
    {
      "epoch": 0.07401,
      "grad_norm": 0.9340892887170521,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 7401
    },
    {
      "epoch": 0.07402,
      "grad_norm": 0.8784569523984895,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 7402
    },
    {
      "epoch": 0.07403,
      "grad_norm": 0.8660676350594653,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 7403
    },
    {
      "epoch": 0.07404,
      "grad_norm": 0.9009419907324966,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 7404
    },
    {
      "epoch": 0.07405,
      "grad_norm": 0.8106715007698777,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 7405
    },
    {
      "epoch": 0.07406,
      "grad_norm": 0.807586011085066,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 7406
    },
    {
      "epoch": 0.07407,
      "grad_norm": 0.8628667672682003,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 7407
    },
    {
      "epoch": 0.07408,
      "grad_norm": 0.9415613388154968,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 7408
    },
    {
      "epoch": 0.07409,
      "grad_norm": 0.9481965047839528,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 7409
    },
    {
      "epoch": 0.0741,
      "grad_norm": 1.0223140305985579,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 7410
    },
    {
      "epoch": 0.07411,
      "grad_norm": 1.1261250470212472,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 7411
    },
    {
      "epoch": 0.07412,
      "grad_norm": 0.8452821990703466,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 7412
    },
    {
      "epoch": 0.07413,
      "grad_norm": 0.7070647319303838,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 7413
    },
    {
      "epoch": 0.07414,
      "grad_norm": 0.6602409738515731,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 7414
    },
    {
      "epoch": 0.07415,
      "grad_norm": 0.6813711085208377,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 7415
    },
    {
      "epoch": 0.07416,
      "grad_norm": 0.7813248550198105,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 7416
    },
    {
      "epoch": 0.07417,
      "grad_norm": 0.7903284361595316,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 7417
    },
    {
      "epoch": 0.07418,
      "grad_norm": 0.8776997591398797,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 7418
    },
    {
      "epoch": 0.07419,
      "grad_norm": 0.9346374935990277,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 7419
    },
    {
      "epoch": 0.0742,
      "grad_norm": 0.8167496983343525,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 7420
    },
    {
      "epoch": 0.07421,
      "grad_norm": 0.7740749507997752,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 7421
    },
    {
      "epoch": 0.07422,
      "grad_norm": 0.7797923880019133,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 7422
    },
    {
      "epoch": 0.07423,
      "grad_norm": 0.7246765372540556,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 7423
    },
    {
      "epoch": 0.07424,
      "grad_norm": 0.7739364594038591,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 7424
    },
    {
      "epoch": 0.07425,
      "grad_norm": 0.8561033384244009,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 7425
    },
    {
      "epoch": 0.07426,
      "grad_norm": 1.0873991072382838,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 7426
    },
    {
      "epoch": 0.07427,
      "grad_norm": 1.0969759409269335,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 7427
    },
    {
      "epoch": 0.07428,
      "grad_norm": 0.7321830844983641,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 7428
    },
    {
      "epoch": 0.07429,
      "grad_norm": 0.6644474298927815,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 7429
    },
    {
      "epoch": 0.0743,
      "grad_norm": 0.6883336057236158,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 7430
    },
    {
      "epoch": 0.07431,
      "grad_norm": 0.7448399088605852,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 7431
    },
    {
      "epoch": 0.07432,
      "grad_norm": 0.8377015855878724,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 7432
    },
    {
      "epoch": 0.07433,
      "grad_norm": 0.8455466117001035,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 7433
    },
    {
      "epoch": 0.07434,
      "grad_norm": 0.8038998713277352,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 7434
    },
    {
      "epoch": 0.07435,
      "grad_norm": 0.7628824415481451,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 7435
    },
    {
      "epoch": 0.07436,
      "grad_norm": 0.8471666681896792,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 7436
    },
    {
      "epoch": 0.07437,
      "grad_norm": 1.022193249171131,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 7437
    },
    {
      "epoch": 0.07438,
      "grad_norm": 0.954279749528698,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 7438
    },
    {
      "epoch": 0.07439,
      "grad_norm": 0.9359744466674262,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7439
    },
    {
      "epoch": 0.0744,
      "grad_norm": 0.985750832333834,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 7440
    },
    {
      "epoch": 0.07441,
      "grad_norm": 1.0387568082669436,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 7441
    },
    {
      "epoch": 0.07442,
      "grad_norm": 0.9354486578980633,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 7442
    },
    {
      "epoch": 0.07443,
      "grad_norm": 0.8937562700947757,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7443
    },
    {
      "epoch": 0.07444,
      "grad_norm": 0.7862746074501281,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 7444
    },
    {
      "epoch": 0.07445,
      "grad_norm": 0.7580214165836215,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 7445
    },
    {
      "epoch": 0.07446,
      "grad_norm": 0.826490781957247,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 7446
    },
    {
      "epoch": 0.07447,
      "grad_norm": 0.78561092743045,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 7447
    },
    {
      "epoch": 0.07448,
      "grad_norm": 0.8720738755090752,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 7448
    },
    {
      "epoch": 0.07449,
      "grad_norm": 1.1356579541218612,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 7449
    },
    {
      "epoch": 0.0745,
      "grad_norm": 1.194362940520485,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 7450
    },
    {
      "epoch": 0.07451,
      "grad_norm": 0.8425811609172174,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 7451
    },
    {
      "epoch": 0.07452,
      "grad_norm": 0.9146392058972959,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 7452
    },
    {
      "epoch": 0.07453,
      "grad_norm": 1.0026955605148151,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 7453
    },
    {
      "epoch": 0.07454,
      "grad_norm": 1.0112038508522885,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 7454
    },
    {
      "epoch": 0.07455,
      "grad_norm": 1.168191155504265,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 7455
    },
    {
      "epoch": 0.07456,
      "grad_norm": 0.7449918001899162,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 7456
    },
    {
      "epoch": 0.07457,
      "grad_norm": 0.7986467621349783,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 7457
    },
    {
      "epoch": 0.07458,
      "grad_norm": 0.6606779110991354,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 7458
    },
    {
      "epoch": 0.07459,
      "grad_norm": 0.74600487207374,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 7459
    },
    {
      "epoch": 0.0746,
      "grad_norm": 0.7705606291540048,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 7460
    },
    {
      "epoch": 0.07461,
      "grad_norm": 0.8369060675928409,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 7461
    },
    {
      "epoch": 0.07462,
      "grad_norm": 0.9604504411667463,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 7462
    },
    {
      "epoch": 0.07463,
      "grad_norm": 1.018164117686845,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 7463
    },
    {
      "epoch": 0.07464,
      "grad_norm": 0.8462415172519894,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 7464
    },
    {
      "epoch": 0.07465,
      "grad_norm": 0.7280333713094582,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 7465
    },
    {
      "epoch": 0.07466,
      "grad_norm": 0.6527117875066514,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 7466
    },
    {
      "epoch": 0.07467,
      "grad_norm": 0.7283692800044493,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 7467
    },
    {
      "epoch": 0.07468,
      "grad_norm": 0.7286232701262282,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 7468
    },
    {
      "epoch": 0.07469,
      "grad_norm": 0.7554001099044895,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 7469
    },
    {
      "epoch": 0.0747,
      "grad_norm": 0.6494769448278457,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 7470
    },
    {
      "epoch": 0.07471,
      "grad_norm": 0.582239848281011,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 7471
    },
    {
      "epoch": 0.07472,
      "grad_norm": 0.6054171618519452,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 7472
    },
    {
      "epoch": 0.07473,
      "grad_norm": 0.5800597140442034,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 7473
    },
    {
      "epoch": 0.07474,
      "grad_norm": 0.6418775672195469,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 7474
    },
    {
      "epoch": 0.07475,
      "grad_norm": 0.8209219098567517,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 7475
    },
    {
      "epoch": 0.07476,
      "grad_norm": 1.0927372326738478,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 7476
    },
    {
      "epoch": 0.07477,
      "grad_norm": 1.1813159129321458,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 7477
    },
    {
      "epoch": 0.07478,
      "grad_norm": 0.7297682403726699,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 7478
    },
    {
      "epoch": 0.07479,
      "grad_norm": 0.6410942452556853,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 7479
    },
    {
      "epoch": 0.0748,
      "grad_norm": 0.8698329147222141,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 7480
    },
    {
      "epoch": 0.07481,
      "grad_norm": 1.0082944565315157,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 7481
    },
    {
      "epoch": 0.07482,
      "grad_norm": 0.9628391840892444,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 7482
    },
    {
      "epoch": 0.07483,
      "grad_norm": 0.8641925302412802,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 7483
    },
    {
      "epoch": 0.07484,
      "grad_norm": 0.7533282082846949,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 7484
    },
    {
      "epoch": 0.07485,
      "grad_norm": 0.7703693706278185,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 7485
    },
    {
      "epoch": 0.07486,
      "grad_norm": 0.8291299861969627,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 7486
    },
    {
      "epoch": 0.07487,
      "grad_norm": 0.9151584554564175,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 7487
    },
    {
      "epoch": 0.07488,
      "grad_norm": 1.0172079023902525,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 7488
    },
    {
      "epoch": 0.07489,
      "grad_norm": 0.985794974411023,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 7489
    },
    {
      "epoch": 0.0749,
      "grad_norm": 0.7878742848581212,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 7490
    },
    {
      "epoch": 0.07491,
      "grad_norm": 0.7562126231512927,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 7491
    },
    {
      "epoch": 0.07492,
      "grad_norm": 0.7316445879371943,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 7492
    },
    {
      "epoch": 0.07493,
      "grad_norm": 0.7015263206527975,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 7493
    },
    {
      "epoch": 0.07494,
      "grad_norm": 0.7094674720974151,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 7494
    },
    {
      "epoch": 0.07495,
      "grad_norm": 0.8005067312090948,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 7495
    },
    {
      "epoch": 0.07496,
      "grad_norm": 1.0221046723780693,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 7496
    },
    {
      "epoch": 0.07497,
      "grad_norm": 1.1782769790208016,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 7497
    },
    {
      "epoch": 0.07498,
      "grad_norm": 0.8031446202765408,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 7498
    },
    {
      "epoch": 0.07499,
      "grad_norm": 0.7861635761156994,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 7499
    },
    {
      "epoch": 0.075,
      "grad_norm": 0.7398205403713403,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 7500
    },
    {
      "epoch": 0.07501,
      "grad_norm": 0.8734378274009751,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 7501
    },
    {
      "epoch": 0.07502,
      "grad_norm": 0.8830879959651403,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 7502
    },
    {
      "epoch": 0.07503,
      "grad_norm": 0.8281346250649004,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 7503
    },
    {
      "epoch": 0.07504,
      "grad_norm": 0.8072671794855871,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 7504
    },
    {
      "epoch": 0.07505,
      "grad_norm": 0.8666820809776766,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 7505
    },
    {
      "epoch": 0.07506,
      "grad_norm": 1.0055552931832379,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 7506
    },
    {
      "epoch": 0.07507,
      "grad_norm": 0.9690975893829946,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 7507
    },
    {
      "epoch": 0.07508,
      "grad_norm": 0.9561250009010052,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 7508
    },
    {
      "epoch": 0.07509,
      "grad_norm": 1.0877559050066379,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7509
    },
    {
      "epoch": 0.0751,
      "grad_norm": 1.0027154307652972,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 7510
    },
    {
      "epoch": 0.07511,
      "grad_norm": 1.0165739974701828,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 7511
    },
    {
      "epoch": 0.07512,
      "grad_norm": 1.0057419295722907,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 7512
    },
    {
      "epoch": 0.07513,
      "grad_norm": 0.9433174355296526,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 7513
    },
    {
      "epoch": 0.07514,
      "grad_norm": 1.0313875668855543,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 7514
    },
    {
      "epoch": 0.07515,
      "grad_norm": 1.0147222353644576,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 7515
    },
    {
      "epoch": 0.07516,
      "grad_norm": 0.9245083215799488,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 7516
    },
    {
      "epoch": 0.07517,
      "grad_norm": 0.8259475443651306,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 7517
    },
    {
      "epoch": 0.07518,
      "grad_norm": 0.7587280215699812,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 7518
    },
    {
      "epoch": 0.07519,
      "grad_norm": 0.8033857629693775,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 7519
    },
    {
      "epoch": 0.0752,
      "grad_norm": 0.6891155253835879,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 7520
    },
    {
      "epoch": 0.07521,
      "grad_norm": 0.6668629114515775,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 7521
    },
    {
      "epoch": 0.07522,
      "grad_norm": 0.6646628095819561,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 7522
    },
    {
      "epoch": 0.07523,
      "grad_norm": 0.6810426992617306,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 7523
    },
    {
      "epoch": 0.07524,
      "grad_norm": 0.5957519856201996,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 7524
    },
    {
      "epoch": 0.07525,
      "grad_norm": 0.6304965471697975,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 7525
    },
    {
      "epoch": 0.07526,
      "grad_norm": 0.7044075744869454,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 7526
    },
    {
      "epoch": 0.07527,
      "grad_norm": 0.8184433815359451,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 7527
    },
    {
      "epoch": 0.07528,
      "grad_norm": 0.8298479343150946,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 7528
    },
    {
      "epoch": 0.07529,
      "grad_norm": 0.7555105418948039,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 7529
    },
    {
      "epoch": 0.0753,
      "grad_norm": 0.62755290323034,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7530
    },
    {
      "epoch": 0.07531,
      "grad_norm": 0.6259328988343698,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 7531
    },
    {
      "epoch": 0.07532,
      "grad_norm": 0.7024264501793085,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 7532
    },
    {
      "epoch": 0.07533,
      "grad_norm": 0.7976246816821072,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 7533
    },
    {
      "epoch": 0.07534,
      "grad_norm": 0.9491149711829591,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 7534
    },
    {
      "epoch": 0.07535,
      "grad_norm": 0.9702246438073734,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 7535
    },
    {
      "epoch": 0.07536,
      "grad_norm": 0.8833175400088759,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 7536
    },
    {
      "epoch": 0.07537,
      "grad_norm": 1.0566640021173932,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 7537
    },
    {
      "epoch": 0.07538,
      "grad_norm": 1.0524050892778398,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 7538
    },
    {
      "epoch": 0.07539,
      "grad_norm": 0.9461058512204453,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 7539
    },
    {
      "epoch": 0.0754,
      "grad_norm": 0.8671896332276945,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 7540
    },
    {
      "epoch": 0.07541,
      "grad_norm": 0.9370741051157787,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 7541
    },
    {
      "epoch": 0.07542,
      "grad_norm": 0.7655033736787794,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 7542
    },
    {
      "epoch": 0.07543,
      "grad_norm": 0.9638854244860205,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 7543
    },
    {
      "epoch": 0.07544,
      "grad_norm": 1.151245583846751,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 7544
    },
    {
      "epoch": 0.07545,
      "grad_norm": 1.045720500701298,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 7545
    },
    {
      "epoch": 0.07546,
      "grad_norm": 0.8508437267643452,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 7546
    },
    {
      "epoch": 0.07547,
      "grad_norm": 0.8758265191573648,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 7547
    },
    {
      "epoch": 0.07548,
      "grad_norm": 1.0308655234346675,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 7548
    },
    {
      "epoch": 0.07549,
      "grad_norm": 1.1222648672525313,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 7549
    },
    {
      "epoch": 0.0755,
      "grad_norm": 1.0347446053481122,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 7550
    },
    {
      "epoch": 0.07551,
      "grad_norm": 0.8951008390390673,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 7551
    },
    {
      "epoch": 0.07552,
      "grad_norm": 0.867925078672574,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 7552
    },
    {
      "epoch": 0.07553,
      "grad_norm": 0.8525722259854441,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 7553
    },
    {
      "epoch": 0.07554,
      "grad_norm": 0.9114938094018381,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 7554
    },
    {
      "epoch": 0.07555,
      "grad_norm": 0.9344191657970983,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 7555
    },
    {
      "epoch": 0.07556,
      "grad_norm": 1.0145169044440256,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 7556
    },
    {
      "epoch": 0.07557,
      "grad_norm": 1.008268955978688,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 7557
    },
    {
      "epoch": 0.07558,
      "grad_norm": 0.9718709803852565,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7558
    },
    {
      "epoch": 0.07559,
      "grad_norm": 1.056325136300485,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 7559
    },
    {
      "epoch": 0.0756,
      "grad_norm": 0.9740547207080791,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 7560
    },
    {
      "epoch": 0.07561,
      "grad_norm": 0.9083193763007056,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 7561
    },
    {
      "epoch": 0.07562,
      "grad_norm": 0.8448439316271258,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 7562
    },
    {
      "epoch": 0.07563,
      "grad_norm": 0.9065754257807024,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 7563
    },
    {
      "epoch": 0.07564,
      "grad_norm": 0.8560571418992496,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 7564
    },
    {
      "epoch": 0.07565,
      "grad_norm": 0.8274853334246817,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 7565
    },
    {
      "epoch": 0.07566,
      "grad_norm": 0.8273427382041143,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 7566
    },
    {
      "epoch": 0.07567,
      "grad_norm": 0.8118867625373324,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 7567
    },
    {
      "epoch": 0.07568,
      "grad_norm": 0.6681343989273923,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 7568
    },
    {
      "epoch": 0.07569,
      "grad_norm": 0.5803814722946369,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 7569
    },
    {
      "epoch": 0.0757,
      "grad_norm": 0.5510764495582413,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 7570
    },
    {
      "epoch": 0.07571,
      "grad_norm": 0.6627010068288498,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 7571
    },
    {
      "epoch": 0.07572,
      "grad_norm": 0.6455742336359287,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 7572
    },
    {
      "epoch": 0.07573,
      "grad_norm": 0.7410211759586838,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 7573
    },
    {
      "epoch": 0.07574,
      "grad_norm": 0.7937533836416206,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 7574
    },
    {
      "epoch": 0.07575,
      "grad_norm": 0.7151578069937585,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 7575
    },
    {
      "epoch": 0.07576,
      "grad_norm": 0.6757485878022385,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 7576
    },
    {
      "epoch": 0.07577,
      "grad_norm": 0.7377828856053927,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 7577
    },
    {
      "epoch": 0.07578,
      "grad_norm": 0.9295354995808791,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 7578
    },
    {
      "epoch": 0.07579,
      "grad_norm": 1.1308441721629587,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 7579
    },
    {
      "epoch": 0.0758,
      "grad_norm": 0.9537849948969392,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 7580
    },
    {
      "epoch": 0.07581,
      "grad_norm": 0.9648991721783183,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 7581
    },
    {
      "epoch": 0.07582,
      "grad_norm": 0.926845739832675,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 7582
    },
    {
      "epoch": 0.07583,
      "grad_norm": 0.894884269204141,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 7583
    },
    {
      "epoch": 0.07584,
      "grad_norm": 0.8924573565922338,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 7584
    },
    {
      "epoch": 0.07585,
      "grad_norm": 0.9888084614023944,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 7585
    },
    {
      "epoch": 0.07586,
      "grad_norm": 1.1114233056422533,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 7586
    },
    {
      "epoch": 0.07587,
      "grad_norm": 0.8144693201284348,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 7587
    },
    {
      "epoch": 0.07588,
      "grad_norm": 0.8305575966505442,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 7588
    },
    {
      "epoch": 0.07589,
      "grad_norm": 0.8498435104769524,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 7589
    },
    {
      "epoch": 0.0759,
      "grad_norm": 0.8188810092794875,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7590
    },
    {
      "epoch": 0.07591,
      "grad_norm": 0.7698614390712419,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 7591
    },
    {
      "epoch": 0.07592,
      "grad_norm": 0.5909542261154095,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 7592
    },
    {
      "epoch": 0.07593,
      "grad_norm": 0.6169915126769421,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 7593
    },
    {
      "epoch": 0.07594,
      "grad_norm": 0.6183226824646945,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 7594
    },
    {
      "epoch": 0.07595,
      "grad_norm": 0.6295258139827735,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 7595
    },
    {
      "epoch": 0.07596,
      "grad_norm": 0.5647538604253549,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 7596
    },
    {
      "epoch": 0.07597,
      "grad_norm": 0.5488719464323173,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 7597
    },
    {
      "epoch": 0.07598,
      "grad_norm": 0.5190318492649775,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 7598
    },
    {
      "epoch": 0.07599,
      "grad_norm": 0.5944308410673468,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 7599
    },
    {
      "epoch": 0.076,
      "grad_norm": 0.6645839697886495,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 7600
    },
    {
      "epoch": 0.07601,
      "grad_norm": 0.9191079894373272,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 7601
    },
    {
      "epoch": 0.07602,
      "grad_norm": 1.2588970299218996,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 7602
    },
    {
      "epoch": 0.07603,
      "grad_norm": 0.9641328600344435,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 7603
    },
    {
      "epoch": 0.07604,
      "grad_norm": 1.0002475506914794,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 7604
    },
    {
      "epoch": 0.07605,
      "grad_norm": 0.9041839404022576,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 7605
    },
    {
      "epoch": 0.07606,
      "grad_norm": 0.8495943134663567,
      "learning_rate": 0.003,
      "loss": 4.1653,
      "step": 7606
    },
    {
      "epoch": 0.07607,
      "grad_norm": 0.800744853771563,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 7607
    },
    {
      "epoch": 0.07608,
      "grad_norm": 0.7315846867215643,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 7608
    },
    {
      "epoch": 0.07609,
      "grad_norm": 0.6443317008016733,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 7609
    },
    {
      "epoch": 0.0761,
      "grad_norm": 0.6907982352452112,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 7610
    },
    {
      "epoch": 0.07611,
      "grad_norm": 0.857723111168813,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 7611
    },
    {
      "epoch": 0.07612,
      "grad_norm": 1.0992578425673403,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 7612
    },
    {
      "epoch": 0.07613,
      "grad_norm": 0.9526866939763878,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 7613
    },
    {
      "epoch": 0.07614,
      "grad_norm": 0.7803177657645249,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 7614
    },
    {
      "epoch": 0.07615,
      "grad_norm": 0.7081390563241489,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 7615
    },
    {
      "epoch": 0.07616,
      "grad_norm": 0.7252993228592656,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 7616
    },
    {
      "epoch": 0.07617,
      "grad_norm": 0.7475989787312242,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 7617
    },
    {
      "epoch": 0.07618,
      "grad_norm": 0.7573711078639473,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 7618
    },
    {
      "epoch": 0.07619,
      "grad_norm": 1.0541262061063688,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 7619
    },
    {
      "epoch": 0.0762,
      "grad_norm": 1.1895056232233716,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 7620
    },
    {
      "epoch": 0.07621,
      "grad_norm": 0.8742054868789058,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 7621
    },
    {
      "epoch": 0.07622,
      "grad_norm": 0.8444291423398331,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 7622
    },
    {
      "epoch": 0.07623,
      "grad_norm": 0.8922019849127748,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 7623
    },
    {
      "epoch": 0.07624,
      "grad_norm": 0.803083822831012,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 7624
    },
    {
      "epoch": 0.07625,
      "grad_norm": 0.7385150275748038,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 7625
    },
    {
      "epoch": 0.07626,
      "grad_norm": 0.6711142915115071,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 7626
    },
    {
      "epoch": 0.07627,
      "grad_norm": 0.812731441964542,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 7627
    },
    {
      "epoch": 0.07628,
      "grad_norm": 0.8485441320364617,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 7628
    },
    {
      "epoch": 0.07629,
      "grad_norm": 0.8442461477013375,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 7629
    },
    {
      "epoch": 0.0763,
      "grad_norm": 0.8784878480520506,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 7630
    },
    {
      "epoch": 0.07631,
      "grad_norm": 0.8831785338162262,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 7631
    },
    {
      "epoch": 0.07632,
      "grad_norm": 0.8267222772195404,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 7632
    },
    {
      "epoch": 0.07633,
      "grad_norm": 0.8474903779702966,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 7633
    },
    {
      "epoch": 0.07634,
      "grad_norm": 0.8527662645558675,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 7634
    },
    {
      "epoch": 0.07635,
      "grad_norm": 0.7539827705409249,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 7635
    },
    {
      "epoch": 0.07636,
      "grad_norm": 0.7979423225524935,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 7636
    },
    {
      "epoch": 0.07637,
      "grad_norm": 0.9896367410513814,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 7637
    },
    {
      "epoch": 0.07638,
      "grad_norm": 1.23693930151839,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7638
    },
    {
      "epoch": 0.07639,
      "grad_norm": 0.8951052515565378,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7639
    },
    {
      "epoch": 0.0764,
      "grad_norm": 0.9661295603094655,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 7640
    },
    {
      "epoch": 0.07641,
      "grad_norm": 1.0381140300663276,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 7641
    },
    {
      "epoch": 0.07642,
      "grad_norm": 0.8657178028651203,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 7642
    },
    {
      "epoch": 0.07643,
      "grad_norm": 0.7585312382956262,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 7643
    },
    {
      "epoch": 0.07644,
      "grad_norm": 0.8424021918604633,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 7644
    },
    {
      "epoch": 0.07645,
      "grad_norm": 1.0547127355883397,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 7645
    },
    {
      "epoch": 0.07646,
      "grad_norm": 1.0501940720210954,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 7646
    },
    {
      "epoch": 0.07647,
      "grad_norm": 1.0089732479233036,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 7647
    },
    {
      "epoch": 0.07648,
      "grad_norm": 0.9700747028421894,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 7648
    },
    {
      "epoch": 0.07649,
      "grad_norm": 0.8831977035938575,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 7649
    },
    {
      "epoch": 0.0765,
      "grad_norm": 0.7985816216412632,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 7650
    },
    {
      "epoch": 0.07651,
      "grad_norm": 0.7812576551419401,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 7651
    },
    {
      "epoch": 0.07652,
      "grad_norm": 0.863619677787746,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 7652
    },
    {
      "epoch": 0.07653,
      "grad_norm": 1.033937451094598,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 7653
    },
    {
      "epoch": 0.07654,
      "grad_norm": 1.0055330888443008,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 7654
    },
    {
      "epoch": 0.07655,
      "grad_norm": 0.961819876357498,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 7655
    },
    {
      "epoch": 0.07656,
      "grad_norm": 0.8931684820162469,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7656
    },
    {
      "epoch": 0.07657,
      "grad_norm": 0.8072740880367048,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 7657
    },
    {
      "epoch": 0.07658,
      "grad_norm": 0.8027134754140552,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 7658
    },
    {
      "epoch": 0.07659,
      "grad_norm": 0.8125620251732755,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 7659
    },
    {
      "epoch": 0.0766,
      "grad_norm": 0.7220469429506398,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 7660
    },
    {
      "epoch": 0.07661,
      "grad_norm": 0.715054421826916,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 7661
    },
    {
      "epoch": 0.07662,
      "grad_norm": 0.7493362311346902,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 7662
    },
    {
      "epoch": 0.07663,
      "grad_norm": 0.8714009180373697,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 7663
    },
    {
      "epoch": 0.07664,
      "grad_norm": 0.9901602291385457,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 7664
    },
    {
      "epoch": 0.07665,
      "grad_norm": 1.0074073013289961,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 7665
    },
    {
      "epoch": 0.07666,
      "grad_norm": 0.953985834092227,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 7666
    },
    {
      "epoch": 0.07667,
      "grad_norm": 0.9324987692217996,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 7667
    },
    {
      "epoch": 0.07668,
      "grad_norm": 0.9181312329680767,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 7668
    },
    {
      "epoch": 0.07669,
      "grad_norm": 1.0636826877445138,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 7669
    },
    {
      "epoch": 0.0767,
      "grad_norm": 1.1446064632763406,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 7670
    },
    {
      "epoch": 0.07671,
      "grad_norm": 1.1178580340464093,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 7671
    },
    {
      "epoch": 0.07672,
      "grad_norm": 1.0178300133444778,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 7672
    },
    {
      "epoch": 0.07673,
      "grad_norm": 1.2016623636923163,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 7673
    },
    {
      "epoch": 0.07674,
      "grad_norm": 0.8240382566249935,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7674
    },
    {
      "epoch": 0.07675,
      "grad_norm": 0.8028023964885046,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 7675
    },
    {
      "epoch": 0.07676,
      "grad_norm": 0.8231955050487717,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 7676
    },
    {
      "epoch": 0.07677,
      "grad_norm": 0.7115485042666164,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 7677
    },
    {
      "epoch": 0.07678,
      "grad_norm": 0.6936604353261299,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 7678
    },
    {
      "epoch": 0.07679,
      "grad_norm": 0.8220436974564044,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 7679
    },
    {
      "epoch": 0.0768,
      "grad_norm": 0.8824999345056386,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 7680
    },
    {
      "epoch": 0.07681,
      "grad_norm": 1.074897714216047,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 7681
    },
    {
      "epoch": 0.07682,
      "grad_norm": 1.1253938246986106,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 7682
    },
    {
      "epoch": 0.07683,
      "grad_norm": 0.8445187360108007,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 7683
    },
    {
      "epoch": 0.07684,
      "grad_norm": 0.6711564131616432,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 7684
    },
    {
      "epoch": 0.07685,
      "grad_norm": 0.8607141168125007,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 7685
    },
    {
      "epoch": 0.07686,
      "grad_norm": 0.990613718786834,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 7686
    },
    {
      "epoch": 0.07687,
      "grad_norm": 1.2572290057789604,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 7687
    },
    {
      "epoch": 0.07688,
      "grad_norm": 0.8215779647578441,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 7688
    },
    {
      "epoch": 0.07689,
      "grad_norm": 0.6891001411750275,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 7689
    },
    {
      "epoch": 0.0769,
      "grad_norm": 0.6166117212029528,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 7690
    },
    {
      "epoch": 0.07691,
      "grad_norm": 0.5677700647993942,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 7691
    },
    {
      "epoch": 0.07692,
      "grad_norm": 0.7724897932547341,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 7692
    },
    {
      "epoch": 0.07693,
      "grad_norm": 1.0012551819046691,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 7693
    },
    {
      "epoch": 0.07694,
      "grad_norm": 1.1907585124420144,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 7694
    },
    {
      "epoch": 0.07695,
      "grad_norm": 0.7010072537738787,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 7695
    },
    {
      "epoch": 0.07696,
      "grad_norm": 0.6208430723305157,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 7696
    },
    {
      "epoch": 0.07697,
      "grad_norm": 0.6196777653103698,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 7697
    },
    {
      "epoch": 0.07698,
      "grad_norm": 0.700128726476087,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 7698
    },
    {
      "epoch": 0.07699,
      "grad_norm": 0.6686751131176947,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 7699
    },
    {
      "epoch": 0.077,
      "grad_norm": 0.7342255589833349,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7700
    },
    {
      "epoch": 0.07701,
      "grad_norm": 0.756854445847,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 7701
    },
    {
      "epoch": 0.07702,
      "grad_norm": 0.88266150116856,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 7702
    },
    {
      "epoch": 0.07703,
      "grad_norm": 0.9388488634222688,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 7703
    },
    {
      "epoch": 0.07704,
      "grad_norm": 0.8712150471498338,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 7704
    },
    {
      "epoch": 0.07705,
      "grad_norm": 0.7905979661669048,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 7705
    },
    {
      "epoch": 0.07706,
      "grad_norm": 0.8554435529838232,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7706
    },
    {
      "epoch": 0.07707,
      "grad_norm": 0.950211545207407,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7707
    },
    {
      "epoch": 0.07708,
      "grad_norm": 0.9941562838099525,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 7708
    },
    {
      "epoch": 0.07709,
      "grad_norm": 0.9405303791524773,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 7709
    },
    {
      "epoch": 0.0771,
      "grad_norm": 0.8363759541957757,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 7710
    },
    {
      "epoch": 0.07711,
      "grad_norm": 0.733875954007421,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 7711
    },
    {
      "epoch": 0.07712,
      "grad_norm": 0.6560253796119527,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 7712
    },
    {
      "epoch": 0.07713,
      "grad_norm": 0.6759517680276719,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 7713
    },
    {
      "epoch": 0.07714,
      "grad_norm": 1.0143647874314274,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 7714
    },
    {
      "epoch": 0.07715,
      "grad_norm": 1.232675077798191,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 7715
    },
    {
      "epoch": 0.07716,
      "grad_norm": 0.8270400357958588,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 7716
    },
    {
      "epoch": 0.07717,
      "grad_norm": 0.7208914160909232,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 7717
    },
    {
      "epoch": 0.07718,
      "grad_norm": 0.7014829688376748,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 7718
    },
    {
      "epoch": 0.07719,
      "grad_norm": 0.7014046527951289,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 7719
    },
    {
      "epoch": 0.0772,
      "grad_norm": 0.746499130996974,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 7720
    },
    {
      "epoch": 0.07721,
      "grad_norm": 0.7004391345875081,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 7721
    },
    {
      "epoch": 0.07722,
      "grad_norm": 0.8408283424603382,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 7722
    },
    {
      "epoch": 0.07723,
      "grad_norm": 1.110388676041958,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 7723
    },
    {
      "epoch": 0.07724,
      "grad_norm": 0.9338754060953655,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 7724
    },
    {
      "epoch": 0.07725,
      "grad_norm": 0.8624819322150851,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 7725
    },
    {
      "epoch": 0.07726,
      "grad_norm": 0.8365430711688485,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 7726
    },
    {
      "epoch": 0.07727,
      "grad_norm": 0.8249395305754468,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 7727
    },
    {
      "epoch": 0.07728,
      "grad_norm": 0.7844147351205248,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 7728
    },
    {
      "epoch": 0.07729,
      "grad_norm": 0.7973591326922866,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 7729
    },
    {
      "epoch": 0.0773,
      "grad_norm": 0.7225338885856214,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 7730
    },
    {
      "epoch": 0.07731,
      "grad_norm": 0.73397785517986,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 7731
    },
    {
      "epoch": 0.07732,
      "grad_norm": 0.7538960206734072,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 7732
    },
    {
      "epoch": 0.07733,
      "grad_norm": 0.9299182384721361,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 7733
    },
    {
      "epoch": 0.07734,
      "grad_norm": 1.1092844376628306,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 7734
    },
    {
      "epoch": 0.07735,
      "grad_norm": 0.9361685548610317,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 7735
    },
    {
      "epoch": 0.07736,
      "grad_norm": 0.8101474819136633,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 7736
    },
    {
      "epoch": 0.07737,
      "grad_norm": 0.8203573852291894,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 7737
    },
    {
      "epoch": 0.07738,
      "grad_norm": 1.169891214994567,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 7738
    },
    {
      "epoch": 0.07739,
      "grad_norm": 1.2019762033842287,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 7739
    },
    {
      "epoch": 0.0774,
      "grad_norm": 0.9191096276408175,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 7740
    },
    {
      "epoch": 0.07741,
      "grad_norm": 0.8019281817921025,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 7741
    },
    {
      "epoch": 0.07742,
      "grad_norm": 0.9593283919942299,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 7742
    },
    {
      "epoch": 0.07743,
      "grad_norm": 1.0190652900336852,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 7743
    },
    {
      "epoch": 0.07744,
      "grad_norm": 1.1716932492490297,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 7744
    },
    {
      "epoch": 0.07745,
      "grad_norm": 1.0937600539399255,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 7745
    },
    {
      "epoch": 0.07746,
      "grad_norm": 0.9602047963501511,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 7746
    },
    {
      "epoch": 0.07747,
      "grad_norm": 0.7943597561692926,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 7747
    },
    {
      "epoch": 0.07748,
      "grad_norm": 0.7757213618554147,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 7748
    },
    {
      "epoch": 0.07749,
      "grad_norm": 0.791302635220645,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 7749
    },
    {
      "epoch": 0.0775,
      "grad_norm": 0.794756302153131,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 7750
    },
    {
      "epoch": 0.07751,
      "grad_norm": 0.8515667360808754,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 7751
    },
    {
      "epoch": 0.07752,
      "grad_norm": 0.8271299496217495,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 7752
    },
    {
      "epoch": 0.07753,
      "grad_norm": 0.8707061104805746,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 7753
    },
    {
      "epoch": 0.07754,
      "grad_norm": 0.8698105492883605,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 7754
    },
    {
      "epoch": 0.07755,
      "grad_norm": 0.9757058824840338,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 7755
    },
    {
      "epoch": 0.07756,
      "grad_norm": 1.0996218275953191,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 7756
    },
    {
      "epoch": 0.07757,
      "grad_norm": 1.0046296348538857,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 7757
    },
    {
      "epoch": 0.07758,
      "grad_norm": 1.088474067480404,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 7758
    },
    {
      "epoch": 0.07759,
      "grad_norm": 0.6838363547835872,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 7759
    },
    {
      "epoch": 0.0776,
      "grad_norm": 0.6387940327331244,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 7760
    },
    {
      "epoch": 0.07761,
      "grad_norm": 0.6801202292166391,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 7761
    },
    {
      "epoch": 0.07762,
      "grad_norm": 0.7524204138228072,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 7762
    },
    {
      "epoch": 0.07763,
      "grad_norm": 0.8422682117877314,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 7763
    },
    {
      "epoch": 0.07764,
      "grad_norm": 0.9464082829191608,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 7764
    },
    {
      "epoch": 0.07765,
      "grad_norm": 1.0400578267063363,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 7765
    },
    {
      "epoch": 0.07766,
      "grad_norm": 0.8570240484946164,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 7766
    },
    {
      "epoch": 0.07767,
      "grad_norm": 0.7886939048717178,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 7767
    },
    {
      "epoch": 0.07768,
      "grad_norm": 0.7327426017118384,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 7768
    },
    {
      "epoch": 0.07769,
      "grad_norm": 0.7224926065792775,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 7769
    },
    {
      "epoch": 0.0777,
      "grad_norm": 0.6812310474755818,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 7770
    },
    {
      "epoch": 0.07771,
      "grad_norm": 0.6297454383181756,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7771
    },
    {
      "epoch": 0.07772,
      "grad_norm": 0.727035568405193,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 7772
    },
    {
      "epoch": 0.07773,
      "grad_norm": 0.7440750703020748,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 7773
    },
    {
      "epoch": 0.07774,
      "grad_norm": 0.6795627443083461,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 7774
    },
    {
      "epoch": 0.07775,
      "grad_norm": 0.6784508298159274,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 7775
    },
    {
      "epoch": 0.07776,
      "grad_norm": 0.6385790918136247,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 7776
    },
    {
      "epoch": 0.07777,
      "grad_norm": 0.6872009152821507,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 7777
    },
    {
      "epoch": 0.07778,
      "grad_norm": 0.6675405617900448,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 7778
    },
    {
      "epoch": 0.07779,
      "grad_norm": 0.7357317176753875,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 7779
    },
    {
      "epoch": 0.0778,
      "grad_norm": 0.8022677370632643,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 7780
    },
    {
      "epoch": 0.07781,
      "grad_norm": 0.9434523581478621,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 7781
    },
    {
      "epoch": 0.07782,
      "grad_norm": 1.5623982009752648,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 7782
    },
    {
      "epoch": 0.07783,
      "grad_norm": 0.7044738718569362,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 7783
    },
    {
      "epoch": 0.07784,
      "grad_norm": 0.7245143997153488,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 7784
    },
    {
      "epoch": 0.07785,
      "grad_norm": 0.716698916381954,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 7785
    },
    {
      "epoch": 0.07786,
      "grad_norm": 0.6921187171985951,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 7786
    },
    {
      "epoch": 0.07787,
      "grad_norm": 0.6045099384961314,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 7787
    },
    {
      "epoch": 0.07788,
      "grad_norm": 0.6083002796628949,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 7788
    },
    {
      "epoch": 0.07789,
      "grad_norm": 0.6450261280850079,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 7789
    },
    {
      "epoch": 0.0779,
      "grad_norm": 0.7723456314060048,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 7790
    },
    {
      "epoch": 0.07791,
      "grad_norm": 0.9626934092811175,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 7791
    },
    {
      "epoch": 0.07792,
      "grad_norm": 1.0213202281359437,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 7792
    },
    {
      "epoch": 0.07793,
      "grad_norm": 0.9823991856862937,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 7793
    },
    {
      "epoch": 0.07794,
      "grad_norm": 1.0324749912970457,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 7794
    },
    {
      "epoch": 0.07795,
      "grad_norm": 1.2976167109237968,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7795
    },
    {
      "epoch": 0.07796,
      "grad_norm": 0.8902788450024939,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 7796
    },
    {
      "epoch": 0.07797,
      "grad_norm": 0.9291964989058917,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 7797
    },
    {
      "epoch": 0.07798,
      "grad_norm": 0.9627682686730562,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 7798
    },
    {
      "epoch": 0.07799,
      "grad_norm": 1.1456962518858615,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 7799
    },
    {
      "epoch": 0.078,
      "grad_norm": 1.0251765436245048,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 7800
    },
    {
      "epoch": 0.07801,
      "grad_norm": 0.9973687128151619,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7801
    },
    {
      "epoch": 0.07802,
      "grad_norm": 0.948086135583704,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 7802
    },
    {
      "epoch": 0.07803,
      "grad_norm": 1.1277990578479882,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 7803
    },
    {
      "epoch": 0.07804,
      "grad_norm": 0.9774623897022733,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 7804
    },
    {
      "epoch": 0.07805,
      "grad_norm": 0.8320985331874762,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 7805
    },
    {
      "epoch": 0.07806,
      "grad_norm": 0.8657196694590319,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 7806
    },
    {
      "epoch": 0.07807,
      "grad_norm": 0.9492463229383535,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 7807
    },
    {
      "epoch": 0.07808,
      "grad_norm": 1.0394516274093006,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 7808
    },
    {
      "epoch": 0.07809,
      "grad_norm": 0.9335393900367771,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 7809
    },
    {
      "epoch": 0.0781,
      "grad_norm": 0.9443482076739536,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 7810
    },
    {
      "epoch": 0.07811,
      "grad_norm": 0.9123874992413225,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 7811
    },
    {
      "epoch": 0.07812,
      "grad_norm": 1.1264587870913032,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 7812
    },
    {
      "epoch": 0.07813,
      "grad_norm": 1.1305952516582176,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 7813
    },
    {
      "epoch": 0.07814,
      "grad_norm": 0.8414805101214488,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 7814
    },
    {
      "epoch": 0.07815,
      "grad_norm": 0.9502914015578153,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 7815
    },
    {
      "epoch": 0.07816,
      "grad_norm": 1.1760246778581924,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 7816
    },
    {
      "epoch": 0.07817,
      "grad_norm": 0.9408621866890481,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 7817
    },
    {
      "epoch": 0.07818,
      "grad_norm": 0.8296428274925994,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 7818
    },
    {
      "epoch": 0.07819,
      "grad_norm": 0.7920801052079187,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 7819
    },
    {
      "epoch": 0.0782,
      "grad_norm": 0.8823722442311102,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 7820
    },
    {
      "epoch": 0.07821,
      "grad_norm": 0.9516339518218203,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 7821
    },
    {
      "epoch": 0.07822,
      "grad_norm": 1.0547110747707245,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 7822
    },
    {
      "epoch": 0.07823,
      "grad_norm": 0.989883022467342,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 7823
    },
    {
      "epoch": 0.07824,
      "grad_norm": 0.8914027725354553,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 7824
    },
    {
      "epoch": 0.07825,
      "grad_norm": 0.8076840807799915,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 7825
    },
    {
      "epoch": 0.07826,
      "grad_norm": 0.7415708695700829,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 7826
    },
    {
      "epoch": 0.07827,
      "grad_norm": 0.7214257388849613,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 7827
    },
    {
      "epoch": 0.07828,
      "grad_norm": 0.8136232450263843,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 7828
    },
    {
      "epoch": 0.07829,
      "grad_norm": 0.9199584413946614,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 7829
    },
    {
      "epoch": 0.0783,
      "grad_norm": 0.8115857466773752,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 7830
    },
    {
      "epoch": 0.07831,
      "grad_norm": 0.7591405213641907,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 7831
    },
    {
      "epoch": 0.07832,
      "grad_norm": 0.6490760526643984,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 7832
    },
    {
      "epoch": 0.07833,
      "grad_norm": 0.703850585915276,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 7833
    },
    {
      "epoch": 0.07834,
      "grad_norm": 0.8284668702922542,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 7834
    },
    {
      "epoch": 0.07835,
      "grad_norm": 0.8475382307293997,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 7835
    },
    {
      "epoch": 0.07836,
      "grad_norm": 0.800674520894362,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 7836
    },
    {
      "epoch": 0.07837,
      "grad_norm": 0.7804816932384945,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 7837
    },
    {
      "epoch": 0.07838,
      "grad_norm": 0.7719832162406756,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 7838
    },
    {
      "epoch": 0.07839,
      "grad_norm": 0.8103102203464976,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 7839
    },
    {
      "epoch": 0.0784,
      "grad_norm": 0.847847320739481,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 7840
    },
    {
      "epoch": 0.07841,
      "grad_norm": 0.9217204569862923,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 7841
    },
    {
      "epoch": 0.07842,
      "grad_norm": 1.0755759756691639,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 7842
    },
    {
      "epoch": 0.07843,
      "grad_norm": 1.0372058631482444,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 7843
    },
    {
      "epoch": 0.07844,
      "grad_norm": 0.8852971276699896,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 7844
    },
    {
      "epoch": 0.07845,
      "grad_norm": 0.7984782327828842,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 7845
    },
    {
      "epoch": 0.07846,
      "grad_norm": 0.8806483710377667,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 7846
    },
    {
      "epoch": 0.07847,
      "grad_norm": 1.0378492921152178,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 7847
    },
    {
      "epoch": 0.07848,
      "grad_norm": 0.895360780588699,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 7848
    },
    {
      "epoch": 0.07849,
      "grad_norm": 0.8289639586531552,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 7849
    },
    {
      "epoch": 0.0785,
      "grad_norm": 0.8305063229582818,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 7850
    },
    {
      "epoch": 0.07851,
      "grad_norm": 0.7058129266076502,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 7851
    },
    {
      "epoch": 0.07852,
      "grad_norm": 0.6418192413394789,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 7852
    },
    {
      "epoch": 0.07853,
      "grad_norm": 0.7531346386515196,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 7853
    },
    {
      "epoch": 0.07854,
      "grad_norm": 0.8398583249558569,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 7854
    },
    {
      "epoch": 0.07855,
      "grad_norm": 0.7774066949599959,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 7855
    },
    {
      "epoch": 0.07856,
      "grad_norm": 1.0200945417324516,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 7856
    },
    {
      "epoch": 0.07857,
      "grad_norm": 1.075009695811694,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 7857
    },
    {
      "epoch": 0.07858,
      "grad_norm": 0.7910404278344594,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 7858
    },
    {
      "epoch": 0.07859,
      "grad_norm": 0.7300829664682413,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 7859
    },
    {
      "epoch": 0.0786,
      "grad_norm": 0.7197539823867891,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 7860
    },
    {
      "epoch": 0.07861,
      "grad_norm": 0.9620546769547311,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 7861
    },
    {
      "epoch": 0.07862,
      "grad_norm": 1.490251553426132,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 7862
    },
    {
      "epoch": 0.07863,
      "grad_norm": 0.7857679852194924,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 7863
    },
    {
      "epoch": 0.07864,
      "grad_norm": 0.8634479975354046,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 7864
    },
    {
      "epoch": 0.07865,
      "grad_norm": 0.7964685970693094,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 7865
    },
    {
      "epoch": 0.07866,
      "grad_norm": 0.6828020542596382,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 7866
    },
    {
      "epoch": 0.07867,
      "grad_norm": 0.6092921432164138,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 7867
    },
    {
      "epoch": 0.07868,
      "grad_norm": 0.5633101550595739,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 7868
    },
    {
      "epoch": 0.07869,
      "grad_norm": 0.6431111329973238,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 7869
    },
    {
      "epoch": 0.0787,
      "grad_norm": 0.901303356812522,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 7870
    },
    {
      "epoch": 0.07871,
      "grad_norm": 1.2191547805657925,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 7871
    },
    {
      "epoch": 0.07872,
      "grad_norm": 0.8262977027782229,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 7872
    },
    {
      "epoch": 0.07873,
      "grad_norm": 0.6724946366349702,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 7873
    },
    {
      "epoch": 0.07874,
      "grad_norm": 0.7244287405483971,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 7874
    },
    {
      "epoch": 0.07875,
      "grad_norm": 0.8227798028339546,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 7875
    },
    {
      "epoch": 0.07876,
      "grad_norm": 0.9349875722236806,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 7876
    },
    {
      "epoch": 0.07877,
      "grad_norm": 1.0794801674884056,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 7877
    },
    {
      "epoch": 0.07878,
      "grad_norm": 0.9276580973578815,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 7878
    },
    {
      "epoch": 0.07879,
      "grad_norm": 0.8251333456721903,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 7879
    },
    {
      "epoch": 0.0788,
      "grad_norm": 0.7060382543658212,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 7880
    },
    {
      "epoch": 0.07881,
      "grad_norm": 0.67303988544837,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 7881
    },
    {
      "epoch": 0.07882,
      "grad_norm": 0.6233971310328372,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 7882
    },
    {
      "epoch": 0.07883,
      "grad_norm": 0.6298409951352602,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 7883
    },
    {
      "epoch": 0.07884,
      "grad_norm": 0.7947655306250354,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 7884
    },
    {
      "epoch": 0.07885,
      "grad_norm": 0.8497398170009447,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 7885
    },
    {
      "epoch": 0.07886,
      "grad_norm": 0.8619830961416387,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 7886
    },
    {
      "epoch": 0.07887,
      "grad_norm": 1.0484039664176932,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 7887
    },
    {
      "epoch": 0.07888,
      "grad_norm": 1.1847395590743321,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 7888
    },
    {
      "epoch": 0.07889,
      "grad_norm": 0.8493272011871066,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 7889
    },
    {
      "epoch": 0.0789,
      "grad_norm": 0.8008301667726648,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 7890
    },
    {
      "epoch": 0.07891,
      "grad_norm": 0.930077598807836,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 7891
    },
    {
      "epoch": 0.07892,
      "grad_norm": 1.0989482964593247,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 7892
    },
    {
      "epoch": 0.07893,
      "grad_norm": 1.1363252018029342,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 7893
    },
    {
      "epoch": 0.07894,
      "grad_norm": 1.0228512594068824,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 7894
    },
    {
      "epoch": 0.07895,
      "grad_norm": 0.9218498361406187,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 7895
    },
    {
      "epoch": 0.07896,
      "grad_norm": 0.9251940784692912,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 7896
    },
    {
      "epoch": 0.07897,
      "grad_norm": 0.931960102854229,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7897
    },
    {
      "epoch": 0.07898,
      "grad_norm": 0.8464573363229492,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 7898
    },
    {
      "epoch": 0.07899,
      "grad_norm": 0.7028868962428811,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 7899
    },
    {
      "epoch": 0.079,
      "grad_norm": 0.7441885103587236,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 7900
    },
    {
      "epoch": 0.07901,
      "grad_norm": 0.8664872116804634,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 7901
    },
    {
      "epoch": 0.07902,
      "grad_norm": 0.9184607477010853,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 7902
    },
    {
      "epoch": 0.07903,
      "grad_norm": 0.7590321607729549,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 7903
    },
    {
      "epoch": 0.07904,
      "grad_norm": 0.7531494902293769,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 7904
    },
    {
      "epoch": 0.07905,
      "grad_norm": 0.70438167321181,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 7905
    },
    {
      "epoch": 0.07906,
      "grad_norm": 0.6511721703071448,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 7906
    },
    {
      "epoch": 0.07907,
      "grad_norm": 0.716900467492343,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 7907
    },
    {
      "epoch": 0.07908,
      "grad_norm": 0.7085281032769419,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 7908
    },
    {
      "epoch": 0.07909,
      "grad_norm": 0.8416217578074326,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7909
    },
    {
      "epoch": 0.0791,
      "grad_norm": 1.1433040371010987,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 7910
    },
    {
      "epoch": 0.07911,
      "grad_norm": 1.1835699988724921,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 7911
    },
    {
      "epoch": 0.07912,
      "grad_norm": 0.9223643758933799,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 7912
    },
    {
      "epoch": 0.07913,
      "grad_norm": 1.089125989023986,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 7913
    },
    {
      "epoch": 0.07914,
      "grad_norm": 0.8390123983066416,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 7914
    },
    {
      "epoch": 0.07915,
      "grad_norm": 0.672914923150026,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 7915
    },
    {
      "epoch": 0.07916,
      "grad_norm": 0.7017447792175915,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 7916
    },
    {
      "epoch": 0.07917,
      "grad_norm": 0.7051019103527106,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 7917
    },
    {
      "epoch": 0.07918,
      "grad_norm": 0.9855647114410215,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 7918
    },
    {
      "epoch": 0.07919,
      "grad_norm": 1.3116973662720455,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 7919
    },
    {
      "epoch": 0.0792,
      "grad_norm": 0.7402493790216023,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 7920
    },
    {
      "epoch": 0.07921,
      "grad_norm": 0.6988160486580384,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 7921
    },
    {
      "epoch": 0.07922,
      "grad_norm": 0.7456591280144469,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 7922
    },
    {
      "epoch": 0.07923,
      "grad_norm": 0.8033173795113849,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 7923
    },
    {
      "epoch": 0.07924,
      "grad_norm": 0.7216426094382513,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 7924
    },
    {
      "epoch": 0.07925,
      "grad_norm": 0.7373410578068488,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7925
    },
    {
      "epoch": 0.07926,
      "grad_norm": 0.9580792267793108,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 7926
    },
    {
      "epoch": 0.07927,
      "grad_norm": 1.1508484267353751,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 7927
    },
    {
      "epoch": 0.07928,
      "grad_norm": 0.8413926446028052,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 7928
    },
    {
      "epoch": 0.07929,
      "grad_norm": 0.8045322762280693,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 7929
    },
    {
      "epoch": 0.0793,
      "grad_norm": 0.7334642856257453,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 7930
    },
    {
      "epoch": 0.07931,
      "grad_norm": 0.7666083653707652,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 7931
    },
    {
      "epoch": 0.07932,
      "grad_norm": 0.845631543538524,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 7932
    },
    {
      "epoch": 0.07933,
      "grad_norm": 1.1053771346780619,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 7933
    },
    {
      "epoch": 0.07934,
      "grad_norm": 1.0697166718848607,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 7934
    },
    {
      "epoch": 0.07935,
      "grad_norm": 0.780788335506962,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 7935
    },
    {
      "epoch": 0.07936,
      "grad_norm": 0.6719369309095249,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 7936
    },
    {
      "epoch": 0.07937,
      "grad_norm": 0.7129234476370259,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 7937
    },
    {
      "epoch": 0.07938,
      "grad_norm": 0.7579710871562575,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 7938
    },
    {
      "epoch": 0.07939,
      "grad_norm": 0.8414949927890002,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 7939
    },
    {
      "epoch": 0.0794,
      "grad_norm": 0.873475867512058,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 7940
    },
    {
      "epoch": 0.07941,
      "grad_norm": 0.7753297409475207,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 7941
    },
    {
      "epoch": 0.07942,
      "grad_norm": 0.7946685821619549,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 7942
    },
    {
      "epoch": 0.07943,
      "grad_norm": 0.9007048298260407,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 7943
    },
    {
      "epoch": 0.07944,
      "grad_norm": 1.071812362091642,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 7944
    },
    {
      "epoch": 0.07945,
      "grad_norm": 1.0693712200612246,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 7945
    },
    {
      "epoch": 0.07946,
      "grad_norm": 0.8795848429684096,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 7946
    },
    {
      "epoch": 0.07947,
      "grad_norm": 0.800378114738204,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 7947
    },
    {
      "epoch": 0.07948,
      "grad_norm": 0.9321588061699683,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 7948
    },
    {
      "epoch": 0.07949,
      "grad_norm": 0.991434583531255,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7949
    },
    {
      "epoch": 0.0795,
      "grad_norm": 0.9816512046959835,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 7950
    },
    {
      "epoch": 0.07951,
      "grad_norm": 0.7677971687728063,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 7951
    },
    {
      "epoch": 0.07952,
      "grad_norm": 0.8051868602715272,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 7952
    },
    {
      "epoch": 0.07953,
      "grad_norm": 0.9001609812725201,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 7953
    },
    {
      "epoch": 0.07954,
      "grad_norm": 1.1106187952067725,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 7954
    },
    {
      "epoch": 0.07955,
      "grad_norm": 0.9649928236706383,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 7955
    },
    {
      "epoch": 0.07956,
      "grad_norm": 1.1226837346899992,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 7956
    },
    {
      "epoch": 0.07957,
      "grad_norm": 1.1705169222423675,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 7957
    },
    {
      "epoch": 0.07958,
      "grad_norm": 0.8899304842940169,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 7958
    },
    {
      "epoch": 0.07959,
      "grad_norm": 0.7210543206553331,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 7959
    },
    {
      "epoch": 0.0796,
      "grad_norm": 0.8474860169251741,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 7960
    },
    {
      "epoch": 0.07961,
      "grad_norm": 0.8510145192053272,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 7961
    },
    {
      "epoch": 0.07962,
      "grad_norm": 0.9477244770512336,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 7962
    },
    {
      "epoch": 0.07963,
      "grad_norm": 0.9694960562391948,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 7963
    },
    {
      "epoch": 0.07964,
      "grad_norm": 1.0096222672390198,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 7964
    },
    {
      "epoch": 0.07965,
      "grad_norm": 0.9843752313745823,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 7965
    },
    {
      "epoch": 0.07966,
      "grad_norm": 0.9595021912517606,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 7966
    },
    {
      "epoch": 0.07967,
      "grad_norm": 1.0016620302362607,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 7967
    },
    {
      "epoch": 0.07968,
      "grad_norm": 0.9784507337152613,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 7968
    },
    {
      "epoch": 0.07969,
      "grad_norm": 0.9283133075078502,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 7969
    },
    {
      "epoch": 0.0797,
      "grad_norm": 0.9854134391658027,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 7970
    },
    {
      "epoch": 0.07971,
      "grad_norm": 1.0491313465828465,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 7971
    },
    {
      "epoch": 0.07972,
      "grad_norm": 1.0232565756589487,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 7972
    },
    {
      "epoch": 0.07973,
      "grad_norm": 0.9389427379261678,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 7973
    },
    {
      "epoch": 0.07974,
      "grad_norm": 0.9196943064318163,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 7974
    },
    {
      "epoch": 0.07975,
      "grad_norm": 0.8757962873318504,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 7975
    },
    {
      "epoch": 0.07976,
      "grad_norm": 0.8223659364709709,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 7976
    },
    {
      "epoch": 0.07977,
      "grad_norm": 0.8719111818094638,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 7977
    },
    {
      "epoch": 0.07978,
      "grad_norm": 0.8824234915401362,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 7978
    },
    {
      "epoch": 0.07979,
      "grad_norm": 0.9247964543100683,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 7979
    },
    {
      "epoch": 0.0798,
      "grad_norm": 0.9397634918432941,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 7980
    },
    {
      "epoch": 0.07981,
      "grad_norm": 1.0543674923134152,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 7981
    },
    {
      "epoch": 0.07982,
      "grad_norm": 0.9626911947278476,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 7982
    },
    {
      "epoch": 0.07983,
      "grad_norm": 1.0080240905133067,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 7983
    },
    {
      "epoch": 0.07984,
      "grad_norm": 1.2190721528512718,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 7984
    },
    {
      "epoch": 0.07985,
      "grad_norm": 0.8286713865087831,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 7985
    },
    {
      "epoch": 0.07986,
      "grad_norm": 0.8677596520468425,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 7986
    },
    {
      "epoch": 0.07987,
      "grad_norm": 0.839953559356892,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 7987
    },
    {
      "epoch": 0.07988,
      "grad_norm": 0.8616747266554645,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 7988
    },
    {
      "epoch": 0.07989,
      "grad_norm": 0.8768670980849567,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 7989
    },
    {
      "epoch": 0.0799,
      "grad_norm": 1.060320785667829,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 7990
    },
    {
      "epoch": 0.07991,
      "grad_norm": 1.2397246116873737,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 7991
    },
    {
      "epoch": 0.07992,
      "grad_norm": 0.8564943829168591,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 7992
    },
    {
      "epoch": 0.07993,
      "grad_norm": 0.6329653057644131,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 7993
    },
    {
      "epoch": 0.07994,
      "grad_norm": 0.5495879883814838,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 7994
    },
    {
      "epoch": 0.07995,
      "grad_norm": 0.6418923191491803,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 7995
    },
    {
      "epoch": 0.07996,
      "grad_norm": 0.6360153656004878,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 7996
    },
    {
      "epoch": 0.07997,
      "grad_norm": 0.6967070433022678,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 7997
    },
    {
      "epoch": 0.07998,
      "grad_norm": 0.8401425534763325,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 7998
    },
    {
      "epoch": 0.07999,
      "grad_norm": 0.9140910976472046,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 7999
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8341075710754534,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 8000
    },
    {
      "epoch": 0.08001,
      "grad_norm": 0.6762636347053551,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 8001
    },
    {
      "epoch": 0.08002,
      "grad_norm": 0.7219182957365917,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 8002
    },
    {
      "epoch": 0.08003,
      "grad_norm": 0.7307059911401975,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 8003
    },
    {
      "epoch": 0.08004,
      "grad_norm": 0.7690134766684371,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8004
    },
    {
      "epoch": 0.08005,
      "grad_norm": 0.7032812824393433,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 8005
    },
    {
      "epoch": 0.08006,
      "grad_norm": 0.8083021905718699,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 8006
    },
    {
      "epoch": 0.08007,
      "grad_norm": 1.0155775443805966,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 8007
    },
    {
      "epoch": 0.08008,
      "grad_norm": 1.2028469175430232,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 8008
    },
    {
      "epoch": 0.08009,
      "grad_norm": 0.7132237916366507,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 8009
    },
    {
      "epoch": 0.0801,
      "grad_norm": 0.5404697854674432,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 8010
    },
    {
      "epoch": 0.08011,
      "grad_norm": 0.5947072885434764,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 8011
    },
    {
      "epoch": 0.08012,
      "grad_norm": 0.7363521666233862,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 8012
    },
    {
      "epoch": 0.08013,
      "grad_norm": 1.0063376322547504,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 8013
    },
    {
      "epoch": 0.08014,
      "grad_norm": 1.1160436669300626,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 8014
    },
    {
      "epoch": 0.08015,
      "grad_norm": 0.7989320079754313,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 8015
    },
    {
      "epoch": 0.08016,
      "grad_norm": 0.7682760666118215,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 8016
    },
    {
      "epoch": 0.08017,
      "grad_norm": 0.7989671209318877,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 8017
    },
    {
      "epoch": 0.08018,
      "grad_norm": 0.8900732101990633,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8018
    },
    {
      "epoch": 0.08019,
      "grad_norm": 0.7809140627865704,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 8019
    },
    {
      "epoch": 0.0802,
      "grad_norm": 0.8370791311704688,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 8020
    },
    {
      "epoch": 0.08021,
      "grad_norm": 0.7244688057921231,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 8021
    },
    {
      "epoch": 0.08022,
      "grad_norm": 0.7032261987394863,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 8022
    },
    {
      "epoch": 0.08023,
      "grad_norm": 0.7746372857782606,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 8023
    },
    {
      "epoch": 0.08024,
      "grad_norm": 0.980871731492932,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 8024
    },
    {
      "epoch": 0.08025,
      "grad_norm": 1.1967229295248476,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 8025
    },
    {
      "epoch": 0.08026,
      "grad_norm": 0.8874537036909782,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 8026
    },
    {
      "epoch": 0.08027,
      "grad_norm": 0.7886044707135144,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 8027
    },
    {
      "epoch": 0.08028,
      "grad_norm": 0.7195647660407796,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 8028
    },
    {
      "epoch": 0.08029,
      "grad_norm": 0.8677533765236772,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 8029
    },
    {
      "epoch": 0.0803,
      "grad_norm": 1.006614001349766,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 8030
    },
    {
      "epoch": 0.08031,
      "grad_norm": 1.1001052865444716,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 8031
    },
    {
      "epoch": 0.08032,
      "grad_norm": 0.9468984767405869,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 8032
    },
    {
      "epoch": 0.08033,
      "grad_norm": 0.7392931914642568,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 8033
    },
    {
      "epoch": 0.08034,
      "grad_norm": 0.8147657234401885,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 8034
    },
    {
      "epoch": 0.08035,
      "grad_norm": 0.9473798784474897,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 8035
    },
    {
      "epoch": 0.08036,
      "grad_norm": 0.9663624382781947,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 8036
    },
    {
      "epoch": 0.08037,
      "grad_norm": 0.9851396854632779,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 8037
    },
    {
      "epoch": 0.08038,
      "grad_norm": 0.9226417242283115,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8038
    },
    {
      "epoch": 0.08039,
      "grad_norm": 0.9417189662091079,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 8039
    },
    {
      "epoch": 0.0804,
      "grad_norm": 1.1112977827509558,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 8040
    },
    {
      "epoch": 0.08041,
      "grad_norm": 0.7739549217316514,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 8041
    },
    {
      "epoch": 0.08042,
      "grad_norm": 0.6673108385892449,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 8042
    },
    {
      "epoch": 0.08043,
      "grad_norm": 0.7412544947569171,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 8043
    },
    {
      "epoch": 0.08044,
      "grad_norm": 0.7934062901474336,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 8044
    },
    {
      "epoch": 0.08045,
      "grad_norm": 0.939659696654444,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 8045
    },
    {
      "epoch": 0.08046,
      "grad_norm": 1.0779216512774807,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 8046
    },
    {
      "epoch": 0.08047,
      "grad_norm": 1.0583855488249307,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 8047
    },
    {
      "epoch": 0.08048,
      "grad_norm": 1.010247891168986,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 8048
    },
    {
      "epoch": 0.08049,
      "grad_norm": 0.8559934548803988,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 8049
    },
    {
      "epoch": 0.0805,
      "grad_norm": 0.785962682018867,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 8050
    },
    {
      "epoch": 0.08051,
      "grad_norm": 0.8243401452192997,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 8051
    },
    {
      "epoch": 0.08052,
      "grad_norm": 0.8118236172127069,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 8052
    },
    {
      "epoch": 0.08053,
      "grad_norm": 0.7919496704600435,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 8053
    },
    {
      "epoch": 0.08054,
      "grad_norm": 0.7931762362063768,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 8054
    },
    {
      "epoch": 0.08055,
      "grad_norm": 0.8667155845252095,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 8055
    },
    {
      "epoch": 0.08056,
      "grad_norm": 1.0124039809076317,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 8056
    },
    {
      "epoch": 0.08057,
      "grad_norm": 1.1336788875699602,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 8057
    },
    {
      "epoch": 0.08058,
      "grad_norm": 0.8138250876786765,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 8058
    },
    {
      "epoch": 0.08059,
      "grad_norm": 0.7527191665858058,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 8059
    },
    {
      "epoch": 0.0806,
      "grad_norm": 0.8398808621055186,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 8060
    },
    {
      "epoch": 0.08061,
      "grad_norm": 0.825612767978505,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 8061
    },
    {
      "epoch": 0.08062,
      "grad_norm": 0.9747793071523012,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 8062
    },
    {
      "epoch": 0.08063,
      "grad_norm": 1.1323711262853373,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 8063
    },
    {
      "epoch": 0.08064,
      "grad_norm": 0.9414405685374689,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8064
    },
    {
      "epoch": 0.08065,
      "grad_norm": 0.877182385950829,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 8065
    },
    {
      "epoch": 0.08066,
      "grad_norm": 0.9823013557480436,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 8066
    },
    {
      "epoch": 0.08067,
      "grad_norm": 1.0076994124987724,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 8067
    },
    {
      "epoch": 0.08068,
      "grad_norm": 0.9543931097825955,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 8068
    },
    {
      "epoch": 0.08069,
      "grad_norm": 0.860639718383474,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 8069
    },
    {
      "epoch": 0.0807,
      "grad_norm": 0.8988959057627193,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8070
    },
    {
      "epoch": 0.08071,
      "grad_norm": 0.9534116382487039,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 8071
    },
    {
      "epoch": 0.08072,
      "grad_norm": 0.8660070014644025,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 8072
    },
    {
      "epoch": 0.08073,
      "grad_norm": 0.8206611525482264,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 8073
    },
    {
      "epoch": 0.08074,
      "grad_norm": 0.965813130304323,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 8074
    },
    {
      "epoch": 0.08075,
      "grad_norm": 1.1442526973314016,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 8075
    },
    {
      "epoch": 0.08076,
      "grad_norm": 0.7971529185891382,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 8076
    },
    {
      "epoch": 0.08077,
      "grad_norm": 0.6441126050923031,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 8077
    },
    {
      "epoch": 0.08078,
      "grad_norm": 0.711624279954515,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 8078
    },
    {
      "epoch": 0.08079,
      "grad_norm": 0.8468946223392866,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 8079
    },
    {
      "epoch": 0.0808,
      "grad_norm": 0.955200727057764,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 8080
    },
    {
      "epoch": 0.08081,
      "grad_norm": 1.153055759299516,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 8081
    },
    {
      "epoch": 0.08082,
      "grad_norm": 0.7784581084262547,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 8082
    },
    {
      "epoch": 0.08083,
      "grad_norm": 0.7994938331488759,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8083
    },
    {
      "epoch": 0.08084,
      "grad_norm": 0.7985218153568882,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 8084
    },
    {
      "epoch": 0.08085,
      "grad_norm": 0.7650441422106858,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 8085
    },
    {
      "epoch": 0.08086,
      "grad_norm": 0.9072526574743619,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 8086
    },
    {
      "epoch": 0.08087,
      "grad_norm": 1.0019552972566608,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 8087
    },
    {
      "epoch": 0.08088,
      "grad_norm": 1.0698186825083669,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 8088
    },
    {
      "epoch": 0.08089,
      "grad_norm": 0.8487971751746695,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 8089
    },
    {
      "epoch": 0.0809,
      "grad_norm": 0.7375163925623175,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 8090
    },
    {
      "epoch": 0.08091,
      "grad_norm": 0.8077965976484879,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 8091
    },
    {
      "epoch": 0.08092,
      "grad_norm": 0.8806655777433894,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 8092
    },
    {
      "epoch": 0.08093,
      "grad_norm": 0.8846095215059361,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 8093
    },
    {
      "epoch": 0.08094,
      "grad_norm": 1.0263143758049644,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 8094
    },
    {
      "epoch": 0.08095,
      "grad_norm": 0.9975182671338138,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 8095
    },
    {
      "epoch": 0.08096,
      "grad_norm": 0.9243866628279602,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 8096
    },
    {
      "epoch": 0.08097,
      "grad_norm": 0.764690705974969,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 8097
    },
    {
      "epoch": 0.08098,
      "grad_norm": 0.8123826720903898,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 8098
    },
    {
      "epoch": 0.08099,
      "grad_norm": 0.8307382789621239,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 8099
    },
    {
      "epoch": 0.081,
      "grad_norm": 0.683295591217391,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 8100
    },
    {
      "epoch": 0.08101,
      "grad_norm": 0.762072793349242,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 8101
    },
    {
      "epoch": 0.08102,
      "grad_norm": 0.8417363332108121,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 8102
    },
    {
      "epoch": 0.08103,
      "grad_norm": 1.0026749696403685,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 8103
    },
    {
      "epoch": 0.08104,
      "grad_norm": 1.3055013665970854,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 8104
    },
    {
      "epoch": 0.08105,
      "grad_norm": 0.7284350470055041,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 8105
    },
    {
      "epoch": 0.08106,
      "grad_norm": 0.7042488841506047,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 8106
    },
    {
      "epoch": 0.08107,
      "grad_norm": 0.6662191545969933,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 8107
    },
    {
      "epoch": 0.08108,
      "grad_norm": 0.7428305715535388,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 8108
    },
    {
      "epoch": 0.08109,
      "grad_norm": 0.7934277909736256,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8109
    },
    {
      "epoch": 0.0811,
      "grad_norm": 0.8818863948526912,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 8110
    },
    {
      "epoch": 0.08111,
      "grad_norm": 1.0271099724103216,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 8111
    },
    {
      "epoch": 0.08112,
      "grad_norm": 1.2432013981414785,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 8112
    },
    {
      "epoch": 0.08113,
      "grad_norm": 0.8995583170658074,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 8113
    },
    {
      "epoch": 0.08114,
      "grad_norm": 1.0954559019247962,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 8114
    },
    {
      "epoch": 0.08115,
      "grad_norm": 1.261265478089719,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 8115
    },
    {
      "epoch": 0.08116,
      "grad_norm": 0.8815832900165901,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 8116
    },
    {
      "epoch": 0.08117,
      "grad_norm": 0.9087111097091439,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 8117
    },
    {
      "epoch": 0.08118,
      "grad_norm": 1.010760403401548,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 8118
    },
    {
      "epoch": 0.08119,
      "grad_norm": 1.056578294217066,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 8119
    },
    {
      "epoch": 0.0812,
      "grad_norm": 0.9182887201534551,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 8120
    },
    {
      "epoch": 0.08121,
      "grad_norm": 0.9041701432881559,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 8121
    },
    {
      "epoch": 0.08122,
      "grad_norm": 0.8561224947474484,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 8122
    },
    {
      "epoch": 0.08123,
      "grad_norm": 0.7614632039050264,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 8123
    },
    {
      "epoch": 0.08124,
      "grad_norm": 0.7907531416628623,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 8124
    },
    {
      "epoch": 0.08125,
      "grad_norm": 0.7300964775523906,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 8125
    },
    {
      "epoch": 0.08126,
      "grad_norm": 0.827569261113533,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 8126
    },
    {
      "epoch": 0.08127,
      "grad_norm": 1.0339006016723764,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 8127
    },
    {
      "epoch": 0.08128,
      "grad_norm": 1.2837248118220423,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 8128
    },
    {
      "epoch": 0.08129,
      "grad_norm": 0.831463644855417,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 8129
    },
    {
      "epoch": 0.0813,
      "grad_norm": 0.7313691024451621,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 8130
    },
    {
      "epoch": 0.08131,
      "grad_norm": 0.7449363031040399,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 8131
    },
    {
      "epoch": 0.08132,
      "grad_norm": 0.8923893884805761,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 8132
    },
    {
      "epoch": 0.08133,
      "grad_norm": 1.00858302579857,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 8133
    },
    {
      "epoch": 0.08134,
      "grad_norm": 1.0839591261263513,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8134
    },
    {
      "epoch": 0.08135,
      "grad_norm": 0.9551176992279036,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 8135
    },
    {
      "epoch": 0.08136,
      "grad_norm": 1.1427717477515151,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 8136
    },
    {
      "epoch": 0.08137,
      "grad_norm": 1.1608247537070675,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 8137
    },
    {
      "epoch": 0.08138,
      "grad_norm": 0.8036743748727992,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8138
    },
    {
      "epoch": 0.08139,
      "grad_norm": 0.7123613175658509,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 8139
    },
    {
      "epoch": 0.0814,
      "grad_norm": 0.7693281819705117,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 8140
    },
    {
      "epoch": 0.08141,
      "grad_norm": 0.6727495096734566,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 8141
    },
    {
      "epoch": 0.08142,
      "grad_norm": 0.6845081706658788,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 8142
    },
    {
      "epoch": 0.08143,
      "grad_norm": 0.7460442553490216,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 8143
    },
    {
      "epoch": 0.08144,
      "grad_norm": 0.8245361382413854,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 8144
    },
    {
      "epoch": 0.08145,
      "grad_norm": 0.8230934857490788,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 8145
    },
    {
      "epoch": 0.08146,
      "grad_norm": 0.7712994219712946,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 8146
    },
    {
      "epoch": 0.08147,
      "grad_norm": 0.7460865071832761,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 8147
    },
    {
      "epoch": 0.08148,
      "grad_norm": 0.7735141713283167,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 8148
    },
    {
      "epoch": 0.08149,
      "grad_norm": 0.7896814609713348,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 8149
    },
    {
      "epoch": 0.0815,
      "grad_norm": 0.7747886996009599,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 8150
    },
    {
      "epoch": 0.08151,
      "grad_norm": 0.8866325001517937,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 8151
    },
    {
      "epoch": 0.08152,
      "grad_norm": 0.9586498492742512,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 8152
    },
    {
      "epoch": 0.08153,
      "grad_norm": 0.8759445141936913,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 8153
    },
    {
      "epoch": 0.08154,
      "grad_norm": 0.8213012411475468,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 8154
    },
    {
      "epoch": 0.08155,
      "grad_norm": 0.7343561260089444,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 8155
    },
    {
      "epoch": 0.08156,
      "grad_norm": 0.7055042559161329,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 8156
    },
    {
      "epoch": 0.08157,
      "grad_norm": 0.6906999012739224,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 8157
    },
    {
      "epoch": 0.08158,
      "grad_norm": 0.6797037737186805,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 8158
    },
    {
      "epoch": 0.08159,
      "grad_norm": 0.6440815279956414,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 8159
    },
    {
      "epoch": 0.0816,
      "grad_norm": 0.6557581564589431,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 8160
    },
    {
      "epoch": 0.08161,
      "grad_norm": 0.6961649850170736,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 8161
    },
    {
      "epoch": 0.08162,
      "grad_norm": 0.7152753880484393,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 8162
    },
    {
      "epoch": 0.08163,
      "grad_norm": 0.7444333284957504,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 8163
    },
    {
      "epoch": 0.08164,
      "grad_norm": 0.8297049679085736,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 8164
    },
    {
      "epoch": 0.08165,
      "grad_norm": 0.9254543973261257,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 8165
    },
    {
      "epoch": 0.08166,
      "grad_norm": 1.135495751545436,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 8166
    },
    {
      "epoch": 0.08167,
      "grad_norm": 1.003264644227776,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 8167
    },
    {
      "epoch": 0.08168,
      "grad_norm": 1.3151569568484487,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 8168
    },
    {
      "epoch": 0.08169,
      "grad_norm": 0.7417217462858907,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 8169
    },
    {
      "epoch": 0.0817,
      "grad_norm": 0.6967246388224767,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 8170
    },
    {
      "epoch": 0.08171,
      "grad_norm": 0.8159574604205326,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 8171
    },
    {
      "epoch": 0.08172,
      "grad_norm": 0.8338929792539445,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 8172
    },
    {
      "epoch": 0.08173,
      "grad_norm": 0.98094890794689,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 8173
    },
    {
      "epoch": 0.08174,
      "grad_norm": 1.0004268057358918,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 8174
    },
    {
      "epoch": 0.08175,
      "grad_norm": 1.1607043587315327,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 8175
    },
    {
      "epoch": 0.08176,
      "grad_norm": 0.9605737836969909,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 8176
    },
    {
      "epoch": 0.08177,
      "grad_norm": 0.8834558793913779,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 8177
    },
    {
      "epoch": 0.08178,
      "grad_norm": 0.9463401693916579,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 8178
    },
    {
      "epoch": 0.08179,
      "grad_norm": 0.9414403348160015,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 8179
    },
    {
      "epoch": 0.0818,
      "grad_norm": 0.9984246787560999,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 8180
    },
    {
      "epoch": 0.08181,
      "grad_norm": 1.1775258869822263,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 8181
    },
    {
      "epoch": 0.08182,
      "grad_norm": 0.8435168559607914,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 8182
    },
    {
      "epoch": 0.08183,
      "grad_norm": 0.8045360298546768,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 8183
    },
    {
      "epoch": 0.08184,
      "grad_norm": 0.7940871155944753,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 8184
    },
    {
      "epoch": 0.08185,
      "grad_norm": 0.7578411281597589,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 8185
    },
    {
      "epoch": 0.08186,
      "grad_norm": 0.7267884989595176,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 8186
    },
    {
      "epoch": 0.08187,
      "grad_norm": 0.9028879506424374,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 8187
    },
    {
      "epoch": 0.08188,
      "grad_norm": 1.0836114502579879,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 8188
    },
    {
      "epoch": 0.08189,
      "grad_norm": 1.2100347103121765,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 8189
    },
    {
      "epoch": 0.0819,
      "grad_norm": 0.7309510071375395,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 8190
    },
    {
      "epoch": 0.08191,
      "grad_norm": 0.6311549475723786,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 8191
    },
    {
      "epoch": 0.08192,
      "grad_norm": 0.7435668383560015,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 8192
    },
    {
      "epoch": 0.08193,
      "grad_norm": 0.8580955152495726,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8193
    },
    {
      "epoch": 0.08194,
      "grad_norm": 0.8737720108493068,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 8194
    },
    {
      "epoch": 0.08195,
      "grad_norm": 0.7368558728662361,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 8195
    },
    {
      "epoch": 0.08196,
      "grad_norm": 0.648364175407553,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 8196
    },
    {
      "epoch": 0.08197,
      "grad_norm": 0.7511314273266356,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8197
    },
    {
      "epoch": 0.08198,
      "grad_norm": 0.9036375613694632,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 8198
    },
    {
      "epoch": 0.08199,
      "grad_norm": 1.0500737128901654,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 8199
    },
    {
      "epoch": 0.082,
      "grad_norm": 1.1226888117313103,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 8200
    },
    {
      "epoch": 0.08201,
      "grad_norm": 0.9559312403994046,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 8201
    },
    {
      "epoch": 0.08202,
      "grad_norm": 0.9864224976849971,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 8202
    },
    {
      "epoch": 0.08203,
      "grad_norm": 1.0665254798248887,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 8203
    },
    {
      "epoch": 0.08204,
      "grad_norm": 0.9332708500174354,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 8204
    },
    {
      "epoch": 0.08205,
      "grad_norm": 0.9114402954224472,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 8205
    },
    {
      "epoch": 0.08206,
      "grad_norm": 0.9578612535805147,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 8206
    },
    {
      "epoch": 0.08207,
      "grad_norm": 0.8491904917863199,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 8207
    },
    {
      "epoch": 0.08208,
      "grad_norm": 0.6965883444740882,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 8208
    },
    {
      "epoch": 0.08209,
      "grad_norm": 0.7953053984407373,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 8209
    },
    {
      "epoch": 0.0821,
      "grad_norm": 0.7485874491455442,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 8210
    },
    {
      "epoch": 0.08211,
      "grad_norm": 0.6959203077939378,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 8211
    },
    {
      "epoch": 0.08212,
      "grad_norm": 0.8449995767744634,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 8212
    },
    {
      "epoch": 0.08213,
      "grad_norm": 1.1015900099708935,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 8213
    },
    {
      "epoch": 0.08214,
      "grad_norm": 1.0315791177365499,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 8214
    },
    {
      "epoch": 0.08215,
      "grad_norm": 0.9380557040904997,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 8215
    },
    {
      "epoch": 0.08216,
      "grad_norm": 0.8923877134844612,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 8216
    },
    {
      "epoch": 0.08217,
      "grad_norm": 0.8575588533424483,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 8217
    },
    {
      "epoch": 0.08218,
      "grad_norm": 0.8232596706369313,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 8218
    },
    {
      "epoch": 0.08219,
      "grad_norm": 0.9081979548545349,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 8219
    },
    {
      "epoch": 0.0822,
      "grad_norm": 0.8155798907038242,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8220
    },
    {
      "epoch": 0.08221,
      "grad_norm": 0.8942358339851308,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 8221
    },
    {
      "epoch": 0.08222,
      "grad_norm": 1.1787055400411643,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 8222
    },
    {
      "epoch": 0.08223,
      "grad_norm": 0.9509702723098798,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 8223
    },
    {
      "epoch": 0.08224,
      "grad_norm": 0.888798577292018,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 8224
    },
    {
      "epoch": 0.08225,
      "grad_norm": 0.8859713724373705,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 8225
    },
    {
      "epoch": 0.08226,
      "grad_norm": 0.9453256114459335,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 8226
    },
    {
      "epoch": 0.08227,
      "grad_norm": 1.0285023285363262,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 8227
    },
    {
      "epoch": 0.08228,
      "grad_norm": 1.0441982441165125,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 8228
    },
    {
      "epoch": 0.08229,
      "grad_norm": 0.9285745096914537,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 8229
    },
    {
      "epoch": 0.0823,
      "grad_norm": 0.8612854507864894,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 8230
    },
    {
      "epoch": 0.08231,
      "grad_norm": 0.9539626109153166,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 8231
    },
    {
      "epoch": 0.08232,
      "grad_norm": 1.1127495391486235,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 8232
    },
    {
      "epoch": 0.08233,
      "grad_norm": 0.9309241874844985,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 8233
    },
    {
      "epoch": 0.08234,
      "grad_norm": 0.8586480291175957,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 8234
    },
    {
      "epoch": 0.08235,
      "grad_norm": 0.8943116440399151,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 8235
    },
    {
      "epoch": 0.08236,
      "grad_norm": 0.8765191739384368,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 8236
    },
    {
      "epoch": 0.08237,
      "grad_norm": 0.8771079726635557,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 8237
    },
    {
      "epoch": 0.08238,
      "grad_norm": 0.983672922309625,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 8238
    },
    {
      "epoch": 0.08239,
      "grad_norm": 1.0592369601456073,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 8239
    },
    {
      "epoch": 0.0824,
      "grad_norm": 1.0147386504894589,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 8240
    },
    {
      "epoch": 0.08241,
      "grad_norm": 0.8937176926338467,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 8241
    },
    {
      "epoch": 0.08242,
      "grad_norm": 0.8211844129412678,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 8242
    },
    {
      "epoch": 0.08243,
      "grad_norm": 0.941400186703893,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 8243
    },
    {
      "epoch": 0.08244,
      "grad_norm": 1.147358753730803,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8244
    },
    {
      "epoch": 0.08245,
      "grad_norm": 1.0164893306751752,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 8245
    },
    {
      "epoch": 0.08246,
      "grad_norm": 0.8548549249810995,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 8246
    },
    {
      "epoch": 0.08247,
      "grad_norm": 0.7706214458931752,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 8247
    },
    {
      "epoch": 0.08248,
      "grad_norm": 0.8475424364775245,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 8248
    },
    {
      "epoch": 0.08249,
      "grad_norm": 0.8469414618345413,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 8249
    },
    {
      "epoch": 0.0825,
      "grad_norm": 0.8229434139477997,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 8250
    },
    {
      "epoch": 0.08251,
      "grad_norm": 0.8011704320030776,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 8251
    },
    {
      "epoch": 0.08252,
      "grad_norm": 0.6866093619498361,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 8252
    },
    {
      "epoch": 0.08253,
      "grad_norm": 0.7062240053088432,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 8253
    },
    {
      "epoch": 0.08254,
      "grad_norm": 0.6263882996334876,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 8254
    },
    {
      "epoch": 0.08255,
      "grad_norm": 0.6406212013874797,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 8255
    },
    {
      "epoch": 0.08256,
      "grad_norm": 0.6332253598165484,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 8256
    },
    {
      "epoch": 0.08257,
      "grad_norm": 0.7298881894270829,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 8257
    },
    {
      "epoch": 0.08258,
      "grad_norm": 0.8061858783513308,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 8258
    },
    {
      "epoch": 0.08259,
      "grad_norm": 0.9631605130652663,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 8259
    },
    {
      "epoch": 0.0826,
      "grad_norm": 1.312679616746302,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 8260
    },
    {
      "epoch": 0.08261,
      "grad_norm": 0.7878732930444392,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 8261
    },
    {
      "epoch": 0.08262,
      "grad_norm": 0.7074636050563899,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 8262
    },
    {
      "epoch": 0.08263,
      "grad_norm": 0.6997390004949057,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 8263
    },
    {
      "epoch": 0.08264,
      "grad_norm": 0.772169379776631,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 8264
    },
    {
      "epoch": 0.08265,
      "grad_norm": 0.7993877169518088,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 8265
    },
    {
      "epoch": 0.08266,
      "grad_norm": 1.0950952503503433,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 8266
    },
    {
      "epoch": 0.08267,
      "grad_norm": 1.1929897335482988,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 8267
    },
    {
      "epoch": 0.08268,
      "grad_norm": 0.8778501205471867,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 8268
    },
    {
      "epoch": 0.08269,
      "grad_norm": 0.892710429777809,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 8269
    },
    {
      "epoch": 0.0827,
      "grad_norm": 1.0149545588000088,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 8270
    },
    {
      "epoch": 0.08271,
      "grad_norm": 1.124408719720435,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 8271
    },
    {
      "epoch": 0.08272,
      "grad_norm": 0.8266826354793325,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 8272
    },
    {
      "epoch": 0.08273,
      "grad_norm": 1.0123352772430743,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 8273
    },
    {
      "epoch": 0.08274,
      "grad_norm": 1.0192036235033302,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 8274
    },
    {
      "epoch": 0.08275,
      "grad_norm": 0.9102355182979404,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8275
    },
    {
      "epoch": 0.08276,
      "grad_norm": 0.7090653355389134,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 8276
    },
    {
      "epoch": 0.08277,
      "grad_norm": 0.6881786586872917,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 8277
    },
    {
      "epoch": 0.08278,
      "grad_norm": 0.7699897377296708,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 8278
    },
    {
      "epoch": 0.08279,
      "grad_norm": 0.8358151974396222,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 8279
    },
    {
      "epoch": 0.0828,
      "grad_norm": 0.9018628671728572,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 8280
    },
    {
      "epoch": 0.08281,
      "grad_norm": 1.1343593814872228,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 8281
    },
    {
      "epoch": 0.08282,
      "grad_norm": 0.9781923188910686,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 8282
    },
    {
      "epoch": 0.08283,
      "grad_norm": 0.9927329287124252,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 8283
    },
    {
      "epoch": 0.08284,
      "grad_norm": 0.98118308584172,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 8284
    },
    {
      "epoch": 0.08285,
      "grad_norm": 0.9263465527068231,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 8285
    },
    {
      "epoch": 0.08286,
      "grad_norm": 0.8673110587039227,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 8286
    },
    {
      "epoch": 0.08287,
      "grad_norm": 0.8419508037976696,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 8287
    },
    {
      "epoch": 0.08288,
      "grad_norm": 0.8220681636415317,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 8288
    },
    {
      "epoch": 0.08289,
      "grad_norm": 0.7505932152286665,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 8289
    },
    {
      "epoch": 0.0829,
      "grad_norm": 0.8010371678478732,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 8290
    },
    {
      "epoch": 0.08291,
      "grad_norm": 0.7956640617851198,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 8291
    },
    {
      "epoch": 0.08292,
      "grad_norm": 0.9655748890689609,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 8292
    },
    {
      "epoch": 0.08293,
      "grad_norm": 1.2599943864429868,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 8293
    },
    {
      "epoch": 0.08294,
      "grad_norm": 0.9706360443217623,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 8294
    },
    {
      "epoch": 0.08295,
      "grad_norm": 1.048320155699331,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8295
    },
    {
      "epoch": 0.08296,
      "grad_norm": 0.9595325746578597,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 8296
    },
    {
      "epoch": 0.08297,
      "grad_norm": 0.9381047024976223,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 8297
    },
    {
      "epoch": 0.08298,
      "grad_norm": 0.863050580789861,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 8298
    },
    {
      "epoch": 0.08299,
      "grad_norm": 0.7240303327199505,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 8299
    },
    {
      "epoch": 0.083,
      "grad_norm": 0.7823461180480966,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 8300
    },
    {
      "epoch": 0.08301,
      "grad_norm": 0.8990430145286981,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 8301
    },
    {
      "epoch": 0.08302,
      "grad_norm": 0.938888405100819,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 8302
    },
    {
      "epoch": 0.08303,
      "grad_norm": 0.8734564423642468,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8303
    },
    {
      "epoch": 0.08304,
      "grad_norm": 0.8158703165239903,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 8304
    },
    {
      "epoch": 0.08305,
      "grad_norm": 0.75620484767062,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 8305
    },
    {
      "epoch": 0.08306,
      "grad_norm": 0.6540375187690641,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 8306
    },
    {
      "epoch": 0.08307,
      "grad_norm": 0.7220457038710596,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 8307
    },
    {
      "epoch": 0.08308,
      "grad_norm": 0.7585842959907338,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 8308
    },
    {
      "epoch": 0.08309,
      "grad_norm": 0.7562956967912909,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 8309
    },
    {
      "epoch": 0.0831,
      "grad_norm": 0.8620656108784794,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 8310
    },
    {
      "epoch": 0.08311,
      "grad_norm": 1.0392515900505281,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 8311
    },
    {
      "epoch": 0.08312,
      "grad_norm": 1.1358410812232829,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 8312
    },
    {
      "epoch": 0.08313,
      "grad_norm": 0.8856911622270947,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 8313
    },
    {
      "epoch": 0.08314,
      "grad_norm": 0.8971402484163674,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 8314
    },
    {
      "epoch": 0.08315,
      "grad_norm": 0.9089314400577564,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 8315
    },
    {
      "epoch": 0.08316,
      "grad_norm": 0.9935520424178989,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 8316
    },
    {
      "epoch": 0.08317,
      "grad_norm": 0.9643025611664917,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 8317
    },
    {
      "epoch": 0.08318,
      "grad_norm": 0.9787470048990922,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 8318
    },
    {
      "epoch": 0.08319,
      "grad_norm": 1.196041471896753,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 8319
    },
    {
      "epoch": 0.0832,
      "grad_norm": 0.7528163771935631,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 8320
    },
    {
      "epoch": 0.08321,
      "grad_norm": 0.6113115162451314,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 8321
    },
    {
      "epoch": 0.08322,
      "grad_norm": 0.6302210737347822,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 8322
    },
    {
      "epoch": 0.08323,
      "grad_norm": 0.6846448753101897,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 8323
    },
    {
      "epoch": 0.08324,
      "grad_norm": 0.81282422542553,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 8324
    },
    {
      "epoch": 0.08325,
      "grad_norm": 0.9233311436381315,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 8325
    },
    {
      "epoch": 0.08326,
      "grad_norm": 0.954642418929674,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 8326
    },
    {
      "epoch": 0.08327,
      "grad_norm": 0.8277691448246721,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 8327
    },
    {
      "epoch": 0.08328,
      "grad_norm": 0.7651088353198385,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 8328
    },
    {
      "epoch": 0.08329,
      "grad_norm": 0.8129905787749772,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 8329
    },
    {
      "epoch": 0.0833,
      "grad_norm": 0.7705105903193307,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 8330
    },
    {
      "epoch": 0.08331,
      "grad_norm": 0.8783516141923371,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 8331
    },
    {
      "epoch": 0.08332,
      "grad_norm": 0.9500635724943229,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 8332
    },
    {
      "epoch": 0.08333,
      "grad_norm": 1.0022282494956223,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 8333
    },
    {
      "epoch": 0.08334,
      "grad_norm": 0.9868081102540992,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 8334
    },
    {
      "epoch": 0.08335,
      "grad_norm": 0.9702541505884114,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 8335
    },
    {
      "epoch": 0.08336,
      "grad_norm": 0.9616697241003944,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 8336
    },
    {
      "epoch": 0.08337,
      "grad_norm": 0.9197392338204791,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 8337
    },
    {
      "epoch": 0.08338,
      "grad_norm": 0.9953727673941519,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 8338
    },
    {
      "epoch": 0.08339,
      "grad_norm": 1.0143098515036326,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 8339
    },
    {
      "epoch": 0.0834,
      "grad_norm": 0.8294567785378771,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 8340
    },
    {
      "epoch": 0.08341,
      "grad_norm": 0.6099348529213143,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 8341
    },
    {
      "epoch": 0.08342,
      "grad_norm": 0.7224131427612864,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 8342
    },
    {
      "epoch": 0.08343,
      "grad_norm": 0.8439433910256732,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 8343
    },
    {
      "epoch": 0.08344,
      "grad_norm": 0.9868340222493676,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 8344
    },
    {
      "epoch": 0.08345,
      "grad_norm": 0.9401831390198352,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 8345
    },
    {
      "epoch": 0.08346,
      "grad_norm": 1.0302321487078914,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 8346
    },
    {
      "epoch": 0.08347,
      "grad_norm": 0.8650226664698788,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 8347
    },
    {
      "epoch": 0.08348,
      "grad_norm": 0.7651090535509265,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 8348
    },
    {
      "epoch": 0.08349,
      "grad_norm": 0.804167869373704,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 8349
    },
    {
      "epoch": 0.0835,
      "grad_norm": 0.907923736734775,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 8350
    },
    {
      "epoch": 0.08351,
      "grad_norm": 0.9166006948131097,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 8351
    },
    {
      "epoch": 0.08352,
      "grad_norm": 0.9742286942493177,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 8352
    },
    {
      "epoch": 0.08353,
      "grad_norm": 1.0443086898250653,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 8353
    },
    {
      "epoch": 0.08354,
      "grad_norm": 1.0482003041090164,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 8354
    },
    {
      "epoch": 0.08355,
      "grad_norm": 0.9948528176450955,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 8355
    },
    {
      "epoch": 0.08356,
      "grad_norm": 0.9930403012130316,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 8356
    },
    {
      "epoch": 0.08357,
      "grad_norm": 1.058290503425188,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 8357
    },
    {
      "epoch": 0.08358,
      "grad_norm": 0.7888495179404684,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 8358
    },
    {
      "epoch": 0.08359,
      "grad_norm": 0.7764006655298166,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 8359
    },
    {
      "epoch": 0.0836,
      "grad_norm": 0.9269785078441903,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 8360
    },
    {
      "epoch": 0.08361,
      "grad_norm": 1.0448000920569085,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 8361
    },
    {
      "epoch": 0.08362,
      "grad_norm": 1.0677595443672623,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 8362
    },
    {
      "epoch": 0.08363,
      "grad_norm": 0.9762108759601814,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 8363
    },
    {
      "epoch": 0.08364,
      "grad_norm": 0.9250190436823729,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 8364
    },
    {
      "epoch": 0.08365,
      "grad_norm": 0.6667015272636464,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 8365
    },
    {
      "epoch": 0.08366,
      "grad_norm": 0.6845211638503894,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 8366
    },
    {
      "epoch": 0.08367,
      "grad_norm": 0.5939518428147165,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 8367
    },
    {
      "epoch": 0.08368,
      "grad_norm": 0.5996368389783523,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 8368
    },
    {
      "epoch": 0.08369,
      "grad_norm": 0.6523282835388826,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 8369
    },
    {
      "epoch": 0.0837,
      "grad_norm": 0.8159208873071091,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8370
    },
    {
      "epoch": 0.08371,
      "grad_norm": 0.9860834398447139,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 8371
    },
    {
      "epoch": 0.08372,
      "grad_norm": 1.0940596921770058,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 8372
    },
    {
      "epoch": 0.08373,
      "grad_norm": 0.9442033751036637,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 8373
    },
    {
      "epoch": 0.08374,
      "grad_norm": 0.8663695859388787,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 8374
    },
    {
      "epoch": 0.08375,
      "grad_norm": 0.7415517737245889,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 8375
    },
    {
      "epoch": 0.08376,
      "grad_norm": 0.8577754980749478,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 8376
    },
    {
      "epoch": 0.08377,
      "grad_norm": 1.0199862392503956,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 8377
    },
    {
      "epoch": 0.08378,
      "grad_norm": 1.2449008964290007,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 8378
    },
    {
      "epoch": 0.08379,
      "grad_norm": 0.9419961904078284,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 8379
    },
    {
      "epoch": 0.0838,
      "grad_norm": 0.9566083866982931,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 8380
    },
    {
      "epoch": 0.08381,
      "grad_norm": 0.921682176354393,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 8381
    },
    {
      "epoch": 0.08382,
      "grad_norm": 0.8446653628794919,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 8382
    },
    {
      "epoch": 0.08383,
      "grad_norm": 0.817980279883183,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 8383
    },
    {
      "epoch": 0.08384,
      "grad_norm": 0.7858769081287543,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 8384
    },
    {
      "epoch": 0.08385,
      "grad_norm": 0.7428564824532166,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 8385
    },
    {
      "epoch": 0.08386,
      "grad_norm": 0.7507352447976503,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 8386
    },
    {
      "epoch": 0.08387,
      "grad_norm": 0.8467430532110711,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 8387
    },
    {
      "epoch": 0.08388,
      "grad_norm": 0.9680594439287274,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 8388
    },
    {
      "epoch": 0.08389,
      "grad_norm": 1.2252522777923969,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 8389
    },
    {
      "epoch": 0.0839,
      "grad_norm": 0.7533692382713882,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 8390
    },
    {
      "epoch": 0.08391,
      "grad_norm": 0.578854889997608,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 8391
    },
    {
      "epoch": 0.08392,
      "grad_norm": 0.7310406618044913,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 8392
    },
    {
      "epoch": 0.08393,
      "grad_norm": 0.8080163613542798,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 8393
    },
    {
      "epoch": 0.08394,
      "grad_norm": 0.9313902145452904,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 8394
    },
    {
      "epoch": 0.08395,
      "grad_norm": 0.8774904667389832,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 8395
    },
    {
      "epoch": 0.08396,
      "grad_norm": 0.7336808643576519,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 8396
    },
    {
      "epoch": 0.08397,
      "grad_norm": 0.6526993931050207,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 8397
    },
    {
      "epoch": 0.08398,
      "grad_norm": 0.6704748477002462,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 8398
    },
    {
      "epoch": 0.08399,
      "grad_norm": 0.7009862189779927,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 8399
    },
    {
      "epoch": 0.084,
      "grad_norm": 0.7959197464027524,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 8400
    },
    {
      "epoch": 0.08401,
      "grad_norm": 0.7942806861956023,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 8401
    },
    {
      "epoch": 0.08402,
      "grad_norm": 0.7901087516905125,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 8402
    },
    {
      "epoch": 0.08403,
      "grad_norm": 0.7878285742447195,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 8403
    },
    {
      "epoch": 0.08404,
      "grad_norm": 0.760912884794821,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 8404
    },
    {
      "epoch": 0.08405,
      "grad_norm": 0.9006130071518619,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 8405
    },
    {
      "epoch": 0.08406,
      "grad_norm": 1.0016144008572414,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 8406
    },
    {
      "epoch": 0.08407,
      "grad_norm": 1.3443814703857433,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 8407
    },
    {
      "epoch": 0.08408,
      "grad_norm": 0.9056353298369414,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 8408
    },
    {
      "epoch": 0.08409,
      "grad_norm": 0.8063158095511546,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 8409
    },
    {
      "epoch": 0.0841,
      "grad_norm": 0.9219305008449818,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 8410
    },
    {
      "epoch": 0.08411,
      "grad_norm": 1.1026094058352998,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 8411
    },
    {
      "epoch": 0.08412,
      "grad_norm": 0.9534553488792078,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 8412
    },
    {
      "epoch": 0.08413,
      "grad_norm": 0.942913373390524,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 8413
    },
    {
      "epoch": 0.08414,
      "grad_norm": 1.0244828961575672,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 8414
    },
    {
      "epoch": 0.08415,
      "grad_norm": 0.9400339571658545,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 8415
    },
    {
      "epoch": 0.08416,
      "grad_norm": 0.8710315727738005,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 8416
    },
    {
      "epoch": 0.08417,
      "grad_norm": 0.7950437016706896,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 8417
    },
    {
      "epoch": 0.08418,
      "grad_norm": 0.8098581252447725,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 8418
    },
    {
      "epoch": 0.08419,
      "grad_norm": 0.9057985377788469,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 8419
    },
    {
      "epoch": 0.0842,
      "grad_norm": 0.9778312033623336,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 8420
    },
    {
      "epoch": 0.08421,
      "grad_norm": 0.8982644728463484,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 8421
    },
    {
      "epoch": 0.08422,
      "grad_norm": 0.9376171311714971,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 8422
    },
    {
      "epoch": 0.08423,
      "grad_norm": 0.9399936667089988,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 8423
    },
    {
      "epoch": 0.08424,
      "grad_norm": 0.8765907774353715,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 8424
    },
    {
      "epoch": 0.08425,
      "grad_norm": 0.8093336331696083,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 8425
    },
    {
      "epoch": 0.08426,
      "grad_norm": 0.7177444716043226,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 8426
    },
    {
      "epoch": 0.08427,
      "grad_norm": 0.8420947027796308,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 8427
    },
    {
      "epoch": 0.08428,
      "grad_norm": 1.0570206327016962,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 8428
    },
    {
      "epoch": 0.08429,
      "grad_norm": 1.0693132771916776,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 8429
    },
    {
      "epoch": 0.0843,
      "grad_norm": 1.037364188618762,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 8430
    },
    {
      "epoch": 0.08431,
      "grad_norm": 0.9849908084466207,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 8431
    },
    {
      "epoch": 0.08432,
      "grad_norm": 0.9291586130000788,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 8432
    },
    {
      "epoch": 0.08433,
      "grad_norm": 0.865132194731464,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 8433
    },
    {
      "epoch": 0.08434,
      "grad_norm": 0.8122049399169294,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 8434
    },
    {
      "epoch": 0.08435,
      "grad_norm": 0.747566312505787,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 8435
    },
    {
      "epoch": 0.08436,
      "grad_norm": 0.7617380609411367,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 8436
    },
    {
      "epoch": 0.08437,
      "grad_norm": 0.7640674198661149,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 8437
    },
    {
      "epoch": 0.08438,
      "grad_norm": 0.7897561671508848,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 8438
    },
    {
      "epoch": 0.08439,
      "grad_norm": 0.9805961893177073,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 8439
    },
    {
      "epoch": 0.0844,
      "grad_norm": 1.2397471986974102,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 8440
    },
    {
      "epoch": 0.08441,
      "grad_norm": 0.7657531909009802,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 8441
    },
    {
      "epoch": 0.08442,
      "grad_norm": 0.6761237170315371,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 8442
    },
    {
      "epoch": 0.08443,
      "grad_norm": 0.6698413991893777,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 8443
    },
    {
      "epoch": 0.08444,
      "grad_norm": 0.6538638151947389,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 8444
    },
    {
      "epoch": 0.08445,
      "grad_norm": 0.7819289887319408,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 8445
    },
    {
      "epoch": 0.08446,
      "grad_norm": 0.7360808672824343,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 8446
    },
    {
      "epoch": 0.08447,
      "grad_norm": 0.8192930714654896,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 8447
    },
    {
      "epoch": 0.08448,
      "grad_norm": 0.9656464757675268,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 8448
    },
    {
      "epoch": 0.08449,
      "grad_norm": 1.1936151486401498,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 8449
    },
    {
      "epoch": 0.0845,
      "grad_norm": 0.685549098243575,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 8450
    },
    {
      "epoch": 0.08451,
      "grad_norm": 0.9139246196778028,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 8451
    },
    {
      "epoch": 0.08452,
      "grad_norm": 1.2202645871735718,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 8452
    },
    {
      "epoch": 0.08453,
      "grad_norm": 1.011747247457234,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 8453
    },
    {
      "epoch": 0.08454,
      "grad_norm": 1.0262226366034506,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 8454
    },
    {
      "epoch": 0.08455,
      "grad_norm": 1.0099082871033074,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 8455
    },
    {
      "epoch": 0.08456,
      "grad_norm": 0.90849087232309,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 8456
    },
    {
      "epoch": 0.08457,
      "grad_norm": 0.8383502937124733,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 8457
    },
    {
      "epoch": 0.08458,
      "grad_norm": 0.8440455792853685,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 8458
    },
    {
      "epoch": 0.08459,
      "grad_norm": 0.7112059922093875,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 8459
    },
    {
      "epoch": 0.0846,
      "grad_norm": 0.649515091750979,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 8460
    },
    {
      "epoch": 0.08461,
      "grad_norm": 0.6333241585768099,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 8461
    },
    {
      "epoch": 0.08462,
      "grad_norm": 0.6384025767344915,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 8462
    },
    {
      "epoch": 0.08463,
      "grad_norm": 0.6400437078939833,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 8463
    },
    {
      "epoch": 0.08464,
      "grad_norm": 0.6759399211580148,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 8464
    },
    {
      "epoch": 0.08465,
      "grad_norm": 0.7413271665798129,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8465
    },
    {
      "epoch": 0.08466,
      "grad_norm": 0.8771869007648114,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 8466
    },
    {
      "epoch": 0.08467,
      "grad_norm": 0.9227181671457924,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 8467
    },
    {
      "epoch": 0.08468,
      "grad_norm": 0.9695793195002803,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 8468
    },
    {
      "epoch": 0.08469,
      "grad_norm": 1.2889696763476837,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 8469
    },
    {
      "epoch": 0.0847,
      "grad_norm": 0.9448980455530938,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 8470
    },
    {
      "epoch": 0.08471,
      "grad_norm": 1.1057721964654461,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 8471
    },
    {
      "epoch": 0.08472,
      "grad_norm": 1.1484149282990468,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 8472
    },
    {
      "epoch": 0.08473,
      "grad_norm": 1.0503023171846668,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 8473
    },
    {
      "epoch": 0.08474,
      "grad_norm": 0.8628642552887439,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 8474
    },
    {
      "epoch": 0.08475,
      "grad_norm": 0.7509423112660792,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 8475
    },
    {
      "epoch": 0.08476,
      "grad_norm": 0.70961640106593,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 8476
    },
    {
      "epoch": 0.08477,
      "grad_norm": 0.7231357262197663,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 8477
    },
    {
      "epoch": 0.08478,
      "grad_norm": 0.804332121451016,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 8478
    },
    {
      "epoch": 0.08479,
      "grad_norm": 0.9887354629105892,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 8479
    },
    {
      "epoch": 0.0848,
      "grad_norm": 0.9881963393542088,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 8480
    },
    {
      "epoch": 0.08481,
      "grad_norm": 0.9379461518994493,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 8481
    },
    {
      "epoch": 0.08482,
      "grad_norm": 0.7805227298805099,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 8482
    },
    {
      "epoch": 0.08483,
      "grad_norm": 0.7856593793741489,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8483
    },
    {
      "epoch": 0.08484,
      "grad_norm": 0.8932525879434696,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 8484
    },
    {
      "epoch": 0.08485,
      "grad_norm": 0.8851276814593144,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 8485
    },
    {
      "epoch": 0.08486,
      "grad_norm": 0.8944513894921937,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 8486
    },
    {
      "epoch": 0.08487,
      "grad_norm": 1.0310888734066477,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 8487
    },
    {
      "epoch": 0.08488,
      "grad_norm": 1.0686836420532142,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 8488
    },
    {
      "epoch": 0.08489,
      "grad_norm": 1.0317207317630044,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 8489
    },
    {
      "epoch": 0.0849,
      "grad_norm": 1.1495485663194194,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 8490
    },
    {
      "epoch": 0.08491,
      "grad_norm": 1.0263416248655877,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 8491
    },
    {
      "epoch": 0.08492,
      "grad_norm": 0.9700428254885138,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 8492
    },
    {
      "epoch": 0.08493,
      "grad_norm": 0.9948556669000941,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 8493
    },
    {
      "epoch": 0.08494,
      "grad_norm": 0.910141922117321,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 8494
    },
    {
      "epoch": 0.08495,
      "grad_norm": 0.9597026790665053,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 8495
    },
    {
      "epoch": 0.08496,
      "grad_norm": 0.9275785612350264,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 8496
    },
    {
      "epoch": 0.08497,
      "grad_norm": 1.01867706095219,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 8497
    },
    {
      "epoch": 0.08498,
      "grad_norm": 1.1590197748031095,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 8498
    },
    {
      "epoch": 0.08499,
      "grad_norm": 0.9880096011902544,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 8499
    },
    {
      "epoch": 0.085,
      "grad_norm": 0.8266327675260843,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 8500
    },
    {
      "epoch": 0.08501,
      "grad_norm": 0.7411428596681484,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 8501
    },
    {
      "epoch": 0.08502,
      "grad_norm": 0.6949138984131208,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 8502
    },
    {
      "epoch": 0.08503,
      "grad_norm": 0.7536288112312667,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 8503
    },
    {
      "epoch": 0.08504,
      "grad_norm": 0.7223239288762903,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 8504
    },
    {
      "epoch": 0.08505,
      "grad_norm": 0.675765276936943,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 8505
    },
    {
      "epoch": 0.08506,
      "grad_norm": 0.7576978694249362,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 8506
    },
    {
      "epoch": 0.08507,
      "grad_norm": 0.8515649801971872,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 8507
    },
    {
      "epoch": 0.08508,
      "grad_norm": 1.057062705401568,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 8508
    },
    {
      "epoch": 0.08509,
      "grad_norm": 1.121630687562224,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 8509
    },
    {
      "epoch": 0.0851,
      "grad_norm": 0.9104321815273234,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 8510
    },
    {
      "epoch": 0.08511,
      "grad_norm": 0.8342488080751641,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 8511
    },
    {
      "epoch": 0.08512,
      "grad_norm": 0.7988070005118341,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 8512
    },
    {
      "epoch": 0.08513,
      "grad_norm": 0.799837965249621,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 8513
    },
    {
      "epoch": 0.08514,
      "grad_norm": 0.835432784668365,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 8514
    },
    {
      "epoch": 0.08515,
      "grad_norm": 0.8955465769825964,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8515
    },
    {
      "epoch": 0.08516,
      "grad_norm": 0.8928688903157641,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 8516
    },
    {
      "epoch": 0.08517,
      "grad_norm": 0.8226476531644744,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 8517
    },
    {
      "epoch": 0.08518,
      "grad_norm": 0.9561707034998627,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 8518
    },
    {
      "epoch": 0.08519,
      "grad_norm": 1.1902153338579717,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 8519
    },
    {
      "epoch": 0.0852,
      "grad_norm": 0.8018095905157278,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 8520
    },
    {
      "epoch": 0.08521,
      "grad_norm": 0.5533864795410459,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 8521
    },
    {
      "epoch": 0.08522,
      "grad_norm": 0.7791152111447774,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 8522
    },
    {
      "epoch": 0.08523,
      "grad_norm": 0.900500602043729,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 8523
    },
    {
      "epoch": 0.08524,
      "grad_norm": 0.9074705833253267,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 8524
    },
    {
      "epoch": 0.08525,
      "grad_norm": 0.9112667601876601,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 8525
    },
    {
      "epoch": 0.08526,
      "grad_norm": 1.0212625794257977,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 8526
    },
    {
      "epoch": 0.08527,
      "grad_norm": 1.0584452877128767,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8527
    },
    {
      "epoch": 0.08528,
      "grad_norm": 0.8561020391089984,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 8528
    },
    {
      "epoch": 0.08529,
      "grad_norm": 0.8298501875520379,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 8529
    },
    {
      "epoch": 0.0853,
      "grad_norm": 0.9089572166905378,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 8530
    },
    {
      "epoch": 0.08531,
      "grad_norm": 0.8653272986614832,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 8531
    },
    {
      "epoch": 0.08532,
      "grad_norm": 0.788926984023568,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 8532
    },
    {
      "epoch": 0.08533,
      "grad_norm": 0.8903035626600532,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 8533
    },
    {
      "epoch": 0.08534,
      "grad_norm": 0.963034125025281,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 8534
    },
    {
      "epoch": 0.08535,
      "grad_norm": 1.2879128764669079,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 8535
    },
    {
      "epoch": 0.08536,
      "grad_norm": 0.8902797022173426,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 8536
    },
    {
      "epoch": 0.08537,
      "grad_norm": 0.8657020447888398,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 8537
    },
    {
      "epoch": 0.08538,
      "grad_norm": 0.849426185382095,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 8538
    },
    {
      "epoch": 0.08539,
      "grad_norm": 0.8702155984928356,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 8539
    },
    {
      "epoch": 0.0854,
      "grad_norm": 0.7764427649934481,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 8540
    },
    {
      "epoch": 0.08541,
      "grad_norm": 0.7058501386903097,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 8541
    },
    {
      "epoch": 0.08542,
      "grad_norm": 0.8663633824046528,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 8542
    },
    {
      "epoch": 0.08543,
      "grad_norm": 0.906265069337281,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 8543
    },
    {
      "epoch": 0.08544,
      "grad_norm": 0.9275930195253327,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 8544
    },
    {
      "epoch": 0.08545,
      "grad_norm": 0.8078784033785711,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 8545
    },
    {
      "epoch": 0.08546,
      "grad_norm": 0.9066354249548785,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 8546
    },
    {
      "epoch": 0.08547,
      "grad_norm": 1.3364123020627683,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 8547
    },
    {
      "epoch": 0.08548,
      "grad_norm": 0.9435393460362433,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 8548
    },
    {
      "epoch": 0.08549,
      "grad_norm": 0.9332959081861278,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 8549
    },
    {
      "epoch": 0.0855,
      "grad_norm": 0.8374993962866415,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 8550
    },
    {
      "epoch": 0.08551,
      "grad_norm": 0.72433894811856,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 8551
    },
    {
      "epoch": 0.08552,
      "grad_norm": 0.7603328491561405,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 8552
    },
    {
      "epoch": 0.08553,
      "grad_norm": 0.8165853461739714,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 8553
    },
    {
      "epoch": 0.08554,
      "grad_norm": 0.8329371861835474,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 8554
    },
    {
      "epoch": 0.08555,
      "grad_norm": 0.8836247642143207,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 8555
    },
    {
      "epoch": 0.08556,
      "grad_norm": 1.0655224233717422,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 8556
    },
    {
      "epoch": 0.08557,
      "grad_norm": 1.1033400346275346,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 8557
    },
    {
      "epoch": 0.08558,
      "grad_norm": 0.9185812974874371,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 8558
    },
    {
      "epoch": 0.08559,
      "grad_norm": 0.8449525867089718,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 8559
    },
    {
      "epoch": 0.0856,
      "grad_norm": 0.8791279440183213,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8560
    },
    {
      "epoch": 0.08561,
      "grad_norm": 1.007009657834021,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 8561
    },
    {
      "epoch": 0.08562,
      "grad_norm": 1.2506411131458524,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 8562
    },
    {
      "epoch": 0.08563,
      "grad_norm": 0.8595663427695304,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 8563
    },
    {
      "epoch": 0.08564,
      "grad_norm": 0.8258432628148261,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 8564
    },
    {
      "epoch": 0.08565,
      "grad_norm": 0.8232028179529909,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 8565
    },
    {
      "epoch": 0.08566,
      "grad_norm": 0.7588530988097654,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 8566
    },
    {
      "epoch": 0.08567,
      "grad_norm": 0.6824147570028768,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 8567
    },
    {
      "epoch": 0.08568,
      "grad_norm": 0.5996215117346594,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 8568
    },
    {
      "epoch": 0.08569,
      "grad_norm": 0.5807584231599277,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 8569
    },
    {
      "epoch": 0.0857,
      "grad_norm": 0.5266785999312754,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 8570
    },
    {
      "epoch": 0.08571,
      "grad_norm": 0.4724117155463532,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8571
    },
    {
      "epoch": 0.08572,
      "grad_norm": 0.49806214531928805,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 8572
    },
    {
      "epoch": 0.08573,
      "grad_norm": 0.5210491883768703,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 8573
    },
    {
      "epoch": 0.08574,
      "grad_norm": 0.580899421582982,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 8574
    },
    {
      "epoch": 0.08575,
      "grad_norm": 0.759338722165245,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 8575
    },
    {
      "epoch": 0.08576,
      "grad_norm": 1.0882621390033222,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 8576
    },
    {
      "epoch": 0.08577,
      "grad_norm": 1.286177814911776,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 8577
    },
    {
      "epoch": 0.08578,
      "grad_norm": 0.649434769796102,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 8578
    },
    {
      "epoch": 0.08579,
      "grad_norm": 0.7955247063051825,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 8579
    },
    {
      "epoch": 0.0858,
      "grad_norm": 0.9332228897387346,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 8580
    },
    {
      "epoch": 0.08581,
      "grad_norm": 0.9286534200000857,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 8581
    },
    {
      "epoch": 0.08582,
      "grad_norm": 0.8515796422164925,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 8582
    },
    {
      "epoch": 0.08583,
      "grad_norm": 0.9496730587144366,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 8583
    },
    {
      "epoch": 0.08584,
      "grad_norm": 1.0376797351944236,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 8584
    },
    {
      "epoch": 0.08585,
      "grad_norm": 1.0733355797768753,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 8585
    },
    {
      "epoch": 0.08586,
      "grad_norm": 1.0898909465211588,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 8586
    },
    {
      "epoch": 0.08587,
      "grad_norm": 0.8541629848679168,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 8587
    },
    {
      "epoch": 0.08588,
      "grad_norm": 0.9259074296023726,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8588
    },
    {
      "epoch": 0.08589,
      "grad_norm": 1.0133166847780468,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 8589
    },
    {
      "epoch": 0.0859,
      "grad_norm": 0.9952675282959716,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 8590
    },
    {
      "epoch": 0.08591,
      "grad_norm": 0.8432804668392331,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 8591
    },
    {
      "epoch": 0.08592,
      "grad_norm": 0.7927775995190869,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 8592
    },
    {
      "epoch": 0.08593,
      "grad_norm": 0.9135931190636065,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 8593
    },
    {
      "epoch": 0.08594,
      "grad_norm": 0.89183466708387,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 8594
    },
    {
      "epoch": 0.08595,
      "grad_norm": 1.0748167455019701,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 8595
    },
    {
      "epoch": 0.08596,
      "grad_norm": 0.9757553423551145,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 8596
    },
    {
      "epoch": 0.08597,
      "grad_norm": 1.1029927439937899,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 8597
    },
    {
      "epoch": 0.08598,
      "grad_norm": 1.2202072285107675,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8598
    },
    {
      "epoch": 0.08599,
      "grad_norm": 1.0472074968226261,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 8599
    },
    {
      "epoch": 0.086,
      "grad_norm": 0.9436651521834845,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 8600
    },
    {
      "epoch": 0.08601,
      "grad_norm": 0.8978086885499934,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 8601
    },
    {
      "epoch": 0.08602,
      "grad_norm": 0.9046422898166396,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 8602
    },
    {
      "epoch": 0.08603,
      "grad_norm": 0.9940609256714368,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 8603
    },
    {
      "epoch": 0.08604,
      "grad_norm": 1.2914507269620887,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 8604
    },
    {
      "epoch": 0.08605,
      "grad_norm": 1.0010719188912025,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 8605
    },
    {
      "epoch": 0.08606,
      "grad_norm": 0.851064370373218,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 8606
    },
    {
      "epoch": 0.08607,
      "grad_norm": 0.8084048446777777,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 8607
    },
    {
      "epoch": 0.08608,
      "grad_norm": 0.8260223223516866,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 8608
    },
    {
      "epoch": 0.08609,
      "grad_norm": 0.800858297527019,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 8609
    },
    {
      "epoch": 0.0861,
      "grad_norm": 0.7751744199950412,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 8610
    },
    {
      "epoch": 0.08611,
      "grad_norm": 0.7850692774747576,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 8611
    },
    {
      "epoch": 0.08612,
      "grad_norm": 0.8034287087644159,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 8612
    },
    {
      "epoch": 0.08613,
      "grad_norm": 0.7907981566162866,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 8613
    },
    {
      "epoch": 0.08614,
      "grad_norm": 0.7648908788973424,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 8614
    },
    {
      "epoch": 0.08615,
      "grad_norm": 0.8058658371668642,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 8615
    },
    {
      "epoch": 0.08616,
      "grad_norm": 1.0179169747185717,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 8616
    },
    {
      "epoch": 0.08617,
      "grad_norm": 1.3401947590337657,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 8617
    },
    {
      "epoch": 0.08618,
      "grad_norm": 0.6850314006469229,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 8618
    },
    {
      "epoch": 0.08619,
      "grad_norm": 0.8070821056107416,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 8619
    },
    {
      "epoch": 0.0862,
      "grad_norm": 1.0123939132301667,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 8620
    },
    {
      "epoch": 0.08621,
      "grad_norm": 1.1347343173557283,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 8621
    },
    {
      "epoch": 0.08622,
      "grad_norm": 0.9266592246218266,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 8622
    },
    {
      "epoch": 0.08623,
      "grad_norm": 0.9044983849980115,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 8623
    },
    {
      "epoch": 0.08624,
      "grad_norm": 0.8761897564386483,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 8624
    },
    {
      "epoch": 0.08625,
      "grad_norm": 0.8620326850964458,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 8625
    },
    {
      "epoch": 0.08626,
      "grad_norm": 0.7720603843449916,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 8626
    },
    {
      "epoch": 0.08627,
      "grad_norm": 0.8313063670205044,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 8627
    },
    {
      "epoch": 0.08628,
      "grad_norm": 0.7788816276172811,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 8628
    },
    {
      "epoch": 0.08629,
      "grad_norm": 0.8225879339157376,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 8629
    },
    {
      "epoch": 0.0863,
      "grad_norm": 0.9503798474325023,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 8630
    },
    {
      "epoch": 0.08631,
      "grad_norm": 1.0311613535749917,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 8631
    },
    {
      "epoch": 0.08632,
      "grad_norm": 1.097072302177679,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 8632
    },
    {
      "epoch": 0.08633,
      "grad_norm": 0.905010614567883,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 8633
    },
    {
      "epoch": 0.08634,
      "grad_norm": 0.9280254071681021,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 8634
    },
    {
      "epoch": 0.08635,
      "grad_norm": 0.9486845314400906,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 8635
    },
    {
      "epoch": 0.08636,
      "grad_norm": 0.965675528971389,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 8636
    },
    {
      "epoch": 0.08637,
      "grad_norm": 0.9874360924938551,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 8637
    },
    {
      "epoch": 0.08638,
      "grad_norm": 0.9015114006705611,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 8638
    },
    {
      "epoch": 0.08639,
      "grad_norm": 0.8776646457013992,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 8639
    },
    {
      "epoch": 0.0864,
      "grad_norm": 1.1686490796312563,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 8640
    },
    {
      "epoch": 0.08641,
      "grad_norm": 1.1835266668134277,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 8641
    },
    {
      "epoch": 0.08642,
      "grad_norm": 0.7789856451482106,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 8642
    },
    {
      "epoch": 0.08643,
      "grad_norm": 0.7159387374682472,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 8643
    },
    {
      "epoch": 0.08644,
      "grad_norm": 0.7127204757723076,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 8644
    },
    {
      "epoch": 0.08645,
      "grad_norm": 0.8068921966237395,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 8645
    },
    {
      "epoch": 0.08646,
      "grad_norm": 0.9115412811447862,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 8646
    },
    {
      "epoch": 0.08647,
      "grad_norm": 0.9852125911616223,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 8647
    },
    {
      "epoch": 0.08648,
      "grad_norm": 0.9203439827975389,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 8648
    },
    {
      "epoch": 0.08649,
      "grad_norm": 0.8551894521735353,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 8649
    },
    {
      "epoch": 0.0865,
      "grad_norm": 0.752190786342537,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 8650
    },
    {
      "epoch": 0.08651,
      "grad_norm": 0.736051245410185,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 8651
    },
    {
      "epoch": 0.08652,
      "grad_norm": 0.6463683221296629,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 8652
    },
    {
      "epoch": 0.08653,
      "grad_norm": 0.6704399865370819,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 8653
    },
    {
      "epoch": 0.08654,
      "grad_norm": 0.7075584672548234,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 8654
    },
    {
      "epoch": 0.08655,
      "grad_norm": 0.7589064409569638,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 8655
    },
    {
      "epoch": 0.08656,
      "grad_norm": 0.8797294460543287,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 8656
    },
    {
      "epoch": 0.08657,
      "grad_norm": 1.0746650206858512,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 8657
    },
    {
      "epoch": 0.08658,
      "grad_norm": 1.0381315037441246,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 8658
    },
    {
      "epoch": 0.08659,
      "grad_norm": 0.9216590731223901,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 8659
    },
    {
      "epoch": 0.0866,
      "grad_norm": 0.9060927251078454,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 8660
    },
    {
      "epoch": 0.08661,
      "grad_norm": 0.8365841556281326,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 8661
    },
    {
      "epoch": 0.08662,
      "grad_norm": 0.940796854284168,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 8662
    },
    {
      "epoch": 0.08663,
      "grad_norm": 1.1158443930232451,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 8663
    },
    {
      "epoch": 0.08664,
      "grad_norm": 1.0186570224186573,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8664
    },
    {
      "epoch": 0.08665,
      "grad_norm": 1.0348438398835733,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 8665
    },
    {
      "epoch": 0.08666,
      "grad_norm": 0.9620834884254315,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8666
    },
    {
      "epoch": 0.08667,
      "grad_norm": 1.006500389771506,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 8667
    },
    {
      "epoch": 0.08668,
      "grad_norm": 0.9850331378360856,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 8668
    },
    {
      "epoch": 0.08669,
      "grad_norm": 0.6894471054181678,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 8669
    },
    {
      "epoch": 0.0867,
      "grad_norm": 0.6694894610529343,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 8670
    },
    {
      "epoch": 0.08671,
      "grad_norm": 0.5962672000226233,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 8671
    },
    {
      "epoch": 0.08672,
      "grad_norm": 0.6763083203036772,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 8672
    },
    {
      "epoch": 0.08673,
      "grad_norm": 0.7676489015109336,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 8673
    },
    {
      "epoch": 0.08674,
      "grad_norm": 0.7787151403749619,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 8674
    },
    {
      "epoch": 0.08675,
      "grad_norm": 0.7891229989108475,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 8675
    },
    {
      "epoch": 0.08676,
      "grad_norm": 0.8927860274538245,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 8676
    },
    {
      "epoch": 0.08677,
      "grad_norm": 1.067464395193778,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 8677
    },
    {
      "epoch": 0.08678,
      "grad_norm": 0.8827877751142491,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 8678
    },
    {
      "epoch": 0.08679,
      "grad_norm": 0.7244826450369652,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 8679
    },
    {
      "epoch": 0.0868,
      "grad_norm": 0.633006551065808,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 8680
    },
    {
      "epoch": 0.08681,
      "grad_norm": 0.7719505462142406,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 8681
    },
    {
      "epoch": 0.08682,
      "grad_norm": 1.0884331781139083,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 8682
    },
    {
      "epoch": 0.08683,
      "grad_norm": 1.1714352580907677,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 8683
    },
    {
      "epoch": 0.08684,
      "grad_norm": 1.0196218025826278,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 8684
    },
    {
      "epoch": 0.08685,
      "grad_norm": 1.0458752270945595,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 8685
    },
    {
      "epoch": 0.08686,
      "grad_norm": 0.9641978455934338,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 8686
    },
    {
      "epoch": 0.08687,
      "grad_norm": 0.9135236516973639,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 8687
    },
    {
      "epoch": 0.08688,
      "grad_norm": 0.8512259905499174,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8688
    },
    {
      "epoch": 0.08689,
      "grad_norm": 0.8190255834805528,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 8689
    },
    {
      "epoch": 0.0869,
      "grad_norm": 0.8398166759029391,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 8690
    },
    {
      "epoch": 0.08691,
      "grad_norm": 0.8574277247566975,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 8691
    },
    {
      "epoch": 0.08692,
      "grad_norm": 0.9518906449211479,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 8692
    },
    {
      "epoch": 0.08693,
      "grad_norm": 0.9069434842808899,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 8693
    },
    {
      "epoch": 0.08694,
      "grad_norm": 0.7908401280606225,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 8694
    },
    {
      "epoch": 0.08695,
      "grad_norm": 0.875710856836608,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 8695
    },
    {
      "epoch": 0.08696,
      "grad_norm": 1.0582108487000337,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 8696
    },
    {
      "epoch": 0.08697,
      "grad_norm": 1.1048858177228473,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 8697
    },
    {
      "epoch": 0.08698,
      "grad_norm": 1.0885309061884358,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 8698
    },
    {
      "epoch": 0.08699,
      "grad_norm": 1.0780097630336325,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 8699
    },
    {
      "epoch": 0.087,
      "grad_norm": 0.8462640972072897,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 8700
    },
    {
      "epoch": 0.08701,
      "grad_norm": 0.7500583295478122,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 8701
    },
    {
      "epoch": 0.08702,
      "grad_norm": 0.8192338624119824,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8702
    },
    {
      "epoch": 0.08703,
      "grad_norm": 0.9761783341836241,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 8703
    },
    {
      "epoch": 0.08704,
      "grad_norm": 1.0637726019049591,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 8704
    },
    {
      "epoch": 0.08705,
      "grad_norm": 1.039691640283121,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 8705
    },
    {
      "epoch": 0.08706,
      "grad_norm": 1.0345969448915813,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 8706
    },
    {
      "epoch": 0.08707,
      "grad_norm": 1.073442317904122,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 8707
    },
    {
      "epoch": 0.08708,
      "grad_norm": 1.0167968758055734,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 8708
    },
    {
      "epoch": 0.08709,
      "grad_norm": 1.1281785403043407,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 8709
    },
    {
      "epoch": 0.0871,
      "grad_norm": 0.856237067855571,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 8710
    },
    {
      "epoch": 0.08711,
      "grad_norm": 0.7245271989076703,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 8711
    },
    {
      "epoch": 0.08712,
      "grad_norm": 0.6348264826820577,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 8712
    },
    {
      "epoch": 0.08713,
      "grad_norm": 0.7213305248825544,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 8713
    },
    {
      "epoch": 0.08714,
      "grad_norm": 0.7267634259694989,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 8714
    },
    {
      "epoch": 0.08715,
      "grad_norm": 0.7226602431965007,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 8715
    },
    {
      "epoch": 0.08716,
      "grad_norm": 0.749512188800938,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 8716
    },
    {
      "epoch": 0.08717,
      "grad_norm": 0.7945844637136416,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 8717
    },
    {
      "epoch": 0.08718,
      "grad_norm": 0.9455334040170957,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 8718
    },
    {
      "epoch": 0.08719,
      "grad_norm": 1.2301600366994065,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 8719
    },
    {
      "epoch": 0.0872,
      "grad_norm": 0.9688272507046074,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 8720
    },
    {
      "epoch": 0.08721,
      "grad_norm": 0.882130678243584,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 8721
    },
    {
      "epoch": 0.08722,
      "grad_norm": 0.7624823510784595,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 8722
    },
    {
      "epoch": 0.08723,
      "grad_norm": 0.8208175149976452,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 8723
    },
    {
      "epoch": 0.08724,
      "grad_norm": 1.0208221289653825,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 8724
    },
    {
      "epoch": 0.08725,
      "grad_norm": 1.1583066056086238,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 8725
    },
    {
      "epoch": 0.08726,
      "grad_norm": 0.8636528049466422,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 8726
    },
    {
      "epoch": 0.08727,
      "grad_norm": 0.7315027868053344,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 8727
    },
    {
      "epoch": 0.08728,
      "grad_norm": 0.7105524649657913,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 8728
    },
    {
      "epoch": 0.08729,
      "grad_norm": 0.8027256735287427,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 8729
    },
    {
      "epoch": 0.0873,
      "grad_norm": 0.7642689263014413,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 8730
    },
    {
      "epoch": 0.08731,
      "grad_norm": 0.8054605331616608,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 8731
    },
    {
      "epoch": 0.08732,
      "grad_norm": 0.8920977986550327,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 8732
    },
    {
      "epoch": 0.08733,
      "grad_norm": 0.8107240409545715,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 8733
    },
    {
      "epoch": 0.08734,
      "grad_norm": 0.8548040032308308,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 8734
    },
    {
      "epoch": 0.08735,
      "grad_norm": 1.2015525237447153,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 8735
    },
    {
      "epoch": 0.08736,
      "grad_norm": 1.0254473175716212,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 8736
    },
    {
      "epoch": 0.08737,
      "grad_norm": 0.9567188089886024,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 8737
    },
    {
      "epoch": 0.08738,
      "grad_norm": 1.0271105193607937,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 8738
    },
    {
      "epoch": 0.08739,
      "grad_norm": 0.9282295321228008,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8739
    },
    {
      "epoch": 0.0874,
      "grad_norm": 0.8558073220103064,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 8740
    },
    {
      "epoch": 0.08741,
      "grad_norm": 0.8429315708538133,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 8741
    },
    {
      "epoch": 0.08742,
      "grad_norm": 0.8121185242286065,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 8742
    },
    {
      "epoch": 0.08743,
      "grad_norm": 0.9197196989946418,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 8743
    },
    {
      "epoch": 0.08744,
      "grad_norm": 1.1348863902424295,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 8744
    },
    {
      "epoch": 0.08745,
      "grad_norm": 0.9755220575900516,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 8745
    },
    {
      "epoch": 0.08746,
      "grad_norm": 1.1617288318988086,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 8746
    },
    {
      "epoch": 0.08747,
      "grad_norm": 0.8515070893905221,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 8747
    },
    {
      "epoch": 0.08748,
      "grad_norm": 0.8210706822748797,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 8748
    },
    {
      "epoch": 0.08749,
      "grad_norm": 0.8533026928432937,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 8749
    },
    {
      "epoch": 0.0875,
      "grad_norm": 0.7247474091640602,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 8750
    },
    {
      "epoch": 0.08751,
      "grad_norm": 0.7613678855591323,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 8751
    },
    {
      "epoch": 0.08752,
      "grad_norm": 0.8298586709080547,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 8752
    },
    {
      "epoch": 0.08753,
      "grad_norm": 0.9510227968156119,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 8753
    },
    {
      "epoch": 0.08754,
      "grad_norm": 1.1661555141798292,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 8754
    },
    {
      "epoch": 0.08755,
      "grad_norm": 0.9742950157697924,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 8755
    },
    {
      "epoch": 0.08756,
      "grad_norm": 1.0948999240878925,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 8756
    },
    {
      "epoch": 0.08757,
      "grad_norm": 1.0747066998581816,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 8757
    },
    {
      "epoch": 0.08758,
      "grad_norm": 0.9803826009713699,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 8758
    },
    {
      "epoch": 0.08759,
      "grad_norm": 0.8978411068790542,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 8759
    },
    {
      "epoch": 0.0876,
      "grad_norm": 0.8777548601783808,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8760
    },
    {
      "epoch": 0.08761,
      "grad_norm": 0.8296530204156158,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 8761
    },
    {
      "epoch": 0.08762,
      "grad_norm": 0.9503211821723436,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 8762
    },
    {
      "epoch": 0.08763,
      "grad_norm": 0.9498831438790756,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 8763
    },
    {
      "epoch": 0.08764,
      "grad_norm": 0.9656491920428545,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 8764
    },
    {
      "epoch": 0.08765,
      "grad_norm": 1.0147165951625998,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 8765
    },
    {
      "epoch": 0.08766,
      "grad_norm": 0.9826351379893413,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 8766
    },
    {
      "epoch": 0.08767,
      "grad_norm": 1.0458878090436103,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 8767
    },
    {
      "epoch": 0.08768,
      "grad_norm": 1.0251752707071955,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 8768
    },
    {
      "epoch": 0.08769,
      "grad_norm": 0.9012853792156551,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8769
    },
    {
      "epoch": 0.0877,
      "grad_norm": 0.8035752435955863,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 8770
    },
    {
      "epoch": 0.08771,
      "grad_norm": 0.7518314170540558,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 8771
    },
    {
      "epoch": 0.08772,
      "grad_norm": 0.8180447348392333,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 8772
    },
    {
      "epoch": 0.08773,
      "grad_norm": 0.7147251860859791,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 8773
    },
    {
      "epoch": 0.08774,
      "grad_norm": 0.5867039223653453,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 8774
    },
    {
      "epoch": 0.08775,
      "grad_norm": 0.5527229194986593,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 8775
    },
    {
      "epoch": 0.08776,
      "grad_norm": 0.5844947741413182,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 8776
    },
    {
      "epoch": 0.08777,
      "grad_norm": 0.6563370684489898,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 8777
    },
    {
      "epoch": 0.08778,
      "grad_norm": 0.7712524036141793,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8778
    },
    {
      "epoch": 0.08779,
      "grad_norm": 0.7807517822630886,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 8779
    },
    {
      "epoch": 0.0878,
      "grad_norm": 0.6904453164163412,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 8780
    },
    {
      "epoch": 0.08781,
      "grad_norm": 0.767764779325477,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 8781
    },
    {
      "epoch": 0.08782,
      "grad_norm": 0.9600307430515294,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 8782
    },
    {
      "epoch": 0.08783,
      "grad_norm": 1.286330289921701,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 8783
    },
    {
      "epoch": 0.08784,
      "grad_norm": 0.9521055086549031,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 8784
    },
    {
      "epoch": 0.08785,
      "grad_norm": 1.3646540084787944,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 8785
    },
    {
      "epoch": 0.08786,
      "grad_norm": 0.9165165437571451,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 8786
    },
    {
      "epoch": 0.08787,
      "grad_norm": 0.8674027071829526,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 8787
    },
    {
      "epoch": 0.08788,
      "grad_norm": 0.8808990624383837,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 8788
    },
    {
      "epoch": 0.08789,
      "grad_norm": 0.9789530665994413,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8789
    },
    {
      "epoch": 0.0879,
      "grad_norm": 1.05003444606044,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 8790
    },
    {
      "epoch": 0.08791,
      "grad_norm": 0.9193191175375216,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 8791
    },
    {
      "epoch": 0.08792,
      "grad_norm": 0.8116887310714239,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 8792
    },
    {
      "epoch": 0.08793,
      "grad_norm": 0.8040951439376696,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8793
    },
    {
      "epoch": 0.08794,
      "grad_norm": 0.8728936717925508,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 8794
    },
    {
      "epoch": 0.08795,
      "grad_norm": 0.9425258213709625,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 8795
    },
    {
      "epoch": 0.08796,
      "grad_norm": 0.9269620549507115,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 8796
    },
    {
      "epoch": 0.08797,
      "grad_norm": 0.9274274055304775,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 8797
    },
    {
      "epoch": 0.08798,
      "grad_norm": 1.0495786479990563,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 8798
    },
    {
      "epoch": 0.08799,
      "grad_norm": 1.0941402572479257,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 8799
    },
    {
      "epoch": 0.088,
      "grad_norm": 1.0409088434122218,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 8800
    },
    {
      "epoch": 0.08801,
      "grad_norm": 1.0008095022296148,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 8801
    },
    {
      "epoch": 0.08802,
      "grad_norm": 0.9260235726665792,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 8802
    },
    {
      "epoch": 0.08803,
      "grad_norm": 0.959188194655516,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 8803
    },
    {
      "epoch": 0.08804,
      "grad_norm": 1.0435507338866736,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 8804
    },
    {
      "epoch": 0.08805,
      "grad_norm": 0.9112940402613895,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 8805
    },
    {
      "epoch": 0.08806,
      "grad_norm": 0.9708013111525423,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 8806
    },
    {
      "epoch": 0.08807,
      "grad_norm": 1.0648214390050283,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 8807
    },
    {
      "epoch": 0.08808,
      "grad_norm": 0.996034423121219,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 8808
    },
    {
      "epoch": 0.08809,
      "grad_norm": 1.049344011106456,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8809
    },
    {
      "epoch": 0.0881,
      "grad_norm": 0.8742010348146958,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 8810
    },
    {
      "epoch": 0.08811,
      "grad_norm": 0.8862670461095695,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 8811
    },
    {
      "epoch": 0.08812,
      "grad_norm": 0.8360246704981421,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 8812
    },
    {
      "epoch": 0.08813,
      "grad_norm": 0.8761385565588603,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 8813
    },
    {
      "epoch": 0.08814,
      "grad_norm": 0.8913146587128812,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 8814
    },
    {
      "epoch": 0.08815,
      "grad_norm": 0.9207388295031492,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 8815
    },
    {
      "epoch": 0.08816,
      "grad_norm": 1.0838936761521594,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 8816
    },
    {
      "epoch": 0.08817,
      "grad_norm": 0.967762286433456,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 8817
    },
    {
      "epoch": 0.08818,
      "grad_norm": 0.8962809858840016,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 8818
    },
    {
      "epoch": 0.08819,
      "grad_norm": 0.7701740177256127,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 8819
    },
    {
      "epoch": 0.0882,
      "grad_norm": 0.7258766499158136,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 8820
    },
    {
      "epoch": 0.08821,
      "grad_norm": 0.7058532115848066,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 8821
    },
    {
      "epoch": 0.08822,
      "grad_norm": 0.7289621370989584,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 8822
    },
    {
      "epoch": 0.08823,
      "grad_norm": 0.9370302189185644,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 8823
    },
    {
      "epoch": 0.08824,
      "grad_norm": 1.1037702165909034,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 8824
    },
    {
      "epoch": 0.08825,
      "grad_norm": 0.8276194064577156,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 8825
    },
    {
      "epoch": 0.08826,
      "grad_norm": 0.7237687402112217,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 8826
    },
    {
      "epoch": 0.08827,
      "grad_norm": 0.6068281322897773,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 8827
    },
    {
      "epoch": 0.08828,
      "grad_norm": 0.6236195415565339,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 8828
    },
    {
      "epoch": 0.08829,
      "grad_norm": 0.6501295996843758,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 8829
    },
    {
      "epoch": 0.0883,
      "grad_norm": 0.7182692035416562,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 8830
    },
    {
      "epoch": 0.08831,
      "grad_norm": 0.825778877664003,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 8831
    },
    {
      "epoch": 0.08832,
      "grad_norm": 0.8409707101378571,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 8832
    },
    {
      "epoch": 0.08833,
      "grad_norm": 0.8688218779665827,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 8833
    },
    {
      "epoch": 0.08834,
      "grad_norm": 1.053985971333072,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 8834
    },
    {
      "epoch": 0.08835,
      "grad_norm": 1.2011317648997946,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 8835
    },
    {
      "epoch": 0.08836,
      "grad_norm": 0.8675347336538084,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 8836
    },
    {
      "epoch": 0.08837,
      "grad_norm": 0.6847724769015546,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 8837
    },
    {
      "epoch": 0.08838,
      "grad_norm": 0.7664193052446167,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8838
    },
    {
      "epoch": 0.08839,
      "grad_norm": 0.7849476506437728,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 8839
    },
    {
      "epoch": 0.0884,
      "grad_norm": 0.8581420457247539,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 8840
    },
    {
      "epoch": 0.08841,
      "grad_norm": 1.1437255891991425,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 8841
    },
    {
      "epoch": 0.08842,
      "grad_norm": 1.0640839104480644,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 8842
    },
    {
      "epoch": 0.08843,
      "grad_norm": 0.8177173898310196,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 8843
    },
    {
      "epoch": 0.08844,
      "grad_norm": 0.7006811082685729,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 8844
    },
    {
      "epoch": 0.08845,
      "grad_norm": 0.6654586857775031,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 8845
    },
    {
      "epoch": 0.08846,
      "grad_norm": 0.7828442286838105,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 8846
    },
    {
      "epoch": 0.08847,
      "grad_norm": 1.002596697776703,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 8847
    },
    {
      "epoch": 0.08848,
      "grad_norm": 1.0769646848289371,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 8848
    },
    {
      "epoch": 0.08849,
      "grad_norm": 0.9588089835995077,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 8849
    },
    {
      "epoch": 0.0885,
      "grad_norm": 0.9736041532230283,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 8850
    },
    {
      "epoch": 0.08851,
      "grad_norm": 1.0038135324969604,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 8851
    },
    {
      "epoch": 0.08852,
      "grad_norm": 0.8144136857171164,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 8852
    },
    {
      "epoch": 0.08853,
      "grad_norm": 0.7492869712538301,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 8853
    },
    {
      "epoch": 0.08854,
      "grad_norm": 0.7964882824236851,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 8854
    },
    {
      "epoch": 0.08855,
      "grad_norm": 0.8940261132351949,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 8855
    },
    {
      "epoch": 0.08856,
      "grad_norm": 0.954650133885222,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 8856
    },
    {
      "epoch": 0.08857,
      "grad_norm": 0.9861048909441799,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 8857
    },
    {
      "epoch": 0.08858,
      "grad_norm": 1.0148434704216471,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 8858
    },
    {
      "epoch": 0.08859,
      "grad_norm": 1.2410690598958254,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 8859
    },
    {
      "epoch": 0.0886,
      "grad_norm": 0.9416605613389235,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 8860
    },
    {
      "epoch": 0.08861,
      "grad_norm": 0.8801276255413449,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 8861
    },
    {
      "epoch": 0.08862,
      "grad_norm": 0.867667458558974,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8862
    },
    {
      "epoch": 0.08863,
      "grad_norm": 0.9625185944766314,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 8863
    },
    {
      "epoch": 0.08864,
      "grad_norm": 1.025113767580424,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 8864
    },
    {
      "epoch": 0.08865,
      "grad_norm": 1.0305392893083485,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 8865
    },
    {
      "epoch": 0.08866,
      "grad_norm": 1.0855251168096982,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 8866
    },
    {
      "epoch": 0.08867,
      "grad_norm": 0.9690901901735053,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 8867
    },
    {
      "epoch": 0.08868,
      "grad_norm": 0.9930947313572007,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 8868
    },
    {
      "epoch": 0.08869,
      "grad_norm": 1.0329659460825737,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 8869
    },
    {
      "epoch": 0.0887,
      "grad_norm": 0.962624300853848,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 8870
    },
    {
      "epoch": 0.08871,
      "grad_norm": 0.8967371579665262,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 8871
    },
    {
      "epoch": 0.08872,
      "grad_norm": 0.8242663665328195,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 8872
    },
    {
      "epoch": 0.08873,
      "grad_norm": 0.8080330802630397,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 8873
    },
    {
      "epoch": 0.08874,
      "grad_norm": 0.8320866327498992,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8874
    },
    {
      "epoch": 0.08875,
      "grad_norm": 0.8237590696825445,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 8875
    },
    {
      "epoch": 0.08876,
      "grad_norm": 0.7670191226856026,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 8876
    },
    {
      "epoch": 0.08877,
      "grad_norm": 0.8150234000038336,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 8877
    },
    {
      "epoch": 0.08878,
      "grad_norm": 0.9188804850843868,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 8878
    },
    {
      "epoch": 0.08879,
      "grad_norm": 1.1505868854127517,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 8879
    },
    {
      "epoch": 0.0888,
      "grad_norm": 1.0381859644112361,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 8880
    },
    {
      "epoch": 0.08881,
      "grad_norm": 0.8126207042081777,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 8881
    },
    {
      "epoch": 0.08882,
      "grad_norm": 0.6375023727555839,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 8882
    },
    {
      "epoch": 0.08883,
      "grad_norm": 0.8121388182245645,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 8883
    },
    {
      "epoch": 0.08884,
      "grad_norm": 1.0502326054062787,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 8884
    },
    {
      "epoch": 0.08885,
      "grad_norm": 1.2283294749042548,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 8885
    },
    {
      "epoch": 0.08886,
      "grad_norm": 0.749971714680993,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 8886
    },
    {
      "epoch": 0.08887,
      "grad_norm": 0.8691386698939016,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 8887
    },
    {
      "epoch": 0.08888,
      "grad_norm": 0.9459570536683495,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 8888
    },
    {
      "epoch": 0.08889,
      "grad_norm": 1.0034071726990192,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 8889
    },
    {
      "epoch": 0.0889,
      "grad_norm": 1.1036828603076412,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 8890
    },
    {
      "epoch": 0.08891,
      "grad_norm": 0.7547248732778705,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 8891
    },
    {
      "epoch": 0.08892,
      "grad_norm": 0.7125455249994503,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 8892
    },
    {
      "epoch": 0.08893,
      "grad_norm": 0.9110389886249487,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 8893
    },
    {
      "epoch": 0.08894,
      "grad_norm": 1.0085325989909744,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 8894
    },
    {
      "epoch": 0.08895,
      "grad_norm": 0.9395630146430425,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 8895
    },
    {
      "epoch": 0.08896,
      "grad_norm": 0.932005587780262,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 8896
    },
    {
      "epoch": 0.08897,
      "grad_norm": 0.9683101599385363,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 8897
    },
    {
      "epoch": 0.08898,
      "grad_norm": 0.8778233714006216,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8898
    },
    {
      "epoch": 0.08899,
      "grad_norm": 0.8843104049066195,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 8899
    },
    {
      "epoch": 0.089,
      "grad_norm": 0.8611736022905074,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 8900
    },
    {
      "epoch": 0.08901,
      "grad_norm": 0.7675741709312146,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 8901
    },
    {
      "epoch": 0.08902,
      "grad_norm": 0.7542143956414811,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 8902
    },
    {
      "epoch": 0.08903,
      "grad_norm": 0.824143498268071,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 8903
    },
    {
      "epoch": 0.08904,
      "grad_norm": 1.0036741311146173,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 8904
    },
    {
      "epoch": 0.08905,
      "grad_norm": 1.1665427882326276,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 8905
    },
    {
      "epoch": 0.08906,
      "grad_norm": 0.9167446591164601,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 8906
    },
    {
      "epoch": 0.08907,
      "grad_norm": 0.8151193889750405,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 8907
    },
    {
      "epoch": 0.08908,
      "grad_norm": 0.8148807348539009,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 8908
    },
    {
      "epoch": 0.08909,
      "grad_norm": 0.8006264904937266,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 8909
    },
    {
      "epoch": 0.0891,
      "grad_norm": 0.7847183342716237,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 8910
    },
    {
      "epoch": 0.08911,
      "grad_norm": 0.871137922082431,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 8911
    },
    {
      "epoch": 0.08912,
      "grad_norm": 0.9186358754285994,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 8912
    },
    {
      "epoch": 0.08913,
      "grad_norm": 0.8792963429484404,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 8913
    },
    {
      "epoch": 0.08914,
      "grad_norm": 0.8389522069822098,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 8914
    },
    {
      "epoch": 0.08915,
      "grad_norm": 0.7438660035575635,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 8915
    },
    {
      "epoch": 0.08916,
      "grad_norm": 0.81031482315623,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8916
    },
    {
      "epoch": 0.08917,
      "grad_norm": 0.9879431715632052,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 8917
    },
    {
      "epoch": 0.08918,
      "grad_norm": 1.1263164984050484,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 8918
    },
    {
      "epoch": 0.08919,
      "grad_norm": 0.8217791465004585,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 8919
    },
    {
      "epoch": 0.0892,
      "grad_norm": 0.838741117076949,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 8920
    },
    {
      "epoch": 0.08921,
      "grad_norm": 0.8013123325103196,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 8921
    },
    {
      "epoch": 0.08922,
      "grad_norm": 0.8829008386490941,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 8922
    },
    {
      "epoch": 0.08923,
      "grad_norm": 0.831010513545994,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 8923
    },
    {
      "epoch": 0.08924,
      "grad_norm": 0.8026665557631941,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8924
    },
    {
      "epoch": 0.08925,
      "grad_norm": 0.7516895419563575,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 8925
    },
    {
      "epoch": 0.08926,
      "grad_norm": 0.6859708086212067,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 8926
    },
    {
      "epoch": 0.08927,
      "grad_norm": 0.705876228465091,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 8927
    },
    {
      "epoch": 0.08928,
      "grad_norm": 0.8197442137549819,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 8928
    },
    {
      "epoch": 0.08929,
      "grad_norm": 1.158197035763089,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 8929
    },
    {
      "epoch": 0.0893,
      "grad_norm": 1.2835489288131519,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 8930
    },
    {
      "epoch": 0.08931,
      "grad_norm": 0.9255470549413722,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 8931
    },
    {
      "epoch": 0.08932,
      "grad_norm": 0.9494109894231941,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 8932
    },
    {
      "epoch": 0.08933,
      "grad_norm": 0.9325764456947372,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 8933
    },
    {
      "epoch": 0.08934,
      "grad_norm": 0.9120790589940708,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 8934
    },
    {
      "epoch": 0.08935,
      "grad_norm": 0.9622123789895787,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 8935
    },
    {
      "epoch": 0.08936,
      "grad_norm": 0.9982089757464768,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 8936
    },
    {
      "epoch": 0.08937,
      "grad_norm": 0.8488183553743273,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 8937
    },
    {
      "epoch": 0.08938,
      "grad_norm": 0.711168048268646,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 8938
    },
    {
      "epoch": 0.08939,
      "grad_norm": 0.7145069662449071,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 8939
    },
    {
      "epoch": 0.0894,
      "grad_norm": 0.9351641300422199,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 8940
    },
    {
      "epoch": 0.08941,
      "grad_norm": 0.9908374405712305,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 8941
    },
    {
      "epoch": 0.08942,
      "grad_norm": 1.0369857459287184,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 8942
    },
    {
      "epoch": 0.08943,
      "grad_norm": 0.9815132194398201,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 8943
    },
    {
      "epoch": 0.08944,
      "grad_norm": 0.8275387834639889,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 8944
    },
    {
      "epoch": 0.08945,
      "grad_norm": 0.7363945481987325,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 8945
    },
    {
      "epoch": 0.08946,
      "grad_norm": 0.7493687013662418,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 8946
    },
    {
      "epoch": 0.08947,
      "grad_norm": 0.8635191868013845,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 8947
    },
    {
      "epoch": 0.08948,
      "grad_norm": 0.8462786471943283,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 8948
    },
    {
      "epoch": 0.08949,
      "grad_norm": 0.8671422196356308,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 8949
    },
    {
      "epoch": 0.0895,
      "grad_norm": 0.9409364514795033,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 8950
    },
    {
      "epoch": 0.08951,
      "grad_norm": 1.090641876171059,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 8951
    },
    {
      "epoch": 0.08952,
      "grad_norm": 1.0606427376589673,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 8952
    },
    {
      "epoch": 0.08953,
      "grad_norm": 0.9918378904457558,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 8953
    },
    {
      "epoch": 0.08954,
      "grad_norm": 0.9774618750373327,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 8954
    },
    {
      "epoch": 0.08955,
      "grad_norm": 1.2355673972258312,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 8955
    },
    {
      "epoch": 0.08956,
      "grad_norm": 0.958337187094943,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 8956
    },
    {
      "epoch": 0.08957,
      "grad_norm": 1.1059278113358553,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 8957
    },
    {
      "epoch": 0.08958,
      "grad_norm": 1.0600369480191711,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 8958
    },
    {
      "epoch": 0.08959,
      "grad_norm": 0.9044461430911768,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8959
    },
    {
      "epoch": 0.0896,
      "grad_norm": 0.751672034162579,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 8960
    },
    {
      "epoch": 0.08961,
      "grad_norm": 0.6890812720881869,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 8961
    },
    {
      "epoch": 0.08962,
      "grad_norm": 0.6420758804789315,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 8962
    },
    {
      "epoch": 0.08963,
      "grad_norm": 0.7715824227745989,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 8963
    },
    {
      "epoch": 0.08964,
      "grad_norm": 0.9769914979893802,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 8964
    },
    {
      "epoch": 0.08965,
      "grad_norm": 1.1055411204729655,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 8965
    },
    {
      "epoch": 0.08966,
      "grad_norm": 0.8858925381363988,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 8966
    },
    {
      "epoch": 0.08967,
      "grad_norm": 0.973405476101832,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 8967
    },
    {
      "epoch": 0.08968,
      "grad_norm": 1.246488899818171,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 8968
    },
    {
      "epoch": 0.08969,
      "grad_norm": 0.7919830012222802,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 8969
    },
    {
      "epoch": 0.0897,
      "grad_norm": 0.783905488272791,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 8970
    },
    {
      "epoch": 0.08971,
      "grad_norm": 0.8587360970358755,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8971
    },
    {
      "epoch": 0.08972,
      "grad_norm": 1.1812204010197107,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 8972
    },
    {
      "epoch": 0.08973,
      "grad_norm": 1.072340183548216,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 8973
    },
    {
      "epoch": 0.08974,
      "grad_norm": 0.9624889318642255,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 8974
    },
    {
      "epoch": 0.08975,
      "grad_norm": 1.103329859592797,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 8975
    },
    {
      "epoch": 0.08976,
      "grad_norm": 0.9226100631400868,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 8976
    },
    {
      "epoch": 0.08977,
      "grad_norm": 0.9050852208024028,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8977
    },
    {
      "epoch": 0.08978,
      "grad_norm": 0.9593748971336933,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8978
    },
    {
      "epoch": 0.08979,
      "grad_norm": 0.917813099791213,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 8979
    },
    {
      "epoch": 0.0898,
      "grad_norm": 0.850359408142017,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 8980
    },
    {
      "epoch": 0.08981,
      "grad_norm": 0.7873512624929737,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 8981
    },
    {
      "epoch": 0.08982,
      "grad_norm": 0.7460904455327514,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 8982
    },
    {
      "epoch": 0.08983,
      "grad_norm": 0.813398130626127,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 8983
    },
    {
      "epoch": 0.08984,
      "grad_norm": 1.0379519181428254,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 8984
    },
    {
      "epoch": 0.08985,
      "grad_norm": 1.2369546672505778,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 8985
    },
    {
      "epoch": 0.08986,
      "grad_norm": 0.7482892312093569,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 8986
    },
    {
      "epoch": 0.08987,
      "grad_norm": 0.6543486223493022,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 8987
    },
    {
      "epoch": 0.08988,
      "grad_norm": 0.6451367251570368,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 8988
    },
    {
      "epoch": 0.08989,
      "grad_norm": 0.601214981373186,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 8989
    },
    {
      "epoch": 0.0899,
      "grad_norm": 0.670157839844488,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 8990
    },
    {
      "epoch": 0.08991,
      "grad_norm": 0.6344666888841743,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 8991
    },
    {
      "epoch": 0.08992,
      "grad_norm": 0.6087663409550896,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 8992
    },
    {
      "epoch": 0.08993,
      "grad_norm": 0.6007106864127649,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 8993
    },
    {
      "epoch": 0.08994,
      "grad_norm": 0.6569404032054016,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 8994
    },
    {
      "epoch": 0.08995,
      "grad_norm": 0.8011788589422568,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8995
    },
    {
      "epoch": 0.08996,
      "grad_norm": 1.090537283208205,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 8996
    },
    {
      "epoch": 0.08997,
      "grad_norm": 1.259139252903192,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 8997
    },
    {
      "epoch": 0.08998,
      "grad_norm": 0.7711113115206871,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 8998
    },
    {
      "epoch": 0.08999,
      "grad_norm": 0.7350565093378323,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 8999
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8212463674203402,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 9000
    },
    {
      "epoch": 0.09001,
      "grad_norm": 0.9909623667562864,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9001
    },
    {
      "epoch": 0.09002,
      "grad_norm": 1.1337359570952938,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 9002
    },
    {
      "epoch": 0.09003,
      "grad_norm": 1.066025616693887,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 9003
    },
    {
      "epoch": 0.09004,
      "grad_norm": 1.022429577767667,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 9004
    },
    {
      "epoch": 0.09005,
      "grad_norm": 0.8748962510761829,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 9005
    },
    {
      "epoch": 0.09006,
      "grad_norm": 0.7098491602250373,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 9006
    },
    {
      "epoch": 0.09007,
      "grad_norm": 0.5692107949898508,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 9007
    },
    {
      "epoch": 0.09008,
      "grad_norm": 0.5772606567632318,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 9008
    },
    {
      "epoch": 0.09009,
      "grad_norm": 0.6669070549022329,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 9009
    },
    {
      "epoch": 0.0901,
      "grad_norm": 0.8245290219344589,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 9010
    },
    {
      "epoch": 0.09011,
      "grad_norm": 0.9106422213480005,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 9011
    },
    {
      "epoch": 0.09012,
      "grad_norm": 0.9526910382185314,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 9012
    },
    {
      "epoch": 0.09013,
      "grad_norm": 0.9182834414849824,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 9013
    },
    {
      "epoch": 0.09014,
      "grad_norm": 0.8705840436567862,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 9014
    },
    {
      "epoch": 0.09015,
      "grad_norm": 0.9199037787555114,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 9015
    },
    {
      "epoch": 0.09016,
      "grad_norm": 0.8979198281963656,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 9016
    },
    {
      "epoch": 0.09017,
      "grad_norm": 0.9760173547686396,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 9017
    },
    {
      "epoch": 0.09018,
      "grad_norm": 1.0721385512227886,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 9018
    },
    {
      "epoch": 0.09019,
      "grad_norm": 0.9082460234466702,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 9019
    },
    {
      "epoch": 0.0902,
      "grad_norm": 0.9401369695786034,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 9020
    },
    {
      "epoch": 0.09021,
      "grad_norm": 0.9948251941982775,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 9021
    },
    {
      "epoch": 0.09022,
      "grad_norm": 1.1842814118486442,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 9022
    },
    {
      "epoch": 0.09023,
      "grad_norm": 0.7972897768988289,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9023
    },
    {
      "epoch": 0.09024,
      "grad_norm": 0.7421033982019426,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 9024
    },
    {
      "epoch": 0.09025,
      "grad_norm": 0.7738259101346501,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 9025
    },
    {
      "epoch": 0.09026,
      "grad_norm": 0.8970150206090364,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 9026
    },
    {
      "epoch": 0.09027,
      "grad_norm": 1.0492341252458535,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9027
    },
    {
      "epoch": 0.09028,
      "grad_norm": 1.2454585937869171,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 9028
    },
    {
      "epoch": 0.09029,
      "grad_norm": 0.9252985895422703,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 9029
    },
    {
      "epoch": 0.0903,
      "grad_norm": 0.9241514183975191,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 9030
    },
    {
      "epoch": 0.09031,
      "grad_norm": 0.8763169048046952,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 9031
    },
    {
      "epoch": 0.09032,
      "grad_norm": 0.9386352079429895,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 9032
    },
    {
      "epoch": 0.09033,
      "grad_norm": 0.9783200260031812,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 9033
    },
    {
      "epoch": 0.09034,
      "grad_norm": 0.956003427831922,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9034
    },
    {
      "epoch": 0.09035,
      "grad_norm": 0.8041920674510835,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 9035
    },
    {
      "epoch": 0.09036,
      "grad_norm": 0.8190103140470621,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 9036
    },
    {
      "epoch": 0.09037,
      "grad_norm": 0.8735918968075451,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 9037
    },
    {
      "epoch": 0.09038,
      "grad_norm": 0.8950344717934043,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 9038
    },
    {
      "epoch": 0.09039,
      "grad_norm": 0.8598737751053982,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 9039
    },
    {
      "epoch": 0.0904,
      "grad_norm": 0.909321180715369,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 9040
    },
    {
      "epoch": 0.09041,
      "grad_norm": 0.8046317272372915,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 9041
    },
    {
      "epoch": 0.09042,
      "grad_norm": 0.7635821438024983,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 9042
    },
    {
      "epoch": 0.09043,
      "grad_norm": 0.9240410411420402,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 9043
    },
    {
      "epoch": 0.09044,
      "grad_norm": 1.15129774565637,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9044
    },
    {
      "epoch": 0.09045,
      "grad_norm": 1.0176507372571526,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 9045
    },
    {
      "epoch": 0.09046,
      "grad_norm": 1.0168928227228877,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 9046
    },
    {
      "epoch": 0.09047,
      "grad_norm": 1.0855871224553029,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 9047
    },
    {
      "epoch": 0.09048,
      "grad_norm": 0.9615447887028904,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 9048
    },
    {
      "epoch": 0.09049,
      "grad_norm": 0.9780305253510888,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 9049
    },
    {
      "epoch": 0.0905,
      "grad_norm": 0.919712660817147,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 9050
    },
    {
      "epoch": 0.09051,
      "grad_norm": 0.9279054815642093,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 9051
    },
    {
      "epoch": 0.09052,
      "grad_norm": 0.9418205046994602,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 9052
    },
    {
      "epoch": 0.09053,
      "grad_norm": 0.9469408829030873,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 9053
    },
    {
      "epoch": 0.09054,
      "grad_norm": 0.9051974077534375,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 9054
    },
    {
      "epoch": 0.09055,
      "grad_norm": 1.024878799010692,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 9055
    },
    {
      "epoch": 0.09056,
      "grad_norm": 0.9922488913468747,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 9056
    },
    {
      "epoch": 0.09057,
      "grad_norm": 0.881282710691066,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 9057
    },
    {
      "epoch": 0.09058,
      "grad_norm": 0.9505568921531226,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 9058
    },
    {
      "epoch": 0.09059,
      "grad_norm": 0.9853808181646652,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 9059
    },
    {
      "epoch": 0.0906,
      "grad_norm": 1.085605727215449,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 9060
    },
    {
      "epoch": 0.09061,
      "grad_norm": 0.9708980429037549,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 9061
    },
    {
      "epoch": 0.09062,
      "grad_norm": 0.9437630749155476,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 9062
    },
    {
      "epoch": 0.09063,
      "grad_norm": 1.15349103657829,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 9063
    },
    {
      "epoch": 0.09064,
      "grad_norm": 0.9708247162873537,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 9064
    },
    {
      "epoch": 0.09065,
      "grad_norm": 0.8902413777847998,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 9065
    },
    {
      "epoch": 0.09066,
      "grad_norm": 0.737181277087745,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9066
    },
    {
      "epoch": 0.09067,
      "grad_norm": 0.6279056212510534,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 9067
    },
    {
      "epoch": 0.09068,
      "grad_norm": 0.6385574242216696,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 9068
    },
    {
      "epoch": 0.09069,
      "grad_norm": 0.6483082983150563,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 9069
    },
    {
      "epoch": 0.0907,
      "grad_norm": 0.8253555420224653,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 9070
    },
    {
      "epoch": 0.09071,
      "grad_norm": 1.0346138256073594,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 9071
    },
    {
      "epoch": 0.09072,
      "grad_norm": 0.9313433051670195,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 9072
    },
    {
      "epoch": 0.09073,
      "grad_norm": 1.0276295171442573,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 9073
    },
    {
      "epoch": 0.09074,
      "grad_norm": 1.1532562018862866,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 9074
    },
    {
      "epoch": 0.09075,
      "grad_norm": 1.0895198666143424,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 9075
    },
    {
      "epoch": 0.09076,
      "grad_norm": 0.9624509207513962,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 9076
    },
    {
      "epoch": 0.09077,
      "grad_norm": 1.0339320521373805,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 9077
    },
    {
      "epoch": 0.09078,
      "grad_norm": 1.0567530098664932,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 9078
    },
    {
      "epoch": 0.09079,
      "grad_norm": 0.9774925721478938,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 9079
    },
    {
      "epoch": 0.0908,
      "grad_norm": 1.0195847641854132,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 9080
    },
    {
      "epoch": 0.09081,
      "grad_norm": 1.0092501737148587,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 9081
    },
    {
      "epoch": 0.09082,
      "grad_norm": 1.2418266763070203,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 9082
    },
    {
      "epoch": 0.09083,
      "grad_norm": 0.7841216118125688,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 9083
    },
    {
      "epoch": 0.09084,
      "grad_norm": 0.6720714850940325,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 9084
    },
    {
      "epoch": 0.09085,
      "grad_norm": 0.8612009953168612,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 9085
    },
    {
      "epoch": 0.09086,
      "grad_norm": 1.0531662786767666,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 9086
    },
    {
      "epoch": 0.09087,
      "grad_norm": 1.0678011915356584,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9087
    },
    {
      "epoch": 0.09088,
      "grad_norm": 0.8076834182848771,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 9088
    },
    {
      "epoch": 0.09089,
      "grad_norm": 0.7750777971721545,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9089
    },
    {
      "epoch": 0.0909,
      "grad_norm": 0.8512525308462379,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 9090
    },
    {
      "epoch": 0.09091,
      "grad_norm": 1.0582316561541651,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 9091
    },
    {
      "epoch": 0.09092,
      "grad_norm": 0.898113597368355,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 9092
    },
    {
      "epoch": 0.09093,
      "grad_norm": 0.8629700654684181,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 9093
    },
    {
      "epoch": 0.09094,
      "grad_norm": 0.8939556429591207,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9094
    },
    {
      "epoch": 0.09095,
      "grad_norm": 0.8038507515784818,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 9095
    },
    {
      "epoch": 0.09096,
      "grad_norm": 0.7779918254246555,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 9096
    },
    {
      "epoch": 0.09097,
      "grad_norm": 0.7898567013017581,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 9097
    },
    {
      "epoch": 0.09098,
      "grad_norm": 0.7469344511651337,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 9098
    },
    {
      "epoch": 0.09099,
      "grad_norm": 0.81782758465319,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 9099
    },
    {
      "epoch": 0.091,
      "grad_norm": 0.8663932475321643,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 9100
    },
    {
      "epoch": 0.09101,
      "grad_norm": 0.9040908007482058,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 9101
    },
    {
      "epoch": 0.09102,
      "grad_norm": 0.8974025166455672,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 9102
    },
    {
      "epoch": 0.09103,
      "grad_norm": 0.8432481209256596,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 9103
    },
    {
      "epoch": 0.09104,
      "grad_norm": 0.7839104837670768,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 9104
    },
    {
      "epoch": 0.09105,
      "grad_norm": 0.74968201955386,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 9105
    },
    {
      "epoch": 0.09106,
      "grad_norm": 0.862192217242314,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 9106
    },
    {
      "epoch": 0.09107,
      "grad_norm": 0.8987410447628759,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 9107
    },
    {
      "epoch": 0.09108,
      "grad_norm": 0.834199691676135,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 9108
    },
    {
      "epoch": 0.09109,
      "grad_norm": 0.8002673934636545,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 9109
    },
    {
      "epoch": 0.0911,
      "grad_norm": 0.8991637907546197,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 9110
    },
    {
      "epoch": 0.09111,
      "grad_norm": 1.0150826030428626,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 9111
    },
    {
      "epoch": 0.09112,
      "grad_norm": 1.0652838006542034,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 9112
    },
    {
      "epoch": 0.09113,
      "grad_norm": 0.8745809726912822,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 9113
    },
    {
      "epoch": 0.09114,
      "grad_norm": 0.891905966917683,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 9114
    },
    {
      "epoch": 0.09115,
      "grad_norm": 1.0521275368839003,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 9115
    },
    {
      "epoch": 0.09116,
      "grad_norm": 0.8815584303045687,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 9116
    },
    {
      "epoch": 0.09117,
      "grad_norm": 1.0372762229109376,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 9117
    },
    {
      "epoch": 0.09118,
      "grad_norm": 0.9807823101415436,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 9118
    },
    {
      "epoch": 0.09119,
      "grad_norm": 1.0463681239596228,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 9119
    },
    {
      "epoch": 0.0912,
      "grad_norm": 1.1558735058794685,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 9120
    },
    {
      "epoch": 0.09121,
      "grad_norm": 0.9471193935945554,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 9121
    },
    {
      "epoch": 0.09122,
      "grad_norm": 0.903406256690084,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 9122
    },
    {
      "epoch": 0.09123,
      "grad_norm": 0.9082666157548727,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 9123
    },
    {
      "epoch": 0.09124,
      "grad_norm": 1.0288774290050668,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 9124
    },
    {
      "epoch": 0.09125,
      "grad_norm": 1.1563589417293623,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 9125
    },
    {
      "epoch": 0.09126,
      "grad_norm": 0.9664619902252881,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 9126
    },
    {
      "epoch": 0.09127,
      "grad_norm": 0.9629651919979498,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 9127
    },
    {
      "epoch": 0.09128,
      "grad_norm": 1.0135807050519996,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 9128
    },
    {
      "epoch": 0.09129,
      "grad_norm": 1.1527226810083393,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 9129
    },
    {
      "epoch": 0.0913,
      "grad_norm": 0.8454458533144666,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 9130
    },
    {
      "epoch": 0.09131,
      "grad_norm": 0.6217315912216276,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 9131
    },
    {
      "epoch": 0.09132,
      "grad_norm": 0.6894799817711807,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 9132
    },
    {
      "epoch": 0.09133,
      "grad_norm": 0.9164420443264122,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 9133
    },
    {
      "epoch": 0.09134,
      "grad_norm": 1.118137516546724,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 9134
    },
    {
      "epoch": 0.09135,
      "grad_norm": 0.9451104479671788,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 9135
    },
    {
      "epoch": 0.09136,
      "grad_norm": 0.8741863870514021,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 9136
    },
    {
      "epoch": 0.09137,
      "grad_norm": 0.6559657083546983,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 9137
    },
    {
      "epoch": 0.09138,
      "grad_norm": 0.6591621056680503,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9138
    },
    {
      "epoch": 0.09139,
      "grad_norm": 0.7172388233296422,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 9139
    },
    {
      "epoch": 0.0914,
      "grad_norm": 0.74698785265184,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9140
    },
    {
      "epoch": 0.09141,
      "grad_norm": 0.7383928212536637,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 9141
    },
    {
      "epoch": 0.09142,
      "grad_norm": 0.8365703977523314,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 9142
    },
    {
      "epoch": 0.09143,
      "grad_norm": 0.8481652164716268,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 9143
    },
    {
      "epoch": 0.09144,
      "grad_norm": 0.7795528073061184,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9144
    },
    {
      "epoch": 0.09145,
      "grad_norm": 0.8962209591026435,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 9145
    },
    {
      "epoch": 0.09146,
      "grad_norm": 0.977738810966034,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 9146
    },
    {
      "epoch": 0.09147,
      "grad_norm": 1.1158176717774442,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 9147
    },
    {
      "epoch": 0.09148,
      "grad_norm": 0.851745175787143,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 9148
    },
    {
      "epoch": 0.09149,
      "grad_norm": 0.7540599482452148,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 9149
    },
    {
      "epoch": 0.0915,
      "grad_norm": 0.8969362552019574,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 9150
    },
    {
      "epoch": 0.09151,
      "grad_norm": 0.9897498994292178,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 9151
    },
    {
      "epoch": 0.09152,
      "grad_norm": 1.0440364187474465,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 9152
    },
    {
      "epoch": 0.09153,
      "grad_norm": 1.1474475308871095,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9153
    },
    {
      "epoch": 0.09154,
      "grad_norm": 0.9705935043490699,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 9154
    },
    {
      "epoch": 0.09155,
      "grad_norm": 0.9298282598002533,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 9155
    },
    {
      "epoch": 0.09156,
      "grad_norm": 0.9723217314936119,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 9156
    },
    {
      "epoch": 0.09157,
      "grad_norm": 1.0948870894323015,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 9157
    },
    {
      "epoch": 0.09158,
      "grad_norm": 1.082303366161037,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 9158
    },
    {
      "epoch": 0.09159,
      "grad_norm": 1.1823492886345546,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 9159
    },
    {
      "epoch": 0.0916,
      "grad_norm": 1.1448872296285921,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9160
    },
    {
      "epoch": 0.09161,
      "grad_norm": 0.7917308786382252,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 9161
    },
    {
      "epoch": 0.09162,
      "grad_norm": 0.7101340083428053,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 9162
    },
    {
      "epoch": 0.09163,
      "grad_norm": 0.5734667662994228,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 9163
    },
    {
      "epoch": 0.09164,
      "grad_norm": 0.6463742170108849,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 9164
    },
    {
      "epoch": 0.09165,
      "grad_norm": 0.8791425623296205,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 9165
    },
    {
      "epoch": 0.09166,
      "grad_norm": 1.2896040065899281,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 9166
    },
    {
      "epoch": 0.09167,
      "grad_norm": 0.777626739229617,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 9167
    },
    {
      "epoch": 0.09168,
      "grad_norm": 0.6059045588929811,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 9168
    },
    {
      "epoch": 0.09169,
      "grad_norm": 0.6211082886708152,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 9169
    },
    {
      "epoch": 0.0917,
      "grad_norm": 0.6201173250342321,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 9170
    },
    {
      "epoch": 0.09171,
      "grad_norm": 0.65420012036481,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 9171
    },
    {
      "epoch": 0.09172,
      "grad_norm": 0.6193382027516,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 9172
    },
    {
      "epoch": 0.09173,
      "grad_norm": 0.726634103434902,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 9173
    },
    {
      "epoch": 0.09174,
      "grad_norm": 0.75555391022193,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 9174
    },
    {
      "epoch": 0.09175,
      "grad_norm": 0.791110324469862,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 9175
    },
    {
      "epoch": 0.09176,
      "grad_norm": 0.785799126594617,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 9176
    },
    {
      "epoch": 0.09177,
      "grad_norm": 0.7612319543811192,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 9177
    },
    {
      "epoch": 0.09178,
      "grad_norm": 0.7565419822439869,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 9178
    },
    {
      "epoch": 0.09179,
      "grad_norm": 0.8420544599698648,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 9179
    },
    {
      "epoch": 0.0918,
      "grad_norm": 1.084438958299728,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 9180
    },
    {
      "epoch": 0.09181,
      "grad_norm": 1.1059617305324483,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 9181
    },
    {
      "epoch": 0.09182,
      "grad_norm": 1.047379795485612,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 9182
    },
    {
      "epoch": 0.09183,
      "grad_norm": 0.9756788460870816,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 9183
    },
    {
      "epoch": 0.09184,
      "grad_norm": 0.8475124797770287,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 9184
    },
    {
      "epoch": 0.09185,
      "grad_norm": 0.7818162898325053,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 9185
    },
    {
      "epoch": 0.09186,
      "grad_norm": 0.7766932782419979,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 9186
    },
    {
      "epoch": 0.09187,
      "grad_norm": 0.8825420516075367,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 9187
    },
    {
      "epoch": 0.09188,
      "grad_norm": 0.9631123401662338,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9188
    },
    {
      "epoch": 0.09189,
      "grad_norm": 1.0225689776098181,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 9189
    },
    {
      "epoch": 0.0919,
      "grad_norm": 1.0782850500259313,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 9190
    },
    {
      "epoch": 0.09191,
      "grad_norm": 0.9063112295904093,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 9191
    },
    {
      "epoch": 0.09192,
      "grad_norm": 1.055089562560554,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 9192
    },
    {
      "epoch": 0.09193,
      "grad_norm": 1.014344237878195,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 9193
    },
    {
      "epoch": 0.09194,
      "grad_norm": 1.3217721747559106,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 9194
    },
    {
      "epoch": 0.09195,
      "grad_norm": 0.9570521535388152,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 9195
    },
    {
      "epoch": 0.09196,
      "grad_norm": 0.9200244611664627,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 9196
    },
    {
      "epoch": 0.09197,
      "grad_norm": 0.962705740113485,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 9197
    },
    {
      "epoch": 0.09198,
      "grad_norm": 1.0994040596641663,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 9198
    },
    {
      "epoch": 0.09199,
      "grad_norm": 0.9897723274209836,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 9199
    },
    {
      "epoch": 0.092,
      "grad_norm": 0.9542387690632056,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 9200
    },
    {
      "epoch": 0.09201,
      "grad_norm": 1.073948721463555,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 9201
    },
    {
      "epoch": 0.09202,
      "grad_norm": 0.9997228282811497,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9202
    },
    {
      "epoch": 0.09203,
      "grad_norm": 0.8556418820489938,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 9203
    },
    {
      "epoch": 0.09204,
      "grad_norm": 0.7789051082427513,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 9204
    },
    {
      "epoch": 0.09205,
      "grad_norm": 0.7717492134597749,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 9205
    },
    {
      "epoch": 0.09206,
      "grad_norm": 0.8428849907020309,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 9206
    },
    {
      "epoch": 0.09207,
      "grad_norm": 0.9040648516741944,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 9207
    },
    {
      "epoch": 0.09208,
      "grad_norm": 1.0913270867403186,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9208
    },
    {
      "epoch": 0.09209,
      "grad_norm": 1.0048770200215928,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 9209
    },
    {
      "epoch": 0.0921,
      "grad_norm": 1.0392328935432598,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 9210
    },
    {
      "epoch": 0.09211,
      "grad_norm": 0.9204701850435161,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 9211
    },
    {
      "epoch": 0.09212,
      "grad_norm": 0.9402955316963719,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 9212
    },
    {
      "epoch": 0.09213,
      "grad_norm": 1.048761874355027,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 9213
    },
    {
      "epoch": 0.09214,
      "grad_norm": 1.0562819020868857,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 9214
    },
    {
      "epoch": 0.09215,
      "grad_norm": 1.085719329634591,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 9215
    },
    {
      "epoch": 0.09216,
      "grad_norm": 0.9924362432989838,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 9216
    },
    {
      "epoch": 0.09217,
      "grad_norm": 0.9416068771195782,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 9217
    },
    {
      "epoch": 0.09218,
      "grad_norm": 0.9216157764389595,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 9218
    },
    {
      "epoch": 0.09219,
      "grad_norm": 1.0578289251402284,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 9219
    },
    {
      "epoch": 0.0922,
      "grad_norm": 0.8494550851933941,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 9220
    },
    {
      "epoch": 0.09221,
      "grad_norm": 0.7004867483817171,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 9221
    },
    {
      "epoch": 0.09222,
      "grad_norm": 0.7348105532675423,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 9222
    },
    {
      "epoch": 0.09223,
      "grad_norm": 0.9417547593730853,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 9223
    },
    {
      "epoch": 0.09224,
      "grad_norm": 1.049648042435812,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 9224
    },
    {
      "epoch": 0.09225,
      "grad_norm": 1.0276279527925067,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 9225
    },
    {
      "epoch": 0.09226,
      "grad_norm": 0.9585527602924722,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 9226
    },
    {
      "epoch": 0.09227,
      "grad_norm": 0.9337166610853146,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 9227
    },
    {
      "epoch": 0.09228,
      "grad_norm": 0.9027418181317508,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9228
    },
    {
      "epoch": 0.09229,
      "grad_norm": 0.9198727309478711,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 9229
    },
    {
      "epoch": 0.0923,
      "grad_norm": 0.9045649479252982,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 9230
    },
    {
      "epoch": 0.09231,
      "grad_norm": 0.9974652662909131,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 9231
    },
    {
      "epoch": 0.09232,
      "grad_norm": 1.0912379899673035,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 9232
    },
    {
      "epoch": 0.09233,
      "grad_norm": 1.0915766025748115,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 9233
    },
    {
      "epoch": 0.09234,
      "grad_norm": 0.9888878322311199,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 9234
    },
    {
      "epoch": 0.09235,
      "grad_norm": 0.8256741524071066,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 9235
    },
    {
      "epoch": 0.09236,
      "grad_norm": 0.7623729808775097,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 9236
    },
    {
      "epoch": 0.09237,
      "grad_norm": 0.784646386208245,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 9237
    },
    {
      "epoch": 0.09238,
      "grad_norm": 0.781447671218314,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 9238
    },
    {
      "epoch": 0.09239,
      "grad_norm": 0.7580100064874866,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 9239
    },
    {
      "epoch": 0.0924,
      "grad_norm": 0.7887138742063122,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 9240
    },
    {
      "epoch": 0.09241,
      "grad_norm": 0.9232343150935588,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 9241
    },
    {
      "epoch": 0.09242,
      "grad_norm": 1.1881392195909175,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 9242
    },
    {
      "epoch": 0.09243,
      "grad_norm": 0.9840929580945444,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 9243
    },
    {
      "epoch": 0.09244,
      "grad_norm": 1.304430856896462,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 9244
    },
    {
      "epoch": 0.09245,
      "grad_norm": 0.9351259127699327,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 9245
    },
    {
      "epoch": 0.09246,
      "grad_norm": 0.9793836268379075,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 9246
    },
    {
      "epoch": 0.09247,
      "grad_norm": 0.9197456699785576,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 9247
    },
    {
      "epoch": 0.09248,
      "grad_norm": 0.8196313202466566,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 9248
    },
    {
      "epoch": 0.09249,
      "grad_norm": 0.7928866730787516,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 9249
    },
    {
      "epoch": 0.0925,
      "grad_norm": 0.7531883062490411,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 9250
    },
    {
      "epoch": 0.09251,
      "grad_norm": 0.7666898332256793,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 9251
    },
    {
      "epoch": 0.09252,
      "grad_norm": 0.7988522492516801,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 9252
    },
    {
      "epoch": 0.09253,
      "grad_norm": 0.7491102789906501,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 9253
    },
    {
      "epoch": 0.09254,
      "grad_norm": 0.6784653406918104,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 9254
    },
    {
      "epoch": 0.09255,
      "grad_norm": 0.5654398621536241,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 9255
    },
    {
      "epoch": 0.09256,
      "grad_norm": 0.6643608382759081,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 9256
    },
    {
      "epoch": 0.09257,
      "grad_norm": 0.7193700387398151,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9257
    },
    {
      "epoch": 0.09258,
      "grad_norm": 0.8759808055403874,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 9258
    },
    {
      "epoch": 0.09259,
      "grad_norm": 1.2012930245710616,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 9259
    },
    {
      "epoch": 0.0926,
      "grad_norm": 0.7418114937003976,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 9260
    },
    {
      "epoch": 0.09261,
      "grad_norm": 0.5474250675303859,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 9261
    },
    {
      "epoch": 0.09262,
      "grad_norm": 0.6829621284506816,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 9262
    },
    {
      "epoch": 0.09263,
      "grad_norm": 0.7768878158809396,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 9263
    },
    {
      "epoch": 0.09264,
      "grad_norm": 0.9996107855545961,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 9264
    },
    {
      "epoch": 0.09265,
      "grad_norm": 1.163827375444901,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 9265
    },
    {
      "epoch": 0.09266,
      "grad_norm": 0.7349870746113485,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 9266
    },
    {
      "epoch": 0.09267,
      "grad_norm": 0.6853508114179676,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 9267
    },
    {
      "epoch": 0.09268,
      "grad_norm": 0.6333065813900194,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 9268
    },
    {
      "epoch": 0.09269,
      "grad_norm": 0.7044393984561579,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 9269
    },
    {
      "epoch": 0.0927,
      "grad_norm": 0.7829797104270343,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 9270
    },
    {
      "epoch": 0.09271,
      "grad_norm": 0.8257575052405663,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 9271
    },
    {
      "epoch": 0.09272,
      "grad_norm": 0.9439694552484319,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 9272
    },
    {
      "epoch": 0.09273,
      "grad_norm": 1.0602176422418417,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 9273
    },
    {
      "epoch": 0.09274,
      "grad_norm": 0.9858527050122837,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 9274
    },
    {
      "epoch": 0.09275,
      "grad_norm": 0.9316290417225153,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 9275
    },
    {
      "epoch": 0.09276,
      "grad_norm": 0.9228744715959912,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 9276
    },
    {
      "epoch": 0.09277,
      "grad_norm": 0.97661764276748,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 9277
    },
    {
      "epoch": 0.09278,
      "grad_norm": 0.9667223874049936,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 9278
    },
    {
      "epoch": 0.09279,
      "grad_norm": 1.0509045802580743,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 9279
    },
    {
      "epoch": 0.0928,
      "grad_norm": 1.199654598508013,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 9280
    },
    {
      "epoch": 0.09281,
      "grad_norm": 0.8190040066327821,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 9281
    },
    {
      "epoch": 0.09282,
      "grad_norm": 0.8005381258667182,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 9282
    },
    {
      "epoch": 0.09283,
      "grad_norm": 0.9731528357234364,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 9283
    },
    {
      "epoch": 0.09284,
      "grad_norm": 1.0699122649133415,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 9284
    },
    {
      "epoch": 0.09285,
      "grad_norm": 0.9736646539104704,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 9285
    },
    {
      "epoch": 0.09286,
      "grad_norm": 0.7534181301780545,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 9286
    },
    {
      "epoch": 0.09287,
      "grad_norm": 0.7039247323481843,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 9287
    },
    {
      "epoch": 0.09288,
      "grad_norm": 0.6153847411143628,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 9288
    },
    {
      "epoch": 0.09289,
      "grad_norm": 0.7439491494868004,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 9289
    },
    {
      "epoch": 0.0929,
      "grad_norm": 1.0211178389308402,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 9290
    },
    {
      "epoch": 0.09291,
      "grad_norm": 1.2912396043684187,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 9291
    },
    {
      "epoch": 0.09292,
      "grad_norm": 1.0329590152181434,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 9292
    },
    {
      "epoch": 0.09293,
      "grad_norm": 0.925671296507913,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 9293
    },
    {
      "epoch": 0.09294,
      "grad_norm": 0.8753174570663704,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 9294
    },
    {
      "epoch": 0.09295,
      "grad_norm": 0.7750429431472656,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9295
    },
    {
      "epoch": 0.09296,
      "grad_norm": 0.8498716958126399,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 9296
    },
    {
      "epoch": 0.09297,
      "grad_norm": 0.9547075919931944,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 9297
    },
    {
      "epoch": 0.09298,
      "grad_norm": 1.0547931478455412,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 9298
    },
    {
      "epoch": 0.09299,
      "grad_norm": 0.9874863882567291,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 9299
    },
    {
      "epoch": 0.093,
      "grad_norm": 1.0320217787422357,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 9300
    },
    {
      "epoch": 0.09301,
      "grad_norm": 0.868256523126305,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 9301
    },
    {
      "epoch": 0.09302,
      "grad_norm": 0.8911189359509369,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 9302
    },
    {
      "epoch": 0.09303,
      "grad_norm": 1.1184167860096568,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 9303
    },
    {
      "epoch": 0.09304,
      "grad_norm": 1.0616222963872175,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 9304
    },
    {
      "epoch": 0.09305,
      "grad_norm": 1.0167967680320442,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 9305
    },
    {
      "epoch": 0.09306,
      "grad_norm": 0.9100621741165005,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 9306
    },
    {
      "epoch": 0.09307,
      "grad_norm": 0.7277198010566454,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 9307
    },
    {
      "epoch": 0.09308,
      "grad_norm": 0.7642184186042998,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 9308
    },
    {
      "epoch": 0.09309,
      "grad_norm": 0.7898375897779486,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 9309
    },
    {
      "epoch": 0.0931,
      "grad_norm": 0.8313627098153431,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 9310
    },
    {
      "epoch": 0.09311,
      "grad_norm": 0.9180361447599314,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 9311
    },
    {
      "epoch": 0.09312,
      "grad_norm": 1.0722183150254507,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 9312
    },
    {
      "epoch": 0.09313,
      "grad_norm": 1.0053080965881522,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 9313
    },
    {
      "epoch": 0.09314,
      "grad_norm": 1.0429150067918742,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 9314
    },
    {
      "epoch": 0.09315,
      "grad_norm": 1.0275188761589462,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 9315
    },
    {
      "epoch": 0.09316,
      "grad_norm": 0.9852252098429989,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 9316
    },
    {
      "epoch": 0.09317,
      "grad_norm": 1.0943356019238426,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 9317
    },
    {
      "epoch": 0.09318,
      "grad_norm": 0.8892700832028245,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 9318
    },
    {
      "epoch": 0.09319,
      "grad_norm": 0.8440332983350739,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 9319
    },
    {
      "epoch": 0.0932,
      "grad_norm": 0.9136370265528279,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 9320
    },
    {
      "epoch": 0.09321,
      "grad_norm": 1.0652363081449419,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 9321
    },
    {
      "epoch": 0.09322,
      "grad_norm": 0.9657802025171371,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9322
    },
    {
      "epoch": 0.09323,
      "grad_norm": 0.9541349646760839,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 9323
    },
    {
      "epoch": 0.09324,
      "grad_norm": 0.940690991603989,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 9324
    },
    {
      "epoch": 0.09325,
      "grad_norm": 0.9773641494477271,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 9325
    },
    {
      "epoch": 0.09326,
      "grad_norm": 1.1123850382527165,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 9326
    },
    {
      "epoch": 0.09327,
      "grad_norm": 0.8756313815561116,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9327
    },
    {
      "epoch": 0.09328,
      "grad_norm": 0.7389172317899884,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 9328
    },
    {
      "epoch": 0.09329,
      "grad_norm": 0.6983046686762022,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 9329
    },
    {
      "epoch": 0.0933,
      "grad_norm": 0.6752597877286427,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 9330
    },
    {
      "epoch": 0.09331,
      "grad_norm": 0.8385408030611338,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 9331
    },
    {
      "epoch": 0.09332,
      "grad_norm": 0.92425769636424,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 9332
    },
    {
      "epoch": 0.09333,
      "grad_norm": 1.1149373386342603,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 9333
    },
    {
      "epoch": 0.09334,
      "grad_norm": 0.9061499154659839,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 9334
    },
    {
      "epoch": 0.09335,
      "grad_norm": 0.7814181455603038,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 9335
    },
    {
      "epoch": 0.09336,
      "grad_norm": 0.6185969547295879,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 9336
    },
    {
      "epoch": 0.09337,
      "grad_norm": 0.570098829353087,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 9337
    },
    {
      "epoch": 0.09338,
      "grad_norm": 0.6307398718763437,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9338
    },
    {
      "epoch": 0.09339,
      "grad_norm": 0.5876765823170635,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 9339
    },
    {
      "epoch": 0.0934,
      "grad_norm": 0.6057455120022778,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 9340
    },
    {
      "epoch": 0.09341,
      "grad_norm": 0.7819289757235965,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 9341
    },
    {
      "epoch": 0.09342,
      "grad_norm": 1.0525239903626875,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 9342
    },
    {
      "epoch": 0.09343,
      "grad_norm": 0.922797767764872,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 9343
    },
    {
      "epoch": 0.09344,
      "grad_norm": 0.7651325270309342,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 9344
    },
    {
      "epoch": 0.09345,
      "grad_norm": 0.8307201489790069,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 9345
    },
    {
      "epoch": 0.09346,
      "grad_norm": 0.9268559341614948,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 9346
    },
    {
      "epoch": 0.09347,
      "grad_norm": 0.8843815142269438,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9347
    },
    {
      "epoch": 0.09348,
      "grad_norm": 0.7511422034725228,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 9348
    },
    {
      "epoch": 0.09349,
      "grad_norm": 0.7681336246456821,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9349
    },
    {
      "epoch": 0.0935,
      "grad_norm": 0.8391732024120159,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 9350
    },
    {
      "epoch": 0.09351,
      "grad_norm": 0.8572535723613135,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 9351
    },
    {
      "epoch": 0.09352,
      "grad_norm": 1.1345129108011323,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 9352
    },
    {
      "epoch": 0.09353,
      "grad_norm": 1.188322335200505,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 9353
    },
    {
      "epoch": 0.09354,
      "grad_norm": 0.7871694631960734,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 9354
    },
    {
      "epoch": 0.09355,
      "grad_norm": 0.7942202356512671,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 9355
    },
    {
      "epoch": 0.09356,
      "grad_norm": 0.8720300671130099,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 9356
    },
    {
      "epoch": 0.09357,
      "grad_norm": 0.9400770356953755,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9357
    },
    {
      "epoch": 0.09358,
      "grad_norm": 0.872908592359249,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 9358
    },
    {
      "epoch": 0.09359,
      "grad_norm": 0.9729065406434361,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 9359
    },
    {
      "epoch": 0.0936,
      "grad_norm": 1.4002450511532056,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 9360
    },
    {
      "epoch": 0.09361,
      "grad_norm": 0.8007667671323881,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 9361
    },
    {
      "epoch": 0.09362,
      "grad_norm": 0.8253755648015588,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 9362
    },
    {
      "epoch": 0.09363,
      "grad_norm": 0.8603986662115061,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 9363
    },
    {
      "epoch": 0.09364,
      "grad_norm": 0.8888437458688937,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 9364
    },
    {
      "epoch": 0.09365,
      "grad_norm": 1.062809120441484,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 9365
    },
    {
      "epoch": 0.09366,
      "grad_norm": 0.9302023322409313,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 9366
    },
    {
      "epoch": 0.09367,
      "grad_norm": 1.1396009232198598,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 9367
    },
    {
      "epoch": 0.09368,
      "grad_norm": 0.9961165393209213,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9368
    },
    {
      "epoch": 0.09369,
      "grad_norm": 0.9522147427617241,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 9369
    },
    {
      "epoch": 0.0937,
      "grad_norm": 1.0784020973346404,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 9370
    },
    {
      "epoch": 0.09371,
      "grad_norm": 1.1071187911940317,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 9371
    },
    {
      "epoch": 0.09372,
      "grad_norm": 0.94833464425284,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 9372
    },
    {
      "epoch": 0.09373,
      "grad_norm": 0.9480880293079869,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 9373
    },
    {
      "epoch": 0.09374,
      "grad_norm": 0.994929884790511,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 9374
    },
    {
      "epoch": 0.09375,
      "grad_norm": 0.988811266369445,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 9375
    },
    {
      "epoch": 0.09376,
      "grad_norm": 0.9324886958686643,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 9376
    },
    {
      "epoch": 0.09377,
      "grad_norm": 0.9548553533271001,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 9377
    },
    {
      "epoch": 0.09378,
      "grad_norm": 0.9562248687261611,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 9378
    },
    {
      "epoch": 0.09379,
      "grad_norm": 1.0222988140200817,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 9379
    },
    {
      "epoch": 0.0938,
      "grad_norm": 0.9124887754159823,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 9380
    },
    {
      "epoch": 0.09381,
      "grad_norm": 1.037020146160175,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 9381
    },
    {
      "epoch": 0.09382,
      "grad_norm": 0.9546631751003297,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 9382
    },
    {
      "epoch": 0.09383,
      "grad_norm": 1.040184676131133,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 9383
    },
    {
      "epoch": 0.09384,
      "grad_norm": 1.029833944161637,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 9384
    },
    {
      "epoch": 0.09385,
      "grad_norm": 0.899541831452752,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 9385
    },
    {
      "epoch": 0.09386,
      "grad_norm": 0.8629312944127356,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 9386
    },
    {
      "epoch": 0.09387,
      "grad_norm": 1.0409426461274158,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 9387
    },
    {
      "epoch": 0.09388,
      "grad_norm": 1.006589469006825,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 9388
    },
    {
      "epoch": 0.09389,
      "grad_norm": 1.1077114456174915,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 9389
    },
    {
      "epoch": 0.0939,
      "grad_norm": 1.011899263145205,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 9390
    },
    {
      "epoch": 0.09391,
      "grad_norm": 0.8545868658356651,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 9391
    },
    {
      "epoch": 0.09392,
      "grad_norm": 0.7530450948177452,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 9392
    },
    {
      "epoch": 0.09393,
      "grad_norm": 0.8389311586343143,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 9393
    },
    {
      "epoch": 0.09394,
      "grad_norm": 0.8905715644273344,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9394
    },
    {
      "epoch": 0.09395,
      "grad_norm": 0.9931804256938302,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 9395
    },
    {
      "epoch": 0.09396,
      "grad_norm": 1.1886533323261048,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 9396
    },
    {
      "epoch": 0.09397,
      "grad_norm": 0.8490709939667962,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 9397
    },
    {
      "epoch": 0.09398,
      "grad_norm": 0.9054704841108464,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 9398
    },
    {
      "epoch": 0.09399,
      "grad_norm": 0.8063846844294164,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 9399
    },
    {
      "epoch": 0.094,
      "grad_norm": 0.69451549125735,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 9400
    },
    {
      "epoch": 0.09401,
      "grad_norm": 0.6661444422351755,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 9401
    },
    {
      "epoch": 0.09402,
      "grad_norm": 0.7717530463401683,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 9402
    },
    {
      "epoch": 0.09403,
      "grad_norm": 0.9162649629518274,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 9403
    },
    {
      "epoch": 0.09404,
      "grad_norm": 1.0198930452293649,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 9404
    },
    {
      "epoch": 0.09405,
      "grad_norm": 0.9665735513307437,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 9405
    },
    {
      "epoch": 0.09406,
      "grad_norm": 0.9242295093220471,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 9406
    },
    {
      "epoch": 0.09407,
      "grad_norm": 0.8899999098808963,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9407
    },
    {
      "epoch": 0.09408,
      "grad_norm": 0.7998331174379482,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 9408
    },
    {
      "epoch": 0.09409,
      "grad_norm": 0.7813089937027473,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 9409
    },
    {
      "epoch": 0.0941,
      "grad_norm": 0.8667120250947528,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 9410
    },
    {
      "epoch": 0.09411,
      "grad_norm": 0.9269402848773428,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 9411
    },
    {
      "epoch": 0.09412,
      "grad_norm": 0.9835401553473229,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 9412
    },
    {
      "epoch": 0.09413,
      "grad_norm": 0.9172083233778977,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 9413
    },
    {
      "epoch": 0.09414,
      "grad_norm": 1.006134207979243,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 9414
    },
    {
      "epoch": 0.09415,
      "grad_norm": 1.132729901318295,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 9415
    },
    {
      "epoch": 0.09416,
      "grad_norm": 0.8404655613518899,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 9416
    },
    {
      "epoch": 0.09417,
      "grad_norm": 0.7194877963786109,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 9417
    },
    {
      "epoch": 0.09418,
      "grad_norm": 0.7730057932177794,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 9418
    },
    {
      "epoch": 0.09419,
      "grad_norm": 1.1292343134560585,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 9419
    },
    {
      "epoch": 0.0942,
      "grad_norm": 1.1774763095295602,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 9420
    },
    {
      "epoch": 0.09421,
      "grad_norm": 0.7857554926583439,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9421
    },
    {
      "epoch": 0.09422,
      "grad_norm": 0.7323939073029049,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 9422
    },
    {
      "epoch": 0.09423,
      "grad_norm": 0.7380592179333626,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 9423
    },
    {
      "epoch": 0.09424,
      "grad_norm": 0.811775438330701,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 9424
    },
    {
      "epoch": 0.09425,
      "grad_norm": 0.7513687697939088,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 9425
    },
    {
      "epoch": 0.09426,
      "grad_norm": 0.7219130870944211,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 9426
    },
    {
      "epoch": 0.09427,
      "grad_norm": 0.7513581740192289,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 9427
    },
    {
      "epoch": 0.09428,
      "grad_norm": 0.8363301820337448,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 9428
    },
    {
      "epoch": 0.09429,
      "grad_norm": 0.8654532174390273,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 9429
    },
    {
      "epoch": 0.0943,
      "grad_norm": 1.071277664835941,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9430
    },
    {
      "epoch": 0.09431,
      "grad_norm": 1.2678205157799658,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 9431
    },
    {
      "epoch": 0.09432,
      "grad_norm": 0.754680518969369,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 9432
    },
    {
      "epoch": 0.09433,
      "grad_norm": 0.8341523731988698,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 9433
    },
    {
      "epoch": 0.09434,
      "grad_norm": 0.9342603548990726,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 9434
    },
    {
      "epoch": 0.09435,
      "grad_norm": 0.9289091636637407,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 9435
    },
    {
      "epoch": 0.09436,
      "grad_norm": 1.107766105780556,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 9436
    },
    {
      "epoch": 0.09437,
      "grad_norm": 0.9085956133024019,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 9437
    },
    {
      "epoch": 0.09438,
      "grad_norm": 0.7903225243057821,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 9438
    },
    {
      "epoch": 0.09439,
      "grad_norm": 0.8100969592812339,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 9439
    },
    {
      "epoch": 0.0944,
      "grad_norm": 0.9111468297674603,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 9440
    },
    {
      "epoch": 0.09441,
      "grad_norm": 0.8331316815484991,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 9441
    },
    {
      "epoch": 0.09442,
      "grad_norm": 0.870678637091265,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 9442
    },
    {
      "epoch": 0.09443,
      "grad_norm": 0.9311560253694179,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 9443
    },
    {
      "epoch": 0.09444,
      "grad_norm": 1.0214536095553628,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 9444
    },
    {
      "epoch": 0.09445,
      "grad_norm": 1.1766483143623587,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 9445
    },
    {
      "epoch": 0.09446,
      "grad_norm": 0.9519894255979084,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 9446
    },
    {
      "epoch": 0.09447,
      "grad_norm": 0.9198353317350884,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 9447
    },
    {
      "epoch": 0.09448,
      "grad_norm": 1.05453779469519,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 9448
    },
    {
      "epoch": 0.09449,
      "grad_norm": 1.0266876101640614,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 9449
    },
    {
      "epoch": 0.0945,
      "grad_norm": 1.029243760537115,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 9450
    },
    {
      "epoch": 0.09451,
      "grad_norm": 1.0088314651093355,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 9451
    },
    {
      "epoch": 0.09452,
      "grad_norm": 0.8598300778559472,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 9452
    },
    {
      "epoch": 0.09453,
      "grad_norm": 0.8624461571472201,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 9453
    },
    {
      "epoch": 0.09454,
      "grad_norm": 1.0851623978196565,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 9454
    },
    {
      "epoch": 0.09455,
      "grad_norm": 1.1706846745804926,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 9455
    },
    {
      "epoch": 0.09456,
      "grad_norm": 0.720398550854311,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 9456
    },
    {
      "epoch": 0.09457,
      "grad_norm": 0.6428844036374872,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 9457
    },
    {
      "epoch": 0.09458,
      "grad_norm": 0.6854122980561818,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 9458
    },
    {
      "epoch": 0.09459,
      "grad_norm": 0.7557470276669376,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 9459
    },
    {
      "epoch": 0.0946,
      "grad_norm": 0.9400745909919421,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 9460
    },
    {
      "epoch": 0.09461,
      "grad_norm": 0.8272490781668856,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 9461
    },
    {
      "epoch": 0.09462,
      "grad_norm": 0.7264239631357388,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 9462
    },
    {
      "epoch": 0.09463,
      "grad_norm": 0.6420461092891769,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 9463
    },
    {
      "epoch": 0.09464,
      "grad_norm": 0.6005295130137699,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 9464
    },
    {
      "epoch": 0.09465,
      "grad_norm": 0.7261508186379931,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 9465
    },
    {
      "epoch": 0.09466,
      "grad_norm": 0.7958128917859779,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 9466
    },
    {
      "epoch": 0.09467,
      "grad_norm": 0.7585869693775182,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 9467
    },
    {
      "epoch": 0.09468,
      "grad_norm": 0.7806606873704798,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 9468
    },
    {
      "epoch": 0.09469,
      "grad_norm": 0.9838410664741073,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 9469
    },
    {
      "epoch": 0.0947,
      "grad_norm": 1.1761299535165681,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 9470
    },
    {
      "epoch": 0.09471,
      "grad_norm": 1.0889544845009487,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 9471
    },
    {
      "epoch": 0.09472,
      "grad_norm": 1.0307683311785911,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 9472
    },
    {
      "epoch": 0.09473,
      "grad_norm": 0.9564223700183074,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 9473
    },
    {
      "epoch": 0.09474,
      "grad_norm": 0.9376809113489067,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 9474
    },
    {
      "epoch": 0.09475,
      "grad_norm": 0.9618524959475816,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9475
    },
    {
      "epoch": 0.09476,
      "grad_norm": 0.9554715312019628,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 9476
    },
    {
      "epoch": 0.09477,
      "grad_norm": 1.000035475540716,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 9477
    },
    {
      "epoch": 0.09478,
      "grad_norm": 1.0839210902354595,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 9478
    },
    {
      "epoch": 0.09479,
      "grad_norm": 0.869487169656702,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 9479
    },
    {
      "epoch": 0.0948,
      "grad_norm": 0.7926358942114886,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 9480
    },
    {
      "epoch": 0.09481,
      "grad_norm": 0.6884497755914634,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 9481
    },
    {
      "epoch": 0.09482,
      "grad_norm": 0.6845896021916732,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 9482
    },
    {
      "epoch": 0.09483,
      "grad_norm": 0.6566997735021866,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 9483
    },
    {
      "epoch": 0.09484,
      "grad_norm": 0.616530773620674,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 9484
    },
    {
      "epoch": 0.09485,
      "grad_norm": 0.7642856486603846,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 9485
    },
    {
      "epoch": 0.09486,
      "grad_norm": 0.8665981874168233,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 9486
    },
    {
      "epoch": 0.09487,
      "grad_norm": 1.0909398291701577,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 9487
    },
    {
      "epoch": 0.09488,
      "grad_norm": 1.1491216476616155,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 9488
    },
    {
      "epoch": 0.09489,
      "grad_norm": 0.8354603918248583,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9489
    },
    {
      "epoch": 0.0949,
      "grad_norm": 0.6895784339169947,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 9490
    },
    {
      "epoch": 0.09491,
      "grad_norm": 0.711524009729495,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 9491
    },
    {
      "epoch": 0.09492,
      "grad_norm": 0.8394489207958683,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 9492
    },
    {
      "epoch": 0.09493,
      "grad_norm": 1.2414418185211027,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 9493
    },
    {
      "epoch": 0.09494,
      "grad_norm": 0.8710054135901388,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 9494
    },
    {
      "epoch": 0.09495,
      "grad_norm": 0.8645492616541143,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 9495
    },
    {
      "epoch": 0.09496,
      "grad_norm": 0.83427100248925,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 9496
    },
    {
      "epoch": 0.09497,
      "grad_norm": 0.7271594302410049,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 9497
    },
    {
      "epoch": 0.09498,
      "grad_norm": 0.641946567264643,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 9498
    },
    {
      "epoch": 0.09499,
      "grad_norm": 0.6170199490496553,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 9499
    },
    {
      "epoch": 0.095,
      "grad_norm": 0.6200424975500217,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9500
    },
    {
      "epoch": 0.09501,
      "grad_norm": 0.809057988961176,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 9501
    },
    {
      "epoch": 0.09502,
      "grad_norm": 1.1020241879146306,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 9502
    },
    {
      "epoch": 0.09503,
      "grad_norm": 1.030306126051077,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 9503
    },
    {
      "epoch": 0.09504,
      "grad_norm": 0.9735652853838207,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 9504
    },
    {
      "epoch": 0.09505,
      "grad_norm": 0.9236443544745729,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9505
    },
    {
      "epoch": 0.09506,
      "grad_norm": 0.8592569358203569,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 9506
    },
    {
      "epoch": 0.09507,
      "grad_norm": 0.9643353093501955,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 9507
    },
    {
      "epoch": 0.09508,
      "grad_norm": 1.0137276574092238,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 9508
    },
    {
      "epoch": 0.09509,
      "grad_norm": 1.1660412011916976,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 9509
    },
    {
      "epoch": 0.0951,
      "grad_norm": 1.082728289928524,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 9510
    },
    {
      "epoch": 0.09511,
      "grad_norm": 0.9624008478957283,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 9511
    },
    {
      "epoch": 0.09512,
      "grad_norm": 1.0923795540142767,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 9512
    },
    {
      "epoch": 0.09513,
      "grad_norm": 1.0328382351919922,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 9513
    },
    {
      "epoch": 0.09514,
      "grad_norm": 1.0089078992511966,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 9514
    },
    {
      "epoch": 0.09515,
      "grad_norm": 0.9704873672647005,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 9515
    },
    {
      "epoch": 0.09516,
      "grad_norm": 0.8985075408487389,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 9516
    },
    {
      "epoch": 0.09517,
      "grad_norm": 0.9130742158343249,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 9517
    },
    {
      "epoch": 0.09518,
      "grad_norm": 0.9244616783687439,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 9518
    },
    {
      "epoch": 0.09519,
      "grad_norm": 0.9668574431326041,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 9519
    },
    {
      "epoch": 0.0952,
      "grad_norm": 0.9637024791097162,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 9520
    },
    {
      "epoch": 0.09521,
      "grad_norm": 1.016108766606305,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 9521
    },
    {
      "epoch": 0.09522,
      "grad_norm": 1.0754104899740695,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 9522
    },
    {
      "epoch": 0.09523,
      "grad_norm": 0.9460439865375799,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 9523
    },
    {
      "epoch": 0.09524,
      "grad_norm": 0.8972736244759736,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 9524
    },
    {
      "epoch": 0.09525,
      "grad_norm": 0.8707423847198635,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 9525
    },
    {
      "epoch": 0.09526,
      "grad_norm": 0.9298574511865992,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 9526
    },
    {
      "epoch": 0.09527,
      "grad_norm": 0.873097150552636,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9527
    },
    {
      "epoch": 0.09528,
      "grad_norm": 0.861692752888038,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 9528
    },
    {
      "epoch": 0.09529,
      "grad_norm": 0.8130110024800541,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 9529
    },
    {
      "epoch": 0.0953,
      "grad_norm": 0.7905443102217827,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 9530
    },
    {
      "epoch": 0.09531,
      "grad_norm": 0.8291019771418112,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 9531
    },
    {
      "epoch": 0.09532,
      "grad_norm": 0.8459154353642524,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 9532
    },
    {
      "epoch": 0.09533,
      "grad_norm": 0.9922895216920463,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 9533
    },
    {
      "epoch": 0.09534,
      "grad_norm": 1.185301105499309,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 9534
    },
    {
      "epoch": 0.09535,
      "grad_norm": 1.0236593208473932,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 9535
    },
    {
      "epoch": 0.09536,
      "grad_norm": 1.1070204390625342,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 9536
    },
    {
      "epoch": 0.09537,
      "grad_norm": 0.9574983262523417,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 9537
    },
    {
      "epoch": 0.09538,
      "grad_norm": 0.8493144641874563,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 9538
    },
    {
      "epoch": 0.09539,
      "grad_norm": 0.835531151551209,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 9539
    },
    {
      "epoch": 0.0954,
      "grad_norm": 0.8283289116606958,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 9540
    },
    {
      "epoch": 0.09541,
      "grad_norm": 0.9341466910775755,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 9541
    },
    {
      "epoch": 0.09542,
      "grad_norm": 1.0314622632534094,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 9542
    },
    {
      "epoch": 0.09543,
      "grad_norm": 1.0142586101903608,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 9543
    },
    {
      "epoch": 0.09544,
      "grad_norm": 1.4582749640876167,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 9544
    },
    {
      "epoch": 0.09545,
      "grad_norm": 0.6203000789802795,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 9545
    },
    {
      "epoch": 0.09546,
      "grad_norm": 0.7933224015182839,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 9546
    },
    {
      "epoch": 0.09547,
      "grad_norm": 0.7738375177165676,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9547
    },
    {
      "epoch": 0.09548,
      "grad_norm": 0.8381207027995874,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 9548
    },
    {
      "epoch": 0.09549,
      "grad_norm": 0.8454264400374799,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 9549
    },
    {
      "epoch": 0.0955,
      "grad_norm": 0.8076905957792103,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 9550
    },
    {
      "epoch": 0.09551,
      "grad_norm": 0.8632802824577473,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 9551
    },
    {
      "epoch": 0.09552,
      "grad_norm": 1.0768644429113419,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 9552
    },
    {
      "epoch": 0.09553,
      "grad_norm": 1.1900364276456132,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 9553
    },
    {
      "epoch": 0.09554,
      "grad_norm": 0.6910437093615704,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 9554
    },
    {
      "epoch": 0.09555,
      "grad_norm": 0.677559408017465,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 9555
    },
    {
      "epoch": 0.09556,
      "grad_norm": 0.8863789977775014,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 9556
    },
    {
      "epoch": 0.09557,
      "grad_norm": 1.2130858830229994,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 9557
    },
    {
      "epoch": 0.09558,
      "grad_norm": 0.8078845755680898,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 9558
    },
    {
      "epoch": 0.09559,
      "grad_norm": 0.6888474933216685,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 9559
    },
    {
      "epoch": 0.0956,
      "grad_norm": 0.7444419985119464,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 9560
    },
    {
      "epoch": 0.09561,
      "grad_norm": 0.7331440318600505,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 9561
    },
    {
      "epoch": 0.09562,
      "grad_norm": 0.6745064106478187,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 9562
    },
    {
      "epoch": 0.09563,
      "grad_norm": 0.7223110614602103,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 9563
    },
    {
      "epoch": 0.09564,
      "grad_norm": 0.7954354073035838,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 9564
    },
    {
      "epoch": 0.09565,
      "grad_norm": 0.9518848575929585,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 9565
    },
    {
      "epoch": 0.09566,
      "grad_norm": 1.1360554053001062,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 9566
    },
    {
      "epoch": 0.09567,
      "grad_norm": 1.0190304984027616,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 9567
    },
    {
      "epoch": 0.09568,
      "grad_norm": 1.1111115745846918,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 9568
    },
    {
      "epoch": 0.09569,
      "grad_norm": 1.060656585628691,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 9569
    },
    {
      "epoch": 0.0957,
      "grad_norm": 0.9102915228109848,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 9570
    },
    {
      "epoch": 0.09571,
      "grad_norm": 1.018964046699271,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 9571
    },
    {
      "epoch": 0.09572,
      "grad_norm": 1.2527068883296468,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 9572
    },
    {
      "epoch": 0.09573,
      "grad_norm": 0.9851106487089002,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 9573
    },
    {
      "epoch": 0.09574,
      "grad_norm": 1.0626664552153582,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 9574
    },
    {
      "epoch": 0.09575,
      "grad_norm": 1.1535018572622826,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 9575
    },
    {
      "epoch": 0.09576,
      "grad_norm": 0.866406401028726,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 9576
    },
    {
      "epoch": 0.09577,
      "grad_norm": 0.7598448539401876,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 9577
    },
    {
      "epoch": 0.09578,
      "grad_norm": 0.76846069859649,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 9578
    },
    {
      "epoch": 0.09579,
      "grad_norm": 0.8043603561549131,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 9579
    },
    {
      "epoch": 0.0958,
      "grad_norm": 0.8020399563357902,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 9580
    },
    {
      "epoch": 0.09581,
      "grad_norm": 0.7682141048439647,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 9581
    },
    {
      "epoch": 0.09582,
      "grad_norm": 0.8015394924731275,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 9582
    },
    {
      "epoch": 0.09583,
      "grad_norm": 0.8226749457094206,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 9583
    },
    {
      "epoch": 0.09584,
      "grad_norm": 0.94933199587187,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 9584
    },
    {
      "epoch": 0.09585,
      "grad_norm": 0.9918679044193051,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 9585
    },
    {
      "epoch": 0.09586,
      "grad_norm": 0.9705822187192004,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 9586
    },
    {
      "epoch": 0.09587,
      "grad_norm": 0.9451328209074761,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 9587
    },
    {
      "epoch": 0.09588,
      "grad_norm": 1.2495169152311112,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 9588
    },
    {
      "epoch": 0.09589,
      "grad_norm": 0.8554058248225164,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 9589
    },
    {
      "epoch": 0.0959,
      "grad_norm": 0.8694247907519619,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 9590
    },
    {
      "epoch": 0.09591,
      "grad_norm": 0.8764253050367747,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 9591
    },
    {
      "epoch": 0.09592,
      "grad_norm": 0.928129824027457,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 9592
    },
    {
      "epoch": 0.09593,
      "grad_norm": 0.9505672638184878,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 9593
    },
    {
      "epoch": 0.09594,
      "grad_norm": 1.0714724754192935,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 9594
    },
    {
      "epoch": 0.09595,
      "grad_norm": 0.9819378603136486,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 9595
    },
    {
      "epoch": 0.09596,
      "grad_norm": 1.026865992142624,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 9596
    },
    {
      "epoch": 0.09597,
      "grad_norm": 0.973648595397517,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 9597
    },
    {
      "epoch": 0.09598,
      "grad_norm": 0.9850517417503974,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 9598
    },
    {
      "epoch": 0.09599,
      "grad_norm": 0.9928584309344902,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9599
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.9952109010549817,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 9600
    },
    {
      "epoch": 0.09601,
      "grad_norm": 0.9715688246997389,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9601
    },
    {
      "epoch": 0.09602,
      "grad_norm": 0.896038866298406,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 9602
    },
    {
      "epoch": 0.09603,
      "grad_norm": 0.9394621805775981,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 9603
    },
    {
      "epoch": 0.09604,
      "grad_norm": 0.8832102278330413,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 9604
    },
    {
      "epoch": 0.09605,
      "grad_norm": 0.8768838219761711,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 9605
    },
    {
      "epoch": 0.09606,
      "grad_norm": 0.8777392831472427,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 9606
    },
    {
      "epoch": 0.09607,
      "grad_norm": 1.076257004258557,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 9607
    },
    {
      "epoch": 0.09608,
      "grad_norm": 0.9490087215698528,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 9608
    },
    {
      "epoch": 0.09609,
      "grad_norm": 0.8957046130890897,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 9609
    },
    {
      "epoch": 0.0961,
      "grad_norm": 0.7984430379991309,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9610
    },
    {
      "epoch": 0.09611,
      "grad_norm": 0.7499563857935081,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 9611
    },
    {
      "epoch": 0.09612,
      "grad_norm": 0.7348626128667639,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 9612
    },
    {
      "epoch": 0.09613,
      "grad_norm": 0.7649035088697457,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 9613
    },
    {
      "epoch": 0.09614,
      "grad_norm": 0.8970754465663441,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 9614
    },
    {
      "epoch": 0.09615,
      "grad_norm": 1.0353388858633117,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9615
    },
    {
      "epoch": 0.09616,
      "grad_norm": 1.0422928251395278,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 9616
    },
    {
      "epoch": 0.09617,
      "grad_norm": 0.9330622686075548,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 9617
    },
    {
      "epoch": 0.09618,
      "grad_norm": 0.8079910426137322,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 9618
    },
    {
      "epoch": 0.09619,
      "grad_norm": 0.7658297024476686,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9619
    },
    {
      "epoch": 0.0962,
      "grad_norm": 0.7549136355891398,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 9620
    },
    {
      "epoch": 0.09621,
      "grad_norm": 0.7392852138332249,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 9621
    },
    {
      "epoch": 0.09622,
      "grad_norm": 0.8832584494387006,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 9622
    },
    {
      "epoch": 0.09623,
      "grad_norm": 1.0636469145202627,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 9623
    },
    {
      "epoch": 0.09624,
      "grad_norm": 1.0064354117819914,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 9624
    },
    {
      "epoch": 0.09625,
      "grad_norm": 0.9371314262078005,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 9625
    },
    {
      "epoch": 0.09626,
      "grad_norm": 0.8253170716175934,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 9626
    },
    {
      "epoch": 0.09627,
      "grad_norm": 0.7030524329715849,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 9627
    },
    {
      "epoch": 0.09628,
      "grad_norm": 0.7250410432973622,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 9628
    },
    {
      "epoch": 0.09629,
      "grad_norm": 0.7346067220792749,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 9629
    },
    {
      "epoch": 0.0963,
      "grad_norm": 0.9046057127262703,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 9630
    },
    {
      "epoch": 0.09631,
      "grad_norm": 0.894854230452782,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 9631
    },
    {
      "epoch": 0.09632,
      "grad_norm": 0.9480273147630979,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 9632
    },
    {
      "epoch": 0.09633,
      "grad_norm": 0.9786150037379322,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 9633
    },
    {
      "epoch": 0.09634,
      "grad_norm": 1.0282461128507772,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 9634
    },
    {
      "epoch": 0.09635,
      "grad_norm": 1.2837836327015333,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 9635
    },
    {
      "epoch": 0.09636,
      "grad_norm": 1.0371256794904447,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 9636
    },
    {
      "epoch": 0.09637,
      "grad_norm": 0.9109444824213261,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 9637
    },
    {
      "epoch": 0.09638,
      "grad_norm": 0.8349568525937817,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 9638
    },
    {
      "epoch": 0.09639,
      "grad_norm": 0.8952363870351221,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 9639
    },
    {
      "epoch": 0.0964,
      "grad_norm": 0.8817057896870121,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 9640
    },
    {
      "epoch": 0.09641,
      "grad_norm": 1.0165123777415066,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 9641
    },
    {
      "epoch": 0.09642,
      "grad_norm": 1.2088791393181662,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 9642
    },
    {
      "epoch": 0.09643,
      "grad_norm": 0.8298242740488613,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 9643
    },
    {
      "epoch": 0.09644,
      "grad_norm": 0.7360761912273628,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 9644
    },
    {
      "epoch": 0.09645,
      "grad_norm": 0.7427508177687894,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 9645
    },
    {
      "epoch": 0.09646,
      "grad_norm": 0.8169170909806851,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 9646
    },
    {
      "epoch": 0.09647,
      "grad_norm": 0.873096623162937,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 9647
    },
    {
      "epoch": 0.09648,
      "grad_norm": 0.9301514895225974,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 9648
    },
    {
      "epoch": 0.09649,
      "grad_norm": 1.0261303129344521,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 9649
    },
    {
      "epoch": 0.0965,
      "grad_norm": 0.9858489832150986,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 9650
    },
    {
      "epoch": 0.09651,
      "grad_norm": 0.9207520107530078,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 9651
    },
    {
      "epoch": 0.09652,
      "grad_norm": 0.8934066228016666,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 9652
    },
    {
      "epoch": 0.09653,
      "grad_norm": 0.8604206377048786,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 9653
    },
    {
      "epoch": 0.09654,
      "grad_norm": 0.9663619779941607,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 9654
    },
    {
      "epoch": 0.09655,
      "grad_norm": 0.91146406489948,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 9655
    },
    {
      "epoch": 0.09656,
      "grad_norm": 0.8975020800835469,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 9656
    },
    {
      "epoch": 0.09657,
      "grad_norm": 1.0608421766317369,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 9657
    },
    {
      "epoch": 0.09658,
      "grad_norm": 0.9751782896772202,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 9658
    },
    {
      "epoch": 0.09659,
      "grad_norm": 0.8507331684637727,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 9659
    },
    {
      "epoch": 0.0966,
      "grad_norm": 0.7797760426559466,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 9660
    },
    {
      "epoch": 0.09661,
      "grad_norm": 0.7280346537301909,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 9661
    },
    {
      "epoch": 0.09662,
      "grad_norm": 0.7687521057717993,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 9662
    },
    {
      "epoch": 0.09663,
      "grad_norm": 0.8765165791530211,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 9663
    },
    {
      "epoch": 0.09664,
      "grad_norm": 1.213269447787462,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 9664
    },
    {
      "epoch": 0.09665,
      "grad_norm": 0.9511980015499266,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 9665
    },
    {
      "epoch": 0.09666,
      "grad_norm": 0.8932643356990735,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 9666
    },
    {
      "epoch": 0.09667,
      "grad_norm": 1.1991501443292183,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 9667
    },
    {
      "epoch": 0.09668,
      "grad_norm": 0.9122924207412219,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 9668
    },
    {
      "epoch": 0.09669,
      "grad_norm": 0.8177666074816453,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 9669
    },
    {
      "epoch": 0.0967,
      "grad_norm": 0.8642756088338996,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 9670
    },
    {
      "epoch": 0.09671,
      "grad_norm": 0.9387142926253513,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 9671
    },
    {
      "epoch": 0.09672,
      "grad_norm": 0.9876148943907853,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 9672
    },
    {
      "epoch": 0.09673,
      "grad_norm": 1.183065157513859,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 9673
    },
    {
      "epoch": 0.09674,
      "grad_norm": 0.8981167917963628,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 9674
    },
    {
      "epoch": 0.09675,
      "grad_norm": 0.7908901876516276,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 9675
    },
    {
      "epoch": 0.09676,
      "grad_norm": 0.7836781101706645,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9676
    },
    {
      "epoch": 0.09677,
      "grad_norm": 0.9396215534952582,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 9677
    },
    {
      "epoch": 0.09678,
      "grad_norm": 1.097205685443494,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 9678
    },
    {
      "epoch": 0.09679,
      "grad_norm": 0.9990488663831774,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 9679
    },
    {
      "epoch": 0.0968,
      "grad_norm": 0.9171443826425181,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 9680
    },
    {
      "epoch": 0.09681,
      "grad_norm": 1.009511628313745,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 9681
    },
    {
      "epoch": 0.09682,
      "grad_norm": 1.2431943193297423,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 9682
    },
    {
      "epoch": 0.09683,
      "grad_norm": 0.9350989569045836,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 9683
    },
    {
      "epoch": 0.09684,
      "grad_norm": 0.9827551754050654,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 9684
    },
    {
      "epoch": 0.09685,
      "grad_norm": 1.0518093494350942,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 9685
    },
    {
      "epoch": 0.09686,
      "grad_norm": 1.1163299364855512,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 9686
    },
    {
      "epoch": 0.09687,
      "grad_norm": 0.8195919781179394,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 9687
    },
    {
      "epoch": 0.09688,
      "grad_norm": 0.8190869432151955,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 9688
    },
    {
      "epoch": 0.09689,
      "grad_norm": 0.916773205951842,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 9689
    },
    {
      "epoch": 0.0969,
      "grad_norm": 0.8066088249172502,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 9690
    },
    {
      "epoch": 0.09691,
      "grad_norm": 0.7401326149708389,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 9691
    },
    {
      "epoch": 0.09692,
      "grad_norm": 0.6424151800103749,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 9692
    },
    {
      "epoch": 0.09693,
      "grad_norm": 0.6299822425823625,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 9693
    },
    {
      "epoch": 0.09694,
      "grad_norm": 0.6362006844765989,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 9694
    },
    {
      "epoch": 0.09695,
      "grad_norm": 0.6132233963780619,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 9695
    },
    {
      "epoch": 0.09696,
      "grad_norm": 0.6660368458929525,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 9696
    },
    {
      "epoch": 0.09697,
      "grad_norm": 0.7544139461670122,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 9697
    },
    {
      "epoch": 0.09698,
      "grad_norm": 0.8497357368030164,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 9698
    },
    {
      "epoch": 0.09699,
      "grad_norm": 0.9145155762891021,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 9699
    },
    {
      "epoch": 0.097,
      "grad_norm": 0.8975185216089709,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 9700
    },
    {
      "epoch": 0.09701,
      "grad_norm": 0.9835477897845034,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 9701
    },
    {
      "epoch": 0.09702,
      "grad_norm": 1.2588913749038257,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 9702
    },
    {
      "epoch": 0.09703,
      "grad_norm": 1.231889122377567,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 9703
    },
    {
      "epoch": 0.09704,
      "grad_norm": 0.9026679969753263,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 9704
    },
    {
      "epoch": 0.09705,
      "grad_norm": 0.917096876663794,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 9705
    },
    {
      "epoch": 0.09706,
      "grad_norm": 1.1210979563105794,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 9706
    },
    {
      "epoch": 0.09707,
      "grad_norm": 1.046684770518191,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 9707
    },
    {
      "epoch": 0.09708,
      "grad_norm": 0.8586665214097383,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 9708
    },
    {
      "epoch": 0.09709,
      "grad_norm": 0.8399659702694784,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 9709
    },
    {
      "epoch": 0.0971,
      "grad_norm": 0.9061619367314612,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 9710
    },
    {
      "epoch": 0.09711,
      "grad_norm": 0.869420827619381,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 9711
    },
    {
      "epoch": 0.09712,
      "grad_norm": 0.9484632391522392,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 9712
    },
    {
      "epoch": 0.09713,
      "grad_norm": 0.9344386880847007,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 9713
    },
    {
      "epoch": 0.09714,
      "grad_norm": 0.8191433536044963,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 9714
    },
    {
      "epoch": 0.09715,
      "grad_norm": 0.9260497381898702,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 9715
    },
    {
      "epoch": 0.09716,
      "grad_norm": 1.2168835302778682,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 9716
    },
    {
      "epoch": 0.09717,
      "grad_norm": 1.1755192129415277,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 9717
    },
    {
      "epoch": 0.09718,
      "grad_norm": 0.8578207270174794,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9718
    },
    {
      "epoch": 0.09719,
      "grad_norm": 0.7982551490113904,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 9719
    },
    {
      "epoch": 0.0972,
      "grad_norm": 0.8181582924013971,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 9720
    },
    {
      "epoch": 0.09721,
      "grad_norm": 0.7803993304194405,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9721
    },
    {
      "epoch": 0.09722,
      "grad_norm": 0.7217245724741701,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 9722
    },
    {
      "epoch": 0.09723,
      "grad_norm": 0.8297726003767051,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 9723
    },
    {
      "epoch": 0.09724,
      "grad_norm": 0.8385216713500718,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 9724
    },
    {
      "epoch": 0.09725,
      "grad_norm": 0.8535894035344416,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 9725
    },
    {
      "epoch": 0.09726,
      "grad_norm": 0.9997627703805871,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 9726
    },
    {
      "epoch": 0.09727,
      "grad_norm": 1.2871027951432406,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 9727
    },
    {
      "epoch": 0.09728,
      "grad_norm": 0.9786561845767965,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 9728
    },
    {
      "epoch": 0.09729,
      "grad_norm": 1.056518217713997,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 9729
    },
    {
      "epoch": 0.0973,
      "grad_norm": 1.0746377651628496,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 9730
    },
    {
      "epoch": 0.09731,
      "grad_norm": 1.1074902816443135,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 9731
    },
    {
      "epoch": 0.09732,
      "grad_norm": 0.9021480346313361,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 9732
    },
    {
      "epoch": 0.09733,
      "grad_norm": 0.8698953887525058,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 9733
    },
    {
      "epoch": 0.09734,
      "grad_norm": 0.7657365794014724,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 9734
    },
    {
      "epoch": 0.09735,
      "grad_norm": 0.7136532065218357,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 9735
    },
    {
      "epoch": 0.09736,
      "grad_norm": 0.6717357230030461,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 9736
    },
    {
      "epoch": 0.09737,
      "grad_norm": 0.5638315775801472,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 9737
    },
    {
      "epoch": 0.09738,
      "grad_norm": 0.6110116723814848,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 9738
    },
    {
      "epoch": 0.09739,
      "grad_norm": 0.6086506139891856,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 9739
    },
    {
      "epoch": 0.0974,
      "grad_norm": 0.7935050152052187,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 9740
    },
    {
      "epoch": 0.09741,
      "grad_norm": 1.1123363599125007,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 9741
    },
    {
      "epoch": 0.09742,
      "grad_norm": 1.0030554366325666,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 9742
    },
    {
      "epoch": 0.09743,
      "grad_norm": 0.8267397016080698,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 9743
    },
    {
      "epoch": 0.09744,
      "grad_norm": 0.7717073469243041,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 9744
    },
    {
      "epoch": 0.09745,
      "grad_norm": 0.798472215693979,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 9745
    },
    {
      "epoch": 0.09746,
      "grad_norm": 0.9589191423737808,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 9746
    },
    {
      "epoch": 0.09747,
      "grad_norm": 1.116733109628546,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 9747
    },
    {
      "epoch": 0.09748,
      "grad_norm": 0.8375628533401334,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 9748
    },
    {
      "epoch": 0.09749,
      "grad_norm": 0.8553001862226853,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 9749
    },
    {
      "epoch": 0.0975,
      "grad_norm": 0.9088453607552779,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 9750
    },
    {
      "epoch": 0.09751,
      "grad_norm": 0.8569486483584623,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 9751
    },
    {
      "epoch": 0.09752,
      "grad_norm": 0.7380045417599184,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 9752
    },
    {
      "epoch": 0.09753,
      "grad_norm": 0.8851823127379947,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 9753
    },
    {
      "epoch": 0.09754,
      "grad_norm": 1.0358214518222049,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 9754
    },
    {
      "epoch": 0.09755,
      "grad_norm": 0.9439521181468254,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9755
    },
    {
      "epoch": 0.09756,
      "grad_norm": 0.8198513848838082,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9756
    },
    {
      "epoch": 0.09757,
      "grad_norm": 0.829860888973659,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 9757
    },
    {
      "epoch": 0.09758,
      "grad_norm": 0.9561785595312966,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 9758
    },
    {
      "epoch": 0.09759,
      "grad_norm": 1.2593905860698673,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 9759
    },
    {
      "epoch": 0.0976,
      "grad_norm": 1.046642575447866,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9760
    },
    {
      "epoch": 0.09761,
      "grad_norm": 1.207472022721697,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9761
    },
    {
      "epoch": 0.09762,
      "grad_norm": 0.9869421396722704,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 9762
    },
    {
      "epoch": 0.09763,
      "grad_norm": 1.0380173466462297,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 9763
    },
    {
      "epoch": 0.09764,
      "grad_norm": 1.1529962238971903,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 9764
    },
    {
      "epoch": 0.09765,
      "grad_norm": 0.8259926007525762,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 9765
    },
    {
      "epoch": 0.09766,
      "grad_norm": 0.8887761626705335,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 9766
    },
    {
      "epoch": 0.09767,
      "grad_norm": 0.8128663843712138,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 9767
    },
    {
      "epoch": 0.09768,
      "grad_norm": 0.7660324180116674,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 9768
    },
    {
      "epoch": 0.09769,
      "grad_norm": 0.7219697561948825,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 9769
    },
    {
      "epoch": 0.0977,
      "grad_norm": 0.7328719015179833,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 9770
    },
    {
      "epoch": 0.09771,
      "grad_norm": 0.6934477189530808,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 9771
    },
    {
      "epoch": 0.09772,
      "grad_norm": 0.8226483433677332,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 9772
    },
    {
      "epoch": 0.09773,
      "grad_norm": 1.0407467139689701,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9773
    },
    {
      "epoch": 0.09774,
      "grad_norm": 1.379965710605147,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 9774
    },
    {
      "epoch": 0.09775,
      "grad_norm": 0.743770947000131,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 9775
    },
    {
      "epoch": 0.09776,
      "grad_norm": 0.7186119882928109,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 9776
    },
    {
      "epoch": 0.09777,
      "grad_norm": 0.7494079596750043,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 9777
    },
    {
      "epoch": 0.09778,
      "grad_norm": 0.8250803113950317,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 9778
    },
    {
      "epoch": 0.09779,
      "grad_norm": 0.9604869205387597,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 9779
    },
    {
      "epoch": 0.0978,
      "grad_norm": 1.2294622769995196,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 9780
    },
    {
      "epoch": 0.09781,
      "grad_norm": 0.8713194051271618,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 9781
    },
    {
      "epoch": 0.09782,
      "grad_norm": 0.8347924061909529,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 9782
    },
    {
      "epoch": 0.09783,
      "grad_norm": 1.1395358663076178,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 9783
    },
    {
      "epoch": 0.09784,
      "grad_norm": 1.1019088831778014,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 9784
    },
    {
      "epoch": 0.09785,
      "grad_norm": 1.169417153963683,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 9785
    },
    {
      "epoch": 0.09786,
      "grad_norm": 1.0034794245848353,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 9786
    },
    {
      "epoch": 0.09787,
      "grad_norm": 1.0284399680460898,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 9787
    },
    {
      "epoch": 0.09788,
      "grad_norm": 0.9057070287775122,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 9788
    },
    {
      "epoch": 0.09789,
      "grad_norm": 0.8560028986393307,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 9789
    },
    {
      "epoch": 0.0979,
      "grad_norm": 0.9296582713729994,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 9790
    },
    {
      "epoch": 0.09791,
      "grad_norm": 1.1827366857116655,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 9791
    },
    {
      "epoch": 0.09792,
      "grad_norm": 1.0204010831990022,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 9792
    },
    {
      "epoch": 0.09793,
      "grad_norm": 0.9153519628354887,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 9793
    },
    {
      "epoch": 0.09794,
      "grad_norm": 0.9068739502438202,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 9794
    },
    {
      "epoch": 0.09795,
      "grad_norm": 0.9496301652018498,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 9795
    },
    {
      "epoch": 0.09796,
      "grad_norm": 0.9958118680271525,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 9796
    },
    {
      "epoch": 0.09797,
      "grad_norm": 0.9557782092056588,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 9797
    },
    {
      "epoch": 0.09798,
      "grad_norm": 0.8533142469787229,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 9798
    },
    {
      "epoch": 0.09799,
      "grad_norm": 0.8190713601356191,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 9799
    },
    {
      "epoch": 0.098,
      "grad_norm": 0.9429777135714619,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 9800
    },
    {
      "epoch": 0.09801,
      "grad_norm": 1.0012945094084325,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 9801
    },
    {
      "epoch": 0.09802,
      "grad_norm": 0.9521250021764239,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 9802
    },
    {
      "epoch": 0.09803,
      "grad_norm": 1.2251473492074867,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 9803
    },
    {
      "epoch": 0.09804,
      "grad_norm": 0.9134016069934916,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 9804
    },
    {
      "epoch": 0.09805,
      "grad_norm": 0.8286943653648622,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 9805
    },
    {
      "epoch": 0.09806,
      "grad_norm": 0.7519693958000754,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 9806
    },
    {
      "epoch": 0.09807,
      "grad_norm": 0.7722973199567947,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 9807
    },
    {
      "epoch": 0.09808,
      "grad_norm": 0.74781146199721,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 9808
    },
    {
      "epoch": 0.09809,
      "grad_norm": 0.7364400647725448,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 9809
    },
    {
      "epoch": 0.0981,
      "grad_norm": 0.7200212527489623,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 9810
    },
    {
      "epoch": 0.09811,
      "grad_norm": 0.7544356048465244,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 9811
    },
    {
      "epoch": 0.09812,
      "grad_norm": 0.7686580462245453,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 9812
    },
    {
      "epoch": 0.09813,
      "grad_norm": 0.984346883906706,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 9813
    },
    {
      "epoch": 0.09814,
      "grad_norm": 1.2700083355049891,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 9814
    },
    {
      "epoch": 0.09815,
      "grad_norm": 0.9442514607634342,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 9815
    },
    {
      "epoch": 0.09816,
      "grad_norm": 1.0077025937052435,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 9816
    },
    {
      "epoch": 0.09817,
      "grad_norm": 0.971550909174614,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 9817
    },
    {
      "epoch": 0.09818,
      "grad_norm": 1.0193923220149106,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 9818
    },
    {
      "epoch": 0.09819,
      "grad_norm": 1.3161986032712076,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 9819
    },
    {
      "epoch": 0.0982,
      "grad_norm": 0.8557332761380659,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 9820
    },
    {
      "epoch": 0.09821,
      "grad_norm": 0.7233644886267359,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 9821
    },
    {
      "epoch": 0.09822,
      "grad_norm": 0.8326436608697041,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 9822
    },
    {
      "epoch": 0.09823,
      "grad_norm": 0.8980854204875377,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 9823
    },
    {
      "epoch": 0.09824,
      "grad_norm": 1.2283773516871788,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 9824
    },
    {
      "epoch": 0.09825,
      "grad_norm": 0.7785325214207814,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 9825
    },
    {
      "epoch": 0.09826,
      "grad_norm": 0.597043372236362,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 9826
    },
    {
      "epoch": 0.09827,
      "grad_norm": 0.6464758597311613,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 9827
    },
    {
      "epoch": 0.09828,
      "grad_norm": 0.8139034532487643,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 9828
    },
    {
      "epoch": 0.09829,
      "grad_norm": 1.244278860848288,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 9829
    },
    {
      "epoch": 0.0983,
      "grad_norm": 0.856108216172038,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 9830
    },
    {
      "epoch": 0.09831,
      "grad_norm": 0.6433844409257128,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 9831
    },
    {
      "epoch": 0.09832,
      "grad_norm": 0.7326548261726062,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 9832
    },
    {
      "epoch": 0.09833,
      "grad_norm": 0.7997173970955105,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 9833
    },
    {
      "epoch": 0.09834,
      "grad_norm": 0.9405671127384347,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 9834
    },
    {
      "epoch": 0.09835,
      "grad_norm": 1.131659672629919,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 9835
    },
    {
      "epoch": 0.09836,
      "grad_norm": 0.997538927089926,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 9836
    },
    {
      "epoch": 0.09837,
      "grad_norm": 1.0175411430188779,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 9837
    },
    {
      "epoch": 0.09838,
      "grad_norm": 0.9799166460402474,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9838
    },
    {
      "epoch": 0.09839,
      "grad_norm": 1.0396590201522609,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 9839
    },
    {
      "epoch": 0.0984,
      "grad_norm": 1.1282086069881085,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 9840
    },
    {
      "epoch": 0.09841,
      "grad_norm": 0.8198231903899644,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 9841
    },
    {
      "epoch": 0.09842,
      "grad_norm": 0.9217921401090745,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 9842
    },
    {
      "epoch": 0.09843,
      "grad_norm": 1.0197186165919299,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 9843
    },
    {
      "epoch": 0.09844,
      "grad_norm": 1.2101350812829232,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 9844
    },
    {
      "epoch": 0.09845,
      "grad_norm": 0.8651149977324143,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 9845
    },
    {
      "epoch": 0.09846,
      "grad_norm": 0.7265952518823648,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 9846
    },
    {
      "epoch": 0.09847,
      "grad_norm": 0.6887147254332203,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 9847
    },
    {
      "epoch": 0.09848,
      "grad_norm": 0.7507853589418222,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 9848
    },
    {
      "epoch": 0.09849,
      "grad_norm": 1.0036628039255613,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 9849
    },
    {
      "epoch": 0.0985,
      "grad_norm": 1.2748454320535545,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 9850
    },
    {
      "epoch": 0.09851,
      "grad_norm": 0.8314804970803586,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 9851
    },
    {
      "epoch": 0.09852,
      "grad_norm": 0.817675377067337,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 9852
    },
    {
      "epoch": 0.09853,
      "grad_norm": 0.7841091435659291,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9853
    },
    {
      "epoch": 0.09854,
      "grad_norm": 0.6372546033409912,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 9854
    },
    {
      "epoch": 0.09855,
      "grad_norm": 0.679618174317228,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 9855
    },
    {
      "epoch": 0.09856,
      "grad_norm": 0.7500362277289165,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 9856
    },
    {
      "epoch": 0.09857,
      "grad_norm": 0.8068873392409563,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 9857
    },
    {
      "epoch": 0.09858,
      "grad_norm": 0.7652166764948078,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 9858
    },
    {
      "epoch": 0.09859,
      "grad_norm": 0.8218029235936964,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 9859
    },
    {
      "epoch": 0.0986,
      "grad_norm": 0.843394170131785,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 9860
    },
    {
      "epoch": 0.09861,
      "grad_norm": 0.9379505462471198,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 9861
    },
    {
      "epoch": 0.09862,
      "grad_norm": 1.32072231867818,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 9862
    },
    {
      "epoch": 0.09863,
      "grad_norm": 1.0523750210753542,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 9863
    },
    {
      "epoch": 0.09864,
      "grad_norm": 0.9406173426354154,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9864
    },
    {
      "epoch": 0.09865,
      "grad_norm": 0.8931432490745203,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 9865
    },
    {
      "epoch": 0.09866,
      "grad_norm": 0.9699347096958343,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 9866
    },
    {
      "epoch": 0.09867,
      "grad_norm": 1.046711651416207,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 9867
    },
    {
      "epoch": 0.09868,
      "grad_norm": 1.0808042490594871,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 9868
    },
    {
      "epoch": 0.09869,
      "grad_norm": 1.1961822678848502,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 9869
    },
    {
      "epoch": 0.0987,
      "grad_norm": 0.9228079748238792,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 9870
    },
    {
      "epoch": 0.09871,
      "grad_norm": 0.8545983336178743,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 9871
    },
    {
      "epoch": 0.09872,
      "grad_norm": 0.8784832785791439,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 9872
    },
    {
      "epoch": 0.09873,
      "grad_norm": 1.070280498696221,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 9873
    },
    {
      "epoch": 0.09874,
      "grad_norm": 1.0439215572833065,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 9874
    },
    {
      "epoch": 0.09875,
      "grad_norm": 1.0814551372745504,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 9875
    },
    {
      "epoch": 0.09876,
      "grad_norm": 0.8511905866205883,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 9876
    },
    {
      "epoch": 0.09877,
      "grad_norm": 0.811275306253899,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 9877
    },
    {
      "epoch": 0.09878,
      "grad_norm": 1.249209967465114,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 9878
    },
    {
      "epoch": 0.09879,
      "grad_norm": 0.9304246999674388,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 9879
    },
    {
      "epoch": 0.0988,
      "grad_norm": 0.9414369214716679,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 9880
    },
    {
      "epoch": 0.09881,
      "grad_norm": 1.0878622028222777,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 9881
    },
    {
      "epoch": 0.09882,
      "grad_norm": 0.9240651030772346,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 9882
    },
    {
      "epoch": 0.09883,
      "grad_norm": 0.7991058410005727,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 9883
    },
    {
      "epoch": 0.09884,
      "grad_norm": 0.9050441156700046,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 9884
    },
    {
      "epoch": 0.09885,
      "grad_norm": 1.1399203818579307,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 9885
    },
    {
      "epoch": 0.09886,
      "grad_norm": 0.8766884786296881,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 9886
    },
    {
      "epoch": 0.09887,
      "grad_norm": 0.801700624949135,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 9887
    },
    {
      "epoch": 0.09888,
      "grad_norm": 0.8628961038287333,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 9888
    },
    {
      "epoch": 0.09889,
      "grad_norm": 0.7467079097142902,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 9889
    },
    {
      "epoch": 0.0989,
      "grad_norm": 0.7726770182387934,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 9890
    },
    {
      "epoch": 0.09891,
      "grad_norm": 0.9462651212194271,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 9891
    },
    {
      "epoch": 0.09892,
      "grad_norm": 1.139679261250144,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 9892
    },
    {
      "epoch": 0.09893,
      "grad_norm": 1.0951119397642,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9893
    },
    {
      "epoch": 0.09894,
      "grad_norm": 0.905062924614346,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 9894
    },
    {
      "epoch": 0.09895,
      "grad_norm": 0.7933902578399411,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 9895
    },
    {
      "epoch": 0.09896,
      "grad_norm": 0.704722792642583,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 9896
    },
    {
      "epoch": 0.09897,
      "grad_norm": 0.6940718562411224,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 9897
    },
    {
      "epoch": 0.09898,
      "grad_norm": 0.8845748690268396,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 9898
    },
    {
      "epoch": 0.09899,
      "grad_norm": 1.0988377109535674,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 9899
    },
    {
      "epoch": 0.099,
      "grad_norm": 1.1105673173122586,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 9900
    },
    {
      "epoch": 0.09901,
      "grad_norm": 0.9810554294045679,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 9901
    },
    {
      "epoch": 0.09902,
      "grad_norm": 1.088761843451906,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 9902
    },
    {
      "epoch": 0.09903,
      "grad_norm": 0.8865852858603028,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 9903
    },
    {
      "epoch": 0.09904,
      "grad_norm": 1.043616089079261,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 9904
    },
    {
      "epoch": 0.09905,
      "grad_norm": 1.068953951671368,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9905
    },
    {
      "epoch": 0.09906,
      "grad_norm": 1.044686025986228,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 9906
    },
    {
      "epoch": 0.09907,
      "grad_norm": 0.8886077091593347,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 9907
    },
    {
      "epoch": 0.09908,
      "grad_norm": 0.8555950937756621,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 9908
    },
    {
      "epoch": 0.09909,
      "grad_norm": 0.7817570855188947,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 9909
    },
    {
      "epoch": 0.0991,
      "grad_norm": 0.8362250403473259,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9910
    },
    {
      "epoch": 0.09911,
      "grad_norm": 0.9500328133313455,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 9911
    },
    {
      "epoch": 0.09912,
      "grad_norm": 1.031805969191571,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 9912
    },
    {
      "epoch": 0.09913,
      "grad_norm": 1.0138965400557471,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 9913
    },
    {
      "epoch": 0.09914,
      "grad_norm": 0.9678001875339941,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 9914
    },
    {
      "epoch": 0.09915,
      "grad_norm": 0.9475987221695131,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9915
    },
    {
      "epoch": 0.09916,
      "grad_norm": 0.9451880390051262,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 9916
    },
    {
      "epoch": 0.09917,
      "grad_norm": 0.9072128377197956,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 9917
    },
    {
      "epoch": 0.09918,
      "grad_norm": 0.9058874365902825,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 9918
    },
    {
      "epoch": 0.09919,
      "grad_norm": 0.9988923576329186,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 9919
    },
    {
      "epoch": 0.0992,
      "grad_norm": 0.972034979186135,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 9920
    },
    {
      "epoch": 0.09921,
      "grad_norm": 0.9407505238435098,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 9921
    },
    {
      "epoch": 0.09922,
      "grad_norm": 1.0709663672999241,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 9922
    },
    {
      "epoch": 0.09923,
      "grad_norm": 1.011390937859059,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 9923
    },
    {
      "epoch": 0.09924,
      "grad_norm": 1.0734808599050338,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 9924
    },
    {
      "epoch": 0.09925,
      "grad_norm": 0.9724775683011624,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 9925
    },
    {
      "epoch": 0.09926,
      "grad_norm": 0.9858754523172633,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 9926
    },
    {
      "epoch": 0.09927,
      "grad_norm": 1.0026350425415425,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 9927
    },
    {
      "epoch": 0.09928,
      "grad_norm": 0.9618825289266347,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9928
    },
    {
      "epoch": 0.09929,
      "grad_norm": 0.8720935243540162,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 9929
    },
    {
      "epoch": 0.0993,
      "grad_norm": 0.8729170421206854,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 9930
    },
    {
      "epoch": 0.09931,
      "grad_norm": 0.9387248454992504,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 9931
    },
    {
      "epoch": 0.09932,
      "grad_norm": 1.0565356457114414,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 9932
    },
    {
      "epoch": 0.09933,
      "grad_norm": 0.8625813163715004,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 9933
    },
    {
      "epoch": 0.09934,
      "grad_norm": 0.7002720895141619,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 9934
    },
    {
      "epoch": 0.09935,
      "grad_norm": 0.7613232133923477,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 9935
    },
    {
      "epoch": 0.09936,
      "grad_norm": 1.0484981670115654,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 9936
    },
    {
      "epoch": 0.09937,
      "grad_norm": 1.2541580708606934,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 9937
    },
    {
      "epoch": 0.09938,
      "grad_norm": 0.7679423053752109,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 9938
    },
    {
      "epoch": 0.09939,
      "grad_norm": 0.7623951555699858,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 9939
    },
    {
      "epoch": 0.0994,
      "grad_norm": 0.743467827926508,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9940
    },
    {
      "epoch": 0.09941,
      "grad_norm": 0.8158817381706474,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 9941
    },
    {
      "epoch": 0.09942,
      "grad_norm": 0.9477928068113599,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 9942
    },
    {
      "epoch": 0.09943,
      "grad_norm": 0.8989443009796476,
      "learning_rate": 0.003,
      "loss": 4.0361,
      "step": 9943
    },
    {
      "epoch": 0.09944,
      "grad_norm": 0.8465160184457518,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 9944
    },
    {
      "epoch": 0.09945,
      "grad_norm": 0.96077143421336,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 9945
    },
    {
      "epoch": 0.09946,
      "grad_norm": 0.982634646562586,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 9946
    },
    {
      "epoch": 0.09947,
      "grad_norm": 0.9829649395908524,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 9947
    },
    {
      "epoch": 0.09948,
      "grad_norm": 1.0904657767502883,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 9948
    },
    {
      "epoch": 0.09949,
      "grad_norm": 0.958754249548164,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 9949
    },
    {
      "epoch": 0.0995,
      "grad_norm": 0.941102523885678,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 9950
    },
    {
      "epoch": 0.09951,
      "grad_norm": 0.9547561697534392,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 9951
    },
    {
      "epoch": 0.09952,
      "grad_norm": 1.0369210267712394,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 9952
    },
    {
      "epoch": 0.09953,
      "grad_norm": 1.1666110255591098,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 9953
    },
    {
      "epoch": 0.09954,
      "grad_norm": 0.8404966208187018,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 9954
    },
    {
      "epoch": 0.09955,
      "grad_norm": 0.6348854161168744,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 9955
    },
    {
      "epoch": 0.09956,
      "grad_norm": 0.6821965250062815,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 9956
    },
    {
      "epoch": 0.09957,
      "grad_norm": 0.6331997868974779,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 9957
    },
    {
      "epoch": 0.09958,
      "grad_norm": 0.777484394684235,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 9958
    },
    {
      "epoch": 0.09959,
      "grad_norm": 1.0101723616966292,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 9959
    },
    {
      "epoch": 0.0996,
      "grad_norm": 1.2396789474457461,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 9960
    },
    {
      "epoch": 0.09961,
      "grad_norm": 0.883683117211385,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 9961
    },
    {
      "epoch": 0.09962,
      "grad_norm": 0.9018658759893716,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 9962
    },
    {
      "epoch": 0.09963,
      "grad_norm": 1.0876146413308398,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 9963
    },
    {
      "epoch": 0.09964,
      "grad_norm": 1.0287603306521411,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 9964
    },
    {
      "epoch": 0.09965,
      "grad_norm": 1.0182673045567394,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 9965
    },
    {
      "epoch": 0.09966,
      "grad_norm": 1.1527631254244322,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 9966
    },
    {
      "epoch": 0.09967,
      "grad_norm": 0.8709975266827947,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 9967
    },
    {
      "epoch": 0.09968,
      "grad_norm": 0.8331778651716775,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 9968
    },
    {
      "epoch": 0.09969,
      "grad_norm": 0.9011465361487814,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 9969
    },
    {
      "epoch": 0.0997,
      "grad_norm": 1.0080570307471883,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 9970
    },
    {
      "epoch": 0.09971,
      "grad_norm": 1.1140173135877225,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 9971
    },
    {
      "epoch": 0.09972,
      "grad_norm": 0.9393638818310338,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 9972
    },
    {
      "epoch": 0.09973,
      "grad_norm": 1.0649082720147478,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 9973
    },
    {
      "epoch": 0.09974,
      "grad_norm": 1.130092450772193,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 9974
    },
    {
      "epoch": 0.09975,
      "grad_norm": 0.88585944847689,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 9975
    },
    {
      "epoch": 0.09976,
      "grad_norm": 0.857801840723658,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9976
    },
    {
      "epoch": 0.09977,
      "grad_norm": 0.8296564805689629,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 9977
    },
    {
      "epoch": 0.09978,
      "grad_norm": 0.8072297066940064,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 9978
    },
    {
      "epoch": 0.09979,
      "grad_norm": 0.8227232337051928,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 9979
    },
    {
      "epoch": 0.0998,
      "grad_norm": 0.8476832328670773,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 9980
    },
    {
      "epoch": 0.09981,
      "grad_norm": 0.9357206908798009,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9981
    },
    {
      "epoch": 0.09982,
      "grad_norm": 1.143225588531109,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 9982
    },
    {
      "epoch": 0.09983,
      "grad_norm": 0.8237550880806321,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 9983
    },
    {
      "epoch": 0.09984,
      "grad_norm": 0.7811421447324085,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 9984
    },
    {
      "epoch": 0.09985,
      "grad_norm": 0.8438712038256704,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 9985
    },
    {
      "epoch": 0.09986,
      "grad_norm": 0.8254292728857884,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 9986
    },
    {
      "epoch": 0.09987,
      "grad_norm": 0.8131007944980652,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 9987
    },
    {
      "epoch": 0.09988,
      "grad_norm": 0.8584836605861299,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 9988
    },
    {
      "epoch": 0.09989,
      "grad_norm": 1.089542214562673,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 9989
    },
    {
      "epoch": 0.0999,
      "grad_norm": 1.2008176887832276,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 9990
    },
    {
      "epoch": 0.09991,
      "grad_norm": 0.9071696842404575,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 9991
    },
    {
      "epoch": 0.09992,
      "grad_norm": 0.9778468657106637,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 9992
    },
    {
      "epoch": 0.09993,
      "grad_norm": 1.018266367771181,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 9993
    },
    {
      "epoch": 0.09994,
      "grad_norm": 0.9996060334571353,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 9994
    },
    {
      "epoch": 0.09995,
      "grad_norm": 0.8639535717682494,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 9995
    },
    {
      "epoch": 0.09996,
      "grad_norm": 0.752574714119443,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 9996
    },
    {
      "epoch": 0.09997,
      "grad_norm": 0.7765764977536432,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 9997
    },
    {
      "epoch": 0.09998,
      "grad_norm": 0.8178576426181308,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 9998
    },
    {
      "epoch": 0.09999,
      "grad_norm": 1.007703696887012,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 9999
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2762926674519628,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 10000
    }
  ],
  "logging_steps": 1,
  "max_steps": 100000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.9643642855424e+17,
  "train_batch_size": 1024,
  "trial_name": null,
  "trial_params": null
}