{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 12500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 3.2635705523930243,
      "learning_rate": 2.6666666666666667e-08,
      "loss": 0.3102,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.543236875321121,
      "learning_rate": 5.3333333333333334e-08,
      "loss": 0.3172,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0655688091283118,
      "learning_rate": 8e-08,
      "loss": 0.2632,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.081711234511395,
      "learning_rate": 1.0666666666666667e-07,
      "loss": 0.8341,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.154098344404168,
      "learning_rate": 1.3333333333333336e-07,
      "loss": 0.3314,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.535508933365584,
      "learning_rate": 1.6e-07,
      "loss": 0.9176,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.402369845867405,
      "learning_rate": 1.866666666666667e-07,
      "loss": 0.3436,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5114298823570063,
      "learning_rate": 2.1333333333333334e-07,
      "loss": 0.3615,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.476628489258035,
      "learning_rate": 2.4000000000000003e-07,
      "loss": 0.9509,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.7138745616249826,
      "learning_rate": 2.666666666666667e-07,
      "loss": 0.3222,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.8609823589881858,
      "learning_rate": 2.9333333333333337e-07,
      "loss": 0.2921,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.075192583236817,
      "learning_rate": 3.2e-07,
      "loss": 0.3245,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.998118821182612,
      "learning_rate": 3.466666666666667e-07,
      "loss": 0.3024,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.045818795124452,
      "learning_rate": 3.733333333333334e-07,
      "loss": 0.3282,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.973244687568935,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.7978,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.9808464485960586,
      "learning_rate": 4.266666666666667e-07,
      "loss": 0.3279,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.4364859666828984,
      "learning_rate": 4.533333333333334e-07,
      "loss": 0.2966,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.430219771069506,
      "learning_rate": 4.800000000000001e-07,
      "loss": 0.3061,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1459749927364595,
      "learning_rate": 5.066666666666667e-07,
      "loss": 0.3716,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.583293257964736,
      "learning_rate": 5.333333333333335e-07,
      "loss": 0.2714,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.820393113611824,
      "learning_rate": 5.6e-07,
      "loss": 0.2903,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.996635641133045,
      "learning_rate": 5.866666666666667e-07,
      "loss": 0.9355,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8556690346481117,
      "learning_rate": 6.133333333333333e-07,
      "loss": 0.3388,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.596995774371009,
      "learning_rate": 6.4e-07,
      "loss": 0.3523,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2831190744445444,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.2908,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.599652572249794,
      "learning_rate": 6.933333333333334e-07,
      "loss": 0.3018,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.690583140200056,
      "learning_rate": 7.2e-07,
      "loss": 0.3214,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.888226221915256,
      "learning_rate": 7.466666666666668e-07,
      "loss": 0.7488,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.248420029290666,
      "learning_rate": 7.733333333333335e-07,
      "loss": 0.292,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.166642960970659,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.8535,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.153730429443561,
      "learning_rate": 8.266666666666668e-07,
      "loss": 0.2999,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.045491499674651,
      "learning_rate": 8.533333333333334e-07,
      "loss": 0.2379,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4058724593030694,
      "learning_rate": 8.8e-07,
      "loss": 0.2759,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1177615847510913,
      "learning_rate": 9.066666666666668e-07,
      "loss": 0.2711,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3992686930361855,
      "learning_rate": 9.333333333333334e-07,
      "loss": 0.2788,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.28546520521007,
      "learning_rate": 9.600000000000001e-07,
      "loss": 0.239,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3080986760989655,
      "learning_rate": 9.866666666666668e-07,
      "loss": 0.2883,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.662264753934099,
      "learning_rate": 1.0133333333333333e-06,
      "loss": 0.2807,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.175493921197926,
      "learning_rate": 1.04e-06,
      "loss": 0.2659,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1416533996119678,
      "learning_rate": 1.066666666666667e-06,
      "loss": 0.308,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9403011675836694,
      "learning_rate": 1.0933333333333334e-06,
      "loss": 0.2768,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.662490627051228,
      "learning_rate": 1.12e-06,
      "loss": 0.7614,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.7966519208521365,
      "learning_rate": 1.1466666666666668e-06,
      "loss": 0.8204,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9809096759810592,
      "learning_rate": 1.1733333333333335e-06,
      "loss": 0.2237,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.47988732627334,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.8607,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.296376364024676,
      "learning_rate": 1.2266666666666666e-06,
      "loss": 0.2883,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.323464926411196,
      "learning_rate": 1.2533333333333333e-06,
      "loss": 0.2503,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.422466514928935,
      "learning_rate": 1.28e-06,
      "loss": 0.2873,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1864023301873354,
      "learning_rate": 1.3066666666666667e-06,
      "loss": 0.2343,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9463373428273896,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.2338,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1405490757192798,
      "learning_rate": 1.3600000000000001e-06,
      "loss": 0.2221,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0836180918551928,
      "learning_rate": 1.3866666666666668e-06,
      "loss": 0.2491,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.100488803126805,
      "learning_rate": 1.4133333333333335e-06,
      "loss": 0.2822,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.032919241222733,
      "learning_rate": 1.44e-06,
      "loss": 0.2195,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.815257401523501,
      "learning_rate": 1.4666666666666669e-06,
      "loss": 0.7812,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1564309626298725,
      "learning_rate": 1.4933333333333336e-06,
      "loss": 0.2469,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9480887603542227,
      "learning_rate": 1.52e-06,
      "loss": 0.231,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.011965218482532,
      "learning_rate": 1.546666666666667e-06,
      "loss": 0.2705,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.622723478044898,
      "learning_rate": 1.5733333333333334e-06,
      "loss": 0.2621,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.14774348381317,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.2304,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1845766304625953,
      "learning_rate": 1.6266666666666666e-06,
      "loss": 0.2012,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.066296298115803,
      "learning_rate": 1.6533333333333335e-06,
      "loss": 0.2328,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.148557387257148,
      "learning_rate": 1.6800000000000002e-06,
      "loss": 0.215,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2476637406213626,
      "learning_rate": 1.7066666666666667e-06,
      "loss": 0.26,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0641999839834813,
      "learning_rate": 1.7333333333333336e-06,
      "loss": 0.2399,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0376118977375186,
      "learning_rate": 1.76e-06,
      "loss": 0.2016,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0274034656824056,
      "learning_rate": 1.7866666666666668e-06,
      "loss": 0.228,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2957221670003958,
      "learning_rate": 1.8133333333333337e-06,
      "loss": 0.2641,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.041701984504308,
      "learning_rate": 1.8400000000000002e-06,
      "loss": 0.1948,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.037906356469153,
      "learning_rate": 1.8666666666666669e-06,
      "loss": 0.242,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0212291033145426,
      "learning_rate": 1.8933333333333333e-06,
      "loss": 0.2195,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.005787621345112,
      "learning_rate": 1.9200000000000003e-06,
      "loss": 0.2436,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.110313066520128,
      "learning_rate": 1.9466666666666665e-06,
      "loss": 0.2097,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1635446281024207,
      "learning_rate": 1.9733333333333336e-06,
      "loss": 0.1888,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.200565820752455,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.7992,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.193104289614479,
      "learning_rate": 2.0266666666666666e-06,
      "loss": 0.2297,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9612852870447144,
      "learning_rate": 2.0533333333333337e-06,
      "loss": 0.208,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.391661786594564,
      "learning_rate": 2.08e-06,
      "loss": 0.7669,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.11332632538686,
      "learning_rate": 2.1066666666666667e-06,
      "loss": 0.2424,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.260362688106463,
      "learning_rate": 2.133333333333334e-06,
      "loss": 0.821,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.803596009402048,
      "learning_rate": 2.16e-06,
      "loss": 0.6818,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.991407922507565,
      "learning_rate": 2.1866666666666668e-06,
      "loss": 0.2306,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.187669567898178,
      "learning_rate": 2.2133333333333335e-06,
      "loss": 0.7662,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2267663175830386,
      "learning_rate": 2.24e-06,
      "loss": 0.2389,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9644457331448772,
      "learning_rate": 2.266666666666667e-06,
      "loss": 0.2018,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0324819383736497,
      "learning_rate": 2.2933333333333335e-06,
      "loss": 0.2407,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.84810247901649,
      "learning_rate": 2.3200000000000002e-06,
      "loss": 0.5149,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.75065201318718,
      "learning_rate": 2.346666666666667e-06,
      "loss": 0.7061,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.470538421286546,
      "learning_rate": 2.3733333333333336e-06,
      "loss": 0.829,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.583734329121602,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.6821,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.099338441420322,
      "learning_rate": 2.426666666666667e-06,
      "loss": 0.9718,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3324966371105424,
      "learning_rate": 2.4533333333333333e-06,
      "loss": 0.2248,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.016810889780768,
      "learning_rate": 2.4800000000000004e-06,
      "loss": 0.2584,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1304257948220604,
      "learning_rate": 2.5066666666666667e-06,
      "loss": 0.2196,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0180392724288096,
      "learning_rate": 2.5333333333333338e-06,
      "loss": 0.2055,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1615753822613946,
      "learning_rate": 2.56e-06,
      "loss": 0.2371,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0803764832105878,
      "learning_rate": 2.5866666666666667e-06,
      "loss": 0.2408,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2928421674371715,
      "learning_rate": 2.6133333333333334e-06,
      "loss": 0.2244,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0523804437065363,
      "learning_rate": 2.64e-06,
      "loss": 0.277,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2286435641079323,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.2225,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2962715373714495,
      "learning_rate": 2.6933333333333335e-06,
      "loss": 0.2735,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.285439769931514,
      "learning_rate": 2.7200000000000002e-06,
      "loss": 0.2372,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9342386464302803,
      "learning_rate": 2.746666666666667e-06,
      "loss": 0.208,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1503197007060293,
      "learning_rate": 2.7733333333333336e-06,
      "loss": 0.2809,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7893235217469743,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.1758,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.906007462604689,
      "learning_rate": 2.826666666666667e-06,
      "loss": 0.2484,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9786598638462265,
      "learning_rate": 2.8533333333333337e-06,
      "loss": 0.2114,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.049152819107196,
      "learning_rate": 2.88e-06,
      "loss": 0.2237,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6661362139299618,
      "learning_rate": 2.906666666666667e-06,
      "loss": 0.2113,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2480093172022495,
      "learning_rate": 2.9333333333333338e-06,
      "loss": 0.2459,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.927738641246031,
      "learning_rate": 2.96e-06,
      "loss": 0.2035,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8857780825353894,
      "learning_rate": 2.986666666666667e-06,
      "loss": 0.1594,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9963991331191024,
      "learning_rate": 3.013333333333334e-06,
      "loss": 0.2328,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7325046061943692,
      "learning_rate": 3.04e-06,
      "loss": 0.2022,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.970097203736795,
      "learning_rate": 3.066666666666667e-06,
      "loss": 0.677,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.926852729565814,
      "learning_rate": 3.093333333333334e-06,
      "loss": 0.5464,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0605927676465607,
      "learning_rate": 3.12e-06,
      "loss": 0.198,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1104917300712245,
      "learning_rate": 3.146666666666667e-06,
      "loss": 0.2352,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8787240286872835,
      "learning_rate": 3.173333333333334e-06,
      "loss": 0.2465,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9218432020497223,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.2031,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1772188471633362,
      "learning_rate": 3.226666666666667e-06,
      "loss": 0.2051,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.328005319469429,
      "learning_rate": 3.2533333333333332e-06,
      "loss": 0.2153,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.925604145136642,
      "learning_rate": 3.2800000000000004e-06,
      "loss": 0.25,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9820026322447761,
      "learning_rate": 3.306666666666667e-06,
      "loss": 0.1943,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0411928414856777,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.1685,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8652352360105509,
      "learning_rate": 3.3600000000000004e-06,
      "loss": 0.1883,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2516584878176875,
      "learning_rate": 3.386666666666667e-06,
      "loss": 0.1727,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.087264742735575,
      "learning_rate": 3.4133333333333334e-06,
      "loss": 0.793,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.996973052396692,
      "learning_rate": 3.44e-06,
      "loss": 0.6537,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3405875842702155,
      "learning_rate": 3.4666666666666672e-06,
      "loss": 0.2419,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.035032668627448,
      "learning_rate": 3.4933333333333335e-06,
      "loss": 0.1946,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.043677962988261,
      "learning_rate": 3.52e-06,
      "loss": 0.2255,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.178344339290289,
      "learning_rate": 3.5466666666666673e-06,
      "loss": 0.2287,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9821123059862547,
      "learning_rate": 3.5733333333333336e-06,
      "loss": 0.1983,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.122660059246671,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.2235,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.862156009599487,
      "learning_rate": 3.6266666666666674e-06,
      "loss": 0.1862,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.209332983222699,
      "learning_rate": 3.6533333333333336e-06,
      "loss": 0.2381,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3148626754621127,
      "learning_rate": 3.6800000000000003e-06,
      "loss": 0.229,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.468563994514957,
      "learning_rate": 3.7066666666666666e-06,
      "loss": 0.5659,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.231979066487987,
      "learning_rate": 3.7333333333333337e-06,
      "loss": 0.2475,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0539829331885677,
      "learning_rate": 3.7600000000000004e-06,
      "loss": 0.2088,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9101853584281967,
      "learning_rate": 3.7866666666666667e-06,
      "loss": 0.1787,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7549572650718648,
      "learning_rate": 3.813333333333334e-06,
      "loss": 0.1745,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9917671770254983,
      "learning_rate": 3.8400000000000005e-06,
      "loss": 0.2373,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9719928027465417,
      "learning_rate": 3.866666666666667e-06,
      "loss": 0.1964,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0200700893787116,
      "learning_rate": 3.893333333333333e-06,
      "loss": 0.2487,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9704671229709034,
      "learning_rate": 3.920000000000001e-06,
      "loss": 0.2275,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9949728915250518,
      "learning_rate": 3.946666666666667e-06,
      "loss": 0.2455,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.183427935948576,
      "learning_rate": 3.973333333333333e-06,
      "loss": 0.4523,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8543991607502426,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.2223,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8723747071389218,
      "learning_rate": 4.026666666666667e-06,
      "loss": 0.1991,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9645684725537398,
      "learning_rate": 4.053333333333333e-06,
      "loss": 0.1986,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9654853771367395,
      "learning_rate": 4.08e-06,
      "loss": 0.2294,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.312535389212955,
      "learning_rate": 4.1066666666666674e-06,
      "loss": 0.2392,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9576843768103063,
      "learning_rate": 4.133333333333333e-06,
      "loss": 0.236,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1319457853387407,
      "learning_rate": 4.16e-06,
      "loss": 0.2175,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.444462809770049,
      "learning_rate": 4.1866666666666675e-06,
      "loss": 0.8498,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0619876961010677,
      "learning_rate": 4.213333333333333e-06,
      "loss": 0.2348,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0771362926774257,
      "learning_rate": 4.24e-06,
      "loss": 0.1975,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0444600557386154,
      "learning_rate": 4.266666666666668e-06,
      "loss": 0.2139,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.130966379197011,
      "learning_rate": 4.2933333333333334e-06,
      "loss": 0.2686,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8898003382530546,
      "learning_rate": 4.32e-06,
      "loss": 0.2184,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2166378958169264,
      "learning_rate": 4.346666666666667e-06,
      "loss": 0.2192,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.055870577216373,
      "learning_rate": 4.3733333333333335e-06,
      "loss": 0.6057,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.127915791381282,
      "learning_rate": 4.4e-06,
      "loss": 0.2716,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.367786962086758,
      "learning_rate": 4.426666666666667e-06,
      "loss": 0.7017,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1641741065311466,
      "learning_rate": 4.453333333333334e-06,
      "loss": 0.2736,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8708115322718382,
      "learning_rate": 4.48e-06,
      "loss": 0.1873,
      "step": 168
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0197250826297384,
      "learning_rate": 4.506666666666667e-06,
      "loss": 0.2603,
      "step": 169
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.12178576784744,
      "learning_rate": 4.533333333333334e-06,
      "loss": 0.245,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0097943872195185,
      "learning_rate": 4.56e-06,
      "loss": 0.2111,
      "step": 171
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.102553739297998,
      "learning_rate": 4.586666666666667e-06,
      "loss": 0.2364,
      "step": 172
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.401209297673534,
      "learning_rate": 4.613333333333334e-06,
      "loss": 0.7021,
      "step": 173
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2145803698424196,
      "learning_rate": 4.6400000000000005e-06,
      "loss": 0.2344,
      "step": 174
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8939923366942533,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.1737,
      "step": 175
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7960576034922373,
      "learning_rate": 4.693333333333334e-06,
      "loss": 0.1825,
      "step": 176
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9298842821684317,
      "learning_rate": 4.7200000000000005e-06,
      "loss": 0.229,
      "step": 177
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1151162193087947,
      "learning_rate": 4.746666666666667e-06,
      "loss": 0.2215,
      "step": 178
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6525488005070617,
      "learning_rate": 4.773333333333334e-06,
      "loss": 0.1643,
      "step": 179
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1236553322274374,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.2302,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7230232090898925,
      "learning_rate": 4.826666666666667e-06,
      "loss": 0.1685,
      "step": 181
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.143435168699805,
      "learning_rate": 4.853333333333334e-06,
      "loss": 0.1821,
      "step": 182
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.024389744687161,
      "learning_rate": 4.880000000000001e-06,
      "loss": 0.2364,
      "step": 183
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1772637853759935,
      "learning_rate": 4.9066666666666666e-06,
      "loss": 0.2518,
      "step": 184
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.940779281076909,
      "learning_rate": 4.933333333333334e-06,
      "loss": 0.2328,
      "step": 185
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9820568269378853,
      "learning_rate": 4.960000000000001e-06,
      "loss": 0.2611,
      "step": 186
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9456026666072652,
      "learning_rate": 4.986666666666667e-06,
      "loss": 0.1822,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9269335908419998,
      "learning_rate": 5.013333333333333e-06,
      "loss": 0.1799,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8155386201438246,
      "learning_rate": 5.04e-06,
      "loss": 0.196,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9008726008303274,
      "learning_rate": 5.0666666666666676e-06,
      "loss": 0.2435,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.827807938534924,
      "learning_rate": 5.093333333333333e-06,
      "loss": 0.4753,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.011283711428876,
      "learning_rate": 5.12e-06,
      "loss": 0.2263,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9527270517091515,
      "learning_rate": 5.146666666666668e-06,
      "loss": 0.2415,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.871668931176077,
      "learning_rate": 5.1733333333333335e-06,
      "loss": 0.1864,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9831175773285001,
      "learning_rate": 5.2e-06,
      "loss": 0.2285,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9719109856763253,
      "learning_rate": 5.226666666666667e-06,
      "loss": 0.252,
      "step": 196
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0807459864555495,
      "learning_rate": 5.2533333333333336e-06,
      "loss": 0.2228,
      "step": 197
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8797229209078343,
      "learning_rate": 5.28e-06,
      "loss": 0.2275,
      "step": 198
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.08389250069845,
      "learning_rate": 5.306666666666667e-06,
      "loss": 0.2272,
      "step": 199
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8425804995374695,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.2299,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0385319681393197,
      "learning_rate": 5.36e-06,
      "loss": 0.2434,
      "step": 201
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1102651455026393,
      "learning_rate": 5.386666666666667e-06,
      "loss": 0.2121,
      "step": 202
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9422624276384832,
      "learning_rate": 5.413333333333334e-06,
      "loss": 0.2092,
      "step": 203
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0720994845225444,
      "learning_rate": 5.4400000000000004e-06,
      "loss": 0.2313,
      "step": 204
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.921416678358997,
      "learning_rate": 5.466666666666667e-06,
      "loss": 0.2433,
      "step": 205
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.797567743632961,
      "learning_rate": 5.493333333333334e-06,
      "loss": 0.1871,
      "step": 206
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9465412885408135,
      "learning_rate": 5.5200000000000005e-06,
      "loss": 0.2206,
      "step": 207
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9013536989372188,
      "learning_rate": 5.546666666666667e-06,
      "loss": 0.2297,
      "step": 208
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9389930349578413,
      "learning_rate": 5.573333333333334e-06,
      "loss": 0.215,
      "step": 209
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.975104964930779,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.2737,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.168711960577819,
      "learning_rate": 5.626666666666667e-06,
      "loss": 0.1783,
      "step": 211
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9704932793281564,
      "learning_rate": 5.653333333333334e-06,
      "loss": 0.227,
      "step": 212
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.376518860409301,
      "learning_rate": 5.68e-06,
      "loss": 0.238,
      "step": 213
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8730662945261776,
      "learning_rate": 5.706666666666667e-06,
      "loss": 0.2194,
      "step": 214
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0232295140895715,
      "learning_rate": 5.733333333333334e-06,
      "loss": 0.2156,
      "step": 215
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5439796923408347,
      "learning_rate": 5.76e-06,
      "loss": 0.2265,
      "step": 216
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8615861537301621,
      "learning_rate": 5.7866666666666674e-06,
      "loss": 0.1879,
      "step": 217
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.200188694629253,
      "learning_rate": 5.813333333333334e-06,
      "loss": 0.2444,
      "step": 218
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8304717092554472,
      "learning_rate": 5.84e-06,
      "loss": 0.2002,
      "step": 219
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1521566647861547,
      "learning_rate": 5.8666666666666675e-06,
      "loss": 0.2095,
      "step": 220
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8073865127216844,
      "learning_rate": 5.893333333333334e-06,
      "loss": 0.2227,
      "step": 221
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.024782578359343,
      "learning_rate": 5.92e-06,
      "loss": 0.2213,
      "step": 222
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9713612411625212,
      "learning_rate": 5.946666666666668e-06,
      "loss": 0.2345,
      "step": 223
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1359137855094446,
      "learning_rate": 5.973333333333334e-06,
      "loss": 0.1907,
      "step": 224
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9426893065862894,
      "learning_rate": 6e-06,
      "loss": 0.2361,
      "step": 225
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0072523715793786,
      "learning_rate": 6.026666666666668e-06,
      "loss": 0.2735,
      "step": 226
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8592934742515896,
      "learning_rate": 6.0533333333333335e-06,
      "loss": 0.211,
      "step": 227
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.548305143394406,
      "learning_rate": 6.08e-06,
      "loss": 0.8559,
      "step": 228
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.11879905460239,
      "learning_rate": 6.106666666666668e-06,
      "loss": 0.7902,
      "step": 229
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9948479210724899,
      "learning_rate": 6.133333333333334e-06,
      "loss": 0.2324,
      "step": 230
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.221209470421269,
      "learning_rate": 6.16e-06,
      "loss": 0.2747,
      "step": 231
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9617241424388598,
      "learning_rate": 6.186666666666668e-06,
      "loss": 0.2064,
      "step": 232
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9546682934109598,
      "learning_rate": 6.213333333333334e-06,
      "loss": 0.2083,
      "step": 233
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8838853737186432,
      "learning_rate": 6.24e-06,
      "loss": 0.2207,
      "step": 234
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.757184860971195,
      "learning_rate": 6.266666666666668e-06,
      "loss": 0.1955,
      "step": 235
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8738247110019026,
      "learning_rate": 6.293333333333334e-06,
      "loss": 0.192,
      "step": 236
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.330828649974647,
      "learning_rate": 6.3200000000000005e-06,
      "loss": 0.5612,
      "step": 237
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8771857797349003,
      "learning_rate": 6.346666666666668e-06,
      "loss": 0.2176,
      "step": 238
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8658177829772218,
      "learning_rate": 6.373333333333334e-06,
      "loss": 0.2408,
      "step": 239
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.862642133730281,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.1968,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8485072293724576,
      "learning_rate": 6.426666666666668e-06,
      "loss": 0.1823,
      "step": 241
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.008109455070311,
      "learning_rate": 6.453333333333334e-06,
      "loss": 0.2123,
      "step": 242
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9646557938344782,
      "learning_rate": 6.480000000000001e-06,
      "loss": 0.239,
      "step": 243
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.802054143981362,
      "learning_rate": 6.5066666666666665e-06,
      "loss": 0.2153,
      "step": 244
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.801969351436421,
      "learning_rate": 6.533333333333334e-06,
      "loss": 0.217,
      "step": 245
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.298119907018182,
      "learning_rate": 6.560000000000001e-06,
      "loss": 0.7515,
      "step": 246
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7330655635007703,
      "learning_rate": 6.5866666666666666e-06,
      "loss": 0.1975,
      "step": 247
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0251944467161525,
      "learning_rate": 6.613333333333334e-06,
      "loss": 0.2445,
      "step": 248
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.794177856475254,
      "learning_rate": 6.640000000000001e-06,
      "loss": 0.2187,
      "step": 249
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0157810651568338,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.2493,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.953270411021334,
      "learning_rate": 6.693333333333334e-06,
      "loss": 0.202,
      "step": 251
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.136550048049724,
      "learning_rate": 6.720000000000001e-06,
      "loss": 0.2787,
      "step": 252
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8213446521605676,
      "learning_rate": 6.746666666666667e-06,
      "loss": 0.211,
      "step": 253
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.094535836932365,
      "learning_rate": 6.773333333333334e-06,
      "loss": 0.1939,
      "step": 254
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9131906863440633,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.2235,
      "step": 255
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.409854210199455,
      "learning_rate": 6.826666666666667e-06,
      "loss": 0.8275,
      "step": 256
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8230651680675103,
      "learning_rate": 6.853333333333334e-06,
      "loss": 0.1901,
      "step": 257
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8762137776791508,
      "learning_rate": 6.88e-06,
      "loss": 0.2608,
      "step": 258
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9068957983795443,
      "learning_rate": 6.906666666666667e-06,
      "loss": 0.2139,
      "step": 259
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8187862767783025,
      "learning_rate": 6.9333333333333344e-06,
      "loss": 0.2404,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8098183193892265,
      "learning_rate": 6.96e-06,
      "loss": 0.2145,
      "step": 261
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8439694892995122,
      "learning_rate": 6.986666666666667e-06,
      "loss": 0.2076,
      "step": 262
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7493929656056368,
      "learning_rate": 7.0133333333333345e-06,
      "loss": 0.2226,
      "step": 263
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8055290320505561,
      "learning_rate": 7.04e-06,
      "loss": 0.2188,
      "step": 264
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.780493758788641,
      "learning_rate": 7.066666666666667e-06,
      "loss": 0.2343,
      "step": 265
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9117221859319986,
      "learning_rate": 7.093333333333335e-06,
      "loss": 0.2715,
      "step": 266
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8683689548762643,
      "learning_rate": 7.1200000000000004e-06,
      "loss": 0.2264,
      "step": 267
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.124299529327855,
      "learning_rate": 7.146666666666667e-06,
      "loss": 0.2838,
      "step": 268
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9898439974394204,
      "learning_rate": 7.173333333333335e-06,
      "loss": 0.2614,
      "step": 269
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8557654889796444,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.2487,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.746708978725774,
      "learning_rate": 7.226666666666667e-06,
      "loss": 0.2112,
      "step": 271
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.78569883174777,
      "learning_rate": 7.253333333333335e-06,
      "loss": 0.3946,
      "step": 272
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.477661416396561,
      "learning_rate": 7.280000000000001e-06,
      "loss": 0.5882,
      "step": 273
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8556111067049568,
      "learning_rate": 7.306666666666667e-06,
      "loss": 0.1944,
      "step": 274
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.883219434344051,
      "learning_rate": 7.333333333333333e-06,
      "loss": 0.1615,
      "step": 275
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.801070208917459,
      "learning_rate": 7.360000000000001e-06,
      "loss": 0.2263,
      "step": 276
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.806011893395799,
      "learning_rate": 7.386666666666667e-06,
      "loss": 0.1902,
      "step": 277
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9652604726727818,
      "learning_rate": 7.413333333333333e-06,
      "loss": 0.2641,
      "step": 278
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9322580861382184,
      "learning_rate": 7.440000000000001e-06,
      "loss": 0.189,
      "step": 279
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9507311818604411,
      "learning_rate": 7.4666666666666675e-06,
      "loss": 0.1964,
      "step": 280
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.604747701390727,
      "learning_rate": 7.493333333333333e-06,
      "loss": 0.7259,
      "step": 281
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.612562328878434,
      "learning_rate": 7.520000000000001e-06,
      "loss": 0.6476,
      "step": 282
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.596977838357216,
      "learning_rate": 7.5466666666666675e-06,
      "loss": 0.2662,
      "step": 283
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9001729147014241,
      "learning_rate": 7.573333333333333e-06,
      "loss": 0.2534,
      "step": 284
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.844177418989628,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.2497,
      "step": 285
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5932126387816608,
      "learning_rate": 7.626666666666668e-06,
      "loss": 0.1896,
      "step": 286
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.170754685838896,
      "learning_rate": 7.653333333333333e-06,
      "loss": 0.2431,
      "step": 287
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.826138504100035,
      "learning_rate": 7.680000000000001e-06,
      "loss": 0.1937,
      "step": 288
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9202338734227242,
      "learning_rate": 7.706666666666669e-06,
      "loss": 0.2727,
      "step": 289
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.920998957025096,
      "learning_rate": 7.733333333333334e-06,
      "loss": 0.7006,
      "step": 290
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8418945575301588,
      "learning_rate": 7.76e-06,
      "loss": 0.2167,
      "step": 291
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.673045546958607,
      "learning_rate": 7.786666666666666e-06,
      "loss": 0.2321,
      "step": 292
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8366811114957107,
      "learning_rate": 7.813333333333334e-06,
      "loss": 0.1899,
      "step": 293
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8869371609251013,
      "learning_rate": 7.840000000000001e-06,
      "loss": 0.259,
      "step": 294
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8667091217240646,
      "learning_rate": 7.866666666666667e-06,
      "loss": 0.2247,
      "step": 295
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5730756464467448,
      "learning_rate": 7.893333333333335e-06,
      "loss": 0.2151,
      "step": 296
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0657362534847246,
      "learning_rate": 7.92e-06,
      "loss": 0.3143,
      "step": 297
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1407957139153857,
      "learning_rate": 7.946666666666666e-06,
      "loss": 0.2284,
      "step": 298
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.549824490588346,
      "learning_rate": 7.973333333333334e-06,
      "loss": 0.1499,
      "step": 299
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.6912225591495655,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.7355,
      "step": 300
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9068102090093282,
      "learning_rate": 8.026666666666667e-06,
      "loss": 0.2488,
      "step": 301
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8057644158054502,
      "learning_rate": 8.053333333333335e-06,
      "loss": 0.2296,
      "step": 302
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.026373947332388,
      "learning_rate": 8.08e-06,
      "loss": 0.597,
      "step": 303
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.787899072762206,
      "learning_rate": 8.106666666666666e-06,
      "loss": 0.2073,
      "step": 304
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6580375831203185,
      "learning_rate": 8.133333333333334e-06,
      "loss": 0.2289,
      "step": 305
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7894920400473349,
      "learning_rate": 8.16e-06,
      "loss": 0.1785,
      "step": 306
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.198390276773755,
      "learning_rate": 8.186666666666667e-06,
      "loss": 0.6606,
      "step": 307
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6930201576413186,
      "learning_rate": 8.213333333333335e-06,
      "loss": 0.2181,
      "step": 308
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.019897622083253,
      "learning_rate": 8.24e-06,
      "loss": 0.2236,
      "step": 309
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7334906676332429,
      "learning_rate": 8.266666666666667e-06,
      "loss": 0.2353,
      "step": 310
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8572114438308505,
      "learning_rate": 8.293333333333334e-06,
      "loss": 0.2772,
      "step": 311
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9469342984931026,
      "learning_rate": 8.32e-06,
      "loss": 0.2296,
      "step": 312
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9630239498130868,
      "learning_rate": 8.346666666666668e-06,
      "loss": 0.262,
      "step": 313
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.27470813972205,
      "learning_rate": 8.373333333333335e-06,
      "loss": 0.5589,
      "step": 314
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0608869422345855,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.2098,
      "step": 315
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.917682143653835,
      "learning_rate": 8.426666666666667e-06,
      "loss": 0.596,
      "step": 316
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7643571133847522,
      "learning_rate": 8.453333333333334e-06,
      "loss": 0.2344,
      "step": 317
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7957129154630194,
      "learning_rate": 8.48e-06,
      "loss": 0.2214,
      "step": 318
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.028080745862539,
      "learning_rate": 8.506666666666668e-06,
      "loss": 0.2543,
      "step": 319
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.778878662216473,
      "learning_rate": 8.533333333333335e-06,
      "loss": 0.2381,
      "step": 320
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7940192538414392,
      "learning_rate": 8.560000000000001e-06,
      "loss": 0.2437,
      "step": 321
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.751415208191758,
      "learning_rate": 8.586666666666667e-06,
      "loss": 0.2373,
      "step": 322
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7286912456876244,
      "learning_rate": 8.613333333333333e-06,
      "loss": 0.1959,
      "step": 323
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.4695403527963204,
      "learning_rate": 8.64e-06,
      "loss": 0.7221,
      "step": 324
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6958831748028036,
      "learning_rate": 8.666666666666668e-06,
      "loss": 0.1976,
      "step": 325
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8568383992999282,
      "learning_rate": 8.693333333333334e-06,
      "loss": 0.2119,
      "step": 326
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7775066097406733,
      "learning_rate": 8.720000000000001e-06,
      "loss": 0.2459,
      "step": 327
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9796336175492852,
      "learning_rate": 8.746666666666667e-06,
      "loss": 0.228,
      "step": 328
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.725337679038237,
      "learning_rate": 8.773333333333333e-06,
      "loss": 0.2803,
      "step": 329
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8607473362530842,
      "learning_rate": 8.8e-06,
      "loss": 0.23,
      "step": 330
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7873655100663122,
      "learning_rate": 8.826666666666668e-06,
      "loss": 0.2475,
      "step": 331
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9470752103117503,
      "learning_rate": 8.853333333333334e-06,
      "loss": 0.2079,
      "step": 332
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9077336629858086,
      "learning_rate": 8.880000000000001e-06,
      "loss": 0.2496,
      "step": 333
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.865253714411325,
      "learning_rate": 8.906666666666667e-06,
      "loss": 0.202,
      "step": 334
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8483104452799306,
      "learning_rate": 8.933333333333333e-06,
      "loss": 0.213,
      "step": 335
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.655696928032451,
      "learning_rate": 8.96e-06,
      "loss": 0.6472,
      "step": 336
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6530692292567741,
      "learning_rate": 8.986666666666666e-06,
      "loss": 0.2225,
      "step": 337
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0226672513208204,
      "learning_rate": 9.013333333333334e-06,
      "loss": 0.227,
      "step": 338
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.738470394388767,
      "learning_rate": 9.040000000000002e-06,
      "loss": 0.2146,
      "step": 339
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.626623339712544,
      "learning_rate": 9.066666666666667e-06,
      "loss": 0.1957,
      "step": 340
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.744696499076202,
      "learning_rate": 9.093333333333333e-06,
      "loss": 0.2002,
      "step": 341
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7051863913858039,
      "learning_rate": 9.12e-06,
      "loss": 0.21,
      "step": 342
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.686387549981206,
      "learning_rate": 9.146666666666667e-06,
      "loss": 0.2006,
      "step": 343
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7793543082196661,
      "learning_rate": 9.173333333333334e-06,
      "loss": 0.2408,
      "step": 344
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7479798580717922,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.2642,
      "step": 345
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7614687271317424,
      "learning_rate": 9.226666666666668e-06,
      "loss": 0.2313,
      "step": 346
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.722511361895267,
      "learning_rate": 9.253333333333333e-06,
      "loss": 0.2102,
      "step": 347
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.161835584502886,
      "learning_rate": 9.280000000000001e-06,
      "loss": 0.2899,
      "step": 348
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.64586602691352,
      "learning_rate": 9.306666666666667e-06,
      "loss": 0.7191,
      "step": 349
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.64111257548122,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.6482,
      "step": 350
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.42583444153681,
      "learning_rate": 9.360000000000002e-06,
      "loss": 0.7959,
      "step": 351
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5424720013071538,
      "learning_rate": 9.386666666666668e-06,
      "loss": 0.1429,
      "step": 352
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.982575233322272,
      "learning_rate": 9.413333333333334e-06,
      "loss": 0.2918,
      "step": 353
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8336683173813957,
      "learning_rate": 9.440000000000001e-06,
      "loss": 0.2107,
      "step": 354
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9179224490424405,
      "learning_rate": 9.466666666666667e-06,
      "loss": 0.2576,
      "step": 355
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7444807373179754,
      "learning_rate": 9.493333333333334e-06,
      "loss": 0.244,
      "step": 356
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8960163999495967,
      "learning_rate": 9.52e-06,
      "loss": 0.246,
      "step": 357
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8008657169690534,
      "learning_rate": 9.546666666666668e-06,
      "loss": 0.2349,
      "step": 358
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.810225823418963,
      "learning_rate": 9.573333333333334e-06,
      "loss": 0.2366,
      "step": 359
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.047361957077407,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.3164,
      "step": 360
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8823083558049156,
      "learning_rate": 9.626666666666667e-06,
      "loss": 0.2385,
      "step": 361
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.698176588727454,
      "learning_rate": 9.653333333333335e-06,
      "loss": 0.1787,
      "step": 362
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8160965938479063,
      "learning_rate": 9.68e-06,
      "loss": 0.2242,
      "step": 363
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9437981118956582,
      "learning_rate": 9.706666666666668e-06,
      "loss": 0.2605,
      "step": 364
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.373439759697795,
      "learning_rate": 9.733333333333334e-06,
      "loss": 0.6327,
      "step": 365
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0998902475697885,
      "learning_rate": 9.760000000000001e-06,
      "loss": 0.2929,
      "step": 366
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8744366599560585,
      "learning_rate": 9.786666666666667e-06,
      "loss": 0.2381,
      "step": 367
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7793901248003852,
      "learning_rate": 9.813333333333333e-06,
      "loss": 0.246,
      "step": 368
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6586529404544306,
      "learning_rate": 9.84e-06,
      "loss": 0.206,
      "step": 369
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9018004543174982,
      "learning_rate": 9.866666666666668e-06,
      "loss": 0.2365,
      "step": 370
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7106291778939413,
      "learning_rate": 9.893333333333334e-06,
      "loss": 0.2469,
      "step": 371
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9474151303100837,
      "learning_rate": 9.920000000000002e-06,
      "loss": 0.3033,
      "step": 372
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.742825501117863,
      "learning_rate": 9.946666666666667e-06,
      "loss": 0.2719,
      "step": 373
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.649110246787641,
      "learning_rate": 9.973333333333333e-06,
      "loss": 0.2706,
      "step": 374
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.433233054289498,
      "learning_rate": 1e-05,
      "loss": 0.7441,
      "step": 375
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7598126079007446,
      "learning_rate": 9.999999832167426e-06,
      "loss": 0.2421,
      "step": 376
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7327741927654245,
      "learning_rate": 9.999999328669713e-06,
      "loss": 0.2139,
      "step": 377
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7952826167116798,
      "learning_rate": 9.999998489506897e-06,
      "loss": 0.2542,
      "step": 378
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9112956335206404,
      "learning_rate": 9.999997314679031e-06,
      "loss": 0.2414,
      "step": 379
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8893839347164407,
      "learning_rate": 9.999995804186196e-06,
      "loss": 0.2711,
      "step": 380
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.693669030375468,
      "learning_rate": 9.999993958028495e-06,
      "loss": 0.239,
      "step": 381
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6549425714155548,
      "learning_rate": 9.999991776206049e-06,
      "loss": 0.1948,
      "step": 382
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2597209046667817,
      "learning_rate": 9.999989258719005e-06,
      "loss": 0.2375,
      "step": 383
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.87451040285667,
      "learning_rate": 9.999986405567535e-06,
      "loss": 0.7829,
      "step": 384
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0777043865702987,
      "learning_rate": 9.999983216751826e-06,
      "loss": 0.2943,
      "step": 385
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.697246082818542,
      "learning_rate": 9.999979692272095e-06,
      "loss": 0.2378,
      "step": 386
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8307078193666357,
      "learning_rate": 9.999975832128578e-06,
      "loss": 0.2814,
      "step": 387
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8804460518214392,
      "learning_rate": 9.999971636321535e-06,
      "loss": 0.2378,
      "step": 388
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9249249089304041,
      "learning_rate": 9.999967104851244e-06,
      "loss": 0.2693,
      "step": 389
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.959547312703074,
      "learning_rate": 9.999962237718015e-06,
      "loss": 0.2515,
      "step": 390
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8687320589379146,
      "learning_rate": 9.99995703492217e-06,
      "loss": 0.2584,
      "step": 391
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7066756693145695,
      "learning_rate": 9.999951496464062e-06,
      "loss": 0.1908,
      "step": 392
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7529062052157671,
      "learning_rate": 9.999945622344058e-06,
      "loss": 0.2449,
      "step": 393
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.411631790011608,
      "learning_rate": 9.999939412562558e-06,
      "loss": 0.7327,
      "step": 394
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9015666314960635,
      "learning_rate": 9.999932867119974e-06,
      "loss": 0.2759,
      "step": 395
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7322147151979983,
      "learning_rate": 9.999925986016748e-06,
      "loss": 0.2217,
      "step": 396
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7866095729204259,
      "learning_rate": 9.99991876925334e-06,
      "loss": 0.2588,
      "step": 397
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.174368380749566,
      "learning_rate": 9.999911216830239e-06,
      "loss": 0.4313,
      "step": 398
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5310984726495105,
      "learning_rate": 9.999903328747946e-06,
      "loss": 0.1944,
      "step": 399
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8519706513959513,
      "learning_rate": 9.999895105006995e-06,
      "loss": 0.2823,
      "step": 400
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6637054641929983,
      "learning_rate": 9.999886545607935e-06,
      "loss": 0.1717,
      "step": 401
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9952085032243967,
      "learning_rate": 9.999877650551344e-06,
      "loss": 0.2397,
      "step": 402
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8545572787682478,
      "learning_rate": 9.999868419837818e-06,
      "loss": 0.3225,
      "step": 403
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8104336215147734,
      "learning_rate": 9.999858853467972e-06,
      "loss": 0.2675,
      "step": 404
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7762845295831875,
      "learning_rate": 9.999848951442455e-06,
      "loss": 0.2207,
      "step": 405
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.797172007569505,
      "learning_rate": 9.99983871376193e-06,
      "loss": 0.2024,
      "step": 406
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.615293018085328,
      "learning_rate": 9.999828140427082e-06,
      "loss": 0.2355,
      "step": 407
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8968112208105452,
      "learning_rate": 9.99981723143862e-06,
      "loss": 0.2597,
      "step": 408
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.754145304264831,
      "learning_rate": 9.99980598679728e-06,
      "loss": 0.2534,
      "step": 409
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.007116910841986,
      "learning_rate": 9.999794406503816e-06,
      "loss": 0.2539,
      "step": 410
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9583626850834912,
      "learning_rate": 9.999782490559004e-06,
      "loss": 0.2394,
      "step": 411
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.817502406871994,
      "learning_rate": 9.999770238963646e-06,
      "loss": 0.269,
      "step": 412
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.796483917544564,
      "learning_rate": 9.999757651718562e-06,
      "loss": 0.276,
      "step": 413
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7443361297939857,
      "learning_rate": 9.999744728824599e-06,
      "loss": 0.2522,
      "step": 414
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9119833716967742,
      "learning_rate": 9.999731470282621e-06,
      "loss": 0.2566,
      "step": 415
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.604759730691754,
      "learning_rate": 9.999717876093525e-06,
      "loss": 0.7022,
      "step": 416
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9478155159981574,
      "learning_rate": 9.999703946258217e-06,
      "loss": 0.2637,
      "step": 417
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8742409053779854,
      "learning_rate": 9.999689680777634e-06,
      "loss": 0.282,
      "step": 418
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.936408504271645,
      "learning_rate": 9.999675079652736e-06,
      "loss": 0.2429,
      "step": 419
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.508756672582905,
      "learning_rate": 9.9996601428845e-06,
      "loss": 0.2627,
      "step": 420
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6844374620864568,
      "learning_rate": 9.99964487047393e-06,
      "loss": 0.2613,
      "step": 421
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.587888438073988,
      "learning_rate": 9.999629262422053e-06,
      "loss": 0.2277,
      "step": 422
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7669357863217947,
      "learning_rate": 9.999613318729915e-06,
      "loss": 0.2395,
      "step": 423
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.272726419813807,
      "learning_rate": 9.999597039398586e-06,
      "loss": 0.5656,
      "step": 424
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5784608076294535,
      "learning_rate": 9.99958042442916e-06,
      "loss": 0.2131,
      "step": 425
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.620535365064519,
      "learning_rate": 9.999563473822752e-06,
      "loss": 0.2197,
      "step": 426
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.817314467358351,
      "learning_rate": 9.9995461875805e-06,
      "loss": 0.3054,
      "step": 427
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.814826032540398,
      "learning_rate": 9.999528565703564e-06,
      "loss": 0.2649,
      "step": 428
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7801458756492414,
      "learning_rate": 9.999510608193128e-06,
      "loss": 0.2512,
      "step": 429
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7476056654752405,
      "learning_rate": 9.999492315050396e-06,
      "loss": 0.2738,
      "step": 430
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6010235466810252,
      "learning_rate": 9.999473686276598e-06,
      "loss": 0.2201,
      "step": 431
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8536584612725224,
      "learning_rate": 9.999454721872983e-06,
      "loss": 0.2709,
      "step": 432
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8135164102945176,
      "learning_rate": 9.999435421840826e-06,
      "loss": 0.2487,
      "step": 433
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5818417174610364,
      "learning_rate": 9.99941578618142e-06,
      "loss": 0.2141,
      "step": 434
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6005212701306177,
      "learning_rate": 9.999395814896086e-06,
      "loss": 0.1943,
      "step": 435
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5933879667459743,
      "learning_rate": 9.999375507986163e-06,
      "loss": 0.2374,
      "step": 436
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.688048880829338,
      "learning_rate": 9.999354865453012e-06,
      "loss": 0.2335,
      "step": 437
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.187328425483692,
      "learning_rate": 9.999333887298025e-06,
      "loss": 0.8429,
      "step": 438
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7960781284703393,
      "learning_rate": 9.999312573522606e-06,
      "loss": 0.2388,
      "step": 439
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7385905500174983,
      "learning_rate": 9.999290924128186e-06,
      "loss": 0.3019,
      "step": 440
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6675120324319856,
      "learning_rate": 9.999268939116218e-06,
      "loss": 0.2551,
      "step": 441
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7654656204783086,
      "learning_rate": 9.999246618488181e-06,
      "loss": 0.2163,
      "step": 442
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6330572866481377,
      "learning_rate": 9.99922396224557e-06,
      "loss": 0.1949,
      "step": 443
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6030756243071447,
      "learning_rate": 9.999200970389909e-06,
      "loss": 0.2331,
      "step": 444
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8210510700229596,
      "learning_rate": 9.999177642922736e-06,
      "loss": 0.295,
      "step": 445
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7002742660671717,
      "learning_rate": 9.999153979845625e-06,
      "loss": 0.2716,
      "step": 446
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.807176114965877,
      "learning_rate": 9.999129981160159e-06,
      "loss": 0.2146,
      "step": 447
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5681821768850328,
      "learning_rate": 9.99910564686795e-06,
      "loss": 0.2238,
      "step": 448
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7555130945026285,
      "learning_rate": 9.999080976970635e-06,
      "loss": 0.2364,
      "step": 449
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8856066109241383,
      "learning_rate": 9.999055971469864e-06,
      "loss": 0.2595,
      "step": 450
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7074668763472476,
      "learning_rate": 9.99903063036732e-06,
      "loss": 0.2229,
      "step": 451
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7397396556855906,
      "learning_rate": 9.999004953664703e-06,
      "loss": 0.2335,
      "step": 452
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7253941594552007,
      "learning_rate": 9.998978941363739e-06,
      "loss": 0.2516,
      "step": 453
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.367070400252453,
      "learning_rate": 9.998952593466171e-06,
      "loss": 0.1554,
      "step": 454
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.777433265413999,
      "learning_rate": 9.998925909973769e-06,
      "loss": 0.6771,
      "step": 455
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.837246019873197,
      "learning_rate": 9.998898890888325e-06,
      "loss": 0.2477,
      "step": 456
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6072135157782246,
      "learning_rate": 9.998871536211652e-06,
      "loss": 0.2286,
      "step": 457
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6051591770866929,
      "learning_rate": 9.998843845945587e-06,
      "loss": 0.2172,
      "step": 458
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6027647348744865,
      "learning_rate": 9.99881582009199e-06,
      "loss": 0.2305,
      "step": 459
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.768135323642207,
      "learning_rate": 9.99878745865274e-06,
      "loss": 0.2705,
      "step": 460
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7930246213753742,
      "learning_rate": 9.99875876162974e-06,
      "loss": 0.2599,
      "step": 461
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7192238909009794,
      "learning_rate": 9.998729729024922e-06,
      "loss": 0.2626,
      "step": 462
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.692627846513932,
      "learning_rate": 9.998700360840231e-06,
      "loss": 0.2517,
      "step": 463
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.646238954943402,
      "learning_rate": 9.998670657077638e-06,
      "loss": 0.1984,
      "step": 464
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6968526535099162,
      "learning_rate": 9.99864061773914e-06,
      "loss": 0.2461,
      "step": 465
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5928989101111992,
      "learning_rate": 9.998610242826752e-06,
      "loss": 0.2378,
      "step": 466
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6016481726238563,
      "learning_rate": 9.998579532342511e-06,
      "loss": 0.2267,
      "step": 467
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.77298901119548,
      "learning_rate": 9.998548486288483e-06,
      "loss": 0.2291,
      "step": 468
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7132993670059347,
      "learning_rate": 9.998517104666749e-06,
      "loss": 0.2264,
      "step": 469
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6817500764056423,
      "learning_rate": 9.998485387479418e-06,
      "loss": 0.2113,
      "step": 470
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8988742098042801,
      "learning_rate": 9.998453334728619e-06,
      "loss": 0.2257,
      "step": 471
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6829839684411732,
      "learning_rate": 9.9984209464165e-06,
      "loss": 0.2189,
      "step": 472
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6160207290097985,
      "learning_rate": 9.998388222545242e-06,
      "loss": 0.2233,
      "step": 473
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6062583538820858,
      "learning_rate": 9.998355163117035e-06,
      "loss": 0.206,
      "step": 474
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.686815420738238,
      "learning_rate": 9.998321768134101e-06,
      "loss": 0.2472,
      "step": 475
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.664740784069062,
      "learning_rate": 9.998288037598684e-06,
      "loss": 0.2344,
      "step": 476
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6842869750212988,
      "learning_rate": 9.998253971513048e-06,
      "loss": 0.2393,
      "step": 477
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6891112601576237,
      "learning_rate": 9.998219569879476e-06,
      "loss": 0.2362,
      "step": 478
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7388089742044242,
      "learning_rate": 9.998184832700282e-06,
      "loss": 0.2438,
      "step": 479
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8036799202626073,
      "learning_rate": 9.998149759977795e-06,
      "loss": 0.2179,
      "step": 480
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7263253675044903,
      "learning_rate": 9.998114351714373e-06,
      "loss": 0.216,
      "step": 481
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9501042184813449,
      "learning_rate": 9.99807860791239e-06,
      "loss": 0.2944,
      "step": 482
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.501912712257805,
      "learning_rate": 9.998042528574246e-06,
      "loss": 0.1826,
      "step": 483
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.744971987478523,
      "learning_rate": 9.998006113702363e-06,
      "loss": 0.2325,
      "step": 484
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7658194754401142,
      "learning_rate": 9.997969363299187e-06,
      "loss": 0.2535,
      "step": 485
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6956324374083949,
      "learning_rate": 9.997932277367183e-06,
      "loss": 0.2549,
      "step": 486
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9675776314228486,
      "learning_rate": 9.997894855908844e-06,
      "loss": 0.2411,
      "step": 487
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.492592773883712,
      "learning_rate": 9.997857098926679e-06,
      "loss": 0.1714,
      "step": 488
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8597328856530386,
      "learning_rate": 9.997819006423227e-06,
      "loss": 0.2189,
      "step": 489
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6858733738383007,
      "learning_rate": 9.99778057840104e-06,
      "loss": 0.2047,
      "step": 490
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.659006252231804,
      "learning_rate": 9.9977418148627e-06,
      "loss": 0.2575,
      "step": 491
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5191581617735919,
      "learning_rate": 9.99770271581081e-06,
      "loss": 0.2221,
      "step": 492
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.566686974966257,
      "learning_rate": 9.997663281247993e-06,
      "loss": 0.2337,
      "step": 493
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7483886664121793,
      "learning_rate": 9.9976235111769e-06,
      "loss": 0.287,
      "step": 494
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5983677886961352,
      "learning_rate": 9.997583405600194e-06,
      "loss": 0.1951,
      "step": 495
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.602791792171724,
      "learning_rate": 9.997542964520576e-06,
      "loss": 0.2432,
      "step": 496
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5582802615572773,
      "learning_rate": 9.997502187940757e-06,
      "loss": 0.2039,
      "step": 497
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6463461419806718,
      "learning_rate": 9.997461075863473e-06,
      "loss": 0.2609,
      "step": 498
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9442539106674928,
      "learning_rate": 9.997419628291485e-06,
      "loss": 0.279,
      "step": 499
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4895204015877028,
      "learning_rate": 9.997377845227577e-06,
      "loss": 0.1967,
      "step": 500
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.635756892914474,
      "learning_rate": 9.99733572667455e-06,
      "loss": 0.2752,
      "step": 501
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7491036468801793,
      "learning_rate": 9.997293272635236e-06,
      "loss": 0.2402,
      "step": 502
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5469829330069942,
      "learning_rate": 9.997250483112483e-06,
      "loss": 0.1876,
      "step": 503
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6720312780346405,
      "learning_rate": 9.997207358109166e-06,
      "loss": 0.2323,
      "step": 504
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8177630203346855,
      "learning_rate": 9.997163897628175e-06,
      "loss": 0.2402,
      "step": 505
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.501670140224173,
      "learning_rate": 9.997120101672434e-06,
      "loss": 0.1858,
      "step": 506
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4159744247278743,
      "learning_rate": 9.997075970244878e-06,
      "loss": 0.2067,
      "step": 507
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7360277144501992,
      "learning_rate": 9.997031503348473e-06,
      "loss": 0.258,
      "step": 508
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6441702779488527,
      "learning_rate": 9.996986700986201e-06,
      "loss": 0.1734,
      "step": 509
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.728726117742014,
      "learning_rate": 9.996941563161071e-06,
      "loss": 0.6285,
      "step": 510
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6457050606364305,
      "learning_rate": 9.996896089876116e-06,
      "loss": 0.2351,
      "step": 511
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.6029027785714085,
      "learning_rate": 9.996850281134385e-06,
      "loss": 0.5682,
      "step": 512
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6916565173802096,
      "learning_rate": 9.996804136938956e-06,
      "loss": 0.2322,
      "step": 513
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.691599731192677,
      "learning_rate": 9.996757657292923e-06,
      "loss": 0.1974,
      "step": 514
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.659840187173813,
      "learning_rate": 9.996710842199412e-06,
      "loss": 0.2344,
      "step": 515
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6493918530999607,
      "learning_rate": 9.99666369166156e-06,
      "loss": 0.2352,
      "step": 516
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3729762300075468,
      "learning_rate": 9.996616205682538e-06,
      "loss": 0.2157,
      "step": 517
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.505073858359476,
      "learning_rate": 9.996568384265529e-06,
      "loss": 0.2292,
      "step": 518
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6759476137826335,
      "learning_rate": 9.996520227413747e-06,
      "loss": 0.2714,
      "step": 519
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9658492138473096,
      "learning_rate": 9.996471735130422e-06,
      "loss": 0.2911,
      "step": 520
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7163456216618151,
      "learning_rate": 9.99642290741881e-06,
      "loss": 0.236,
      "step": 521
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.764867585471799,
      "learning_rate": 9.99637374428219e-06,
      "loss": 0.2337,
      "step": 522
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6346477893821554,
      "learning_rate": 9.996324245723863e-06,
      "loss": 0.2375,
      "step": 523
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.558777251804133,
      "learning_rate": 9.99627441174715e-06,
      "loss": 0.2538,
      "step": 524
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6284371865490614,
      "learning_rate": 9.9962242423554e-06,
      "loss": 0.2217,
      "step": 525
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.832394420792578,
      "learning_rate": 9.996173737551976e-06,
      "loss": 0.2616,
      "step": 526
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6409982686137434,
      "learning_rate": 9.996122897340273e-06,
      "loss": 0.2353,
      "step": 527
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4367469841502498,
      "learning_rate": 9.9960717217237e-06,
      "loss": 0.2171,
      "step": 528
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5379249205662373,
      "learning_rate": 9.996020210705697e-06,
      "loss": 0.2215,
      "step": 529
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.658894243914053,
      "learning_rate": 9.995968364289719e-06,
      "loss": 0.5064,
      "step": 530
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.240879290750167,
      "learning_rate": 9.995916182479248e-06,
      "loss": 0.5841,
      "step": 531
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6287031206021532,
      "learning_rate": 9.995863665277787e-06,
      "loss": 0.2332,
      "step": 532
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.079833609184642,
      "learning_rate": 9.99581081268886e-06,
      "loss": 0.2507,
      "step": 533
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6297785420124187,
      "learning_rate": 9.995757624716019e-06,
      "loss": 0.2446,
      "step": 534
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4917721777721014,
      "learning_rate": 9.99570410136283e-06,
      "loss": 0.2636,
      "step": 535
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6260663646793445,
      "learning_rate": 9.995650242632887e-06,
      "loss": 0.2065,
      "step": 536
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6490972581637198,
      "learning_rate": 9.99559604852981e-06,
      "loss": 0.249,
      "step": 537
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.568733558786324,
      "learning_rate": 9.995541519057231e-06,
      "loss": 0.2165,
      "step": 538
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.59945261733457,
      "learning_rate": 9.995486654218815e-06,
      "loss": 0.2263,
      "step": 539
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.526253766435983,
      "learning_rate": 9.995431454018246e-06,
      "loss": 0.2131,
      "step": 540
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5989519470593818,
      "learning_rate": 9.995375918459227e-06,
      "loss": 0.2746,
      "step": 541
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7991778962413925,
      "learning_rate": 9.995320047545488e-06,
      "loss": 0.2969,
      "step": 542
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.633462776227588,
      "learning_rate": 9.995263841280776e-06,
      "loss": 0.2716,
      "step": 543
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8007509659264962,
      "learning_rate": 9.99520729966887e-06,
      "loss": 0.291,
      "step": 544
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7498919144157523,
      "learning_rate": 9.995150422713561e-06,
      "loss": 0.2152,
      "step": 545
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7115026792009813,
      "learning_rate": 9.995093210418672e-06,
      "loss": 0.2372,
      "step": 546
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.920929337976908,
      "learning_rate": 9.995035662788039e-06,
      "loss": 0.2256,
      "step": 547
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6368308690184818,
      "learning_rate": 9.99497777982553e-06,
      "loss": 0.2228,
      "step": 548
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.587645451987956,
      "learning_rate": 9.994919561535026e-06,
      "loss": 0.2231,
      "step": 549
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7628630883457435,
      "learning_rate": 9.99486100792044e-06,
      "loss": 0.2523,
      "step": 550
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.85604783302287,
      "learning_rate": 9.9948021189857e-06,
      "loss": 0.2487,
      "step": 551
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6386254313357684,
      "learning_rate": 9.994742894734759e-06,
      "loss": 0.2116,
      "step": 552
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6351585775425137,
      "learning_rate": 9.994683335171594e-06,
      "loss": 0.2206,
      "step": 553
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6512105214169943,
      "learning_rate": 9.994623440300205e-06,
      "loss": 0.2698,
      "step": 554
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6491951674575047,
      "learning_rate": 9.99456321012461e-06,
      "loss": 0.2166,
      "step": 555
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5276431680172764,
      "learning_rate": 9.994502644648854e-06,
      "loss": 0.2473,
      "step": 556
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7477318881062665,
      "learning_rate": 9.994441743877003e-06,
      "loss": 0.2155,
      "step": 557
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7290047662759773,
      "learning_rate": 9.994380507813146e-06,
      "loss": 0.265,
      "step": 558
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5903361943985859,
      "learning_rate": 9.994318936461393e-06,
      "loss": 0.2171,
      "step": 559
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7378313280195632,
      "learning_rate": 9.994257029825876e-06,
      "loss": 0.2195,
      "step": 560
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.363053125638119,
      "learning_rate": 9.994194787910754e-06,
      "loss": 0.5256,
      "step": 561
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7225026488301578,
      "learning_rate": 9.994132210720204e-06,
      "loss": 0.2306,
      "step": 562
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7367531611479357,
      "learning_rate": 9.994069298258427e-06,
      "loss": 0.2675,
      "step": 563
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8272753889086069,
      "learning_rate": 9.994006050529645e-06,
      "loss": 0.2763,
      "step": 564
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6823963619766302,
      "learning_rate": 9.993942467538107e-06,
      "loss": 0.2054,
      "step": 565
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.676880147327587,
      "learning_rate": 9.99387854928808e-06,
      "loss": 0.3209,
      "step": 566
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4441124262864924,
      "learning_rate": 9.993814295783855e-06,
      "loss": 0.1592,
      "step": 567
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6419816105732825,
      "learning_rate": 9.993749707029746e-06,
      "loss": 0.2401,
      "step": 568
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8621215355585214,
      "learning_rate": 9.99368478303009e-06,
      "loss": 0.2736,
      "step": 569
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4834496089438753,
      "learning_rate": 9.993619523789241e-06,
      "loss": 0.201,
      "step": 570
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.637398050899101,
      "learning_rate": 9.993553929311587e-06,
      "loss": 0.6305,
      "step": 571
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.9423650310906915,
      "learning_rate": 9.993487999601522e-06,
      "loss": 0.5231,
      "step": 572
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4707362256197447,
      "learning_rate": 9.993421734663484e-06,
      "loss": 0.2245,
      "step": 573
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5734565332666623,
      "learning_rate": 9.993355134501914e-06,
      "loss": 0.2185,
      "step": 574
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8655602777043687,
      "learning_rate": 9.993288199121283e-06,
      "loss": 0.2637,
      "step": 575
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.541634910203471,
      "learning_rate": 9.993220928526086e-06,
      "loss": 0.2297,
      "step": 576
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.218874888642801,
      "learning_rate": 9.993153322720841e-06,
      "loss": 0.7584,
      "step": 577
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6847881833008065,
      "learning_rate": 9.993085381710083e-06,
      "loss": 0.2531,
      "step": 578
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.648901477630294,
      "learning_rate": 9.993017105498378e-06,
      "loss": 0.2649,
      "step": 579
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.812625406157889,
      "learning_rate": 9.992948494090303e-06,
      "loss": 0.2152,
      "step": 580
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6954501957093269,
      "learning_rate": 9.992879547490469e-06,
      "loss": 0.2617,
      "step": 581
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.721341560188178,
      "learning_rate": 9.992810265703503e-06,
      "loss": 0.2604,
      "step": 582
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.84219526134802,
      "learning_rate": 9.992740648734057e-06,
      "loss": 0.2507,
      "step": 583
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5922860407384904,
      "learning_rate": 9.992670696586802e-06,
      "loss": 0.21,
      "step": 584
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.939742249964271,
      "learning_rate": 9.992600409266437e-06,
      "loss": 0.3023,
      "step": 585
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6440231924238256,
      "learning_rate": 9.99252978677768e-06,
      "loss": 0.2561,
      "step": 586
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.510332088821589,
      "learning_rate": 9.992458829125271e-06,
      "loss": 0.2418,
      "step": 587
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.615846261148202,
      "learning_rate": 9.992387536313975e-06,
      "loss": 0.2105,
      "step": 588
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6226165591908892,
      "learning_rate": 9.992315908348578e-06,
      "loss": 0.2219,
      "step": 589
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.716925910548682,
      "learning_rate": 9.992243945233886e-06,
      "loss": 0.8111,
      "step": 590
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4470812954837977,
      "learning_rate": 9.992171646974734e-06,
      "loss": 0.2093,
      "step": 591
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9229297576856135,
      "learning_rate": 9.99209901357597e-06,
      "loss": 0.2341,
      "step": 592
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6953907196044202,
      "learning_rate": 9.992026045042478e-06,
      "loss": 0.2448,
      "step": 593
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5659509369156763,
      "learning_rate": 9.99195274137915e-06,
      "loss": 0.2571,
      "step": 594
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5864828262557804,
      "learning_rate": 9.991879102590912e-06,
      "loss": 0.2444,
      "step": 595
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6379102039520295,
      "learning_rate": 9.9918051286827e-06,
      "loss": 0.2355,
      "step": 596
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7257890430894567,
      "learning_rate": 9.99173081965949e-06,
      "loss": 0.2721,
      "step": 597
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7318963219098946,
      "learning_rate": 9.991656175526264e-06,
      "loss": 0.2847,
      "step": 598
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5231759153164557,
      "learning_rate": 9.991581196288035e-06,
      "loss": 0.2373,
      "step": 599
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4847070433892302,
      "learning_rate": 9.991505881949837e-06,
      "loss": 0.2098,
      "step": 600
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6001821953972448,
      "learning_rate": 9.991430232516725e-06,
      "loss": 0.2182,
      "step": 601
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.802972270465234,
      "learning_rate": 9.991354247993776e-06,
      "loss": 0.2574,
      "step": 602
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.74547706467968,
      "learning_rate": 9.991277928386095e-06,
      "loss": 0.2135,
      "step": 603
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6257433356625492,
      "learning_rate": 9.991201273698805e-06,
      "loss": 0.2207,
      "step": 604
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.589040562884,
      "learning_rate": 9.991124283937049e-06,
      "loss": 0.2356,
      "step": 605
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5511165109105016,
      "learning_rate": 9.991046959105998e-06,
      "loss": 0.2341,
      "step": 606
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.011595514303622,
      "learning_rate": 9.990969299210843e-06,
      "loss": 0.5676,
      "step": 607
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6447541800466938,
      "learning_rate": 9.990891304256796e-06,
      "loss": 0.2691,
      "step": 608
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5651861030216716,
      "learning_rate": 9.990812974249094e-06,
      "loss": 0.2191,
      "step": 609
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7103707894771236,
      "learning_rate": 9.990734309192995e-06,
      "loss": 0.2309,
      "step": 610
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.395402475841665,
      "learning_rate": 9.99065530909378e-06,
      "loss": 0.2385,
      "step": 611
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5322543278660479,
      "learning_rate": 9.990575973956754e-06,
      "loss": 0.2451,
      "step": 612
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8074170798515372,
      "learning_rate": 9.990496303787243e-06,
      "loss": 0.2867,
      "step": 613
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7953475087350166,
      "learning_rate": 9.990416298590593e-06,
      "loss": 0.249,
      "step": 614
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7398426315939088,
      "learning_rate": 9.990335958372178e-06,
      "loss": 0.2566,
      "step": 615
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.694536819583506,
      "learning_rate": 9.990255283137388e-06,
      "loss": 0.2401,
      "step": 616
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.57965968100311,
      "learning_rate": 9.990174272891642e-06,
      "loss": 0.2128,
      "step": 617
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8619530161861824,
      "learning_rate": 9.990092927640378e-06,
      "loss": 0.2363,
      "step": 618
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7732123527437331,
      "learning_rate": 9.990011247389055e-06,
      "loss": 0.2123,
      "step": 619
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.680198178950124,
      "learning_rate": 9.989929232143159e-06,
      "loss": 0.2809,
      "step": 620
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4884085087328633,
      "learning_rate": 9.989846881908194e-06,
      "loss": 0.2391,
      "step": 621
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4396744748564938,
      "learning_rate": 9.98976419668969e-06,
      "loss": 0.2096,
      "step": 622
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5968857523302402,
      "learning_rate": 9.989681176493197e-06,
      "loss": 0.2536,
      "step": 623
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.615719943427046,
      "learning_rate": 9.989597821324288e-06,
      "loss": 0.7034,
      "step": 624
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.614385891347682,
      "learning_rate": 9.98951413118856e-06,
      "loss": 0.2293,
      "step": 625
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.191075367956811,
      "learning_rate": 9.989430106091629e-06,
      "loss": 0.612,
      "step": 626
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.603659924226727,
      "learning_rate": 9.989345746039138e-06,
      "loss": 0.2174,
      "step": 627
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6804506211375603,
      "learning_rate": 9.989261051036752e-06,
      "loss": 0.2836,
      "step": 628
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6564991052749514,
      "learning_rate": 9.989176021090155e-06,
      "loss": 0.2338,
      "step": 629
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.62722867667748,
      "learning_rate": 9.989090656205052e-06,
      "loss": 0.2266,
      "step": 630
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4852472228714626,
      "learning_rate": 9.989004956387179e-06,
      "loss": 0.2264,
      "step": 631
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6813906031094794,
      "learning_rate": 9.988918921642287e-06,
      "loss": 0.2327,
      "step": 632
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.808867267957172,
      "learning_rate": 9.988832551976151e-06,
      "loss": 0.2772,
      "step": 633
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2115923601291123,
      "learning_rate": 9.988745847394572e-06,
      "loss": 0.2219,
      "step": 634
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6525129481359726,
      "learning_rate": 9.988658807903369e-06,
      "loss": 0.2703,
      "step": 635
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6657086181445584,
      "learning_rate": 9.988571433508383e-06,
      "loss": 0.2713,
      "step": 636
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5380327534887346,
      "learning_rate": 9.988483724215483e-06,
      "loss": 0.188,
      "step": 637
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4789487314289715,
      "learning_rate": 9.988395680030556e-06,
      "loss": 0.2108,
      "step": 638
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.535315629330471,
      "learning_rate": 9.988307300959513e-06,
      "loss": 0.216,
      "step": 639
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5164162701768915,
      "learning_rate": 9.988218587008287e-06,
      "loss": 0.239,
      "step": 640
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6307974968305041,
      "learning_rate": 9.988129538182833e-06,
      "loss": 0.2109,
      "step": 641
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.765092766661041,
      "learning_rate": 9.98804015448913e-06,
      "loss": 0.2883,
      "step": 642
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4852765359981503,
      "learning_rate": 9.987950435933179e-06,
      "loss": 0.2209,
      "step": 643
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3738595567429717,
      "learning_rate": 9.987860382521003e-06,
      "loss": 0.1841,
      "step": 644
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7597230144739753,
      "learning_rate": 9.987769994258645e-06,
      "loss": 0.2699,
      "step": 645
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6374916651117752,
      "learning_rate": 9.987679271152175e-06,
      "loss": 0.291,
      "step": 646
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4744783927154217,
      "learning_rate": 9.987588213207684e-06,
      "loss": 0.2349,
      "step": 647
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.604462093974148,
      "learning_rate": 9.987496820431284e-06,
      "loss": 0.257,
      "step": 648
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.17027017060943,
      "learning_rate": 9.987405092829113e-06,
      "loss": 0.7866,
      "step": 649
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6545224837022368,
      "learning_rate": 9.987313030407325e-06,
      "loss": 0.2556,
      "step": 650
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.770167038492094,
      "learning_rate": 9.987220633172101e-06,
      "loss": 0.5711,
      "step": 651
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4945050815715162,
      "learning_rate": 9.987127901129647e-06,
      "loss": 0.1835,
      "step": 652
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5209064188816457,
      "learning_rate": 9.987034834286186e-06,
      "loss": 0.2482,
      "step": 653
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.574506628240721,
      "learning_rate": 9.986941432647968e-06,
      "loss": 0.1818,
      "step": 654
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7695495539151342,
      "learning_rate": 9.98684769622126e-06,
      "loss": 0.2404,
      "step": 655
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8528949420014384,
      "learning_rate": 9.986753625012358e-06,
      "loss": 0.2419,
      "step": 656
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9018601086016529,
      "learning_rate": 9.986659219027575e-06,
      "loss": 0.2273,
      "step": 657
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4649290227741514,
      "learning_rate": 9.986564478273249e-06,
      "loss": 0.2356,
      "step": 658
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5766365636580522,
      "learning_rate": 9.986469402755742e-06,
      "loss": 0.2483,
      "step": 659
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5410862867976602,
      "learning_rate": 9.986373992481434e-06,
      "loss": 0.2093,
      "step": 660
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5896560643961246,
      "learning_rate": 9.986278247456735e-06,
      "loss": 0.2081,
      "step": 661
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5720182589287917,
      "learning_rate": 9.986182167688066e-06,
      "loss": 0.2292,
      "step": 662
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7676532617931748,
      "learning_rate": 9.986085753181883e-06,
      "loss": 0.2674,
      "step": 663
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.964118047196627,
      "learning_rate": 9.985989003944655e-06,
      "loss": 0.7795,
      "step": 664
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4443487250329152,
      "learning_rate": 9.985891919982878e-06,
      "loss": 0.2276,
      "step": 665
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6024950006097898,
      "learning_rate": 9.98579450130307e-06,
      "loss": 0.2298,
      "step": 666
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.499163223693087,
      "learning_rate": 9.98569674791177e-06,
      "loss": 0.2468,
      "step": 667
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4996555762851789,
      "learning_rate": 9.985598659815543e-06,
      "loss": 0.2315,
      "step": 668
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.589601832900611,
      "learning_rate": 9.985500237020972e-06,
      "loss": 0.1994,
      "step": 669
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.511346317111211,
      "learning_rate": 9.985401479534664e-06,
      "loss": 0.1664,
      "step": 670
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.510992851798693,
      "learning_rate": 9.985302387363249e-06,
      "loss": 0.1495,
      "step": 671
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5637961680474897,
      "learning_rate": 9.985202960513381e-06,
      "loss": 0.2005,
      "step": 672
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.809633882925538,
      "learning_rate": 9.985103198991733e-06,
      "loss": 0.6506,
      "step": 673
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6417080526020813,
      "learning_rate": 9.985003102805004e-06,
      "loss": 0.2702,
      "step": 674
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.612573023316388,
      "learning_rate": 9.984902671959911e-06,
      "loss": 0.226,
      "step": 675
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4991800013606413,
      "learning_rate": 9.984801906463199e-06,
      "loss": 0.2415,
      "step": 676
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.571089584243743,
      "learning_rate": 9.984700806321631e-06,
      "loss": 0.2078,
      "step": 677
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.767553442670523,
      "learning_rate": 9.984599371541995e-06,
      "loss": 0.2548,
      "step": 678
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6380892120623323,
      "learning_rate": 9.984497602131101e-06,
      "loss": 0.2697,
      "step": 679
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6872874689547022,
      "learning_rate": 9.98439549809578e-06,
      "loss": 0.2405,
      "step": 680
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7309222951178749,
      "learning_rate": 9.984293059442888e-06,
      "loss": 0.2714,
      "step": 681
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4928730463120055,
      "learning_rate": 9.9841902861793e-06,
      "loss": 0.2357,
      "step": 682
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7283480791607235,
      "learning_rate": 9.984087178311917e-06,
      "loss": 0.2318,
      "step": 683
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5683098080694315,
      "learning_rate": 9.98398373584766e-06,
      "loss": 0.1564,
      "step": 684
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.483321651287427,
      "learning_rate": 9.983879958793476e-06,
      "loss": 0.5989,
      "step": 685
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.73557280843961,
      "learning_rate": 9.983775847156327e-06,
      "loss": 0.7552,
      "step": 686
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5430137528961039,
      "learning_rate": 9.983671400943206e-06,
      "loss": 0.2078,
      "step": 687
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6108688052305573,
      "learning_rate": 9.983566620161126e-06,
      "loss": 0.2167,
      "step": 688
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5866760723152404,
      "learning_rate": 9.983461504817119e-06,
      "loss": 0.2159,
      "step": 689
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6287221689970799,
      "learning_rate": 9.983356054918238e-06,
      "loss": 0.2298,
      "step": 690
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.309985381252691,
      "learning_rate": 9.983250270471569e-06,
      "loss": 0.6667,
      "step": 691
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.46354666321353,
      "learning_rate": 9.98314415148421e-06,
      "loss": 0.1661,
      "step": 692
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7442231640817898,
      "learning_rate": 9.983037697963287e-06,
      "loss": 0.2756,
      "step": 693
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5942400155331855,
      "learning_rate": 9.982930909915944e-06,
      "loss": 0.2374,
      "step": 694
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.967596940971285,
      "learning_rate": 9.982823787349352e-06,
      "loss": 0.4513,
      "step": 695
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5748029913355504,
      "learning_rate": 9.982716330270701e-06,
      "loss": 0.2721,
      "step": 696
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7528853554171635,
      "learning_rate": 9.982608538687208e-06,
      "loss": 0.3013,
      "step": 697
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8242073167411055,
      "learning_rate": 9.982500412606105e-06,
      "loss": 0.2718,
      "step": 698
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.877471434195713,
      "learning_rate": 9.982391952034653e-06,
      "loss": 0.2788,
      "step": 699
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.431516573750643,
      "learning_rate": 9.982283156980133e-06,
      "loss": 0.1923,
      "step": 700
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5305076944664961,
      "learning_rate": 9.982174027449849e-06,
      "loss": 0.24,
      "step": 701
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5516135495338292,
      "learning_rate": 9.982064563451128e-06,
      "loss": 0.2038,
      "step": 702
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8808013155712298,
      "learning_rate": 9.981954764991318e-06,
      "loss": 0.2414,
      "step": 703
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9105853038326197,
      "learning_rate": 9.981844632077788e-06,
      "loss": 0.2307,
      "step": 704
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.070204946011062,
      "learning_rate": 9.981734164717936e-06,
      "loss": 0.7087,
      "step": 705
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6145858457686906,
      "learning_rate": 9.981623362919173e-06,
      "loss": 0.2177,
      "step": 706
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7589534705172456,
      "learning_rate": 9.981512226688943e-06,
      "loss": 0.299,
      "step": 707
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7872652226825059,
      "learning_rate": 9.981400756034701e-06,
      "loss": 0.2177,
      "step": 708
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.595467585425629,
      "learning_rate": 9.981288950963935e-06,
      "loss": 0.6067,
      "step": 709
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7039434754853582,
      "learning_rate": 9.981176811484148e-06,
      "loss": 0.2768,
      "step": 710
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.936703805348352,
      "learning_rate": 9.981064337602869e-06,
      "loss": 0.7969,
      "step": 711
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5417335536639822,
      "learning_rate": 9.980951529327649e-06,
      "loss": 0.2288,
      "step": 712
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6031228450364754,
      "learning_rate": 9.980838386666063e-06,
      "loss": 0.2112,
      "step": 713
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5275420117561587,
      "learning_rate": 9.980724909625704e-06,
      "loss": 0.1867,
      "step": 714
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.58178778497284,
      "learning_rate": 9.98061109821419e-06,
      "loss": 0.2188,
      "step": 715
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6153527043651708,
      "learning_rate": 9.980496952439162e-06,
      "loss": 0.2413,
      "step": 716
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7786101072483669,
      "learning_rate": 9.980382472308283e-06,
      "loss": 0.2484,
      "step": 717
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5491416040421198,
      "learning_rate": 9.980267657829241e-06,
      "loss": 0.2427,
      "step": 718
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.631354661525792,
      "learning_rate": 9.98015250900974e-06,
      "loss": 0.5361,
      "step": 719
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5936177552114652,
      "learning_rate": 9.980037025857511e-06,
      "loss": 0.2743,
      "step": 720
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.75208958421876,
      "learning_rate": 9.979921208380308e-06,
      "loss": 0.263,
      "step": 721
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.542580678026301,
      "learning_rate": 9.979805056585907e-06,
      "loss": 0.2257,
      "step": 722
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5516303181404911,
      "learning_rate": 9.979688570482102e-06,
      "loss": 0.2321,
      "step": 723
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5122807139154193,
      "learning_rate": 9.979571750076717e-06,
      "loss": 0.2383,
      "step": 724
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7023205749186943,
      "learning_rate": 9.979454595377594e-06,
      "loss": 0.2492,
      "step": 725
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7133166051736621,
      "learning_rate": 9.979337106392596e-06,
      "loss": 0.2321,
      "step": 726
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3762061613889318,
      "learning_rate": 9.97921928312961e-06,
      "loss": 0.1812,
      "step": 727
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.700459300745377,
      "learning_rate": 9.979101125596548e-06,
      "loss": 0.2862,
      "step": 728
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4414095884976699,
      "learning_rate": 9.978982633801342e-06,
      "loss": 0.2094,
      "step": 729
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.682819877335394,
      "learning_rate": 9.978863807751944e-06,
      "loss": 0.2483,
      "step": 730
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.9730690074405635,
      "learning_rate": 9.978744647456335e-06,
      "loss": 0.6216,
      "step": 731
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8055672341248263,
      "learning_rate": 9.978625152922511e-06,
      "loss": 0.2933,
      "step": 732
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.62835706808305,
      "learning_rate": 9.978505324158499e-06,
      "loss": 0.2548,
      "step": 733
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.548947469338119,
      "learning_rate": 9.97838516117234e-06,
      "loss": 0.2634,
      "step": 734
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8679351330490013,
      "learning_rate": 9.978264663972099e-06,
      "loss": 0.3044,
      "step": 735
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5976705469811023,
      "learning_rate": 9.978143832565868e-06,
      "loss": 0.2224,
      "step": 736
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5369696680646707,
      "learning_rate": 9.97802266696176e-06,
      "loss": 0.2587,
      "step": 737
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.55808208457302,
      "learning_rate": 9.977901167167904e-06,
      "loss": 0.2193,
      "step": 738
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6041090402085885,
      "learning_rate": 9.977779333192464e-06,
      "loss": 0.226,
      "step": 739
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6009131481741816,
      "learning_rate": 9.977657165043613e-06,
      "loss": 0.2288,
      "step": 740
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6068170310437762,
      "learning_rate": 9.977534662729556e-06,
      "loss": 0.2091,
      "step": 741
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5875563244585098,
      "learning_rate": 9.977411826258516e-06,
      "loss": 0.2704,
      "step": 742
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5382575067972875,
      "learning_rate": 9.977288655638737e-06,
      "loss": 0.2476,
      "step": 743
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6527039859361619,
      "learning_rate": 9.977165150878492e-06,
      "loss": 0.2223,
      "step": 744
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5500088874233902,
      "learning_rate": 9.977041311986072e-06,
      "loss": 0.2236,
      "step": 745
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.671063292017438,
      "learning_rate": 9.976917138969784e-06,
      "loss": 0.2244,
      "step": 746
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5894419089902876,
      "learning_rate": 9.976792631837973e-06,
      "loss": 0.2645,
      "step": 747
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.748680435043252,
      "learning_rate": 9.976667790598991e-06,
      "loss": 0.2851,
      "step": 748
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.086589648319484,
      "learning_rate": 9.976542615261223e-06,
      "loss": 0.7588,
      "step": 749
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5242147921331002,
      "learning_rate": 9.97641710583307e-06,
      "loss": 0.2097,
      "step": 750
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.88729303714169,
      "learning_rate": 9.97629126232296e-06,
      "loss": 0.5644,
      "step": 751
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5971660429260632,
      "learning_rate": 9.97616508473934e-06,
      "loss": 0.1859,
      "step": 752
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5200953564780697,
      "learning_rate": 9.976038573090679e-06,
      "loss": 0.2198,
      "step": 753
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3229233995214702,
      "learning_rate": 9.975911727385473e-06,
      "loss": 0.1937,
      "step": 754
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7107874577662805,
      "learning_rate": 9.975784547632237e-06,
      "loss": 0.279,
      "step": 755
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.590861964069376,
      "learning_rate": 9.975657033839506e-06,
      "loss": 0.2154,
      "step": 756
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5852799064118221,
      "learning_rate": 9.975529186015844e-06,
      "loss": 0.2438,
      "step": 757
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.833299840709122,
      "learning_rate": 9.975401004169834e-06,
      "loss": 0.2529,
      "step": 758
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6261263969385116,
      "learning_rate": 9.975272488310077e-06,
      "loss": 0.2953,
      "step": 759
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.134755177981726,
      "learning_rate": 9.975143638445205e-06,
      "loss": 0.4565,
      "step": 760
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4838292118798306,
      "learning_rate": 9.975014454583867e-06,
      "loss": 0.1897,
      "step": 761
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5086098099853702,
      "learning_rate": 9.974884936734734e-06,
      "loss": 0.2216,
      "step": 762
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5807894380856151,
      "learning_rate": 9.974755084906503e-06,
      "loss": 0.2387,
      "step": 763
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6034485373287957,
      "learning_rate": 9.974624899107887e-06,
      "loss": 0.1979,
      "step": 764
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6666110968479153,
      "learning_rate": 9.974494379347632e-06,
      "loss": 0.2971,
      "step": 765
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3425496727150754,
      "learning_rate": 9.974363525634496e-06,
      "loss": 0.1679,
      "step": 766
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6816606853728424,
      "learning_rate": 9.974232337977265e-06,
      "loss": 0.2389,
      "step": 767
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6097517949514568,
      "learning_rate": 9.974100816384746e-06,
      "loss": 0.2436,
      "step": 768
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.2533419767079295,
      "learning_rate": 9.97396896086577e-06,
      "loss": 0.3795,
      "step": 769
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6474270440333585,
      "learning_rate": 9.973836771429185e-06,
      "loss": 0.2474,
      "step": 770
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5292543960921259,
      "learning_rate": 9.973704248083868e-06,
      "loss": 0.2144,
      "step": 771
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.91669239814435,
      "learning_rate": 9.973571390838715e-06,
      "loss": 0.5486,
      "step": 772
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6010615356715285,
      "learning_rate": 9.973438199702645e-06,
      "loss": 0.2737,
      "step": 773
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6150235863971005,
      "learning_rate": 9.9733046746846e-06,
      "loss": 0.2341,
      "step": 774
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.549145978996915,
      "learning_rate": 9.973170815793543e-06,
      "loss": 0.2201,
      "step": 775
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.854928860857746,
      "learning_rate": 9.973036623038462e-06,
      "loss": 0.7801,
      "step": 776
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5339319218427738,
      "learning_rate": 9.972902096428365e-06,
      "loss": 0.2196,
      "step": 777
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3591211197368716,
      "learning_rate": 9.972767235972283e-06,
      "loss": 0.1778,
      "step": 778
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.721549849234291,
      "learning_rate": 9.972632041679268e-06,
      "loss": 0.3117,
      "step": 779
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5052800770726014,
      "learning_rate": 9.972496513558399e-06,
      "loss": 0.2019,
      "step": 780
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8045207991279795,
      "learning_rate": 9.972360651618772e-06,
      "loss": 0.2673,
      "step": 781
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7356180445400264,
      "learning_rate": 9.972224455869508e-06,
      "loss": 0.2204,
      "step": 782
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5275042560215937,
      "learning_rate": 9.972087926319753e-06,
      "loss": 0.2035,
      "step": 783
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6674975865562465,
      "learning_rate": 9.971951062978671e-06,
      "loss": 0.2243,
      "step": 784
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6735786014976555,
      "learning_rate": 9.971813865855448e-06,
      "loss": 0.2393,
      "step": 785
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.485054417507819,
      "learning_rate": 9.971676334959297e-06,
      "loss": 0.2449,
      "step": 786
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4598758873716757,
      "learning_rate": 9.971538470299452e-06,
      "loss": 0.2146,
      "step": 787
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4500147360744786,
      "learning_rate": 9.971400271885163e-06,
      "loss": 0.2205,
      "step": 788
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.564290257911547,
      "learning_rate": 9.971261739725713e-06,
      "loss": 0.2403,
      "step": 789
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5888183187840743,
      "learning_rate": 9.971122873830398e-06,
      "loss": 0.2161,
      "step": 790
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.399120216199401,
      "learning_rate": 9.970983674208546e-06,
      "loss": 0.2339,
      "step": 791
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.616206822487652,
      "learning_rate": 9.970844140869495e-06,
      "loss": 0.2423,
      "step": 792
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4704532267575672,
      "learning_rate": 9.970704273822618e-06,
      "loss": 0.186,
      "step": 793
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.510036611301035,
      "learning_rate": 9.9705640730773e-06,
      "loss": 0.2159,
      "step": 794
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5833938812424388,
      "learning_rate": 9.970423538642959e-06,
      "loss": 0.2063,
      "step": 795
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.783645778096462,
      "learning_rate": 9.970282670529024e-06,
      "loss": 0.1971,
      "step": 796
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6602911300475403,
      "learning_rate": 9.970141468744953e-06,
      "loss": 0.2835,
      "step": 797
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.4237699713685235,
      "learning_rate": 9.969999933300229e-06,
      "loss": 0.7389,
      "step": 798
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.146900448883108,
      "learning_rate": 9.96985806420435e-06,
      "loss": 0.7412,
      "step": 799
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.651043306863088,
      "learning_rate": 9.969715861466839e-06,
      "loss": 0.204,
      "step": 800
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5515993963549433,
      "learning_rate": 9.969573325097247e-06,
      "loss": 0.2184,
      "step": 801
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.580301351811492,
      "learning_rate": 9.96943045510514e-06,
      "loss": 0.2038,
      "step": 802
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.503931124932624,
      "learning_rate": 9.969287251500109e-06,
      "loss": 0.2203,
      "step": 803
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5395751313167916,
      "learning_rate": 9.96914371429177e-06,
      "loss": 0.232,
      "step": 804
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7578679879406942,
      "learning_rate": 9.968999843489755e-06,
      "loss": 0.2638,
      "step": 805
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4629681709189775,
      "learning_rate": 9.968855639103727e-06,
      "loss": 0.2363,
      "step": 806
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.823305579292933,
      "learning_rate": 9.968711101143364e-06,
      "loss": 0.5712,
      "step": 807
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6129816277901692,
      "learning_rate": 9.96856622961837e-06,
      "loss": 0.2085,
      "step": 808
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.675313435355806,
      "learning_rate": 9.968421024538473e-06,
      "loss": 0.2144,
      "step": 809
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.643725522989435,
      "learning_rate": 9.968275485913417e-06,
      "loss": 0.5537,
      "step": 810
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6736518701220606,
      "learning_rate": 9.968129613752975e-06,
      "loss": 0.2294,
      "step": 811
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5608030907843273,
      "learning_rate": 9.967983408066939e-06,
      "loss": 0.239,
      "step": 812
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.692421120277671,
      "learning_rate": 9.967836868865125e-06,
      "loss": 0.2152,
      "step": 813
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.490998666464393,
      "learning_rate": 9.967689996157368e-06,
      "loss": 0.2281,
      "step": 814
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5911104377564773,
      "learning_rate": 9.967542789953532e-06,
      "loss": 0.1956,
      "step": 815
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7486329006113877,
      "learning_rate": 9.967395250263496e-06,
      "loss": 0.2615,
      "step": 816
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.918347572210306,
      "learning_rate": 9.967247377097168e-06,
      "loss": 0.5843,
      "step": 817
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7712335600413032,
      "learning_rate": 9.967099170464473e-06,
      "loss": 0.222,
      "step": 818
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4287793366941868,
      "learning_rate": 9.966950630375361e-06,
      "loss": 0.1941,
      "step": 819
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6253770319429273,
      "learning_rate": 9.966801756839802e-06,
      "loss": 0.232,
      "step": 820
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.620658044413487,
      "learning_rate": 9.966652549867795e-06,
      "loss": 0.2458,
      "step": 821
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5390088992562736,
      "learning_rate": 9.966503009469352e-06,
      "loss": 0.1821,
      "step": 822
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7820222092675864,
      "learning_rate": 9.966353135654513e-06,
      "loss": 0.2764,
      "step": 823
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7563027287758286,
      "learning_rate": 9.966202928433344e-06,
      "loss": 0.2619,
      "step": 824
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6412047958781044,
      "learning_rate": 9.966052387815923e-06,
      "loss": 0.2296,
      "step": 825
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8113012568420899,
      "learning_rate": 9.96590151381236e-06,
      "loss": 0.2548,
      "step": 826
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8036425669711458,
      "learning_rate": 9.965750306432782e-06,
      "loss": 0.2735,
      "step": 827
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7605930651182087,
      "learning_rate": 9.965598765687338e-06,
      "loss": 0.2326,
      "step": 828
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7939421554999757,
      "learning_rate": 9.965446891586208e-06,
      "loss": 0.2651,
      "step": 829
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5273136745259914,
      "learning_rate": 9.96529468413958e-06,
      "loss": 0.2342,
      "step": 830
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7647983712606425,
      "learning_rate": 9.965142143357677e-06,
      "loss": 0.2703,
      "step": 831
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4905150487397107,
      "learning_rate": 9.964989269250737e-06,
      "loss": 0.2231,
      "step": 832
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6319494378773265,
      "learning_rate": 9.964836061829026e-06,
      "loss": 0.2739,
      "step": 833
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.545059339126701,
      "learning_rate": 9.964682521102827e-06,
      "loss": 0.2391,
      "step": 834
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4755142109745647,
      "learning_rate": 9.964528647082447e-06,
      "loss": 0.2064,
      "step": 835
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6133247402182394,
      "learning_rate": 9.964374439778217e-06,
      "loss": 0.2633,
      "step": 836
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.79839837211012,
      "learning_rate": 9.964219899200489e-06,
      "loss": 0.6151,
      "step": 837
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6126352316833688,
      "learning_rate": 9.964065025359639e-06,
      "loss": 0.2248,
      "step": 838
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5443897893326974,
      "learning_rate": 9.963909818266063e-06,
      "loss": 0.223,
      "step": 839
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6891128038040328,
      "learning_rate": 9.96375427793018e-06,
      "loss": 0.2333,
      "step": 840
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5128042488233393,
      "learning_rate": 9.963598404362435e-06,
      "loss": 0.2204,
      "step": 841
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6184011045283073,
      "learning_rate": 9.963442197573288e-06,
      "loss": 0.1955,
      "step": 842
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5777606878059403,
      "learning_rate": 9.96328565757323e-06,
      "loss": 0.2442,
      "step": 843
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5594988761353863,
      "learning_rate": 9.963128784372765e-06,
      "loss": 0.2371,
      "step": 844
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5431292321025256,
      "learning_rate": 9.962971577982428e-06,
      "loss": 0.231,
      "step": 845
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3876276464270902,
      "learning_rate": 9.962814038412772e-06,
      "loss": 0.1801,
      "step": 846
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5547517341145136,
      "learning_rate": 9.96265616567437e-06,
      "loss": 0.2167,
      "step": 847
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5748061768874722,
      "learning_rate": 9.962497959777828e-06,
      "loss": 0.2515,
      "step": 848
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.507620857460194,
      "learning_rate": 9.96233942073376e-06,
      "loss": 0.251,
      "step": 849
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.786641707498006,
      "learning_rate": 9.962180548552812e-06,
      "loss": 0.2228,
      "step": 850
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5734565870069346,
      "learning_rate": 9.96202134324565e-06,
      "loss": 0.2255,
      "step": 851
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.530993640017462,
      "learning_rate": 9.961861804822958e-06,
      "loss": 0.2184,
      "step": 852
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7994293277832776,
      "learning_rate": 9.961701933295451e-06,
      "loss": 0.2226,
      "step": 853
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5517399666080043,
      "learning_rate": 9.961541728673859e-06,
      "loss": 0.1879,
      "step": 854
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5249486387514,
      "learning_rate": 9.96138119096894e-06,
      "loss": 0.2072,
      "step": 855
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6528342596573378,
      "learning_rate": 9.961220320191466e-06,
      "loss": 0.2421,
      "step": 856
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3783025751873852,
      "learning_rate": 9.961059116352242e-06,
      "loss": 0.2071,
      "step": 857
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6393079519715634,
      "learning_rate": 9.960897579462088e-06,
      "loss": 0.2504,
      "step": 858
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5175184471151877,
      "learning_rate": 9.960735709531848e-06,
      "loss": 0.235,
      "step": 859
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.517128423630176,
      "learning_rate": 9.960573506572391e-06,
      "loss": 0.7902,
      "step": 860
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.652309784477271,
      "learning_rate": 9.960410970594603e-06,
      "loss": 0.5837,
      "step": 861
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6105868569186426,
      "learning_rate": 9.960248101609396e-06,
      "loss": 0.261,
      "step": 862
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.733488267489593,
      "learning_rate": 9.960084899627707e-06,
      "loss": 0.2248,
      "step": 863
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5597977486668484,
      "learning_rate": 9.95992136466049e-06,
      "loss": 0.1748,
      "step": 864
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5795473051518327,
      "learning_rate": 9.959757496718723e-06,
      "loss": 0.2708,
      "step": 865
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2374724235499004,
      "learning_rate": 9.959593295813409e-06,
      "loss": 0.1666,
      "step": 866
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6591074051281063,
      "learning_rate": 9.959428761955569e-06,
      "loss": 0.2571,
      "step": 867
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.903684773595323,
      "learning_rate": 9.95926389515625e-06,
      "loss": 0.6636,
      "step": 868
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4582363686787694,
      "learning_rate": 9.959098695426518e-06,
      "loss": 0.2195,
      "step": 869
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.657775124388049,
      "learning_rate": 9.958933162777468e-06,
      "loss": 0.3255,
      "step": 870
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7430338508342884,
      "learning_rate": 9.958767297220209e-06,
      "loss": 0.2759,
      "step": 871
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5635963750475803,
      "learning_rate": 9.958601098765877e-06,
      "loss": 0.2379,
      "step": 872
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.663777620810148,
      "learning_rate": 9.958434567425627e-06,
      "loss": 0.7178,
      "step": 873
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.593180292621246,
      "learning_rate": 9.958267703210645e-06,
      "loss": 0.2592,
      "step": 874
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4938445708827173,
      "learning_rate": 9.958100506132127e-06,
      "loss": 0.1859,
      "step": 875
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4001428069220136,
      "learning_rate": 9.957932976201298e-06,
      "loss": 0.2169,
      "step": 876
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.6482041393412645,
      "learning_rate": 9.95776511342941e-06,
      "loss": 0.6046,
      "step": 877
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6632024693049707,
      "learning_rate": 9.957596917827726e-06,
      "loss": 0.2463,
      "step": 878
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7549458091522492,
      "learning_rate": 9.95742838940754e-06,
      "loss": 0.1821,
      "step": 879
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.336607146784815,
      "learning_rate": 9.957259528180166e-06,
      "loss": 0.6606,
      "step": 880
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.596421033384349,
      "learning_rate": 9.95709033415694e-06,
      "loss": 0.2209,
      "step": 881
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7031625720340766,
      "learning_rate": 9.956920807349222e-06,
      "loss": 0.2214,
      "step": 882
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.614354261023114,
      "learning_rate": 9.95675094776839e-06,
      "loss": 0.2226,
      "step": 883
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5391995482599397,
      "learning_rate": 9.956580755425847e-06,
      "loss": 0.2109,
      "step": 884
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6158173858056841,
      "learning_rate": 9.956410230333023e-06,
      "loss": 0.2377,
      "step": 885
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6555947275471679,
      "learning_rate": 9.956239372501361e-06,
      "loss": 0.2236,
      "step": 886
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6681512449709777,
      "learning_rate": 9.956068181942333e-06,
      "loss": 0.2782,
      "step": 887
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5652856598780753,
      "learning_rate": 9.955896658667433e-06,
      "loss": 0.2397,
      "step": 888
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.454534919169001,
      "learning_rate": 9.955724802688173e-06,
      "loss": 0.2301,
      "step": 889
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5952741803703179,
      "learning_rate": 9.955552614016093e-06,
      "loss": 0.2475,
      "step": 890
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4927589801345362,
      "learning_rate": 9.955380092662751e-06,
      "loss": 0.2365,
      "step": 891
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.685133760573064,
      "learning_rate": 9.955207238639729e-06,
      "loss": 0.2577,
      "step": 892
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5541686942396116,
      "learning_rate": 9.955034051958632e-06,
      "loss": 0.2164,
      "step": 893
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.35315888932159,
      "learning_rate": 9.954860532631086e-06,
      "loss": 0.6672,
      "step": 894
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5424407308739105,
      "learning_rate": 9.954686680668737e-06,
      "loss": 0.186,
      "step": 895
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3933068540170461,
      "learning_rate": 9.954512496083262e-06,
      "loss": 0.2166,
      "step": 896
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5958090136772076,
      "learning_rate": 9.95433797888635e-06,
      "loss": 0.2378,
      "step": 897
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.411236747990016,
      "learning_rate": 9.95416312908972e-06,
      "loss": 0.5844,
      "step": 898
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4064977250970212,
      "learning_rate": 9.953987946705108e-06,
      "loss": 0.186,
      "step": 899
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6459686346692761,
      "learning_rate": 9.953812431744274e-06,
      "loss": 0.2663,
      "step": 900
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6815411637250797,
      "learning_rate": 9.953636584219004e-06,
      "loss": 0.2335,
      "step": 901
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5351604977918487,
      "learning_rate": 9.9534604041411e-06,
      "loss": 0.232,
      "step": 902
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7265725215370527,
      "learning_rate": 9.953283891522393e-06,
      "loss": 0.2608,
      "step": 903
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.070918595194136,
      "learning_rate": 9.953107046374726e-06,
      "loss": 0.6555,
      "step": 904
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6266271289217815,
      "learning_rate": 9.95292986870998e-06,
      "loss": 0.2466,
      "step": 905
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6221893894908572,
      "learning_rate": 9.95275235854004e-06,
      "loss": 0.264,
      "step": 906
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.740166860557886,
      "learning_rate": 9.952574515876833e-06,
      "loss": 0.2098,
      "step": 907
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6429813550569097,
      "learning_rate": 9.952396340732292e-06,
      "loss": 0.2645,
      "step": 908
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3973402353889148,
      "learning_rate": 9.952217833118377e-06,
      "loss": 0.1947,
      "step": 909
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2841636408910444,
      "learning_rate": 9.952038993047076e-06,
      "loss": 0.2188,
      "step": 910
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6062633476128698,
      "learning_rate": 9.951859820530394e-06,
      "loss": 0.2003,
      "step": 911
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6050190748438495,
      "learning_rate": 9.951680315580356e-06,
      "loss": 0.2755,
      "step": 912
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8415823509569391,
      "learning_rate": 9.951500478209018e-06,
      "loss": 0.2936,
      "step": 913
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6127471457348062,
      "learning_rate": 9.951320308428449e-06,
      "loss": 0.2425,
      "step": 914
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5751057869021148,
      "learning_rate": 9.951139806250747e-06,
      "loss": 0.2193,
      "step": 915
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.386367079834223,
      "learning_rate": 9.950958971688028e-06,
      "loss": 0.7378,
      "step": 916
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5491473017552104,
      "learning_rate": 9.950777804752432e-06,
      "loss": 0.2147,
      "step": 917
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6049453296205214,
      "learning_rate": 9.950596305456124e-06,
      "loss": 0.2489,
      "step": 918
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.643927653364026,
      "learning_rate": 9.950414473811283e-06,
      "loss": 0.7063,
      "step": 919
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.921814508291732,
      "learning_rate": 9.950232309830121e-06,
      "loss": 0.4937,
      "step": 920
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5538309090102893,
      "learning_rate": 9.950049813524865e-06,
      "loss": 0.2254,
      "step": 921
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7942758741770277,
      "learning_rate": 9.949866984907768e-06,
      "loss": 0.2549,
      "step": 922
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6934101926310348,
      "learning_rate": 9.9496838239911e-06,
      "loss": 0.2501,
      "step": 923
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5839252274937599,
      "learning_rate": 9.949500330787162e-06,
      "loss": 0.2323,
      "step": 924
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.579175450399343,
      "learning_rate": 9.94931650530827e-06,
      "loss": 0.2117,
      "step": 925
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.316547008170586,
      "learning_rate": 9.949132347566765e-06,
      "loss": 0.7896,
      "step": 926
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5327134758391607,
      "learning_rate": 9.948947857575012e-06,
      "loss": 0.2325,
      "step": 927
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.468447725809474,
      "learning_rate": 9.948763035345393e-06,
      "loss": 0.7521,
      "step": 928
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4508076923260516,
      "learning_rate": 9.948577880890318e-06,
      "loss": 0.2227,
      "step": 929
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.612063571957598,
      "learning_rate": 9.948392394222214e-06,
      "loss": 0.2197,
      "step": 930
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.583841051716503,
      "learning_rate": 9.948206575353539e-06,
      "loss": 0.5356,
      "step": 931
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5291051527881614,
      "learning_rate": 9.948020424296762e-06,
      "loss": 0.2391,
      "step": 932
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6544556640373385,
      "learning_rate": 9.947833941064382e-06,
      "loss": 0.2765,
      "step": 933
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.492941213120045,
      "learning_rate": 9.94764712566892e-06,
      "loss": 0.2283,
      "step": 934
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.476821365229082,
      "learning_rate": 9.947459978122912e-06,
      "loss": 0.8646,
      "step": 935
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.524828204377511,
      "learning_rate": 9.947272498438929e-06,
      "loss": 0.2004,
      "step": 936
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4975819681283637,
      "learning_rate": 9.947084686629552e-06,
      "loss": 0.2203,
      "step": 937
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5338876062166038,
      "learning_rate": 9.946896542707391e-06,
      "loss": 0.2312,
      "step": 938
    },
    {
      "epoch": 0.08,
      "grad_norm": 30.750702403951543,
      "learning_rate": 9.946708066685077e-06,
      "loss": 0.7122,
      "step": 939
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5084990224143429,
      "learning_rate": 9.946519258575263e-06,
      "loss": 0.2541,
      "step": 940
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6274106660844325,
      "learning_rate": 9.946330118390622e-06,
      "loss": 0.231,
      "step": 941
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7300014426029913,
      "learning_rate": 9.946140646143856e-06,
      "loss": 0.2565,
      "step": 942
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.09768129921992,
      "learning_rate": 9.94595084184768e-06,
      "loss": 0.6627,
      "step": 943
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5832460693725294,
      "learning_rate": 9.945760705514839e-06,
      "loss": 0.2582,
      "step": 944
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7698903199669065,
      "learning_rate": 9.945570237158098e-06,
      "loss": 0.2576,
      "step": 945
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.497249738308736,
      "learning_rate": 9.94537943679024e-06,
      "loss": 0.2269,
      "step": 946
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3900691512748142,
      "learning_rate": 9.945188304424078e-06,
      "loss": 0.2249,
      "step": 947
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.926543979555571,
      "learning_rate": 9.944996840072442e-06,
      "loss": 0.8144,
      "step": 948
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5465396857631586,
      "learning_rate": 9.944805043748185e-06,
      "loss": 0.2537,
      "step": 949
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.034809549310933,
      "learning_rate": 9.944612915464183e-06,
      "loss": 0.5551,
      "step": 950
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.562651876822184,
      "learning_rate": 9.944420455233335e-06,
      "loss": 0.689,
      "step": 951
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6360519621442735,
      "learning_rate": 9.94422766306856e-06,
      "loss": 0.2617,
      "step": 952
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7375542378308855,
      "learning_rate": 9.944034538982804e-06,
      "loss": 0.2667,
      "step": 953
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6005154233453405,
      "learning_rate": 9.943841082989027e-06,
      "loss": 0.2065,
      "step": 954
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5790467979664349,
      "learning_rate": 9.943647295100219e-06,
      "loss": 0.2606,
      "step": 955
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5743554333429108,
      "learning_rate": 9.94345317532939e-06,
      "loss": 0.2743,
      "step": 956
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.60837717216851,
      "learning_rate": 9.94325872368957e-06,
      "loss": 0.2692,
      "step": 957
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.518814478187395,
      "learning_rate": 9.943063940193817e-06,
      "loss": 0.2245,
      "step": 958
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5019157165089836,
      "learning_rate": 9.942868824855202e-06,
      "loss": 0.2129,
      "step": 959
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.541331492481318,
      "learning_rate": 9.94267337768683e-06,
      "loss": 0.2374,
      "step": 960
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4329875568597175,
      "learning_rate": 9.942477598701815e-06,
      "loss": 0.1988,
      "step": 961
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3895280585930996,
      "learning_rate": 9.942281487913306e-06,
      "loss": 0.2209,
      "step": 962
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4109001670295296,
      "learning_rate": 9.942085045334464e-06,
      "loss": 0.1857,
      "step": 963
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4512056507607836,
      "learning_rate": 9.941888270978482e-06,
      "loss": 0.2243,
      "step": 964
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.597993735500171,
      "learning_rate": 9.941691164858565e-06,
      "loss": 0.2587,
      "step": 965
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5330797032064054,
      "learning_rate": 9.941493726987947e-06,
      "loss": 0.2375,
      "step": 966
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6283274387387057,
      "learning_rate": 9.941295957379884e-06,
      "loss": 0.2321,
      "step": 967
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6170550052541643,
      "learning_rate": 9.941097856047652e-06,
      "loss": 0.2231,
      "step": 968
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6310327950769177,
      "learning_rate": 9.940899423004548e-06,
      "loss": 0.2249,
      "step": 969
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7714358577666947,
      "learning_rate": 9.940700658263897e-06,
      "loss": 0.2927,
      "step": 970
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.494587341201075,
      "learning_rate": 9.940501561839043e-06,
      "loss": 0.6137,
      "step": 971
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5307607267643362,
      "learning_rate": 9.940302133743347e-06,
      "loss": 0.1963,
      "step": 972
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.41109477015087,
      "learning_rate": 9.940102373990202e-06,
      "loss": 0.2398,
      "step": 973
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6673305237552476,
      "learning_rate": 9.939902282593015e-06,
      "loss": 0.2671,
      "step": 974
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3937229651279954,
      "learning_rate": 9.93970185956522e-06,
      "loss": 0.2063,
      "step": 975
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.801030692531414,
      "learning_rate": 9.939501104920275e-06,
      "loss": 0.6237,
      "step": 976
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6131397479118943,
      "learning_rate": 9.939300018671654e-06,
      "loss": 0.1828,
      "step": 977
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5320730722523492,
      "learning_rate": 9.939098600832857e-06,
      "loss": 0.2219,
      "step": 978
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6972839039006973,
      "learning_rate": 9.938896851417406e-06,
      "loss": 0.2499,
      "step": 979
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7300102797747343,
      "learning_rate": 9.938694770438843e-06,
      "loss": 0.2306,
      "step": 980
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6926207038849739,
      "learning_rate": 9.93849235791074e-06,
      "loss": 0.2472,
      "step": 981
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5464901577367662,
      "learning_rate": 9.93828961384668e-06,
      "loss": 0.2208,
      "step": 982
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4750972717226003,
      "learning_rate": 9.938086538260277e-06,
      "loss": 0.1851,
      "step": 983
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6370078355023407,
      "learning_rate": 9.937883131165163e-06,
      "loss": 0.238,
      "step": 984
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.648554332459701,
      "learning_rate": 9.937679392574991e-06,
      "loss": 0.2452,
      "step": 985
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3602112441409195,
      "learning_rate": 9.937475322503442e-06,
      "loss": 0.1765,
      "step": 986
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.834108864437724,
      "learning_rate": 9.937270920964214e-06,
      "loss": 0.5768,
      "step": 987
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5723169396135572,
      "learning_rate": 9.937066187971031e-06,
      "loss": 0.2409,
      "step": 988
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.449766155435059,
      "learning_rate": 9.936861123537636e-06,
      "loss": 0.2064,
      "step": 989
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.569226779964195,
      "learning_rate": 9.936655727677795e-06,
      "loss": 0.216,
      "step": 990
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.675949214876949,
      "learning_rate": 9.936450000405297e-06,
      "loss": 0.2592,
      "step": 991
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.75156627668863,
      "learning_rate": 9.936243941733956e-06,
      "loss": 0.2665,
      "step": 992
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5423023160586935,
      "learning_rate": 9.9360375516776e-06,
      "loss": 0.1817,
      "step": 993
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.396136235643332,
      "learning_rate": 9.93583083025009e-06,
      "loss": 0.2003,
      "step": 994
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8085942243824484,
      "learning_rate": 9.9356237774653e-06,
      "loss": 0.2397,
      "step": 995
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4024269055644474,
      "learning_rate": 9.935416393337132e-06,
      "loss": 0.2203,
      "step": 996
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6147659812331727,
      "learning_rate": 9.935208677879508e-06,
      "loss": 0.2273,
      "step": 997
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5653745315197816,
      "learning_rate": 9.935000631106372e-06,
      "loss": 0.2583,
      "step": 998
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4414956885664476,
      "learning_rate": 9.93479225303169e-06,
      "loss": 0.2295,
      "step": 999
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.513450440125113,
      "learning_rate": 9.934583543669454e-06,
      "loss": 0.2709,
      "step": 1000
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5560148017202278,
      "learning_rate": 9.934374503033672e-06,
      "loss": 0.2461,
      "step": 1001
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.92353045907564,
      "learning_rate": 9.934165131138381e-06,
      "loss": 0.7145,
      "step": 1002
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.527946841404391,
      "learning_rate": 9.933955427997634e-06,
      "loss": 0.2363,
      "step": 1003
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5162096594084293,
      "learning_rate": 9.933745393625509e-06,
      "loss": 0.2213,
      "step": 1004
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8043073181529599,
      "learning_rate": 9.933535028036108e-06,
      "loss": 0.3097,
      "step": 1005
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4571863545941868,
      "learning_rate": 9.933324331243553e-06,
      "loss": 0.2479,
      "step": 1006
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4390003929230886,
      "learning_rate": 9.933113303261987e-06,
      "loss": 0.2439,
      "step": 1007
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.348203148223269,
      "learning_rate": 9.932901944105578e-06,
      "loss": 0.2131,
      "step": 1008
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.327973066127189,
      "learning_rate": 9.932690253788516e-06,
      "loss": 0.2242,
      "step": 1009
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7254541944995574,
      "learning_rate": 9.932478232325013e-06,
      "loss": 0.2537,
      "step": 1010
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4601674357344157,
      "learning_rate": 9.9322658797293e-06,
      "loss": 0.2241,
      "step": 1011
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6868388967371297,
      "learning_rate": 9.932053196015634e-06,
      "loss": 0.2437,
      "step": 1012
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8984128773301088,
      "learning_rate": 9.931840181198296e-06,
      "loss": 0.2901,
      "step": 1013
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.552741733811618,
      "learning_rate": 9.931626835291581e-06,
      "loss": 0.211,
      "step": 1014
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5708978741476807,
      "learning_rate": 9.931413158309816e-06,
      "loss": 0.2368,
      "step": 1015
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5167620261031745,
      "learning_rate": 9.931199150267343e-06,
      "loss": 0.2588,
      "step": 1016
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6334059992666325,
      "learning_rate": 9.93098481117853e-06,
      "loss": 0.2955,
      "step": 1017
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.510915928012243,
      "learning_rate": 9.930770141057767e-06,
      "loss": 0.2119,
      "step": 1018
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6512566941178652,
      "learning_rate": 9.930555139919465e-06,
      "loss": 0.2926,
      "step": 1019
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0137385934719614,
      "learning_rate": 9.930339807778056e-06,
      "loss": 0.2583,
      "step": 1020
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5162046675913567,
      "learning_rate": 9.930124144647998e-06,
      "loss": 0.2114,
      "step": 1021
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4523137789073988,
      "learning_rate": 9.929908150543769e-06,
      "loss": 0.2331,
      "step": 1022
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.997674020135156,
      "learning_rate": 9.929691825479868e-06,
      "loss": 0.8094,
      "step": 1023
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5994899779486356,
      "learning_rate": 9.929475169470819e-06,
      "loss": 0.2274,
      "step": 1024
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5258480728416541,
      "learning_rate": 9.929258182531167e-06,
      "loss": 0.2642,
      "step": 1025
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4987306543976613,
      "learning_rate": 9.929040864675477e-06,
      "loss": 0.2957,
      "step": 1026
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7077739460073242,
      "learning_rate": 9.92882321591834e-06,
      "loss": 0.2652,
      "step": 1027
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.075228388987073,
      "learning_rate": 9.928605236274368e-06,
      "loss": 0.5457,
      "step": 1028
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4727007291628043,
      "learning_rate": 9.928386925758191e-06,
      "loss": 0.2081,
      "step": 1029
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4970191667197104,
      "learning_rate": 9.928168284384468e-06,
      "loss": 0.2044,
      "step": 1030
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.859127292495222,
      "learning_rate": 9.927949312167876e-06,
      "loss": 0.6173,
      "step": 1031
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.663958350712528,
      "learning_rate": 9.927730009123116e-06,
      "loss": 0.7604,
      "step": 1032
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6049887541285104,
      "learning_rate": 9.92751037526491e-06,
      "loss": 0.2041,
      "step": 1033
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.408856198512203,
      "learning_rate": 9.927290410608003e-06,
      "loss": 0.2053,
      "step": 1034
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4552447095142726,
      "learning_rate": 9.927070115167161e-06,
      "loss": 0.2137,
      "step": 1035
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3913669250596103,
      "learning_rate": 9.926849488957176e-06,
      "loss": 0.2497,
      "step": 1036
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4662104773724527,
      "learning_rate": 9.926628531992855e-06,
      "loss": 0.2686,
      "step": 1037
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4533843449797672,
      "learning_rate": 9.926407244289033e-06,
      "loss": 0.1831,
      "step": 1038
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3069315357941098,
      "learning_rate": 9.926185625860567e-06,
      "loss": 0.165,
      "step": 1039
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8197806524288562,
      "learning_rate": 9.925963676722335e-06,
      "loss": 0.2582,
      "step": 1040
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.636391482369491,
      "learning_rate": 9.925741396889235e-06,
      "loss": 0.2414,
      "step": 1041
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6667961366980302,
      "learning_rate": 9.925518786376192e-06,
      "loss": 0.2002,
      "step": 1042
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.627358044434155,
      "learning_rate": 9.925295845198148e-06,
      "loss": 0.2518,
      "step": 1043
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5305873633266944,
      "learning_rate": 9.92507257337007e-06,
      "loss": 0.1877,
      "step": 1044
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.589233215509505,
      "learning_rate": 9.92484897090695e-06,
      "loss": 0.5678,
      "step": 1045
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5282959217170176,
      "learning_rate": 9.924625037823797e-06,
      "loss": 0.2133,
      "step": 1046
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9613381692978133,
      "learning_rate": 9.924400774135641e-06,
      "loss": 0.248,
      "step": 1047
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5984806319273763,
      "learning_rate": 9.924176179857543e-06,
      "loss": 0.2563,
      "step": 1048
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4152145838407604,
      "learning_rate": 9.923951255004577e-06,
      "loss": 0.2027,
      "step": 1049
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.967197837263107,
      "learning_rate": 9.923725999591846e-06,
      "loss": 0.5974,
      "step": 1050
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.947466420008018,
      "learning_rate": 9.92350041363447e-06,
      "loss": 0.7772,
      "step": 1051
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4447722431281742,
      "learning_rate": 9.923274497147595e-06,
      "loss": 0.2329,
      "step": 1052
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.468874504154234,
      "learning_rate": 9.923048250146383e-06,
      "loss": 0.1986,
      "step": 1053
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5855376368991068,
      "learning_rate": 9.922821672646028e-06,
      "loss": 0.2888,
      "step": 1054
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6317941171123513,
      "learning_rate": 9.922594764661737e-06,
      "loss": 0.2516,
      "step": 1055
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5453199820230867,
      "learning_rate": 9.922367526208746e-06,
      "loss": 0.221,
      "step": 1056
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5356831504505377,
      "learning_rate": 9.922139957302308e-06,
      "loss": 0.2432,
      "step": 1057
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6677105545506716,
      "learning_rate": 9.921912057957701e-06,
      "loss": 0.2428,
      "step": 1058
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.697788112722738,
      "learning_rate": 9.921683828190225e-06,
      "loss": 0.4556,
      "step": 1059
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.563437534868239,
      "learning_rate": 9.9214552680152e-06,
      "loss": 0.2682,
      "step": 1060
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5172546439591739,
      "learning_rate": 9.921226377447975e-06,
      "loss": 0.2367,
      "step": 1061
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.586214624701855,
      "learning_rate": 9.920997156503912e-06,
      "loss": 0.2635,
      "step": 1062
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4906773460514615,
      "learning_rate": 9.920767605198396e-06,
      "loss": 0.2007,
      "step": 1063
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4065150667127742,
      "learning_rate": 9.920537723546843e-06,
      "loss": 0.2087,
      "step": 1064
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.570240754506179,
      "learning_rate": 9.920307511564686e-06,
      "loss": 0.2326,
      "step": 1065
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4435832403868885,
      "learning_rate": 9.920076969267375e-06,
      "loss": 0.2256,
      "step": 1066
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.599550093240707,
      "learning_rate": 9.919846096670393e-06,
      "loss": 0.2598,
      "step": 1067
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5539707437993482,
      "learning_rate": 9.919614893789234e-06,
      "loss": 0.2378,
      "step": 1068
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4931619450507125,
      "learning_rate": 9.919383360639423e-06,
      "loss": 0.2044,
      "step": 1069
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4940578508063573,
      "learning_rate": 9.9191514972365e-06,
      "loss": 0.1937,
      "step": 1070
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5171987980143786,
      "learning_rate": 9.918919303596034e-06,
      "loss": 0.2397,
      "step": 1071
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.045229593322809,
      "learning_rate": 9.918686779733608e-06,
      "loss": 0.5765,
      "step": 1072
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.014158078044732,
      "learning_rate": 9.91845392566484e-06,
      "loss": 0.7735,
      "step": 1073
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5891342340719457,
      "learning_rate": 9.918220741405356e-06,
      "loss": 0.28,
      "step": 1074
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6345710308036756,
      "learning_rate": 9.917987226970811e-06,
      "loss": 0.2543,
      "step": 1075
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5146539662657386,
      "learning_rate": 9.917753382376883e-06,
      "loss": 0.2567,
      "step": 1076
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5358136489579914,
      "learning_rate": 9.91751920763927e-06,
      "loss": 0.2751,
      "step": 1077
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5356927672818956,
      "learning_rate": 9.917284702773692e-06,
      "loss": 0.2322,
      "step": 1078
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.386307629619825,
      "learning_rate": 9.917049867795896e-06,
      "loss": 0.2154,
      "step": 1079
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.162002919949642,
      "learning_rate": 9.916814702721641e-06,
      "loss": 0.2277,
      "step": 1080
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4808403162173098,
      "learning_rate": 9.916579207566721e-06,
      "loss": 0.2316,
      "step": 1081
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7920413639979027,
      "learning_rate": 9.916343382346942e-06,
      "loss": 0.2953,
      "step": 1082
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.52418176156363,
      "learning_rate": 9.916107227078133e-06,
      "loss": 0.1971,
      "step": 1083
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5789887755440322,
      "learning_rate": 9.915870741776153e-06,
      "loss": 0.2452,
      "step": 1084
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5674536441991773,
      "learning_rate": 9.915633926456874e-06,
      "loss": 0.2332,
      "step": 1085
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4787930528596724,
      "learning_rate": 9.915396781136197e-06,
      "loss": 0.305,
      "step": 1086
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6145443680635934,
      "learning_rate": 9.91515930583004e-06,
      "loss": 0.2433,
      "step": 1087
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6283076020039178,
      "learning_rate": 9.914921500554347e-06,
      "loss": 0.272,
      "step": 1088
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5844388695709666,
      "learning_rate": 9.914683365325083e-06,
      "loss": 0.2277,
      "step": 1089
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5718160599454403,
      "learning_rate": 9.914444900158234e-06,
      "loss": 0.193,
      "step": 1090
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.581305145458717,
      "learning_rate": 9.914206105069806e-06,
      "loss": 0.2368,
      "step": 1091
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3677432212787979,
      "learning_rate": 9.913966980075834e-06,
      "loss": 0.1884,
      "step": 1092
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.251517553439886,
      "learning_rate": 9.91372752519237e-06,
      "loss": 0.6546,
      "step": 1093
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6329209880909659,
      "learning_rate": 9.91348774043549e-06,
      "loss": 0.246,
      "step": 1094
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.575897279826717,
      "learning_rate": 9.91324762582129e-06,
      "loss": 0.5309,
      "step": 1095
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8458113687611641,
      "learning_rate": 9.91300718136589e-06,
      "loss": 0.2899,
      "step": 1096
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.748777461531222,
      "learning_rate": 9.912766407085432e-06,
      "loss": 0.2746,
      "step": 1097
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7240686198227673,
      "learning_rate": 9.912525302996081e-06,
      "loss": 0.2397,
      "step": 1098
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6165923715996868,
      "learning_rate": 9.91228386911402e-06,
      "loss": 0.2159,
      "step": 1099
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6999843109138466,
      "learning_rate": 9.912042105455462e-06,
      "loss": 0.2232,
      "step": 1100
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5562556534579395,
      "learning_rate": 9.911800012036633e-06,
      "loss": 0.2144,
      "step": 1101
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4283147342679852,
      "learning_rate": 9.911557588873787e-06,
      "loss": 0.2013,
      "step": 1102
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4331252199175264,
      "learning_rate": 9.911314835983202e-06,
      "loss": 0.2028,
      "step": 1103
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6814503495401072,
      "learning_rate": 9.911071753381168e-06,
      "loss": 0.225,
      "step": 1104
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.502283281418432,
      "learning_rate": 9.910828341084006e-06,
      "loss": 0.5931,
      "step": 1105
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4318182870228882,
      "learning_rate": 9.91058459910806e-06,
      "loss": 0.2269,
      "step": 1106
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3852522308820963,
      "learning_rate": 9.910340527469692e-06,
      "loss": 0.2036,
      "step": 1107
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7521162616308004,
      "learning_rate": 9.910096126185286e-06,
      "loss": 0.2478,
      "step": 1108
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5242575950324693,
      "learning_rate": 9.90985139527125e-06,
      "loss": 0.2257,
      "step": 1109
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6071380500848118,
      "learning_rate": 9.909606334744013e-06,
      "loss": 0.2346,
      "step": 1110
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6690425820703312,
      "learning_rate": 9.909360944620027e-06,
      "loss": 0.2265,
      "step": 1111
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6188746078753145,
      "learning_rate": 9.909115224915768e-06,
      "loss": 0.2438,
      "step": 1112
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.3935367242329555,
      "learning_rate": 9.90886917564773e-06,
      "loss": 0.481,
      "step": 1113
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5557804504612123,
      "learning_rate": 9.908622796832427e-06,
      "loss": 0.2091,
      "step": 1114
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4295384589046842,
      "learning_rate": 9.908376088486407e-06,
      "loss": 0.2112,
      "step": 1115
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5630822073770563,
      "learning_rate": 9.908129050626228e-06,
      "loss": 0.2611,
      "step": 1116
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3386470670812902,
      "learning_rate": 9.907881683268472e-06,
      "loss": 0.1919,
      "step": 1117
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.122830265056606,
      "learning_rate": 9.90763398642975e-06,
      "loss": 0.5886,
      "step": 1118
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.9099362011156655,
      "learning_rate": 9.907385960126689e-06,
      "loss": 0.697,
      "step": 1119
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4619422154511335,
      "learning_rate": 9.907137604375941e-06,
      "loss": 0.2362,
      "step": 1120
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4332301330619177,
      "learning_rate": 9.906888919194178e-06,
      "loss": 0.2288,
      "step": 1121
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4929799912449913,
      "learning_rate": 9.906639904598092e-06,
      "loss": 0.2188,
      "step": 1122
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4861937576996784,
      "learning_rate": 9.906390560604404e-06,
      "loss": 0.2267,
      "step": 1123
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5621958857376876,
      "learning_rate": 9.906140887229852e-06,
      "loss": 0.2273,
      "step": 1124
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.449037565170767,
      "learning_rate": 9.905890884491196e-06,
      "loss": 0.2305,
      "step": 1125
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4342417119363762,
      "learning_rate": 9.905640552405222e-06,
      "loss": 0.2007,
      "step": 1126
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5815516766780038,
      "learning_rate": 9.905389890988734e-06,
      "loss": 0.2644,
      "step": 1127
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5211514049346904,
      "learning_rate": 9.90513890025856e-06,
      "loss": 0.2577,
      "step": 1128
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5769451735251425,
      "learning_rate": 9.904887580231548e-06,
      "loss": 0.2211,
      "step": 1129
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4833729839778171,
      "learning_rate": 9.904635930924573e-06,
      "loss": 0.2235,
      "step": 1130
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.662068518411611,
      "learning_rate": 9.904383952354528e-06,
      "loss": 0.6652,
      "step": 1131
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7770467061461308,
      "learning_rate": 9.904131644538327e-06,
      "loss": 0.2525,
      "step": 1132
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5486244175079165,
      "learning_rate": 9.903879007492912e-06,
      "loss": 0.2054,
      "step": 1133
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.643524940309224,
      "learning_rate": 9.90362604123524e-06,
      "loss": 0.6821,
      "step": 1134
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5080254709256975,
      "learning_rate": 9.903372745782294e-06,
      "loss": 0.1879,
      "step": 1135
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5889033904886767,
      "learning_rate": 9.903119121151079e-06,
      "loss": 0.2837,
      "step": 1136
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.07897584118699,
      "learning_rate": 9.90286516735862e-06,
      "loss": 0.2376,
      "step": 1137
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5901485119856444,
      "learning_rate": 9.90261088442197e-06,
      "loss": 0.2386,
      "step": 1138
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6169484316909901,
      "learning_rate": 9.902356272358196e-06,
      "loss": 0.2426,
      "step": 1139
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4562920020039767,
      "learning_rate": 9.902101331184391e-06,
      "loss": 0.1921,
      "step": 1140
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6275751987498905,
      "learning_rate": 9.901846060917673e-06,
      "loss": 0.2669,
      "step": 1141
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4483419800946988,
      "learning_rate": 9.901590461575175e-06,
      "loss": 0.283,
      "step": 1142
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5581625117799525,
      "learning_rate": 9.901334533174058e-06,
      "loss": 0.2438,
      "step": 1143
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.845540790063638,
      "learning_rate": 9.901078275731504e-06,
      "loss": 0.7201,
      "step": 1144
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2663777535983152,
      "learning_rate": 9.900821689264715e-06,
      "loss": 0.1933,
      "step": 1145
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4199211927022155,
      "learning_rate": 9.90056477379092e-06,
      "loss": 0.1677,
      "step": 1146
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4331803505883982,
      "learning_rate": 9.90030752932736e-06,
      "loss": 0.2008,
      "step": 1147
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.140823995480581,
      "learning_rate": 9.90004995589131e-06,
      "loss": 0.6989,
      "step": 1148
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2880289710375101,
      "learning_rate": 9.899792053500059e-06,
      "loss": 0.1854,
      "step": 1149
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.179045516523009,
      "learning_rate": 9.899533822170922e-06,
      "loss": 0.6106,
      "step": 1150
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.248541362098338,
      "learning_rate": 9.899275261921236e-06,
      "loss": 0.7359,
      "step": 1151
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.709322350555803,
      "learning_rate": 9.899016372768355e-06,
      "loss": 0.565,
      "step": 1152
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9086156275210808,
      "learning_rate": 9.898757154729663e-06,
      "loss": 0.2523,
      "step": 1153
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.750415728950451,
      "learning_rate": 9.898497607822561e-06,
      "loss": 0.2634,
      "step": 1154
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6254634038989189,
      "learning_rate": 9.898237732064472e-06,
      "loss": 0.227,
      "step": 1155
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6893065904451008,
      "learning_rate": 9.897977527472842e-06,
      "loss": 0.2077,
      "step": 1156
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8202402491061556,
      "learning_rate": 9.89771699406514e-06,
      "loss": 0.2616,
      "step": 1157
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.274171275975403,
      "learning_rate": 9.89745613185886e-06,
      "loss": 0.5854,
      "step": 1158
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6841254522318398,
      "learning_rate": 9.897194940871509e-06,
      "loss": 0.2797,
      "step": 1159
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.520906972182786,
      "learning_rate": 9.896933421120623e-06,
      "loss": 0.233,
      "step": 1160
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.663735323311016,
      "learning_rate": 9.89667157262376e-06,
      "loss": 0.2845,
      "step": 1161
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5702075679085052,
      "learning_rate": 9.896409395398499e-06,
      "loss": 0.2511,
      "step": 1162
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.67545391789473,
      "learning_rate": 9.896146889462438e-06,
      "loss": 0.2065,
      "step": 1163
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5343816317798418,
      "learning_rate": 9.895884054833202e-06,
      "loss": 0.242,
      "step": 1164
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5656425485755339,
      "learning_rate": 9.895620891528437e-06,
      "loss": 0.2223,
      "step": 1165
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.640067511117483,
      "learning_rate": 9.895357399565806e-06,
      "loss": 0.7804,
      "step": 1166
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4143070131597346,
      "learning_rate": 9.895093578963002e-06,
      "loss": 0.2002,
      "step": 1167
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.519246262172307,
      "learning_rate": 9.894829429737734e-06,
      "loss": 0.2609,
      "step": 1168
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4767782603853812,
      "learning_rate": 9.894564951907737e-06,
      "loss": 0.1614,
      "step": 1169
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7026333192013057,
      "learning_rate": 9.894300145490763e-06,
      "loss": 0.188,
      "step": 1170
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6776679740499743,
      "learning_rate": 9.894035010504592e-06,
      "loss": 0.1955,
      "step": 1171
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4927297898718945,
      "learning_rate": 9.893769546967023e-06,
      "loss": 0.2389,
      "step": 1172
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5806633855964354,
      "learning_rate": 9.893503754895874e-06,
      "loss": 0.2034,
      "step": 1173
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6572760535927649,
      "learning_rate": 9.893237634308995e-06,
      "loss": 0.2273,
      "step": 1174
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4167153890754696,
      "learning_rate": 9.892971185224244e-06,
      "loss": 0.1986,
      "step": 1175
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7277837602673367,
      "learning_rate": 9.892704407659514e-06,
      "loss": 0.2473,
      "step": 1176
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8087353905467967,
      "learning_rate": 9.892437301632713e-06,
      "loss": 0.2522,
      "step": 1177
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3446262351005338,
      "learning_rate": 9.892169867161774e-06,
      "loss": 0.1729,
      "step": 1178
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3965088154997665,
      "learning_rate": 9.891902104264646e-06,
      "loss": 0.173,
      "step": 1179
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5968698423880505,
      "learning_rate": 9.891634012959311e-06,
      "loss": 0.2746,
      "step": 1180
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.90291674074756,
      "learning_rate": 9.891365593263761e-06,
      "loss": 0.6693,
      "step": 1181
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.595639180848154,
      "learning_rate": 9.891096845196019e-06,
      "loss": 0.2098,
      "step": 1182
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4603592896229445,
      "learning_rate": 9.890827768774127e-06,
      "loss": 0.2356,
      "step": 1183
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5904912869445982,
      "learning_rate": 9.890558364016148e-06,
      "loss": 0.2794,
      "step": 1184
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6741059309875053,
      "learning_rate": 9.890288630940168e-06,
      "loss": 0.3042,
      "step": 1185
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4537682558757834,
      "learning_rate": 9.890018569564298e-06,
      "loss": 0.2207,
      "step": 1186
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5335758109164082,
      "learning_rate": 9.889748179906661e-06,
      "loss": 0.2129,
      "step": 1187
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.629553283848535,
      "learning_rate": 9.889477461985415e-06,
      "loss": 0.2292,
      "step": 1188
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4604753078543957,
      "learning_rate": 9.889206415818733e-06,
      "loss": 0.2264,
      "step": 1189
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4606776420321377,
      "learning_rate": 9.88893504142481e-06,
      "loss": 0.2162,
      "step": 1190
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1869958052882126,
      "learning_rate": 9.888663338821864e-06,
      "loss": 0.254,
      "step": 1191
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.560041437595407,
      "learning_rate": 9.888391308028138e-06,
      "loss": 0.2399,
      "step": 1192
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4748190213329984,
      "learning_rate": 9.888118949061891e-06,
      "loss": 0.271,
      "step": 1193
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3723794852552889,
      "learning_rate": 9.887846261941408e-06,
      "loss": 0.2061,
      "step": 1194
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.589319000495709,
      "learning_rate": 9.887573246684998e-06,
      "loss": 0.2008,
      "step": 1195
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.748045907144685,
      "learning_rate": 9.887299903310985e-06,
      "loss": 0.5921,
      "step": 1196
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6656061947855736,
      "learning_rate": 9.887026231837722e-06,
      "loss": 0.2534,
      "step": 1197
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5509884640833587,
      "learning_rate": 9.886752232283582e-06,
      "loss": 0.1893,
      "step": 1198
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.479228476949519,
      "learning_rate": 9.886477904666958e-06,
      "loss": 0.2075,
      "step": 1199
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5427920471763739,
      "learning_rate": 9.886203249006265e-06,
      "loss": 0.2431,
      "step": 1200
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4742759367376064,
      "learning_rate": 9.885928265319946e-06,
      "loss": 0.2351,
      "step": 1201
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5855799865481581,
      "learning_rate": 9.885652953626456e-06,
      "loss": 0.2062,
      "step": 1202
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6698415751519944,
      "learning_rate": 9.885377313944284e-06,
      "loss": 0.2245,
      "step": 1203
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3541746200863356,
      "learning_rate": 9.885101346291928e-06,
      "loss": 0.1951,
      "step": 1204
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4599680798670263,
      "learning_rate": 9.884825050687918e-06,
      "loss": 0.2622,
      "step": 1205
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5050691344117746,
      "learning_rate": 9.884548427150802e-06,
      "loss": 0.2155,
      "step": 1206
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6376583010376282,
      "learning_rate": 9.88427147569915e-06,
      "loss": 0.197,
      "step": 1207
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5941627822361515,
      "learning_rate": 9.883994196351555e-06,
      "loss": 0.2478,
      "step": 1208
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3789988160489755,
      "learning_rate": 9.883716589126633e-06,
      "loss": 0.1998,
      "step": 1209
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.687964050380182,
      "learning_rate": 9.883438654043019e-06,
      "loss": 0.2439,
      "step": 1210
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.119080356055434,
      "learning_rate": 9.88316039111937e-06,
      "loss": 0.832,
      "step": 1211
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.477191608850536,
      "learning_rate": 9.88288180037437e-06,
      "loss": 0.212,
      "step": 1212
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5368308370797685,
      "learning_rate": 9.882602881826721e-06,
      "loss": 0.2404,
      "step": 1213
    },
    {
      "epoch": 0.1,
      "grad_norm": 14.876471593532495,
      "learning_rate": 9.882323635495145e-06,
      "loss": 0.6998,
      "step": 1214
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4796850485491542,
      "learning_rate": 9.882044061398393e-06,
      "loss": 0.2424,
      "step": 1215
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5376896210134923,
      "learning_rate": 9.88176415955523e-06,
      "loss": 0.2467,
      "step": 1216
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5474008286208663,
      "learning_rate": 9.881483929984446e-06,
      "loss": 0.2429,
      "step": 1217
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2319481864748325,
      "learning_rate": 9.881203372704857e-06,
      "loss": 0.1809,
      "step": 1218
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3864755229772052,
      "learning_rate": 9.880922487735295e-06,
      "loss": 0.1924,
      "step": 1219
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.363777296476471,
      "learning_rate": 9.88064127509462e-06,
      "loss": 0.1499,
      "step": 1220
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4378195642987939,
      "learning_rate": 9.880359734801708e-06,
      "loss": 0.2205,
      "step": 1221
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7000927531120587,
      "learning_rate": 9.880077866875459e-06,
      "loss": 0.2608,
      "step": 1222
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.58547703977809,
      "learning_rate": 9.879795671334798e-06,
      "loss": 0.2806,
      "step": 1223
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5204246237789443,
      "learning_rate": 9.879513148198668e-06,
      "loss": 0.2339,
      "step": 1224
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.506551857601678,
      "learning_rate": 9.879230297486034e-06,
      "loss": 0.2114,
      "step": 1225
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.296145757799375,
      "learning_rate": 9.878947119215889e-06,
      "loss": 0.7773,
      "step": 1226
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4541957464519022,
      "learning_rate": 9.87866361340724e-06,
      "loss": 0.216,
      "step": 1227
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5021421019592718,
      "learning_rate": 9.878379780079122e-06,
      "loss": 0.2256,
      "step": 1228
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5568659697802163,
      "learning_rate": 9.878095619250588e-06,
      "loss": 0.1983,
      "step": 1229
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5945908895129153,
      "learning_rate": 9.877811130940715e-06,
      "loss": 0.2526,
      "step": 1230
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6393224547703271,
      "learning_rate": 9.8775263151686e-06,
      "loss": 0.2415,
      "step": 1231
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3882238251581343,
      "learning_rate": 9.877241171953367e-06,
      "loss": 0.1852,
      "step": 1232
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5142479666155222,
      "learning_rate": 9.876955701314157e-06,
      "loss": 0.2546,
      "step": 1233
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.656882863325662,
      "learning_rate": 9.876669903270133e-06,
      "loss": 0.6298,
      "step": 1234
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3810587229994047,
      "learning_rate": 9.876383777840484e-06,
      "loss": 0.1984,
      "step": 1235
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6805618016968162,
      "learning_rate": 9.876097325044416e-06,
      "loss": 0.2811,
      "step": 1236
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2961923253721568,
      "learning_rate": 9.87581054490116e-06,
      "loss": 0.2063,
      "step": 1237
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6457488650821899,
      "learning_rate": 9.87552343742997e-06,
      "loss": 0.3113,
      "step": 1238
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5313417194840961,
      "learning_rate": 9.875236002650119e-06,
      "loss": 0.2159,
      "step": 1239
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.353590915230694,
      "learning_rate": 9.874948240580903e-06,
      "loss": 0.1823,
      "step": 1240
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.507847020727077,
      "learning_rate": 9.874660151241644e-06,
      "loss": 0.2216,
      "step": 1241
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.4961928046658075,
      "learning_rate": 9.874371734651678e-06,
      "loss": 0.3865,
      "step": 1242
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7216839037715528,
      "learning_rate": 9.874082990830366e-06,
      "loss": 0.2561,
      "step": 1243
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.53656246277392,
      "learning_rate": 9.873793919797099e-06,
      "loss": 0.214,
      "step": 1244
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4365897369027967,
      "learning_rate": 9.873504521571278e-06,
      "loss": 0.2261,
      "step": 1245
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5472450575095278,
      "learning_rate": 9.87321479617233e-06,
      "loss": 0.2678,
      "step": 1246
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6981838623952037,
      "learning_rate": 9.87292474361971e-06,
      "loss": 0.2521,
      "step": 1247
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.639554932401539,
      "learning_rate": 9.872634363932887e-06,
      "loss": 0.2632,
      "step": 1248
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4520043958718585,
      "learning_rate": 9.872343657131355e-06,
      "loss": 0.2183,
      "step": 1249
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3424890791931532,
      "learning_rate": 9.872052623234632e-06,
      "loss": 0.2004,
      "step": 1250
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4162601855887524,
      "learning_rate": 9.871761262262252e-06,
      "loss": 0.2306,
      "step": 1251
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5179272074142474,
      "learning_rate": 9.871469574233781e-06,
      "loss": 0.229,
      "step": 1252
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.520275733372812,
      "learning_rate": 9.871177559168795e-06,
      "loss": 0.224,
      "step": 1253
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5287062683156647,
      "learning_rate": 9.8708852170869e-06,
      "loss": 0.2204,
      "step": 1254
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.66286348628413,
      "learning_rate": 9.870592548007725e-06,
      "loss": 0.2682,
      "step": 1255
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5805089451503687,
      "learning_rate": 9.870299551950912e-06,
      "loss": 0.2079,
      "step": 1256
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.468115576201194,
      "learning_rate": 9.870006228936135e-06,
      "loss": 0.2112,
      "step": 1257
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5114860861271562,
      "learning_rate": 9.869712578983085e-06,
      "loss": 0.2061,
      "step": 1258
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5385575720939868,
      "learning_rate": 9.869418602111475e-06,
      "loss": 0.1992,
      "step": 1259
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.616794660571654,
      "learning_rate": 9.869124298341039e-06,
      "loss": 0.2449,
      "step": 1260
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.179466654330429,
      "learning_rate": 9.868829667691538e-06,
      "loss": 0.8159,
      "step": 1261
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6874042904014803,
      "learning_rate": 9.868534710182747e-06,
      "loss": 0.212,
      "step": 1262
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6645803845864144,
      "learning_rate": 9.868239425834472e-06,
      "loss": 0.2672,
      "step": 1263
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.627170207393232,
      "learning_rate": 9.867943814666533e-06,
      "loss": 0.257,
      "step": 1264
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5782840856787934,
      "learning_rate": 9.867647876698776e-06,
      "loss": 0.2831,
      "step": 1265
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.07227715987488,
      "learning_rate": 9.867351611951071e-06,
      "loss": 0.4821,
      "step": 1266
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.558220330838703,
      "learning_rate": 9.867055020443302e-06,
      "loss": 0.2328,
      "step": 1267
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6552393255471114,
      "learning_rate": 9.866758102195384e-06,
      "loss": 0.2035,
      "step": 1268
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6111335874381598,
      "learning_rate": 9.86646085722725e-06,
      "loss": 0.2478,
      "step": 1269
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.568783215246882,
      "learning_rate": 9.866163285558851e-06,
      "loss": 0.2278,
      "step": 1270
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5369253250884514,
      "learning_rate": 9.865865387210169e-06,
      "loss": 0.206,
      "step": 1271
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.601777102209474,
      "learning_rate": 9.8655671622012e-06,
      "loss": 0.2607,
      "step": 1272
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.524083001926831,
      "learning_rate": 9.865268610551966e-06,
      "loss": 0.276,
      "step": 1273
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6077407560378083,
      "learning_rate": 9.864969732282507e-06,
      "loss": 0.2338,
      "step": 1274
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.544743916225948,
      "learning_rate": 9.864670527412891e-06,
      "loss": 0.2236,
      "step": 1275
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5352981573532034,
      "learning_rate": 9.864370995963204e-06,
      "loss": 0.2372,
      "step": 1276
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6596131906689564,
      "learning_rate": 9.864071137953552e-06,
      "loss": 0.2142,
      "step": 1277
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.401168154543283,
      "learning_rate": 9.863770953404068e-06,
      "loss": 0.618,
      "step": 1278
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6881806004593924,
      "learning_rate": 9.863470442334904e-06,
      "loss": 0.2489,
      "step": 1279
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.270452032620281,
      "learning_rate": 9.863169604766231e-06,
      "loss": 0.5501,
      "step": 1280
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.921263308335348,
      "learning_rate": 9.86286844071825e-06,
      "loss": 0.5891,
      "step": 1281
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6326002299720224,
      "learning_rate": 9.862566950211175e-06,
      "loss": 0.2087,
      "step": 1282
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1613978573095491,
      "learning_rate": 9.862265133265248e-06,
      "loss": 0.1747,
      "step": 1283
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7024187966796624,
      "learning_rate": 9.861962989900732e-06,
      "loss": 0.2747,
      "step": 1284
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5568553565742755,
      "learning_rate": 9.861660520137908e-06,
      "loss": 0.2461,
      "step": 1285
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7646407229744727,
      "learning_rate": 9.861357723997082e-06,
      "loss": 0.2798,
      "step": 1286
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5202771441452385,
      "learning_rate": 9.861054601498586e-06,
      "loss": 0.2126,
      "step": 1287
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.495939184098469,
      "learning_rate": 9.860751152662762e-06,
      "loss": 0.2268,
      "step": 1288
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.600047661816389,
      "learning_rate": 9.860447377509989e-06,
      "loss": 0.2031,
      "step": 1289
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3784986845475653,
      "learning_rate": 9.860143276060655e-06,
      "loss": 0.2008,
      "step": 1290
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4369934389728147,
      "learning_rate": 9.859838848335178e-06,
      "loss": 0.204,
      "step": 1291
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5247123382267085,
      "learning_rate": 9.859534094353994e-06,
      "loss": 0.2388,
      "step": 1292
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5214889060063344,
      "learning_rate": 9.859229014137564e-06,
      "loss": 0.2478,
      "step": 1293
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5837295621394827,
      "learning_rate": 9.858923607706366e-06,
      "loss": 0.233,
      "step": 1294
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4327328130112271,
      "learning_rate": 9.858617875080904e-06,
      "loss": 0.2551,
      "step": 1295
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.712899783157388,
      "learning_rate": 9.858311816281703e-06,
      "loss": 0.2308,
      "step": 1296
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6467437562054286,
      "learning_rate": 9.858005431329309e-06,
      "loss": 0.2525,
      "step": 1297
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4168690711237977,
      "learning_rate": 9.857698720244294e-06,
      "loss": 0.189,
      "step": 1298
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4449466105721798,
      "learning_rate": 9.857391683047244e-06,
      "loss": 0.2031,
      "step": 1299
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.554423339121032,
      "learning_rate": 9.857084319758772e-06,
      "loss": 0.2293,
      "step": 1300
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.51829907476096,
      "learning_rate": 9.856776630399514e-06,
      "loss": 0.4161,
      "step": 1301
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5682420260526695,
      "learning_rate": 9.856468614990127e-06,
      "loss": 0.2804,
      "step": 1302
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4309682589456088,
      "learning_rate": 9.856160273551285e-06,
      "loss": 0.2023,
      "step": 1303
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4918386482698014,
      "learning_rate": 9.855851606103691e-06,
      "loss": 0.2125,
      "step": 1304
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3802167648011263,
      "learning_rate": 9.855542612668066e-06,
      "loss": 0.2315,
      "step": 1305
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6220486547133197,
      "learning_rate": 9.855233293265153e-06,
      "loss": 0.261,
      "step": 1306
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.043414255382233,
      "learning_rate": 9.85492364791572e-06,
      "loss": 0.4871,
      "step": 1307
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.503732509748684,
      "learning_rate": 9.854613676640551e-06,
      "loss": 0.2246,
      "step": 1308
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6049608124902606,
      "learning_rate": 9.854303379460458e-06,
      "loss": 0.2403,
      "step": 1309
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4643366977450483,
      "learning_rate": 9.853992756396272e-06,
      "loss": 0.2279,
      "step": 1310
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6054568792744657,
      "learning_rate": 9.853681807468845e-06,
      "loss": 0.2901,
      "step": 1311
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5145162783211317,
      "learning_rate": 9.853370532699052e-06,
      "loss": 0.2082,
      "step": 1312
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6343484845675553,
      "learning_rate": 9.853058932107789e-06,
      "loss": 0.207,
      "step": 1313
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5182533982877526,
      "learning_rate": 9.852747005715976e-06,
      "loss": 0.2309,
      "step": 1314
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6377338184254033,
      "learning_rate": 9.852434753544552e-06,
      "loss": 0.2157,
      "step": 1315
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7927427690368656,
      "learning_rate": 9.852122175614484e-06,
      "loss": 0.2108,
      "step": 1316
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5413721558837432,
      "learning_rate": 9.85180927194675e-06,
      "loss": 0.2269,
      "step": 1317
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.604072206692453,
      "learning_rate": 9.85149604256236e-06,
      "loss": 0.7057,
      "step": 1318
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6518681112436822,
      "learning_rate": 9.851182487482342e-06,
      "loss": 0.2321,
      "step": 1319
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6176386535258809,
      "learning_rate": 9.850868606727745e-06,
      "loss": 0.2977,
      "step": 1320
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5163833273166403,
      "learning_rate": 9.85055440031964e-06,
      "loss": 0.2142,
      "step": 1321
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4593753709401616,
      "learning_rate": 9.850239868279123e-06,
      "loss": 0.213,
      "step": 1322
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3948648754092454,
      "learning_rate": 9.849925010627308e-06,
      "loss": 0.1798,
      "step": 1323
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6110357416701928,
      "learning_rate": 9.84960982738533e-06,
      "loss": 0.2632,
      "step": 1324
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.114977292061194,
      "learning_rate": 9.849294318574353e-06,
      "loss": 0.6103,
      "step": 1325
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7754173369612274,
      "learning_rate": 9.848978484215554e-06,
      "loss": 0.2336,
      "step": 1326
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5000966746851923,
      "learning_rate": 9.848662324330139e-06,
      "loss": 0.2238,
      "step": 1327
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5459688275196888,
      "learning_rate": 9.848345838939329e-06,
      "loss": 0.2535,
      "step": 1328
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.445380477310344,
      "learning_rate": 9.848029028064374e-06,
      "loss": 0.2384,
      "step": 1329
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3401213196546187,
      "learning_rate": 9.847711891726543e-06,
      "loss": 0.1965,
      "step": 1330
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.544165120521194,
      "learning_rate": 9.847394429947124e-06,
      "loss": 0.2514,
      "step": 1331
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5187169729622347,
      "learning_rate": 9.847076642747429e-06,
      "loss": 0.2059,
      "step": 1332
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.450421403812484,
      "learning_rate": 9.846758530148793e-06,
      "loss": 0.1944,
      "step": 1333
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.378706091145846,
      "learning_rate": 9.84644009217257e-06,
      "loss": 0.1828,
      "step": 1334
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6106828071763193,
      "learning_rate": 9.846121328840143e-06,
      "loss": 0.2306,
      "step": 1335
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5249295372187086,
      "learning_rate": 9.845802240172908e-06,
      "loss": 0.2268,
      "step": 1336
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7422706892672513,
      "learning_rate": 9.845482826192284e-06,
      "loss": 0.2568,
      "step": 1337
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5630870214770054,
      "learning_rate": 9.845163086919718e-06,
      "loss": 0.2682,
      "step": 1338
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4423051259483073,
      "learning_rate": 9.844843022376673e-06,
      "loss": 0.2132,
      "step": 1339
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4945468977657284,
      "learning_rate": 9.844522632584636e-06,
      "loss": 0.2286,
      "step": 1340
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4811032687179086,
      "learning_rate": 9.844201917565119e-06,
      "loss": 0.2474,
      "step": 1341
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7351672465107948,
      "learning_rate": 9.843880877339648e-06,
      "loss": 0.2685,
      "step": 1342
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4098817650089928,
      "learning_rate": 9.843559511929777e-06,
      "loss": 0.2058,
      "step": 1343
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6030518121214643,
      "learning_rate": 9.843237821357082e-06,
      "loss": 0.2336,
      "step": 1344
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.492303424991774,
      "learning_rate": 9.842915805643156e-06,
      "loss": 0.2196,
      "step": 1345
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4148967167561444,
      "learning_rate": 9.84259346480962e-06,
      "loss": 0.2448,
      "step": 1346
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4332362316202147,
      "learning_rate": 9.842270798878111e-06,
      "loss": 0.202,
      "step": 1347
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.46852464059966,
      "learning_rate": 9.841947807870293e-06,
      "loss": 0.261,
      "step": 1348
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5477143630663952,
      "learning_rate": 9.841624491807846e-06,
      "loss": 0.2476,
      "step": 1349
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7733072106554304,
      "learning_rate": 9.841300850712479e-06,
      "loss": 0.2717,
      "step": 1350
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8632456287428134,
      "learning_rate": 9.840976884605916e-06,
      "loss": 0.2143,
      "step": 1351
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0008596053827428,
      "learning_rate": 9.840652593509909e-06,
      "loss": 0.2373,
      "step": 1352
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.652903898881646,
      "learning_rate": 9.840327977446226e-06,
      "loss": 0.6222,
      "step": 1353
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.846933979826645,
      "learning_rate": 9.840003036436661e-06,
      "loss": 0.585,
      "step": 1354
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3586585374788638,
      "learning_rate": 9.839677770503028e-06,
      "loss": 0.1728,
      "step": 1355
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6568698252455725,
      "learning_rate": 9.839352179667162e-06,
      "loss": 0.2092,
      "step": 1356
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5432193582207006,
      "learning_rate": 9.83902626395092e-06,
      "loss": 0.231,
      "step": 1357
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5448501122645313,
      "learning_rate": 9.838700023376184e-06,
      "loss": 0.2109,
      "step": 1358
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.342338369806452,
      "learning_rate": 9.838373457964856e-06,
      "loss": 0.6459,
      "step": 1359
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.20287931982747,
      "learning_rate": 9.838046567738856e-06,
      "loss": 0.4659,
      "step": 1360
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.297105111468454,
      "learning_rate": 9.837719352720133e-06,
      "loss": 0.6487,
      "step": 1361
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.619963408462713,
      "learning_rate": 9.83739181293065e-06,
      "loss": 0.2942,
      "step": 1362
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4349394650160046,
      "learning_rate": 9.837063948392401e-06,
      "loss": 0.2021,
      "step": 1363
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2944483742902744,
      "learning_rate": 9.836735759127391e-06,
      "loss": 0.1804,
      "step": 1364
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.442794120773668,
      "learning_rate": 9.836407245157656e-06,
      "loss": 0.2095,
      "step": 1365
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6739206900543397,
      "learning_rate": 9.836078406505249e-06,
      "loss": 0.2272,
      "step": 1366
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4887511961771356,
      "learning_rate": 9.835749243192245e-06,
      "loss": 0.2296,
      "step": 1367
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.33437153770658,
      "learning_rate": 9.835419755240743e-06,
      "loss": 0.4742,
      "step": 1368
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4765551147247555,
      "learning_rate": 9.835089942672862e-06,
      "loss": 0.2583,
      "step": 1369
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4567166693949587,
      "learning_rate": 9.834759805510742e-06,
      "loss": 0.2246,
      "step": 1370
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6546526285836909,
      "learning_rate": 9.834429343776551e-06,
      "loss": 0.2808,
      "step": 1371
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.460627681075815,
      "learning_rate": 9.834098557492467e-06,
      "loss": 0.2526,
      "step": 1372
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4086192899309873,
      "learning_rate": 9.8337674466807e-06,
      "loss": 0.2243,
      "step": 1373
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5091255987601904,
      "learning_rate": 9.833436011363482e-06,
      "loss": 0.2042,
      "step": 1374
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6167079637260537,
      "learning_rate": 9.833104251563058e-06,
      "loss": 0.2604,
      "step": 1375
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4291148760164998,
      "learning_rate": 9.832772167301701e-06,
      "loss": 0.208,
      "step": 1376
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.755968490273507,
      "learning_rate": 9.832439758601706e-06,
      "loss": 0.5376,
      "step": 1377
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.184471085781334,
      "learning_rate": 9.83210702548539e-06,
      "loss": 0.5408,
      "step": 1378
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5989298590823278,
      "learning_rate": 9.831773967975085e-06,
      "loss": 0.2302,
      "step": 1379
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.3485455377460465,
      "learning_rate": 9.831440586093157e-06,
      "loss": 0.4865,
      "step": 1380
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.773040260322894,
      "learning_rate": 9.831106879861982e-06,
      "loss": 0.6059,
      "step": 1381
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4972312019509375,
      "learning_rate": 9.830772849303967e-06,
      "loss": 0.248,
      "step": 1382
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5611347541717178,
      "learning_rate": 9.830438494441533e-06,
      "loss": 0.2491,
      "step": 1383
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2940051640247117,
      "learning_rate": 9.830103815297126e-06,
      "loss": 0.1694,
      "step": 1384
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.551032344180417,
      "learning_rate": 9.829768811893214e-06,
      "loss": 0.2361,
      "step": 1385
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5416562841956336,
      "learning_rate": 9.829433484252292e-06,
      "loss": 0.2104,
      "step": 1386
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.585628610240876,
      "learning_rate": 9.829097832396864e-06,
      "loss": 0.2189,
      "step": 1387
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6828530326464681,
      "learning_rate": 9.82876185634947e-06,
      "loss": 0.2551,
      "step": 1388
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4639213732594658,
      "learning_rate": 9.828425556132659e-06,
      "loss": 0.2113,
      "step": 1389
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5468244789615329,
      "learning_rate": 9.828088931769012e-06,
      "loss": 0.2484,
      "step": 1390
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7052255095411852,
      "learning_rate": 9.827751983281126e-06,
      "loss": 0.2284,
      "step": 1391
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3946262694016962,
      "learning_rate": 9.827414710691624e-06,
      "loss": 0.2139,
      "step": 1392
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5706864110951968,
      "learning_rate": 9.827077114023145e-06,
      "loss": 0.2303,
      "step": 1393
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.8467701436448705,
      "learning_rate": 9.826739193298353e-06,
      "loss": 0.4653,
      "step": 1394
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.377439390234066,
      "learning_rate": 9.826400948539935e-06,
      "loss": 0.1939,
      "step": 1395
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3667716905354454,
      "learning_rate": 9.826062379770598e-06,
      "loss": 0.1824,
      "step": 1396
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.826233858029706,
      "learning_rate": 9.82572348701307e-06,
      "loss": 0.529,
      "step": 1397
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.38865560960334,
      "learning_rate": 9.825384270290104e-06,
      "loss": 0.2135,
      "step": 1398
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5217516109068419,
      "learning_rate": 9.825044729624472e-06,
      "loss": 0.2249,
      "step": 1399
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5573089448173083,
      "learning_rate": 9.824704865038967e-06,
      "loss": 0.2469,
      "step": 1400
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4939652856714627,
      "learning_rate": 9.824364676556406e-06,
      "loss": 0.1969,
      "step": 1401
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.417313856655264,
      "learning_rate": 9.824024164199627e-06,
      "loss": 0.5295,
      "step": 1402
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5893032806163254,
      "learning_rate": 9.823683327991492e-06,
      "loss": 0.2426,
      "step": 1403
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.530048069706559,
      "learning_rate": 9.82334216795488e-06,
      "loss": 0.2411,
      "step": 1404
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5691111762677317,
      "learning_rate": 9.823000684112691e-06,
      "loss": 0.2371,
      "step": 1405
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5740814661995475,
      "learning_rate": 9.822658876487854e-06,
      "loss": 0.2373,
      "step": 1406
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4544963366575885,
      "learning_rate": 9.822316745103316e-06,
      "loss": 0.2155,
      "step": 1407
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.78182010890901,
      "learning_rate": 9.821974289982042e-06,
      "loss": 0.2717,
      "step": 1408
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3542350353714456,
      "learning_rate": 9.821631511147025e-06,
      "loss": 0.2226,
      "step": 1409
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.630477332425947,
      "learning_rate": 9.821288408621276e-06,
      "loss": 0.2675,
      "step": 1410
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.487475758372642,
      "learning_rate": 9.820944982427826e-06,
      "loss": 0.2359,
      "step": 1411
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.471869211768212,
      "learning_rate": 9.820601232589735e-06,
      "loss": 0.189,
      "step": 1412
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3654992248220554,
      "learning_rate": 9.820257159130076e-06,
      "loss": 0.196,
      "step": 1413
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4416320816628994,
      "learning_rate": 9.81991276207195e-06,
      "loss": 0.214,
      "step": 1414
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3953485014430456,
      "learning_rate": 9.819568041438477e-06,
      "loss": 0.218,
      "step": 1415
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4661686834911387,
      "learning_rate": 9.819222997252798e-06,
      "loss": 0.1914,
      "step": 1416
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5321069518731856,
      "learning_rate": 9.818877629538077e-06,
      "loss": 0.29,
      "step": 1417
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.610913915200411,
      "learning_rate": 9.818531938317499e-06,
      "loss": 0.6348,
      "step": 1418
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5830589045520829,
      "learning_rate": 9.818185923614274e-06,
      "loss": 0.2326,
      "step": 1419
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4443262685480311,
      "learning_rate": 9.817839585451629e-06,
      "loss": 0.2303,
      "step": 1420
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4140632364202714,
      "learning_rate": 9.817492923852817e-06,
      "loss": 0.2143,
      "step": 1421
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.50414358071682,
      "learning_rate": 9.817145938841106e-06,
      "loss": 0.2229,
      "step": 1422
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5038105686779784,
      "learning_rate": 9.816798630439794e-06,
      "loss": 0.2289,
      "step": 1423
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5601935200054282,
      "learning_rate": 9.816450998672195e-06,
      "loss": 0.248,
      "step": 1424
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4408127622582956,
      "learning_rate": 9.816103043561648e-06,
      "loss": 0.2338,
      "step": 1425
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.770503382532695,
      "learning_rate": 9.815754765131511e-06,
      "loss": 0.2582,
      "step": 1426
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5456056631965465,
      "learning_rate": 9.815406163405165e-06,
      "loss": 0.2285,
      "step": 1427
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5105420645521708,
      "learning_rate": 9.815057238406015e-06,
      "loss": 0.2499,
      "step": 1428
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2713021279102137,
      "learning_rate": 9.814707990157482e-06,
      "loss": 0.1608,
      "step": 1429
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6944330333303748,
      "learning_rate": 9.814358418683014e-06,
      "loss": 0.2508,
      "step": 1430
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4603186588033277,
      "learning_rate": 9.814008524006077e-06,
      "loss": 0.216,
      "step": 1431
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7791581265113101,
      "learning_rate": 9.813658306150164e-06,
      "loss": 0.2485,
      "step": 1432
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6145543915863054,
      "learning_rate": 9.813307765138784e-06,
      "loss": 0.2598,
      "step": 1433
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6239851367449045,
      "learning_rate": 9.81295690099547e-06,
      "loss": 0.2501,
      "step": 1434
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3407668059144913,
      "learning_rate": 9.812605713743775e-06,
      "loss": 0.2135,
      "step": 1435
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6451002778773953,
      "learning_rate": 9.812254203407278e-06,
      "loss": 0.2957,
      "step": 1436
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3738871085555746,
      "learning_rate": 9.811902370009576e-06,
      "loss": 0.2228,
      "step": 1437
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2018390035675401,
      "learning_rate": 9.811550213574287e-06,
      "loss": 0.1826,
      "step": 1438
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4340336457735143,
      "learning_rate": 9.811197734125055e-06,
      "loss": 0.2009,
      "step": 1439
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.601991205647889,
      "learning_rate": 9.810844931685542e-06,
      "loss": 0.2543,
      "step": 1440
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6402376946753459,
      "learning_rate": 9.810491806279432e-06,
      "loss": 0.2725,
      "step": 1441
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7149124662662572,
      "learning_rate": 9.81013835793043e-06,
      "loss": 0.2511,
      "step": 1442
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.581572668740855,
      "learning_rate": 9.809784586662268e-06,
      "loss": 0.2466,
      "step": 1443
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4928160289781764,
      "learning_rate": 9.809430492498693e-06,
      "loss": 0.21,
      "step": 1444
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7698665858693028,
      "learning_rate": 9.809076075463476e-06,
      "loss": 0.2557,
      "step": 1445
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.276943382726287,
      "learning_rate": 9.808721335580414e-06,
      "loss": 0.1847,
      "step": 1446
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.405350957138464,
      "learning_rate": 9.808366272873317e-06,
      "loss": 0.661,
      "step": 1447
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.824441567134671,
      "learning_rate": 9.808010887366024e-06,
      "loss": 0.2549,
      "step": 1448
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3207735091632264,
      "learning_rate": 9.807655179082392e-06,
      "loss": 0.1645,
      "step": 1449
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3159725071161843,
      "learning_rate": 9.807299148046301e-06,
      "loss": 0.1507,
      "step": 1450
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5284544676265406,
      "learning_rate": 9.806942794281654e-06,
      "loss": 0.2146,
      "step": 1451
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3646302306878015,
      "learning_rate": 9.80658611781237e-06,
      "loss": 0.23,
      "step": 1452
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4778281106495794,
      "learning_rate": 9.806229118662398e-06,
      "loss": 0.2029,
      "step": 1453
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4820536704594791,
      "learning_rate": 9.805871796855704e-06,
      "loss": 0.216,
      "step": 1454
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4880666126792284,
      "learning_rate": 9.805514152416274e-06,
      "loss": 0.225,
      "step": 1455
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5381085612039622,
      "learning_rate": 9.80515618536812e-06,
      "loss": 0.2384,
      "step": 1456
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6104170851049522,
      "learning_rate": 9.80479789573527e-06,
      "loss": 0.26,
      "step": 1457
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.792092933624506,
      "learning_rate": 9.804439283541781e-06,
      "loss": 0.6425,
      "step": 1458
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.200474311649978,
      "learning_rate": 9.804080348811725e-06,
      "loss": 0.7019,
      "step": 1459
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7435265336733736,
      "learning_rate": 9.803721091569201e-06,
      "loss": 0.2592,
      "step": 1460
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.89886811005992,
      "learning_rate": 9.803361511838324e-06,
      "loss": 0.6341,
      "step": 1461
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.365379346529135,
      "learning_rate": 9.803001609643234e-06,
      "loss": 0.2099,
      "step": 1462
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3343566353898706,
      "learning_rate": 9.802641385008096e-06,
      "loss": 0.1868,
      "step": 1463
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.515053796543866,
      "learning_rate": 9.80228083795709e-06,
      "loss": 0.2034,
      "step": 1464
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6269161043224634,
      "learning_rate": 9.80191996851442e-06,
      "loss": 0.211,
      "step": 1465
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4292789078983759,
      "learning_rate": 9.801558776704315e-06,
      "loss": 0.2349,
      "step": 1466
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5690316736296728,
      "learning_rate": 9.801197262551019e-06,
      "loss": 0.2026,
      "step": 1467
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.7698241411753,
      "learning_rate": 9.800835426078804e-06,
      "loss": 0.614,
      "step": 1468
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.367120240972899,
      "learning_rate": 9.800473267311962e-06,
      "loss": 0.2342,
      "step": 1469
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4721811650345396,
      "learning_rate": 9.800110786274803e-06,
      "loss": 0.2247,
      "step": 1470
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.055228939688422,
      "learning_rate": 9.799747982991665e-06,
      "loss": 0.7436,
      "step": 1471
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4344985045654202,
      "learning_rate": 9.799384857486902e-06,
      "loss": 0.2091,
      "step": 1472
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3500369241321077,
      "learning_rate": 9.799021409784892e-06,
      "loss": 0.206,
      "step": 1473
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4754295176965464,
      "learning_rate": 9.798657639910033e-06,
      "loss": 0.2614,
      "step": 1474
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5538437626574393,
      "learning_rate": 9.798293547886748e-06,
      "loss": 0.2396,
      "step": 1475
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4867721451291298,
      "learning_rate": 9.79792913373948e-06,
      "loss": 0.2459,
      "step": 1476
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4662795235305603,
      "learning_rate": 9.79756439749269e-06,
      "loss": 0.2131,
      "step": 1477
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5728136361085496,
      "learning_rate": 9.797199339170866e-06,
      "loss": 0.2354,
      "step": 1478
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5014828919693486,
      "learning_rate": 9.796833958798517e-06,
      "loss": 0.1893,
      "step": 1479
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5643366021672795,
      "learning_rate": 9.79646825640017e-06,
      "loss": 0.154,
      "step": 1480
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3849792979232767,
      "learning_rate": 9.796102232000378e-06,
      "loss": 0.1897,
      "step": 1481
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3507529912131202,
      "learning_rate": 9.795735885623708e-06,
      "loss": 0.199,
      "step": 1482
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5222848732322185,
      "learning_rate": 9.795369217294759e-06,
      "loss": 0.2461,
      "step": 1483
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.095961786416891,
      "learning_rate": 9.795002227038146e-06,
      "loss": 0.6301,
      "step": 1484
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6102115047438696,
      "learning_rate": 9.794634914878505e-06,
      "loss": 0.2131,
      "step": 1485
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.384163364613551,
      "learning_rate": 9.794267280840494e-06,
      "loss": 0.202,
      "step": 1486
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.980561450118831,
      "learning_rate": 9.793899324948795e-06,
      "loss": 0.5201,
      "step": 1487
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7045279102154405,
      "learning_rate": 9.79353104722811e-06,
      "loss": 0.2522,
      "step": 1488
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4721532606486294,
      "learning_rate": 9.793162447703161e-06,
      "loss": 0.2297,
      "step": 1489
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5583306675273565,
      "learning_rate": 9.792793526398694e-06,
      "loss": 0.2572,
      "step": 1490
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6535601309701211,
      "learning_rate": 9.792424283339477e-06,
      "loss": 0.2349,
      "step": 1491
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6238342068170448,
      "learning_rate": 9.792054718550297e-06,
      "loss": 0.3,
      "step": 1492
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5620780458702739,
      "learning_rate": 9.791684832055962e-06,
      "loss": 0.2115,
      "step": 1493
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4673245768704077,
      "learning_rate": 9.79131462388131e-06,
      "loss": 0.2303,
      "step": 1494
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5205843174141263,
      "learning_rate": 9.790944094051188e-06,
      "loss": 0.1741,
      "step": 1495
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.617888816866139,
      "learning_rate": 9.790573242590473e-06,
      "loss": 0.2517,
      "step": 1496
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6754660394184404,
      "learning_rate": 9.790202069524061e-06,
      "loss": 0.2622,
      "step": 1497
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5456154240653932,
      "learning_rate": 9.789830574876873e-06,
      "loss": 0.2715,
      "step": 1498
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.2915401357377725,
      "learning_rate": 9.789458758673843e-06,
      "loss": 0.606,
      "step": 1499
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7166910882738635,
      "learning_rate": 9.789086620939936e-06,
      "loss": 0.2637,
      "step": 1500
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5005129863346887,
      "learning_rate": 9.788714161700135e-06,
      "loss": 0.2689,
      "step": 1501
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3478767869455088,
      "learning_rate": 9.78834138097944e-06,
      "loss": 0.2587,
      "step": 1502
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3746996540686276,
      "learning_rate": 9.787968278802883e-06,
      "loss": 0.2149,
      "step": 1503
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.44408894601983,
      "learning_rate": 9.787594855195509e-06,
      "loss": 0.2194,
      "step": 1504
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.376583673021218,
      "learning_rate": 9.787221110182384e-06,
      "loss": 0.2454,
      "step": 1505
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.397941276924308,
      "learning_rate": 9.786847043788601e-06,
      "loss": 0.6664,
      "step": 1506
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6146693771277414,
      "learning_rate": 9.786472656039275e-06,
      "loss": 0.2427,
      "step": 1507
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6544197733435781,
      "learning_rate": 9.786097946959534e-06,
      "loss": 0.2751,
      "step": 1508
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5076062634960605,
      "learning_rate": 9.785722916574539e-06,
      "loss": 0.167,
      "step": 1509
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4068663414249152,
      "learning_rate": 9.785347564909464e-06,
      "loss": 0.1855,
      "step": 1510
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.927890560023354,
      "learning_rate": 9.784971891989508e-06,
      "loss": 0.6761,
      "step": 1511
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.473618323153323,
      "learning_rate": 9.78459589783989e-06,
      "loss": 0.5578,
      "step": 1512
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4826105830934497,
      "learning_rate": 9.784219582485853e-06,
      "loss": 0.2144,
      "step": 1513
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4421224805808652,
      "learning_rate": 9.78384294595266e-06,
      "loss": 0.224,
      "step": 1514
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4868483772962446,
      "learning_rate": 9.783465988265594e-06,
      "loss": 0.2246,
      "step": 1515
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6342851575571882,
      "learning_rate": 9.783088709449967e-06,
      "loss": 0.2121,
      "step": 1516
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5792765679981189,
      "learning_rate": 9.7827111095311e-06,
      "loss": 0.2372,
      "step": 1517
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6149526191605383,
      "learning_rate": 9.782333188534345e-06,
      "loss": 0.2339,
      "step": 1518
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4198423148094563,
      "learning_rate": 9.781954946485072e-06,
      "loss": 0.2085,
      "step": 1519
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.417247206892955,
      "learning_rate": 9.781576383408678e-06,
      "loss": 0.1626,
      "step": 1520
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4668535926425135,
      "learning_rate": 9.781197499330572e-06,
      "loss": 0.1678,
      "step": 1521
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5649086717820815,
      "learning_rate": 9.78081829427619e-06,
      "loss": 0.2426,
      "step": 1522
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.472924529567563,
      "learning_rate": 9.780438768270992e-06,
      "loss": 0.6806,
      "step": 1523
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.370004457147356,
      "learning_rate": 9.780058921340456e-06,
      "loss": 0.2394,
      "step": 1524
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.459787777712049,
      "learning_rate": 9.779678753510082e-06,
      "loss": 0.668,
      "step": 1525
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4576166496865888,
      "learning_rate": 9.77929826480539e-06,
      "loss": 0.1966,
      "step": 1526
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4629814851551488,
      "learning_rate": 9.778917455251924e-06,
      "loss": 0.2657,
      "step": 1527
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.55569845763157,
      "learning_rate": 9.778536324875252e-06,
      "loss": 0.2848,
      "step": 1528
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3684044127238737,
      "learning_rate": 9.778154873700956e-06,
      "loss": 0.2078,
      "step": 1529
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5414397128211301,
      "learning_rate": 9.777773101754648e-06,
      "loss": 0.2835,
      "step": 1530
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3758651514609745,
      "learning_rate": 9.777391009061954e-06,
      "loss": 0.208,
      "step": 1531
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.9448235189303915,
      "learning_rate": 9.777008595648527e-06,
      "loss": 0.6133,
      "step": 1532
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5865145755859187,
      "learning_rate": 9.77662586154004e-06,
      "loss": 0.2833,
      "step": 1533
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5179452983165114,
      "learning_rate": 9.776242806762187e-06,
      "loss": 0.2791,
      "step": 1534
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6299484277744316,
      "learning_rate": 9.775859431340681e-06,
      "loss": 0.1934,
      "step": 1535
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4114162741359657,
      "learning_rate": 9.775475735301261e-06,
      "loss": 0.2088,
      "step": 1536
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3334542333679187,
      "learning_rate": 9.775091718669688e-06,
      "loss": 0.1901,
      "step": 1537
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5763527019236323,
      "learning_rate": 9.774707381471737e-06,
      "loss": 0.2211,
      "step": 1538
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7234397146890839,
      "learning_rate": 9.774322723733216e-06,
      "loss": 0.2823,
      "step": 1539
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6165818412590764,
      "learning_rate": 9.773937745479942e-06,
      "loss": 0.2321,
      "step": 1540
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2934389720843433,
      "learning_rate": 9.773552446737764e-06,
      "loss": 0.2037,
      "step": 1541
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4628982851816554,
      "learning_rate": 9.773166827532548e-06,
      "loss": 0.217,
      "step": 1542
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5175023983306994,
      "learning_rate": 9.77278088789018e-06,
      "loss": 0.2786,
      "step": 1543
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5691679386620183,
      "learning_rate": 9.77239462783657e-06,
      "loss": 0.2438,
      "step": 1544
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5579485036314944,
      "learning_rate": 9.772008047397647e-06,
      "loss": 0.2372,
      "step": 1545
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4464378185654634,
      "learning_rate": 9.771621146599369e-06,
      "loss": 0.1584,
      "step": 1546
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.393102295164832,
      "learning_rate": 9.771233925467703e-06,
      "loss": 0.6021,
      "step": 1547
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3279935690715412,
      "learning_rate": 9.770846384028647e-06,
      "loss": 0.2526,
      "step": 1548
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3988597764851942,
      "learning_rate": 9.77045852230822e-06,
      "loss": 0.2023,
      "step": 1549
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4144040639472815,
      "learning_rate": 9.770070340332457e-06,
      "loss": 0.2144,
      "step": 1550
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.861902363026179,
      "learning_rate": 9.769681838127421e-06,
      "loss": 0.2369,
      "step": 1551
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.389033659250747,
      "learning_rate": 9.76929301571919e-06,
      "loss": 0.198,
      "step": 1552
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.596142413144707,
      "learning_rate": 9.76890387313387e-06,
      "loss": 0.5471,
      "step": 1553
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.669517148887409,
      "learning_rate": 9.768514410397583e-06,
      "loss": 0.644,
      "step": 1554
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.410359637946326,
      "learning_rate": 9.768124627536474e-06,
      "loss": 0.1952,
      "step": 1555
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.727014588356394,
      "learning_rate": 9.767734524576714e-06,
      "loss": 0.8061,
      "step": 1556
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4138314962122738,
      "learning_rate": 9.767344101544489e-06,
      "loss": 0.2108,
      "step": 1557
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4932571770122638,
      "learning_rate": 9.76695335846601e-06,
      "loss": 0.2282,
      "step": 1558
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6449023745881455,
      "learning_rate": 9.76656229536751e-06,
      "loss": 0.2151,
      "step": 1559
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4537766467283442,
      "learning_rate": 9.76617091227524e-06,
      "loss": 0.2191,
      "step": 1560
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.571360027749933,
      "learning_rate": 9.765779209215474e-06,
      "loss": 0.1808,
      "step": 1561
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.103660751294775,
      "learning_rate": 9.765387186214512e-06,
      "loss": 0.6537,
      "step": 1562
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4616768983692543,
      "learning_rate": 9.76499484329867e-06,
      "loss": 0.2342,
      "step": 1563
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3466355581617389,
      "learning_rate": 9.764602180494285e-06,
      "loss": 0.2078,
      "step": 1564
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5461759338537195,
      "learning_rate": 9.764209197827721e-06,
      "loss": 0.2502,
      "step": 1565
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.714268542928138,
      "learning_rate": 9.763815895325357e-06,
      "loss": 0.2027,
      "step": 1566
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3260968907471,
      "learning_rate": 9.7634222730136e-06,
      "loss": 0.1785,
      "step": 1567
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.9517203985932,
      "learning_rate": 9.763028330918874e-06,
      "loss": 0.5397,
      "step": 1568
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5405970638816762,
      "learning_rate": 9.762634069067622e-06,
      "loss": 0.2385,
      "step": 1569
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6015938995679906,
      "learning_rate": 9.762239487486316e-06,
      "loss": 0.2111,
      "step": 1570
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.174633082700354,
      "learning_rate": 9.761844586201444e-06,
      "loss": 0.5955,
      "step": 1571
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5709485893560706,
      "learning_rate": 9.761449365239518e-06,
      "loss": 0.2141,
      "step": 1572
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.937243962845362,
      "learning_rate": 9.761053824627068e-06,
      "loss": 0.5774,
      "step": 1573
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.301274228241699,
      "learning_rate": 9.76065796439065e-06,
      "loss": 0.7388,
      "step": 1574
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.382179165977071,
      "learning_rate": 9.76026178455684e-06,
      "loss": 0.1987,
      "step": 1575
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.389248017784347,
      "learning_rate": 9.759865285152231e-06,
      "loss": 0.2112,
      "step": 1576
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4429928705861934,
      "learning_rate": 9.759468466203444e-06,
      "loss": 0.2001,
      "step": 1577
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5304857577376287,
      "learning_rate": 9.75907132773712e-06,
      "loss": 0.1979,
      "step": 1578
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4636827269179473,
      "learning_rate": 9.758673869779915e-06,
      "loss": 0.2474,
      "step": 1579
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.679234629073408,
      "learning_rate": 9.758276092358518e-06,
      "loss": 0.6342,
      "step": 1580
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.86152090252494,
      "learning_rate": 9.75787799549963e-06,
      "loss": 0.5772,
      "step": 1581
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3741318618094576,
      "learning_rate": 9.757479579229974e-06,
      "loss": 0.2698,
      "step": 1582
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.8542829757183465,
      "learning_rate": 9.757080843576301e-06,
      "loss": 0.4931,
      "step": 1583
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3755555674464939,
      "learning_rate": 9.756681788565379e-06,
      "loss": 0.2309,
      "step": 1584
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4222154739746977,
      "learning_rate": 9.756282414223995e-06,
      "loss": 0.2278,
      "step": 1585
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6003354242457708,
      "learning_rate": 9.75588272057896e-06,
      "loss": 0.2788,
      "step": 1586
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6106017498133456,
      "learning_rate": 9.755482707657109e-06,
      "loss": 0.2263,
      "step": 1587
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.773172038508745,
      "learning_rate": 9.755082375485296e-06,
      "loss": 0.234,
      "step": 1588
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4545373227217495,
      "learning_rate": 9.754681724090396e-06,
      "loss": 0.1929,
      "step": 1589
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.609205148072724,
      "learning_rate": 9.754280753499306e-06,
      "loss": 0.2449,
      "step": 1590
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5666283507519787,
      "learning_rate": 9.753879463738942e-06,
      "loss": 0.229,
      "step": 1591
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3230971638142615,
      "learning_rate": 9.753477854836248e-06,
      "loss": 0.2137,
      "step": 1592
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5369705493136865,
      "learning_rate": 9.753075926818182e-06,
      "loss": 0.1978,
      "step": 1593
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4575961176763852,
      "learning_rate": 9.752673679711728e-06,
      "loss": 0.2461,
      "step": 1594
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6064120304746499,
      "learning_rate": 9.75227111354389e-06,
      "loss": 0.2334,
      "step": 1595
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5829676572043545,
      "learning_rate": 9.751868228341695e-06,
      "loss": 0.2469,
      "step": 1596
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.543565400196061,
      "learning_rate": 9.751465024132184e-06,
      "loss": 0.2054,
      "step": 1597
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4902531190359736,
      "learning_rate": 9.751061500942434e-06,
      "loss": 0.2588,
      "step": 1598
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4464916939465176,
      "learning_rate": 9.750657658799528e-06,
      "loss": 0.2597,
      "step": 1599
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7689443638192823,
      "learning_rate": 9.75025349773058e-06,
      "loss": 0.216,
      "step": 1600
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4004095866677506,
      "learning_rate": 9.749849017762723e-06,
      "loss": 0.1793,
      "step": 1601
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.546138023549135,
      "learning_rate": 9.749444218923108e-06,
      "loss": 0.233,
      "step": 1602
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.456426824446879,
      "learning_rate": 9.749039101238914e-06,
      "loss": 0.1756,
      "step": 1603
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4063896715796795,
      "learning_rate": 9.748633664737334e-06,
      "loss": 0.2183,
      "step": 1604
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.705206337718719,
      "learning_rate": 9.74822790944559e-06,
      "loss": 0.2464,
      "step": 1605
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4840002082776158,
      "learning_rate": 9.74782183539092e-06,
      "loss": 0.2327,
      "step": 1606
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.511477359022925,
      "learning_rate": 9.747415442600585e-06,
      "loss": 0.2129,
      "step": 1607
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4895690426055241,
      "learning_rate": 9.747008731101865e-06,
      "loss": 0.2399,
      "step": 1608
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3378153449320915,
      "learning_rate": 9.74660170092207e-06,
      "loss": 0.2232,
      "step": 1609
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.381222187473741,
      "learning_rate": 9.746194352088518e-06,
      "loss": 0.2227,
      "step": 1610
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.660979463835364,
      "learning_rate": 9.74578668462856e-06,
      "loss": 0.2261,
      "step": 1611
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7241854923008417,
      "learning_rate": 9.745378698569562e-06,
      "loss": 0.2813,
      "step": 1612
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4115039208506461,
      "learning_rate": 9.744970393938915e-06,
      "loss": 0.2285,
      "step": 1613
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4776320597948998,
      "learning_rate": 9.744561770764027e-06,
      "loss": 0.237,
      "step": 1614
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.706970959684569,
      "learning_rate": 9.744152829072333e-06,
      "loss": 0.2548,
      "step": 1615
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5499903134757889,
      "learning_rate": 9.743743568891287e-06,
      "loss": 0.2173,
      "step": 1616
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4368205494799517,
      "learning_rate": 9.743333990248359e-06,
      "loss": 0.2198,
      "step": 1617
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.436406501976359,
      "learning_rate": 9.742924093171051e-06,
      "loss": 0.2456,
      "step": 1618
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3677361623959812,
      "learning_rate": 9.742513877686877e-06,
      "loss": 0.202,
      "step": 1619
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7417420260182053,
      "learning_rate": 9.742103343823376e-06,
      "loss": 0.2174,
      "step": 1620
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5239655330579545,
      "learning_rate": 9.741692491608112e-06,
      "loss": 0.2389,
      "step": 1621
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.521519013274887,
      "learning_rate": 9.741281321068663e-06,
      "loss": 0.2058,
      "step": 1622
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3958180231477604,
      "learning_rate": 9.740869832232634e-06,
      "loss": 0.2353,
      "step": 1623
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4866407509368103,
      "learning_rate": 9.740458025127649e-06,
      "loss": 0.2141,
      "step": 1624
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6068765752354282,
      "learning_rate": 9.740045899781353e-06,
      "loss": 0.2858,
      "step": 1625
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.601379070633747,
      "learning_rate": 9.739633456221415e-06,
      "loss": 0.2811,
      "step": 1626
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.39640397491769,
      "learning_rate": 9.739220694475522e-06,
      "loss": 0.506,
      "step": 1627
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5505128105701302,
      "learning_rate": 9.738807614571384e-06,
      "loss": 0.2244,
      "step": 1628
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3832767673421447,
      "learning_rate": 9.738394216536733e-06,
      "loss": 0.2245,
      "step": 1629
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5330333737615514,
      "learning_rate": 9.737980500399322e-06,
      "loss": 0.2428,
      "step": 1630
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5955024672564626,
      "learning_rate": 9.737566466186922e-06,
      "loss": 0.2338,
      "step": 1631
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4270569297287,
      "learning_rate": 9.737152113927335e-06,
      "loss": 0.2547,
      "step": 1632
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3095426311807545,
      "learning_rate": 9.736737443648372e-06,
      "loss": 0.1955,
      "step": 1633
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.528110671765845,
      "learning_rate": 9.73632245537787e-06,
      "loss": 0.5162,
      "step": 1634
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4692389088093838,
      "learning_rate": 9.735907149143695e-06,
      "loss": 0.2305,
      "step": 1635
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.457093574168827,
      "learning_rate": 9.735491524973723e-06,
      "loss": 0.689,
      "step": 1636
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.628693405302554,
      "learning_rate": 9.735075582895856e-06,
      "loss": 0.2378,
      "step": 1637
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4240199248006942,
      "learning_rate": 9.734659322938018e-06,
      "loss": 0.1779,
      "step": 1638
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6793055074558272,
      "learning_rate": 9.734242745128156e-06,
      "loss": 0.2467,
      "step": 1639
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.432814034423324,
      "learning_rate": 9.733825849494232e-06,
      "loss": 0.2152,
      "step": 1640
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4151679170555584,
      "learning_rate": 9.733408636064236e-06,
      "loss": 0.2328,
      "step": 1641
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4883539018594545,
      "learning_rate": 9.732991104866179e-06,
      "loss": 0.1827,
      "step": 1642
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.8699234101320785,
      "learning_rate": 9.732573255928086e-06,
      "loss": 0.7119,
      "step": 1643
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.177567743065299,
      "learning_rate": 9.732155089278013e-06,
      "loss": 0.411,
      "step": 1644
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7471970828172905,
      "learning_rate": 9.731736604944031e-06,
      "loss": 0.245,
      "step": 1645
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7554415383245885,
      "learning_rate": 9.731317802954233e-06,
      "loss": 0.2649,
      "step": 1646
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3657823793349615,
      "learning_rate": 9.730898683336735e-06,
      "loss": 0.19,
      "step": 1647
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8227522296779513,
      "learning_rate": 9.730479246119677e-06,
      "loss": 0.2257,
      "step": 1648
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.938908800728655,
      "learning_rate": 9.730059491331214e-06,
      "loss": 0.658,
      "step": 1649
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5611551436068443,
      "learning_rate": 9.729639418999524e-06,
      "loss": 0.2595,
      "step": 1650
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.547685064912496,
      "learning_rate": 9.72921902915281e-06,
      "loss": 0.2508,
      "step": 1651
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5374636961496102,
      "learning_rate": 9.728798321819294e-06,
      "loss": 0.215,
      "step": 1652
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4988286417003958,
      "learning_rate": 9.728377297027218e-06,
      "loss": 0.2437,
      "step": 1653
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4196981853732336,
      "learning_rate": 9.727955954804848e-06,
      "loss": 0.2515,
      "step": 1654
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5171443314270052,
      "learning_rate": 9.727534295180471e-06,
      "loss": 0.2713,
      "step": 1655
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.41126335629371,
      "learning_rate": 9.727112318182392e-06,
      "loss": 0.1994,
      "step": 1656
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4592917440724393,
      "learning_rate": 9.72669002383894e-06,
      "loss": 0.1927,
      "step": 1657
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4958176694534637,
      "learning_rate": 9.726267412178467e-06,
      "loss": 0.2435,
      "step": 1658
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3312267431402287,
      "learning_rate": 9.725844483229342e-06,
      "loss": 0.1796,
      "step": 1659
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.1085467511902705,
      "learning_rate": 9.725421237019957e-06,
      "loss": 0.5534,
      "step": 1660
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5264526841901913,
      "learning_rate": 9.724997673578727e-06,
      "loss": 0.2521,
      "step": 1661
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6653291942087425,
      "learning_rate": 9.724573792934089e-06,
      "loss": 0.2472,
      "step": 1662
    },
    {
      "epoch": 0.13,
      "grad_norm": 25.64071668094937,
      "learning_rate": 9.724149595114496e-06,
      "loss": 0.5309,
      "step": 1663
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.308603189700793,
      "learning_rate": 9.723725080148426e-06,
      "loss": 0.5843,
      "step": 1664
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5883693224922457,
      "learning_rate": 9.72330024806438e-06,
      "loss": 0.271,
      "step": 1665
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4579251605038241,
      "learning_rate": 9.722875098890878e-06,
      "loss": 0.1772,
      "step": 1666
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.8562630052979365,
      "learning_rate": 9.72244963265646e-06,
      "loss": 0.5757,
      "step": 1667
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6744308512167032,
      "learning_rate": 9.72202384938969e-06,
      "loss": 0.2761,
      "step": 1668
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.51270892421273,
      "learning_rate": 9.721597749119151e-06,
      "loss": 0.8495,
      "step": 1669
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4967390457973082,
      "learning_rate": 9.721171331873452e-06,
      "loss": 0.2468,
      "step": 1670
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2368237259695152,
      "learning_rate": 9.720744597681213e-06,
      "loss": 0.1914,
      "step": 1671
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3838599243408192,
      "learning_rate": 9.720317546571088e-06,
      "loss": 0.2095,
      "step": 1672
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6013841684581451,
      "learning_rate": 9.719890178571744e-06,
      "loss": 0.2215,
      "step": 1673
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5315634146774177,
      "learning_rate": 9.719462493711873e-06,
      "loss": 0.1962,
      "step": 1674
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5392839013199366,
      "learning_rate": 9.719034492020183e-06,
      "loss": 0.2497,
      "step": 1675
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7014341258207781,
      "learning_rate": 9.718606173525411e-06,
      "loss": 0.2738,
      "step": 1676
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3250246879392782,
      "learning_rate": 9.718177538256309e-06,
      "loss": 0.2267,
      "step": 1677
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.536491524014943,
      "learning_rate": 9.717748586241653e-06,
      "loss": 0.2146,
      "step": 1678
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.293486948058062,
      "learning_rate": 9.71731931751024e-06,
      "loss": 0.6151,
      "step": 1679
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.556929727772488,
      "learning_rate": 9.71688973209089e-06,
      "loss": 0.2693,
      "step": 1680
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3622827255245316,
      "learning_rate": 9.716459830012439e-06,
      "loss": 0.1706,
      "step": 1681
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.318877973805238,
      "learning_rate": 9.71602961130375e-06,
      "loss": 0.515,
      "step": 1682
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2336719882953078,
      "learning_rate": 9.715599075993705e-06,
      "loss": 0.2294,
      "step": 1683
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.047145570888419,
      "learning_rate": 9.715168224111205e-06,
      "loss": 0.644,
      "step": 1684
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.474036620858353,
      "learning_rate": 9.714737055685176e-06,
      "loss": 0.2409,
      "step": 1685
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5778056575076935,
      "learning_rate": 9.714305570744564e-06,
      "loss": 0.2329,
      "step": 1686
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2285047991412406,
      "learning_rate": 9.713873769318333e-06,
      "loss": 0.1891,
      "step": 1687
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5607622134746675,
      "learning_rate": 9.713441651435477e-06,
      "loss": 0.2421,
      "step": 1688
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4513717505441104,
      "learning_rate": 9.713009217125e-06,
      "loss": 0.2218,
      "step": 1689
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5635359649151226,
      "learning_rate": 9.712576466415935e-06,
      "loss": 0.2262,
      "step": 1690
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4433240441328847,
      "learning_rate": 9.712143399337333e-06,
      "loss": 0.2212,
      "step": 1691
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3881155947417456,
      "learning_rate": 9.711710015918266e-06,
      "loss": 0.2209,
      "step": 1692
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4800033949726952,
      "learning_rate": 9.71127631618783e-06,
      "loss": 0.1839,
      "step": 1693
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5028345570390487,
      "learning_rate": 9.710842300175141e-06,
      "loss": 0.2038,
      "step": 1694
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5009735742830224,
      "learning_rate": 9.710407967909336e-06,
      "loss": 0.2705,
      "step": 1695
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3972579493609312,
      "learning_rate": 9.709973319419572e-06,
      "loss": 0.1921,
      "step": 1696
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.523936841142965,
      "learning_rate": 9.709538354735026e-06,
      "loss": 0.54,
      "step": 1697
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6991294389441396,
      "learning_rate": 9.709103073884905e-06,
      "loss": 0.2528,
      "step": 1698
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5639811472812528,
      "learning_rate": 9.708667476898423e-06,
      "loss": 0.2532,
      "step": 1699
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3540175628340287,
      "learning_rate": 9.708231563804828e-06,
      "loss": 0.1779,
      "step": 1700
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6095982351583809,
      "learning_rate": 9.707795334633383e-06,
      "loss": 0.1874,
      "step": 1701
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4729378250201564,
      "learning_rate": 9.707358789413373e-06,
      "loss": 0.216,
      "step": 1702
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5807371727896857,
      "learning_rate": 9.706921928174105e-06,
      "loss": 0.2195,
      "step": 1703
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4800662770005686,
      "learning_rate": 9.706484750944905e-06,
      "loss": 0.2513,
      "step": 1704
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6159367502238662,
      "learning_rate": 9.706047257755124e-06,
      "loss": 0.3019,
      "step": 1705
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4629890329393809,
      "learning_rate": 9.705609448634133e-06,
      "loss": 0.2201,
      "step": 1706
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.225631699616758,
      "learning_rate": 9.705171323611322e-06,
      "loss": 0.5263,
      "step": 1707
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7799884269786705,
      "learning_rate": 9.704732882716104e-06,
      "loss": 0.2766,
      "step": 1708
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4597224490605745,
      "learning_rate": 9.704294125977912e-06,
      "loss": 0.2501,
      "step": 1709
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.466299676143188,
      "learning_rate": 9.703855053426202e-06,
      "loss": 0.2116,
      "step": 1710
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.529599025438097,
      "learning_rate": 9.703415665090452e-06,
      "loss": 0.273,
      "step": 1711
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4503870188316348,
      "learning_rate": 9.702975961000155e-06,
      "loss": 0.1992,
      "step": 1712
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4734372417685295,
      "learning_rate": 9.702535941184833e-06,
      "loss": 0.2385,
      "step": 1713
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2495034589875202,
      "learning_rate": 9.702095605674027e-06,
      "loss": 0.2228,
      "step": 1714
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2685117477059433,
      "learning_rate": 9.701654954497294e-06,
      "loss": 0.2231,
      "step": 1715
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.400857054435006,
      "learning_rate": 9.70121398768422e-06,
      "loss": 0.1931,
      "step": 1716
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4933470651291618,
      "learning_rate": 9.700772705264405e-06,
      "loss": 0.2293,
      "step": 1717
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3789365329228507,
      "learning_rate": 9.700331107267477e-06,
      "loss": 0.2196,
      "step": 1718
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.161258222313218,
      "learning_rate": 9.69988919372308e-06,
      "loss": 0.7293,
      "step": 1719
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5914113306653817,
      "learning_rate": 9.699446964660882e-06,
      "loss": 0.2295,
      "step": 1720
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.815576091636923,
      "learning_rate": 9.69900442011057e-06,
      "loss": 0.5205,
      "step": 1721
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5653434655502219,
      "learning_rate": 9.698561560101853e-06,
      "loss": 0.2874,
      "step": 1722
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5361753284881046,
      "learning_rate": 9.698118384664464e-06,
      "loss": 0.2424,
      "step": 1723
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6057505288675846,
      "learning_rate": 9.69767489382815e-06,
      "loss": 0.2169,
      "step": 1724
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6105553864379374,
      "learning_rate": 9.697231087622691e-06,
      "loss": 0.2481,
      "step": 1725
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4405817181475413,
      "learning_rate": 9.696786966077875e-06,
      "loss": 0.2457,
      "step": 1726
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4830495415977467,
      "learning_rate": 9.69634252922352e-06,
      "loss": 0.2373,
      "step": 1727
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4006478621622342,
      "learning_rate": 9.69589777708946e-06,
      "loss": 0.2405,
      "step": 1728
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4885861658830843,
      "learning_rate": 9.695452709705555e-06,
      "loss": 0.2203,
      "step": 1729
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3404527013588188,
      "learning_rate": 9.695007327101685e-06,
      "loss": 0.2169,
      "step": 1730
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4890546172742003,
      "learning_rate": 9.694561629307745e-06,
      "loss": 0.2582,
      "step": 1731
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6100396888615591,
      "learning_rate": 9.694115616353662e-06,
      "loss": 0.2302,
      "step": 1732
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.946664602535032,
      "learning_rate": 9.693669288269371e-06,
      "loss": 0.6134,
      "step": 1733
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3256763124198974,
      "learning_rate": 9.69322264508484e-06,
      "loss": 0.2222,
      "step": 1734
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.61350332273475,
      "learning_rate": 9.692775686830057e-06,
      "loss": 0.2356,
      "step": 1735
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5383885625029967,
      "learning_rate": 9.69232841353502e-06,
      "loss": 0.2372,
      "step": 1736
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5613793907693219,
      "learning_rate": 9.69188082522976e-06,
      "loss": 0.2467,
      "step": 1737
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.341786850136851,
      "learning_rate": 9.691432921944325e-06,
      "loss": 0.2541,
      "step": 1738
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5164759420816685,
      "learning_rate": 9.690984703708783e-06,
      "loss": 0.2234,
      "step": 1739
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3333153286507795,
      "learning_rate": 9.690536170553226e-06,
      "loss": 0.1712,
      "step": 1740
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2479864033253565,
      "learning_rate": 9.690087322507763e-06,
      "loss": 0.2337,
      "step": 1741
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4575191194921926,
      "learning_rate": 9.689638159602527e-06,
      "loss": 0.2519,
      "step": 1742
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.608497112628641,
      "learning_rate": 9.689188681867675e-06,
      "loss": 0.2299,
      "step": 1743
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3864261547311492,
      "learning_rate": 9.688738889333376e-06,
      "loss": 0.2051,
      "step": 1744
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1792271922396842,
      "learning_rate": 9.68828878202983e-06,
      "loss": 0.1767,
      "step": 1745
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4219191134336626,
      "learning_rate": 9.687838359987254e-06,
      "loss": 0.2431,
      "step": 1746
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4753799795627052,
      "learning_rate": 9.687387623235885e-06,
      "loss": 0.2374,
      "step": 1747
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3470168408588994,
      "learning_rate": 9.686936571805982e-06,
      "loss": 0.1749,
      "step": 1748
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4509181163443456,
      "learning_rate": 9.686485205727827e-06,
      "loss": 0.1974,
      "step": 1749
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4787658377822652,
      "learning_rate": 9.68603352503172e-06,
      "loss": 0.2642,
      "step": 1750
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3933065630299974,
      "learning_rate": 9.685581529747982e-06,
      "loss": 0.2168,
      "step": 1751
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5075548508439822,
      "learning_rate": 9.685129219906964e-06,
      "loss": 0.2385,
      "step": 1752
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4706416043161044,
      "learning_rate": 9.684676595539023e-06,
      "loss": 0.2406,
      "step": 1753
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6457101424970801,
      "learning_rate": 9.684223656674548e-06,
      "loss": 0.2719,
      "step": 1754
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6129606462913444,
      "learning_rate": 9.683770403343947e-06,
      "loss": 0.1874,
      "step": 1755
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.874525838877892,
      "learning_rate": 9.683316835577648e-06,
      "loss": 0.6454,
      "step": 1756
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.533704318351654,
      "learning_rate": 9.6828629534061e-06,
      "loss": 0.2124,
      "step": 1757
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4424987723684324,
      "learning_rate": 9.682408756859772e-06,
      "loss": 0.2383,
      "step": 1758
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2855393696998263,
      "learning_rate": 9.681954245969158e-06,
      "loss": 0.2333,
      "step": 1759
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.74254894544208,
      "learning_rate": 9.681499420764771e-06,
      "loss": 0.7153,
      "step": 1760
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5713553570242775,
      "learning_rate": 9.681044281277141e-06,
      "loss": 0.229,
      "step": 1761
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.587260698866443,
      "learning_rate": 9.680588827536828e-06,
      "loss": 0.2201,
      "step": 1762
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5569514938301812,
      "learning_rate": 9.680133059574403e-06,
      "loss": 0.2452,
      "step": 1763
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.454284750554401,
      "learning_rate": 9.679676977420467e-06,
      "loss": 0.2105,
      "step": 1764
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.801300159525248,
      "learning_rate": 9.679220581105636e-06,
      "loss": 0.5018,
      "step": 1765
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.364627286606356,
      "learning_rate": 9.67876387066055e-06,
      "loss": 0.2129,
      "step": 1766
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.018976042950309,
      "learning_rate": 9.67830684611587e-06,
      "loss": 0.6771,
      "step": 1767
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.447175655512617,
      "learning_rate": 9.677849507502275e-06,
      "loss": 0.2346,
      "step": 1768
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3158108160667845,
      "learning_rate": 9.67739185485047e-06,
      "loss": 0.2065,
      "step": 1769
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.476034659343413,
      "learning_rate": 9.676933888191178e-06,
      "loss": 0.2854,
      "step": 1770
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3606615645319498,
      "learning_rate": 9.676475607555145e-06,
      "loss": 0.2547,
      "step": 1771
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3675545766153718,
      "learning_rate": 9.676017012973133e-06,
      "loss": 0.2279,
      "step": 1772
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.727456697665223,
      "learning_rate": 9.675558104475933e-06,
      "loss": 0.6477,
      "step": 1773
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.515171015126078,
      "learning_rate": 9.67509888209435e-06,
      "loss": 0.3094,
      "step": 1774
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5394367436950924,
      "learning_rate": 9.674639345859213e-06,
      "loss": 0.2638,
      "step": 1775
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5962039124052536,
      "learning_rate": 9.674179495801375e-06,
      "loss": 0.2064,
      "step": 1776
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.240197954443099,
      "learning_rate": 9.673719331951706e-06,
      "loss": 0.4859,
      "step": 1777
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.656502508423278,
      "learning_rate": 9.673258854341094e-06,
      "loss": 0.2062,
      "step": 1778
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.552797027294706,
      "learning_rate": 9.672798063000458e-06,
      "loss": 0.2396,
      "step": 1779
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3931331279779118,
      "learning_rate": 9.67233695796073e-06,
      "loss": 0.2041,
      "step": 1780
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4568924059894364,
      "learning_rate": 9.671875539252865e-06,
      "loss": 0.2392,
      "step": 1781
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4802489232587634,
      "learning_rate": 9.67141380690784e-06,
      "loss": 0.1852,
      "step": 1782
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.639534657927216,
      "learning_rate": 9.670951760956653e-06,
      "loss": 0.6627,
      "step": 1783
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3268191120135426,
      "learning_rate": 9.670489401430322e-06,
      "loss": 0.1699,
      "step": 1784
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.282712151151691,
      "learning_rate": 9.670026728359884e-06,
      "loss": 0.6576,
      "step": 1785
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4822379096998657,
      "learning_rate": 9.669563741776405e-06,
      "loss": 0.252,
      "step": 1786
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4795322220388172,
      "learning_rate": 9.669100441710962e-06,
      "loss": 0.2574,
      "step": 1787
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6200600945979102,
      "learning_rate": 9.66863682819466e-06,
      "loss": 0.2862,
      "step": 1788
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4324867009525744,
      "learning_rate": 9.668172901258623e-06,
      "loss": 0.2259,
      "step": 1789
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5968672444934398,
      "learning_rate": 9.667708660933994e-06,
      "loss": 0.2563,
      "step": 1790
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3424385409778248,
      "learning_rate": 9.66724410725194e-06,
      "loss": 0.2113,
      "step": 1791
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5133813264514135,
      "learning_rate": 9.66677924024365e-06,
      "loss": 0.2227,
      "step": 1792
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.935970297876555,
      "learning_rate": 9.666314059940326e-06,
      "loss": 0.6504,
      "step": 1793
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4143739506100628,
      "learning_rate": 9.665848566373204e-06,
      "loss": 0.2215,
      "step": 1794
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3364310700549318,
      "learning_rate": 9.665382759573529e-06,
      "loss": 0.2022,
      "step": 1795
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3541510201119917,
      "learning_rate": 9.664916639572574e-06,
      "loss": 0.2016,
      "step": 1796
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4388248897716867,
      "learning_rate": 9.664450206401633e-06,
      "loss": 0.2302,
      "step": 1797
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3651287386259812,
      "learning_rate": 9.663983460092015e-06,
      "loss": 0.2042,
      "step": 1798
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2350813275361885,
      "learning_rate": 9.663516400675057e-06,
      "loss": 0.1715,
      "step": 1799
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4219278792127619,
      "learning_rate": 9.663049028182112e-06,
      "loss": 0.1833,
      "step": 1800
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5984418880959148,
      "learning_rate": 9.662581342644557e-06,
      "loss": 0.2481,
      "step": 1801
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5474821964505603,
      "learning_rate": 9.662113344093791e-06,
      "loss": 0.2505,
      "step": 1802
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.440962164400899,
      "learning_rate": 9.66164503256123e-06,
      "loss": 0.2435,
      "step": 1803
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.576455989915794,
      "learning_rate": 9.661176408078315e-06,
      "loss": 0.2246,
      "step": 1804
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.414412901157168,
      "learning_rate": 9.660707470676503e-06,
      "loss": 0.5895,
      "step": 1805
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5228796606428752,
      "learning_rate": 9.660238220387277e-06,
      "loss": 0.2213,
      "step": 1806
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4916439094235168,
      "learning_rate": 9.659768657242138e-06,
      "loss": 0.2333,
      "step": 1807
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.273099535753963,
      "learning_rate": 9.659298781272615e-06,
      "loss": 0.6208,
      "step": 1808
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.845409289409724,
      "learning_rate": 9.658828592510243e-06,
      "loss": 0.2603,
      "step": 1809
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4958124122288672,
      "learning_rate": 9.658358090986594e-06,
      "loss": 0.2757,
      "step": 1810
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4821414328281577,
      "learning_rate": 9.657887276733254e-06,
      "loss": 0.2513,
      "step": 1811
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4463573145481676,
      "learning_rate": 9.657416149781826e-06,
      "loss": 0.2355,
      "step": 1812
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3962703248241843,
      "learning_rate": 9.65694471016394e-06,
      "loss": 0.185,
      "step": 1813
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5092840818926048,
      "learning_rate": 9.656472957911247e-06,
      "loss": 0.2272,
      "step": 1814
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5675223724328375,
      "learning_rate": 9.656000893055416e-06,
      "loss": 0.2116,
      "step": 1815
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.264692402207976,
      "learning_rate": 9.655528515628136e-06,
      "loss": 0.1975,
      "step": 1816
    },
    {
      "epoch": 0.15,
      "grad_norm": 25.45217738128297,
      "learning_rate": 9.655055825661122e-06,
      "loss": 0.7673,
      "step": 1817
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4165821059514272,
      "learning_rate": 9.654582823186107e-06,
      "loss": 0.238,
      "step": 1818
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5698541977874627,
      "learning_rate": 9.654109508234843e-06,
      "loss": 0.2435,
      "step": 1819
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6561874069663314,
      "learning_rate": 9.653635880839107e-06,
      "loss": 0.2296,
      "step": 1820
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7544145649718108,
      "learning_rate": 9.653161941030695e-06,
      "loss": 0.3023,
      "step": 1821
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6215254166246602,
      "learning_rate": 9.652687688841422e-06,
      "loss": 0.2399,
      "step": 1822
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.529883773675546,
      "learning_rate": 9.652213124303126e-06,
      "loss": 0.2131,
      "step": 1823
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4077640443300024,
      "learning_rate": 9.65173824744767e-06,
      "loss": 0.2164,
      "step": 1824
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.950152782950908,
      "learning_rate": 9.651263058306932e-06,
      "loss": 0.5494,
      "step": 1825
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4457866648370847,
      "learning_rate": 9.650787556912811e-06,
      "loss": 0.2144,
      "step": 1826
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4620074494844246,
      "learning_rate": 9.650311743297229e-06,
      "loss": 0.2424,
      "step": 1827
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.382121137459981,
      "learning_rate": 9.64983561749213e-06,
      "loss": 0.6872,
      "step": 1828
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.883283566820824,
      "learning_rate": 9.649359179529477e-06,
      "loss": 0.6622,
      "step": 1829
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3915846225953552,
      "learning_rate": 9.648882429441258e-06,
      "loss": 0.2305,
      "step": 1830
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3348083555934644,
      "learning_rate": 9.648405367259475e-06,
      "loss": 0.2074,
      "step": 1831
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3593364059817745,
      "learning_rate": 9.647927993016154e-06,
      "loss": 0.1948,
      "step": 1832
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.283341574165253,
      "learning_rate": 9.647450306743345e-06,
      "loss": 0.1597,
      "step": 1833
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.424624696304144,
      "learning_rate": 9.646972308473115e-06,
      "loss": 0.2212,
      "step": 1834
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4161912210982865,
      "learning_rate": 9.646493998237557e-06,
      "loss": 0.2152,
      "step": 1835
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.528444093794733,
      "learning_rate": 9.646015376068776e-06,
      "loss": 0.2639,
      "step": 1836
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4918759334076668,
      "learning_rate": 9.645536441998907e-06,
      "loss": 0.294,
      "step": 1837
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6963649947054689,
      "learning_rate": 9.6450571960601e-06,
      "loss": 0.2598,
      "step": 1838
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.358260605009517,
      "learning_rate": 9.64457763828453e-06,
      "loss": 0.2321,
      "step": 1839
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4477344401864383,
      "learning_rate": 9.64409776870439e-06,
      "loss": 0.2039,
      "step": 1840
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.412542886639437,
      "learning_rate": 9.643617587351897e-06,
      "loss": 0.2201,
      "step": 1841
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4171031095198496,
      "learning_rate": 9.643137094259285e-06,
      "loss": 0.2512,
      "step": 1842
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.474123267364693,
      "learning_rate": 9.642656289458812e-06,
      "loss": 0.2183,
      "step": 1843
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.002951251146631,
      "learning_rate": 9.642175172982755e-06,
      "loss": 0.4185,
      "step": 1844
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4805643562694681,
      "learning_rate": 9.641693744863413e-06,
      "loss": 0.209,
      "step": 1845
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6013647707949583,
      "learning_rate": 9.641212005133107e-06,
      "loss": 0.2495,
      "step": 1846
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.204465097782585,
      "learning_rate": 9.640729953824178e-06,
      "loss": 0.6387,
      "step": 1847
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.469859351823903,
      "learning_rate": 9.640247590968985e-06,
      "loss": 0.602,
      "step": 1848
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4040876612257354,
      "learning_rate": 9.639764916599913e-06,
      "loss": 0.2249,
      "step": 1849
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4923479942416529,
      "learning_rate": 9.639281930749363e-06,
      "loss": 0.2602,
      "step": 1850
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3751367266363022,
      "learning_rate": 9.63879863344976e-06,
      "loss": 0.2317,
      "step": 1851
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4436553306092534,
      "learning_rate": 9.638315024733552e-06,
      "loss": 0.2361,
      "step": 1852
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5383488585043816,
      "learning_rate": 9.6378311046332e-06,
      "loss": 0.2462,
      "step": 1853
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4334003279016496,
      "learning_rate": 9.637346873181194e-06,
      "loss": 0.2143,
      "step": 1854
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1673812802913086,
      "learning_rate": 9.636862330410043e-06,
      "loss": 0.1625,
      "step": 1855
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.58793702787862,
      "learning_rate": 9.636377476352277e-06,
      "loss": 0.2532,
      "step": 1856
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3925343785715028,
      "learning_rate": 9.63589231104044e-06,
      "loss": 0.2634,
      "step": 1857
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.42520859198846,
      "learning_rate": 9.635406834507108e-06,
      "loss": 0.2289,
      "step": 1858
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5713187342892687,
      "learning_rate": 9.63492104678487e-06,
      "loss": 0.1898,
      "step": 1859
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5081517942789442,
      "learning_rate": 9.634434947906337e-06,
      "loss": 0.2367,
      "step": 1860
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5185103167867047,
      "learning_rate": 9.633948537904145e-06,
      "loss": 0.2268,
      "step": 1861
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8004943209709992,
      "learning_rate": 9.633461816810949e-06,
      "loss": 0.2933,
      "step": 1862
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5849704524915296,
      "learning_rate": 9.632974784659421e-06,
      "loss": 0.2445,
      "step": 1863
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4691035787116464,
      "learning_rate": 9.632487441482258e-06,
      "loss": 0.3007,
      "step": 1864
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4209399595229988,
      "learning_rate": 9.631999787312179e-06,
      "loss": 0.2427,
      "step": 1865
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4991589331775699,
      "learning_rate": 9.631511822181918e-06,
      "loss": 0.2006,
      "step": 1866
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5548685053767215,
      "learning_rate": 9.631023546124236e-06,
      "loss": 0.2207,
      "step": 1867
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5006229270384637,
      "learning_rate": 9.630534959171912e-06,
      "loss": 0.2339,
      "step": 1868
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4997565938429933,
      "learning_rate": 9.630046061357745e-06,
      "loss": 0.2284,
      "step": 1869
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4548194027415642,
      "learning_rate": 9.62955685271456e-06,
      "loss": 0.2252,
      "step": 1870
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3836536278467995,
      "learning_rate": 9.629067333275195e-06,
      "loss": 0.1946,
      "step": 1871
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3601950686464495,
      "learning_rate": 9.628577503072513e-06,
      "loss": 0.2329,
      "step": 1872
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3874831891846593,
      "learning_rate": 9.628087362139402e-06,
      "loss": 0.2025,
      "step": 1873
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4435317874573304,
      "learning_rate": 9.627596910508763e-06,
      "loss": 0.1901,
      "step": 1874
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6043322653590923,
      "learning_rate": 9.627106148213521e-06,
      "loss": 0.2372,
      "step": 1875
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.439157766200496,
      "learning_rate": 9.626615075286626e-06,
      "loss": 0.1894,
      "step": 1876
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5111564304011444,
      "learning_rate": 9.62612369176104e-06,
      "loss": 0.2367,
      "step": 1877
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7662817959135402,
      "learning_rate": 9.625631997669757e-06,
      "loss": 0.2176,
      "step": 1878
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.648335523316646,
      "learning_rate": 9.62513999304578e-06,
      "loss": 0.2244,
      "step": 1879
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3874137180612482,
      "learning_rate": 9.624647677922143e-06,
      "loss": 0.1973,
      "step": 1880
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.567280988811608,
      "learning_rate": 9.624155052331896e-06,
      "loss": 0.2375,
      "step": 1881
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5353112251887062,
      "learning_rate": 9.623662116308108e-06,
      "loss": 0.2741,
      "step": 1882
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5896201454120966,
      "learning_rate": 9.623168869883874e-06,
      "loss": 0.2532,
      "step": 1883
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.436006516667036,
      "learning_rate": 9.622675313092307e-06,
      "loss": 0.2445,
      "step": 1884
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6584035529200605,
      "learning_rate": 9.622181445966539e-06,
      "loss": 0.2372,
      "step": 1885
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3677471570809905,
      "learning_rate": 9.621687268539725e-06,
      "loss": 0.1718,
      "step": 1886
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.532037447855381,
      "learning_rate": 9.621192780845044e-06,
      "loss": 0.2068,
      "step": 1887
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5706662625356325,
      "learning_rate": 9.620697982915688e-06,
      "loss": 0.238,
      "step": 1888
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.584125466194905,
      "learning_rate": 9.620202874784878e-06,
      "loss": 0.2652,
      "step": 1889
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5165891746076328,
      "learning_rate": 9.619707456485848e-06,
      "loss": 0.2173,
      "step": 1890
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3596238825379194,
      "learning_rate": 9.61921172805186e-06,
      "loss": 0.1858,
      "step": 1891
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3782633084287081,
      "learning_rate": 9.618715689516194e-06,
      "loss": 0.2025,
      "step": 1892
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6404819007168323,
      "learning_rate": 9.61821934091215e-06,
      "loss": 0.2759,
      "step": 1893
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4671685267446017,
      "learning_rate": 9.617722682273048e-06,
      "loss": 0.2588,
      "step": 1894
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.802647054447075,
      "learning_rate": 9.61722571363223e-06,
      "loss": 0.2539,
      "step": 1895
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3088073009265881,
      "learning_rate": 9.616728435023061e-06,
      "loss": 0.2152,
      "step": 1896
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7437712948100854,
      "learning_rate": 9.616230846478925e-06,
      "loss": 0.2605,
      "step": 1897
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.552820013193392,
      "learning_rate": 9.615732948033225e-06,
      "loss": 0.2466,
      "step": 1898
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3554777950492651,
      "learning_rate": 9.615234739719387e-06,
      "loss": 0.2383,
      "step": 1899
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4223252507208866,
      "learning_rate": 9.61473622157086e-06,
      "loss": 0.2225,
      "step": 1900
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2953329567490843,
      "learning_rate": 9.614237393621104e-06,
      "loss": 0.2114,
      "step": 1901
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5240746242103045,
      "learning_rate": 9.613738255903613e-06,
      "loss": 0.2022,
      "step": 1902
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4085098899521087,
      "learning_rate": 9.613238808451894e-06,
      "loss": 0.2233,
      "step": 1903
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5415613357623623,
      "learning_rate": 9.612739051299477e-06,
      "loss": 0.224,
      "step": 1904
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3701432841120822,
      "learning_rate": 9.61223898447991e-06,
      "loss": 0.2147,
      "step": 1905
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.141515132267844,
      "learning_rate": 9.611738608026765e-06,
      "loss": 0.5456,
      "step": 1906
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.94252467747008,
      "learning_rate": 9.611237921973637e-06,
      "loss": 0.5873,
      "step": 1907
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5175182247666472,
      "learning_rate": 9.610736926354133e-06,
      "loss": 0.2015,
      "step": 1908
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6791296685318229,
      "learning_rate": 9.61023562120189e-06,
      "loss": 0.2576,
      "step": 1909
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.540540676075431,
      "learning_rate": 9.609734006550562e-06,
      "loss": 0.2803,
      "step": 1910
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3932638652674971,
      "learning_rate": 9.609232082433824e-06,
      "loss": 0.2491,
      "step": 1911
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.0673565846479125,
      "learning_rate": 9.608729848885369e-06,
      "loss": 0.5702,
      "step": 1912
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4628770278104222,
      "learning_rate": 9.608227305938915e-06,
      "loss": 0.2505,
      "step": 1913
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4556220972378124,
      "learning_rate": 9.6077244536282e-06,
      "loss": 0.2165,
      "step": 1914
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3776072477629275,
      "learning_rate": 9.607221291986983e-06,
      "loss": 0.2051,
      "step": 1915
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6109416961035818,
      "learning_rate": 9.606717821049042e-06,
      "loss": 0.2318,
      "step": 1916
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.564011078240842,
      "learning_rate": 9.606214040848174e-06,
      "loss": 0.6679,
      "step": 1917
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4734363571316462,
      "learning_rate": 9.605709951418201e-06,
      "loss": 0.2191,
      "step": 1918
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.415947605045889,
      "learning_rate": 9.605205552792964e-06,
      "loss": 0.2592,
      "step": 1919
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3553576209548763,
      "learning_rate": 9.604700845006326e-06,
      "loss": 0.1879,
      "step": 1920
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.770096408550739,
      "learning_rate": 9.604195828092169e-06,
      "loss": 0.6087,
      "step": 1921
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.436176122355819,
      "learning_rate": 9.603690502084396e-06,
      "loss": 0.2166,
      "step": 1922
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.532933079910172,
      "learning_rate": 9.603184867016929e-06,
      "loss": 0.2502,
      "step": 1923
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6724162619718177,
      "learning_rate": 9.602678922923716e-06,
      "loss": 0.2565,
      "step": 1924
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4617272488796216,
      "learning_rate": 9.602172669838721e-06,
      "loss": 0.175,
      "step": 1925
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4826564657118184,
      "learning_rate": 9.60166610779593e-06,
      "loss": 0.2258,
      "step": 1926
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5029089606156945,
      "learning_rate": 9.601159236829353e-06,
      "loss": 0.2181,
      "step": 1927
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3315757135444566,
      "learning_rate": 9.600652056973013e-06,
      "loss": 0.2024,
      "step": 1928
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7744990086828667,
      "learning_rate": 9.600144568260962e-06,
      "loss": 0.205,
      "step": 1929
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.464807782821732,
      "learning_rate": 9.59963677072727e-06,
      "loss": 0.2625,
      "step": 1930
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4557584757300288,
      "learning_rate": 9.599128664406023e-06,
      "loss": 0.2229,
      "step": 1931
    },
    {
      "epoch": 0.15,
      "grad_norm": 20.049400808877856,
      "learning_rate": 9.598620249331334e-06,
      "loss": 0.5854,
      "step": 1932
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5335096183442203,
      "learning_rate": 9.598111525537336e-06,
      "loss": 0.2668,
      "step": 1933
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.728029813280104,
      "learning_rate": 9.597602493058178e-06,
      "loss": 0.2158,
      "step": 1934
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2578209304611148,
      "learning_rate": 9.597093151928035e-06,
      "loss": 0.1605,
      "step": 1935
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.445192653172929,
      "learning_rate": 9.5965835021811e-06,
      "loss": 0.6865,
      "step": 1936
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.155428312150652,
      "learning_rate": 9.596073543851587e-06,
      "loss": 0.7152,
      "step": 1937
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2743830337819548,
      "learning_rate": 9.595563276973732e-06,
      "loss": 0.166,
      "step": 1938
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4978080161236704,
      "learning_rate": 9.59505270158179e-06,
      "loss": 0.2852,
      "step": 1939
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2297690136207637,
      "learning_rate": 9.594541817710039e-06,
      "loss": 0.1969,
      "step": 1940
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5917862593006424,
      "learning_rate": 9.594030625392772e-06,
      "loss": 0.2255,
      "step": 1941
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4976831586923893,
      "learning_rate": 9.593519124664313e-06,
      "loss": 0.1637,
      "step": 1942
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4850846003247933,
      "learning_rate": 9.593007315558996e-06,
      "loss": 0.2559,
      "step": 1943
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.497901466411517,
      "learning_rate": 9.59249519811118e-06,
      "loss": 0.2143,
      "step": 1944
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8133958958486953,
      "learning_rate": 9.591982772355248e-06,
      "loss": 0.2308,
      "step": 1945
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5158881167945921,
      "learning_rate": 9.591470038325599e-06,
      "loss": 0.2385,
      "step": 1946
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5931407945641844,
      "learning_rate": 9.590956996056656e-06,
      "loss": 0.2045,
      "step": 1947
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.875339324321946,
      "learning_rate": 9.590443645582859e-06,
      "loss": 0.575,
      "step": 1948
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.413503531708616,
      "learning_rate": 9.58992998693867e-06,
      "loss": 0.5894,
      "step": 1949
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3899183016364773,
      "learning_rate": 9.589416020158577e-06,
      "loss": 0.224,
      "step": 1950
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.616174180479946,
      "learning_rate": 9.58890174527708e-06,
      "loss": 0.238,
      "step": 1951
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2557217304251171,
      "learning_rate": 9.588387162328705e-06,
      "loss": 0.1765,
      "step": 1952
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.241650721238723,
      "learning_rate": 9.587872271347996e-06,
      "loss": 0.1956,
      "step": 1953
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.182374014235162,
      "learning_rate": 9.587357072369522e-06,
      "loss": 0.4694,
      "step": 1954
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.495638385794956,
      "learning_rate": 9.586841565427869e-06,
      "loss": 0.6953,
      "step": 1955
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5234717979887953,
      "learning_rate": 9.586325750557643e-06,
      "loss": 0.232,
      "step": 1956
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3231156939189745,
      "learning_rate": 9.585809627793475e-06,
      "loss": 0.2181,
      "step": 1957
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6643409253493608,
      "learning_rate": 9.58529319717001e-06,
      "loss": 0.2483,
      "step": 1958
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.220945638880487,
      "learning_rate": 9.584776458721922e-06,
      "loss": 0.5609,
      "step": 1959
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5189513806058108,
      "learning_rate": 9.584259412483899e-06,
      "loss": 0.2371,
      "step": 1960
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3992559758248637,
      "learning_rate": 9.58374205849065e-06,
      "loss": 0.2309,
      "step": 1961
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.457210197620917,
      "learning_rate": 9.58322439677691e-06,
      "loss": 0.6165,
      "step": 1962
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.6775682801719665,
      "learning_rate": 9.58270642737743e-06,
      "loss": 0.5421,
      "step": 1963
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4032638645107387,
      "learning_rate": 9.582188150326981e-06,
      "loss": 0.2113,
      "step": 1964
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.702253330389171,
      "learning_rate": 9.58166956566036e-06,
      "loss": 0.28,
      "step": 1965
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.902093164871039,
      "learning_rate": 9.581150673412376e-06,
      "loss": 0.8045,
      "step": 1966
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4321327890486761,
      "learning_rate": 9.58063147361787e-06,
      "loss": 0.2108,
      "step": 1967
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3360277324608862,
      "learning_rate": 9.580111966311692e-06,
      "loss": 0.1979,
      "step": 1968
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5780497789719112,
      "learning_rate": 9.579592151528721e-06,
      "loss": 0.2494,
      "step": 1969
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4902731643983842,
      "learning_rate": 9.579072029303855e-06,
      "loss": 0.2485,
      "step": 1970
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.843284786117048,
      "learning_rate": 9.578551599672008e-06,
      "loss": 0.649,
      "step": 1971
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4033494890839562,
      "learning_rate": 9.57803086266812e-06,
      "loss": 0.1878,
      "step": 1972
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2682486523733965,
      "learning_rate": 9.57750981832715e-06,
      "loss": 0.1982,
      "step": 1973
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4589166214005198,
      "learning_rate": 9.576988466684077e-06,
      "loss": 0.2242,
      "step": 1974
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5320888780628783,
      "learning_rate": 9.5764668077739e-06,
      "loss": 0.2255,
      "step": 1975
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.44560694881027,
      "learning_rate": 9.575944841631636e-06,
      "loss": 0.2258,
      "step": 1976
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2640671370808514,
      "learning_rate": 9.575422568292336e-06,
      "loss": 0.1815,
      "step": 1977
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4448004433267245,
      "learning_rate": 9.574899987791054e-06,
      "loss": 0.1947,
      "step": 1978
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.53999420291616,
      "learning_rate": 9.574377100162874e-06,
      "loss": 0.2613,
      "step": 1979
    },
    {
      "epoch": 0.16,
      "grad_norm": 21.608890307131713,
      "learning_rate": 9.573853905442899e-06,
      "loss": 0.6941,
      "step": 1980
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4569898264642869,
      "learning_rate": 9.573330403666254e-06,
      "loss": 0.1742,
      "step": 1981
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3107620253197376,
      "learning_rate": 9.572806594868082e-06,
      "loss": 0.2263,
      "step": 1982
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5460657706140506,
      "learning_rate": 9.572282479083548e-06,
      "loss": 0.2216,
      "step": 1983
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4814982356585722,
      "learning_rate": 9.571758056347839e-06,
      "loss": 0.2374,
      "step": 1984
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5328835440939412,
      "learning_rate": 9.571233326696159e-06,
      "loss": 0.207,
      "step": 1985
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8486166299418112,
      "learning_rate": 9.570708290163735e-06,
      "loss": 0.2552,
      "step": 1986
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4918509186981561,
      "learning_rate": 9.570182946785816e-06,
      "loss": 0.208,
      "step": 1987
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3786445868384574,
      "learning_rate": 9.569657296597668e-06,
      "loss": 0.2046,
      "step": 1988
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5473488657885672,
      "learning_rate": 9.569131339634578e-06,
      "loss": 0.2318,
      "step": 1989
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.033039266230712,
      "learning_rate": 9.56860507593186e-06,
      "loss": 0.5917,
      "step": 1990
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.369799275992181,
      "learning_rate": 9.56807850552484e-06,
      "loss": 0.2377,
      "step": 1991
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.552840651691266,
      "learning_rate": 9.56755162844887e-06,
      "loss": 0.1961,
      "step": 1992
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5105102349890958,
      "learning_rate": 9.567024444739319e-06,
      "loss": 0.2288,
      "step": 1993
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3884346756602388,
      "learning_rate": 9.566496954431581e-06,
      "loss": 0.2376,
      "step": 1994
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.548709650092197,
      "learning_rate": 9.565969157561066e-06,
      "loss": 0.1823,
      "step": 1995
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.545572645035,
      "learning_rate": 9.565441054163205e-06,
      "loss": 0.2296,
      "step": 1996
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.50076542549949,
      "learning_rate": 9.564912644273456e-06,
      "loss": 0.2596,
      "step": 1997
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4720231006663815,
      "learning_rate": 9.564383927927289e-06,
      "loss": 0.2588,
      "step": 1998
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5452056101398999,
      "learning_rate": 9.5638549051602e-06,
      "loss": 0.2464,
      "step": 1999
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5219195446778333,
      "learning_rate": 9.563325576007702e-06,
      "loss": 0.2344,
      "step": 2000
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.557654921930902,
      "learning_rate": 9.562795940505332e-06,
      "loss": 0.2622,
      "step": 2001
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6739214624234389,
      "learning_rate": 9.562265998688648e-06,
      "loss": 0.2572,
      "step": 2002
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4127068911803078,
      "learning_rate": 9.561735750593221e-06,
      "loss": 0.2076,
      "step": 2003
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2916223613666138,
      "learning_rate": 9.561205196254652e-06,
      "loss": 0.188,
      "step": 2004
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5532007439461248,
      "learning_rate": 9.56067433570856e-06,
      "loss": 0.2443,
      "step": 2005
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4665407842945997,
      "learning_rate": 9.56014316899058e-06,
      "loss": 0.2112,
      "step": 2006
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3616723481387611,
      "learning_rate": 9.559611696136375e-06,
      "loss": 0.2328,
      "step": 2007
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3875531285528953,
      "learning_rate": 9.55907991718162e-06,
      "loss": 0.2432,
      "step": 2008
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3337642493496045,
      "learning_rate": 9.558547832162017e-06,
      "loss": 0.192,
      "step": 2009
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5039344008795261,
      "learning_rate": 9.558015441113285e-06,
      "loss": 0.2533,
      "step": 2010
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4983109918612854,
      "learning_rate": 9.557482744071166e-06,
      "loss": 0.2314,
      "step": 2011
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4345967592725362,
      "learning_rate": 9.556949741071423e-06,
      "loss": 0.2179,
      "step": 2012
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4163479549051405,
      "learning_rate": 9.556416432149838e-06,
      "loss": 0.2055,
      "step": 2013
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5432022347313818,
      "learning_rate": 9.555882817342212e-06,
      "loss": 0.1913,
      "step": 2014
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4667905992905719,
      "learning_rate": 9.555348896684366e-06,
      "loss": 0.2213,
      "step": 2015
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.525694519663082,
      "learning_rate": 9.55481467021215e-06,
      "loss": 0.2265,
      "step": 2016
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4885313463161238,
      "learning_rate": 9.554280137961423e-06,
      "loss": 0.1936,
      "step": 2017
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7979582509944723,
      "learning_rate": 9.553745299968071e-06,
      "loss": 0.2786,
      "step": 2018
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.911634735118961,
      "learning_rate": 9.553210156268e-06,
      "loss": 0.8003,
      "step": 2019
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4381219281716529,
      "learning_rate": 9.552674706897136e-06,
      "loss": 0.232,
      "step": 2020
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4646740875389157,
      "learning_rate": 9.552138951891425e-06,
      "loss": 0.189,
      "step": 2021
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4032173156466463,
      "learning_rate": 9.551602891286833e-06,
      "loss": 0.2177,
      "step": 2022
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3365001270540964,
      "learning_rate": 9.551066525119349e-06,
      "loss": 0.2423,
      "step": 2023
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6235414420461314,
      "learning_rate": 9.550529853424979e-06,
      "loss": 0.2659,
      "step": 2024
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5801991810269655,
      "learning_rate": 9.549992876239753e-06,
      "loss": 0.2185,
      "step": 2025
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4236622626202178,
      "learning_rate": 9.54945559359972e-06,
      "loss": 0.1621,
      "step": 2026
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2526788461689256,
      "learning_rate": 9.548918005540948e-06,
      "loss": 0.2026,
      "step": 2027
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.359433016465041,
      "learning_rate": 9.548380112099527e-06,
      "loss": 0.2424,
      "step": 2028
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.412417354573437,
      "learning_rate": 9.547841913311567e-06,
      "loss": 0.1905,
      "step": 2029
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5736360950442416,
      "learning_rate": 9.547303409213202e-06,
      "loss": 0.2728,
      "step": 2030
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3870607442195202,
      "learning_rate": 9.546764599840581e-06,
      "loss": 0.231,
      "step": 2031
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.685973324326073,
      "learning_rate": 9.546225485229876e-06,
      "loss": 0.6292,
      "step": 2032
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4805895927061385,
      "learning_rate": 9.545686065417279e-06,
      "loss": 0.2318,
      "step": 2033
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4034043598475996,
      "learning_rate": 9.545146340439005e-06,
      "loss": 0.1979,
      "step": 2034
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.493837060678656,
      "learning_rate": 9.544606310331284e-06,
      "loss": 0.4933,
      "step": 2035
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.392240166641213,
      "learning_rate": 9.544065975130375e-06,
      "loss": 0.1941,
      "step": 2036
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4290041631367258,
      "learning_rate": 9.543525334872546e-06,
      "loss": 0.2174,
      "step": 2037
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4726713094571027,
      "learning_rate": 9.542984389594096e-06,
      "loss": 0.2417,
      "step": 2038
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5966915430033186,
      "learning_rate": 9.54244313933134e-06,
      "loss": 0.2323,
      "step": 2039
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3987276596032185,
      "learning_rate": 9.541901584120612e-06,
      "loss": 0.216,
      "step": 2040
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5736230177539783,
      "learning_rate": 9.541359723998268e-06,
      "loss": 0.275,
      "step": 2041
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.159315491315751,
      "learning_rate": 9.540817559000688e-06,
      "loss": 0.2021,
      "step": 2042
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4957615294768731,
      "learning_rate": 9.540275089164266e-06,
      "loss": 0.2567,
      "step": 2043
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4694328640662935,
      "learning_rate": 9.539732314525421e-06,
      "loss": 0.2056,
      "step": 2044
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4735681707955224,
      "learning_rate": 9.539189235120591e-06,
      "loss": 0.2198,
      "step": 2045
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.370865028932725,
      "learning_rate": 9.538645850986235e-06,
      "loss": 0.2182,
      "step": 2046
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4694703789697863,
      "learning_rate": 9.538102162158832e-06,
      "loss": 0.2018,
      "step": 2047
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6008991062691,
      "learning_rate": 9.53755816867488e-06,
      "loss": 0.25,
      "step": 2048
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4290205669845732,
      "learning_rate": 9.5370138705709e-06,
      "loss": 0.2134,
      "step": 2049
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.524218497245832,
      "learning_rate": 9.536469267883432e-06,
      "loss": 0.5738,
      "step": 2050
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5826078574426712,
      "learning_rate": 9.535924360649038e-06,
      "loss": 0.2056,
      "step": 2051
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4485455108272356,
      "learning_rate": 9.535379148904297e-06,
      "loss": 0.2206,
      "step": 2052
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4123381063334186,
      "learning_rate": 9.534833632685813e-06,
      "loss": 0.2076,
      "step": 2053
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4042409562289215,
      "learning_rate": 9.534287812030207e-06,
      "loss": 0.2568,
      "step": 2054
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5448729375586674,
      "learning_rate": 9.533741686974122e-06,
      "loss": 0.2278,
      "step": 2055
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3972955293158567,
      "learning_rate": 9.53319525755422e-06,
      "loss": 0.1769,
      "step": 2056
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3729575846573594,
      "learning_rate": 9.532648523807186e-06,
      "loss": 0.2158,
      "step": 2057
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.45973639599006,
      "learning_rate": 9.532101485769723e-06,
      "loss": 0.5855,
      "step": 2058
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3411493720748247,
      "learning_rate": 9.531554143478556e-06,
      "loss": 0.2057,
      "step": 2059
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1776009501329519,
      "learning_rate": 9.53100649697043e-06,
      "loss": 0.1607,
      "step": 2060
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5401496818725513,
      "learning_rate": 9.530458546282108e-06,
      "loss": 0.2706,
      "step": 2061
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.508100732135895,
      "learning_rate": 9.529910291450377e-06,
      "loss": 0.2664,
      "step": 2062
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5642677004810706,
      "learning_rate": 9.529361732512044e-06,
      "loss": 0.232,
      "step": 2063
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.495726002865652,
      "learning_rate": 9.528812869503934e-06,
      "loss": 0.1994,
      "step": 2064
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.309080135006772,
      "learning_rate": 9.528263702462894e-06,
      "loss": 0.631,
      "step": 2065
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6086823594169135,
      "learning_rate": 9.527714231425793e-06,
      "loss": 0.2446,
      "step": 2066
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3925787350849426,
      "learning_rate": 9.527164456429517e-06,
      "loss": 0.2425,
      "step": 2067
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.252624648772561,
      "learning_rate": 9.526614377510971e-06,
      "loss": 0.5082,
      "step": 2068
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.406857417188421,
      "learning_rate": 9.526063994707091e-06,
      "loss": 0.2145,
      "step": 2069
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2559383225222025,
      "learning_rate": 9.525513308054818e-06,
      "loss": 0.1841,
      "step": 2070
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6739316440920449,
      "learning_rate": 9.524962317591128e-06,
      "loss": 0.2422,
      "step": 2071
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4333957233835004,
      "learning_rate": 9.524411023353007e-06,
      "loss": 0.2087,
      "step": 2072
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.159383298981887,
      "learning_rate": 9.523859425377464e-06,
      "loss": 0.6348,
      "step": 2073
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5466566084137177,
      "learning_rate": 9.523307523701532e-06,
      "loss": 0.2704,
      "step": 2074
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.433444494139256,
      "learning_rate": 9.52275531836226e-06,
      "loss": 0.1978,
      "step": 2075
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4712908303896224,
      "learning_rate": 9.522202809396721e-06,
      "loss": 0.2111,
      "step": 2076
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4193624016751663,
      "learning_rate": 9.521649996842006e-06,
      "loss": 0.2157,
      "step": 2077
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.407578954997005,
      "learning_rate": 9.521096880735226e-06,
      "loss": 0.5117,
      "step": 2078
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3940715489580375,
      "learning_rate": 9.520543461113514e-06,
      "loss": 0.2256,
      "step": 2079
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5481874725980844,
      "learning_rate": 9.519989738014022e-06,
      "loss": 0.2672,
      "step": 2080
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.459479265706027,
      "learning_rate": 9.519435711473926e-06,
      "loss": 0.2803,
      "step": 2081
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6033144472552365,
      "learning_rate": 9.518881381530415e-06,
      "loss": 0.1971,
      "step": 2082
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3915288260580332,
      "learning_rate": 9.518326748220707e-06,
      "loss": 0.2283,
      "step": 2083
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3302037641307136,
      "learning_rate": 9.517771811582033e-06,
      "loss": 0.2342,
      "step": 2084
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3679423428982893,
      "learning_rate": 9.51721657165165e-06,
      "loss": 0.2202,
      "step": 2085
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3855493969089079,
      "learning_rate": 9.51666102846683e-06,
      "loss": 0.1825,
      "step": 2086
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6501583709617222,
      "learning_rate": 9.516105182064872e-06,
      "loss": 0.2231,
      "step": 2087
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4201177790419752,
      "learning_rate": 9.515549032483091e-06,
      "loss": 0.2075,
      "step": 2088
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.341568882933472,
      "learning_rate": 9.51499257975882e-06,
      "loss": 0.1821,
      "step": 2089
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.091598968673305,
      "learning_rate": 9.514435823929418e-06,
      "loss": 0.4889,
      "step": 2090
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4453792949540785,
      "learning_rate": 9.51387876503226e-06,
      "loss": 0.2102,
      "step": 2091
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.213033142639318,
      "learning_rate": 9.513321403104745e-06,
      "loss": 0.1777,
      "step": 2092
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3778318686899453,
      "learning_rate": 9.512763738184289e-06,
      "loss": 0.1802,
      "step": 2093
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3576336873725359,
      "learning_rate": 9.51220577030833e-06,
      "loss": 0.2045,
      "step": 2094
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2853360472502422,
      "learning_rate": 9.511647499514327e-06,
      "loss": 0.1939,
      "step": 2095
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3877022394839709,
      "learning_rate": 9.511088925839758e-06,
      "loss": 0.2376,
      "step": 2096
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3181303775408801,
      "learning_rate": 9.51053004932212e-06,
      "loss": 0.2082,
      "step": 2097
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4625666963559458,
      "learning_rate": 9.509970869998933e-06,
      "loss": 0.273,
      "step": 2098
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4917515500282716,
      "learning_rate": 9.509411387907738e-06,
      "loss": 0.1954,
      "step": 2099
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.419326449804372,
      "learning_rate": 9.508851603086094e-06,
      "loss": 0.5762,
      "step": 2100
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4734182074506141,
      "learning_rate": 9.50829151557158e-06,
      "loss": 0.2343,
      "step": 2101
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4262186852452459,
      "learning_rate": 9.507731125401795e-06,
      "loss": 0.1929,
      "step": 2102
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.310389441278163,
      "learning_rate": 9.507170432614364e-06,
      "loss": 0.1959,
      "step": 2103
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4456713785030906,
      "learning_rate": 9.506609437246924e-06,
      "loss": 0.2209,
      "step": 2104
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5400771409153984,
      "learning_rate": 9.506048139337142e-06,
      "loss": 0.2259,
      "step": 2105
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.372996562194523,
      "learning_rate": 9.50548653892269e-06,
      "loss": 0.5097,
      "step": 2106
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3971378017901916,
      "learning_rate": 9.504924636041279e-06,
      "loss": 0.1831,
      "step": 2107
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5533099875500511,
      "learning_rate": 9.504362430730627e-06,
      "loss": 0.2745,
      "step": 2108
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3941673192517146,
      "learning_rate": 9.503799923028478e-06,
      "loss": 0.2061,
      "step": 2109
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5649265055545385,
      "learning_rate": 9.503237112972594e-06,
      "loss": 0.2367,
      "step": 2110
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2898029001234865,
      "learning_rate": 9.502674000600758e-06,
      "loss": 0.1878,
      "step": 2111
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.109576612651166,
      "learning_rate": 9.502110585950773e-06,
      "loss": 0.6511,
      "step": 2112
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5188392150368415,
      "learning_rate": 9.501546869060466e-06,
      "loss": 0.2595,
      "step": 2113
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5330604673304256,
      "learning_rate": 9.500982849967674e-06,
      "loss": 0.1868,
      "step": 2114
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4843998264605276,
      "learning_rate": 9.50041852871027e-06,
      "loss": 0.2272,
      "step": 2115
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.352627389988722,
      "learning_rate": 9.499853905326133e-06,
      "loss": 0.2411,
      "step": 2116
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4824838907866142,
      "learning_rate": 9.49928897985317e-06,
      "loss": 0.2476,
      "step": 2117
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2518045357658587,
      "learning_rate": 9.498723752329304e-06,
      "loss": 0.1926,
      "step": 2118
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2994409096323183,
      "learning_rate": 9.498158222792484e-06,
      "loss": 0.2033,
      "step": 2119
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.319524117655798,
      "learning_rate": 9.497592391280672e-06,
      "loss": 0.2061,
      "step": 2120
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4988872726996796,
      "learning_rate": 9.497026257831856e-06,
      "loss": 0.2258,
      "step": 2121
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6105837276298889,
      "learning_rate": 9.496459822484043e-06,
      "loss": 0.194,
      "step": 2122
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.550800217821765,
      "learning_rate": 9.495893085275256e-06,
      "loss": 0.7505,
      "step": 2123
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5692469587642102,
      "learning_rate": 9.495326046243547e-06,
      "loss": 0.2545,
      "step": 2124
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.54748852508291,
      "learning_rate": 9.494758705426978e-06,
      "loss": 0.2978,
      "step": 2125
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.49079564934414,
      "learning_rate": 9.494191062863638e-06,
      "loss": 0.2052,
      "step": 2126
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3271125793021075,
      "learning_rate": 9.493623118591638e-06,
      "loss": 0.2153,
      "step": 2127
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.680213730970512,
      "learning_rate": 9.4930548726491e-06,
      "loss": 0.5648,
      "step": 2128
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5242805027692423,
      "learning_rate": 9.492486325074177e-06,
      "loss": 0.2018,
      "step": 2129
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4846476766560741,
      "learning_rate": 9.491917475905034e-06,
      "loss": 0.234,
      "step": 2130
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.525320937546331,
      "learning_rate": 9.491348325179861e-06,
      "loss": 0.2051,
      "step": 2131
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.820613291417657,
      "learning_rate": 9.490778872936867e-06,
      "loss": 0.1938,
      "step": 2132
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.288141869809796,
      "learning_rate": 9.490209119214282e-06,
      "loss": 0.1612,
      "step": 2133
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5047683918927903,
      "learning_rate": 9.489639064050353e-06,
      "loss": 0.2347,
      "step": 2134
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.346380564933354,
      "learning_rate": 9.48906870748335e-06,
      "loss": 0.6837,
      "step": 2135
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4627085858360793,
      "learning_rate": 9.488498049551563e-06,
      "loss": 0.2368,
      "step": 2136
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.333896780750312,
      "learning_rate": 9.487927090293302e-06,
      "loss": 0.5522,
      "step": 2137
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4753513405176741,
      "learning_rate": 9.4873558297469e-06,
      "loss": 0.2504,
      "step": 2138
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4863404734720786,
      "learning_rate": 9.486784267950704e-06,
      "loss": 0.289,
      "step": 2139
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4997659279544855,
      "learning_rate": 9.486212404943084e-06,
      "loss": 0.2192,
      "step": 2140
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3558126149183962,
      "learning_rate": 9.485640240762434e-06,
      "loss": 0.2306,
      "step": 2141
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6283515566627347,
      "learning_rate": 9.485067775447164e-06,
      "loss": 0.26,
      "step": 2142
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6117645681041304,
      "learning_rate": 9.484495009035705e-06,
      "loss": 0.3015,
      "step": 2143
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5317196743038302,
      "learning_rate": 9.483921941566508e-06,
      "loss": 0.2423,
      "step": 2144
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.504027963909363,
      "learning_rate": 9.483348573078046e-06,
      "loss": 0.2268,
      "step": 2145
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3338812570267844,
      "learning_rate": 9.48277490360881e-06,
      "loss": 0.2343,
      "step": 2146
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8553412194552557,
      "learning_rate": 9.482200933197312e-06,
      "loss": 0.2446,
      "step": 2147
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4586064422482663,
      "learning_rate": 9.481626661882084e-06,
      "loss": 0.211,
      "step": 2148
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5060889303335863,
      "learning_rate": 9.48105208970168e-06,
      "loss": 0.2029,
      "step": 2149
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4789353744808005,
      "learning_rate": 9.480477216694674e-06,
      "loss": 0.2446,
      "step": 2150
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.536782965808731,
      "learning_rate": 9.479902042899655e-06,
      "loss": 0.2579,
      "step": 2151
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4569749023819965,
      "learning_rate": 9.47932656835524e-06,
      "loss": 0.2493,
      "step": 2152
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.26424453942703,
      "learning_rate": 9.47875079310006e-06,
      "loss": 0.4321,
      "step": 2153
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4323574846328135,
      "learning_rate": 9.478174717172768e-06,
      "loss": 0.2283,
      "step": 2154
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.558275622189435,
      "learning_rate": 9.47759834061204e-06,
      "loss": 0.2467,
      "step": 2155
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5004812858716223,
      "learning_rate": 9.477021663456569e-06,
      "loss": 0.2363,
      "step": 2156
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.413430893216606,
      "learning_rate": 9.476444685745067e-06,
      "loss": 0.7435,
      "step": 2157
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.552531712318251,
      "learning_rate": 9.475867407516272e-06,
      "loss": 0.2535,
      "step": 2158
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.091096127791431,
      "learning_rate": 9.475289828808937e-06,
      "loss": 0.682,
      "step": 2159
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2992279651051872,
      "learning_rate": 9.474711949661835e-06,
      "loss": 0.1956,
      "step": 2160
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5254447756918428,
      "learning_rate": 9.474133770113763e-06,
      "loss": 0.2428,
      "step": 2161
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.56535565203796,
      "learning_rate": 9.473555290203534e-06,
      "loss": 0.2003,
      "step": 2162
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4146840246637304,
      "learning_rate": 9.472976509969984e-06,
      "loss": 0.1744,
      "step": 2163
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.972813395348676,
      "learning_rate": 9.47239742945197e-06,
      "loss": 0.603,
      "step": 2164
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.391583479936229,
      "learning_rate": 9.471818048688364e-06,
      "loss": 0.7076,
      "step": 2165
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5273605663717216,
      "learning_rate": 9.471238367718064e-06,
      "loss": 0.2475,
      "step": 2166
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5058994448809844,
      "learning_rate": 9.470658386579983e-06,
      "loss": 0.2424,
      "step": 2167
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4124827274546083,
      "learning_rate": 9.470078105313062e-06,
      "loss": 0.194,
      "step": 2168
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.294539221616167,
      "learning_rate": 9.469497523956253e-06,
      "loss": 0.1832,
      "step": 2169
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4673909426449274,
      "learning_rate": 9.468916642548534e-06,
      "loss": 0.2666,
      "step": 2170
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.307828627227948,
      "learning_rate": 9.468335461128898e-06,
      "loss": 0.6339,
      "step": 2171
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.096639182785343,
      "learning_rate": 9.467753979736365e-06,
      "loss": 0.7023,
      "step": 2172
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.286089408773351,
      "learning_rate": 9.467172198409971e-06,
      "loss": 0.1935,
      "step": 2173
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3258447360043475,
      "learning_rate": 9.466590117188773e-06,
      "loss": 0.1917,
      "step": 2174
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6636379762354336,
      "learning_rate": 9.466007736111846e-06,
      "loss": 0.2243,
      "step": 2175
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2789750861085092,
      "learning_rate": 9.465425055218289e-06,
      "loss": 0.1868,
      "step": 2176
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4551335578557851,
      "learning_rate": 9.464842074547218e-06,
      "loss": 0.24,
      "step": 2177
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3278702168404326,
      "learning_rate": 9.464258794137771e-06,
      "loss": 0.2532,
      "step": 2178
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.437916550773484,
      "learning_rate": 9.463675214029104e-06,
      "loss": 0.2822,
      "step": 2179
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.129141880050032,
      "learning_rate": 9.463091334260397e-06,
      "loss": 0.5549,
      "step": 2180
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2691856699432087,
      "learning_rate": 9.462507154870846e-06,
      "loss": 0.223,
      "step": 2181
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5007735901182446,
      "learning_rate": 9.461922675899668e-06,
      "loss": 0.2059,
      "step": 2182
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4137550893693016,
      "learning_rate": 9.4613378973861e-06,
      "loss": 0.181,
      "step": 2183
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.255164882730478,
      "learning_rate": 9.460752819369405e-06,
      "loss": 0.2046,
      "step": 2184
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4414349439819372,
      "learning_rate": 9.460167441888855e-06,
      "loss": 0.1944,
      "step": 2185
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3020724216845385,
      "learning_rate": 9.459581764983752e-06,
      "loss": 0.2041,
      "step": 2186
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4215039690706694,
      "learning_rate": 9.458995788693414e-06,
      "loss": 0.2284,
      "step": 2187
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3489758337410664,
      "learning_rate": 9.458409513057176e-06,
      "loss": 0.2143,
      "step": 2188
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5903309201331473,
      "learning_rate": 9.457822938114401e-06,
      "loss": 0.2578,
      "step": 2189
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.295301861322975,
      "learning_rate": 9.457236063904465e-06,
      "loss": 0.2437,
      "step": 2190
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.964882309439014,
      "learning_rate": 9.456648890466767e-06,
      "loss": 0.7316,
      "step": 2191
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.963241638325417,
      "learning_rate": 9.456061417840727e-06,
      "loss": 0.604,
      "step": 2192
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.417310052415083,
      "learning_rate": 9.455473646065782e-06,
      "loss": 0.2019,
      "step": 2193
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3665399739551956,
      "learning_rate": 9.454885575181391e-06,
      "loss": 0.1901,
      "step": 2194
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3937418077228976,
      "learning_rate": 9.454297205227034e-06,
      "loss": 0.2287,
      "step": 2195
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7255294675640807,
      "learning_rate": 9.45370853624221e-06,
      "loss": 0.2604,
      "step": 2196
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5374609973935005,
      "learning_rate": 9.453119568266435e-06,
      "loss": 0.2092,
      "step": 2197
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3604098807649905,
      "learning_rate": 9.452530301339254e-06,
      "loss": 0.1765,
      "step": 2198
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5340144135385738,
      "learning_rate": 9.451940735500222e-06,
      "loss": 0.207,
      "step": 2199
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4878568701443773,
      "learning_rate": 9.451350870788922e-06,
      "loss": 0.177,
      "step": 2200
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5957510906960977,
      "learning_rate": 9.450760707244948e-06,
      "loss": 0.2267,
      "step": 2201
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5303839463381181,
      "learning_rate": 9.450170244907924e-06,
      "loss": 0.1811,
      "step": 2202
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.048135481378952,
      "learning_rate": 9.449579483817487e-06,
      "loss": 0.4253,
      "step": 2203
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3785729636903892,
      "learning_rate": 9.448988424013298e-06,
      "loss": 0.2227,
      "step": 2204
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3720288581033389,
      "learning_rate": 9.448397065535037e-06,
      "loss": 0.2146,
      "step": 2205
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.32810794469314,
      "learning_rate": 9.4478054084224e-06,
      "loss": 0.2136,
      "step": 2206
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5341222755225963,
      "learning_rate": 9.447213452715114e-06,
      "loss": 0.2383,
      "step": 2207
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.560362947103128,
      "learning_rate": 9.446621198452912e-06,
      "loss": 0.5618,
      "step": 2208
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.343812869594025,
      "learning_rate": 9.446028645675556e-06,
      "loss": 0.2459,
      "step": 2209
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.472817462672581,
      "learning_rate": 9.445435794422826e-06,
      "loss": 0.2418,
      "step": 2210
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3431266951288083,
      "learning_rate": 9.444842644734521e-06,
      "loss": 0.225,
      "step": 2211
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3341464195510009,
      "learning_rate": 9.444249196650465e-06,
      "loss": 0.1882,
      "step": 2212
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2475789038924463,
      "learning_rate": 9.443655450210494e-06,
      "loss": 0.1259,
      "step": 2213
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.475017705770073,
      "learning_rate": 9.443061405454468e-06,
      "loss": 0.2344,
      "step": 2214
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3255628066564327,
      "learning_rate": 9.442467062422267e-06,
      "loss": 0.2448,
      "step": 2215
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5377187216748018,
      "learning_rate": 9.441872421153792e-06,
      "loss": 0.3013,
      "step": 2216
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3072848885932593,
      "learning_rate": 9.441277481688964e-06,
      "loss": 0.2189,
      "step": 2217
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6497872471436115,
      "learning_rate": 9.440682244067724e-06,
      "loss": 0.2129,
      "step": 2218
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3108276495834656,
      "learning_rate": 9.440086708330028e-06,
      "loss": 0.1855,
      "step": 2219
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4549609342170728,
      "learning_rate": 9.439490874515859e-06,
      "loss": 0.2126,
      "step": 2220
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.697861787304837,
      "learning_rate": 9.438894742665217e-06,
      "loss": 0.6835,
      "step": 2221
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5347843136283408,
      "learning_rate": 9.43829831281812e-06,
      "loss": 0.2544,
      "step": 2222
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.425218775855989,
      "learning_rate": 9.437701585014608e-06,
      "loss": 0.2594,
      "step": 2223
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3888331029979648,
      "learning_rate": 9.437104559294748e-06,
      "loss": 0.2194,
      "step": 2224
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7729432926220772,
      "learning_rate": 9.436507235698613e-06,
      "loss": 0.2157,
      "step": 2225
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5589357900648926,
      "learning_rate": 9.435909614266303e-06,
      "loss": 0.2417,
      "step": 2226
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.170936968921524,
      "learning_rate": 9.435311695037943e-06,
      "loss": 0.6219,
      "step": 2227
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4257273523695209,
      "learning_rate": 9.43471347805367e-06,
      "loss": 0.2331,
      "step": 2228
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7165684337153404,
      "learning_rate": 9.434114963353644e-06,
      "loss": 0.2414,
      "step": 2229
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.499368571544321,
      "learning_rate": 9.433516150978045e-06,
      "loss": 0.2321,
      "step": 2230
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4124288828783422,
      "learning_rate": 9.432917040967074e-06,
      "loss": 0.2332,
      "step": 2231
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.286425194443099,
      "learning_rate": 9.432317633360952e-06,
      "loss": 0.198,
      "step": 2232
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3630367902161011,
      "learning_rate": 9.431717928199916e-06,
      "loss": 0.2211,
      "step": 2233
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.047492976071003,
      "learning_rate": 9.43111792552423e-06,
      "loss": 0.5947,
      "step": 2234
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3783653260827184,
      "learning_rate": 9.430517625374171e-06,
      "loss": 0.201,
      "step": 2235
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.9191294011434685,
      "learning_rate": 9.42991702779004e-06,
      "loss": 0.5277,
      "step": 2236
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5429923558207153,
      "learning_rate": 9.429316132812156e-06,
      "loss": 0.2278,
      "step": 2237
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2044915910082314,
      "learning_rate": 9.428714940480861e-06,
      "loss": 0.2013,
      "step": 2238
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.424683103662734,
      "learning_rate": 9.428113450836514e-06,
      "loss": 0.2208,
      "step": 2239
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2379693625984003,
      "learning_rate": 9.427511663919492e-06,
      "loss": 0.1844,
      "step": 2240
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2824278145572752,
      "learning_rate": 9.426909579770197e-06,
      "loss": 0.1897,
      "step": 2241
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3567667380975004,
      "learning_rate": 9.426307198429053e-06,
      "loss": 0.1793,
      "step": 2242
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2598350002252732,
      "learning_rate": 9.425704519936492e-06,
      "loss": 0.1734,
      "step": 2243
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.430557818807667,
      "learning_rate": 9.425101544332979e-06,
      "loss": 0.2091,
      "step": 2244
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3784187713522942,
      "learning_rate": 9.424498271658991e-06,
      "loss": 0.2168,
      "step": 2245
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4158224674815283,
      "learning_rate": 9.423894701955028e-06,
      "loss": 0.2056,
      "step": 2246
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4250845308816862,
      "learning_rate": 9.42329083526161e-06,
      "loss": 0.2617,
      "step": 2247
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.556938676349873,
      "learning_rate": 9.422686671619277e-06,
      "loss": 0.2485,
      "step": 2248
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5543447924649205,
      "learning_rate": 9.422082211068586e-06,
      "loss": 0.2606,
      "step": 2249
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.500565891959398,
      "learning_rate": 9.421477453650118e-06,
      "loss": 0.2185,
      "step": 2250
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.431173797130157,
      "learning_rate": 9.420872399404473e-06,
      "loss": 0.2197,
      "step": 2251
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4146063595885454,
      "learning_rate": 9.42026704837227e-06,
      "loss": 0.2163,
      "step": 2252
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4833239989390457,
      "learning_rate": 9.419661400594145e-06,
      "loss": 0.2208,
      "step": 2253
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2566977666801733,
      "learning_rate": 9.41905545611076e-06,
      "loss": 0.2243,
      "step": 2254
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4782805197782598,
      "learning_rate": 9.418449214962793e-06,
      "loss": 0.2279,
      "step": 2255
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4540732160502314,
      "learning_rate": 9.417842677190944e-06,
      "loss": 0.2369,
      "step": 2256
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6929880675940345,
      "learning_rate": 9.417235842835929e-06,
      "loss": 0.2256,
      "step": 2257
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4899751865141602,
      "learning_rate": 9.416628711938489e-06,
      "loss": 0.2432,
      "step": 2258
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3691820456239192,
      "learning_rate": 9.41602128453938e-06,
      "loss": 0.2161,
      "step": 2259
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.354782503148032,
      "learning_rate": 9.415413560679385e-06,
      "loss": 0.5143,
      "step": 2260
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7227183919347557,
      "learning_rate": 9.414805540399298e-06,
      "loss": 0.2459,
      "step": 2261
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6625817310054618,
      "learning_rate": 9.414197223739939e-06,
      "loss": 0.2561,
      "step": 2262
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.482168686816286,
      "learning_rate": 9.413588610742146e-06,
      "loss": 0.2514,
      "step": 2263
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5591362269793154,
      "learning_rate": 9.412979701446776e-06,
      "loss": 0.2555,
      "step": 2264
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.370238468176538,
      "learning_rate": 9.412370495894708e-06,
      "loss": 0.2231,
      "step": 2265
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6680891909785376,
      "learning_rate": 9.41176099412684e-06,
      "loss": 0.2612,
      "step": 2266
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6364112331692031,
      "learning_rate": 9.41115119618409e-06,
      "loss": 0.2633,
      "step": 2267
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2245051535314588,
      "learning_rate": 9.410541102107394e-06,
      "loss": 0.2207,
      "step": 2268
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.169399850155293,
      "learning_rate": 9.40993071193771e-06,
      "loss": 0.6819,
      "step": 2269
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3548142848762887,
      "learning_rate": 9.409320025716018e-06,
      "loss": 0.2348,
      "step": 2270
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2966628787183774,
      "learning_rate": 9.40870904348331e-06,
      "loss": 0.2281,
      "step": 2271
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.308631681397771,
      "learning_rate": 9.408097765280608e-06,
      "loss": 0.1797,
      "step": 2272
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5015021105944997,
      "learning_rate": 9.407486191148947e-06,
      "loss": 0.2369,
      "step": 2273
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2750236753729862,
      "learning_rate": 9.406874321129384e-06,
      "loss": 0.1499,
      "step": 2274
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3288722225950578,
      "learning_rate": 9.406262155262995e-06,
      "loss": 0.2315,
      "step": 2275
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4320337509544816,
      "learning_rate": 9.405649693590877e-06,
      "loss": 0.266,
      "step": 2276
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.778047758296228,
      "learning_rate": 9.405036936154146e-06,
      "loss": 0.6217,
      "step": 2277
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.496391342001259,
      "learning_rate": 9.40442388299394e-06,
      "loss": 0.2237,
      "step": 2278
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3519231223068302,
      "learning_rate": 9.403810534151411e-06,
      "loss": 0.1814,
      "step": 2279
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3952159726439315,
      "learning_rate": 9.403196889667742e-06,
      "loss": 0.2253,
      "step": 2280
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2837361425891198,
      "learning_rate": 9.402582949584122e-06,
      "loss": 0.1717,
      "step": 2281
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5225567177379653,
      "learning_rate": 9.40196871394177e-06,
      "loss": 0.2192,
      "step": 2282
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4204188714945731,
      "learning_rate": 9.401354182781921e-06,
      "loss": 0.2132,
      "step": 2283
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4897745662555772,
      "learning_rate": 9.400739356145829e-06,
      "loss": 0.2193,
      "step": 2284
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2669533932800552,
      "learning_rate": 9.400124234074772e-06,
      "loss": 0.1891,
      "step": 2285
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6126075346202686,
      "learning_rate": 9.399508816610042e-06,
      "loss": 0.2235,
      "step": 2286
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.48447667500245,
      "learning_rate": 9.398893103792956e-06,
      "loss": 0.2237,
      "step": 2287
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4336585313324313,
      "learning_rate": 9.398277095664848e-06,
      "loss": 0.2264,
      "step": 2288
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.452047312255996,
      "learning_rate": 9.397660792267072e-06,
      "loss": 0.5344,
      "step": 2289
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5354463918934596,
      "learning_rate": 9.397044193641e-06,
      "loss": 0.2436,
      "step": 2290
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.89052187198876,
      "learning_rate": 9.396427299828033e-06,
      "loss": 0.6043,
      "step": 2291
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.6291210718842715,
      "learning_rate": 9.395810110869579e-06,
      "loss": 0.5204,
      "step": 2292
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5019113663227364,
      "learning_rate": 9.395192626807072e-06,
      "loss": 0.2559,
      "step": 2293
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4394877217179112,
      "learning_rate": 9.39457484768197e-06,
      "loss": 0.2089,
      "step": 2294
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2876524414471524,
      "learning_rate": 9.393956773535742e-06,
      "loss": 0.189,
      "step": 2295
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3627823598710707,
      "learning_rate": 9.393338404409881e-06,
      "loss": 0.223,
      "step": 2296
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3623140768346185,
      "learning_rate": 9.392719740345904e-06,
      "loss": 0.2059,
      "step": 2297
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3030991324587635,
      "learning_rate": 9.39210078138534e-06,
      "loss": 0.2202,
      "step": 2298
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5096265512260567,
      "learning_rate": 9.391481527569744e-06,
      "loss": 0.2177,
      "step": 2299
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5496217167619466,
      "learning_rate": 9.390861978940687e-06,
      "loss": 0.2018,
      "step": 2300
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3387154039716214,
      "learning_rate": 9.390242135539761e-06,
      "loss": 0.1685,
      "step": 2301
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4027077482872332,
      "learning_rate": 9.38962199740858e-06,
      "loss": 0.2063,
      "step": 2302
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.000993102823497,
      "learning_rate": 9.389001564588773e-06,
      "loss": 0.5878,
      "step": 2303
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5151366592570807,
      "learning_rate": 9.388380837121993e-06,
      "loss": 0.2121,
      "step": 2304
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.580363034275209,
      "learning_rate": 9.387759815049911e-06,
      "loss": 0.24,
      "step": 2305
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.106921095293227,
      "learning_rate": 9.38713849841422e-06,
      "loss": 0.6016,
      "step": 2306
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6172350327484344,
      "learning_rate": 9.386516887256627e-06,
      "loss": 0.2794,
      "step": 2307
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.452730447957342,
      "learning_rate": 9.385894981618866e-06,
      "loss": 0.2421,
      "step": 2308
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3796145348848519,
      "learning_rate": 9.385272781542686e-06,
      "loss": 0.2298,
      "step": 2309
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4445220872318818,
      "learning_rate": 9.384650287069856e-06,
      "loss": 0.1849,
      "step": 2310
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4470763287639767,
      "learning_rate": 9.384027498242168e-06,
      "loss": 0.191,
      "step": 2311
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4777095847548318,
      "learning_rate": 9.38340441510143e-06,
      "loss": 0.2171,
      "step": 2312
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.365387148393817,
      "learning_rate": 9.382781037689475e-06,
      "loss": 0.1972,
      "step": 2313
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.848563998496717,
      "learning_rate": 9.382157366048146e-06,
      "loss": 0.5938,
      "step": 2314
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.1728152841064245,
      "learning_rate": 9.381533400219319e-06,
      "loss": 0.6796,
      "step": 2315
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.423788544392309,
      "learning_rate": 9.380909140244878e-06,
      "loss": 0.2499,
      "step": 2316
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5508273829340398,
      "learning_rate": 9.380284586166732e-06,
      "loss": 0.2699,
      "step": 2317
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5861860823822622,
      "learning_rate": 9.379659738026812e-06,
      "loss": 0.294,
      "step": 2318
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2863509195412839,
      "learning_rate": 9.379034595867062e-06,
      "loss": 0.2004,
      "step": 2319
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3201300150206785,
      "learning_rate": 9.378409159729454e-06,
      "loss": 0.178,
      "step": 2320
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3339717944515472,
      "learning_rate": 9.37778342965597e-06,
      "loss": 0.2096,
      "step": 2321
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.530623793273264,
      "learning_rate": 9.377157405688622e-06,
      "loss": 0.2359,
      "step": 2322
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5981000538984693,
      "learning_rate": 9.376531087869435e-06,
      "loss": 0.2488,
      "step": 2323
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.562309712028729,
      "learning_rate": 9.375904476240457e-06,
      "loss": 0.4018,
      "step": 2324
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4080207935052003,
      "learning_rate": 9.37527757084375e-06,
      "loss": 0.2659,
      "step": 2325
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4912483791855708,
      "learning_rate": 9.374650371721405e-06,
      "loss": 0.2034,
      "step": 2326
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3779006290994664,
      "learning_rate": 9.374022878915525e-06,
      "loss": 0.2517,
      "step": 2327
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5894219295573149,
      "learning_rate": 9.373395092468238e-06,
      "loss": 0.2478,
      "step": 2328
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3681202981992624,
      "learning_rate": 9.372767012421687e-06,
      "loss": 0.2095,
      "step": 2329
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6157463860274324,
      "learning_rate": 9.372138638818036e-06,
      "loss": 0.242,
      "step": 2330
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.171450320342379,
      "learning_rate": 9.371509971699471e-06,
      "loss": 0.6706,
      "step": 2331
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3579047532050663,
      "learning_rate": 9.370881011108198e-06,
      "loss": 0.1917,
      "step": 2332
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6272202671548452,
      "learning_rate": 9.370251757086439e-06,
      "loss": 0.2427,
      "step": 2333
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.43516458105959,
      "learning_rate": 9.36962220967644e-06,
      "loss": 0.1965,
      "step": 2334
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5127317609894841,
      "learning_rate": 9.36899236892046e-06,
      "loss": 0.1895,
      "step": 2335
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.479339381212489,
      "learning_rate": 9.368362234860785e-06,
      "loss": 0.2413,
      "step": 2336
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3589742055044967,
      "learning_rate": 9.36773180753972e-06,
      "loss": 0.2012,
      "step": 2337
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3593057048126937,
      "learning_rate": 9.367101086999582e-06,
      "loss": 0.2046,
      "step": 2338
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.781549597069835,
      "learning_rate": 9.366470073282718e-06,
      "loss": 0.6956,
      "step": 2339
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4917263355565447,
      "learning_rate": 9.365838766431487e-06,
      "loss": 0.2498,
      "step": 2340
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4735055991746548,
      "learning_rate": 9.365207166488273e-06,
      "loss": 0.2203,
      "step": 2341
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.390455833402385,
      "learning_rate": 9.364575273495475e-06,
      "loss": 0.1999,
      "step": 2342
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3652222123307436,
      "learning_rate": 9.363943087495515e-06,
      "loss": 0.2049,
      "step": 2343
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5163772394091823,
      "learning_rate": 9.363310608530834e-06,
      "loss": 0.2051,
      "step": 2344
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2999708797126925,
      "learning_rate": 9.36267783664389e-06,
      "loss": 0.2099,
      "step": 2345
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3123219392463945,
      "learning_rate": 9.362044771877164e-06,
      "loss": 0.2149,
      "step": 2346
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3367676114336713,
      "learning_rate": 9.361411414273159e-06,
      "loss": 0.1979,
      "step": 2347
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3605132964306763,
      "learning_rate": 9.360777763874389e-06,
      "loss": 0.1637,
      "step": 2348
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5266572955723654,
      "learning_rate": 9.360143820723395e-06,
      "loss": 0.2209,
      "step": 2349
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5178431709728113,
      "learning_rate": 9.359509584862735e-06,
      "loss": 0.232,
      "step": 2350
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4796228238231883,
      "learning_rate": 9.35887505633499e-06,
      "loss": 0.2516,
      "step": 2351
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.176693194648145,
      "learning_rate": 9.358240235182754e-06,
      "loss": 0.765,
      "step": 2352
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4913809824643796,
      "learning_rate": 9.357605121448648e-06,
      "loss": 0.2191,
      "step": 2353
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4568052241267102,
      "learning_rate": 9.356969715175305e-06,
      "loss": 0.2404,
      "step": 2354
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3235636555415984,
      "learning_rate": 9.356334016405383e-06,
      "loss": 0.2379,
      "step": 2355
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3861266394940404,
      "learning_rate": 9.355698025181561e-06,
      "loss": 0.2152,
      "step": 2356
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4010636261726943,
      "learning_rate": 9.355061741546533e-06,
      "loss": 0.2076,
      "step": 2357
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.615820801241209,
      "learning_rate": 9.354425165543018e-06,
      "loss": 0.2133,
      "step": 2358
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4619290477397322,
      "learning_rate": 9.353788297213743e-06,
      "loss": 0.2533,
      "step": 2359
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5089855513715553,
      "learning_rate": 9.35315113660147e-06,
      "loss": 0.1853,
      "step": 2360
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4269663744798267,
      "learning_rate": 9.352513683748974e-06,
      "loss": 0.2541,
      "step": 2361
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4722174695678938,
      "learning_rate": 9.351875938699045e-06,
      "loss": 0.1752,
      "step": 2362
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.518522418805461,
      "learning_rate": 9.351237901494498e-06,
      "loss": 0.2423,
      "step": 2363
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5362570190042253,
      "learning_rate": 9.35059957217817e-06,
      "loss": 0.2129,
      "step": 2364
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.605574892206847,
      "learning_rate": 9.349960950792907e-06,
      "loss": 0.8395,
      "step": 2365
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.543283674797875,
      "learning_rate": 9.349322037381587e-06,
      "loss": 0.2219,
      "step": 2366
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3981806754507369,
      "learning_rate": 9.348682831987101e-06,
      "loss": 0.25,
      "step": 2367
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.909370059832323,
      "learning_rate": 9.348043334652362e-06,
      "loss": 0.7322,
      "step": 2368
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4107707031599575,
      "learning_rate": 9.347403545420298e-06,
      "loss": 0.2354,
      "step": 2369
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3231843110150259,
      "learning_rate": 9.346763464333862e-06,
      "loss": 0.1833,
      "step": 2370
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4673679706484994,
      "learning_rate": 9.346123091436024e-06,
      "loss": 0.2214,
      "step": 2371
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.538997517146819,
      "learning_rate": 9.345482426769774e-06,
      "loss": 0.6032,
      "step": 2372
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.265264994960151,
      "learning_rate": 9.344841470378125e-06,
      "loss": 0.1959,
      "step": 2373
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.642179043237784,
      "learning_rate": 9.344200222304103e-06,
      "loss": 0.6994,
      "step": 2374
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.462389335451799,
      "learning_rate": 9.343558682590757e-06,
      "loss": 0.7546,
      "step": 2375
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.406716242992927,
      "learning_rate": 9.342916851281155e-06,
      "loss": 0.2443,
      "step": 2376
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6166901826026026,
      "learning_rate": 9.342274728418388e-06,
      "loss": 0.2683,
      "step": 2377
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4185659848446213,
      "learning_rate": 9.34163231404556e-06,
      "loss": 0.2835,
      "step": 2378
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5418184941791382,
      "learning_rate": 9.340989608205803e-06,
      "loss": 0.2707,
      "step": 2379
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.370730856207796,
      "learning_rate": 9.34034661094226e-06,
      "loss": 0.2198,
      "step": 2380
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5013327735301913,
      "learning_rate": 9.339703322298098e-06,
      "loss": 0.2614,
      "step": 2381
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.693488511571179,
      "learning_rate": 9.339059742316501e-06,
      "loss": 0.1875,
      "step": 2382
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2448175965719483,
      "learning_rate": 9.33841587104068e-06,
      "loss": 0.17,
      "step": 2383
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.814135101473496,
      "learning_rate": 9.337771708513854e-06,
      "loss": 0.579,
      "step": 2384
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.395920181242542,
      "learning_rate": 9.337127254779272e-06,
      "loss": 0.2448,
      "step": 2385
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4566858342891218,
      "learning_rate": 9.336482509880195e-06,
      "loss": 0.2559,
      "step": 2386
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4415964968113857,
      "learning_rate": 9.33583747385991e-06,
      "loss": 0.2558,
      "step": 2387
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2779400455329433,
      "learning_rate": 9.335192146761717e-06,
      "loss": 0.2052,
      "step": 2388
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5236088643087702,
      "learning_rate": 9.334546528628942e-06,
      "loss": 0.2316,
      "step": 2389
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4265296924486541,
      "learning_rate": 9.333900619504923e-06,
      "loss": 0.2248,
      "step": 2390
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3575410438730116,
      "learning_rate": 9.333254419433026e-06,
      "loss": 0.2378,
      "step": 2391
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.392596317042315,
      "learning_rate": 9.332607928456629e-06,
      "loss": 0.5604,
      "step": 2392
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3420621854942743,
      "learning_rate": 9.331961146619135e-06,
      "loss": 0.2204,
      "step": 2393
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5032934531910074,
      "learning_rate": 9.331314073963964e-06,
      "loss": 0.2629,
      "step": 2394
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.400943558165862,
      "learning_rate": 9.330666710534556e-06,
      "loss": 0.1968,
      "step": 2395
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3863295774574946,
      "learning_rate": 9.33001905637437e-06,
      "loss": 0.2261,
      "step": 2396
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.379746568261968,
      "learning_rate": 9.329371111526887e-06,
      "loss": 0.2214,
      "step": 2397
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.478170615036254,
      "learning_rate": 9.3287228760356e-06,
      "loss": 0.4555,
      "step": 2398
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3270583989416682,
      "learning_rate": 9.328074349944034e-06,
      "loss": 0.1948,
      "step": 2399
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.583722019108332,
      "learning_rate": 9.327425533295725e-06,
      "loss": 0.2086,
      "step": 2400
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5418714702927594,
      "learning_rate": 9.326776426134223e-06,
      "loss": 0.2411,
      "step": 2401
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5368044413403081,
      "learning_rate": 9.326127028503114e-06,
      "loss": 0.2847,
      "step": 2402
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2989173478732807,
      "learning_rate": 9.325477340445989e-06,
      "loss": 0.1856,
      "step": 2403
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4757640132810568,
      "learning_rate": 9.324827362006464e-06,
      "loss": 0.2596,
      "step": 2404
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.411318082042781,
      "learning_rate": 9.324177093228175e-06,
      "loss": 0.2246,
      "step": 2405
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.251331714601607,
      "learning_rate": 9.323526534154775e-06,
      "loss": 0.1533,
      "step": 2406
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2692964951694725,
      "learning_rate": 9.32287568482994e-06,
      "loss": 0.163,
      "step": 2407
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3492730199877705,
      "learning_rate": 9.322224545297363e-06,
      "loss": 0.2036,
      "step": 2408
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.550025497252298,
      "learning_rate": 9.321573115600755e-06,
      "loss": 0.6656,
      "step": 2409
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3919978756719757,
      "learning_rate": 9.32092139578385e-06,
      "loss": 0.2217,
      "step": 2410
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4727147054007221,
      "learning_rate": 9.3202693858904e-06,
      "loss": 0.2358,
      "step": 2411
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5640289860016316,
      "learning_rate": 9.319617085964177e-06,
      "loss": 0.2474,
      "step": 2412
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2609489212262928,
      "learning_rate": 9.31896449604897e-06,
      "loss": 0.1669,
      "step": 2413
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3274323230598555,
      "learning_rate": 9.318311616188592e-06,
      "loss": 0.2419,
      "step": 2414
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6269479979697206,
      "learning_rate": 9.31765844642687e-06,
      "loss": 0.231,
      "step": 2415
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.434332059359609,
      "learning_rate": 9.317004986807656e-06,
      "loss": 0.2378,
      "step": 2416
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4898402002445745,
      "learning_rate": 9.316351237374816e-06,
      "loss": 0.2019,
      "step": 2417
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.340695080988302,
      "learning_rate": 9.315697198172238e-06,
      "loss": 0.2032,
      "step": 2418
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3219477445287229,
      "learning_rate": 9.315042869243833e-06,
      "loss": 0.2007,
      "step": 2419
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4046389398510306,
      "learning_rate": 9.314388250633526e-06,
      "loss": 0.2562,
      "step": 2420
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.021375975183638,
      "learning_rate": 9.313733342385263e-06,
      "loss": 0.5803,
      "step": 2421
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.897434045292936,
      "learning_rate": 9.313078144543012e-06,
      "loss": 0.6076,
      "step": 2422
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4713017046220924,
      "learning_rate": 9.312422657150755e-06,
      "loss": 0.2091,
      "step": 2423
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4982149910403204,
      "learning_rate": 9.3117668802525e-06,
      "loss": 0.236,
      "step": 2424
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4617716221070527,
      "learning_rate": 9.31111081389227e-06,
      "loss": 0.2523,
      "step": 2425
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.298663219149016,
      "learning_rate": 9.31045445811411e-06,
      "loss": 0.1994,
      "step": 2426
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4004326854239149,
      "learning_rate": 9.30979781296208e-06,
      "loss": 0.2307,
      "step": 2427
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3994074400141987,
      "learning_rate": 9.309140878480267e-06,
      "loss": 0.2443,
      "step": 2428
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.381963114550332,
      "learning_rate": 9.308483654712769e-06,
      "loss": 0.2398,
      "step": 2429
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3301105150205204,
      "learning_rate": 9.30782614170371e-06,
      "loss": 0.2047,
      "step": 2430
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6639510373136035,
      "learning_rate": 9.30716833949723e-06,
      "loss": 0.2514,
      "step": 2431
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4682586644623121,
      "learning_rate": 9.306510248137488e-06,
      "loss": 0.2118,
      "step": 2432
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4232291572592344,
      "learning_rate": 9.305851867668665e-06,
      "loss": 0.2084,
      "step": 2433
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.379806377695181,
      "learning_rate": 9.305193198134962e-06,
      "loss": 0.2139,
      "step": 2434
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5516557437166802,
      "learning_rate": 9.304534239580591e-06,
      "loss": 0.2596,
      "step": 2435
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.983878102860816,
      "learning_rate": 9.303874992049797e-06,
      "loss": 0.5263,
      "step": 2436
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5214073112722155,
      "learning_rate": 9.303215455586835e-06,
      "loss": 0.2192,
      "step": 2437
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.280652021015761,
      "learning_rate": 9.30255563023598e-06,
      "loss": 0.7436,
      "step": 2438
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.044753879829271,
      "learning_rate": 9.301895516041531e-06,
      "loss": 0.5821,
      "step": 2439
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6366410619375933,
      "learning_rate": 9.301235113047801e-06,
      "loss": 0.2705,
      "step": 2440
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4917982027921104,
      "learning_rate": 9.300574421299127e-06,
      "loss": 0.2394,
      "step": 2441
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3845547234589197,
      "learning_rate": 9.299913440839859e-06,
      "loss": 0.2604,
      "step": 2442
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4219534833849874,
      "learning_rate": 9.299252171714374e-06,
      "loss": 0.2287,
      "step": 2443
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.349023961813594,
      "learning_rate": 9.298590613967067e-06,
      "loss": 0.194,
      "step": 2444
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5584438281215978,
      "learning_rate": 9.297928767642346e-06,
      "loss": 0.2296,
      "step": 2445
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.443518436770763,
      "learning_rate": 9.297266632784646e-06,
      "loss": 0.1946,
      "step": 2446
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4453125077668862,
      "learning_rate": 9.296604209438414e-06,
      "loss": 0.2371,
      "step": 2447
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.481652828024523,
      "learning_rate": 9.295941497648125e-06,
      "loss": 0.2513,
      "step": 2448
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3661590584993701,
      "learning_rate": 9.295278497458266e-06,
      "loss": 0.2215,
      "step": 2449
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4871967471239442,
      "learning_rate": 9.29461520891335e-06,
      "loss": 0.2394,
      "step": 2450
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3775037658924878,
      "learning_rate": 9.2939516320579e-06,
      "loss": 0.2006,
      "step": 2451
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5973772895003122,
      "learning_rate": 9.293287766936469e-06,
      "loss": 0.2695,
      "step": 2452
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.26604038601165,
      "learning_rate": 9.29262361359362e-06,
      "loss": 0.8148,
      "step": 2453
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5326213212465247,
      "learning_rate": 9.291959172073943e-06,
      "loss": 0.2379,
      "step": 2454
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5089898058918714,
      "learning_rate": 9.291294442422043e-06,
      "loss": 0.2548,
      "step": 2455
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3671669653549101,
      "learning_rate": 9.290629424682543e-06,
      "loss": 0.1762,
      "step": 2456
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.5413106007252715,
      "learning_rate": 9.289964118900092e-06,
      "loss": 0.5311,
      "step": 2457
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2930332947379293,
      "learning_rate": 9.28929852511935e-06,
      "loss": 0.2378,
      "step": 2458
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5489373036725944,
      "learning_rate": 9.288632643385002e-06,
      "loss": 0.2963,
      "step": 2459
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5187989938325832,
      "learning_rate": 9.287966473741752e-06,
      "loss": 0.2195,
      "step": 2460
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4999921469494621,
      "learning_rate": 9.28730001623432e-06,
      "loss": 0.2291,
      "step": 2461
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4293275251756687,
      "learning_rate": 9.286633270907448e-06,
      "loss": 0.261,
      "step": 2462
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4513720467703264,
      "learning_rate": 9.285966237805895e-06,
      "loss": 0.2387,
      "step": 2463
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.5353417128643265,
      "learning_rate": 9.285298916974443e-06,
      "loss": 0.5325,
      "step": 2464
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.386332466672197,
      "learning_rate": 9.284631308457892e-06,
      "loss": 0.2425,
      "step": 2465
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.296393099229109,
      "learning_rate": 9.283963412301058e-06,
      "loss": 0.2506,
      "step": 2466
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5184022975963396,
      "learning_rate": 9.28329522854878e-06,
      "loss": 0.2394,
      "step": 2467
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4006631701857264,
      "learning_rate": 9.282626757245918e-06,
      "loss": 0.2258,
      "step": 2468
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5533267007148088,
      "learning_rate": 9.281957998437345e-06,
      "loss": 0.2375,
      "step": 2469
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4336727371214546,
      "learning_rate": 9.281288952167957e-06,
      "loss": 0.2341,
      "step": 2470
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4939816028297657,
      "learning_rate": 9.28061961848267e-06,
      "loss": 0.2024,
      "step": 2471
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2816606239333097,
      "learning_rate": 9.279949997426417e-06,
      "loss": 0.1958,
      "step": 2472
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4047626002456695,
      "learning_rate": 9.279280089044155e-06,
      "loss": 0.2161,
      "step": 2473
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4218538824100064,
      "learning_rate": 9.278609893380855e-06,
      "loss": 0.2159,
      "step": 2474
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4709704124435377,
      "learning_rate": 9.277939410481507e-06,
      "loss": 0.2203,
      "step": 2475
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6435936336197325,
      "learning_rate": 9.277268640391126e-06,
      "loss": 0.2541,
      "step": 2476
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4611238593241196,
      "learning_rate": 9.27659758315474e-06,
      "loss": 0.1793,
      "step": 2477
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.168397303915047,
      "learning_rate": 9.275926238817404e-06,
      "loss": 0.5371,
      "step": 2478
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3939953515914356,
      "learning_rate": 9.275254607424182e-06,
      "loss": 0.2667,
      "step": 2479
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5090912411164825,
      "learning_rate": 9.274582689020164e-06,
      "loss": 0.2199,
      "step": 2480
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.470917441778964,
      "learning_rate": 9.273910483650461e-06,
      "loss": 0.2479,
      "step": 2481
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.60024886734015,
      "learning_rate": 9.273237991360196e-06,
      "loss": 0.5251,
      "step": 2482
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5415147891786058,
      "learning_rate": 9.272565212194517e-06,
      "loss": 0.2728,
      "step": 2483
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2115858625178058,
      "learning_rate": 9.271892146198591e-06,
      "loss": 0.1948,
      "step": 2484
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5446657033659177,
      "learning_rate": 9.271218793417601e-06,
      "loss": 0.2336,
      "step": 2485
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2484030724617678,
      "learning_rate": 9.270545153896752e-06,
      "loss": 0.1666,
      "step": 2486
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3862724651901703,
      "learning_rate": 9.269871227681268e-06,
      "loss": 0.1854,
      "step": 2487
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4581185061074284,
      "learning_rate": 9.269197014816393e-06,
      "loss": 0.2508,
      "step": 2488
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.435474486172341,
      "learning_rate": 9.268522515347384e-06,
      "loss": 0.2383,
      "step": 2489
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5731549428676292,
      "learning_rate": 9.267847729319528e-06,
      "loss": 0.2536,
      "step": 2490
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5507697834875012,
      "learning_rate": 9.267172656778122e-06,
      "loss": 0.2129,
      "step": 2491
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4830658508463754,
      "learning_rate": 9.266497297768487e-06,
      "loss": 0.2002,
      "step": 2492
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5309953148650814,
      "learning_rate": 9.265821652335961e-06,
      "loss": 0.2475,
      "step": 2493
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.295659132966027,
      "learning_rate": 9.265145720525902e-06,
      "loss": 0.1987,
      "step": 2494
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3116497712159687,
      "learning_rate": 9.264469502383689e-06,
      "loss": 0.2062,
      "step": 2495
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4758729449247856,
      "learning_rate": 9.263792997954717e-06,
      "loss": 0.2136,
      "step": 2496
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6559882876214413,
      "learning_rate": 9.263116207284402e-06,
      "loss": 0.2954,
      "step": 2497
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.583062641804702,
      "learning_rate": 9.26243913041818e-06,
      "loss": 0.2388,
      "step": 2498
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4827167848800258,
      "learning_rate": 9.261761767401503e-06,
      "loss": 0.2595,
      "step": 2499
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4492652267253305,
      "learning_rate": 9.261084118279846e-06,
      "loss": 0.2302,
      "step": 2500
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4481010435432558,
      "learning_rate": 9.260406183098704e-06,
      "loss": 0.2253,
      "step": 2501
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4762427947961392,
      "learning_rate": 9.259727961903584e-06,
      "loss": 0.2946,
      "step": 2502
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4258686627160804,
      "learning_rate": 9.25904945474002e-06,
      "loss": 0.2424,
      "step": 2503
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4322695557753122,
      "learning_rate": 9.258370661653563e-06,
      "loss": 0.2496,
      "step": 2504
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3557002571171304,
      "learning_rate": 9.25769158268978e-06,
      "loss": 0.2079,
      "step": 2505
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.503981765288857,
      "learning_rate": 9.257012217894261e-06,
      "loss": 0.2397,
      "step": 2506
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.804371997039915,
      "learning_rate": 9.256332567312614e-06,
      "loss": 0.7001,
      "step": 2507
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3999608098437997,
      "learning_rate": 9.255652630990464e-06,
      "loss": 0.2017,
      "step": 2508
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.567890353838836,
      "learning_rate": 9.25497240897346e-06,
      "loss": 0.2553,
      "step": 2509
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.398474103937861,
      "learning_rate": 9.254291901307267e-06,
      "loss": 0.4961,
      "step": 2510
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4145036263527753,
      "learning_rate": 9.253611108037566e-06,
      "loss": 0.2337,
      "step": 2511
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.482965372595709,
      "learning_rate": 9.252930029210066e-06,
      "loss": 0.6682,
      "step": 2512
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4316685741927988,
      "learning_rate": 9.252248664870486e-06,
      "loss": 0.2423,
      "step": 2513
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5229130817369538,
      "learning_rate": 9.25156701506457e-06,
      "loss": 0.2223,
      "step": 2514
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.520685788776464,
      "learning_rate": 9.250885079838079e-06,
      "loss": 0.2363,
      "step": 2515
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.442688848302474,
      "learning_rate": 9.250202859236792e-06,
      "loss": 0.1794,
      "step": 2516
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.465943062395698,
      "learning_rate": 9.249520353306509e-06,
      "loss": 0.225,
      "step": 2517
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6340684080426102,
      "learning_rate": 9.248837562093049e-06,
      "loss": 0.2246,
      "step": 2518
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5159266674273277,
      "learning_rate": 9.24815448564225e-06,
      "loss": 0.2088,
      "step": 2519
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4442845542231468,
      "learning_rate": 9.24747112399997e-06,
      "loss": 0.2357,
      "step": 2520
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4838230499467093,
      "learning_rate": 9.246787477212085e-06,
      "loss": 0.2464,
      "step": 2521
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.652805673648436,
      "learning_rate": 9.246103545324488e-06,
      "loss": 0.6228,
      "step": 2522
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.879983798196003,
      "learning_rate": 9.245419328383095e-06,
      "loss": 0.5753,
      "step": 2523
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3775244886299194,
      "learning_rate": 9.244734826433839e-06,
      "loss": 0.1972,
      "step": 2524
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.381158484000899,
      "learning_rate": 9.244050039522673e-06,
      "loss": 0.2453,
      "step": 2525
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3231891360993122,
      "learning_rate": 9.243364967695569e-06,
      "loss": 0.2025,
      "step": 2526
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.68637547598664,
      "learning_rate": 9.242679610998519e-06,
      "loss": 0.4897,
      "step": 2527
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.413382560534626,
      "learning_rate": 9.241993969477531e-06,
      "loss": 0.2552,
      "step": 2528
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3935434394329702,
      "learning_rate": 9.241308043178635e-06,
      "loss": 0.2256,
      "step": 2529
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5074881432870122,
      "learning_rate": 9.24062183214788e-06,
      "loss": 0.2063,
      "step": 2530
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4281207441465014,
      "learning_rate": 9.239935336431331e-06,
      "loss": 0.1833,
      "step": 2531
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.451110614104323,
      "learning_rate": 9.23924855607508e-06,
      "loss": 0.683,
      "step": 2532
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3517574261227974,
      "learning_rate": 9.238561491125225e-06,
      "loss": 0.1941,
      "step": 2533
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5224681150369972,
      "learning_rate": 9.237874141627896e-06,
      "loss": 0.2488,
      "step": 2534
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3467341771054333,
      "learning_rate": 9.237186507629236e-06,
      "loss": 0.2296,
      "step": 2535
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.905310591288163,
      "learning_rate": 9.236498589175408e-06,
      "loss": 0.4675,
      "step": 2536
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4362406417222504,
      "learning_rate": 9.235810386312594e-06,
      "loss": 0.2257,
      "step": 2537
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3801983119520267,
      "learning_rate": 9.235121899086994e-06,
      "loss": 0.1867,
      "step": 2538
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4807110335585845,
      "learning_rate": 9.23443312754483e-06,
      "loss": 0.1957,
      "step": 2539
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4066779495411443,
      "learning_rate": 9.23374407173234e-06,
      "loss": 0.2146,
      "step": 2540
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.5091301509973105,
      "learning_rate": 9.233054731695782e-06,
      "loss": 0.6844,
      "step": 2541
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4348934563408595,
      "learning_rate": 9.232365107481433e-06,
      "loss": 0.2199,
      "step": 2542
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.433580207343984,
      "learning_rate": 9.231675199135593e-06,
      "loss": 0.2568,
      "step": 2543
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3547260450710197,
      "learning_rate": 9.230985006704575e-06,
      "loss": 0.2678,
      "step": 2544
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4039145897607916,
      "learning_rate": 9.230294530234714e-06,
      "loss": 0.2108,
      "step": 2545
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2697657102508415,
      "learning_rate": 9.229603769772364e-06,
      "loss": 0.1939,
      "step": 2546
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3245371471331209,
      "learning_rate": 9.228912725363897e-06,
      "loss": 0.1914,
      "step": 2547
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3372273214265753,
      "learning_rate": 9.228221397055705e-06,
      "loss": 0.2196,
      "step": 2548
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3373591056905139,
      "learning_rate": 9.2275297848942e-06,
      "loss": 0.26,
      "step": 2549
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.393030613664345,
      "learning_rate": 9.226837888925813e-06,
      "loss": 0.1993,
      "step": 2550
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5892414372315793,
      "learning_rate": 9.226145709196991e-06,
      "loss": 0.2455,
      "step": 2551
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.636981981972357,
      "learning_rate": 9.225453245754204e-06,
      "loss": 0.1862,
      "step": 2552
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3937873285388753,
      "learning_rate": 9.224760498643936e-06,
      "loss": 0.2497,
      "step": 2553
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2882393114518456,
      "learning_rate": 9.224067467912696e-06,
      "loss": 0.1903,
      "step": 2554
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3443596230487904,
      "learning_rate": 9.22337415360701e-06,
      "loss": 0.2054,
      "step": 2555
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3289269251060674,
      "learning_rate": 9.222680555773417e-06,
      "loss": 0.153,
      "step": 2556
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.484344652326302,
      "learning_rate": 9.221986674458488e-06,
      "loss": 0.2689,
      "step": 2557
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.334389007614257,
      "learning_rate": 9.221292509708799e-06,
      "loss": 0.6401,
      "step": 2558
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6878209637215262,
      "learning_rate": 9.220598061570956e-06,
      "loss": 0.2422,
      "step": 2559
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5287373601179137,
      "learning_rate": 9.219903330091575e-06,
      "loss": 0.213,
      "step": 2560
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3116513394070635,
      "learning_rate": 9.219208315317298e-06,
      "loss": 0.2148,
      "step": 2561
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.378407896851258,
      "learning_rate": 9.218513017294784e-06,
      "loss": 0.192,
      "step": 2562
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3877394896723168,
      "learning_rate": 9.217817436070708e-06,
      "loss": 0.2003,
      "step": 2563
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4921635436978344,
      "learning_rate": 9.21712157169177e-06,
      "loss": 0.2569,
      "step": 2564
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4276512213316017,
      "learning_rate": 9.216425424204683e-06,
      "loss": 0.2309,
      "step": 2565
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.413728043637103,
      "learning_rate": 9.21572899365618e-06,
      "loss": 0.2317,
      "step": 2566
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4703808340792581,
      "learning_rate": 9.21503228009302e-06,
      "loss": 0.2278,
      "step": 2567
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.465010652761568,
      "learning_rate": 9.21433528356197e-06,
      "loss": 0.5517,
      "step": 2568
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2985605525229909,
      "learning_rate": 9.213638004109824e-06,
      "loss": 0.2071,
      "step": 2569
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3773735148111539,
      "learning_rate": 9.212940441783392e-06,
      "loss": 0.1815,
      "step": 2570
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4510572683152825,
      "learning_rate": 9.212242596629504e-06,
      "loss": 0.2322,
      "step": 2571
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4896401749045909,
      "learning_rate": 9.211544468695006e-06,
      "loss": 0.2305,
      "step": 2572
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3620070874134844,
      "learning_rate": 9.210846058026767e-06,
      "loss": 0.1904,
      "step": 2573
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5302365974957681,
      "learning_rate": 9.210147364671677e-06,
      "loss": 0.1975,
      "step": 2574
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6015145485481903,
      "learning_rate": 9.209448388676636e-06,
      "loss": 0.2202,
      "step": 2575
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6048312340272386,
      "learning_rate": 9.20874913008857e-06,
      "loss": 0.2767,
      "step": 2576
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3804191550252483,
      "learning_rate": 9.208049588954424e-06,
      "loss": 0.2175,
      "step": 2577
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2798978137197412,
      "learning_rate": 9.20734976532116e-06,
      "loss": 0.1978,
      "step": 2578
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.452676177917455,
      "learning_rate": 9.206649659235756e-06,
      "loss": 0.2097,
      "step": 2579
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.340129562355555,
      "learning_rate": 9.205949270745217e-06,
      "loss": 0.1935,
      "step": 2580
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4160690244536491,
      "learning_rate": 9.20524859989656e-06,
      "loss": 0.2046,
      "step": 2581
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3459733460501493,
      "learning_rate": 9.204547646736821e-06,
      "loss": 0.2059,
      "step": 2582
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3240502547597786,
      "learning_rate": 9.20384641131306e-06,
      "loss": 0.203,
      "step": 2583
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2869660305761765,
      "learning_rate": 9.203144893672354e-06,
      "loss": 0.2396,
      "step": 2584
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4802086491720519,
      "learning_rate": 9.202443093861796e-06,
      "loss": 0.2241,
      "step": 2585
    },
    {
      "epoch": 0.21,
      "grad_norm": 17.120619615915853,
      "learning_rate": 9.201741011928498e-06,
      "loss": 0.5523,
      "step": 2586
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3476667881077649,
      "learning_rate": 9.201038647919595e-06,
      "loss": 0.2597,
      "step": 2587
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.367052989921326,
      "learning_rate": 9.20033600188224e-06,
      "loss": 0.8873,
      "step": 2588
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3255516252692998,
      "learning_rate": 9.199633073863603e-06,
      "loss": 0.2093,
      "step": 2589
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3447462584321312,
      "learning_rate": 9.198929863910874e-06,
      "loss": 0.1762,
      "step": 2590
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2864388853799342,
      "learning_rate": 9.198226372071259e-06,
      "loss": 0.1788,
      "step": 2591
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6518110873473089,
      "learning_rate": 9.19752259839199e-06,
      "loss": 0.2497,
      "step": 2592
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4182444108706607,
      "learning_rate": 9.19681854292031e-06,
      "loss": 0.1983,
      "step": 2593
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.434158326272205,
      "learning_rate": 9.196114205703484e-06,
      "loss": 0.2311,
      "step": 2594
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4521995767763962,
      "learning_rate": 9.1954095867888e-06,
      "loss": 0.1879,
      "step": 2595
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.552286919863675,
      "learning_rate": 9.194704686223557e-06,
      "loss": 0.2356,
      "step": 2596
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.286665369890514,
      "learning_rate": 9.19399950405508e-06,
      "loss": 0.1801,
      "step": 2597
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3806476549559905,
      "learning_rate": 9.193294040330709e-06,
      "loss": 0.1546,
      "step": 2598
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7417330117177041,
      "learning_rate": 9.192588295097801e-06,
      "loss": 0.2417,
      "step": 2599
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3854065599438838,
      "learning_rate": 9.191882268403743e-06,
      "loss": 0.2121,
      "step": 2600
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.484645674153059,
      "learning_rate": 9.191175960295924e-06,
      "loss": 0.2646,
      "step": 2601
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.594624200499128,
      "learning_rate": 9.190469370821764e-06,
      "loss": 0.2721,
      "step": 2602
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4728863401872578,
      "learning_rate": 9.189762500028698e-06,
      "loss": 0.1982,
      "step": 2603
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0971855512730813,
      "learning_rate": 9.189055347964182e-06,
      "loss": 0.1681,
      "step": 2604
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.082576177722389,
      "learning_rate": 9.188347914675689e-06,
      "loss": 0.6596,
      "step": 2605
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3919091198065385,
      "learning_rate": 9.18764020021071e-06,
      "loss": 0.276,
      "step": 2606
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1776139775140204,
      "learning_rate": 9.186932204616756e-06,
      "loss": 0.1789,
      "step": 2607
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.439106220327655,
      "learning_rate": 9.186223927941356e-06,
      "loss": 0.2201,
      "step": 2608
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.39652171188706,
      "learning_rate": 9.185515370232062e-06,
      "loss": 0.1916,
      "step": 2609
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5215548583810834,
      "learning_rate": 9.184806531536438e-06,
      "loss": 0.2205,
      "step": 2610
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7014454615516577,
      "learning_rate": 9.184097411902072e-06,
      "loss": 0.2583,
      "step": 2611
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3717443332297468,
      "learning_rate": 9.18338801137657e-06,
      "loss": 0.1937,
      "step": 2612
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2721018329044964,
      "learning_rate": 9.182678330007556e-06,
      "loss": 0.2183,
      "step": 2613
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3835540132382411,
      "learning_rate": 9.181968367842674e-06,
      "loss": 0.2103,
      "step": 2614
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.511757867687406,
      "learning_rate": 9.181258124929582e-06,
      "loss": 0.2402,
      "step": 2615
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.462402674354687,
      "learning_rate": 9.180547601315963e-06,
      "loss": 0.6583,
      "step": 2616
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4358484770462532,
      "learning_rate": 9.17983679704952e-06,
      "loss": 0.2219,
      "step": 2617
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4710993390172478,
      "learning_rate": 9.179125712177965e-06,
      "loss": 0.1975,
      "step": 2618
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.6978926691234,
      "learning_rate": 9.17841434674904e-06,
      "loss": 0.4826,
      "step": 2619
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6587083948761059,
      "learning_rate": 9.1777027008105e-06,
      "loss": 0.2437,
      "step": 2620
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6282070275551552,
      "learning_rate": 9.17699077441012e-06,
      "loss": 0.2709,
      "step": 2621
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4686786274886592,
      "learning_rate": 9.176278567595696e-06,
      "loss": 0.2582,
      "step": 2622
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4275957389410845,
      "learning_rate": 9.175566080415036e-06,
      "loss": 0.2079,
      "step": 2623
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.175196419461104,
      "learning_rate": 9.174853312915972e-06,
      "loss": 0.7449,
      "step": 2624
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.459219148183228,
      "learning_rate": 9.174140265146356e-06,
      "loss": 0.2057,
      "step": 2625
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4367320719948615,
      "learning_rate": 9.173426937154058e-06,
      "loss": 0.2265,
      "step": 2626
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4388902604254572,
      "learning_rate": 9.172713328986965e-06,
      "loss": 0.2387,
      "step": 2627
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5414812851652706,
      "learning_rate": 9.171999440692982e-06,
      "loss": 0.2482,
      "step": 2628
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3528337947362474,
      "learning_rate": 9.17128527232004e-06,
      "loss": 0.1658,
      "step": 2629
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.179402885857567,
      "learning_rate": 9.170570823916074e-06,
      "loss": 0.1978,
      "step": 2630
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.481990131493022,
      "learning_rate": 9.169856095529055e-06,
      "loss": 0.2221,
      "step": 2631
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4284265168141148,
      "learning_rate": 9.16914108720696e-06,
      "loss": 0.2059,
      "step": 2632
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5474644091835987,
      "learning_rate": 9.168425798997794e-06,
      "loss": 0.2911,
      "step": 2633
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3544932070032691,
      "learning_rate": 9.167710230949573e-06,
      "loss": 0.2489,
      "step": 2634
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4038325987323266,
      "learning_rate": 9.166994383110338e-06,
      "loss": 0.2426,
      "step": 2635
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4927236396525574,
      "learning_rate": 9.166278255528143e-06,
      "loss": 0.2086,
      "step": 2636
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4786421653615467,
      "learning_rate": 9.165561848251066e-06,
      "loss": 0.257,
      "step": 2637
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.494964907518733,
      "learning_rate": 9.164845161327203e-06,
      "loss": 0.2439,
      "step": 2638
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4914512590800875,
      "learning_rate": 9.164128194804663e-06,
      "loss": 0.2487,
      "step": 2639
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.213346824093523,
      "learning_rate": 9.16341094873158e-06,
      "loss": 0.4918,
      "step": 2640
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.8976864828360265,
      "learning_rate": 9.162693423156106e-06,
      "loss": 0.7302,
      "step": 2641
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.293228759486824,
      "learning_rate": 9.161975618126411e-06,
      "loss": 0.1933,
      "step": 2642
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5208125629577405,
      "learning_rate": 9.161257533690682e-06,
      "loss": 0.2141,
      "step": 2643
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.058882638648034,
      "learning_rate": 9.160539169897126e-06,
      "loss": 0.777,
      "step": 2644
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.7070982969173425,
      "learning_rate": 9.15982052679397e-06,
      "loss": 0.7961,
      "step": 2645
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3151563738096888,
      "learning_rate": 9.15910160442946e-06,
      "loss": 0.2186,
      "step": 2646
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4053964091377618,
      "learning_rate": 9.158382402851854e-06,
      "loss": 0.2008,
      "step": 2647
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2994142241834374,
      "learning_rate": 9.15766292210944e-06,
      "loss": 0.1868,
      "step": 2648
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.268345972289518,
      "learning_rate": 9.156943162250516e-06,
      "loss": 0.1615,
      "step": 2649
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5665268742002818,
      "learning_rate": 9.156223123323405e-06,
      "loss": 0.2451,
      "step": 2650
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4240915616702214,
      "learning_rate": 9.155502805376439e-06,
      "loss": 0.2301,
      "step": 2651
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.917390607353714,
      "learning_rate": 9.154782208457981e-06,
      "loss": 0.4922,
      "step": 2652
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3034028824088442,
      "learning_rate": 9.154061332616407e-06,
      "loss": 0.2254,
      "step": 2653
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3376834824067334,
      "learning_rate": 9.153340177900108e-06,
      "loss": 0.2231,
      "step": 2654
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5048346088825546,
      "learning_rate": 9.152618744357498e-06,
      "loss": 0.212,
      "step": 2655
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3953346603606998,
      "learning_rate": 9.15189703203701e-06,
      "loss": 0.2528,
      "step": 2656
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5183701229182094,
      "learning_rate": 9.151175040987094e-06,
      "loss": 0.2235,
      "step": 2657
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3060666326298913,
      "learning_rate": 9.15045277125622e-06,
      "loss": 0.1975,
      "step": 2658
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.246969409520784,
      "learning_rate": 9.149730222892876e-06,
      "loss": 0.668,
      "step": 2659
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3545291327540572,
      "learning_rate": 9.149007395945569e-06,
      "loss": 0.1894,
      "step": 2660
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.322329299306653,
      "learning_rate": 9.148284290462825e-06,
      "loss": 0.1973,
      "step": 2661
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4586395960497665,
      "learning_rate": 9.147560906493189e-06,
      "loss": 0.2445,
      "step": 2662
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3192365454784591,
      "learning_rate": 9.14683724408522e-06,
      "loss": 0.194,
      "step": 2663
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.412074552048598,
      "learning_rate": 9.146113303287503e-06,
      "loss": 0.2587,
      "step": 2664
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.272525701462181,
      "learning_rate": 9.14538908414864e-06,
      "loss": 0.1981,
      "step": 2665
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.390918664885163,
      "learning_rate": 9.144664586717246e-06,
      "loss": 0.2275,
      "step": 2666
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.878339899546916,
      "learning_rate": 9.14393981104196e-06,
      "loss": 0.6393,
      "step": 2667
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3502878561911034,
      "learning_rate": 9.14321475717144e-06,
      "loss": 0.2382,
      "step": 2668
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.301719968096471,
      "learning_rate": 9.14248942515436e-06,
      "loss": 0.1852,
      "step": 2669
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.743392928059241,
      "learning_rate": 9.141763815039413e-06,
      "loss": 0.705,
      "step": 2670
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4748406355115191,
      "learning_rate": 9.141037926875312e-06,
      "loss": 0.1962,
      "step": 2671
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.565445256685354,
      "learning_rate": 9.140311760710788e-06,
      "loss": 0.7296,
      "step": 2672
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3193624040827834,
      "learning_rate": 9.139585316594592e-06,
      "loss": 0.1926,
      "step": 2673
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5039408737590663,
      "learning_rate": 9.13885859457549e-06,
      "loss": 0.216,
      "step": 2674
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.101394223123094,
      "learning_rate": 9.13813159470227e-06,
      "loss": 0.6383,
      "step": 2675
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5459468309760096,
      "learning_rate": 9.137404317023738e-06,
      "loss": 0.2287,
      "step": 2676
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3902110268349865,
      "learning_rate": 9.13667676158872e-06,
      "loss": 0.2405,
      "step": 2677
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.373716524210969,
      "learning_rate": 9.135948928446057e-06,
      "loss": 0.2256,
      "step": 2678
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4329366195571578,
      "learning_rate": 9.13522081764461e-06,
      "loss": 0.2288,
      "step": 2679
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5049253420446151,
      "learning_rate": 9.134492429233262e-06,
      "loss": 0.2167,
      "step": 2680
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.607352752390637,
      "learning_rate": 9.133763763260907e-06,
      "loss": 0.5486,
      "step": 2681
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.347140340316864,
      "learning_rate": 9.133034819776469e-06,
      "loss": 0.2268,
      "step": 2682
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3720230861670513,
      "learning_rate": 9.13230559882888e-06,
      "loss": 0.2061,
      "step": 2683
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3937876334363148,
      "learning_rate": 9.131576100467095e-06,
      "loss": 0.2057,
      "step": 2684
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.577816701365336,
      "learning_rate": 9.130846324740087e-06,
      "loss": 0.2294,
      "step": 2685
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5576769536011323,
      "learning_rate": 9.130116271696851e-06,
      "loss": 0.2429,
      "step": 2686
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3760401761764731,
      "learning_rate": 9.129385941386397e-06,
      "loss": 0.1705,
      "step": 2687
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.421557200319374,
      "learning_rate": 9.128655333857751e-06,
      "loss": 0.2205,
      "step": 2688
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.333311866529666,
      "learning_rate": 9.127924449159966e-06,
      "loss": 0.1973,
      "step": 2689
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.489628096390768,
      "learning_rate": 9.127193287342103e-06,
      "loss": 0.2113,
      "step": 2690
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2733000801837795,
      "learning_rate": 9.12646184845325e-06,
      "loss": 0.1972,
      "step": 2691
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.298630408900462,
      "learning_rate": 9.125730132542511e-06,
      "loss": 0.1678,
      "step": 2692
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.460921832458138,
      "learning_rate": 9.124998139659009e-06,
      "loss": 0.2351,
      "step": 2693
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5015495967938832,
      "learning_rate": 9.124265869851882e-06,
      "loss": 0.2006,
      "step": 2694
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3391767970264414,
      "learning_rate": 9.12353332317029e-06,
      "loss": 0.1874,
      "step": 2695
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5178712198316444,
      "learning_rate": 9.122800499663414e-06,
      "loss": 0.2272,
      "step": 2696
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5116629528869265,
      "learning_rate": 9.12206739938045e-06,
      "loss": 0.2093,
      "step": 2697
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3410534843112707,
      "learning_rate": 9.12133402237061e-06,
      "loss": 0.2153,
      "step": 2698
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4019882394154262,
      "learning_rate": 9.12060036868313e-06,
      "loss": 0.2519,
      "step": 2699
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2002464621642435,
      "learning_rate": 9.119866438367263e-06,
      "loss": 0.1515,
      "step": 2700
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6261783024552916,
      "learning_rate": 9.119132231472278e-06,
      "loss": 0.2194,
      "step": 2701
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3336394962402387,
      "learning_rate": 9.118397748047467e-06,
      "loss": 0.1889,
      "step": 2702
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4605709263858393,
      "learning_rate": 9.117662988142138e-06,
      "loss": 0.2525,
      "step": 2703
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6723157566839806,
      "learning_rate": 9.116927951805615e-06,
      "loss": 0.2185,
      "step": 2704
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6023138177065948,
      "learning_rate": 9.116192639087245e-06,
      "loss": 0.2313,
      "step": 2705
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3621626671413896,
      "learning_rate": 9.115457050036393e-06,
      "loss": 0.197,
      "step": 2706
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3707857273152984,
      "learning_rate": 9.11472118470244e-06,
      "loss": 0.1982,
      "step": 2707
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.441471835179643,
      "learning_rate": 9.113985043134784e-06,
      "loss": 0.7088,
      "step": 2708
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.445595614736569,
      "learning_rate": 9.113248625382849e-06,
      "loss": 0.2784,
      "step": 2709
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.51766381544587,
      "learning_rate": 9.11251193149607e-06,
      "loss": 0.2239,
      "step": 2710
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4026549068643028,
      "learning_rate": 9.111774961523906e-06,
      "loss": 0.2117,
      "step": 2711
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9154104309470954,
      "learning_rate": 9.11103771551583e-06,
      "loss": 0.2648,
      "step": 2712
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3223474030414537,
      "learning_rate": 9.110300193521336e-06,
      "loss": 0.1826,
      "step": 2713
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6005665556405795,
      "learning_rate": 9.109562395589937e-06,
      "loss": 0.1989,
      "step": 2714
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6073390073266425,
      "learning_rate": 9.108824321771163e-06,
      "loss": 0.2288,
      "step": 2715
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4154704609763002,
      "learning_rate": 9.108085972114563e-06,
      "loss": 0.2162,
      "step": 2716
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4622945902637294,
      "learning_rate": 9.107347346669705e-06,
      "loss": 0.2121,
      "step": 2717
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.898084998466211,
      "learning_rate": 9.106608445486171e-06,
      "loss": 0.4668,
      "step": 2718
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4654514280282613,
      "learning_rate": 9.105869268613574e-06,
      "loss": 0.1899,
      "step": 2719
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.6518450146102275,
      "learning_rate": 9.105129816101531e-06,
      "loss": 0.726,
      "step": 2720
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.340745068837695,
      "learning_rate": 9.104390087999686e-06,
      "loss": 0.1607,
      "step": 2721
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4525163388831515,
      "learning_rate": 9.103650084357697e-06,
      "loss": 0.2404,
      "step": 2722
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2417255609281581,
      "learning_rate": 9.102909805225246e-06,
      "loss": 0.2193,
      "step": 2723
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3662320404428347,
      "learning_rate": 9.102169250652029e-06,
      "loss": 0.2499,
      "step": 2724
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5065714633685114,
      "learning_rate": 9.101428420687759e-06,
      "loss": 0.2605,
      "step": 2725
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4756191962821648,
      "learning_rate": 9.100687315382174e-06,
      "loss": 0.2636,
      "step": 2726
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5593777191465357,
      "learning_rate": 9.099945934785026e-06,
      "loss": 0.2616,
      "step": 2727
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5186098200696208,
      "learning_rate": 9.099204278946083e-06,
      "loss": 0.2327,
      "step": 2728
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1860620081515083,
      "learning_rate": 9.098462347915136e-06,
      "loss": 0.2039,
      "step": 2729
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.346188150374557,
      "learning_rate": 9.097720141741994e-06,
      "loss": 0.2221,
      "step": 2730
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.136065294341917,
      "learning_rate": 9.096977660476485e-06,
      "loss": 0.4554,
      "step": 2731
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3584808130383115,
      "learning_rate": 9.096234904168451e-06,
      "loss": 0.2206,
      "step": 2732
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.393026357935776,
      "learning_rate": 9.095491872867757e-06,
      "loss": 0.2319,
      "step": 2733
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4446792389077936,
      "learning_rate": 9.094748566624285e-06,
      "loss": 0.2537,
      "step": 2734
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2915641268691058,
      "learning_rate": 9.094004985487935e-06,
      "loss": 0.2263,
      "step": 2735
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5681978616500116,
      "learning_rate": 9.093261129508625e-06,
      "loss": 0.2655,
      "step": 2736
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6978472680341348,
      "learning_rate": 9.092516998736296e-06,
      "loss": 0.2595,
      "step": 2737
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5187774255232824,
      "learning_rate": 9.0917725932209e-06,
      "loss": 0.1866,
      "step": 2738
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.405692630637023,
      "learning_rate": 9.091027913012411e-06,
      "loss": 0.2198,
      "step": 2739
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2653642747439644,
      "learning_rate": 9.090282958160823e-06,
      "loss": 0.1838,
      "step": 2740
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.787358460986167,
      "learning_rate": 9.089537728716147e-06,
      "loss": 0.5551,
      "step": 2741
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2631310893127194,
      "learning_rate": 9.088792224728413e-06,
      "loss": 0.2308,
      "step": 2742
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2396355531791612,
      "learning_rate": 9.08804644624767e-06,
      "loss": 0.181,
      "step": 2743
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3729885545740517,
      "learning_rate": 9.08730039332398e-06,
      "loss": 0.1992,
      "step": 2744
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3585934430800366,
      "learning_rate": 9.08655406600743e-06,
      "loss": 0.2067,
      "step": 2745
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4222544190520328,
      "learning_rate": 9.085807464348127e-06,
      "loss": 0.2706,
      "step": 2746
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6052385396727238,
      "learning_rate": 9.085060588396188e-06,
      "loss": 0.2578,
      "step": 2747
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.121526491056624,
      "learning_rate": 9.084313438201754e-06,
      "loss": 0.5746,
      "step": 2748
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3475408266941615,
      "learning_rate": 9.083566013814985e-06,
      "loss": 0.2352,
      "step": 2749
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3000050120529731,
      "learning_rate": 9.082818315286054e-06,
      "loss": 0.1872,
      "step": 2750
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.354369059324886,
      "learning_rate": 9.082070342665163e-06,
      "loss": 0.7369,
      "step": 2751
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.374025744809407,
      "learning_rate": 9.08132209600252e-06,
      "loss": 0.2324,
      "step": 2752
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6888991493437802,
      "learning_rate": 9.080573575348358e-06,
      "loss": 0.285,
      "step": 2753
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3115675050413185,
      "learning_rate": 9.079824780752929e-06,
      "loss": 0.2233,
      "step": 2754
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.935306431439466,
      "learning_rate": 9.079075712266501e-06,
      "loss": 0.4932,
      "step": 2755
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6274482135732544,
      "learning_rate": 9.078326369939361e-06,
      "loss": 0.2835,
      "step": 2756
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.312514077491614,
      "learning_rate": 9.077576753821815e-06,
      "loss": 0.65,
      "step": 2757
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2132416216205233,
      "learning_rate": 9.076826863964188e-06,
      "loss": 0.1717,
      "step": 2758
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3067087425137986,
      "learning_rate": 9.07607670041682e-06,
      "loss": 0.1795,
      "step": 2759
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3659413034086476,
      "learning_rate": 9.075326263230073e-06,
      "loss": 0.208,
      "step": 2760
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.432916099636624,
      "learning_rate": 9.074575552454325e-06,
      "loss": 0.2197,
      "step": 2761
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3704222847454821,
      "learning_rate": 9.073824568139979e-06,
      "loss": 0.1894,
      "step": 2762
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4754252279492717,
      "learning_rate": 9.073073310337443e-06,
      "loss": 0.2448,
      "step": 2763
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.780182301212309,
      "learning_rate": 9.072321779097155e-06,
      "loss": 0.4945,
      "step": 2764
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5005067063616255,
      "learning_rate": 9.071569974469569e-06,
      "loss": 0.2177,
      "step": 2765
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5749056665739491,
      "learning_rate": 9.070817896505153e-06,
      "loss": 0.2624,
      "step": 2766
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.594612179109651,
      "learning_rate": 9.0700655452544e-06,
      "loss": 0.2314,
      "step": 2767
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3433587189141731,
      "learning_rate": 9.06931292076781e-06,
      "loss": 0.2086,
      "step": 2768
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4497876571013442,
      "learning_rate": 9.068560023095917e-06,
      "loss": 0.2195,
      "step": 2769
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7234763093412258,
      "learning_rate": 9.067806852289262e-06,
      "loss": 0.2493,
      "step": 2770
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3520959895920677,
      "learning_rate": 9.067053408398409e-06,
      "loss": 0.1883,
      "step": 2771
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4488528219338068,
      "learning_rate": 9.06629969147394e-06,
      "loss": 0.2466,
      "step": 2772
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4125044326631508,
      "learning_rate": 9.065545701566448e-06,
      "loss": 0.2241,
      "step": 2773
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.613366786154387,
      "learning_rate": 9.064791438726557e-06,
      "loss": 0.6188,
      "step": 2774
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5570319551335765,
      "learning_rate": 9.0640369030049e-06,
      "loss": 0.2984,
      "step": 2775
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7640627598685126,
      "learning_rate": 9.063282094452133e-06,
      "loss": 0.2718,
      "step": 2776
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5415451461254195,
      "learning_rate": 9.062527013118926e-06,
      "loss": 0.2708,
      "step": 2777
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2685295948939568,
      "learning_rate": 9.061771659055974e-06,
      "loss": 0.175,
      "step": 2778
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5650799437982696,
      "learning_rate": 9.061016032313984e-06,
      "loss": 0.216,
      "step": 2779
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3140040358820873,
      "learning_rate": 9.060260132943682e-06,
      "loss": 0.2223,
      "step": 2780
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2813241792749517,
      "learning_rate": 9.059503960995816e-06,
      "loss": 0.2133,
      "step": 2781
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3287037191082507,
      "learning_rate": 9.058747516521149e-06,
      "loss": 0.2113,
      "step": 2782
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.158487620129103,
      "learning_rate": 9.057990799570464e-06,
      "loss": 0.6308,
      "step": 2783
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2535091261317577,
      "learning_rate": 9.05723381019456e-06,
      "loss": 0.185,
      "step": 2784
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4418232645320088,
      "learning_rate": 9.056476548444258e-06,
      "loss": 0.2286,
      "step": 2785
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2710842933275697,
      "learning_rate": 9.055719014370396e-06,
      "loss": 0.2066,
      "step": 2786
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3767040865775415,
      "learning_rate": 9.054961208023827e-06,
      "loss": 0.1812,
      "step": 2787
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4680325387215052,
      "learning_rate": 9.054203129455425e-06,
      "loss": 0.2061,
      "step": 2788
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4768127429690432,
      "learning_rate": 9.053444778716085e-06,
      "loss": 0.2298,
      "step": 2789
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3149869910181915,
      "learning_rate": 9.052686155856716e-06,
      "loss": 0.2091,
      "step": 2790
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5389201150829421,
      "learning_rate": 9.051927260928243e-06,
      "loss": 0.2414,
      "step": 2791
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4859793752387178,
      "learning_rate": 9.051168093981619e-06,
      "loss": 0.2539,
      "step": 2792
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.491323541728569,
      "learning_rate": 9.050408655067806e-06,
      "loss": 0.6121,
      "step": 2793
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3732632355151762,
      "learning_rate": 9.049648944237788e-06,
      "loss": 0.1904,
      "step": 2794
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5624289356446526,
      "learning_rate": 9.048888961542565e-06,
      "loss": 0.2429,
      "step": 2795
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3400308202123299,
      "learning_rate": 9.048128707033159e-06,
      "loss": 0.2525,
      "step": 2796
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4765966949134453,
      "learning_rate": 9.04736818076061e-06,
      "loss": 0.2564,
      "step": 2797
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2927968756760015,
      "learning_rate": 9.04660738277597e-06,
      "loss": 0.2248,
      "step": 2798
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3771446452894096,
      "learning_rate": 9.045846313130313e-06,
      "loss": 0.164,
      "step": 2799
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2228823830858666,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.2032,
      "step": 2800
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5615435130701139,
      "learning_rate": 9.044323359060352e-06,
      "loss": 0.2282,
      "step": 2801
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2836631873300806,
      "learning_rate": 9.043561474738285e-06,
      "loss": 0.216,
      "step": 2802
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4032398860712012,
      "learning_rate": 9.042799318959684e-06,
      "loss": 0.2559,
      "step": 2803
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.439838926176643,
      "learning_rate": 9.042036891775715e-06,
      "loss": 0.2166,
      "step": 2804
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3455688786043316,
      "learning_rate": 9.041274193237565e-06,
      "loss": 0.2225,
      "step": 2805
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6534876106769998,
      "learning_rate": 9.040511223396432e-06,
      "loss": 0.2757,
      "step": 2806
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.52418003261114,
      "learning_rate": 9.039747982303539e-06,
      "loss": 0.2419,
      "step": 2807
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5657551020028315,
      "learning_rate": 9.038984470010123e-06,
      "loss": 0.2019,
      "step": 2808
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2971651078265394,
      "learning_rate": 9.038220686567443e-06,
      "loss": 0.2228,
      "step": 2809
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4426981816203392,
      "learning_rate": 9.037456632026774e-06,
      "loss": 0.197,
      "step": 2810
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.467132715805502,
      "learning_rate": 9.036692306439406e-06,
      "loss": 0.2612,
      "step": 2811
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5428826911405988,
      "learning_rate": 9.035927709856654e-06,
      "loss": 0.2073,
      "step": 2812
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7316056738360845,
      "learning_rate": 9.035162842329845e-06,
      "loss": 0.2619,
      "step": 2813
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4074668426840706,
      "learning_rate": 9.034397703910328e-06,
      "loss": 0.1829,
      "step": 2814
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4573635293256941,
      "learning_rate": 9.033632294649473e-06,
      "loss": 0.2717,
      "step": 2815
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5978212214492764,
      "learning_rate": 9.032866614598658e-06,
      "loss": 0.2781,
      "step": 2816
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3990474932649402,
      "learning_rate": 9.032100663809288e-06,
      "loss": 0.1865,
      "step": 2817
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.478225377470301,
      "learning_rate": 9.031334442332784e-06,
      "loss": 0.2209,
      "step": 2818
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.324947628861508,
      "learning_rate": 9.030567950220586e-06,
      "loss": 0.1901,
      "step": 2819
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5447661822890375,
      "learning_rate": 9.029801187524147e-06,
      "loss": 0.2205,
      "step": 2820
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.520956277663838,
      "learning_rate": 9.029034154294945e-06,
      "loss": 0.2232,
      "step": 2821
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6054121043114649,
      "learning_rate": 9.028266850584473e-06,
      "loss": 0.1992,
      "step": 2822
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4074866472708532,
      "learning_rate": 9.027499276444242e-06,
      "loss": 0.2425,
      "step": 2823
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5562234465102696,
      "learning_rate": 9.026731431925784e-06,
      "loss": 0.2364,
      "step": 2824
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4378675591134906,
      "learning_rate": 9.025963317080641e-06,
      "loss": 0.1817,
      "step": 2825
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3776608566050383,
      "learning_rate": 9.025194931960385e-06,
      "loss": 0.2358,
      "step": 2826
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4326468347676402,
      "learning_rate": 9.024426276616595e-06,
      "loss": 0.2475,
      "step": 2827
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5151669557882912,
      "learning_rate": 9.023657351100878e-06,
      "loss": 0.2024,
      "step": 2828
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0865485697602366,
      "learning_rate": 9.02288815546485e-06,
      "loss": 0.1421,
      "step": 2829
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4004083982157023,
      "learning_rate": 9.022118689760153e-06,
      "loss": 0.1914,
      "step": 2830
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5341325603388152,
      "learning_rate": 9.02134895403844e-06,
      "loss": 0.217,
      "step": 2831
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6584298425465849,
      "learning_rate": 9.020578948351389e-06,
      "loss": 0.2646,
      "step": 2832
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3658798680635684,
      "learning_rate": 9.01980867275069e-06,
      "loss": 0.2486,
      "step": 2833
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.242840670162026,
      "learning_rate": 9.019038127288056e-06,
      "loss": 0.5635,
      "step": 2834
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.2319528036005165,
      "learning_rate": 9.018267312015214e-06,
      "loss": 0.6663,
      "step": 2835
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4873446778978956,
      "learning_rate": 9.017496226983915e-06,
      "loss": 0.2237,
      "step": 2836
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.34754222432654,
      "learning_rate": 9.01672487224592e-06,
      "loss": 0.2205,
      "step": 2837
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3877551988307628,
      "learning_rate": 9.015953247853014e-06,
      "loss": 0.2484,
      "step": 2838
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2750056098688927,
      "learning_rate": 9.015181353856998e-06,
      "loss": 0.1852,
      "step": 2839
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5586050695209854,
      "learning_rate": 9.014409190309695e-06,
      "loss": 0.2909,
      "step": 2840
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.957771991515571,
      "learning_rate": 9.013636757262938e-06,
      "loss": 0.6443,
      "step": 2841
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.978687030343337,
      "learning_rate": 9.012864054768584e-06,
      "loss": 0.694,
      "step": 2842
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.68732340687833,
      "learning_rate": 9.01209108287851e-06,
      "loss": 0.2597,
      "step": 2843
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2070138246231235,
      "learning_rate": 9.011317841644602e-06,
      "loss": 0.1918,
      "step": 2844
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.818747401966009,
      "learning_rate": 9.010544331118776e-06,
      "loss": 0.2171,
      "step": 2845
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.909821446912314,
      "learning_rate": 9.009770551352957e-06,
      "loss": 0.6904,
      "step": 2846
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4625678309129038,
      "learning_rate": 9.008996502399092e-06,
      "loss": 0.1937,
      "step": 2847
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3403198462042074,
      "learning_rate": 9.008222184309145e-06,
      "loss": 0.2403,
      "step": 2848
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3078904259706645,
      "learning_rate": 9.007447597135097e-06,
      "loss": 0.1849,
      "step": 2849
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4700985389589813,
      "learning_rate": 9.006672740928952e-06,
      "loss": 0.2278,
      "step": 2850
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5413570894979507,
      "learning_rate": 9.005897615742723e-06,
      "loss": 0.2277,
      "step": 2851
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.186934536845539,
      "learning_rate": 9.005122221628452e-06,
      "loss": 0.6427,
      "step": 2852
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4584957023587855,
      "learning_rate": 9.00434655863819e-06,
      "loss": 0.2496,
      "step": 2853
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.047544675111224,
      "learning_rate": 9.003570626824013e-06,
      "loss": 0.3899,
      "step": 2854
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4842509091263645,
      "learning_rate": 9.002794426238009e-06,
      "loss": 0.2496,
      "step": 2855
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.316297901270813,
      "learning_rate": 9.002017956932285e-06,
      "loss": 0.7612,
      "step": 2856
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3873258790927314,
      "learning_rate": 9.001241218958972e-06,
      "loss": 0.1843,
      "step": 2857
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3310898189891012,
      "learning_rate": 9.00046421237021e-06,
      "loss": 0.2229,
      "step": 2858
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5319725040429875,
      "learning_rate": 8.999686937218168e-06,
      "loss": 0.2215,
      "step": 2859
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5651787938288904,
      "learning_rate": 8.998909393555022e-06,
      "loss": 0.1776,
      "step": 2860
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5884858878347807,
      "learning_rate": 8.998131581432972e-06,
      "loss": 0.2042,
      "step": 2861
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2677565136802358,
      "learning_rate": 8.997353500904234e-06,
      "loss": 0.2085,
      "step": 2862
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.270508258368694,
      "learning_rate": 8.996575152021045e-06,
      "loss": 0.5937,
      "step": 2863
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3566657617792557,
      "learning_rate": 8.995796534835656e-06,
      "loss": 0.2173,
      "step": 2864
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3821513087646573,
      "learning_rate": 8.995017649400341e-06,
      "loss": 0.1928,
      "step": 2865
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4789968006256975,
      "learning_rate": 8.994238495767385e-06,
      "loss": 0.2704,
      "step": 2866
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.302866811475751,
      "learning_rate": 8.993459073989098e-06,
      "loss": 0.1911,
      "step": 2867
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3687427729722517,
      "learning_rate": 8.992679384117802e-06,
      "loss": 0.256,
      "step": 2868
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.466415513897628,
      "learning_rate": 8.991899426205844e-06,
      "loss": 0.2006,
      "step": 2869
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3901368692979712,
      "learning_rate": 8.99111920030558e-06,
      "loss": 0.2073,
      "step": 2870
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.515818563088692,
      "learning_rate": 8.990338706469393e-06,
      "loss": 0.2283,
      "step": 2871
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4387095170975521,
      "learning_rate": 8.989557944749677e-06,
      "loss": 0.2169,
      "step": 2872
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.318235840370533,
      "learning_rate": 8.988776915198849e-06,
      "loss": 0.2323,
      "step": 2873
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.770769206536903,
      "learning_rate": 8.987995617869341e-06,
      "loss": 0.7652,
      "step": 2874
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3194625550872465,
      "learning_rate": 8.987214052813605e-06,
      "loss": 0.2072,
      "step": 2875
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5375410732303605,
      "learning_rate": 8.986432220084108e-06,
      "loss": 0.1963,
      "step": 2876
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3580321144035887,
      "learning_rate": 8.985650119733338e-06,
      "loss": 0.2058,
      "step": 2877
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4223747414901433,
      "learning_rate": 8.9848677518138e-06,
      "loss": 0.2097,
      "step": 2878
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3684062925926486,
      "learning_rate": 8.984085116378015e-06,
      "loss": 0.2448,
      "step": 2879
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5209167131297223,
      "learning_rate": 8.983302213478525e-06,
      "loss": 0.2365,
      "step": 2880
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4976221426758953,
      "learning_rate": 8.982519043167888e-06,
      "loss": 0.2089,
      "step": 2881
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4877203565527528,
      "learning_rate": 8.981735605498683e-06,
      "loss": 0.2168,
      "step": 2882
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4102290530863708,
      "learning_rate": 8.9809519005235e-06,
      "loss": 0.1862,
      "step": 2883
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.686511939581189,
      "learning_rate": 8.980167928294956e-06,
      "loss": 0.6404,
      "step": 2884
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3116354613445917,
      "learning_rate": 8.97938368886568e-06,
      "loss": 0.2243,
      "step": 2885
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3796307310451612,
      "learning_rate": 8.978599182288319e-06,
      "loss": 0.2369,
      "step": 2886
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3931589519539738,
      "learning_rate": 8.97781440861554e-06,
      "loss": 0.191,
      "step": 2887
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.596134309935943,
      "learning_rate": 8.977029367900028e-06,
      "loss": 0.5147,
      "step": 2888
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6230139815261981,
      "learning_rate": 8.976244060194484e-06,
      "loss": 0.2381,
      "step": 2889
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3575497466361595,
      "learning_rate": 8.97545848555163e-06,
      "loss": 0.194,
      "step": 2890
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5174563498139526,
      "learning_rate": 8.9746726440242e-06,
      "loss": 0.211,
      "step": 2891
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.451068308935107,
      "learning_rate": 8.973886535664954e-06,
      "loss": 0.2456,
      "step": 2892
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3813627297283715,
      "learning_rate": 8.973100160526666e-06,
      "loss": 0.1608,
      "step": 2893
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4438076529003137,
      "learning_rate": 8.972313518662125e-06,
      "loss": 0.2236,
      "step": 2894
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.942817792351023,
      "learning_rate": 8.971526610124142e-06,
      "loss": 0.5243,
      "step": 2895
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3242244202244997,
      "learning_rate": 8.970739434965544e-06,
      "loss": 0.1937,
      "step": 2896
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2694900034886134,
      "learning_rate": 8.969951993239177e-06,
      "loss": 0.1772,
      "step": 2897
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6052848770131556,
      "learning_rate": 8.969164284997905e-06,
      "loss": 0.2432,
      "step": 2898
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4316202837576615,
      "learning_rate": 8.968376310294608e-06,
      "loss": 0.1867,
      "step": 2899
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2331308581215907,
      "learning_rate": 8.967588069182184e-06,
      "loss": 0.1828,
      "step": 2900
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.450441644844448,
      "learning_rate": 8.966799561713556e-06,
      "loss": 0.3339,
      "step": 2901
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3921228475439893,
      "learning_rate": 8.96601078794165e-06,
      "loss": 0.1921,
      "step": 2902
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5342836828338007,
      "learning_rate": 8.965221747919424e-06,
      "loss": 0.2616,
      "step": 2903
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3307995401038824,
      "learning_rate": 8.964432441699848e-06,
      "loss": 0.1908,
      "step": 2904
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4488465674166047,
      "learning_rate": 8.963642869335913e-06,
      "loss": 0.2138,
      "step": 2905
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2844722805684925,
      "learning_rate": 8.96285303088062e-06,
      "loss": 0.1935,
      "step": 2906
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4756924621700949,
      "learning_rate": 8.962062926386998e-06,
      "loss": 0.2027,
      "step": 2907
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4807284157803684,
      "learning_rate": 8.961272555908084e-06,
      "loss": 0.2689,
      "step": 2908
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4348991029678073,
      "learning_rate": 8.960481919496944e-06,
      "loss": 0.2156,
      "step": 2909
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3105098253428948,
      "learning_rate": 8.959691017206653e-06,
      "loss": 0.2353,
      "step": 2910
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.563290091567969,
      "learning_rate": 8.958899849090306e-06,
      "loss": 0.2219,
      "step": 2911
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.534905143277963,
      "learning_rate": 8.958108415201017e-06,
      "loss": 0.2411,
      "step": 2912
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3994036775656344,
      "learning_rate": 8.957316715591918e-06,
      "loss": 0.1846,
      "step": 2913
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5292688357309916,
      "learning_rate": 8.956524750316158e-06,
      "loss": 0.2362,
      "step": 2914
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4516891801375305,
      "learning_rate": 8.955732519426902e-06,
      "loss": 0.2289,
      "step": 2915
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.141907338968659,
      "learning_rate": 8.954940022977338e-06,
      "loss": 0.6361,
      "step": 2916
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.412193324937404,
      "learning_rate": 8.954147261020667e-06,
      "loss": 0.2478,
      "step": 2917
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5285618242672612,
      "learning_rate": 8.95335423361011e-06,
      "loss": 0.1927,
      "step": 2918
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4654694347043942,
      "learning_rate": 8.952560940798905e-06,
      "loss": 0.212,
      "step": 2919
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4790847429447933,
      "learning_rate": 8.951767382640308e-06,
      "loss": 0.2413,
      "step": 2920
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3821051837903717,
      "learning_rate": 8.950973559187593e-06,
      "loss": 0.1719,
      "step": 2921
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.236684201702628,
      "learning_rate": 8.950179470494051e-06,
      "loss": 0.1936,
      "step": 2922
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.578188524537762,
      "learning_rate": 8.949385116612994e-06,
      "loss": 0.6843,
      "step": 2923
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.439585453891126,
      "learning_rate": 8.948590497597749e-06,
      "loss": 0.6007,
      "step": 2924
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.344027656495916,
      "learning_rate": 8.947795613501658e-06,
      "loss": 0.2326,
      "step": 2925
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4299267085226401,
      "learning_rate": 8.947000464378088e-06,
      "loss": 0.2543,
      "step": 2926
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4909179275638866,
      "learning_rate": 8.946205050280417e-06,
      "loss": 0.2421,
      "step": 2927
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.179269879670114,
      "learning_rate": 8.945409371262044e-06,
      "loss": 0.5028,
      "step": 2928
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4427496160945599,
      "learning_rate": 8.944613427376385e-06,
      "loss": 0.2538,
      "step": 2929
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4328889098681798,
      "learning_rate": 8.943817218676877e-06,
      "loss": 0.2338,
      "step": 2930
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.061666144203288,
      "learning_rate": 8.943020745216968e-06,
      "loss": 0.4691,
      "step": 2931
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.438352709413,
      "learning_rate": 8.942224007050131e-06,
      "loss": 0.2503,
      "step": 2932
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4303951356733517,
      "learning_rate": 8.941427004229851e-06,
      "loss": 0.2265,
      "step": 2933
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4124173642321693,
      "learning_rate": 8.940629736809635e-06,
      "loss": 0.2468,
      "step": 2934
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.235961248443977,
      "learning_rate": 8.939832204843003e-06,
      "loss": 0.2217,
      "step": 2935
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2548051161104445,
      "learning_rate": 8.939034408383502e-06,
      "loss": 0.1945,
      "step": 2936
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.490504818831335,
      "learning_rate": 8.938236347484684e-06,
      "loss": 0.2447,
      "step": 2937
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3978936944130946,
      "learning_rate": 8.937438022200126e-06,
      "loss": 0.2327,
      "step": 2938
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3862364734631518,
      "learning_rate": 8.936639432583424e-06,
      "loss": 0.2009,
      "step": 2939
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3877759778435041,
      "learning_rate": 8.935840578688191e-06,
      "loss": 0.2141,
      "step": 2940
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4200776610196197,
      "learning_rate": 8.935041460568055e-06,
      "loss": 0.2228,
      "step": 2941
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.475752988629331,
      "learning_rate": 8.934242078276662e-06,
      "loss": 0.1913,
      "step": 2942
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2676661732762509,
      "learning_rate": 8.933442431867678e-06,
      "loss": 0.1868,
      "step": 2943
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.009595564956046,
      "learning_rate": 8.932642521394786e-06,
      "loss": 0.6896,
      "step": 2944
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.923425990247967,
      "learning_rate": 8.931842346911688e-06,
      "loss": 0.7366,
      "step": 2945
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5496012799458294,
      "learning_rate": 8.931041908472098e-06,
      "loss": 0.1953,
      "step": 2946
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4694107537187513,
      "learning_rate": 8.930241206129754e-06,
      "loss": 0.2124,
      "step": 2947
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.156646506399253,
      "learning_rate": 8.929440239938409e-06,
      "loss": 0.6602,
      "step": 2948
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.913468333245076,
      "learning_rate": 8.928639009951837e-06,
      "loss": 0.6046,
      "step": 2949
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.465254233573523,
      "learning_rate": 8.927837516223824e-06,
      "loss": 0.2328,
      "step": 2950
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6024147367264667,
      "learning_rate": 8.927035758808178e-06,
      "loss": 0.2904,
      "step": 2951
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.097798826496697,
      "learning_rate": 8.926233737758722e-06,
      "loss": 0.6091,
      "step": 2952
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.368522890277418,
      "learning_rate": 8.9254314531293e-06,
      "loss": 0.1999,
      "step": 2953
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4722783020652448,
      "learning_rate": 8.924628904973771e-06,
      "loss": 0.2357,
      "step": 2954
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4069748491539367,
      "learning_rate": 8.923826093346013e-06,
      "loss": 0.2291,
      "step": 2955
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5562678957802336,
      "learning_rate": 8.92302301829992e-06,
      "loss": 0.1873,
      "step": 2956
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.28741795386435,
      "learning_rate": 8.922219679889406e-06,
      "loss": 0.1802,
      "step": 2957
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5606671655602036,
      "learning_rate": 8.9214160781684e-06,
      "loss": 0.1958,
      "step": 2958
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.789161722564476,
      "learning_rate": 8.92061221319085e-06,
      "loss": 0.7222,
      "step": 2959
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3413033486236767,
      "learning_rate": 8.919808085010726e-06,
      "loss": 0.18,
      "step": 2960
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3772492899815412,
      "learning_rate": 8.919003693682008e-06,
      "loss": 0.2054,
      "step": 2961
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.142316489802902,
      "learning_rate": 8.918199039258697e-06,
      "loss": 0.5801,
      "step": 2962
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2335058545110258,
      "learning_rate": 8.917394121794814e-06,
      "loss": 0.2602,
      "step": 2963
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3579528788615296,
      "learning_rate": 8.916588941344393e-06,
      "loss": 0.2342,
      "step": 2964
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3640735825962527,
      "learning_rate": 8.915783497961492e-06,
      "loss": 0.1982,
      "step": 2965
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.350564116499505,
      "learning_rate": 8.914977791700178e-06,
      "loss": 0.1843,
      "step": 2966
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.3814629638578175,
      "learning_rate": 8.914171822614543e-06,
      "loss": 0.6641,
      "step": 2967
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3047470511677055,
      "learning_rate": 8.913365590758695e-06,
      "loss": 0.1865,
      "step": 2968
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2583590551135229,
      "learning_rate": 8.912559096186759e-06,
      "loss": 0.1654,
      "step": 2969
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4165514591125556,
      "learning_rate": 8.911752338952875e-06,
      "loss": 0.2117,
      "step": 2970
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.52258832302981,
      "learning_rate": 8.910945319111204e-06,
      "loss": 0.4941,
      "step": 2971
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4804116563106005,
      "learning_rate": 8.910138036715924e-06,
      "loss": 0.2239,
      "step": 2972
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.582575054319406,
      "learning_rate": 8.90933049182123e-06,
      "loss": 0.1789,
      "step": 2973
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3345109313951915,
      "learning_rate": 8.908522684481336e-06,
      "loss": 0.1995,
      "step": 2974
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3138534937746604,
      "learning_rate": 8.907714614750473e-06,
      "loss": 0.1914,
      "step": 2975
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5972900555668406,
      "learning_rate": 8.906906282682886e-06,
      "loss": 0.2433,
      "step": 2976
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4077482613342298,
      "learning_rate": 8.906097688332844e-06,
      "loss": 0.2262,
      "step": 2977
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3154436412799022,
      "learning_rate": 8.905288831754628e-06,
      "loss": 0.1838,
      "step": 2978
    },
    {
      "epoch": 0.24,
      "grad_norm": 75.61717082444883,
      "learning_rate": 8.904479713002542e-06,
      "loss": 0.5783,
      "step": 2979
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.591026012209408,
      "learning_rate": 8.9036703321309e-06,
      "loss": 0.2478,
      "step": 2980
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4679314210011754,
      "learning_rate": 8.902860689194044e-06,
      "loss": 0.2087,
      "step": 2981
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.309384892215417,
      "learning_rate": 8.902050784246324e-06,
      "loss": 0.2452,
      "step": 2982
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4098651448350585,
      "learning_rate": 8.901240617342111e-06,
      "loss": 0.215,
      "step": 2983
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5564166714820489,
      "learning_rate": 8.900430188535796e-06,
      "loss": 0.2637,
      "step": 2984
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3766444097573167,
      "learning_rate": 8.899619497881784e-06,
      "loss": 0.2014,
      "step": 2985
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3376982352528928,
      "learning_rate": 8.8988085454345e-06,
      "loss": 0.1921,
      "step": 2986
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.423813842254584,
      "learning_rate": 8.897997331248384e-06,
      "loss": 0.5051,
      "step": 2987
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.08711338762155,
      "learning_rate": 8.8971858553779e-06,
      "loss": 0.635,
      "step": 2988
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4570796969531588,
      "learning_rate": 8.896374117877519e-06,
      "loss": 0.2543,
      "step": 2989
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.439104010044396,
      "learning_rate": 8.895562118801739e-06,
      "loss": 0.2395,
      "step": 2990
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4181778375549878,
      "learning_rate": 8.89474985820507e-06,
      "loss": 0.2624,
      "step": 2991
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3671217247810508,
      "learning_rate": 8.893937336142043e-06,
      "loss": 0.2422,
      "step": 2992
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.518766375842734,
      "learning_rate": 8.893124552667203e-06,
      "loss": 0.1654,
      "step": 2993
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4632857869933948,
      "learning_rate": 8.892311507835118e-06,
      "loss": 0.2439,
      "step": 2994
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.9888690892946785,
      "learning_rate": 8.891498201700368e-06,
      "loss": 0.6135,
      "step": 2995
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.431313920502848,
      "learning_rate": 8.890684634317552e-06,
      "loss": 0.265,
      "step": 2996
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.53475140058515,
      "learning_rate": 8.889870805741288e-06,
      "loss": 0.2525,
      "step": 2997
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2841874124414565,
      "learning_rate": 8.889056716026213e-06,
      "loss": 0.2127,
      "step": 2998
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.218451917714868,
      "learning_rate": 8.888242365226975e-06,
      "loss": 0.6813,
      "step": 2999
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.325017544413177,
      "learning_rate": 8.887427753398249e-06,
      "loss": 0.2094,
      "step": 3000
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.759196484638667,
      "learning_rate": 8.886612880594715e-06,
      "loss": 0.7537,
      "step": 3001
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2921386736848803,
      "learning_rate": 8.885797746871085e-06,
      "loss": 0.2751,
      "step": 3002
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3224681819846724,
      "learning_rate": 8.884982352282078e-06,
      "loss": 0.2141,
      "step": 3003
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.359173149780651,
      "learning_rate": 8.884166696882436e-06,
      "loss": 0.1897,
      "step": 3004
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5703107608514189,
      "learning_rate": 8.883350780726915e-06,
      "loss": 0.2272,
      "step": 3005
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5044476234016868,
      "learning_rate": 8.88253460387029e-06,
      "loss": 0.2273,
      "step": 3006
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4696345431868707,
      "learning_rate": 8.881718166367353e-06,
      "loss": 0.1896,
      "step": 3007
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4119886546584697,
      "learning_rate": 8.880901468272913e-06,
      "loss": 0.2214,
      "step": 3008
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.460589931004384,
      "learning_rate": 8.8800845096418e-06,
      "loss": 0.252,
      "step": 3009
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3064104514517634,
      "learning_rate": 8.87926729052886e-06,
      "loss": 0.1851,
      "step": 3010
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3861896346737865,
      "learning_rate": 8.87844981098895e-06,
      "loss": 0.2041,
      "step": 3011
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.207236718994992,
      "learning_rate": 8.877632071076952e-06,
      "loss": 0.4273,
      "step": 3012
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3922746374429682,
      "learning_rate": 8.876814070847766e-06,
      "loss": 0.205,
      "step": 3013
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.368476905950356,
      "learning_rate": 8.875995810356306e-06,
      "loss": 0.6352,
      "step": 3014
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.627212978570021,
      "learning_rate": 8.875177289657502e-06,
      "loss": 0.299,
      "step": 3015
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.688386773371714,
      "learning_rate": 8.874358508806306e-06,
      "loss": 0.4555,
      "step": 3016
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2601481112015893,
      "learning_rate": 8.873539467857683e-06,
      "loss": 0.1944,
      "step": 3017
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.489229566544062,
      "learning_rate": 8.872720166866623e-06,
      "loss": 0.1905,
      "step": 3018
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3897923476764766,
      "learning_rate": 8.871900605888121e-06,
      "loss": 0.2158,
      "step": 3019
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5560663938837846,
      "learning_rate": 8.8710807849772e-06,
      "loss": 0.2506,
      "step": 3020
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4041101343418005,
      "learning_rate": 8.870260704188897e-06,
      "loss": 0.2276,
      "step": 3021
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.833725070068477,
      "learning_rate": 8.869440363578267e-06,
      "loss": 0.3166,
      "step": 3022
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3935390529440868,
      "learning_rate": 8.868619763200384e-06,
      "loss": 0.2166,
      "step": 3023
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4822744324400534,
      "learning_rate": 8.867798903110331e-06,
      "loss": 0.221,
      "step": 3024
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2986363180338931,
      "learning_rate": 8.866977783363219e-06,
      "loss": 0.2096,
      "step": 3025
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3982047314246508,
      "learning_rate": 8.866156404014175e-06,
      "loss": 0.1789,
      "step": 3026
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5540256850621013,
      "learning_rate": 8.865334765118335e-06,
      "loss": 0.1916,
      "step": 3027
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5869607996291173,
      "learning_rate": 8.864512866730862e-06,
      "loss": 0.2259,
      "step": 3028
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4862388865873297,
      "learning_rate": 8.863690708906931e-06,
      "loss": 0.1961,
      "step": 3029
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4146381242860455,
      "learning_rate": 8.862868291701735e-06,
      "loss": 0.2056,
      "step": 3030
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2485742860820093,
      "learning_rate": 8.862045615170487e-06,
      "loss": 0.1799,
      "step": 3031
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4809931990909388,
      "learning_rate": 8.861222679368416e-06,
      "loss": 0.1888,
      "step": 3032
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3767173889362387,
      "learning_rate": 8.860399484350768e-06,
      "loss": 0.2001,
      "step": 3033
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.43973407582441,
      "learning_rate": 8.859576030172804e-06,
      "loss": 0.2414,
      "step": 3034
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.399524638135557,
      "learning_rate": 8.858752316889809e-06,
      "loss": 0.6516,
      "step": 3035
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5671356376652963,
      "learning_rate": 8.857928344557079e-06,
      "loss": 0.2693,
      "step": 3036
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.866467922566804,
      "learning_rate": 8.857104113229929e-06,
      "loss": 0.6296,
      "step": 3037
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3960610482910183,
      "learning_rate": 8.856279622963694e-06,
      "loss": 0.2196,
      "step": 3038
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3575735353763143,
      "learning_rate": 8.855454873813724e-06,
      "loss": 0.2346,
      "step": 3039
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3132629173162411,
      "learning_rate": 8.854629865835387e-06,
      "loss": 0.2147,
      "step": 3040
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.919591987793346,
      "learning_rate": 8.853804599084068e-06,
      "loss": 0.5371,
      "step": 3041
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2964773220704293,
      "learning_rate": 8.852979073615172e-06,
      "loss": 0.206,
      "step": 3042
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3785364084279794,
      "learning_rate": 8.852153289484114e-06,
      "loss": 0.2209,
      "step": 3043
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4406195320583162,
      "learning_rate": 8.851327246746334e-06,
      "loss": 0.2299,
      "step": 3044
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3430280094850309,
      "learning_rate": 8.850500945457286e-06,
      "loss": 0.1909,
      "step": 3045
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3193913896377703,
      "learning_rate": 8.849674385672444e-06,
      "loss": 0.2253,
      "step": 3046
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4175866522258325,
      "learning_rate": 8.848847567447298e-06,
      "loss": 0.1864,
      "step": 3047
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4087022432975331,
      "learning_rate": 8.848020490837352e-06,
      "loss": 0.2492,
      "step": 3048
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2203036704738788,
      "learning_rate": 8.84719315589813e-06,
      "loss": 0.1998,
      "step": 3049
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2001359681554475,
      "learning_rate": 8.846365562685178e-06,
      "loss": 0.171,
      "step": 3050
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5737706668660134,
      "learning_rate": 8.845537711254048e-06,
      "loss": 0.3049,
      "step": 3051
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.506734592547174,
      "learning_rate": 8.844709601660323e-06,
      "loss": 0.6631,
      "step": 3052
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3029872973433998,
      "learning_rate": 8.843881233959592e-06,
      "loss": 0.2318,
      "step": 3053
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5730809171370128,
      "learning_rate": 8.843052608207468e-06,
      "loss": 0.2274,
      "step": 3054
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3845940713402958,
      "learning_rate": 8.842223724459578e-06,
      "loss": 0.1717,
      "step": 3055
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.473633082083225,
      "learning_rate": 8.841394582771568e-06,
      "loss": 0.1881,
      "step": 3056
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.402038116612309,
      "learning_rate": 8.840565183199102e-06,
      "loss": 0.4127,
      "step": 3057
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4322500257667174,
      "learning_rate": 8.839735525797857e-06,
      "loss": 0.2449,
      "step": 3058
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3700656809863976,
      "learning_rate": 8.838905610623532e-06,
      "loss": 0.2159,
      "step": 3059
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3712450480118383,
      "learning_rate": 8.838075437731844e-06,
      "loss": 0.214,
      "step": 3060
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4756749356418306,
      "learning_rate": 8.837245007178522e-06,
      "loss": 0.2307,
      "step": 3061
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1130716994336955,
      "learning_rate": 8.836414319019314e-06,
      "loss": 0.1668,
      "step": 3062
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3863741235710332,
      "learning_rate": 8.83558337330999e-06,
      "loss": 0.2495,
      "step": 3063
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1567591748237191,
      "learning_rate": 8.834752170106334e-06,
      "loss": 0.1738,
      "step": 3064
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4198624477188664,
      "learning_rate": 8.833920709464146e-06,
      "loss": 0.2311,
      "step": 3065
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2219532907420627,
      "learning_rate": 8.833088991439245e-06,
      "loss": 0.1676,
      "step": 3066
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3332142493078298,
      "learning_rate": 8.832257016087464e-06,
      "loss": 0.1981,
      "step": 3067
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3883019204283449,
      "learning_rate": 8.83142478346466e-06,
      "loss": 0.2167,
      "step": 3068
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4719396635417274,
      "learning_rate": 8.830592293626702e-06,
      "loss": 0.2073,
      "step": 3069
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3366231586690822,
      "learning_rate": 8.829759546629474e-06,
      "loss": 0.2236,
      "step": 3070
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.251079760118595,
      "learning_rate": 8.828926542528888e-06,
      "loss": 0.685,
      "step": 3071
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6107929707361612,
      "learning_rate": 8.828093281380859e-06,
      "loss": 0.2302,
      "step": 3072
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2683237733034751,
      "learning_rate": 8.82725976324133e-06,
      "loss": 0.1966,
      "step": 3073
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2554818995645787,
      "learning_rate": 8.826425988166259e-06,
      "loss": 0.178,
      "step": 3074
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.555182200381952,
      "learning_rate": 8.825591956211614e-06,
      "loss": 0.2474,
      "step": 3075
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3657277195365287,
      "learning_rate": 8.824757667433392e-06,
      "loss": 0.2156,
      "step": 3076
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3262778738154213,
      "learning_rate": 8.8239231218876e-06,
      "loss": 0.2073,
      "step": 3077
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3479177643160096,
      "learning_rate": 8.823088319630262e-06,
      "loss": 0.2461,
      "step": 3078
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1996438746816949,
      "learning_rate": 8.822253260717422e-06,
      "loss": 0.1436,
      "step": 3079
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4847628824101158,
      "learning_rate": 8.82141794520514e-06,
      "loss": 0.2517,
      "step": 3080
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4384736220957275,
      "learning_rate": 8.820582373149491e-06,
      "loss": 0.2243,
      "step": 3081
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3950071344974906,
      "learning_rate": 8.819746544606573e-06,
      "loss": 0.2355,
      "step": 3082
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3244181944903652,
      "learning_rate": 8.818910459632495e-06,
      "loss": 0.1694,
      "step": 3083
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4289850554357482,
      "learning_rate": 8.818074118283389e-06,
      "loss": 0.2537,
      "step": 3084
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5370309517445233,
      "learning_rate": 8.817237520615398e-06,
      "loss": 0.2272,
      "step": 3085
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4312273013700494,
      "learning_rate": 8.816400666684685e-06,
      "loss": 0.1803,
      "step": 3086
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6924668865681711,
      "learning_rate": 8.815563556547434e-06,
      "loss": 0.2733,
      "step": 3087
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2630884716859379,
      "learning_rate": 8.81472619025984e-06,
      "loss": 0.1854,
      "step": 3088
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.838163449231494,
      "learning_rate": 8.81388856787812e-06,
      "loss": 0.7649,
      "step": 3089
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2981509880149225,
      "learning_rate": 8.813050689458502e-06,
      "loss": 0.1963,
      "step": 3090
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5487570226763947,
      "learning_rate": 8.81221255505724e-06,
      "loss": 0.2355,
      "step": 3091
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5859963702678124,
      "learning_rate": 8.811374164730599e-06,
      "loss": 0.2526,
      "step": 3092
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3696084796315757,
      "learning_rate": 8.810535518534862e-06,
      "loss": 0.1768,
      "step": 3093
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3563731310390237,
      "learning_rate": 8.80969661652633e-06,
      "loss": 0.2235,
      "step": 3094
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3636269619546166,
      "learning_rate": 8.80885745876132e-06,
      "loss": 0.2011,
      "step": 3095
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4851621407302664,
      "learning_rate": 8.80801804529617e-06,
      "loss": 0.2356,
      "step": 3096
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4372345325521587,
      "learning_rate": 8.80717837618723e-06,
      "loss": 0.2592,
      "step": 3097
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3441734166989745,
      "learning_rate": 8.80633845149087e-06,
      "loss": 0.2015,
      "step": 3098
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5511142489781486,
      "learning_rate": 8.805498271263477e-06,
      "loss": 0.225,
      "step": 3099
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3048253057726924,
      "learning_rate": 8.804657835561456e-06,
      "loss": 0.1466,
      "step": 3100
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.424801274070495,
      "learning_rate": 8.803817144441227e-06,
      "loss": 0.1925,
      "step": 3101
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4577061990860265,
      "learning_rate": 8.802976197959228e-06,
      "loss": 0.2214,
      "step": 3102
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2734805013100003,
      "learning_rate": 8.802134996171913e-06,
      "loss": 0.2177,
      "step": 3103
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5259337578505816,
      "learning_rate": 8.801293539135755e-06,
      "loss": 0.2543,
      "step": 3104
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2175613357682593,
      "learning_rate": 8.800451826907245e-06,
      "loss": 0.1846,
      "step": 3105
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5235447681275573,
      "learning_rate": 8.79960985954289e-06,
      "loss": 0.1922,
      "step": 3106
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3490605782380212,
      "learning_rate": 8.798767637099212e-06,
      "loss": 0.2244,
      "step": 3107
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.467369010492928,
      "learning_rate": 8.797925159632753e-06,
      "loss": 0.1963,
      "step": 3108
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.017788081316822,
      "learning_rate": 8.79708242720007e-06,
      "loss": 0.7436,
      "step": 3109
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.861642620861123,
      "learning_rate": 8.79623943985774e-06,
      "loss": 0.6662,
      "step": 3110
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5083832891400009,
      "learning_rate": 8.795396197662355e-06,
      "loss": 0.2602,
      "step": 3111
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.418825262351778,
      "learning_rate": 8.794552700670522e-06,
      "loss": 0.2081,
      "step": 3112
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4549352327666223,
      "learning_rate": 8.79370894893887e-06,
      "loss": 0.2038,
      "step": 3113
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3306174978595904,
      "learning_rate": 8.792864942524042e-06,
      "loss": 0.1953,
      "step": 3114
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4253736787129068,
      "learning_rate": 8.792020681482698e-06,
      "loss": 0.1935,
      "step": 3115
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3800218047018245,
      "learning_rate": 8.791176165871515e-06,
      "loss": 0.2051,
      "step": 3116
    },
    {
      "epoch": 0.25,
      "grad_norm": 28.457137505952314,
      "learning_rate": 8.79033139574719e-06,
      "loss": 0.7992,
      "step": 3117
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2151133082444072,
      "learning_rate": 8.789486371166435e-06,
      "loss": 0.2043,
      "step": 3118
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.9588462300076515,
      "learning_rate": 8.788641092185978e-06,
      "loss": 0.6187,
      "step": 3119
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2413584015854517,
      "learning_rate": 8.787795558862566e-06,
      "loss": 0.1588,
      "step": 3120
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.281017448431733,
      "learning_rate": 8.786949771252961e-06,
      "loss": 0.2258,
      "step": 3121
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2979125104901477,
      "learning_rate": 8.786103729413944e-06,
      "loss": 0.2264,
      "step": 3122
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.9464230852850775,
      "learning_rate": 8.785257433402311e-06,
      "loss": 0.5349,
      "step": 3123
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.496713733707055,
      "learning_rate": 8.784410883274879e-06,
      "loss": 0.5529,
      "step": 3124
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4078929472123278,
      "learning_rate": 8.783564079088478e-06,
      "loss": 0.2158,
      "step": 3125
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3314309201145713,
      "learning_rate": 8.782717020899957e-06,
      "loss": 0.1986,
      "step": 3126
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.268109254509458,
      "learning_rate": 8.781869708766179e-06,
      "loss": 0.2266,
      "step": 3127
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3703667159421709,
      "learning_rate": 8.781022142744028e-06,
      "loss": 0.2077,
      "step": 3128
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.221546913753352,
      "learning_rate": 8.78017432289041e-06,
      "loss": 0.2124,
      "step": 3129
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2365090103815157,
      "learning_rate": 8.779326249262232e-06,
      "loss": 0.1499,
      "step": 3130
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2882794311642867,
      "learning_rate": 8.778477921916431e-06,
      "loss": 0.176,
      "step": 3131
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.742150743840332,
      "learning_rate": 8.777629340909963e-06,
      "loss": 0.7087,
      "step": 3132
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.339115561041797,
      "learning_rate": 8.77678050629979e-06,
      "loss": 0.1957,
      "step": 3133
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4837091657309796,
      "learning_rate": 8.775931418142895e-06,
      "loss": 0.2384,
      "step": 3134
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2602912463680471,
      "learning_rate": 8.775082076496287e-06,
      "loss": 0.2068,
      "step": 3135
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3588876192886485,
      "learning_rate": 8.77423248141698e-06,
      "loss": 0.2287,
      "step": 3136
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3376299926336737,
      "learning_rate": 8.773382632962011e-06,
      "loss": 0.1878,
      "step": 3137
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.44436557411278,
      "learning_rate": 8.772532531188434e-06,
      "loss": 0.7156,
      "step": 3138
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5568207927876347,
      "learning_rate": 8.771682176153317e-06,
      "loss": 0.2903,
      "step": 3139
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4733707072180862,
      "learning_rate": 8.770831567913747e-06,
      "loss": 0.2201,
      "step": 3140
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3133798351375443,
      "learning_rate": 8.76998070652683e-06,
      "loss": 0.2162,
      "step": 3141
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3526197884802227,
      "learning_rate": 8.769129592049685e-06,
      "loss": 0.2099,
      "step": 3142
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4184424008701173,
      "learning_rate": 8.768278224539451e-06,
      "loss": 0.2103,
      "step": 3143
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.217361278573516,
      "learning_rate": 8.767426604053282e-06,
      "loss": 0.1753,
      "step": 3144
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.302061094553151,
      "learning_rate": 8.76657473064835e-06,
      "loss": 0.1923,
      "step": 3145
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.365656133293616,
      "learning_rate": 8.765722604381843e-06,
      "loss": 0.242,
      "step": 3146
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2807699810964557,
      "learning_rate": 8.76487022531097e-06,
      "loss": 0.1839,
      "step": 3147
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5415908616585046,
      "learning_rate": 8.764017593492951e-06,
      "loss": 0.2195,
      "step": 3148
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5790375896224849,
      "learning_rate": 8.763164708985026e-06,
      "loss": 0.2339,
      "step": 3149
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.361680637626458,
      "learning_rate": 8.762311571844453e-06,
      "loss": 0.2208,
      "step": 3150
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.346827836075606,
      "learning_rate": 8.761458182128503e-06,
      "loss": 0.2217,
      "step": 3151
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3468208727814432,
      "learning_rate": 8.76060453989447e-06,
      "loss": 0.1544,
      "step": 3152
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4906687463832031,
      "learning_rate": 8.75975064519966e-06,
      "loss": 0.2555,
      "step": 3153
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5512415912968707,
      "learning_rate": 8.758896498101397e-06,
      "loss": 0.2384,
      "step": 3154
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4653316490156025,
      "learning_rate": 8.758042098657022e-06,
      "loss": 0.2101,
      "step": 3155
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0852536519977536,
      "learning_rate": 8.757187446923896e-06,
      "loss": 0.1414,
      "step": 3156
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4128764081558944,
      "learning_rate": 8.756332542959394e-06,
      "loss": 0.2217,
      "step": 3157
    },
    {
      "epoch": 0.25,
      "grad_norm": 17.88489633311662,
      "learning_rate": 8.755477386820906e-06,
      "loss": 0.7123,
      "step": 3158
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.685680913869717,
      "learning_rate": 8.75462197856584e-06,
      "loss": 0.6202,
      "step": 3159
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.222345531876853,
      "learning_rate": 8.753766318251628e-06,
      "loss": 0.2005,
      "step": 3160
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3017713733607912,
      "learning_rate": 8.752910405935708e-06,
      "loss": 0.1999,
      "step": 3161
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.549801876056592,
      "learning_rate": 8.752054241675543e-06,
      "loss": 0.2419,
      "step": 3162
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5763183339308362,
      "learning_rate": 8.751197825528607e-06,
      "loss": 0.2578,
      "step": 3163
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3534450558695814,
      "learning_rate": 8.750341157552396e-06,
      "loss": 0.1649,
      "step": 3164
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3728107862610204,
      "learning_rate": 8.74948423780442e-06,
      "loss": 0.1753,
      "step": 3165
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4813510512455477,
      "learning_rate": 8.748627066342206e-06,
      "loss": 0.1773,
      "step": 3166
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4417247244486429,
      "learning_rate": 8.7477696432233e-06,
      "loss": 0.2012,
      "step": 3167
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.0524160577338835,
      "learning_rate": 8.746911968505262e-06,
      "loss": 0.7107,
      "step": 3168
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3789545658785332,
      "learning_rate": 8.74605404224567e-06,
      "loss": 0.1825,
      "step": 3169
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5069480115976879,
      "learning_rate": 8.745195864502121e-06,
      "loss": 0.2639,
      "step": 3170
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5186721642131027,
      "learning_rate": 8.744337435332226e-06,
      "loss": 0.2471,
      "step": 3171
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.25716461902238,
      "learning_rate": 8.743478754793616e-06,
      "loss": 0.1874,
      "step": 3172
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.234128400739854,
      "learning_rate": 8.742619822943932e-06,
      "loss": 0.5618,
      "step": 3173
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5324637931168346,
      "learning_rate": 8.74176063984084e-06,
      "loss": 0.2442,
      "step": 3174
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3239359974531444,
      "learning_rate": 8.74090120554202e-06,
      "loss": 0.2067,
      "step": 3175
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.538776837609694,
      "learning_rate": 8.740041520105168e-06,
      "loss": 0.2108,
      "step": 3176
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.267553913591337,
      "learning_rate": 8.739181583587997e-06,
      "loss": 0.2147,
      "step": 3177
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.305862617351721,
      "learning_rate": 8.738321396048235e-06,
      "loss": 0.6456,
      "step": 3178
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2922659852864014,
      "learning_rate": 8.737460957543633e-06,
      "loss": 0.1655,
      "step": 3179
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.317768387210999,
      "learning_rate": 8.736600268131953e-06,
      "loss": 0.2115,
      "step": 3180
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2521343704787657,
      "learning_rate": 8.735739327870974e-06,
      "loss": 0.1555,
      "step": 3181
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3399821987123812,
      "learning_rate": 8.734878136818496e-06,
      "loss": 0.1957,
      "step": 3182
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.179258389900639,
      "learning_rate": 8.734016695032333e-06,
      "loss": 0.1981,
      "step": 3183
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3254075134029584,
      "learning_rate": 8.733155002570315e-06,
      "loss": 0.205,
      "step": 3184
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3581819390716958,
      "learning_rate": 8.73229305949029e-06,
      "loss": 0.1856,
      "step": 3185
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3108554332551716,
      "learning_rate": 8.731430865850124e-06,
      "loss": 0.1962,
      "step": 3186
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3900741661961677,
      "learning_rate": 8.730568421707699e-06,
      "loss": 0.2199,
      "step": 3187
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5162537619511862,
      "learning_rate": 8.729705727120911e-06,
      "loss": 0.253,
      "step": 3188
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.509490908177284,
      "learning_rate": 8.728842782147679e-06,
      "loss": 0.2479,
      "step": 3189
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6280437143260313,
      "learning_rate": 8.727979586845931e-06,
      "loss": 0.28,
      "step": 3190
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3007564642704696,
      "learning_rate": 8.727116141273619e-06,
      "loss": 0.1756,
      "step": 3191
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4623809912460453,
      "learning_rate": 8.726252445488708e-06,
      "loss": 0.2224,
      "step": 3192
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5595365044856657,
      "learning_rate": 8.725388499549182e-06,
      "loss": 0.2474,
      "step": 3193
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3048982681056154,
      "learning_rate": 8.724524303513035e-06,
      "loss": 0.2206,
      "step": 3194
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.299657512967296,
      "learning_rate": 8.72365985743829e-06,
      "loss": 0.2047,
      "step": 3195
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2961884266114125,
      "learning_rate": 8.722795161382974e-06,
      "loss": 0.2667,
      "step": 3196
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.394083392692424,
      "learning_rate": 8.72193021540514e-06,
      "loss": 0.2428,
      "step": 3197
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3633135534989351,
      "learning_rate": 8.721065019562854e-06,
      "loss": 0.235,
      "step": 3198
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.359248119005776,
      "learning_rate": 8.720199573914196e-06,
      "loss": 0.2502,
      "step": 3199
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3594908019782461,
      "learning_rate": 8.719333878517274e-06,
      "loss": 0.2312,
      "step": 3200
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3976886597561842,
      "learning_rate": 8.718467933430195e-06,
      "loss": 0.2236,
      "step": 3201
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1994411275995478,
      "learning_rate": 8.7176017387111e-06,
      "loss": 0.1895,
      "step": 3202
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4821448683220442,
      "learning_rate": 8.716735294418136e-06,
      "loss": 0.2107,
      "step": 3203
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5329378979756643,
      "learning_rate": 8.71586860060947e-06,
      "loss": 0.2458,
      "step": 3204
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.417391308935592,
      "learning_rate": 8.715001657343285e-06,
      "loss": 0.2458,
      "step": 3205
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.3636924614899515,
      "learning_rate": 8.714134464677784e-06,
      "loss": 0.6404,
      "step": 3206
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5088706790184852,
      "learning_rate": 8.713267022671182e-06,
      "loss": 0.2895,
      "step": 3207
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4782466248225636,
      "learning_rate": 8.712399331381715e-06,
      "loss": 0.2062,
      "step": 3208
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4091674634684748,
      "learning_rate": 8.711531390867634e-06,
      "loss": 0.2489,
      "step": 3209
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3080483404843495,
      "learning_rate": 8.710663201187203e-06,
      "loss": 0.2044,
      "step": 3210
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.450289452016168,
      "learning_rate": 8.709794762398709e-06,
      "loss": 0.2374,
      "step": 3211
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.649676077644956,
      "learning_rate": 8.708926074560453e-06,
      "loss": 0.5833,
      "step": 3212
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.397211157496579,
      "learning_rate": 8.708057137730752e-06,
      "loss": 0.1881,
      "step": 3213
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.985061637005715,
      "learning_rate": 8.707187951967939e-06,
      "loss": 0.6008,
      "step": 3214
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.535248458843302,
      "learning_rate": 8.706318517330368e-06,
      "loss": 0.2215,
      "step": 3215
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.428511056438544,
      "learning_rate": 8.705448833876404e-06,
      "loss": 0.2083,
      "step": 3216
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.839412785357798,
      "learning_rate": 8.704578901664434e-06,
      "loss": 0.698,
      "step": 3217
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.92139593858278,
      "learning_rate": 8.703708720752857e-06,
      "loss": 0.6828,
      "step": 3218
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5203192792008802,
      "learning_rate": 8.702838291200093e-06,
      "loss": 0.208,
      "step": 3219
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.1366176428658195,
      "learning_rate": 8.701967613064575e-06,
      "loss": 0.5873,
      "step": 3220
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4010690989462742,
      "learning_rate": 8.701096686404753e-06,
      "loss": 0.2426,
      "step": 3221
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5149715350301836,
      "learning_rate": 8.700225511279097e-06,
      "loss": 0.2155,
      "step": 3222
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4854697173396687,
      "learning_rate": 8.699354087746092e-06,
      "loss": 0.2329,
      "step": 3223
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.29676813740157,
      "learning_rate": 8.69848241586424e-06,
      "loss": 0.1519,
      "step": 3224
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.43750592682205,
      "learning_rate": 8.697610495692055e-06,
      "loss": 0.5887,
      "step": 3225
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2839545550256226,
      "learning_rate": 8.696738327288074e-06,
      "loss": 0.2136,
      "step": 3226
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3520771454702623,
      "learning_rate": 8.69586591071085e-06,
      "loss": 0.1927,
      "step": 3227
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.190971887095076,
      "learning_rate": 8.694993246018949e-06,
      "loss": 0.6942,
      "step": 3228
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2547823970784193,
      "learning_rate": 8.694120333270954e-06,
      "loss": 0.1846,
      "step": 3229
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3824993840843107,
      "learning_rate": 8.693247172525472e-06,
      "loss": 0.2233,
      "step": 3230
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4901238254156066,
      "learning_rate": 8.692373763841115e-06,
      "loss": 0.2535,
      "step": 3231
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4575915479558135,
      "learning_rate": 8.691500107276521e-06,
      "loss": 0.2497,
      "step": 3232
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2454481386423009,
      "learning_rate": 8.690626202890338e-06,
      "loss": 0.1967,
      "step": 3233
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2732372990709095,
      "learning_rate": 8.689752050741237e-06,
      "loss": 0.1939,
      "step": 3234
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.324830849361413,
      "learning_rate": 8.6888776508879e-06,
      "loss": 0.2228,
      "step": 3235
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4958712825898532,
      "learning_rate": 8.688003003389032e-06,
      "loss": 0.2566,
      "step": 3236
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3618273852074925,
      "learning_rate": 8.687128108303347e-06,
      "loss": 0.1707,
      "step": 3237
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4122647443594907,
      "learning_rate": 8.686252965689579e-06,
      "loss": 0.2377,
      "step": 3238
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.27874265177543,
      "learning_rate": 8.685377575606481e-06,
      "loss": 0.219,
      "step": 3239
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4321965753032337,
      "learning_rate": 8.684501938112822e-06,
      "loss": 0.2257,
      "step": 3240
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.568292891239202,
      "learning_rate": 8.683626053267385e-06,
      "loss": 0.2205,
      "step": 3241
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5098104800261514,
      "learning_rate": 8.682749921128968e-06,
      "loss": 0.2557,
      "step": 3242
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4799834535525536,
      "learning_rate": 8.68187354175639e-06,
      "loss": 0.235,
      "step": 3243
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3899808174423203,
      "learning_rate": 8.680996915208486e-06,
      "loss": 0.2312,
      "step": 3244
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2958942764425196,
      "learning_rate": 8.680120041544106e-06,
      "loss": 0.19,
      "step": 3245
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.2469686513214775,
      "learning_rate": 8.679242920822119e-06,
      "loss": 0.6509,
      "step": 3246
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4472804927666902,
      "learning_rate": 8.678365553101406e-06,
      "loss": 0.2507,
      "step": 3247
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.613440696545433,
      "learning_rate": 8.677487938440868e-06,
      "loss": 0.8339,
      "step": 3248
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3763756818865895,
      "learning_rate": 8.676610076899423e-06,
      "loss": 0.1891,
      "step": 3249
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.250798261444277,
      "learning_rate": 8.675731968536004e-06,
      "loss": 0.2024,
      "step": 3250
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4927831683983277,
      "learning_rate": 8.67485361340956e-06,
      "loss": 0.2548,
      "step": 3251
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2870034010066054,
      "learning_rate": 8.673975011579058e-06,
      "loss": 0.1894,
      "step": 3252
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3527719379323089,
      "learning_rate": 8.673096163103483e-06,
      "loss": 0.2163,
      "step": 3253
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4974541870822855,
      "learning_rate": 8.672217068041833e-06,
      "loss": 0.2261,
      "step": 3254
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.592232583908154,
      "learning_rate": 8.671337726453126e-06,
      "loss": 0.2372,
      "step": 3255
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.400763616752801,
      "learning_rate": 8.670458138396391e-06,
      "loss": 0.223,
      "step": 3256
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2239321633047564,
      "learning_rate": 8.669578303930684e-06,
      "loss": 0.2193,
      "step": 3257
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.634241103174643,
      "learning_rate": 8.668698223115063e-06,
      "loss": 0.551,
      "step": 3258
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3945335063867428,
      "learning_rate": 8.667817896008618e-06,
      "loss": 0.2073,
      "step": 3259
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.334345418638459,
      "learning_rate": 8.666937322670443e-06,
      "loss": 0.2215,
      "step": 3260
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3719246270184962,
      "learning_rate": 8.666056503159654e-06,
      "loss": 0.2125,
      "step": 3261
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4195729416911402,
      "learning_rate": 8.665175437535387e-06,
      "loss": 0.2641,
      "step": 3262
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4270259610696958,
      "learning_rate": 8.664294125856786e-06,
      "loss": 0.2388,
      "step": 3263
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.918064465575382,
      "learning_rate": 8.663412568183018e-06,
      "loss": 0.5098,
      "step": 3264
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3573804030741383,
      "learning_rate": 8.662530764573264e-06,
      "loss": 0.2267,
      "step": 3265
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.43384358147485,
      "learning_rate": 8.661648715086725e-06,
      "loss": 0.213,
      "step": 3266
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4165597620522519,
      "learning_rate": 8.66076641978261e-06,
      "loss": 0.1996,
      "step": 3267
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.224044680424154,
      "learning_rate": 8.659883878720158e-06,
      "loss": 0.1864,
      "step": 3268
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5335445345848624,
      "learning_rate": 8.65900109195861e-06,
      "loss": 0.2451,
      "step": 3269
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5067058038654433,
      "learning_rate": 8.658118059557233e-06,
      "loss": 0.1939,
      "step": 3270
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4186756181655202,
      "learning_rate": 8.657234781575305e-06,
      "loss": 0.245,
      "step": 3271
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.12107141288473,
      "learning_rate": 8.656351258072126e-06,
      "loss": 0.1781,
      "step": 3272
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3783289562807386,
      "learning_rate": 8.65546748910701e-06,
      "loss": 0.2305,
      "step": 3273
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5891777073731126,
      "learning_rate": 8.654583474739284e-06,
      "loss": 0.24,
      "step": 3274
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3931356556078667,
      "learning_rate": 8.653699215028298e-06,
      "loss": 0.2132,
      "step": 3275
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.38622832440785,
      "learning_rate": 8.652814710033413e-06,
      "loss": 0.1984,
      "step": 3276
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.414963206934076,
      "learning_rate": 8.651929959814009e-06,
      "loss": 0.2438,
      "step": 3277
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.227173360929713,
      "learning_rate": 8.65104496442948e-06,
      "loss": 0.202,
      "step": 3278
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.695814781365407,
      "learning_rate": 8.650159723939241e-06,
      "loss": 0.6169,
      "step": 3279
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4872483855349818,
      "learning_rate": 8.649274238402723e-06,
      "loss": 0.2199,
      "step": 3280
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5501930447321255,
      "learning_rate": 8.648388507879367e-06,
      "loss": 0.2207,
      "step": 3281
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5059417901422238,
      "learning_rate": 8.647502532428636e-06,
      "loss": 0.2149,
      "step": 3282
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.41956730295782,
      "learning_rate": 8.646616312110008e-06,
      "loss": 0.4973,
      "step": 3283
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.163982146026823,
      "learning_rate": 8.64572984698298e-06,
      "loss": 0.4652,
      "step": 3284
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2477808648153987,
      "learning_rate": 8.644843137107058e-06,
      "loss": 0.1771,
      "step": 3285
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.245524489759415,
      "learning_rate": 8.643956182541775e-06,
      "loss": 0.1906,
      "step": 3286
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.423737742889783,
      "learning_rate": 8.643068983346672e-06,
      "loss": 0.2435,
      "step": 3287
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.392556203965379,
      "learning_rate": 8.64218153958131e-06,
      "loss": 0.2354,
      "step": 3288
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3092858580540458,
      "learning_rate": 8.641293851305269e-06,
      "loss": 0.2235,
      "step": 3289
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3974007528559544,
      "learning_rate": 8.640405918578134e-06,
      "loss": 0.2297,
      "step": 3290
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.440515656995989,
      "learning_rate": 8.639517741459522e-06,
      "loss": 0.2206,
      "step": 3291
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.384342029141178,
      "learning_rate": 8.638629320009058e-06,
      "loss": 0.219,
      "step": 3292
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5994519012809414,
      "learning_rate": 8.637740654286382e-06,
      "loss": 0.2071,
      "step": 3293
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3160360288749673,
      "learning_rate": 8.636851744351154e-06,
      "loss": 0.1902,
      "step": 3294
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3152805292253702,
      "learning_rate": 8.635962590263047e-06,
      "loss": 0.2375,
      "step": 3295
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4591324320680432,
      "learning_rate": 8.635073192081757e-06,
      "loss": 0.2089,
      "step": 3296
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.486571901267544,
      "learning_rate": 8.63418354986699e-06,
      "loss": 0.2268,
      "step": 3297
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.502231562836528,
      "learning_rate": 8.63329366367847e-06,
      "loss": 0.2065,
      "step": 3298
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3750469036192092,
      "learning_rate": 8.632403533575939e-06,
      "loss": 0.2068,
      "step": 3299
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.340091720239436,
      "learning_rate": 8.63151315961915e-06,
      "loss": 0.1986,
      "step": 3300
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4287659176814846,
      "learning_rate": 8.630622541867881e-06,
      "loss": 0.2083,
      "step": 3301
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2741602949274455,
      "learning_rate": 8.629731680381922e-06,
      "loss": 0.1877,
      "step": 3302
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.392968593165019,
      "learning_rate": 8.628840575221076e-06,
      "loss": 0.2211,
      "step": 3303
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3948138917276385,
      "learning_rate": 8.627949226445166e-06,
      "loss": 0.2276,
      "step": 3304
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3767623543680214,
      "learning_rate": 8.627057634114036e-06,
      "loss": 0.2281,
      "step": 3305
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.905700854127078,
      "learning_rate": 8.626165798287534e-06,
      "loss": 0.5206,
      "step": 3306
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.60099476138202,
      "learning_rate": 8.625273719025538e-06,
      "loss": 0.2849,
      "step": 3307
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.474792122507455,
      "learning_rate": 8.62438139638793e-06,
      "loss": 0.2413,
      "step": 3308
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.603070691935612,
      "learning_rate": 8.623488830434619e-06,
      "loss": 0.2436,
      "step": 3309
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5154523071098147,
      "learning_rate": 8.622596021225524e-06,
      "loss": 0.2215,
      "step": 3310
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3714916702675344,
      "learning_rate": 8.621702968820582e-06,
      "loss": 0.2175,
      "step": 3311
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.845900021469104,
      "learning_rate": 8.620809673279746e-06,
      "loss": 0.5095,
      "step": 3312
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3717316383161107,
      "learning_rate": 8.619916134662986e-06,
      "loss": 0.1835,
      "step": 3313
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3760865241075366,
      "learning_rate": 8.619022353030288e-06,
      "loss": 0.2346,
      "step": 3314
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2612030683693312,
      "learning_rate": 8.618128328441655e-06,
      "loss": 0.1943,
      "step": 3315
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4474170736837229,
      "learning_rate": 8.617234060957105e-06,
      "loss": 0.205,
      "step": 3316
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4672323025300713,
      "learning_rate": 8.616339550636672e-06,
      "loss": 0.2528,
      "step": 3317
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.404668787934941,
      "learning_rate": 8.615444797540408e-06,
      "loss": 0.1926,
      "step": 3318
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2932740104839244,
      "learning_rate": 8.61454980172838e-06,
      "loss": 0.1799,
      "step": 3319
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4199019012557053,
      "learning_rate": 8.613654563260673e-06,
      "loss": 0.2321,
      "step": 3320
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4417218287588973,
      "learning_rate": 8.612759082197386e-06,
      "loss": 0.2296,
      "step": 3321
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.381046292822886,
      "learning_rate": 8.611863358598635e-06,
      "loss": 0.2136,
      "step": 3322
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4630894314542768,
      "learning_rate": 8.610967392524554e-06,
      "loss": 0.221,
      "step": 3323
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4330271561023198,
      "learning_rate": 8.610071184035292e-06,
      "loss": 0.225,
      "step": 3324
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4096339550465822,
      "learning_rate": 8.609174733191012e-06,
      "loss": 0.2309,
      "step": 3325
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.677255327444758,
      "learning_rate": 8.608278040051895e-06,
      "loss": 0.6186,
      "step": 3326
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.725038603019051,
      "learning_rate": 8.607381104678142e-06,
      "loss": 0.6055,
      "step": 3327
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6543551893989619,
      "learning_rate": 8.606483927129967e-06,
      "loss": 0.2506,
      "step": 3328
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4641475236957175,
      "learning_rate": 8.605586507467597e-06,
      "loss": 0.2462,
      "step": 3329
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.271741681012463,
      "learning_rate": 8.604688845751283e-06,
      "loss": 0.2333,
      "step": 3330
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2944789322117929,
      "learning_rate": 8.603790942041282e-06,
      "loss": 0.2062,
      "step": 3331
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2496626638288242,
      "learning_rate": 8.602892796397877e-06,
      "loss": 0.2301,
      "step": 3332
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3839262639360257,
      "learning_rate": 8.601994408881362e-06,
      "loss": 0.1944,
      "step": 3333
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.960056264056906,
      "learning_rate": 8.60109577955205e-06,
      "loss": 0.5522,
      "step": 3334
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4845123372797735,
      "learning_rate": 8.600196908470265e-06,
      "loss": 0.2525,
      "step": 3335
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4904222270857068,
      "learning_rate": 8.599297795696355e-06,
      "loss": 0.2868,
      "step": 3336
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4336924661490809,
      "learning_rate": 8.598398441290679e-06,
      "loss": 0.2235,
      "step": 3337
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2862098601757985,
      "learning_rate": 8.597498845313612e-06,
      "loss": 0.205,
      "step": 3338
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3756513263966703,
      "learning_rate": 8.596599007825548e-06,
      "loss": 0.2204,
      "step": 3339
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3290166602454174,
      "learning_rate": 8.595698928886894e-06,
      "loss": 0.2541,
      "step": 3340
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3623803438445348,
      "learning_rate": 8.594798608558077e-06,
      "loss": 0.2251,
      "step": 3341
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.523019867969133,
      "learning_rate": 8.593898046899539e-06,
      "loss": 0.2287,
      "step": 3342
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.413000962049789,
      "learning_rate": 8.592997243971734e-06,
      "loss": 0.2034,
      "step": 3343
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4159815233966042,
      "learning_rate": 8.592096199835141e-06,
      "loss": 0.201,
      "step": 3344
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4366309363516205,
      "learning_rate": 8.591194914550242e-06,
      "loss": 0.178,
      "step": 3345
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.234425552683678,
      "learning_rate": 8.590293388177552e-06,
      "loss": 0.6904,
      "step": 3346
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5826727940453578,
      "learning_rate": 8.589391620777586e-06,
      "loss": 0.2201,
      "step": 3347
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4249936449054619,
      "learning_rate": 8.588489612410885e-06,
      "loss": 0.2393,
      "step": 3348
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.761381874608512,
      "learning_rate": 8.587587363138003e-06,
      "loss": 0.4948,
      "step": 3349
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.355617614554123,
      "learning_rate": 8.586684873019513e-06,
      "loss": 0.2216,
      "step": 3350
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.438418273720014,
      "learning_rate": 8.585782142116e-06,
      "loss": 0.255,
      "step": 3351
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.419152071624994,
      "learning_rate": 8.584879170488068e-06,
      "loss": 0.2129,
      "step": 3352
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4212169669942178,
      "learning_rate": 8.583975958196333e-06,
      "loss": 0.2333,
      "step": 3353
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5804855519775207,
      "learning_rate": 8.583072505301436e-06,
      "loss": 0.2425,
      "step": 3354
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2835645409457161,
      "learning_rate": 8.582168811864022e-06,
      "loss": 0.2021,
      "step": 3355
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2681078675620716,
      "learning_rate": 8.581264877944764e-06,
      "loss": 0.1902,
      "step": 3356
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8395633418399882,
      "learning_rate": 8.580360703604344e-06,
      "loss": 0.2469,
      "step": 3357
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.389459556585165,
      "learning_rate": 8.579456288903463e-06,
      "loss": 0.2617,
      "step": 3358
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.420210456918626,
      "learning_rate": 8.578551633902836e-06,
      "loss": 0.1819,
      "step": 3359
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3741896065552794,
      "learning_rate": 8.577646738663193e-06,
      "loss": 0.1787,
      "step": 3360
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3429191571009946,
      "learning_rate": 8.576741603245285e-06,
      "loss": 0.2267,
      "step": 3361
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3266588368008274,
      "learning_rate": 8.575836227709877e-06,
      "loss": 0.2054,
      "step": 3362
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2665890815488017,
      "learning_rate": 8.574930612117748e-06,
      "loss": 0.1812,
      "step": 3363
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3263880385488571,
      "learning_rate": 8.574024756529698e-06,
      "loss": 0.1919,
      "step": 3364
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5457286664498076,
      "learning_rate": 8.573118661006535e-06,
      "loss": 0.2355,
      "step": 3365
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.337918393069023,
      "learning_rate": 8.572212325609089e-06,
      "loss": 0.2198,
      "step": 3366
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2859309581970662,
      "learning_rate": 8.571305750398207e-06,
      "loss": 0.1786,
      "step": 3367
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1747443336171957,
      "learning_rate": 8.57039893543475e-06,
      "loss": 0.1778,
      "step": 3368
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.152163072155325,
      "learning_rate": 8.569491880779594e-06,
      "loss": 0.5469,
      "step": 3369
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4790229884346278,
      "learning_rate": 8.568584586493635e-06,
      "loss": 0.2172,
      "step": 3370
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4516696240568987,
      "learning_rate": 8.567677052637778e-06,
      "loss": 0.2453,
      "step": 3371
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3448540211877822,
      "learning_rate": 8.566769279272952e-06,
      "loss": 0.2307,
      "step": 3372
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.476500251821296,
      "learning_rate": 8.565861266460096e-06,
      "loss": 0.6898,
      "step": 3373
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4391748969000988,
      "learning_rate": 8.56495301426017e-06,
      "loss": 0.197,
      "step": 3374
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.911363134126704,
      "learning_rate": 8.564044522734147e-06,
      "loss": 0.6137,
      "step": 3375
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.500628730199422,
      "learning_rate": 8.563135791943015e-06,
      "loss": 0.2276,
      "step": 3376
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3924291241373836,
      "learning_rate": 8.562226821947784e-06,
      "loss": 0.2342,
      "step": 3377
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4233917667748524,
      "learning_rate": 8.561317612809473e-06,
      "loss": 0.2018,
      "step": 3378
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4540760032553721,
      "learning_rate": 8.560408164589117e-06,
      "loss": 0.1726,
      "step": 3379
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5823740699067297,
      "learning_rate": 8.559498477347777e-06,
      "loss": 0.2246,
      "step": 3380
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.708874900183982,
      "learning_rate": 8.558588551146519e-06,
      "loss": 0.6208,
      "step": 3381
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4483844586385217,
      "learning_rate": 8.557678386046429e-06,
      "loss": 0.2343,
      "step": 3382
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3770788834879915,
      "learning_rate": 8.556767982108607e-06,
      "loss": 0.182,
      "step": 3383
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2930451226061221,
      "learning_rate": 8.555857339394175e-06,
      "loss": 0.1766,
      "step": 3384
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.475730834805216,
      "learning_rate": 8.554946457964268e-06,
      "loss": 0.2194,
      "step": 3385
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.843550203319319,
      "learning_rate": 8.554035337880034e-06,
      "loss": 0.605,
      "step": 3386
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4479539473973146,
      "learning_rate": 8.553123979202638e-06,
      "loss": 0.2207,
      "step": 3387
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4883021518656074,
      "learning_rate": 8.552212381993264e-06,
      "loss": 0.219,
      "step": 3388
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3753707166415994,
      "learning_rate": 8.55130054631311e-06,
      "loss": 0.207,
      "step": 3389
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.564812415267844,
      "learning_rate": 8.550388472223391e-06,
      "loss": 0.2682,
      "step": 3390
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5335671407695977,
      "learning_rate": 8.549476159785336e-06,
      "loss": 0.2075,
      "step": 3391
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.306855644100196,
      "learning_rate": 8.548563609060191e-06,
      "loss": 0.1748,
      "step": 3392
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.377123622988194,
      "learning_rate": 8.547650820109222e-06,
      "loss": 0.2418,
      "step": 3393
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3783118718988048,
      "learning_rate": 8.546737792993702e-06,
      "loss": 0.1888,
      "step": 3394
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3861471381945925,
      "learning_rate": 8.54582452777493e-06,
      "loss": 0.1702,
      "step": 3395
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.529905020367175,
      "learning_rate": 8.544911024514211e-06,
      "loss": 0.2316,
      "step": 3396
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4085107236134933,
      "learning_rate": 8.543997283272876e-06,
      "loss": 0.2133,
      "step": 3397
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.293103098033083,
      "learning_rate": 8.543083304112265e-06,
      "loss": 0.7167,
      "step": 3398
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3188822304384575,
      "learning_rate": 8.542169087093738e-06,
      "loss": 0.1921,
      "step": 3399
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.305343679466047,
      "learning_rate": 8.541254632278667e-06,
      "loss": 0.7709,
      "step": 3400
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4854512425162207,
      "learning_rate": 8.540339939728442e-06,
      "loss": 0.2361,
      "step": 3401
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4234052592768027,
      "learning_rate": 8.539425009504472e-06,
      "loss": 0.2235,
      "step": 3402
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.479243328820142,
      "learning_rate": 8.538509841668175e-06,
      "loss": 0.203,
      "step": 3403
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4166251003820498,
      "learning_rate": 8.537594436280992e-06,
      "loss": 0.2301,
      "step": 3404
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5124227499727871,
      "learning_rate": 8.536678793404376e-06,
      "loss": 0.2501,
      "step": 3405
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4816169682267926,
      "learning_rate": 8.535762913099798e-06,
      "loss": 0.2128,
      "step": 3406
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.597090458228408,
      "learning_rate": 8.534846795428741e-06,
      "loss": 0.2825,
      "step": 3407
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.285115107151196,
      "learning_rate": 8.53393044045271e-06,
      "loss": 0.1923,
      "step": 3408
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3830473688227631,
      "learning_rate": 8.533013848233218e-06,
      "loss": 0.2099,
      "step": 3409
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.457934509785978,
      "learning_rate": 8.532097018831805e-06,
      "loss": 0.2141,
      "step": 3410
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5704413315177805,
      "learning_rate": 8.531179952310016e-06,
      "loss": 0.2453,
      "step": 3411
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4384697804738853,
      "learning_rate": 8.530262648729419e-06,
      "loss": 0.2234,
      "step": 3412
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.262343225066612,
      "learning_rate": 8.529345108151594e-06,
      "loss": 0.1781,
      "step": 3413
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2904855478551327,
      "learning_rate": 8.528427330638137e-06,
      "loss": 0.2577,
      "step": 3414
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4385361437785058,
      "learning_rate": 8.527509316250663e-06,
      "loss": 0.1713,
      "step": 3415
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5522998355833384,
      "learning_rate": 8.5265910650508e-06,
      "loss": 0.2217,
      "step": 3416
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2393137561563283,
      "learning_rate": 8.525672577100195e-06,
      "loss": 0.1701,
      "step": 3417
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3029039763986576,
      "learning_rate": 8.524753852460507e-06,
      "loss": 0.1799,
      "step": 3418
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3139639179579707,
      "learning_rate": 8.523834891193412e-06,
      "loss": 0.2001,
      "step": 3419
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2268069778300748,
      "learning_rate": 8.522915693360607e-06,
      "loss": 0.1682,
      "step": 3420
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3268490906493136,
      "learning_rate": 8.521996259023795e-06,
      "loss": 0.1576,
      "step": 3421
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1152832258383347,
      "learning_rate": 8.521076588244705e-06,
      "loss": 0.1447,
      "step": 3422
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.354401385770746,
      "learning_rate": 8.520156681085073e-06,
      "loss": 0.229,
      "step": 3423
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4277168865077374,
      "learning_rate": 8.519236537606658e-06,
      "loss": 0.2054,
      "step": 3424
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6231363602590048,
      "learning_rate": 8.518316157871232e-06,
      "loss": 0.2951,
      "step": 3425
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2673811426144943,
      "learning_rate": 8.517395541940582e-06,
      "loss": 0.1918,
      "step": 3426
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.56496349195988,
      "learning_rate": 8.516474689876512e-06,
      "loss": 0.4743,
      "step": 3427
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.384655440165043,
      "learning_rate": 8.51555360174084e-06,
      "loss": 0.2152,
      "step": 3428
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4811281676721388,
      "learning_rate": 8.514632277595405e-06,
      "loss": 0.24,
      "step": 3429
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.8618187422190005,
      "learning_rate": 8.513710717502057e-06,
      "loss": 0.6745,
      "step": 3430
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4836628821142488,
      "learning_rate": 8.51278892152266e-06,
      "loss": 0.2427,
      "step": 3431
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.346042823960647,
      "learning_rate": 8.5118668897191e-06,
      "loss": 0.1746,
      "step": 3432
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3900841993862985,
      "learning_rate": 8.510944622153276e-06,
      "loss": 0.2242,
      "step": 3433
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3842055573971987,
      "learning_rate": 8.510022118887102e-06,
      "loss": 0.2107,
      "step": 3434
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.521602464599794,
      "learning_rate": 8.509099379982509e-06,
      "loss": 0.4941,
      "step": 3435
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5441403836233059,
      "learning_rate": 8.50817640550144e-06,
      "loss": 0.1896,
      "step": 3436
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3526867369859892,
      "learning_rate": 8.50725319550586e-06,
      "loss": 0.2127,
      "step": 3437
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3006985648901674,
      "learning_rate": 8.506329750057747e-06,
      "loss": 0.1805,
      "step": 3438
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3444669617533247,
      "learning_rate": 8.505406069219095e-06,
      "loss": 0.2159,
      "step": 3439
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4643151722564194,
      "learning_rate": 8.504482153051912e-06,
      "loss": 0.2355,
      "step": 3440
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3136783778631522,
      "learning_rate": 8.503558001618224e-06,
      "loss": 0.2046,
      "step": 3441
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4819907086055457,
      "learning_rate": 8.502633614980071e-06,
      "loss": 0.2213,
      "step": 3442
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4795731060927866,
      "learning_rate": 8.501708993199511e-06,
      "loss": 0.2312,
      "step": 3443
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4472907899262863,
      "learning_rate": 8.500784136338618e-06,
      "loss": 0.2787,
      "step": 3444
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.30392472860371,
      "learning_rate": 8.499859044459478e-06,
      "loss": 0.2431,
      "step": 3445
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3472270330454186,
      "learning_rate": 8.498933717624197e-06,
      "loss": 0.1774,
      "step": 3446
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.443513622475838,
      "learning_rate": 8.498008155894892e-06,
      "loss": 0.2344,
      "step": 3447
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4144248404312476,
      "learning_rate": 8.497082359333703e-06,
      "loss": 0.2919,
      "step": 3448
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.904353366374629,
      "learning_rate": 8.49615632800278e-06,
      "loss": 0.7658,
      "step": 3449
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5023091555648453,
      "learning_rate": 8.495230061964289e-06,
      "loss": 0.237,
      "step": 3450
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2514074271087992,
      "learning_rate": 8.494303561280412e-06,
      "loss": 0.2078,
      "step": 3451
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4294497086902498,
      "learning_rate": 8.49337682601335e-06,
      "loss": 0.2412,
      "step": 3452
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.400063897613412,
      "learning_rate": 8.49244985622532e-06,
      "loss": 0.2715,
      "step": 3453
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4086700584117429,
      "learning_rate": 8.491522651978548e-06,
      "loss": 0.243,
      "step": 3454
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4266615227254944,
      "learning_rate": 8.49059521333528e-06,
      "loss": 0.2628,
      "step": 3455
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3772307678633362,
      "learning_rate": 8.489667540357781e-06,
      "loss": 0.2102,
      "step": 3456
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.278040474668398,
      "learning_rate": 8.488739633108328e-06,
      "loss": 0.2149,
      "step": 3457
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4619945053111885,
      "learning_rate": 8.487811491649209e-06,
      "loss": 0.2228,
      "step": 3458
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3253640656608443,
      "learning_rate": 8.48688311604274e-06,
      "loss": 0.1673,
      "step": 3459
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.605505927493333,
      "learning_rate": 8.485954506351241e-06,
      "loss": 0.6058,
      "step": 3460
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3320851338773465,
      "learning_rate": 8.485025662637055e-06,
      "loss": 0.2065,
      "step": 3461
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3683581897303136,
      "learning_rate": 8.484096584962537e-06,
      "loss": 0.2064,
      "step": 3462
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.291190066384424,
      "learning_rate": 8.483167273390058e-06,
      "loss": 0.1986,
      "step": 3463
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.206622654362899,
      "learning_rate": 8.482237727982007e-06,
      "loss": 0.2033,
      "step": 3464
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5512908714215459,
      "learning_rate": 8.481307948800787e-06,
      "loss": 0.2392,
      "step": 3465
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5434148859438048,
      "learning_rate": 8.480377935908817e-06,
      "loss": 0.2308,
      "step": 3466
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3367491214929064,
      "learning_rate": 8.479447689368529e-06,
      "loss": 0.1787,
      "step": 3467
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2605416557649196,
      "learning_rate": 8.478517209242376e-06,
      "loss": 0.185,
      "step": 3468
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4204735594366522,
      "learning_rate": 8.477586495592823e-06,
      "loss": 0.2498,
      "step": 3469
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2940385014474238,
      "learning_rate": 8.476655548482353e-06,
      "loss": 0.2088,
      "step": 3470
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3590296422013841,
      "learning_rate": 8.475724367973462e-06,
      "loss": 0.2211,
      "step": 3471
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.881278190700596,
      "learning_rate": 8.474792954128661e-06,
      "loss": 0.5251,
      "step": 3472
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2956080157986372,
      "learning_rate": 8.473861307010484e-06,
      "loss": 0.1907,
      "step": 3473
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3374040318634932,
      "learning_rate": 8.47292942668147e-06,
      "loss": 0.1892,
      "step": 3474
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5164451403399195,
      "learning_rate": 8.471997313204183e-06,
      "loss": 0.2639,
      "step": 3475
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2832037661818128,
      "learning_rate": 8.471064966641193e-06,
      "loss": 0.2228,
      "step": 3476
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5984064652252163,
      "learning_rate": 8.470132387055098e-06,
      "loss": 0.2604,
      "step": 3477
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.437946545289577,
      "learning_rate": 8.469199574508502e-06,
      "loss": 0.2286,
      "step": 3478
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3509247943561975,
      "learning_rate": 8.468266529064025e-06,
      "loss": 0.2189,
      "step": 3479
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.507517567796907,
      "learning_rate": 8.467333250784309e-06,
      "loss": 0.2279,
      "step": 3480
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.316622863669533,
      "learning_rate": 8.466399739732006e-06,
      "loss": 0.1977,
      "step": 3481
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.088557709046349,
      "learning_rate": 8.465465995969786e-06,
      "loss": 0.6017,
      "step": 3482
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5641777773018952,
      "learning_rate": 8.464532019560335e-06,
      "loss": 0.2291,
      "step": 3483
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5310945209829467,
      "learning_rate": 8.463597810566349e-06,
      "loss": 0.212,
      "step": 3484
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.72140498309021,
      "learning_rate": 8.46266336905055e-06,
      "loss": 0.6118,
      "step": 3485
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.364825616409186,
      "learning_rate": 8.461728695075667e-06,
      "loss": 0.1985,
      "step": 3486
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4458929057102348,
      "learning_rate": 8.460793788704449e-06,
      "loss": 0.2447,
      "step": 3487
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4119090971058033,
      "learning_rate": 8.459858649999657e-06,
      "loss": 0.2056,
      "step": 3488
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4737843444468024,
      "learning_rate": 8.458923279024072e-06,
      "loss": 0.2616,
      "step": 3489
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2897049765546598,
      "learning_rate": 8.457987675840484e-06,
      "loss": 0.1942,
      "step": 3490
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.32461648629269,
      "learning_rate": 8.45705184051171e-06,
      "loss": 0.2071,
      "step": 3491
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5500957466339325,
      "learning_rate": 8.45611577310057e-06,
      "loss": 0.2524,
      "step": 3492
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3673654279746756,
      "learning_rate": 8.455179473669905e-06,
      "loss": 0.2227,
      "step": 3493
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4310744505165995,
      "learning_rate": 8.454242942282576e-06,
      "loss": 0.1618,
      "step": 3494
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.752270089548542,
      "learning_rate": 8.45330617900145e-06,
      "loss": 0.6575,
      "step": 3495
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4296196779878696,
      "learning_rate": 8.452369183889419e-06,
      "loss": 0.2503,
      "step": 3496
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.356048869214277,
      "learning_rate": 8.451431957009384e-06,
      "loss": 0.1914,
      "step": 3497
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4793535135672018,
      "learning_rate": 8.450494498424263e-06,
      "loss": 0.2223,
      "step": 3498
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1105474716643329,
      "learning_rate": 8.449556808196992e-06,
      "loss": 0.1364,
      "step": 3499
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.260996978187563,
      "learning_rate": 8.448618886390523e-06,
      "loss": 0.6377,
      "step": 3500
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4290444155480104,
      "learning_rate": 8.447680733067816e-06,
      "loss": 0.2143,
      "step": 3501
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3916454864739236,
      "learning_rate": 8.446742348291857e-06,
      "loss": 0.2415,
      "step": 3502
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.377311538756095,
      "learning_rate": 8.445803732125641e-06,
      "loss": 0.1977,
      "step": 3503
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3803536348826295,
      "learning_rate": 8.44486488463218e-06,
      "loss": 0.2433,
      "step": 3504
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3594423362030763,
      "learning_rate": 8.443925805874502e-06,
      "loss": 0.216,
      "step": 3505
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.601575872708146,
      "learning_rate": 8.44298649591565e-06,
      "loss": 0.6162,
      "step": 3506
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4177905703529006,
      "learning_rate": 8.442046954818682e-06,
      "loss": 0.2009,
      "step": 3507
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5388420691266567,
      "learning_rate": 8.441107182646674e-06,
      "loss": 0.257,
      "step": 3508
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4269021335765721,
      "learning_rate": 8.440167179462713e-06,
      "loss": 0.219,
      "step": 3509
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.43282075620933,
      "learning_rate": 8.439226945329908e-06,
      "loss": 0.2231,
      "step": 3510
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7349687841648163,
      "learning_rate": 8.438286480311375e-06,
      "loss": 0.3251,
      "step": 3511
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4967416340105482,
      "learning_rate": 8.437345784470255e-06,
      "loss": 0.24,
      "step": 3512
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5108032644382643,
      "learning_rate": 8.436404857869697e-06,
      "loss": 0.2033,
      "step": 3513
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3491116587272378,
      "learning_rate": 8.43546370057287e-06,
      "loss": 0.2242,
      "step": 3514
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3380681770008118,
      "learning_rate": 8.434522312642955e-06,
      "loss": 0.2447,
      "step": 3515
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5057853560513264,
      "learning_rate": 8.433580694143152e-06,
      "loss": 0.248,
      "step": 3516
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.872842745170944,
      "learning_rate": 8.432638845136671e-06,
      "loss": 0.65,
      "step": 3517
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3743038220856165,
      "learning_rate": 8.431696765686746e-06,
      "loss": 0.2137,
      "step": 3518
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.162384675673378,
      "learning_rate": 8.430754455856618e-06,
      "loss": 0.1899,
      "step": 3519
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.632299215587384,
      "learning_rate": 8.42981191570955e-06,
      "loss": 0.4724,
      "step": 3520
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3684654691444889,
      "learning_rate": 8.428869145308817e-06,
      "loss": 0.2446,
      "step": 3521
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5727901526492336,
      "learning_rate": 8.427926144717707e-06,
      "loss": 0.2583,
      "step": 3522
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.9526626926514785,
      "learning_rate": 8.426982913999531e-06,
      "loss": 0.5841,
      "step": 3523
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3820512686571191,
      "learning_rate": 8.426039453217606e-06,
      "loss": 0.2331,
      "step": 3524
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3234939687259617,
      "learning_rate": 8.425095762435274e-06,
      "loss": 0.1872,
      "step": 3525
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2023558747677952,
      "learning_rate": 8.424151841715885e-06,
      "loss": 0.1341,
      "step": 3526
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3970676847709804,
      "learning_rate": 8.423207691122808e-06,
      "loss": 0.1943,
      "step": 3527
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3012580623644552,
      "learning_rate": 8.422263310719428e-06,
      "loss": 0.2298,
      "step": 3528
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.390800120561444,
      "learning_rate": 8.421318700569142e-06,
      "loss": 0.1861,
      "step": 3529
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4388607884217528,
      "learning_rate": 8.420373860735366e-06,
      "loss": 0.1829,
      "step": 3530
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.402792161876903,
      "learning_rate": 8.419428791281528e-06,
      "loss": 0.6133,
      "step": 3531
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5024921698480707,
      "learning_rate": 8.418483492271077e-06,
      "loss": 0.2466,
      "step": 3532
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.114020113468879,
      "learning_rate": 8.417537963767469e-06,
      "loss": 0.1565,
      "step": 3533
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.443472378494259,
      "learning_rate": 8.416592205834185e-06,
      "loss": 0.2148,
      "step": 3534
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2028106971309198,
      "learning_rate": 8.415646218534713e-06,
      "loss": 0.1697,
      "step": 3535
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3894622733695372,
      "learning_rate": 8.41470000193256e-06,
      "loss": 0.2133,
      "step": 3536
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5286291867068054,
      "learning_rate": 8.413753556091254e-06,
      "loss": 0.2355,
      "step": 3537
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2642081670084637,
      "learning_rate": 8.412806881074325e-06,
      "loss": 0.1874,
      "step": 3538
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.147546051377303,
      "learning_rate": 8.41185997694533e-06,
      "loss": 0.5358,
      "step": 3539
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.417027916612012,
      "learning_rate": 8.410912843767837e-06,
      "loss": 0.2353,
      "step": 3540
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4790168596047897,
      "learning_rate": 8.409965481605432e-06,
      "loss": 0.213,
      "step": 3541
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3235210719830932,
      "learning_rate": 8.409017890521711e-06,
      "loss": 0.1862,
      "step": 3542
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4680285760893097,
      "learning_rate": 8.40807007058029e-06,
      "loss": 0.2261,
      "step": 3543
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.931338706852394,
      "learning_rate": 8.407122021844802e-06,
      "loss": 0.6239,
      "step": 3544
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.458212901951154,
      "learning_rate": 8.406173744378887e-06,
      "loss": 0.1877,
      "step": 3545
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2961430426800935,
      "learning_rate": 8.405225238246208e-06,
      "loss": 0.1623,
      "step": 3546
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4483675940464704,
      "learning_rate": 8.404276503510443e-06,
      "loss": 0.227,
      "step": 3547
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3631285114069418,
      "learning_rate": 8.40332754023528e-06,
      "loss": 0.2195,
      "step": 3548
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1651284862412983,
      "learning_rate": 8.402378348484427e-06,
      "loss": 0.154,
      "step": 3549
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.990265354834624,
      "learning_rate": 8.401428928321607e-06,
      "loss": 0.6829,
      "step": 3550
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4117491347143682,
      "learning_rate": 8.400479279810559e-06,
      "loss": 0.1837,
      "step": 3551
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3673114271466706,
      "learning_rate": 8.399529403015031e-06,
      "loss": 0.1915,
      "step": 3552
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3317961161774117,
      "learning_rate": 8.398579297998798e-06,
      "loss": 0.1553,
      "step": 3553
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2843120715717005,
      "learning_rate": 8.397628964825637e-06,
      "loss": 0.1837,
      "step": 3554
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.345890360919074,
      "learning_rate": 8.396678403559348e-06,
      "loss": 0.6522,
      "step": 3555
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2782432092336076,
      "learning_rate": 8.395727614263749e-06,
      "loss": 0.1931,
      "step": 3556
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.273421143340095,
      "learning_rate": 8.394776597002666e-06,
      "loss": 0.1833,
      "step": 3557
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.991776810442344,
      "learning_rate": 8.393825351839942e-06,
      "loss": 0.5452,
      "step": 3558
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2994101098371522,
      "learning_rate": 8.392873878839441e-06,
      "loss": 0.1826,
      "step": 3559
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2754644429793767,
      "learning_rate": 8.391922178065037e-06,
      "loss": 0.1934,
      "step": 3560
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3071973979309786,
      "learning_rate": 8.390970249580619e-06,
      "loss": 0.1894,
      "step": 3561
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3061638562278641,
      "learning_rate": 8.390018093450094e-06,
      "loss": 0.1679,
      "step": 3562
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5446246473592506,
      "learning_rate": 8.389065709737384e-06,
      "loss": 0.2652,
      "step": 3563
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3650648397202592,
      "learning_rate": 8.388113098506422e-06,
      "loss": 0.2014,
      "step": 3564
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7396085664091716,
      "learning_rate": 8.387160259821165e-06,
      "loss": 0.2882,
      "step": 3565
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3332989805657023,
      "learning_rate": 8.386207193745577e-06,
      "loss": 0.203,
      "step": 3566
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.370976767771929,
      "learning_rate": 8.385253900343638e-06,
      "loss": 0.2428,
      "step": 3567
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4026268781131952,
      "learning_rate": 8.38430037967935e-06,
      "loss": 0.1929,
      "step": 3568
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.315817361523236,
      "learning_rate": 8.383346631816723e-06,
      "loss": 0.2324,
      "step": 3569
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2044036728363916,
      "learning_rate": 8.382392656819784e-06,
      "loss": 0.1854,
      "step": 3570
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4770274346571084,
      "learning_rate": 8.38143845475258e-06,
      "loss": 0.2469,
      "step": 3571
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6201309967754525,
      "learning_rate": 8.380484025679166e-06,
      "loss": 0.2489,
      "step": 3572
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4587136809526096,
      "learning_rate": 8.379529369663616e-06,
      "loss": 0.2317,
      "step": 3573
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4329762793107164,
      "learning_rate": 8.37857448677002e-06,
      "loss": 0.2344,
      "step": 3574
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.139150836369118,
      "learning_rate": 8.377619377062483e-06,
      "loss": 0.5683,
      "step": 3575
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3209228526348853,
      "learning_rate": 8.376664040605122e-06,
      "loss": 0.2337,
      "step": 3576
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.300426186948551,
      "learning_rate": 8.375708477462075e-06,
      "loss": 0.5826,
      "step": 3577
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5571356487365462,
      "learning_rate": 8.37475268769749e-06,
      "loss": 0.2294,
      "step": 3578
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2982539749910906,
      "learning_rate": 8.373796671375531e-06,
      "loss": 0.1805,
      "step": 3579
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3027740968560202,
      "learning_rate": 8.372840428560379e-06,
      "loss": 0.2109,
      "step": 3580
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5498852892032748,
      "learning_rate": 8.37188395931623e-06,
      "loss": 0.2192,
      "step": 3581
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.584910515408539,
      "learning_rate": 8.370927263707293e-06,
      "loss": 0.7473,
      "step": 3582
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3808792874416347,
      "learning_rate": 8.369970341797797e-06,
      "loss": 0.2156,
      "step": 3583
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2559392269000558,
      "learning_rate": 8.369013193651982e-06,
      "loss": 0.1762,
      "step": 3584
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.506684861025489,
      "learning_rate": 8.368055819334101e-06,
      "loss": 0.2397,
      "step": 3585
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3035072648123753,
      "learning_rate": 8.367098218908431e-06,
      "loss": 0.1999,
      "step": 3586
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.208307351585978,
      "learning_rate": 8.366140392439255e-06,
      "loss": 0.3899,
      "step": 3587
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2769741335788356,
      "learning_rate": 8.365182339990875e-06,
      "loss": 0.1986,
      "step": 3588
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4492665330628032,
      "learning_rate": 8.364224061627608e-06,
      "loss": 0.2346,
      "step": 3589
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2593445140495894,
      "learning_rate": 8.363265557413786e-06,
      "loss": 0.2077,
      "step": 3590
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5151863512719281,
      "learning_rate": 8.36230682741376e-06,
      "loss": 0.2351,
      "step": 3591
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5491872618165108,
      "learning_rate": 8.361347871691885e-06,
      "loss": 0.2361,
      "step": 3592
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3795351336744357,
      "learning_rate": 8.360388690312546e-06,
      "loss": 0.1909,
      "step": 3593
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.654255473682728,
      "learning_rate": 8.35942928334013e-06,
      "loss": 0.2544,
      "step": 3594
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.503308348173627,
      "learning_rate": 8.358469650839049e-06,
      "loss": 0.6242,
      "step": 3595
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4328297608911573,
      "learning_rate": 8.357509792873725e-06,
      "loss": 0.251,
      "step": 3596
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.356706270059027,
      "learning_rate": 8.356549709508596e-06,
      "loss": 0.194,
      "step": 3597
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4589583616897974,
      "learning_rate": 8.355589400808115e-06,
      "loss": 0.183,
      "step": 3598
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.32716123412696,
      "learning_rate": 8.35462886683675e-06,
      "loss": 0.5073,
      "step": 3599
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2816688133159695,
      "learning_rate": 8.353668107658984e-06,
      "loss": 0.2204,
      "step": 3600
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3436927550294877,
      "learning_rate": 8.352707123339318e-06,
      "loss": 0.1832,
      "step": 3601
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4049583297428472,
      "learning_rate": 8.351745913942265e-06,
      "loss": 0.2157,
      "step": 3602
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.330778287490357,
      "learning_rate": 8.35078447953235e-06,
      "loss": 0.7936,
      "step": 3603
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3753206991155558,
      "learning_rate": 8.349822820174125e-06,
      "loss": 0.2253,
      "step": 3604
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2499544592639533,
      "learning_rate": 8.348860935932143e-06,
      "loss": 0.1838,
      "step": 3605
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3981161905262895,
      "learning_rate": 8.34789882687098e-06,
      "loss": 0.199,
      "step": 3606
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4575574110933573,
      "learning_rate": 8.346936493055224e-06,
      "loss": 0.1866,
      "step": 3607
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3597440436401644,
      "learning_rate": 8.34597393454948e-06,
      "loss": 0.2013,
      "step": 3608
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3188029845421754,
      "learning_rate": 8.345011151418368e-06,
      "loss": 0.2064,
      "step": 3609
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2753653036203787,
      "learning_rate": 8.344048143726524e-06,
      "loss": 0.207,
      "step": 3610
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.242345704634693,
      "learning_rate": 8.343084911538595e-06,
      "loss": 0.1924,
      "step": 3611
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.260163749419984,
      "learning_rate": 8.342121454919246e-06,
      "loss": 0.2158,
      "step": 3612
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.476982778057887,
      "learning_rate": 8.341157773933158e-06,
      "loss": 0.2291,
      "step": 3613
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3896105238425212,
      "learning_rate": 8.340193868645026e-06,
      "loss": 0.1854,
      "step": 3614
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3509730901125538,
      "learning_rate": 8.339229739119558e-06,
      "loss": 0.2194,
      "step": 3615
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.391433033324576,
      "learning_rate": 8.338265385421482e-06,
      "loss": 0.2075,
      "step": 3616
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3746226792363845,
      "learning_rate": 8.337300807615536e-06,
      "loss": 0.2206,
      "step": 3617
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3516295263782565,
      "learning_rate": 8.336336005766475e-06,
      "loss": 0.2005,
      "step": 3618
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4282317438389422,
      "learning_rate": 8.335370979939067e-06,
      "loss": 0.2028,
      "step": 3619
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3501104463948796,
      "learning_rate": 8.334405730198101e-06,
      "loss": 0.2015,
      "step": 3620
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.574640350031064,
      "learning_rate": 8.333440256608376e-06,
      "loss": 0.2372,
      "step": 3621
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4295925903521247,
      "learning_rate": 8.332474559234705e-06,
      "loss": 0.2382,
      "step": 3622
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4372995187617252,
      "learning_rate": 8.33150863814192e-06,
      "loss": 0.1975,
      "step": 3623
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3406626991679254,
      "learning_rate": 8.330542493394868e-06,
      "loss": 0.2469,
      "step": 3624
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2863909604227424,
      "learning_rate": 8.329576125058406e-06,
      "loss": 0.1674,
      "step": 3625
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4563004339514694,
      "learning_rate": 8.328609533197411e-06,
      "loss": 0.2477,
      "step": 3626
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4536198930009567,
      "learning_rate": 8.327642717876772e-06,
      "loss": 0.2213,
      "step": 3627
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4656670854878002,
      "learning_rate": 8.326675679161396e-06,
      "loss": 0.2497,
      "step": 3628
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.285532404428391,
      "learning_rate": 8.3257084171162e-06,
      "loss": 0.196,
      "step": 3629
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4659319420783707,
      "learning_rate": 8.324740931806125e-06,
      "loss": 0.2025,
      "step": 3630
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6033861839326882,
      "learning_rate": 8.323773223296117e-06,
      "loss": 0.2294,
      "step": 3631
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2932123187943596,
      "learning_rate": 8.32280529165114e-06,
      "loss": 0.2279,
      "step": 3632
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3860152862379742,
      "learning_rate": 8.321837136936176e-06,
      "loss": 0.2086,
      "step": 3633
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3582526039781295,
      "learning_rate": 8.320868759216221e-06,
      "loss": 0.1841,
      "step": 3634
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.317588757132336,
      "learning_rate": 8.319900158556285e-06,
      "loss": 0.5561,
      "step": 3635
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.399881478807535,
      "learning_rate": 8.318931335021393e-06,
      "loss": 0.235,
      "step": 3636
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4720100751917324,
      "learning_rate": 8.317962288676582e-06,
      "loss": 0.2192,
      "step": 3637
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2736464299322818,
      "learning_rate": 8.31699301958691e-06,
      "loss": 0.2075,
      "step": 3638
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5199475689639685,
      "learning_rate": 8.316023527817447e-06,
      "loss": 0.1997,
      "step": 3639
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.52945565095852,
      "learning_rate": 8.315053813433279e-06,
      "loss": 0.2421,
      "step": 3640
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5733506511274244,
      "learning_rate": 8.314083876499503e-06,
      "loss": 0.216,
      "step": 3641
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3359313514771205,
      "learning_rate": 8.313113717081234e-06,
      "loss": 0.2095,
      "step": 3642
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2796021653276421,
      "learning_rate": 8.312143335243605e-06,
      "loss": 0.2302,
      "step": 3643
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.563369572541649,
      "learning_rate": 8.311172731051758e-06,
      "loss": 0.2525,
      "step": 3644
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3346987265034822,
      "learning_rate": 8.310201904570853e-06,
      "loss": 0.2063,
      "step": 3645
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.420818497383828,
      "learning_rate": 8.309230855866065e-06,
      "loss": 0.2219,
      "step": 3646
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4621177291913883,
      "learning_rate": 8.308259585002581e-06,
      "loss": 0.2035,
      "step": 3647
    },
    {
      "epoch": 0.29,
      "grad_norm": 79.83605653195225,
      "learning_rate": 8.307288092045608e-06,
      "loss": 0.6522,
      "step": 3648
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4830585883208247,
      "learning_rate": 8.306316377060366e-06,
      "loss": 0.2691,
      "step": 3649
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3036038323488608,
      "learning_rate": 8.305344440112089e-06,
      "loss": 0.1788,
      "step": 3650
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4897912133373863,
      "learning_rate": 8.304372281266023e-06,
      "loss": 0.2429,
      "step": 3651
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4163330999523593,
      "learning_rate": 8.303399900587433e-06,
      "loss": 0.2476,
      "step": 3652
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4271150760742066,
      "learning_rate": 8.3024272981416e-06,
      "loss": 0.2063,
      "step": 3653
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2786276349988321,
      "learning_rate": 8.301454473993817e-06,
      "loss": 0.1857,
      "step": 3654
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.489550263694434,
      "learning_rate": 8.300481428209391e-06,
      "loss": 0.5646,
      "step": 3655
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5996304194049025,
      "learning_rate": 8.299508160853648e-06,
      "loss": 0.2017,
      "step": 3656
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2146250171642388,
      "learning_rate": 8.298534671991925e-06,
      "loss": 0.1867,
      "step": 3657
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3454815762676198,
      "learning_rate": 8.297560961689574e-06,
      "loss": 0.2066,
      "step": 3658
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3649164808778247,
      "learning_rate": 8.296587030011965e-06,
      "loss": 0.2245,
      "step": 3659
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.455109035751038,
      "learning_rate": 8.295612877024482e-06,
      "loss": 0.2107,
      "step": 3660
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.381462131841822,
      "learning_rate": 8.294638502792517e-06,
      "loss": 0.2174,
      "step": 3661
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3088356961123524,
      "learning_rate": 8.29366390738149e-06,
      "loss": 0.2631,
      "step": 3662
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3723464164030308,
      "learning_rate": 8.292689090856826e-06,
      "loss": 0.2356,
      "step": 3663
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.427429774474268,
      "learning_rate": 8.291714053283965e-06,
      "loss": 0.2122,
      "step": 3664
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3695769686363157,
      "learning_rate": 8.290738794728369e-06,
      "loss": 0.2171,
      "step": 3665
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4393291405342103,
      "learning_rate": 8.289763315255506e-06,
      "loss": 0.2141,
      "step": 3666
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.681014742635568,
      "learning_rate": 8.288787614930862e-06,
      "loss": 0.2284,
      "step": 3667
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5775208107337915,
      "learning_rate": 8.287811693819943e-06,
      "loss": 0.215,
      "step": 3668
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3944643343603251,
      "learning_rate": 8.286835551988262e-06,
      "loss": 0.2025,
      "step": 3669
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3641289414013542,
      "learning_rate": 8.285859189501353e-06,
      "loss": 0.1872,
      "step": 3670
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3591705302643022,
      "learning_rate": 8.28488260642476e-06,
      "loss": 0.2247,
      "step": 3671
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4712739259040275,
      "learning_rate": 8.283905802824044e-06,
      "loss": 0.1944,
      "step": 3672
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2743427209148437,
      "learning_rate": 8.282928778764783e-06,
      "loss": 0.1663,
      "step": 3673
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.302232184843105,
      "learning_rate": 8.281951534312566e-06,
      "loss": 0.2181,
      "step": 3674
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2098277345307213,
      "learning_rate": 8.280974069532999e-06,
      "loss": 0.1878,
      "step": 3675
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6433614979568003,
      "learning_rate": 8.2799963844917e-06,
      "loss": 0.219,
      "step": 3676
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6470605389946293,
      "learning_rate": 8.279018479254307e-06,
      "loss": 0.2516,
      "step": 3677
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3535455443861624,
      "learning_rate": 8.278040353886468e-06,
      "loss": 0.1803,
      "step": 3678
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.388689154054311,
      "learning_rate": 8.277062008453848e-06,
      "loss": 0.2127,
      "step": 3679
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.286672203501988,
      "learning_rate": 8.276083443022127e-06,
      "loss": 0.1903,
      "step": 3680
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2099200395071816,
      "learning_rate": 8.275104657656998e-06,
      "loss": 0.1932,
      "step": 3681
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.34566525864894,
      "learning_rate": 8.274125652424169e-06,
      "loss": 0.2064,
      "step": 3682
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.812300159171215,
      "learning_rate": 8.273146427389365e-06,
      "loss": 0.7962,
      "step": 3683
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2614664935664717,
      "learning_rate": 8.272166982618326e-06,
      "loss": 0.1853,
      "step": 3684
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4009811097030007,
      "learning_rate": 8.271187318176803e-06,
      "loss": 0.2172,
      "step": 3685
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4703533836664249,
      "learning_rate": 8.270207434130562e-06,
      "loss": 0.2143,
      "step": 3686
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3942150400953677,
      "learning_rate": 8.269227330545388e-06,
      "loss": 0.2737,
      "step": 3687
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4359373033324636,
      "learning_rate": 8.26824700748708e-06,
      "loss": 0.2042,
      "step": 3688
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4024022106210114,
      "learning_rate": 8.267266465021445e-06,
      "loss": 0.2344,
      "step": 3689
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3645021143953837,
      "learning_rate": 8.266285703214315e-06,
      "loss": 0.1971,
      "step": 3690
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2983707680147094,
      "learning_rate": 8.265304722131528e-06,
      "loss": 0.222,
      "step": 3691
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4190517889433056,
      "learning_rate": 8.264323521838945e-06,
      "loss": 0.2315,
      "step": 3692
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.399609214833165,
      "learning_rate": 8.26334210240243e-06,
      "loss": 0.2265,
      "step": 3693
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4849815667250656,
      "learning_rate": 8.262360463887871e-06,
      "loss": 0.1911,
      "step": 3694
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4065921565172728,
      "learning_rate": 8.261378606361173e-06,
      "loss": 0.2306,
      "step": 3695
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3978970327573963,
      "learning_rate": 8.260396529888244e-06,
      "loss": 0.2423,
      "step": 3696
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4786677270683066,
      "learning_rate": 8.25941423453502e-06,
      "loss": 0.2306,
      "step": 3697
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2950361479029586,
      "learning_rate": 8.25843172036744e-06,
      "loss": 0.2029,
      "step": 3698
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.529412594745916,
      "learning_rate": 8.25744898745147e-06,
      "loss": 0.2194,
      "step": 3699
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3973395692805393,
      "learning_rate": 8.256466035853077e-06,
      "loss": 0.2054,
      "step": 3700
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.9423219727902445,
      "learning_rate": 8.255482865638253e-06,
      "loss": 0.7162,
      "step": 3701
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4680897725275448,
      "learning_rate": 8.254499476873001e-06,
      "loss": 0.1768,
      "step": 3702
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3666915719498542,
      "learning_rate": 8.253515869623339e-06,
      "loss": 0.1748,
      "step": 3703
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.480952367741595,
      "learning_rate": 8.252532043955296e-06,
      "loss": 0.263,
      "step": 3704
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3093405784125278,
      "learning_rate": 8.251547999934924e-06,
      "loss": 0.182,
      "step": 3705
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3902644640422197,
      "learning_rate": 8.250563737628284e-06,
      "loss": 0.212,
      "step": 3706
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4616072349454408,
      "learning_rate": 8.24957925710145e-06,
      "loss": 0.2108,
      "step": 3707
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.376501002638323,
      "learning_rate": 8.248594558420515e-06,
      "loss": 0.2016,
      "step": 3708
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5387570187128823,
      "learning_rate": 8.247609641651584e-06,
      "loss": 0.2254,
      "step": 3709
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5489918495288462,
      "learning_rate": 8.246624506860779e-06,
      "loss": 0.2697,
      "step": 3710
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3097732406823503,
      "learning_rate": 8.245639154114234e-06,
      "loss": 0.1699,
      "step": 3711
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2622903227246631,
      "learning_rate": 8.244653583478097e-06,
      "loss": 0.227,
      "step": 3712
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3749743408834962,
      "learning_rate": 8.243667795018534e-06,
      "loss": 0.2164,
      "step": 3713
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4549180829773172,
      "learning_rate": 8.242681788801727e-06,
      "loss": 0.231,
      "step": 3714
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5138124701086562,
      "learning_rate": 8.241695564893863e-06,
      "loss": 0.199,
      "step": 3715
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4729945768951331,
      "learning_rate": 8.240709123361157e-06,
      "loss": 0.1819,
      "step": 3716
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4379992296055528,
      "learning_rate": 8.239722464269826e-06,
      "loss": 0.238,
      "step": 3717
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5441518070877207,
      "learning_rate": 8.238735587686111e-06,
      "loss": 0.2034,
      "step": 3718
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2635229442489286,
      "learning_rate": 8.237748493676264e-06,
      "loss": 0.2089,
      "step": 3719
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4433181528455978,
      "learning_rate": 8.23676118230655e-06,
      "loss": 0.2225,
      "step": 3720
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3295000548166718,
      "learning_rate": 8.23577365364325e-06,
      "loss": 0.1927,
      "step": 3721
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3299863091429769,
      "learning_rate": 8.23478590775266e-06,
      "loss": 0.2058,
      "step": 3722
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.32929603006992,
      "learning_rate": 8.233797944701093e-06,
      "loss": 0.1697,
      "step": 3723
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3850555964034388,
      "learning_rate": 8.23280976455487e-06,
      "loss": 0.1985,
      "step": 3724
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.6459866806684795,
      "learning_rate": 8.231821367380335e-06,
      "loss": 0.5332,
      "step": 3725
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4984089277641566,
      "learning_rate": 8.230832753243838e-06,
      "loss": 0.244,
      "step": 3726
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3791108475428366,
      "learning_rate": 8.22984392221175e-06,
      "loss": 0.2206,
      "step": 3727
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3937207997634669,
      "learning_rate": 8.228854874350452e-06,
      "loss": 0.2785,
      "step": 3728
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.347972058847856,
      "learning_rate": 8.227865609726344e-06,
      "loss": 0.1893,
      "step": 3729
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4610277082582304,
      "learning_rate": 8.226876128405838e-06,
      "loss": 0.2318,
      "step": 3730
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4266060858796787,
      "learning_rate": 8.22588643045536e-06,
      "loss": 0.2127,
      "step": 3731
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3381218278042828,
      "learning_rate": 8.224896515941352e-06,
      "loss": 0.2172,
      "step": 3732
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.059647950561012,
      "learning_rate": 8.22390638493027e-06,
      "loss": 0.751,
      "step": 3733
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3438928045051468,
      "learning_rate": 8.222916037488585e-06,
      "loss": 0.2327,
      "step": 3734
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6371557991669559,
      "learning_rate": 8.22192547368278e-06,
      "loss": 0.2752,
      "step": 3735
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3556432457610368,
      "learning_rate": 8.220934693579357e-06,
      "loss": 0.1792,
      "step": 3736
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3950380065500485,
      "learning_rate": 8.219943697244828e-06,
      "loss": 0.2156,
      "step": 3737
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5186624761909908,
      "learning_rate": 8.218952484745723e-06,
      "loss": 0.2469,
      "step": 3738
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3430299980915081,
      "learning_rate": 8.217961056148584e-06,
      "loss": 0.1695,
      "step": 3739
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5473682006150449,
      "learning_rate": 8.216969411519971e-06,
      "loss": 0.2411,
      "step": 3740
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.339528098739716,
      "learning_rate": 8.215977550926452e-06,
      "loss": 0.2373,
      "step": 3741
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2857071520583583,
      "learning_rate": 8.214985474434618e-06,
      "loss": 0.1797,
      "step": 3742
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.160616836812752,
      "learning_rate": 8.213993182111066e-06,
      "loss": 0.1471,
      "step": 3743
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1978952469314283,
      "learning_rate": 8.213000674022415e-06,
      "loss": 0.1481,
      "step": 3744
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3652798963890227,
      "learning_rate": 8.212007950235295e-06,
      "loss": 0.1952,
      "step": 3745
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3333484544166128,
      "learning_rate": 8.211015010816347e-06,
      "loss": 0.2063,
      "step": 3746
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.268932675839391,
      "learning_rate": 8.210021855832234e-06,
      "loss": 0.1888,
      "step": 3747
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4098247194853812,
      "learning_rate": 8.209028485349627e-06,
      "loss": 0.244,
      "step": 3748
    },
    {
      "epoch": 0.3,
      "grad_norm": 21.99483528163536,
      "learning_rate": 8.208034899435216e-06,
      "loss": 0.5939,
      "step": 3749
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.28325668313735,
      "learning_rate": 8.207041098155701e-06,
      "loss": 0.1718,
      "step": 3750
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.320616648218833,
      "learning_rate": 8.2060470815778e-06,
      "loss": 0.193,
      "step": 3751
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4483438505468544,
      "learning_rate": 8.205052849768244e-06,
      "loss": 0.2602,
      "step": 3752
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2273375653978682,
      "learning_rate": 8.204058402793782e-06,
      "loss": 0.2313,
      "step": 3753
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.335381397708339,
      "learning_rate": 8.20306374072117e-06,
      "loss": 0.1754,
      "step": 3754
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5015469322437207,
      "learning_rate": 8.202068863617183e-06,
      "loss": 0.2072,
      "step": 3755
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5663185400489383,
      "learning_rate": 8.201073771548611e-06,
      "loss": 0.2716,
      "step": 3756
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2099649688181984,
      "learning_rate": 8.20007846458226e-06,
      "loss": 0.1854,
      "step": 3757
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2170989499886844,
      "learning_rate": 8.199082942784943e-06,
      "loss": 0.2103,
      "step": 3758
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3745667809485924,
      "learning_rate": 8.198087206223498e-06,
      "loss": 0.2073,
      "step": 3759
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.509638521605373,
      "learning_rate": 8.19709125496477e-06,
      "loss": 0.2558,
      "step": 3760
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1512066954533253,
      "learning_rate": 8.196095089075615e-06,
      "loss": 0.1845,
      "step": 3761
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3621491829710406,
      "learning_rate": 8.195098708622917e-06,
      "loss": 0.2229,
      "step": 3762
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6281650486868589,
      "learning_rate": 8.194102113673558e-06,
      "loss": 0.2679,
      "step": 3763
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1191402777196033,
      "learning_rate": 8.193105304294452e-06,
      "loss": 0.1792,
      "step": 3764
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4479889098365322,
      "learning_rate": 8.192108280552507e-06,
      "loss": 0.2576,
      "step": 3765
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4873604791788069,
      "learning_rate": 8.191111042514663e-06,
      "loss": 0.2157,
      "step": 3766
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5418804154420083,
      "learning_rate": 8.190113590247866e-06,
      "loss": 0.2147,
      "step": 3767
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2470064232035385,
      "learning_rate": 8.189115923819078e-06,
      "loss": 0.2021,
      "step": 3768
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3102997450068385,
      "learning_rate": 8.188118043295276e-06,
      "loss": 0.2018,
      "step": 3769
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3357024427014244,
      "learning_rate": 8.18711994874345e-06,
      "loss": 0.1948,
      "step": 3770
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3251568560369325,
      "learning_rate": 8.186121640230604e-06,
      "loss": 0.1874,
      "step": 3771
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.373837088366283,
      "learning_rate": 8.18512311782376e-06,
      "loss": 0.2203,
      "step": 3772
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2400577554799947,
      "learning_rate": 8.184124381589951e-06,
      "loss": 0.1883,
      "step": 3773
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1858150849851898,
      "learning_rate": 8.183125431596224e-06,
      "loss": 0.2094,
      "step": 3774
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3700138523272412,
      "learning_rate": 8.182126267909642e-06,
      "loss": 0.2411,
      "step": 3775
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3167487756659753,
      "learning_rate": 8.181126890597281e-06,
      "loss": 0.1921,
      "step": 3776
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.443382472707996,
      "learning_rate": 8.180127299726236e-06,
      "loss": 0.2667,
      "step": 3777
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3151528661559193,
      "learning_rate": 8.179127495363607e-06,
      "loss": 0.2337,
      "step": 3778
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.576240110368662,
      "learning_rate": 8.178127477576521e-06,
      "loss": 0.2624,
      "step": 3779
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3715344336381463,
      "learning_rate": 8.177127246432105e-06,
      "loss": 0.2065,
      "step": 3780
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.452805563399023,
      "learning_rate": 8.176126801997512e-06,
      "loss": 0.1897,
      "step": 3781
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.178345529572568,
      "learning_rate": 8.175126144339901e-06,
      "loss": 0.1943,
      "step": 3782
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3576207740697785,
      "learning_rate": 8.174125273526455e-06,
      "loss": 0.2137,
      "step": 3783
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6846658531351133,
      "learning_rate": 8.173124189624362e-06,
      "loss": 0.275,
      "step": 3784
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.208878340863701,
      "learning_rate": 8.172122892700826e-06,
      "loss": 0.2088,
      "step": 3785
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2254991042253245,
      "learning_rate": 8.171121382823069e-06,
      "loss": 0.1546,
      "step": 3786
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3434993431008382,
      "learning_rate": 8.170119660058327e-06,
      "loss": 0.2129,
      "step": 3787
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.034028722742464,
      "learning_rate": 8.169117724473847e-06,
      "loss": 0.5966,
      "step": 3788
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3228752110650597,
      "learning_rate": 8.168115576136891e-06,
      "loss": 0.1617,
      "step": 3789
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5694246028967147,
      "learning_rate": 8.167113215114738e-06,
      "loss": 0.2845,
      "step": 3790
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4130274092295652,
      "learning_rate": 8.166110641474679e-06,
      "loss": 0.2349,
      "step": 3791
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4367740492720582,
      "learning_rate": 8.165107855284019e-06,
      "loss": 0.2401,
      "step": 3792
    },
    {
      "epoch": 0.3,
      "grad_norm": 12.643956673191243,
      "learning_rate": 8.16410485661008e-06,
      "loss": 0.6212,
      "step": 3793
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2734615238494864,
      "learning_rate": 8.163101645520193e-06,
      "loss": 0.2043,
      "step": 3794
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4149364356908314,
      "learning_rate": 8.16209822208171e-06,
      "loss": 0.1979,
      "step": 3795
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2582820217914774,
      "learning_rate": 8.161094586361992e-06,
      "loss": 0.2225,
      "step": 3796
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.478590770963363,
      "learning_rate": 8.160090738428418e-06,
      "loss": 0.604,
      "step": 3797
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.391850948496934,
      "learning_rate": 8.159086678348378e-06,
      "loss": 0.1988,
      "step": 3798
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5561336336706055,
      "learning_rate": 8.158082406189278e-06,
      "loss": 0.2571,
      "step": 3799
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3930409160327242,
      "learning_rate": 8.157077922018537e-06,
      "loss": 0.2242,
      "step": 3800
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5124187551910635,
      "learning_rate": 8.156073225903588e-06,
      "loss": 0.2507,
      "step": 3801
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3272023628567406,
      "learning_rate": 8.155068317911882e-06,
      "loss": 0.2024,
      "step": 3802
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3523427424580936,
      "learning_rate": 8.15406319811088e-06,
      "loss": 0.1956,
      "step": 3803
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3571419779991183,
      "learning_rate": 8.153057866568059e-06,
      "loss": 0.1982,
      "step": 3804
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3532630537981003,
      "learning_rate": 8.152052323350909e-06,
      "loss": 0.2466,
      "step": 3805
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3399469802869355,
      "learning_rate": 8.151046568526938e-06,
      "loss": 0.178,
      "step": 3806
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.268732818577074,
      "learning_rate": 8.150040602163665e-06,
      "loss": 0.6794,
      "step": 3807
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.471598565280061,
      "learning_rate": 8.14903442432862e-06,
      "loss": 0.1827,
      "step": 3808
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3741470379487557,
      "learning_rate": 8.148028035089353e-06,
      "loss": 0.2025,
      "step": 3809
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.320980023199665,
      "learning_rate": 8.147021434513425e-06,
      "loss": 0.1797,
      "step": 3810
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2158344135043195,
      "learning_rate": 8.146014622668415e-06,
      "loss": 0.182,
      "step": 3811
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.254493537630793,
      "learning_rate": 8.14500759962191e-06,
      "loss": 0.2174,
      "step": 3812
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5143944226286659,
      "learning_rate": 8.144000365441515e-06,
      "loss": 0.1746,
      "step": 3813
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.352600148430449,
      "learning_rate": 8.142992920194852e-06,
      "loss": 0.208,
      "step": 3814
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3309887233938618,
      "learning_rate": 8.14198526394955e-06,
      "loss": 0.1861,
      "step": 3815
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3395657168842534,
      "learning_rate": 8.140977396773257e-06,
      "loss": 0.2254,
      "step": 3816
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.828579637503647,
      "learning_rate": 8.139969318733636e-06,
      "loss": 0.1971,
      "step": 3817
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4154001019319817,
      "learning_rate": 8.138961029898358e-06,
      "loss": 0.2162,
      "step": 3818
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.722554168417846,
      "learning_rate": 8.137952530335119e-06,
      "loss": 0.4921,
      "step": 3819
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.39348253248178,
      "learning_rate": 8.136943820111615e-06,
      "loss": 0.2143,
      "step": 3820
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.402773358244312,
      "learning_rate": 8.135934899295572e-06,
      "loss": 0.2185,
      "step": 3821
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.435000871013821,
      "learning_rate": 8.134925767954716e-06,
      "loss": 0.2292,
      "step": 3822
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3145847619822002,
      "learning_rate": 8.133916426156795e-06,
      "loss": 0.2218,
      "step": 3823
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3795694584195648,
      "learning_rate": 8.132906873969568e-06,
      "loss": 0.2014,
      "step": 3824
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.426083799878832,
      "learning_rate": 8.13189711146081e-06,
      "loss": 0.1869,
      "step": 3825
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5010355448842745,
      "learning_rate": 8.130887138698311e-06,
      "loss": 0.1965,
      "step": 3826
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3447163614327744,
      "learning_rate": 8.129876955749871e-06,
      "loss": 0.2063,
      "step": 3827
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3723379728908258,
      "learning_rate": 8.128866562683309e-06,
      "loss": 0.2094,
      "step": 3828
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1235312923257483,
      "learning_rate": 8.127855959566452e-06,
      "loss": 0.1542,
      "step": 3829
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2770919731434849,
      "learning_rate": 8.126845146467151e-06,
      "loss": 0.158,
      "step": 3830
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5168597774387853,
      "learning_rate": 8.12583412345326e-06,
      "loss": 0.2203,
      "step": 3831
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5001805143384344,
      "learning_rate": 8.124822890592652e-06,
      "loss": 0.2773,
      "step": 3832
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5648243265884805,
      "learning_rate": 8.123811447953218e-06,
      "loss": 0.2417,
      "step": 3833
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3648919281105643,
      "learning_rate": 8.122799795602855e-06,
      "loss": 0.1761,
      "step": 3834
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5458571796780096,
      "learning_rate": 8.12178793360948e-06,
      "loss": 0.2449,
      "step": 3835
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.271115662251034,
      "learning_rate": 8.120775862041023e-06,
      "loss": 0.1921,
      "step": 3836
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3516787335878178,
      "learning_rate": 8.119763580965425e-06,
      "loss": 0.2417,
      "step": 3837
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.417710205515965,
      "learning_rate": 8.118751090450647e-06,
      "loss": 0.2281,
      "step": 3838
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.372317266512435,
      "learning_rate": 8.117738390564658e-06,
      "loss": 0.1675,
      "step": 3839
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1982225719895114,
      "learning_rate": 8.116725481375448e-06,
      "loss": 0.1544,
      "step": 3840
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2010840745297853,
      "learning_rate": 8.11571236295101e-06,
      "loss": 0.1774,
      "step": 3841
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.414962284996212,
      "learning_rate": 8.114699035359361e-06,
      "loss": 0.2266,
      "step": 3842
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3693482179846685,
      "learning_rate": 8.113685498668527e-06,
      "loss": 0.2407,
      "step": 3843
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2687671228040156,
      "learning_rate": 8.112671752946554e-06,
      "loss": 0.2121,
      "step": 3844
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3241716119899425,
      "learning_rate": 8.111657798261495e-06,
      "loss": 0.2235,
      "step": 3845
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2164656639018097,
      "learning_rate": 8.110643634681419e-06,
      "loss": 0.2039,
      "step": 3846
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.516246851677116,
      "learning_rate": 8.10962926227441e-06,
      "loss": 0.219,
      "step": 3847
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3117082523676236,
      "learning_rate": 8.108614681108568e-06,
      "loss": 0.1849,
      "step": 3848
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.509941156196788,
      "learning_rate": 8.107599891252005e-06,
      "loss": 0.2072,
      "step": 3849
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2132483981011335,
      "learning_rate": 8.106584892772844e-06,
      "loss": 0.198,
      "step": 3850
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3858259817769758,
      "learning_rate": 8.105569685739227e-06,
      "loss": 0.2519,
      "step": 3851
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.512459936470633,
      "learning_rate": 8.104554270219307e-06,
      "loss": 0.2304,
      "step": 3852
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3263916266519096,
      "learning_rate": 8.103538646281253e-06,
      "loss": 0.1505,
      "step": 3853
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.294783366535686,
      "learning_rate": 8.102522813993247e-06,
      "loss": 0.203,
      "step": 3854
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.216788746044837,
      "learning_rate": 8.101506773423484e-06,
      "loss": 0.1856,
      "step": 3855
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4940954150927188,
      "learning_rate": 8.100490524640172e-06,
      "loss": 0.2634,
      "step": 3856
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.338140377670126,
      "learning_rate": 8.09947406771154e-06,
      "loss": 0.184,
      "step": 3857
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.6104205638483835,
      "learning_rate": 8.098457402705822e-06,
      "loss": 0.6341,
      "step": 3858
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.274708201948737,
      "learning_rate": 8.09744052969127e-06,
      "loss": 0.1976,
      "step": 3859
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3494122180929107,
      "learning_rate": 8.09642344873615e-06,
      "loss": 0.2015,
      "step": 3860
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.347851965256785,
      "learning_rate": 8.095406159908741e-06,
      "loss": 0.177,
      "step": 3861
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.253097303071293,
      "learning_rate": 8.094388663277339e-06,
      "loss": 0.2085,
      "step": 3862
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4515645437875249,
      "learning_rate": 8.093370958910252e-06,
      "loss": 0.2646,
      "step": 3863
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3888438508248435,
      "learning_rate": 8.092353046875798e-06,
      "loss": 0.1876,
      "step": 3864
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.72749920782667,
      "learning_rate": 8.091334927242315e-06,
      "loss": 0.4353,
      "step": 3865
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4557629869975592,
      "learning_rate": 8.090316600078152e-06,
      "loss": 0.2413,
      "step": 3866
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.10867396928844,
      "learning_rate": 8.089298065451673e-06,
      "loss": 0.5667,
      "step": 3867
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3324617322697836,
      "learning_rate": 8.088279323431255e-06,
      "loss": 0.1712,
      "step": 3868
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6462473953992076,
      "learning_rate": 8.087260374085286e-06,
      "loss": 0.2406,
      "step": 3869
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.667452322103438,
      "learning_rate": 8.086241217482177e-06,
      "loss": 0.2574,
      "step": 3870
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3138267296210167,
      "learning_rate": 8.085221853690344e-06,
      "loss": 0.2088,
      "step": 3871
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.546490218014016,
      "learning_rate": 8.084202282778218e-06,
      "loss": 0.5648,
      "step": 3872
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2583768178123669,
      "learning_rate": 8.083182504814249e-06,
      "loss": 0.1319,
      "step": 3873
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.378382834789108,
      "learning_rate": 8.082162519866898e-06,
      "loss": 0.2296,
      "step": 3874
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.444753271743106,
      "learning_rate": 8.081142328004638e-06,
      "loss": 0.2023,
      "step": 3875
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.559102318618348,
      "learning_rate": 8.080121929295957e-06,
      "loss": 0.2534,
      "step": 3876
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.620181786986073,
      "learning_rate": 8.07910132380936e-06,
      "loss": 0.6777,
      "step": 3877
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3915794860198896,
      "learning_rate": 8.07808051161336e-06,
      "loss": 0.2318,
      "step": 3878
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.150878766324051,
      "learning_rate": 8.07705949277649e-06,
      "loss": 0.1812,
      "step": 3879
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3357387381515504,
      "learning_rate": 8.076038267367292e-06,
      "loss": 0.1817,
      "step": 3880
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2328356037374417,
      "learning_rate": 8.075016835454327e-06,
      "loss": 0.1972,
      "step": 3881
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.50137873260632,
      "learning_rate": 8.073995197106163e-06,
      "loss": 0.2,
      "step": 3882
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4156423292367393,
      "learning_rate": 8.07297335239139e-06,
      "loss": 0.1815,
      "step": 3883
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2383994634532183,
      "learning_rate": 8.071951301378604e-06,
      "loss": 0.187,
      "step": 3884
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2539914277041322,
      "learning_rate": 8.070929044136419e-06,
      "loss": 0.2061,
      "step": 3885
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4900851522767355,
      "learning_rate": 8.069906580733461e-06,
      "loss": 0.2451,
      "step": 3886
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2343465521259052,
      "learning_rate": 8.068883911238377e-06,
      "loss": 0.2053,
      "step": 3887
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2943545835185957,
      "learning_rate": 8.067861035719816e-06,
      "loss": 0.224,
      "step": 3888
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.300324359689798,
      "learning_rate": 8.066837954246447e-06,
      "loss": 0.6481,
      "step": 3889
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3778856431987445,
      "learning_rate": 8.065814666886954e-06,
      "loss": 0.2133,
      "step": 3890
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2015676728607707,
      "learning_rate": 8.064791173710033e-06,
      "loss": 0.1927,
      "step": 3891
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.231435034184277,
      "learning_rate": 8.063767474784397e-06,
      "loss": 0.1655,
      "step": 3892
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3684226485653346,
      "learning_rate": 8.062743570178767e-06,
      "loss": 0.2022,
      "step": 3893
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.623355058271424,
      "learning_rate": 8.06171945996188e-06,
      "loss": 0.2597,
      "step": 3894
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.900743990962824,
      "learning_rate": 8.06069514420249e-06,
      "loss": 0.4829,
      "step": 3895
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5129030600681552,
      "learning_rate": 8.059670622969363e-06,
      "loss": 0.2184,
      "step": 3896
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4583743399132925,
      "learning_rate": 8.058645896331274e-06,
      "loss": 0.1809,
      "step": 3897
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5219000938628264,
      "learning_rate": 8.057620964357018e-06,
      "loss": 0.2333,
      "step": 3898
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.272767208035992,
      "learning_rate": 8.056595827115404e-06,
      "loss": 0.5816,
      "step": 3899
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3408453091881496,
      "learning_rate": 8.055570484675252e-06,
      "loss": 0.1869,
      "step": 3900
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1830891435623785,
      "learning_rate": 8.054544937105393e-06,
      "loss": 0.1992,
      "step": 3901
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5870902756894514,
      "learning_rate": 8.053519184474679e-06,
      "loss": 0.2175,
      "step": 3902
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5366996290517843,
      "learning_rate": 8.052493226851971e-06,
      "loss": 0.2005,
      "step": 3903
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3383593295558844,
      "learning_rate": 8.051467064306142e-06,
      "loss": 0.1871,
      "step": 3904
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4643097722439373,
      "learning_rate": 8.050440696906086e-06,
      "loss": 0.2025,
      "step": 3905
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3122153673218235,
      "learning_rate": 8.049414124720702e-06,
      "loss": 0.1813,
      "step": 3906
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5189376024119374,
      "learning_rate": 8.048387347818908e-06,
      "loss": 0.2198,
      "step": 3907
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2335353404978318,
      "learning_rate": 8.047360366269636e-06,
      "loss": 0.1717,
      "step": 3908
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2212660078410205,
      "learning_rate": 8.046333180141827e-06,
      "loss": 0.1876,
      "step": 3909
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5880997170269437,
      "learning_rate": 8.045305789504446e-06,
      "loss": 0.2225,
      "step": 3910
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2372582583447032,
      "learning_rate": 8.044278194426459e-06,
      "loss": 0.151,
      "step": 3911
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.455127719565864,
      "learning_rate": 8.043250394976852e-06,
      "loss": 0.2678,
      "step": 3912
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3381907926194567,
      "learning_rate": 8.042222391224624e-06,
      "loss": 0.2333,
      "step": 3913
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.349875936995672,
      "learning_rate": 8.04119418323879e-06,
      "loss": 0.5374,
      "step": 3914
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2824233812787074,
      "learning_rate": 8.040165771088377e-06,
      "loss": 0.226,
      "step": 3915
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4698355742895297,
      "learning_rate": 8.039137154842424e-06,
      "loss": 0.1962,
      "step": 3916
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2825614472697844,
      "learning_rate": 8.038108334569986e-06,
      "loss": 0.2204,
      "step": 3917
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.583614757035065,
      "learning_rate": 8.03707931034013e-06,
      "loss": 0.6275,
      "step": 3918
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.64898838428393,
      "learning_rate": 8.03605008222194e-06,
      "loss": 0.7141,
      "step": 3919
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.687256771767737,
      "learning_rate": 8.035020650284507e-06,
      "loss": 0.1977,
      "step": 3920
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4359456165979352,
      "learning_rate": 8.03399101459694e-06,
      "loss": 0.1947,
      "step": 3921
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.501620848749396,
      "learning_rate": 8.032961175228366e-06,
      "loss": 0.2594,
      "step": 3922
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4129752316209603,
      "learning_rate": 8.031931132247918e-06,
      "loss": 0.2337,
      "step": 3923
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.173079752936573,
      "learning_rate": 8.030900885724748e-06,
      "loss": 0.1759,
      "step": 3924
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.358015484779165,
      "learning_rate": 8.029870435728018e-06,
      "loss": 0.7442,
      "step": 3925
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3117396566032236,
      "learning_rate": 8.028839782326905e-06,
      "loss": 0.1646,
      "step": 3926
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3842003877459141,
      "learning_rate": 8.0278089255906e-06,
      "loss": 0.2263,
      "step": 3927
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5118005228721116,
      "learning_rate": 8.026777865588308e-06,
      "loss": 0.2495,
      "step": 3928
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.591346622603105,
      "learning_rate": 8.025746602389248e-06,
      "loss": 0.2477,
      "step": 3929
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2546252596318728,
      "learning_rate": 8.02471513606265e-06,
      "loss": 0.2042,
      "step": 3930
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.75292150213567,
      "learning_rate": 8.023683466677762e-06,
      "loss": 0.7022,
      "step": 3931
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2940007025010616,
      "learning_rate": 8.02265159430384e-06,
      "loss": 0.196,
      "step": 3932
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.297520267530138,
      "learning_rate": 8.021619519010158e-06,
      "loss": 0.159,
      "step": 3933
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2361737642107624,
      "learning_rate": 8.020587240866004e-06,
      "loss": 0.2157,
      "step": 3934
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3442300348071081,
      "learning_rate": 8.019554759940675e-06,
      "loss": 0.188,
      "step": 3935
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4386745021893488,
      "learning_rate": 8.018522076303487e-06,
      "loss": 0.2509,
      "step": 3936
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2366331257831256,
      "learning_rate": 8.017489190023767e-06,
      "loss": 0.2058,
      "step": 3937
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4474858879275714,
      "learning_rate": 8.016456101170853e-06,
      "loss": 0.2203,
      "step": 3938
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3533307490578559,
      "learning_rate": 8.015422809814102e-06,
      "loss": 0.1949,
      "step": 3939
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5336873402792435,
      "learning_rate": 8.01438931602288e-06,
      "loss": 0.2174,
      "step": 3940
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4467875984137255,
      "learning_rate": 8.013355619866573e-06,
      "loss": 0.2155,
      "step": 3941
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3170092877132251,
      "learning_rate": 8.012321721414572e-06,
      "loss": 0.166,
      "step": 3942
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2957565469038352,
      "learning_rate": 8.011287620736288e-06,
      "loss": 0.2385,
      "step": 3943
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4190068889190592,
      "learning_rate": 8.01025331790114e-06,
      "loss": 0.1994,
      "step": 3944
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3431785937654042,
      "learning_rate": 8.009218812978567e-06,
      "loss": 0.1857,
      "step": 3945
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3630350792240267,
      "learning_rate": 8.008184106038017e-06,
      "loss": 0.2079,
      "step": 3946
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2200559031101863,
      "learning_rate": 8.007149197148954e-06,
      "loss": 0.1826,
      "step": 3947
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3701955395318743,
      "learning_rate": 8.006114086380855e-06,
      "loss": 0.1887,
      "step": 3948
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.9197906091642905,
      "learning_rate": 8.005078773803207e-06,
      "loss": 0.5194,
      "step": 3949
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3318217094219627,
      "learning_rate": 8.004043259485519e-06,
      "loss": 0.2202,
      "step": 3950
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3461664106334252,
      "learning_rate": 8.003007543497303e-06,
      "loss": 0.1647,
      "step": 3951
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.474177600164627,
      "learning_rate": 8.001971625908091e-06,
      "loss": 0.244,
      "step": 3952
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4031588350502715,
      "learning_rate": 8.000935506787431e-06,
      "loss": 0.1959,
      "step": 3953
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2978934812693592,
      "learning_rate": 7.999899186204876e-06,
      "loss": 0.2118,
      "step": 3954
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2610008584659997,
      "learning_rate": 7.99886266423e-06,
      "loss": 0.2133,
      "step": 3955
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4112411612048457,
      "learning_rate": 7.997825940932387e-06,
      "loss": 0.2556,
      "step": 3956
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2416601441206263,
      "learning_rate": 7.996789016381633e-06,
      "loss": 0.215,
      "step": 3957
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4814599174413965,
      "learning_rate": 7.995751890647356e-06,
      "loss": 0.261,
      "step": 3958
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3593429679006102,
      "learning_rate": 7.994714563799177e-06,
      "loss": 0.1821,
      "step": 3959
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.99668364553336,
      "learning_rate": 7.993677035906734e-06,
      "loss": 0.634,
      "step": 3960
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6311457431206913,
      "learning_rate": 7.992639307039683e-06,
      "loss": 0.2369,
      "step": 3961
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3452305034566447,
      "learning_rate": 7.991601377267688e-06,
      "loss": 0.2156,
      "step": 3962
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5013586172301305,
      "learning_rate": 7.990563246660427e-06,
      "loss": 0.2081,
      "step": 3963
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4708830704824098,
      "learning_rate": 7.989524915287595e-06,
      "loss": 0.248,
      "step": 3964
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5071056098392208,
      "learning_rate": 7.988486383218898e-06,
      "loss": 0.2336,
      "step": 3965
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3028345996066666,
      "learning_rate": 7.987447650524054e-06,
      "loss": 0.2233,
      "step": 3966
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.142034713267671,
      "learning_rate": 7.9864087172728e-06,
      "loss": 0.7164,
      "step": 3967
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1477877146706996,
      "learning_rate": 7.985369583534877e-06,
      "loss": 0.1793,
      "step": 3968
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2934143979115635,
      "learning_rate": 7.98433024938005e-06,
      "loss": 0.1668,
      "step": 3969
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.290238257160589,
      "learning_rate": 7.98329071487809e-06,
      "loss": 0.1973,
      "step": 3970
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.548102246802242,
      "learning_rate": 7.982250980098789e-06,
      "loss": 0.7221,
      "step": 3971
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4850484042084637,
      "learning_rate": 7.981211045111942e-06,
      "loss": 0.2362,
      "step": 3972
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4825843799523257,
      "learning_rate": 7.980170909987363e-06,
      "loss": 0.2269,
      "step": 3973
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5342327252589767,
      "learning_rate": 7.979130574794884e-06,
      "loss": 0.2135,
      "step": 3974
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3032135002126601,
      "learning_rate": 7.978090039604342e-06,
      "loss": 0.1955,
      "step": 3975
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3368386284837364,
      "learning_rate": 7.97704930448559e-06,
      "loss": 0.1971,
      "step": 3976
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4700572540687062,
      "learning_rate": 7.9760083695085e-06,
      "loss": 0.2252,
      "step": 3977
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.918500265559613,
      "learning_rate": 7.974967234742954e-06,
      "loss": 0.5652,
      "step": 3978
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.990893084110975,
      "learning_rate": 7.973925900258841e-06,
      "loss": 0.7468,
      "step": 3979
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3898168052542716,
      "learning_rate": 7.972884366126072e-06,
      "loss": 0.2566,
      "step": 3980
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.600702829975893,
      "learning_rate": 7.971842632414569e-06,
      "loss": 0.5795,
      "step": 3981
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.426458912575459,
      "learning_rate": 7.970800699194263e-06,
      "loss": 0.762,
      "step": 3982
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3591071123757004,
      "learning_rate": 7.969758566535106e-06,
      "loss": 0.1839,
      "step": 3983
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2299600097155978,
      "learning_rate": 7.96871623450706e-06,
      "loss": 0.2245,
      "step": 3984
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.381872173161071,
      "learning_rate": 7.967673703180096e-06,
      "loss": 0.2212,
      "step": 3985
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.336387902708184,
      "learning_rate": 7.966630972624208e-06,
      "loss": 0.212,
      "step": 3986
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.268092437514557,
      "learning_rate": 7.96558804290939e-06,
      "loss": 0.1906,
      "step": 3987
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3688463127849748,
      "learning_rate": 7.964544914105665e-06,
      "loss": 0.2043,
      "step": 3988
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5133901464003763,
      "learning_rate": 7.963501586283055e-06,
      "loss": 0.236,
      "step": 3989
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1561157597935667,
      "learning_rate": 7.962458059511607e-06,
      "loss": 0.1973,
      "step": 3990
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.400514810286149,
      "learning_rate": 7.961414333861373e-06,
      "loss": 0.2454,
      "step": 3991
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1753817993493918,
      "learning_rate": 7.96037040940242e-06,
      "loss": 0.1808,
      "step": 3992
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.188775357970155,
      "learning_rate": 7.959326286204833e-06,
      "loss": 0.2349,
      "step": 3993
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4044122369734588,
      "learning_rate": 7.958281964338706e-06,
      "loss": 0.2197,
      "step": 3994
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.382862706820919,
      "learning_rate": 7.957237443874148e-06,
      "loss": 0.196,
      "step": 3995
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4953246061369465,
      "learning_rate": 7.95619272488128e-06,
      "loss": 0.2021,
      "step": 3996
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4610163211226375,
      "learning_rate": 7.955147807430238e-06,
      "loss": 0.2219,
      "step": 3997
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4841565402265473,
      "learning_rate": 7.954102691591171e-06,
      "loss": 0.198,
      "step": 3998
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.69635422259678,
      "learning_rate": 7.953057377434237e-06,
      "loss": 0.7829,
      "step": 3999
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.281165321518233,
      "learning_rate": 7.952011865029614e-06,
      "loss": 0.171,
      "step": 4000
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.168749381051066,
      "learning_rate": 7.950966154447492e-06,
      "loss": 0.6267,
      "step": 4001
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.240793874016181,
      "learning_rate": 7.94992024575807e-06,
      "loss": 0.1861,
      "step": 4002
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4054228000952058,
      "learning_rate": 7.948874139031564e-06,
      "loss": 0.1995,
      "step": 4003
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3283706346551687,
      "learning_rate": 7.947827834338203e-06,
      "loss": 0.2136,
      "step": 4004
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.363874966015505,
      "learning_rate": 7.946781331748226e-06,
      "loss": 0.2034,
      "step": 4005
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.380403245357081,
      "learning_rate": 7.945734631331891e-06,
      "loss": 0.1808,
      "step": 4006
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4358422606141732,
      "learning_rate": 7.944687733159466e-06,
      "loss": 0.2221,
      "step": 4007
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2421013953647664,
      "learning_rate": 7.94364063730123e-06,
      "loss": 0.208,
      "step": 4008
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1639220080902029,
      "learning_rate": 7.94259334382748e-06,
      "loss": 0.2045,
      "step": 4009
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4221892213425364,
      "learning_rate": 7.941545852808523e-06,
      "loss": 0.2037,
      "step": 4010
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3656781835471623,
      "learning_rate": 7.94049816431468e-06,
      "loss": 0.1942,
      "step": 4011
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5351563075092955,
      "learning_rate": 7.939450278416288e-06,
      "loss": 0.2247,
      "step": 4012
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3095625745197526,
      "learning_rate": 7.93840219518369e-06,
      "loss": 0.1823,
      "step": 4013
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2836290807238213,
      "learning_rate": 7.937353914687249e-06,
      "loss": 0.1844,
      "step": 4014
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4352008247571864,
      "learning_rate": 7.936305436997343e-06,
      "loss": 0.2277,
      "step": 4015
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.506104855577261,
      "learning_rate": 7.935256762184354e-06,
      "loss": 0.2291,
      "step": 4016
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3501947784815256,
      "learning_rate": 7.934207890318686e-06,
      "loss": 0.1985,
      "step": 4017
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4608276714556927,
      "learning_rate": 7.933158821470752e-06,
      "loss": 0.2124,
      "step": 4018
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4296851955667318,
      "learning_rate": 7.932109555710979e-06,
      "loss": 0.2213,
      "step": 4019
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5187092033792717,
      "learning_rate": 7.931060093109807e-06,
      "loss": 0.245,
      "step": 4020
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3785124563433033,
      "learning_rate": 7.930010433737692e-06,
      "loss": 0.1676,
      "step": 4021
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4544041941697485,
      "learning_rate": 7.928960577665096e-06,
      "loss": 0.2563,
      "step": 4022
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6048481141989983,
      "learning_rate": 7.927910524962506e-06,
      "loss": 0.2297,
      "step": 4023
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3572033122224483,
      "learning_rate": 7.926860275700408e-06,
      "loss": 0.1917,
      "step": 4024
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5781350418753863,
      "learning_rate": 7.925809829949312e-06,
      "loss": 0.2579,
      "step": 4025
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3552825211617143,
      "learning_rate": 7.924759187779737e-06,
      "loss": 0.1972,
      "step": 4026
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4569620142731758,
      "learning_rate": 7.923708349262218e-06,
      "loss": 0.2348,
      "step": 4027
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3094162763446615,
      "learning_rate": 7.922657314467297e-06,
      "loss": 0.2003,
      "step": 4028
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3841680587079115,
      "learning_rate": 7.921606083465538e-06,
      "loss": 0.2241,
      "step": 4029
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4067911724993933,
      "learning_rate": 7.920554656327509e-06,
      "loss": 0.2016,
      "step": 4030
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2691942745031113,
      "learning_rate": 7.919503033123795e-06,
      "loss": 0.1971,
      "step": 4031
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1884472633314014,
      "learning_rate": 7.918451213925e-06,
      "loss": 0.1776,
      "step": 4032
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.276012405701036,
      "learning_rate": 7.91739919880173e-06,
      "loss": 0.1557,
      "step": 4033
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.873425089766336,
      "learning_rate": 7.916346987824612e-06,
      "loss": 0.4123,
      "step": 4034
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.468295867731718,
      "learning_rate": 7.915294581064287e-06,
      "loss": 0.2411,
      "step": 4035
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.487222314913909,
      "learning_rate": 7.914241978591403e-06,
      "loss": 0.256,
      "step": 4036
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.386196938454332,
      "learning_rate": 7.913189180476623e-06,
      "loss": 0.2289,
      "step": 4037
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.366204466731206,
      "learning_rate": 7.912136186790628e-06,
      "loss": 0.2247,
      "step": 4038
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2281103362720471,
      "learning_rate": 7.91108299760411e-06,
      "loss": 0.1686,
      "step": 4039
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3332936179328003,
      "learning_rate": 7.910029612987766e-06,
      "loss": 0.1908,
      "step": 4040
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3579630446771767,
      "learning_rate": 7.90897603301232e-06,
      "loss": 0.2195,
      "step": 4041
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3742289140500026,
      "learning_rate": 7.907922257748498e-06,
      "loss": 0.2244,
      "step": 4042
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4705197476751664,
      "learning_rate": 7.906868287267044e-06,
      "loss": 0.2219,
      "step": 4043
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3978853583772395,
      "learning_rate": 7.905814121638715e-06,
      "loss": 0.2464,
      "step": 4044
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3425902814367763,
      "learning_rate": 7.90475976093428e-06,
      "loss": 0.1795,
      "step": 4045
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.999084275607514,
      "learning_rate": 7.903705205224518e-06,
      "loss": 0.4826,
      "step": 4046
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3197182804567718,
      "learning_rate": 7.902650454580233e-06,
      "loss": 0.1817,
      "step": 4047
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.082998522186344,
      "learning_rate": 7.901595509072224e-06,
      "loss": 0.161,
      "step": 4048
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4531715688974516,
      "learning_rate": 7.900540368771319e-06,
      "loss": 0.2281,
      "step": 4049
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3410596430698283,
      "learning_rate": 7.89948503374835e-06,
      "loss": 0.1993,
      "step": 4050
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.493736311943534,
      "learning_rate": 7.898429504074165e-06,
      "loss": 0.2638,
      "step": 4051
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3549260568043828,
      "learning_rate": 7.897373779819627e-06,
      "loss": 0.1893,
      "step": 4052
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.657542154263301,
      "learning_rate": 7.896317861055609e-06,
      "loss": 0.587,
      "step": 4053
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.137440944391673,
      "learning_rate": 7.895261747852996e-06,
      "loss": 0.6967,
      "step": 4054
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2340759483895059,
      "learning_rate": 7.89420544028269e-06,
      "loss": 0.1923,
      "step": 4055
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.045243153512769,
      "learning_rate": 7.893148938415602e-06,
      "loss": 0.7346,
      "step": 4056
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1748107632597535,
      "learning_rate": 7.892092242322662e-06,
      "loss": 0.1753,
      "step": 4057
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.300679857611565,
      "learning_rate": 7.891035352074808e-06,
      "loss": 0.7109,
      "step": 4058
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.277711060694452,
      "learning_rate": 7.88997826774299e-06,
      "loss": 0.1973,
      "step": 4059
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.348231432666873,
      "learning_rate": 7.888920989398174e-06,
      "loss": 0.2195,
      "step": 4060
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.470930614121796,
      "learning_rate": 7.887863517111337e-06,
      "loss": 0.2141,
      "step": 4061
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4371714376575915,
      "learning_rate": 7.886805850953476e-06,
      "loss": 0.2188,
      "step": 4062
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.432188247191454,
      "learning_rate": 7.88574799099559e-06,
      "loss": 0.2474,
      "step": 4063
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4404755853856661,
      "learning_rate": 7.884689937308698e-06,
      "loss": 0.1924,
      "step": 4064
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4152935209153108,
      "learning_rate": 7.883631689963831e-06,
      "loss": 0.1867,
      "step": 4065
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3348794602290117,
      "learning_rate": 7.882573249032031e-06,
      "loss": 0.1985,
      "step": 4066
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3164126258258746,
      "learning_rate": 7.881514614584356e-06,
      "loss": 0.2092,
      "step": 4067
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.441138815362133,
      "learning_rate": 7.880455786691872e-06,
      "loss": 0.2156,
      "step": 4068
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4826858508841896,
      "learning_rate": 7.879396765425665e-06,
      "loss": 0.2119,
      "step": 4069
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2620397541727186,
      "learning_rate": 7.878337550856829e-06,
      "loss": 0.1889,
      "step": 4070
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.376404947090972,
      "learning_rate": 7.87727814305647e-06,
      "loss": 0.2096,
      "step": 4071
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2892368752426073,
      "learning_rate": 7.876218542095715e-06,
      "loss": 0.1798,
      "step": 4072
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4684951972136515,
      "learning_rate": 7.875158748045691e-06,
      "loss": 0.2332,
      "step": 4073
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.73732110642301,
      "learning_rate": 7.874098760977552e-06,
      "loss": 0.7673,
      "step": 4074
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3295259617834692,
      "learning_rate": 7.873038580962453e-06,
      "loss": 0.2591,
      "step": 4075
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5145936257203483,
      "learning_rate": 7.871978208071572e-06,
      "loss": 0.1971,
      "step": 4076
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.403915698666203,
      "learning_rate": 7.870917642376088e-06,
      "loss": 0.2294,
      "step": 4077
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5436568345254131,
      "learning_rate": 7.869856883947208e-06,
      "loss": 0.2057,
      "step": 4078
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4180047816894987,
      "learning_rate": 7.868795932856138e-06,
      "loss": 0.2276,
      "step": 4079
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0968950017144357,
      "learning_rate": 7.867734789174104e-06,
      "loss": 0.132,
      "step": 4080
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4881437193403455,
      "learning_rate": 7.866673452972346e-06,
      "loss": 0.1926,
      "step": 4081
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.540178649001633,
      "learning_rate": 7.865611924322113e-06,
      "loss": 0.5771,
      "step": 4082
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.275561595975137,
      "learning_rate": 7.864550203294669e-06,
      "loss": 0.2215,
      "step": 4083
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3102470688598118,
      "learning_rate": 7.863488289961291e-06,
      "loss": 0.1838,
      "step": 4084
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.401617961777475,
      "learning_rate": 7.86242618439327e-06,
      "loss": 0.2073,
      "step": 4085
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4639585447551864,
      "learning_rate": 7.861363886661903e-06,
      "loss": 0.2268,
      "step": 4086
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.492680339583702,
      "learning_rate": 7.86030139683851e-06,
      "loss": 0.6386,
      "step": 4087
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.549542900902339,
      "learning_rate": 7.859238714994419e-06,
      "loss": 0.5361,
      "step": 4088
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.304386027337992,
      "learning_rate": 7.858175841200968e-06,
      "loss": 0.177,
      "step": 4089
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2991924800561592,
      "learning_rate": 7.857112775529513e-06,
      "loss": 0.1952,
      "step": 4090
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.348850933197621,
      "learning_rate": 7.856049518051423e-06,
      "loss": 0.2301,
      "step": 4091
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1420891247625766,
      "learning_rate": 7.854986068838073e-06,
      "loss": 0.1436,
      "step": 4092
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3347477451301466,
      "learning_rate": 7.85392242796086e-06,
      "loss": 0.231,
      "step": 4093
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.394754992691177,
      "learning_rate": 7.852858595491186e-06,
      "loss": 0.7561,
      "step": 4094
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.313125991920316,
      "learning_rate": 7.85179457150047e-06,
      "loss": 0.6511,
      "step": 4095
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3106698991438515,
      "learning_rate": 7.850730356060145e-06,
      "loss": 0.1593,
      "step": 4096
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.363134227599685,
      "learning_rate": 7.849665949241653e-06,
      "loss": 0.2012,
      "step": 4097
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4383464626993177,
      "learning_rate": 7.848601351116454e-06,
      "loss": 0.2121,
      "step": 4098
    },
    {
      "epoch": 0.33,
      "grad_norm": 15.498277113151147,
      "learning_rate": 7.847536561756012e-06,
      "loss": 0.7387,
      "step": 4099
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3875819764255877,
      "learning_rate": 7.846471581231814e-06,
      "loss": 0.224,
      "step": 4100
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5177600431702303,
      "learning_rate": 7.845406409615354e-06,
      "loss": 0.2387,
      "step": 4101
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3216211886130689,
      "learning_rate": 7.844341046978144e-06,
      "loss": 0.1944,
      "step": 4102
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2881404785662474,
      "learning_rate": 7.843275493391698e-06,
      "loss": 0.2161,
      "step": 4103
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5227554331459012,
      "learning_rate": 7.842209748927554e-06,
      "loss": 0.2647,
      "step": 4104
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4383841940976994,
      "learning_rate": 7.841143813657257e-06,
      "loss": 0.1985,
      "step": 4105
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5166287636946287,
      "learning_rate": 7.840077687652368e-06,
      "loss": 0.2496,
      "step": 4106
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3601324961236294,
      "learning_rate": 7.83901137098446e-06,
      "loss": 0.2131,
      "step": 4107
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.347209417009386,
      "learning_rate": 7.837944863725116e-06,
      "loss": 0.196,
      "step": 4108
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.160420508312375,
      "learning_rate": 7.836878165945934e-06,
      "loss": 0.1973,
      "step": 4109
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3362402130212154,
      "learning_rate": 7.835811277718528e-06,
      "loss": 0.2084,
      "step": 4110
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3933196723668342,
      "learning_rate": 7.834744199114517e-06,
      "loss": 0.1796,
      "step": 4111
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3662873983944175,
      "learning_rate": 7.833676930205538e-06,
      "loss": 0.2152,
      "step": 4112
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3095816993456044,
      "learning_rate": 7.832609471063242e-06,
      "loss": 0.1969,
      "step": 4113
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.130632417059272,
      "learning_rate": 7.831541821759288e-06,
      "loss": 0.7041,
      "step": 4114
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.180047106402712,
      "learning_rate": 7.830473982365355e-06,
      "loss": 0.6027,
      "step": 4115
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3333731514574172,
      "learning_rate": 7.829405952953127e-06,
      "loss": 0.1815,
      "step": 4116
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3323515731898747,
      "learning_rate": 7.828337733594304e-06,
      "loss": 0.2165,
      "step": 4117
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2779810112881456,
      "learning_rate": 7.827269324360599e-06,
      "loss": 0.2039,
      "step": 4118
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4811841669676526,
      "learning_rate": 7.826200725323738e-06,
      "loss": 0.2117,
      "step": 4119
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3076168826317172,
      "learning_rate": 7.82513193655546e-06,
      "loss": 0.2223,
      "step": 4120
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3929572930277456,
      "learning_rate": 7.824062958127515e-06,
      "loss": 0.2546,
      "step": 4121
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.298777907252214,
      "learning_rate": 7.822993790111668e-06,
      "loss": 0.2204,
      "step": 4122
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.60516313466515,
      "learning_rate": 7.821924432579693e-06,
      "loss": 0.6862,
      "step": 4123
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.308252922836962,
      "learning_rate": 7.820854885603381e-06,
      "loss": 0.1684,
      "step": 4124
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5399384307295758,
      "learning_rate": 7.819785149254534e-06,
      "loss": 0.1939,
      "step": 4125
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4965111177440602,
      "learning_rate": 7.818715223604966e-06,
      "loss": 0.2097,
      "step": 4126
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3250007089055158,
      "learning_rate": 7.817645108726504e-06,
      "loss": 0.2331,
      "step": 4127
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4254926725775519,
      "learning_rate": 7.816574804690991e-06,
      "loss": 0.2194,
      "step": 4128
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.442442451184468,
      "learning_rate": 7.815504311570276e-06,
      "loss": 0.2207,
      "step": 4129
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3819579153450376,
      "learning_rate": 7.814433629436225e-06,
      "loss": 0.2405,
      "step": 4130
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5168482279016906,
      "learning_rate": 7.813362758360719e-06,
      "loss": 0.2447,
      "step": 4131
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3545507943858213,
      "learning_rate": 7.812291698415647e-06,
      "loss": 0.2258,
      "step": 4132
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4443173695540255,
      "learning_rate": 7.811220449672909e-06,
      "loss": 0.2425,
      "step": 4133
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.530496486389679,
      "learning_rate": 7.810149012204427e-06,
      "loss": 0.232,
      "step": 4134
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2913255442481155,
      "learning_rate": 7.809077386082129e-06,
      "loss": 0.1883,
      "step": 4135
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2609003405147097,
      "learning_rate": 7.808005571377952e-06,
      "loss": 0.1707,
      "step": 4136
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.443199968020226,
      "learning_rate": 7.806933568163855e-06,
      "loss": 0.6255,
      "step": 4137
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.320971000452345,
      "learning_rate": 7.805861376511802e-06,
      "loss": 0.213,
      "step": 4138
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.714154643071379,
      "learning_rate": 7.804788996493773e-06,
      "loss": 0.4573,
      "step": 4139
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3335379499933375,
      "learning_rate": 7.803716428181762e-06,
      "loss": 0.2148,
      "step": 4140
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3919067920752246,
      "learning_rate": 7.802643671647772e-06,
      "loss": 0.2381,
      "step": 4141
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3709071941724613,
      "learning_rate": 7.80157072696382e-06,
      "loss": 0.2249,
      "step": 4142
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3575278358919654,
      "learning_rate": 7.800497594201935e-06,
      "loss": 0.1865,
      "step": 4143
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.40489892921925,
      "learning_rate": 7.799424273434164e-06,
      "loss": 0.6067,
      "step": 4144
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.231895430554824,
      "learning_rate": 7.79835076473256e-06,
      "loss": 0.1867,
      "step": 4145
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.446794622329574,
      "learning_rate": 7.797277068169187e-06,
      "loss": 0.26,
      "step": 4146
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.336387924672816,
      "learning_rate": 7.796203183816131e-06,
      "loss": 0.2368,
      "step": 4147
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2615560496544245,
      "learning_rate": 7.795129111745484e-06,
      "loss": 0.1968,
      "step": 4148
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2577797643190718,
      "learning_rate": 7.79405485202935e-06,
      "loss": 0.2177,
      "step": 4149
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5176143521946903,
      "learning_rate": 7.792980404739849e-06,
      "loss": 0.2117,
      "step": 4150
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3115485988816853,
      "learning_rate": 7.791905769949108e-06,
      "loss": 0.2457,
      "step": 4151
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.314607596902828,
      "learning_rate": 7.790830947729278e-06,
      "loss": 0.1873,
      "step": 4152
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.447361498007139,
      "learning_rate": 7.789755938152508e-06,
      "loss": 0.2162,
      "step": 4153
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.377479347802542,
      "learning_rate": 7.78868074129097e-06,
      "loss": 0.2413,
      "step": 4154
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.458719204776982,
      "learning_rate": 7.787605357216843e-06,
      "loss": 0.2137,
      "step": 4155
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2802137334720627,
      "learning_rate": 7.786529786002324e-06,
      "loss": 0.202,
      "step": 4156
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.9385986703183,
      "learning_rate": 7.785454027719617e-06,
      "loss": 0.686,
      "step": 4157
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2031168936930932,
      "learning_rate": 7.78437808244094e-06,
      "loss": 0.1537,
      "step": 4158
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4563418943929554,
      "learning_rate": 7.783301950238528e-06,
      "loss": 0.2734,
      "step": 4159
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4541863287729577,
      "learning_rate": 7.782225631184624e-06,
      "loss": 0.2081,
      "step": 4160
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.296192840361664,
      "learning_rate": 7.78114912535148e-06,
      "loss": 0.2081,
      "step": 4161
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5302300013841614,
      "learning_rate": 7.780072432811371e-06,
      "loss": 0.2453,
      "step": 4162
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4080875155902433,
      "learning_rate": 7.778995553636576e-06,
      "loss": 0.1932,
      "step": 4163
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3765035226588807,
      "learning_rate": 7.77791848789939e-06,
      "loss": 0.2012,
      "step": 4164
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4731695362636956,
      "learning_rate": 7.776841235672119e-06,
      "loss": 0.2338,
      "step": 4165
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3910677970213865,
      "learning_rate": 7.775763797027081e-06,
      "loss": 0.1959,
      "step": 4166
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5911570004789106,
      "learning_rate": 7.77468617203661e-06,
      "loss": 0.1896,
      "step": 4167
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6529736922760596,
      "learning_rate": 7.77360836077305e-06,
      "loss": 0.3104,
      "step": 4168
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2277596731451668,
      "learning_rate": 7.772530363308756e-06,
      "loss": 0.1744,
      "step": 4169
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4497968508973773,
      "learning_rate": 7.771452179716099e-06,
      "loss": 0.2368,
      "step": 4170
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3258176159054198,
      "learning_rate": 7.77037381006746e-06,
      "loss": 0.1764,
      "step": 4171
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.600599428436095,
      "learning_rate": 7.769295254435235e-06,
      "loss": 0.6881,
      "step": 4172
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2189255987784076,
      "learning_rate": 7.768216512891824e-06,
      "loss": 0.1765,
      "step": 4173
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.485076744012145,
      "learning_rate": 7.767137585509655e-06,
      "loss": 0.6165,
      "step": 4174
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.142160338125644,
      "learning_rate": 7.766058472361154e-06,
      "loss": 0.6272,
      "step": 4175
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5289245801088447,
      "learning_rate": 7.764979173518767e-06,
      "loss": 0.2158,
      "step": 4176
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.627641625367301,
      "learning_rate": 7.763899689054952e-06,
      "loss": 0.5275,
      "step": 4177
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3863225139843434,
      "learning_rate": 7.762820019042176e-06,
      "loss": 0.2015,
      "step": 4178
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4651849976812208,
      "learning_rate": 7.761740163552921e-06,
      "loss": 0.2462,
      "step": 4179
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5377969727360974,
      "learning_rate": 7.760660122659682e-06,
      "loss": 0.2168,
      "step": 4180
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4711604137385677,
      "learning_rate": 7.759579896434963e-06,
      "loss": 0.2328,
      "step": 4181
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2744588930710363,
      "learning_rate": 7.758499484951285e-06,
      "loss": 0.1913,
      "step": 4182
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4197186979163279,
      "learning_rate": 7.757418888281179e-06,
      "loss": 0.1993,
      "step": 4183
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.472656766566373,
      "learning_rate": 7.756338106497188e-06,
      "loss": 0.2157,
      "step": 4184
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3779435982915498,
      "learning_rate": 7.755257139671868e-06,
      "loss": 0.2071,
      "step": 4185
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3599978928521972,
      "learning_rate": 7.754175987877788e-06,
      "loss": 0.1999,
      "step": 4186
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.985531377875093,
      "learning_rate": 7.75309465118753e-06,
      "loss": 0.4293,
      "step": 4187
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.691334258919299,
      "learning_rate": 7.752013129673685e-06,
      "loss": 0.6981,
      "step": 4188
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.347249148822668,
      "learning_rate": 7.750931423408864e-06,
      "loss": 0.1939,
      "step": 4189
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4037889948654199,
      "learning_rate": 7.749849532465677e-06,
      "loss": 0.2203,
      "step": 4190
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1972210802439345,
      "learning_rate": 7.74876745691676e-06,
      "loss": 0.1603,
      "step": 4191
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3691166158292847,
      "learning_rate": 7.747685196834757e-06,
      "loss": 0.2132,
      "step": 4192
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3614275795980018,
      "learning_rate": 7.746602752292322e-06,
      "loss": 0.2551,
      "step": 4193
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.427300788012132,
      "learning_rate": 7.745520123362121e-06,
      "loss": 0.2217,
      "step": 4194
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4468402668620954,
      "learning_rate": 7.744437310116837e-06,
      "loss": 0.1896,
      "step": 4195
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3194101849500879,
      "learning_rate": 7.743354312629158e-06,
      "loss": 0.1857,
      "step": 4196
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5017080054448577,
      "learning_rate": 7.742271130971795e-06,
      "loss": 0.2524,
      "step": 4197
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2433922557027257,
      "learning_rate": 7.741187765217461e-06,
      "loss": 0.2219,
      "step": 4198
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.34213668757596,
      "learning_rate": 7.740104215438888e-06,
      "loss": 0.2197,
      "step": 4199
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.320091687636802,
      "learning_rate": 7.739020481708816e-06,
      "loss": 0.1928,
      "step": 4200
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4003029227778818,
      "learning_rate": 7.7379365641e-06,
      "loss": 0.1841,
      "step": 4201
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.395411463616382,
      "learning_rate": 7.736852462685206e-06,
      "loss": 0.2353,
      "step": 4202
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4040396488127749,
      "learning_rate": 7.735768177537215e-06,
      "loss": 0.2294,
      "step": 4203
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3414786318578837,
      "learning_rate": 7.734683708728816e-06,
      "loss": 0.2331,
      "step": 4204
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4308723445125975,
      "learning_rate": 7.733599056332816e-06,
      "loss": 0.239,
      "step": 4205
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.883879224848194,
      "learning_rate": 7.732514220422028e-06,
      "loss": 0.5157,
      "step": 4206
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2920345164391571,
      "learning_rate": 7.73142920106928e-06,
      "loss": 0.1917,
      "step": 4207
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4398283330785886,
      "learning_rate": 7.730343998347415e-06,
      "loss": 0.2286,
      "step": 4208
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2944976893335767,
      "learning_rate": 7.729258612329285e-06,
      "loss": 0.1818,
      "step": 4209
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2575464171243669,
      "learning_rate": 7.728173043087754e-06,
      "loss": 0.1944,
      "step": 4210
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4131519379810098,
      "learning_rate": 7.7270872906957e-06,
      "loss": 0.2345,
      "step": 4211
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2329081662113242,
      "learning_rate": 7.726001355226016e-06,
      "loss": 0.1974,
      "step": 4212
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.957081903516495,
      "learning_rate": 7.7249152367516e-06,
      "loss": 0.5569,
      "step": 4213
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3522454115235596,
      "learning_rate": 7.723828935345368e-06,
      "loss": 0.199,
      "step": 4214
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4427884642224482,
      "learning_rate": 7.722742451080247e-06,
      "loss": 0.2192,
      "step": 4215
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2897373625936364,
      "learning_rate": 7.721655784029174e-06,
      "loss": 0.2135,
      "step": 4216
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5266920245615034,
      "learning_rate": 7.720568934265104e-06,
      "loss": 0.2324,
      "step": 4217
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5006556877155306,
      "learning_rate": 7.719481901860996e-06,
      "loss": 0.2212,
      "step": 4218
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.317082212468535,
      "learning_rate": 7.718394686889831e-06,
      "loss": 0.2002,
      "step": 4219
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3895741235982342,
      "learning_rate": 7.717307289424594e-06,
      "loss": 0.2147,
      "step": 4220
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3944440536557305,
      "learning_rate": 7.716219709538285e-06,
      "loss": 0.2169,
      "step": 4221
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.336767905467579,
      "learning_rate": 7.715131947303916e-06,
      "loss": 0.2117,
      "step": 4222
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2850502037655405,
      "learning_rate": 7.714044002794514e-06,
      "loss": 0.2228,
      "step": 4223
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2626411034111444,
      "learning_rate": 7.712955876083115e-06,
      "loss": 0.2213,
      "step": 4224
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.627504520972727,
      "learning_rate": 7.711867567242769e-06,
      "loss": 0.6374,
      "step": 4225
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3911205672149252,
      "learning_rate": 7.710779076346534e-06,
      "loss": 0.1892,
      "step": 4226
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5127681709757852,
      "learning_rate": 7.709690403467486e-06,
      "loss": 0.297,
      "step": 4227
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4654532049612483,
      "learning_rate": 7.708601548678714e-06,
      "loss": 0.2177,
      "step": 4228
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5993654707404812,
      "learning_rate": 7.707512512053312e-06,
      "loss": 0.1898,
      "step": 4229
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3062088485113348,
      "learning_rate": 7.70642329366439e-06,
      "loss": 0.2015,
      "step": 4230
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.386291832646128,
      "learning_rate": 7.705333893585076e-06,
      "loss": 0.2153,
      "step": 4231
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.369131830370373,
      "learning_rate": 7.704244311888499e-06,
      "loss": 0.2028,
      "step": 4232
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.404820069553254,
      "learning_rate": 7.703154548647806e-06,
      "loss": 0.2067,
      "step": 4233
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8180872088476219,
      "learning_rate": 7.702064603936162e-06,
      "loss": 0.2566,
      "step": 4234
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3819647206585368,
      "learning_rate": 7.70097447782673e-06,
      "loss": 0.2277,
      "step": 4235
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.572663553318597,
      "learning_rate": 7.6998841703927e-06,
      "loss": 0.5376,
      "step": 4236
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3985544874096796,
      "learning_rate": 7.698793681707263e-06,
      "loss": 0.2124,
      "step": 4237
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.301509388981198,
      "learning_rate": 7.697703011843632e-06,
      "loss": 0.4187,
      "step": 4238
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.42064472596067,
      "learning_rate": 7.696612160875023e-06,
      "loss": 0.214,
      "step": 4239
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.25355065984926,
      "learning_rate": 7.695521128874669e-06,
      "loss": 0.1846,
      "step": 4240
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.549583995529812,
      "learning_rate": 7.694429915915816e-06,
      "loss": 0.2272,
      "step": 4241
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2982164778276364,
      "learning_rate": 7.693338522071717e-06,
      "loss": 0.1767,
      "step": 4242
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.008084087006375,
      "learning_rate": 7.692246947415643e-06,
      "loss": 0.4301,
      "step": 4243
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5457529540847386,
      "learning_rate": 7.691155192020874e-06,
      "loss": 0.2506,
      "step": 4244
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3205918049903738,
      "learning_rate": 7.690063255960702e-06,
      "loss": 0.2392,
      "step": 4245
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4408328904197585,
      "learning_rate": 7.688971139308435e-06,
      "loss": 0.229,
      "step": 4246
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.329064139277016,
      "learning_rate": 7.687878842137388e-06,
      "loss": 0.2044,
      "step": 4247
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5669282394333617,
      "learning_rate": 7.686786364520889e-06,
      "loss": 0.1937,
      "step": 4248
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4271242160587696,
      "learning_rate": 7.685693706532283e-06,
      "loss": 0.2323,
      "step": 4249
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.211928811424097,
      "learning_rate": 7.68460086824492e-06,
      "loss": 0.2033,
      "step": 4250
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2953847062829094,
      "learning_rate": 7.683507849732166e-06,
      "loss": 0.2177,
      "step": 4251
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2532318780310638,
      "learning_rate": 7.6824146510674e-06,
      "loss": 0.2414,
      "step": 4252
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.428110191115999,
      "learning_rate": 7.681321272324011e-06,
      "loss": 0.4797,
      "step": 4253
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3439137040781126,
      "learning_rate": 7.6802277135754e-06,
      "loss": 0.2193,
      "step": 4254
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.464793473879814,
      "learning_rate": 7.679133974894984e-06,
      "loss": 0.2227,
      "step": 4255
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.341722815372195,
      "learning_rate": 7.678040056356185e-06,
      "loss": 0.2224,
      "step": 4256
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3974479372533142,
      "learning_rate": 7.676945958032445e-06,
      "loss": 0.1857,
      "step": 4257
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.3813999041193545,
      "learning_rate": 7.67585167999721e-06,
      "loss": 0.5687,
      "step": 4258
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3136805477651075,
      "learning_rate": 7.674757222323943e-06,
      "loss": 0.1885,
      "step": 4259
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3494551601006548,
      "learning_rate": 7.673662585086123e-06,
      "loss": 0.2292,
      "step": 4260
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4420032562388025,
      "learning_rate": 7.67256776835723e-06,
      "loss": 0.2031,
      "step": 4261
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4446056609067182,
      "learning_rate": 7.671472772210766e-06,
      "loss": 0.1752,
      "step": 4262
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2063830384558318,
      "learning_rate": 7.67037759672024e-06,
      "loss": 0.1826,
      "step": 4263
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3633372454871888,
      "learning_rate": 7.669282241959177e-06,
      "loss": 0.2134,
      "step": 4264
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.908639598652913,
      "learning_rate": 7.668186708001106e-06,
      "loss": 0.7562,
      "step": 4265
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2381373928276707,
      "learning_rate": 7.66709099491958e-06,
      "loss": 0.1831,
      "step": 4266
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3637384554901681,
      "learning_rate": 7.665995102788153e-06,
      "loss": 0.2374,
      "step": 4267
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3032998469499624,
      "learning_rate": 7.6648990316804e-06,
      "loss": 0.1822,
      "step": 4268
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3981222173006953,
      "learning_rate": 7.663802781669898e-06,
      "loss": 0.2396,
      "step": 4269
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.348116124783053,
      "learning_rate": 7.662706352830244e-06,
      "loss": 0.5654,
      "step": 4270
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.977317041593801,
      "learning_rate": 7.661609745235046e-06,
      "loss": 0.6015,
      "step": 4271
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2858802799413296,
      "learning_rate": 7.66051295895792e-06,
      "loss": 0.2026,
      "step": 4272
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.326866905988052,
      "learning_rate": 7.6594159940725e-06,
      "loss": 0.2052,
      "step": 4273
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6222980943077077,
      "learning_rate": 7.658318850652426e-06,
      "loss": 0.2466,
      "step": 4274
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.53765424393116,
      "learning_rate": 7.657221528771352e-06,
      "loss": 0.2113,
      "step": 4275
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.284850830434102,
      "learning_rate": 7.656124028502947e-06,
      "loss": 0.2082,
      "step": 4276
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4226945246952964,
      "learning_rate": 7.65502634992089e-06,
      "loss": 0.2425,
      "step": 4277
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.182864885487913,
      "learning_rate": 7.653928493098866e-06,
      "loss": 0.1677,
      "step": 4278
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3584948278801785,
      "learning_rate": 7.652830458110581e-06,
      "loss": 0.2243,
      "step": 4279
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.700720776837866,
      "learning_rate": 7.651732245029753e-06,
      "loss": 0.5615,
      "step": 4280
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.466290762444148,
      "learning_rate": 7.650633853930102e-06,
      "loss": 0.2376,
      "step": 4281
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.363050277550056,
      "learning_rate": 7.64953528488537e-06,
      "loss": 0.2087,
      "step": 4282
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.16950897355178,
      "learning_rate": 7.648436537969308e-06,
      "loss": 0.1985,
      "step": 4283
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.219778568413987,
      "learning_rate": 7.647337613255677e-06,
      "loss": 0.1911,
      "step": 4284
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2918953787679484,
      "learning_rate": 7.646238510818249e-06,
      "loss": 0.2015,
      "step": 4285
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1406604945409573,
      "learning_rate": 7.645139230730813e-06,
      "loss": 0.1739,
      "step": 4286
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.414104612293941,
      "learning_rate": 7.644039773067166e-06,
      "loss": 0.2061,
      "step": 4287
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.254493107967551,
      "learning_rate": 7.642940137901116e-06,
      "loss": 0.1917,
      "step": 4288
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3427715656788006,
      "learning_rate": 7.64184032530649e-06,
      "loss": 0.2202,
      "step": 4289
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3484078980661294,
      "learning_rate": 7.640740335357116e-06,
      "loss": 0.2169,
      "step": 4290
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4802304452975208,
      "learning_rate": 7.639640168126843e-06,
      "loss": 0.1956,
      "step": 4291
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3842962527860643,
      "learning_rate": 7.638539823689529e-06,
      "loss": 0.2252,
      "step": 4292
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4228930219645821,
      "learning_rate": 7.637439302119038e-06,
      "loss": 0.2402,
      "step": 4293
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2984993477898443,
      "learning_rate": 7.636338603489261e-06,
      "loss": 0.1867,
      "step": 4294
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.37283854069665,
      "learning_rate": 7.635237727874082e-06,
      "loss": 0.229,
      "step": 4295
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.255637379407906,
      "learning_rate": 7.634136675347413e-06,
      "loss": 0.4579,
      "step": 4296
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2760678841634427,
      "learning_rate": 7.633035445983164e-06,
      "loss": 0.203,
      "step": 4297
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3838105608371336,
      "learning_rate": 7.63193403985527e-06,
      "loss": 0.2499,
      "step": 4298
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2707668585548204,
      "learning_rate": 7.63083245703767e-06,
      "loss": 0.2023,
      "step": 4299
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.386579180623586,
      "learning_rate": 7.629730697604314e-06,
      "loss": 0.2086,
      "step": 4300
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3375171439837847,
      "learning_rate": 7.62862876162917e-06,
      "loss": 0.1991,
      "step": 4301
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.402240528665343,
      "learning_rate": 7.627526649186214e-06,
      "loss": 0.2041,
      "step": 4302
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3354579346682385,
      "learning_rate": 7.6264243603494305e-06,
      "loss": 0.2054,
      "step": 4303
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.365193097943103,
      "learning_rate": 7.625321895192823e-06,
      "loss": 0.2223,
      "step": 4304
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4003792133502433,
      "learning_rate": 7.624219253790403e-06,
      "loss": 0.212,
      "step": 4305
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4537694196828728,
      "learning_rate": 7.623116436216191e-06,
      "loss": 0.2337,
      "step": 4306
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.0904077358819,
      "learning_rate": 7.6220134425442274e-06,
      "loss": 0.5892,
      "step": 4307
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4131412816300277,
      "learning_rate": 7.620910272848556e-06,
      "loss": 0.209,
      "step": 4308
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.402169315827073,
      "learning_rate": 7.619806927203237e-06,
      "loss": 0.1972,
      "step": 4309
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5688591956381426,
      "learning_rate": 7.618703405682341e-06,
      "loss": 0.2798,
      "step": 4310
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.353040254018066,
      "learning_rate": 7.6175997083599525e-06,
      "loss": 0.2282,
      "step": 4311
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.226449187028031,
      "learning_rate": 7.616495835310162e-06,
      "loss": 0.5595,
      "step": 4312
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5367011747437263,
      "learning_rate": 7.61539178660708e-06,
      "loss": 0.2278,
      "step": 4313
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4629270891683064,
      "learning_rate": 7.614287562324823e-06,
      "loss": 0.1602,
      "step": 4314
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3577875456828694,
      "learning_rate": 7.613183162537521e-06,
      "loss": 0.1924,
      "step": 4315
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.693249268018882,
      "learning_rate": 7.612078587319316e-06,
      "loss": 0.3669,
      "step": 4316
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4713893208560636,
      "learning_rate": 7.61097383674436e-06,
      "loss": 0.231,
      "step": 4317
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3897757044193606,
      "learning_rate": 7.60986891088682e-06,
      "loss": 0.2047,
      "step": 4318
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4483935743298884,
      "learning_rate": 7.6087638098208726e-06,
      "loss": 0.1933,
      "step": 4319
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4108536513798686,
      "learning_rate": 7.607658533620708e-06,
      "loss": 0.1987,
      "step": 4320
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2787779176795473,
      "learning_rate": 7.606553082360523e-06,
      "loss": 0.1507,
      "step": 4321
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4929284022756466,
      "learning_rate": 7.605447456114532e-06,
      "loss": 0.2367,
      "step": 4322
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3028605340556978,
      "learning_rate": 7.604341654956959e-06,
      "loss": 0.168,
      "step": 4323
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5882068021196865,
      "learning_rate": 7.60323567896204e-06,
      "loss": 0.2384,
      "step": 4324
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.47487033904761,
      "learning_rate": 7.602129528204023e-06,
      "loss": 0.2238,
      "step": 4325
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.624881947891497,
      "learning_rate": 7.601023202757168e-06,
      "loss": 0.7792,
      "step": 4326
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3313743574496752,
      "learning_rate": 7.599916702695742e-06,
      "loss": 0.1796,
      "step": 4327
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.314746299253457,
      "learning_rate": 7.59881002809403e-06,
      "loss": 0.1999,
      "step": 4328
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.417733681669009,
      "learning_rate": 7.59770317902633e-06,
      "loss": 0.5444,
      "step": 4329
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5598066496222407,
      "learning_rate": 7.596596155566942e-06,
      "loss": 0.2465,
      "step": 4330
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3962840354819928,
      "learning_rate": 7.595488957790188e-06,
      "loss": 0.2156,
      "step": 4331
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4797845153123654,
      "learning_rate": 7.594381585770396e-06,
      "loss": 0.185,
      "step": 4332
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4181708315186943,
      "learning_rate": 7.593274039581908e-06,
      "loss": 0.1856,
      "step": 4333
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.936770487574651,
      "learning_rate": 7.592166319299075e-06,
      "loss": 0.5457,
      "step": 4334
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5557926978635392,
      "learning_rate": 7.591058424996264e-06,
      "loss": 0.2604,
      "step": 4335
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.412072310837266,
      "learning_rate": 7.5899503567478506e-06,
      "loss": 0.2563,
      "step": 4336
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.387054261331683,
      "learning_rate": 7.5888421146282225e-06,
      "loss": 0.1954,
      "step": 4337
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2969096808661633,
      "learning_rate": 7.587733698711779e-06,
      "loss": 0.2059,
      "step": 4338
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.429373153227334,
      "learning_rate": 7.586625109072931e-06,
      "loss": 0.1852,
      "step": 4339
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1825909849279925,
      "learning_rate": 7.585516345786103e-06,
      "loss": 0.152,
      "step": 4340
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2239825512306242,
      "learning_rate": 7.5844074089257295e-06,
      "loss": 0.1809,
      "step": 4341
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2915997510271018,
      "learning_rate": 7.583298298566255e-06,
      "loss": 0.1946,
      "step": 4342
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2181684015092504,
      "learning_rate": 7.582189014782139e-06,
      "loss": 0.2273,
      "step": 4343
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3142569522346708,
      "learning_rate": 7.581079557647853e-06,
      "loss": 0.1833,
      "step": 4344
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4453547020285573,
      "learning_rate": 7.5799699272378715e-06,
      "loss": 0.2333,
      "step": 4345
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2683677869451335,
      "learning_rate": 7.578860123626696e-06,
      "loss": 0.1546,
      "step": 4346
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5343711634776158,
      "learning_rate": 7.577750146888824e-06,
      "loss": 0.268,
      "step": 4347
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.355791554163768,
      "learning_rate": 7.576639997098776e-06,
      "loss": 0.2431,
      "step": 4348
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.460623424433426,
      "learning_rate": 7.575529674331077e-06,
      "loss": 0.6729,
      "step": 4349
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6035309037596033,
      "learning_rate": 7.574419178660269e-06,
      "loss": 0.2243,
      "step": 4350
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4166656432654923,
      "learning_rate": 7.573308510160899e-06,
      "loss": 0.2091,
      "step": 4351
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2395376282516506,
      "learning_rate": 7.572197668907533e-06,
      "loss": 0.1706,
      "step": 4352
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.255248531298815,
      "learning_rate": 7.571086654974744e-06,
      "loss": 0.1996,
      "step": 4353
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4932687803936877,
      "learning_rate": 7.569975468437119e-06,
      "loss": 0.221,
      "step": 4354
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3402318856151096,
      "learning_rate": 7.568864109369252e-06,
      "loss": 0.1868,
      "step": 4355
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4762789770001095,
      "learning_rate": 7.567752577845755e-06,
      "loss": 0.2061,
      "step": 4356
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4009751628905103,
      "learning_rate": 7.566640873941248e-06,
      "loss": 0.2464,
      "step": 4357
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4578742350587837,
      "learning_rate": 7.565528997730362e-06,
      "loss": 0.2423,
      "step": 4358
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.920170714618471,
      "learning_rate": 7.564416949287741e-06,
      "loss": 0.6145,
      "step": 4359
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5096005704088873,
      "learning_rate": 7.56330472868804e-06,
      "loss": 0.2188,
      "step": 4360
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.381664120103105,
      "learning_rate": 7.562192336005927e-06,
      "loss": 0.2415,
      "step": 4361
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1842900794828364,
      "learning_rate": 7.561079771316078e-06,
      "loss": 0.1622,
      "step": 4362
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3626779383158645,
      "learning_rate": 7.559967034693186e-06,
      "loss": 0.1935,
      "step": 4363
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.484710156623623,
      "learning_rate": 7.55885412621195e-06,
      "loss": 0.2022,
      "step": 4364
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1138028358137637,
      "learning_rate": 7.557741045947082e-06,
      "loss": 0.1543,
      "step": 4365
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2894063886904183,
      "learning_rate": 7.556627793973309e-06,
      "loss": 0.2424,
      "step": 4366
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5934110116631923,
      "learning_rate": 7.555514370365367e-06,
      "loss": 0.2367,
      "step": 4367
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2625960266815925,
      "learning_rate": 7.554400775198e-06,
      "loss": 0.1936,
      "step": 4368
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.298927840330535,
      "learning_rate": 7.553287008545971e-06,
      "loss": 0.5638,
      "step": 4369
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.323300025029743,
      "learning_rate": 7.552173070484048e-06,
      "loss": 0.2402,
      "step": 4370
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4177886523114012,
      "learning_rate": 7.551058961087016e-06,
      "loss": 0.2547,
      "step": 4371
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.428894422623511,
      "learning_rate": 7.549944680429665e-06,
      "loss": 0.2318,
      "step": 4372
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4766800229153574,
      "learning_rate": 7.548830228586802e-06,
      "loss": 0.1845,
      "step": 4373
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2902147089384044,
      "learning_rate": 7.547715605633244e-06,
      "loss": 0.2143,
      "step": 4374
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3100434821869236,
      "learning_rate": 7.546600811643816e-06,
      "loss": 0.166,
      "step": 4375
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3688743154375775,
      "learning_rate": 7.54548584669336e-06,
      "loss": 0.1798,
      "step": 4376
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2147949320316396,
      "learning_rate": 7.544370710856728e-06,
      "loss": 0.1601,
      "step": 4377
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4781062551402164,
      "learning_rate": 7.54325540420878e-06,
      "loss": 0.1788,
      "step": 4378
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.478746977300905,
      "learning_rate": 7.542139926824391e-06,
      "loss": 0.2496,
      "step": 4379
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4278261139750605,
      "learning_rate": 7.541024278778446e-06,
      "loss": 0.2083,
      "step": 4380
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3173428410391212,
      "learning_rate": 7.539908460145844e-06,
      "loss": 0.1957,
      "step": 4381
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5169638296411432,
      "learning_rate": 7.53879247100149e-06,
      "loss": 0.2372,
      "step": 4382
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.246390859079121,
      "learning_rate": 7.537676311420305e-06,
      "loss": 0.1622,
      "step": 4383
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1803885613119514,
      "learning_rate": 7.536559981477221e-06,
      "loss": 0.1809,
      "step": 4384
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.393438701477145,
      "learning_rate": 7.53544348124718e-06,
      "loss": 0.2102,
      "step": 4385
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3470319236418413,
      "learning_rate": 7.534326810805135e-06,
      "loss": 0.2042,
      "step": 4386
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2060190386125407,
      "learning_rate": 7.533209970226054e-06,
      "loss": 0.217,
      "step": 4387
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5594906111192852,
      "learning_rate": 7.532092959584912e-06,
      "loss": 0.2583,
      "step": 4388
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.696617810064743,
      "learning_rate": 7.530975778956697e-06,
      "loss": 0.7041,
      "step": 4389
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2986880518500625,
      "learning_rate": 7.52985842841641e-06,
      "loss": 0.2131,
      "step": 4390
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2613738484088162,
      "learning_rate": 7.528740908039062e-06,
      "loss": 0.1975,
      "step": 4391
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3899847671131442,
      "learning_rate": 7.527623217899673e-06,
      "loss": 0.2284,
      "step": 4392
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3385992625893894,
      "learning_rate": 7.526505358073282e-06,
      "loss": 0.2255,
      "step": 4393
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.70737911324868,
      "learning_rate": 7.525387328634929e-06,
      "loss": 0.5338,
      "step": 4394
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6002234435781306,
      "learning_rate": 7.524269129659674e-06,
      "loss": 0.2256,
      "step": 4395
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.741025361112017,
      "learning_rate": 7.5231507612225814e-06,
      "loss": 0.5213,
      "step": 4396
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2689030575416727,
      "learning_rate": 7.522032223398736e-06,
      "loss": 0.1957,
      "step": 4397
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3010050325158051,
      "learning_rate": 7.520913516263229e-06,
      "loss": 0.1878,
      "step": 4398
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3775197801340422,
      "learning_rate": 7.519794639891153e-06,
      "loss": 0.2217,
      "step": 4399
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3418745999799693,
      "learning_rate": 7.5186755943576324e-06,
      "loss": 0.2221,
      "step": 4400
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.041694730428965,
      "learning_rate": 7.517556379737787e-06,
      "loss": 0.6515,
      "step": 4401
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6790298924946974,
      "learning_rate": 7.5164369961067555e-06,
      "loss": 0.2029,
      "step": 4402
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.550829115945779,
      "learning_rate": 7.51531744353968e-06,
      "loss": 0.1829,
      "step": 4403
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.275127835390056,
      "learning_rate": 7.514197722111727e-06,
      "loss": 0.1653,
      "step": 4404
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.4518442474388475,
      "learning_rate": 7.5130778318980614e-06,
      "loss": 0.6706,
      "step": 4405
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5544527304712086,
      "learning_rate": 7.511957772973868e-06,
      "loss": 0.2302,
      "step": 4406
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0893096500926531,
      "learning_rate": 7.5108375454143375e-06,
      "loss": 0.1694,
      "step": 4407
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2604724638638392,
      "learning_rate": 7.509717149294675e-06,
      "loss": 0.19,
      "step": 4408
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5297797021274266,
      "learning_rate": 7.508596584690096e-06,
      "loss": 0.2385,
      "step": 4409
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2810236237157906,
      "learning_rate": 7.5074758516758276e-06,
      "loss": 0.1811,
      "step": 4410
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2863407868743817,
      "learning_rate": 7.506354950327108e-06,
      "loss": 0.2402,
      "step": 4411
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3715753637275603,
      "learning_rate": 7.5052338807191875e-06,
      "loss": 0.199,
      "step": 4412
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3348736405952355,
      "learning_rate": 7.504112642927324e-06,
      "loss": 0.1711,
      "step": 4413
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.632058193267838,
      "learning_rate": 7.502991237026794e-06,
      "loss": 0.1985,
      "step": 4414
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4867185206000277,
      "learning_rate": 7.501869663092875e-06,
      "loss": 0.2013,
      "step": 4415
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.500792796518845,
      "learning_rate": 7.5007479212008685e-06,
      "loss": 0.1965,
      "step": 4416
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.432326021842077,
      "learning_rate": 7.499626011426077e-06,
      "loss": 0.2196,
      "step": 4417
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3485624477554385,
      "learning_rate": 7.498503933843816e-06,
      "loss": 0.1963,
      "step": 4418
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5399879571735526,
      "learning_rate": 7.497381688529417e-06,
      "loss": 0.2517,
      "step": 4419
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2981810327993155,
      "learning_rate": 7.496259275558218e-06,
      "loss": 0.2044,
      "step": 4420
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3588820883825423,
      "learning_rate": 7.4951366950055695e-06,
      "loss": 0.1934,
      "step": 4421
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.249244737874179,
      "learning_rate": 7.4940139469468364e-06,
      "loss": 0.1831,
      "step": 4422
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.4673879008021204,
      "learning_rate": 7.49289103145739e-06,
      "loss": 0.7449,
      "step": 4423
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3034845124653658,
      "learning_rate": 7.491767948612616e-06,
      "loss": 0.1974,
      "step": 4424
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1637042439452236,
      "learning_rate": 7.490644698487909e-06,
      "loss": 0.1844,
      "step": 4425
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4068526917031556,
      "learning_rate": 7.489521281158677e-06,
      "loss": 0.1695,
      "step": 4426
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.132983317815409,
      "learning_rate": 7.488397696700338e-06,
      "loss": 0.1771,
      "step": 4427
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4014519145012532,
      "learning_rate": 7.487273945188323e-06,
      "loss": 0.2177,
      "step": 4428
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2796188291075352,
      "learning_rate": 7.486150026698071e-06,
      "loss": 0.2217,
      "step": 4429
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.323096300133733,
      "learning_rate": 7.485025941305036e-06,
      "loss": 0.1713,
      "step": 4430
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4254883352883738,
      "learning_rate": 7.48390168908468e-06,
      "loss": 0.2323,
      "step": 4431
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.415031413768985,
      "learning_rate": 7.482777270112477e-06,
      "loss": 0.1785,
      "step": 4432
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4311722680785817,
      "learning_rate": 7.481652684463914e-06,
      "loss": 0.2344,
      "step": 4433
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3419537919599347,
      "learning_rate": 7.480527932214487e-06,
      "loss": 0.1723,
      "step": 4434
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3861655832970259,
      "learning_rate": 7.4794030134397055e-06,
      "loss": 0.19,
      "step": 4435
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5363110987871995,
      "learning_rate": 7.478277928215085e-06,
      "loss": 0.2127,
      "step": 4436
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3468114136713882,
      "learning_rate": 7.47715267661616e-06,
      "loss": 0.216,
      "step": 4437
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.397027404279438,
      "learning_rate": 7.476027258718469e-06,
      "loss": 0.2355,
      "step": 4438
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3836591345656608,
      "learning_rate": 7.4749016745975675e-06,
      "loss": 0.2317,
      "step": 4439
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3216636414724485,
      "learning_rate": 7.473775924329018e-06,
      "loss": 0.1963,
      "step": 4440
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.476499902791249,
      "learning_rate": 7.4726500079883935e-06,
      "loss": 0.1972,
      "step": 4441
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4891834781552717,
      "learning_rate": 7.471523925651285e-06,
      "loss": 0.2504,
      "step": 4442
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5221223543218547,
      "learning_rate": 7.4703976773932855e-06,
      "loss": 0.2362,
      "step": 4443
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3954100546000903,
      "learning_rate": 7.469271263290003e-06,
      "loss": 0.177,
      "step": 4444
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3337702311794086,
      "learning_rate": 7.468144683417061e-06,
      "loss": 0.2635,
      "step": 4445
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3478397701041323,
      "learning_rate": 7.467017937850088e-06,
      "loss": 0.21,
      "step": 4446
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.042519453332044,
      "learning_rate": 7.465891026664726e-06,
      "loss": 0.5139,
      "step": 4447
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3536591577352095,
      "learning_rate": 7.4647639499366275e-06,
      "loss": 0.1852,
      "step": 4448
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.467150668110183,
      "learning_rate": 7.463636707741458e-06,
      "loss": 0.2346,
      "step": 4449
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.294432452525009,
      "learning_rate": 7.462509300154892e-06,
      "loss": 0.1964,
      "step": 4450
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1554747495352153,
      "learning_rate": 7.461381727252615e-06,
      "loss": 0.1698,
      "step": 4451
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4096047780387313,
      "learning_rate": 7.4602539891103254e-06,
      "loss": 0.2462,
      "step": 4452
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3405073739596516,
      "learning_rate": 7.459126085803731e-06,
      "loss": 0.219,
      "step": 4453
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.221749352731346,
      "learning_rate": 7.4579980174085506e-06,
      "loss": 0.1588,
      "step": 4454
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3178386026120088,
      "learning_rate": 7.456869784000517e-06,
      "loss": 0.2234,
      "step": 4455
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3544038502193925,
      "learning_rate": 7.455741385655371e-06,
      "loss": 0.186,
      "step": 4456
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3475199779534466,
      "learning_rate": 7.454612822448865e-06,
      "loss": 0.1945,
      "step": 4457
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3011295094363582,
      "learning_rate": 7.453484094456761e-06,
      "loss": 0.1825,
      "step": 4458
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.213867702758694,
      "learning_rate": 7.45235520175484e-06,
      "loss": 0.4681,
      "step": 4459
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.306280923665334,
      "learning_rate": 7.4512261444188805e-06,
      "loss": 0.1911,
      "step": 4460
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3994551049174764,
      "learning_rate": 7.450096922524683e-06,
      "loss": 0.2697,
      "step": 4461
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.933099316204779,
      "learning_rate": 7.448967536148057e-06,
      "loss": 0.5556,
      "step": 4462
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.231002763921294,
      "learning_rate": 7.447837985364821e-06,
      "loss": 0.1646,
      "step": 4463
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.8783328863749436,
      "learning_rate": 7.446708270250803e-06,
      "loss": 0.609,
      "step": 4464
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.307839553343257,
      "learning_rate": 7.445578390881846e-06,
      "loss": 0.1977,
      "step": 4465
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6286465022839878,
      "learning_rate": 7.444448347333802e-06,
      "loss": 0.1956,
      "step": 4466
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3541853826039754,
      "learning_rate": 7.4433181396825334e-06,
      "loss": 0.236,
      "step": 4467
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3256808977320573,
      "learning_rate": 7.442187768003916e-06,
      "loss": 0.2146,
      "step": 4468
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6127973932410316,
      "learning_rate": 7.441057232373834e-06,
      "loss": 0.2438,
      "step": 4469
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.289192808470553,
      "learning_rate": 7.439926532868183e-06,
      "loss": 0.1672,
      "step": 4470
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3835273122625376,
      "learning_rate": 7.438795669562873e-06,
      "loss": 0.2173,
      "step": 4471
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2330883130982486,
      "learning_rate": 7.437664642533817e-06,
      "loss": 0.1971,
      "step": 4472
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.33202662008548,
      "learning_rate": 7.436533451856951e-06,
      "loss": 0.2005,
      "step": 4473
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3826294644597923,
      "learning_rate": 7.435402097608211e-06,
      "loss": 0.2169,
      "step": 4474
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4883545797424773,
      "learning_rate": 7.434270579863549e-06,
      "loss": 0.241,
      "step": 4475
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2649737180228018,
      "learning_rate": 7.433138898698927e-06,
      "loss": 0.1761,
      "step": 4476
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.378872533239398,
      "learning_rate": 7.43200705419032e-06,
      "loss": 0.2255,
      "step": 4477
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.332321703964259,
      "learning_rate": 7.430875046413709e-06,
      "loss": 0.1996,
      "step": 4478
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1082217396660838,
      "learning_rate": 7.429742875445092e-06,
      "loss": 0.1566,
      "step": 4479
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2631845816238365,
      "learning_rate": 7.428610541360475e-06,
      "loss": 0.1538,
      "step": 4480
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.288802401433073,
      "learning_rate": 7.427478044235872e-06,
      "loss": 0.1967,
      "step": 4481
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.37517899911686,
      "learning_rate": 7.426345384147313e-06,
      "loss": 0.2017,
      "step": 4482
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2862295081156347,
      "learning_rate": 7.425212561170838e-06,
      "loss": 0.1734,
      "step": 4483
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.262494639198926,
      "learning_rate": 7.424079575382495e-06,
      "loss": 0.1633,
      "step": 4484
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.60514519373858,
      "learning_rate": 7.422946426858346e-06,
      "loss": 0.6855,
      "step": 4485
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.282869454889024,
      "learning_rate": 7.421813115674461e-06,
      "loss": 0.1988,
      "step": 4486
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4138422959112926,
      "learning_rate": 7.420679641906925e-06,
      "loss": 0.1994,
      "step": 4487
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.405506923011502,
      "learning_rate": 7.4195460056318304e-06,
      "loss": 0.1993,
      "step": 4488
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2642286165144958,
      "learning_rate": 7.41841220692528e-06,
      "loss": 0.1658,
      "step": 4489
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3616416866587953,
      "learning_rate": 7.417278245863391e-06,
      "loss": 0.211,
      "step": 4490
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3285358714578297,
      "learning_rate": 7.4161441225222905e-06,
      "loss": 0.1575,
      "step": 4491
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4567799596116144,
      "learning_rate": 7.415009836978113e-06,
      "loss": 0.2361,
      "step": 4492
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5233089215006175,
      "learning_rate": 7.413875389307008e-06,
      "loss": 0.198,
      "step": 4493
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3885142700008979,
      "learning_rate": 7.412740779585137e-06,
      "loss": 0.2041,
      "step": 4494
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2577220357179733,
      "learning_rate": 7.411606007888665e-06,
      "loss": 0.1737,
      "step": 4495
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3335098064591036,
      "learning_rate": 7.410471074293774e-06,
      "loss": 0.2097,
      "step": 4496
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1894209347843252,
      "learning_rate": 7.409335978876657e-06,
      "loss": 0.2024,
      "step": 4497
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.298540103583728,
      "learning_rate": 7.408200721713517e-06,
      "loss": 0.1924,
      "step": 4498
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3886252595589323,
      "learning_rate": 7.407065302880564e-06,
      "loss": 0.2372,
      "step": 4499
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.398388547903395,
      "learning_rate": 7.405929722454026e-06,
      "loss": 0.1855,
      "step": 4500
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.52892238362788,
      "learning_rate": 7.404793980510135e-06,
      "loss": 0.1994,
      "step": 4501
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3153486417070632,
      "learning_rate": 7.4036580771251375e-06,
      "loss": 0.1828,
      "step": 4502
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2497810285844588,
      "learning_rate": 7.402522012375292e-06,
      "loss": 0.2037,
      "step": 4503
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5388946422378202,
      "learning_rate": 7.401385786336863e-06,
      "loss": 0.257,
      "step": 4504
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3151955903898773,
      "learning_rate": 7.4002493990861314e-06,
      "loss": 0.209,
      "step": 4505
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.412572195019964,
      "learning_rate": 7.399112850699384e-06,
      "loss": 0.1998,
      "step": 4506
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4964431110948937,
      "learning_rate": 7.397976141252923e-06,
      "loss": 0.244,
      "step": 4507
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4186408474924375,
      "learning_rate": 7.396839270823057e-06,
      "loss": 0.2129,
      "step": 4508
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5772421167161794,
      "learning_rate": 7.395702239486109e-06,
      "loss": 0.2617,
      "step": 4509
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.471832027951409,
      "learning_rate": 7.39456504731841e-06,
      "loss": 0.2346,
      "step": 4510
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3487777087599147,
      "learning_rate": 7.3934276943963065e-06,
      "loss": 0.2483,
      "step": 4511
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1724074669821147,
      "learning_rate": 7.39229018079615e-06,
      "loss": 0.1791,
      "step": 4512
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4691126837807114,
      "learning_rate": 7.391152506594304e-06,
      "loss": 0.2012,
      "step": 4513
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.121554013558425,
      "learning_rate": 7.390014671867145e-06,
      "loss": 0.1788,
      "step": 4514
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5676402516846408,
      "learning_rate": 7.3888766766910605e-06,
      "loss": 0.2129,
      "step": 4515
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4040702081770413,
      "learning_rate": 7.387738521142445e-06,
      "loss": 0.2457,
      "step": 4516
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4514816758232465,
      "learning_rate": 7.3866002052977096e-06,
      "loss": 0.2072,
      "step": 4517
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.454291681952188,
      "learning_rate": 7.385461729233271e-06,
      "loss": 0.2487,
      "step": 4518
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3710475888999527,
      "learning_rate": 7.384323093025558e-06,
      "loss": 0.2349,
      "step": 4519
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5358186265975866,
      "learning_rate": 7.383184296751014e-06,
      "loss": 0.215,
      "step": 4520
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.744461193022722,
      "learning_rate": 7.382045340486086e-06,
      "loss": 0.5168,
      "step": 4521
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2789260824127995,
      "learning_rate": 7.380906224307236e-06,
      "loss": 0.1946,
      "step": 4522
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3433354534259556,
      "learning_rate": 7.379766948290937e-06,
      "loss": 0.1858,
      "step": 4523
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.279715167027864,
      "learning_rate": 7.3786275125136745e-06,
      "loss": 0.1659,
      "step": 4524
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1274172792952344,
      "learning_rate": 7.3774879170519386e-06,
      "loss": 0.1359,
      "step": 4525
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2365332901154997,
      "learning_rate": 7.376348161982236e-06,
      "loss": 0.1399,
      "step": 4526
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1202514511877528,
      "learning_rate": 7.37520824738108e-06,
      "loss": 0.1812,
      "step": 4527
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2439707639750857,
      "learning_rate": 7.374068173325e-06,
      "loss": 0.1781,
      "step": 4528
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3935584089382431,
      "learning_rate": 7.37292793989053e-06,
      "loss": 0.1785,
      "step": 4529
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2242613495163874,
      "learning_rate": 7.371787547154215e-06,
      "loss": 0.1761,
      "step": 4530
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2769778507428013,
      "learning_rate": 7.370646995192618e-06,
      "loss": 0.1771,
      "step": 4531
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.816999362916801,
      "learning_rate": 7.369506284082306e-06,
      "loss": 0.5785,
      "step": 4532
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3920909405915207,
      "learning_rate": 7.3683654138998576e-06,
      "loss": 0.1887,
      "step": 4533
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.289094729493865,
      "learning_rate": 7.367224384721861e-06,
      "loss": 0.1824,
      "step": 4534
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.374885122074753,
      "learning_rate": 7.36608319662492e-06,
      "loss": 0.1966,
      "step": 4535
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.353939476108601,
      "learning_rate": 7.364941849685646e-06,
      "loss": 0.2219,
      "step": 4536
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.54695126040923,
      "learning_rate": 7.363800343980659e-06,
      "loss": 0.2192,
      "step": 4537
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4553934383941032,
      "learning_rate": 7.3626586795865936e-06,
      "loss": 0.2387,
      "step": 4538
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3398652503622337,
      "learning_rate": 7.361516856580093e-06,
      "loss": 0.1965,
      "step": 4539
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.239725004914783,
      "learning_rate": 7.36037487503781e-06,
      "loss": 0.1796,
      "step": 4540
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3907444317559783,
      "learning_rate": 7.35923273503641e-06,
      "loss": 0.2056,
      "step": 4541
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.790279110492094,
      "learning_rate": 7.358090436652568e-06,
      "loss": 0.7062,
      "step": 4542
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3839464487762783,
      "learning_rate": 7.356947979962972e-06,
      "loss": 0.2289,
      "step": 4543
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6018934204000703,
      "learning_rate": 7.355805365044314e-06,
      "loss": 0.2699,
      "step": 4544
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.131398515008712,
      "learning_rate": 7.3546625919733065e-06,
      "loss": 0.6112,
      "step": 4545
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4232478849224517,
      "learning_rate": 7.353519660826665e-06,
      "loss": 0.2216,
      "step": 4546
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4422033385752677,
      "learning_rate": 7.352376571681114e-06,
      "loss": 0.2172,
      "step": 4547
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.401986030227101,
      "learning_rate": 7.3512333246134e-06,
      "loss": 0.1858,
      "step": 4548
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2844177905785152,
      "learning_rate": 7.350089919700268e-06,
      "loss": 0.2014,
      "step": 4549
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3023427377276269,
      "learning_rate": 7.348946357018479e-06,
      "loss": 0.2172,
      "step": 4550
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2719449461518417,
      "learning_rate": 7.347802636644804e-06,
      "loss": 0.1875,
      "step": 4551
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.427846722096851,
      "learning_rate": 7.3466587586560245e-06,
      "loss": 0.2561,
      "step": 4552
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.380321257060329,
      "learning_rate": 7.345514723128932e-06,
      "loss": 0.2049,
      "step": 4553
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3924644565521185,
      "learning_rate": 7.344370530140331e-06,
      "loss": 0.187,
      "step": 4554
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3426902500988527,
      "learning_rate": 7.343226179767034e-06,
      "loss": 0.2185,
      "step": 4555
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4944494104212183,
      "learning_rate": 7.342081672085861e-06,
      "loss": 0.25,
      "step": 4556
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5009986963432371,
      "learning_rate": 7.340937007173651e-06,
      "loss": 0.2286,
      "step": 4557
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3147383534369999,
      "learning_rate": 7.339792185107245e-06,
      "loss": 0.2548,
      "step": 4558
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.721786663233292,
      "learning_rate": 7.338647205963502e-06,
      "loss": 0.5509,
      "step": 4559
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3787574653497223,
      "learning_rate": 7.337502069819285e-06,
      "loss": 0.2197,
      "step": 4560
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.66993794136623,
      "learning_rate": 7.336356776751472e-06,
      "loss": 0.6364,
      "step": 4561
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6200029166625864,
      "learning_rate": 7.335211326836951e-06,
      "loss": 0.2453,
      "step": 4562
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3473429751541997,
      "learning_rate": 7.334065720152617e-06,
      "loss": 0.2371,
      "step": 4563
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.296567216967196,
      "learning_rate": 7.3329199567753796e-06,
      "loss": 0.2246,
      "step": 4564
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4202835188076974,
      "learning_rate": 7.331774036782158e-06,
      "loss": 0.2257,
      "step": 4565
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.966647827600356,
      "learning_rate": 7.3306279602498785e-06,
      "loss": 0.5718,
      "step": 4566
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2209317147261978,
      "learning_rate": 7.329481727255484e-06,
      "loss": 0.2024,
      "step": 4567
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4261913915726283,
      "learning_rate": 7.328335337875921e-06,
      "loss": 0.169,
      "step": 4568
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.076045592292064,
      "learning_rate": 7.327188792188155e-06,
      "loss": 0.6056,
      "step": 4569
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2241143578854154,
      "learning_rate": 7.326042090269152e-06,
      "loss": 0.209,
      "step": 4570
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3288193003278035,
      "learning_rate": 7.324895232195898e-06,
      "loss": 0.2081,
      "step": 4571
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4444692075856738,
      "learning_rate": 7.32374821804538e-06,
      "loss": 0.2201,
      "step": 4572
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3493685407625815,
      "learning_rate": 7.322601047894607e-06,
      "loss": 0.2055,
      "step": 4573
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3575542573181467,
      "learning_rate": 7.321453721820586e-06,
      "loss": 0.1771,
      "step": 4574
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.301417514470607,
      "learning_rate": 7.320306239900343e-06,
      "loss": 0.1945,
      "step": 4575
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6454748602880356,
      "learning_rate": 7.319158602210913e-06,
      "loss": 0.2267,
      "step": 4576
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.128095455240721,
      "learning_rate": 7.318010808829339e-06,
      "loss": 0.6972,
      "step": 4577
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.781483942919324,
      "learning_rate": 7.316862859832675e-06,
      "loss": 0.5377,
      "step": 4578
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2912530661845676,
      "learning_rate": 7.315714755297988e-06,
      "loss": 0.1943,
      "step": 4579
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.401066836588219,
      "learning_rate": 7.314566495302353e-06,
      "loss": 0.2702,
      "step": 4580
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2645527815254818,
      "learning_rate": 7.313418079922858e-06,
      "loss": 0.1811,
      "step": 4581
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7328561206365583,
      "learning_rate": 7.3122695092365955e-06,
      "loss": 0.2325,
      "step": 4582
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.463693038641647,
      "learning_rate": 7.311120783320676e-06,
      "loss": 0.2039,
      "step": 4583
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2337237203382514,
      "learning_rate": 7.309971902252216e-06,
      "loss": 0.2036,
      "step": 4584
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4682135718410099,
      "learning_rate": 7.308822866108343e-06,
      "loss": 0.235,
      "step": 4585
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4977696648088044,
      "learning_rate": 7.307673674966196e-06,
      "loss": 0.2319,
      "step": 4586
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2840219102661012,
      "learning_rate": 7.306524328902923e-06,
      "loss": 0.1814,
      "step": 4587
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2594684362184525,
      "learning_rate": 7.305374827995684e-06,
      "loss": 0.1915,
      "step": 4588
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5247346409080598,
      "learning_rate": 7.304225172321646e-06,
      "loss": 0.2003,
      "step": 4589
    },
    {
      "epoch": 0.37,
      "grad_norm": 16.94837189189484,
      "learning_rate": 7.303075361957992e-06,
      "loss": 0.6169,
      "step": 4590
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3024394632314393,
      "learning_rate": 7.301925396981912e-06,
      "loss": 0.2076,
      "step": 4591
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4407981513835169,
      "learning_rate": 7.300775277470602e-06,
      "loss": 0.1901,
      "step": 4592
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.603719147363319,
      "learning_rate": 7.29962500350128e-06,
      "loss": 0.6748,
      "step": 4593
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.46696115351919,
      "learning_rate": 7.298474575151162e-06,
      "loss": 0.2155,
      "step": 4594
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3030749933965713,
      "learning_rate": 7.297323992497483e-06,
      "loss": 0.1432,
      "step": 4595
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3259812525285162,
      "learning_rate": 7.2961732556174826e-06,
      "loss": 0.1954,
      "step": 4596
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.141472081338822,
      "learning_rate": 7.295022364588414e-06,
      "loss": 0.6461,
      "step": 4597
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2892807824588552,
      "learning_rate": 7.293871319487542e-06,
      "loss": 0.1816,
      "step": 4598
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4347591466152476,
      "learning_rate": 7.292720120392137e-06,
      "loss": 0.2466,
      "step": 4599
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.386323116832039,
      "learning_rate": 7.291568767379484e-06,
      "loss": 0.5947,
      "step": 4600
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.9477206071544195,
      "learning_rate": 7.290417260526877e-06,
      "loss": 0.6105,
      "step": 4601
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4981837006776133,
      "learning_rate": 7.289265599911619e-06,
      "loss": 0.1852,
      "step": 4602
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4307276165576726,
      "learning_rate": 7.288113785611025e-06,
      "loss": 0.2225,
      "step": 4603
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2524610704747874,
      "learning_rate": 7.2869618177024204e-06,
      "loss": 0.1617,
      "step": 4604
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.468193183133097,
      "learning_rate": 7.2858096962631395e-06,
      "loss": 0.1569,
      "step": 4605
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2935021472727608,
      "learning_rate": 7.284657421370528e-06,
      "loss": 0.2136,
      "step": 4606
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.8583769118854763,
      "learning_rate": 7.283504993101942e-06,
      "loss": 0.3823,
      "step": 4607
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.501302182949338,
      "learning_rate": 7.282352411534747e-06,
      "loss": 0.2129,
      "step": 4608
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.401695867666399,
      "learning_rate": 7.281199676746318e-06,
      "loss": 0.2456,
      "step": 4609
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3906372708228258,
      "learning_rate": 7.280046788814045e-06,
      "loss": 0.194,
      "step": 4610
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4424034482372718,
      "learning_rate": 7.278893747815321e-06,
      "loss": 0.2461,
      "step": 4611
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4343741351432249,
      "learning_rate": 7.277740553827556e-06,
      "loss": 0.2205,
      "step": 4612
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5053663128084247,
      "learning_rate": 7.2765872069281655e-06,
      "loss": 0.2181,
      "step": 4613
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3327550329491824,
      "learning_rate": 7.275433707194579e-06,
      "loss": 0.2259,
      "step": 4614
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.489346800317658,
      "learning_rate": 7.274280054704232e-06,
      "loss": 0.6711,
      "step": 4615
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3095953620270868,
      "learning_rate": 7.273126249534576e-06,
      "loss": 0.2013,
      "step": 4616
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.765202329043797,
      "learning_rate": 7.271972291763066e-06,
      "loss": 0.2085,
      "step": 4617
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.309552408526066,
      "learning_rate": 7.270818181467174e-06,
      "loss": 0.2327,
      "step": 4618
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.369767798704067,
      "learning_rate": 7.269663918724376e-06,
      "loss": 0.1949,
      "step": 4619
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2143494109955377,
      "learning_rate": 7.268509503612162e-06,
      "loss": 0.2107,
      "step": 4620
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4333956883802197,
      "learning_rate": 7.267354936208032e-06,
      "loss": 0.2305,
      "step": 4621
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5204917150270796,
      "learning_rate": 7.266200216589495e-06,
      "loss": 0.211,
      "step": 4622
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3521035931488967,
      "learning_rate": 7.265045344834072e-06,
      "loss": 0.2046,
      "step": 4623
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4821187381043048,
      "learning_rate": 7.263890321019291e-06,
      "loss": 0.2448,
      "step": 4624
    },
    {
      "epoch": 0.37,
      "grad_norm": 13.472179561909549,
      "learning_rate": 7.262735145222696e-06,
      "loss": 0.6137,
      "step": 4625
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.348783700749346,
      "learning_rate": 7.2615798175218325e-06,
      "loss": 0.2154,
      "step": 4626
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4442483710464578,
      "learning_rate": 7.260424337994263e-06,
      "loss": 0.1952,
      "step": 4627
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4294802902806512,
      "learning_rate": 7.2592687067175605e-06,
      "loss": 0.2687,
      "step": 4628
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.404617299354495,
      "learning_rate": 7.2581129237693045e-06,
      "loss": 0.2042,
      "step": 4629
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3925394151137753,
      "learning_rate": 7.256956989227084e-06,
      "loss": 0.2181,
      "step": 4630
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3489965299530309,
      "learning_rate": 7.255800903168505e-06,
      "loss": 0.1824,
      "step": 4631
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4067110607303004,
      "learning_rate": 7.254644665671176e-06,
      "loss": 0.2007,
      "step": 4632
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2825144348655453,
      "learning_rate": 7.253488276812719e-06,
      "loss": 0.218,
      "step": 4633
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4814934847448538,
      "learning_rate": 7.252331736670765e-06,
      "loss": 0.2164,
      "step": 4634
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3947338488963392,
      "learning_rate": 7.251175045322959e-06,
      "loss": 0.2225,
      "step": 4635
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2306781525582993,
      "learning_rate": 7.250018202846951e-06,
      "loss": 0.2002,
      "step": 4636
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2884757484626046,
      "learning_rate": 7.2488612093204025e-06,
      "loss": 0.2077,
      "step": 4637
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.44123447967655,
      "learning_rate": 7.247704064820989e-06,
      "loss": 0.2385,
      "step": 4638
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.413843015550043,
      "learning_rate": 7.24654676942639e-06,
      "loss": 0.2326,
      "step": 4639
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2895788492650495,
      "learning_rate": 7.245389323214301e-06,
      "loss": 0.2288,
      "step": 4640
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2367232140865478,
      "learning_rate": 7.244231726262422e-06,
      "loss": 0.2112,
      "step": 4641
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.512118395217802,
      "learning_rate": 7.243073978648469e-06,
      "loss": 0.2033,
      "step": 4642
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.250700487481848,
      "learning_rate": 7.241916080450163e-06,
      "loss": 0.1657,
      "step": 4643
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1311745688189312,
      "learning_rate": 7.240758031745237e-06,
      "loss": 0.1688,
      "step": 4644
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4022151033737413,
      "learning_rate": 7.2395998326114345e-06,
      "loss": 0.2099,
      "step": 4645
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3712520212644563,
      "learning_rate": 7.238441483126512e-06,
      "loss": 0.2309,
      "step": 4646
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3075430382988562,
      "learning_rate": 7.237282983368228e-06,
      "loss": 0.2121,
      "step": 4647
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.300659953498008,
      "learning_rate": 7.236124333414359e-06,
      "loss": 0.183,
      "step": 4648
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4419638416751694,
      "learning_rate": 7.2349655333426905e-06,
      "loss": 0.2182,
      "step": 4649
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3356492464479306,
      "learning_rate": 7.233806583231012e-06,
      "loss": 0.2018,
      "step": 4650
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3249028716378075,
      "learning_rate": 7.23264748315713e-06,
      "loss": 0.2071,
      "step": 4651
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3828948765923208,
      "learning_rate": 7.231488233198858e-06,
      "loss": 0.2024,
      "step": 4652
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3766039057922546,
      "learning_rate": 7.2303288334340205e-06,
      "loss": 0.2122,
      "step": 4653
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5144341514830022,
      "learning_rate": 7.2291692839404495e-06,
      "loss": 0.253,
      "step": 4654
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2494676684544648,
      "learning_rate": 7.22800958479599e-06,
      "loss": 0.1987,
      "step": 4655
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2612132673659266,
      "learning_rate": 7.226849736078497e-06,
      "loss": 0.2011,
      "step": 4656
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3699986513963234,
      "learning_rate": 7.225689737865835e-06,
      "loss": 0.2314,
      "step": 4657
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.558499070842386,
      "learning_rate": 7.224529590235876e-06,
      "loss": 0.2141,
      "step": 4658
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3528678116321762,
      "learning_rate": 7.223369293266507e-06,
      "loss": 0.1664,
      "step": 4659
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3962024985135608,
      "learning_rate": 7.222208847035621e-06,
      "loss": 0.2215,
      "step": 4660
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4055564085365457,
      "learning_rate": 7.22104825162112e-06,
      "loss": 0.2087,
      "step": 4661
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4351363404772675,
      "learning_rate": 7.219887507100921e-06,
      "loss": 0.2296,
      "step": 4662
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.366196893853091,
      "learning_rate": 7.218726613552949e-06,
      "loss": 0.2065,
      "step": 4663
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3763920825696134,
      "learning_rate": 7.217565571055137e-06,
      "loss": 0.2253,
      "step": 4664
    },
    {
      "epoch": 0.37,
      "grad_norm": 14.146982778521922,
      "learning_rate": 7.216404379685427e-06,
      "loss": 0.5625,
      "step": 4665
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4421753066478902,
      "learning_rate": 7.2152430395217765e-06,
      "loss": 0.229,
      "step": 4666
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4464948217931586,
      "learning_rate": 7.214081550642149e-06,
      "loss": 0.2479,
      "step": 4667
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4656849620299957,
      "learning_rate": 7.212919913124519e-06,
      "loss": 0.1908,
      "step": 4668
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.398624504276929,
      "learning_rate": 7.211758127046871e-06,
      "loss": 0.2151,
      "step": 4669
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.856083965265127,
      "learning_rate": 7.210596192487198e-06,
      "loss": 0.6346,
      "step": 4670
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3977669365205103,
      "learning_rate": 7.2094341095235035e-06,
      "loss": 0.2578,
      "step": 4671
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3688436603958962,
      "learning_rate": 7.208271878233804e-06,
      "loss": 0.2382,
      "step": 4672
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4066932807716008,
      "learning_rate": 7.207109498696122e-06,
      "loss": 0.2074,
      "step": 4673
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3922291847099941,
      "learning_rate": 7.205946970988493e-06,
      "loss": 0.1953,
      "step": 4674
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.466403899970326,
      "learning_rate": 7.204784295188959e-06,
      "loss": 0.2406,
      "step": 4675
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.537515448389777,
      "learning_rate": 7.203621471375576e-06,
      "loss": 0.5576,
      "step": 4676
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3274543529146274,
      "learning_rate": 7.202458499626408e-06,
      "loss": 0.2067,
      "step": 4677
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3956618814986548,
      "learning_rate": 7.201295380019525e-06,
      "loss": 0.2263,
      "step": 4678
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.377930505504033,
      "learning_rate": 7.200132112633015e-06,
      "loss": 0.6434,
      "step": 4679
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4321209132549046,
      "learning_rate": 7.19896869754497e-06,
      "loss": 0.2185,
      "step": 4680
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4922829482083375,
      "learning_rate": 7.197805134833493e-06,
      "loss": 0.2409,
      "step": 4681
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3111282877814505,
      "learning_rate": 7.196641424576698e-06,
      "loss": 0.1975,
      "step": 4682
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2440097550478715,
      "learning_rate": 7.19547756685271e-06,
      "loss": 0.2211,
      "step": 4683
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3409633376468209,
      "learning_rate": 7.194313561739661e-06,
      "loss": 0.1877,
      "step": 4684
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3690105212856414,
      "learning_rate": 7.193149409315694e-06,
      "loss": 0.2287,
      "step": 4685
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2689817910497139,
      "learning_rate": 7.191985109658961e-06,
      "loss": 0.213,
      "step": 4686
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3767022262520805,
      "learning_rate": 7.190820662847627e-06,
      "loss": 0.2426,
      "step": 4687
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4424232108778996,
      "learning_rate": 7.189656068959863e-06,
      "loss": 0.2068,
      "step": 4688
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4860548701268863,
      "learning_rate": 7.188491328073853e-06,
      "loss": 0.2362,
      "step": 4689
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5343218494259938,
      "learning_rate": 7.18732644026779e-06,
      "loss": 0.1956,
      "step": 4690
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2628542747039755,
      "learning_rate": 7.186161405619876e-06,
      "loss": 0.1619,
      "step": 4691
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5452611344458735,
      "learning_rate": 7.1849962242083215e-06,
      "loss": 0.2289,
      "step": 4692
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.290376781748268,
      "learning_rate": 7.183830896111351e-06,
      "loss": 0.2129,
      "step": 4693
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1708075625985892,
      "learning_rate": 7.182665421407196e-06,
      "loss": 0.2029,
      "step": 4694
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2738470024004076,
      "learning_rate": 7.181499800174099e-06,
      "loss": 0.174,
      "step": 4695
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2877942716957311,
      "learning_rate": 7.1803340324903096e-06,
      "loss": 0.2388,
      "step": 4696
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4709086675119016,
      "learning_rate": 7.17916811843409e-06,
      "loss": 0.1986,
      "step": 4697
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3994313067924706,
      "learning_rate": 7.1780020580837125e-06,
      "loss": 0.2029,
      "step": 4698
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2911034152124485,
      "learning_rate": 7.176835851517458e-06,
      "loss": 0.189,
      "step": 4699
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3727867959853184,
      "learning_rate": 7.1756694988136165e-06,
      "loss": 0.2316,
      "step": 4700
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.43053224914512,
      "learning_rate": 7.17450300005049e-06,
      "loss": 0.2163,
      "step": 4701
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.110308618160761,
      "learning_rate": 7.17333635530639e-06,
      "loss": 0.1577,
      "step": 4702
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2549224998408561,
      "learning_rate": 7.1721695646596335e-06,
      "loss": 0.2215,
      "step": 4703
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3723128666652684,
      "learning_rate": 7.171002628188554e-06,
      "loss": 0.2601,
      "step": 4704
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2545082552861848,
      "learning_rate": 7.16983554597149e-06,
      "loss": 0.2007,
      "step": 4705
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5267378614042626,
      "learning_rate": 7.16866831808679e-06,
      "loss": 0.2346,
      "step": 4706
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3893049688525259,
      "learning_rate": 7.167500944612815e-06,
      "loss": 0.2244,
      "step": 4707
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.256496702377986,
      "learning_rate": 7.166333425627936e-06,
      "loss": 0.1872,
      "step": 4708
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3895559819080734,
      "learning_rate": 7.165165761210529e-06,
      "loss": 0.2506,
      "step": 4709
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4527652774161193,
      "learning_rate": 7.163997951438986e-06,
      "loss": 0.2436,
      "step": 4710
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.5130824959754845,
      "learning_rate": 7.162829996391702e-06,
      "loss": 0.7399,
      "step": 4711
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3606716588125523,
      "learning_rate": 7.161661896147088e-06,
      "loss": 0.2269,
      "step": 4712
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4586771135839935,
      "learning_rate": 7.160493650783562e-06,
      "loss": 0.2084,
      "step": 4713
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5036539250450593,
      "learning_rate": 7.159325260379551e-06,
      "loss": 0.2258,
      "step": 4714
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5168111329496314,
      "learning_rate": 7.158156725013493e-06,
      "loss": 0.2066,
      "step": 4715
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.323394576935599,
      "learning_rate": 7.156988044763833e-06,
      "loss": 0.1945,
      "step": 4716
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3229632078055302,
      "learning_rate": 7.155819219709032e-06,
      "loss": 0.209,
      "step": 4717
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.772106145246674,
      "learning_rate": 7.154650249927555e-06,
      "loss": 0.4629,
      "step": 4718
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4597883991305334,
      "learning_rate": 7.153481135497878e-06,
      "loss": 0.2366,
      "step": 4719
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3743141889587922,
      "learning_rate": 7.152311876498487e-06,
      "loss": 0.2011,
      "step": 4720
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4131271019310823,
      "learning_rate": 7.15114247300788e-06,
      "loss": 0.2255,
      "step": 4721
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5259833897546178,
      "learning_rate": 7.14997292510456e-06,
      "loss": 0.252,
      "step": 4722
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3028279028756722,
      "learning_rate": 7.148803232867043e-06,
      "loss": 0.1944,
      "step": 4723
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.21179511273452,
      "learning_rate": 7.147633396373855e-06,
      "loss": 0.2178,
      "step": 4724
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.390477745753487,
      "learning_rate": 7.14646341570353e-06,
      "loss": 0.2154,
      "step": 4725
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3760584100121604,
      "learning_rate": 7.145293290934613e-06,
      "loss": 0.2051,
      "step": 4726
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3856957853375935,
      "learning_rate": 7.144123022145655e-06,
      "loss": 0.1745,
      "step": 4727
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.182377080155907,
      "learning_rate": 7.1429526094152244e-06,
      "loss": 0.1751,
      "step": 4728
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.093060848013929,
      "learning_rate": 7.141782052821892e-06,
      "loss": 0.5751,
      "step": 4729
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3337009627534544,
      "learning_rate": 7.14061135244424e-06,
      "loss": 0.1973,
      "step": 4730
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.565877356770094,
      "learning_rate": 7.139440508360863e-06,
      "loss": 0.4938,
      "step": 4731
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.426887224584748,
      "learning_rate": 7.138269520650362e-06,
      "loss": 0.7291,
      "step": 4732
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3691091137941518,
      "learning_rate": 7.137098389391349e-06,
      "loss": 0.2296,
      "step": 4733
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3661731354440267,
      "learning_rate": 7.135927114662448e-06,
      "loss": 0.1818,
      "step": 4734
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.258596824728972,
      "learning_rate": 7.134755696542286e-06,
      "loss": 0.1695,
      "step": 4735
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.774615072980027,
      "learning_rate": 7.133584135109507e-06,
      "loss": 0.47,
      "step": 4736
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1692197188097482,
      "learning_rate": 7.132412430442759e-06,
      "loss": 0.2076,
      "step": 4737
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2800107703228183,
      "learning_rate": 7.131240582620706e-06,
      "loss": 0.2037,
      "step": 4738
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3722555093795468,
      "learning_rate": 7.130068591722015e-06,
      "loss": 0.1921,
      "step": 4739
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2223374021938294,
      "learning_rate": 7.128896457825364e-06,
      "loss": 0.1446,
      "step": 4740
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2453630118627566,
      "learning_rate": 7.127724181009444e-06,
      "loss": 0.2061,
      "step": 4741
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4541515528343412,
      "learning_rate": 7.126551761352953e-06,
      "loss": 0.2114,
      "step": 4742
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4615846206925147,
      "learning_rate": 7.1253791989346e-06,
      "loss": 0.1727,
      "step": 4743
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3042964921822728,
      "learning_rate": 7.1242064938331e-06,
      "loss": 0.164,
      "step": 4744
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4439156921952279,
      "learning_rate": 7.123033646127183e-06,
      "loss": 0.201,
      "step": 4745
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.328002126416615,
      "learning_rate": 7.121860655895585e-06,
      "loss": 0.2016,
      "step": 4746
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3894458085701467,
      "learning_rate": 7.120687523217052e-06,
      "loss": 0.1808,
      "step": 4747
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.369086271262368,
      "learning_rate": 7.11951424817034e-06,
      "loss": 0.2171,
      "step": 4748
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2821568227180404,
      "learning_rate": 7.118340830834216e-06,
      "loss": 0.1929,
      "step": 4749
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2088893799608302,
      "learning_rate": 7.117167271287453e-06,
      "loss": 0.181,
      "step": 4750
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4390116330684632,
      "learning_rate": 7.1159935696088346e-06,
      "loss": 0.2371,
      "step": 4751
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4307431707713578,
      "learning_rate": 7.114819725877158e-06,
      "loss": 0.1923,
      "step": 4752
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2753073620623494,
      "learning_rate": 7.1136457401712255e-06,
      "loss": 0.1893,
      "step": 4753
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.549165364491072,
      "learning_rate": 7.112471612569849e-06,
      "loss": 0.219,
      "step": 4754
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3879968085204364,
      "learning_rate": 7.111297343151854e-06,
      "loss": 0.2104,
      "step": 4755
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5127874215402446,
      "learning_rate": 7.110122931996073e-06,
      "loss": 0.2197,
      "step": 4756
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.764811609584851,
      "learning_rate": 7.108948379181343e-06,
      "loss": 0.5671,
      "step": 4757
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3628419901696498,
      "learning_rate": 7.107773684786521e-06,
      "loss": 0.2182,
      "step": 4758
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2471538620607514,
      "learning_rate": 7.106598848890464e-06,
      "loss": 0.1976,
      "step": 4759
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1299104828149948,
      "learning_rate": 7.105423871572043e-06,
      "loss": 0.1588,
      "step": 4760
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2555138247463418,
      "learning_rate": 7.104248752910139e-06,
      "loss": 0.1995,
      "step": 4761
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4604070456468663,
      "learning_rate": 7.1030734929836405e-06,
      "loss": 0.1835,
      "step": 4762
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3672598271617655,
      "learning_rate": 7.101898091871447e-06,
      "loss": 0.1907,
      "step": 4763
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1692413804108257,
      "learning_rate": 7.100722549652465e-06,
      "loss": 0.1414,
      "step": 4764
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.421086934454808,
      "learning_rate": 7.0995468664056135e-06,
      "loss": 0.2412,
      "step": 4765
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.301400006914409,
      "learning_rate": 7.0983710422098205e-06,
      "loss": 0.6559,
      "step": 4766
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2478468570576833,
      "learning_rate": 7.097195077144021e-06,
      "loss": 0.201,
      "step": 4767
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4117594439537449,
      "learning_rate": 7.096018971287161e-06,
      "loss": 0.2772,
      "step": 4768
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3213733827334329,
      "learning_rate": 7.094842724718197e-06,
      "loss": 0.1842,
      "step": 4769
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3543254643213234,
      "learning_rate": 7.093666337516094e-06,
      "loss": 0.1704,
      "step": 4770
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5336305097787688,
      "learning_rate": 7.0924898097598276e-06,
      "loss": 0.2247,
      "step": 4771
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3011424030111844,
      "learning_rate": 7.091313141528378e-06,
      "loss": 0.2299,
      "step": 4772
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3583807176306357,
      "learning_rate": 7.0901363329007435e-06,
      "loss": 0.2197,
      "step": 4773
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3458212775468819,
      "learning_rate": 7.0889593839559215e-06,
      "loss": 0.2267,
      "step": 4774
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2855340372452824,
      "learning_rate": 7.0877822947729265e-06,
      "loss": 0.227,
      "step": 4775
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2530404128884884,
      "learning_rate": 7.0866050654307824e-06,
      "loss": 0.1683,
      "step": 4776
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3471158778308971,
      "learning_rate": 7.085427696008517e-06,
      "loss": 0.189,
      "step": 4777
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.162521903821187,
      "learning_rate": 7.084250186585172e-06,
      "loss": 0.172,
      "step": 4778
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2259662055855378,
      "learning_rate": 7.0830725372397965e-06,
      "loss": 0.1842,
      "step": 4779
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2164696392224137,
      "learning_rate": 7.081894748051451e-06,
      "loss": 0.1582,
      "step": 4780
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.29867431099379,
      "learning_rate": 7.080716819099204e-06,
      "loss": 0.2423,
      "step": 4781
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3378007552101296,
      "learning_rate": 7.07953875046213e-06,
      "loss": 0.1966,
      "step": 4782
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6797732265410092,
      "learning_rate": 7.0783605422193215e-06,
      "loss": 0.2729,
      "step": 4783
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2926198085286513,
      "learning_rate": 7.077182194449872e-06,
      "loss": 0.1895,
      "step": 4784
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2952148211658299,
      "learning_rate": 7.0760037072328855e-06,
      "loss": 0.2561,
      "step": 4785
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6157629065690955,
      "learning_rate": 7.074825080647483e-06,
      "loss": 0.1943,
      "step": 4786
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4344836758916617,
      "learning_rate": 7.073646314772787e-06,
      "loss": 0.214,
      "step": 4787
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2526146504703244,
      "learning_rate": 7.072467409687928e-06,
      "loss": 0.1859,
      "step": 4788
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2979609127143175,
      "learning_rate": 7.071288365472054e-06,
      "loss": 0.2045,
      "step": 4789
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.290462777531266,
      "learning_rate": 7.070109182204317e-06,
      "loss": 0.1904,
      "step": 4790
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3377588283983977,
      "learning_rate": 7.068929859963878e-06,
      "loss": 0.262,
      "step": 4791
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3476943852152492,
      "learning_rate": 7.067750398829908e-06,
      "loss": 0.2208,
      "step": 4792
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.6480179868341045,
      "learning_rate": 7.066570798881589e-06,
      "loss": 0.69,
      "step": 4793
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2624266267338566,
      "learning_rate": 7.065391060198111e-06,
      "loss": 0.1934,
      "step": 4794
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4228929032171018,
      "learning_rate": 7.064211182858673e-06,
      "loss": 0.2257,
      "step": 4795
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2566747916226606,
      "learning_rate": 7.063031166942485e-06,
      "loss": 0.1736,
      "step": 4796
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.791828552918496,
      "learning_rate": 7.0618510125287636e-06,
      "loss": 0.5563,
      "step": 4797
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3325220313677701,
      "learning_rate": 7.0606707196967365e-06,
      "loss": 0.1984,
      "step": 4798
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6180852709667057,
      "learning_rate": 7.05949028852564e-06,
      "loss": 0.2418,
      "step": 4799
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1923736574734867,
      "learning_rate": 7.05830971909472e-06,
      "loss": 0.1984,
      "step": 4800
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2093112425977857,
      "learning_rate": 7.057129011483233e-06,
      "loss": 0.194,
      "step": 4801
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1465677515140582,
      "learning_rate": 7.055948165770442e-06,
      "loss": 0.174,
      "step": 4802
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.32073674114426,
      "learning_rate": 7.054767182035622e-06,
      "loss": 0.7384,
      "step": 4803
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.245348169906903,
      "learning_rate": 7.053586060358056e-06,
      "loss": 0.1633,
      "step": 4804
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.680001709333162,
      "learning_rate": 7.0524048008170345e-06,
      "loss": 0.6267,
      "step": 4805
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3283813129000381,
      "learning_rate": 7.051223403491859e-06,
      "loss": 0.1814,
      "step": 4806
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3457247207207927,
      "learning_rate": 7.050041868461843e-06,
      "loss": 0.1754,
      "step": 4807
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2312384006678254,
      "learning_rate": 7.048860195806306e-06,
      "loss": 0.1778,
      "step": 4808
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5873004543021405,
      "learning_rate": 7.047678385604575e-06,
      "loss": 0.2056,
      "step": 4809
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4385135595250433,
      "learning_rate": 7.04649643793599e-06,
      "loss": 0.2038,
      "step": 4810
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.36835698222397,
      "learning_rate": 7.045314352879898e-06,
      "loss": 0.6407,
      "step": 4811
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.355633131986204,
      "learning_rate": 7.044132130515657e-06,
      "loss": 0.5287,
      "step": 4812
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.494963153879723,
      "learning_rate": 7.042949770922631e-06,
      "loss": 0.2181,
      "step": 4813
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.233453748166965,
      "learning_rate": 7.041767274180199e-06,
      "loss": 0.2066,
      "step": 4814
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5486545767832707,
      "learning_rate": 7.040584640367744e-06,
      "loss": 0.235,
      "step": 4815
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.406417972631745,
      "learning_rate": 7.039401869564657e-06,
      "loss": 0.7878,
      "step": 4816
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3015302768207249,
      "learning_rate": 7.038218961850346e-06,
      "loss": 0.2044,
      "step": 4817
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3787175176541557,
      "learning_rate": 7.037035917304219e-06,
      "loss": 0.2128,
      "step": 4818
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3499394036573094,
      "learning_rate": 7.035852736005698e-06,
      "loss": 0.1867,
      "step": 4819
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3702685278899334,
      "learning_rate": 7.034669418034217e-06,
      "loss": 0.2448,
      "step": 4820
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4141019983180052,
      "learning_rate": 7.033485963469211e-06,
      "loss": 0.1797,
      "step": 4821
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2591278873935836,
      "learning_rate": 7.032302372390132e-06,
      "loss": 0.167,
      "step": 4822
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3123881388421341,
      "learning_rate": 7.031118644876436e-06,
      "loss": 0.2309,
      "step": 4823
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.935948876499589,
      "learning_rate": 7.029934781007592e-06,
      "loss": 0.5871,
      "step": 4824
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3543901889964152,
      "learning_rate": 7.028750780863078e-06,
      "loss": 0.2025,
      "step": 4825
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3473654162878381,
      "learning_rate": 7.0275666445223745e-06,
      "loss": 0.2106,
      "step": 4826
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1615918620167034,
      "learning_rate": 7.0263823720649795e-06,
      "loss": 0.167,
      "step": 4827
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4936536838354497,
      "learning_rate": 7.025197963570396e-06,
      "loss": 0.2358,
      "step": 4828
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4285922221885665,
      "learning_rate": 7.024013419118138e-06,
      "loss": 0.1781,
      "step": 4829
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2629307023708323,
      "learning_rate": 7.022828738787725e-06,
      "loss": 0.224,
      "step": 4830
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4305422882932475,
      "learning_rate": 7.021643922658691e-06,
      "loss": 0.2011,
      "step": 4831
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5108733104670857,
      "learning_rate": 7.020458970810575e-06,
      "loss": 0.2459,
      "step": 4832
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7227817553121723,
      "learning_rate": 7.0192738833229255e-06,
      "loss": 0.2525,
      "step": 4833
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2802231535873252,
      "learning_rate": 7.018088660275304e-06,
      "loss": 0.2019,
      "step": 4834
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2786367616500365,
      "learning_rate": 7.016903301747275e-06,
      "loss": 0.1825,
      "step": 4835
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2597425755505227,
      "learning_rate": 7.015717807818417e-06,
      "loss": 0.2016,
      "step": 4836
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4209919238692725,
      "learning_rate": 7.014532178568314e-06,
      "loss": 0.2134,
      "step": 4837
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3585054663107867,
      "learning_rate": 7.013346414076563e-06,
      "loss": 0.2049,
      "step": 4838
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.069209179081909,
      "learning_rate": 7.012160514422767e-06,
      "loss": 0.5267,
      "step": 4839
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5546811833164766,
      "learning_rate": 7.010974479686538e-06,
      "loss": 0.244,
      "step": 4840
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.507484296014732,
      "learning_rate": 7.0097883099475e-06,
      "loss": 0.2602,
      "step": 4841
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1623343477439738,
      "learning_rate": 7.008602005285283e-06,
      "loss": 0.1758,
      "step": 4842
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2558409497821823,
      "learning_rate": 7.007415565779528e-06,
      "loss": 0.2332,
      "step": 4843
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.346866654961198,
      "learning_rate": 7.006228991509885e-06,
      "loss": 0.1975,
      "step": 4844
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4330432859433306,
      "learning_rate": 7.005042282556009e-06,
      "loss": 0.1796,
      "step": 4845
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2223240963082276,
      "learning_rate": 7.003855438997571e-06,
      "loss": 0.1671,
      "step": 4846
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3870883713136453,
      "learning_rate": 7.002668460914244e-06,
      "loss": 0.2097,
      "step": 4847
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4439850347518288,
      "learning_rate": 7.001481348385718e-06,
      "loss": 0.2389,
      "step": 4848
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.287013176096537,
      "learning_rate": 7.000294101491683e-06,
      "loss": 0.206,
      "step": 4849
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3347672372789836,
      "learning_rate": 6.999106720311846e-06,
      "loss": 0.1841,
      "step": 4850
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.303674612482775,
      "learning_rate": 6.997919204925916e-06,
      "loss": 0.1595,
      "step": 4851
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2929481381648444,
      "learning_rate": 6.9967315554136186e-06,
      "loss": 0.2169,
      "step": 4852
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.441351601225843,
      "learning_rate": 6.995543771854681e-06,
      "loss": 0.2194,
      "step": 4853
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3851054356495633,
      "learning_rate": 6.994355854328842e-06,
      "loss": 0.1956,
      "step": 4854
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3476074353276983,
      "learning_rate": 6.993167802915854e-06,
      "loss": 0.1843,
      "step": 4855
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3048652297547558,
      "learning_rate": 6.99197961769547e-06,
      "loss": 0.2116,
      "step": 4856
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4561127090797057,
      "learning_rate": 6.990791298747461e-06,
      "loss": 0.2263,
      "step": 4857
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.474774556199081,
      "learning_rate": 6.989602846151599e-06,
      "loss": 0.2593,
      "step": 4858
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4782006444187838,
      "learning_rate": 6.98841425998767e-06,
      "loss": 0.2089,
      "step": 4859
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2483069291308928,
      "learning_rate": 6.987225540335467e-06,
      "loss": 0.1546,
      "step": 4860
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.301715452053872,
      "learning_rate": 6.986036687274793e-06,
      "loss": 0.1896,
      "step": 4861
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4783805081667416,
      "learning_rate": 6.984847700885459e-06,
      "loss": 0.2194,
      "step": 4862
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5851414386900278,
      "learning_rate": 6.983658581247285e-06,
      "loss": 0.251,
      "step": 4863
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.04631605323419,
      "learning_rate": 6.982469328440098e-06,
      "loss": 0.5197,
      "step": 4864
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2272960603223113,
      "learning_rate": 6.981279942543741e-06,
      "loss": 0.1622,
      "step": 4865
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2746854881956649,
      "learning_rate": 6.980090423638057e-06,
      "loss": 0.209,
      "step": 4866
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.206302545876208,
      "learning_rate": 6.978900771802903e-06,
      "loss": 0.6047,
      "step": 4867
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2198123179720712,
      "learning_rate": 6.9777109871181444e-06,
      "loss": 0.1895,
      "step": 4868
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2951570853463479,
      "learning_rate": 6.976521069663657e-06,
      "loss": 0.2139,
      "step": 4869
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.337000024808618,
      "learning_rate": 6.975331019519322e-06,
      "loss": 0.2112,
      "step": 4870
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4741104816399193,
      "learning_rate": 6.974140836765028e-06,
      "loss": 0.2158,
      "step": 4871
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.2876117875760857,
      "learning_rate": 6.972950521480679e-06,
      "loss": 0.6419,
      "step": 4872
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.261432132399821,
      "learning_rate": 6.971760073746183e-06,
      "loss": 0.167,
      "step": 4873
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4445220320859973,
      "learning_rate": 6.9705694936414614e-06,
      "loss": 0.2412,
      "step": 4874
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.69626174492105,
      "learning_rate": 6.969378781246436e-06,
      "loss": 0.7072,
      "step": 4875
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.93499438990244,
      "learning_rate": 6.968187936641049e-06,
      "loss": 0.6157,
      "step": 4876
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3284789234946335,
      "learning_rate": 6.96699695990524e-06,
      "loss": 0.1861,
      "step": 4877
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3700035370714105,
      "learning_rate": 6.965805851118967e-06,
      "loss": 0.2203,
      "step": 4878
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.292692702839124,
      "learning_rate": 6.96461461036219e-06,
      "loss": 0.1876,
      "step": 4879
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5352402568945485,
      "learning_rate": 6.9634232377148835e-06,
      "loss": 0.2081,
      "step": 4880
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3871809061173839,
      "learning_rate": 6.9622317332570235e-06,
      "loss": 0.2425,
      "step": 4881
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4876367943419888,
      "learning_rate": 6.961040097068603e-06,
      "loss": 0.2569,
      "step": 4882
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4130185792218803,
      "learning_rate": 6.959848329229619e-06,
      "loss": 0.2315,
      "step": 4883
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3432689621067724,
      "learning_rate": 6.95865642982008e-06,
      "loss": 0.2241,
      "step": 4884
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1628376790757289,
      "learning_rate": 6.957464398919998e-06,
      "loss": 0.1642,
      "step": 4885
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2912117106395649,
      "learning_rate": 6.956272236609402e-06,
      "loss": 0.2048,
      "step": 4886
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.2493178095540864,
      "learning_rate": 6.955079942968324e-06,
      "loss": 0.5668,
      "step": 4887
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4202703870738962,
      "learning_rate": 6.9538875180768025e-06,
      "loss": 0.1906,
      "step": 4888
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.970964089357048,
      "learning_rate": 6.952694962014894e-06,
      "loss": 0.5516,
      "step": 4889
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3803562728062753,
      "learning_rate": 6.951502274862656e-06,
      "loss": 0.2229,
      "step": 4890
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.871589193090764,
      "learning_rate": 6.950309456700158e-06,
      "loss": 0.4558,
      "step": 4891
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4185740673601595,
      "learning_rate": 6.949116507607476e-06,
      "loss": 0.1956,
      "step": 4892
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4590087545002777,
      "learning_rate": 6.947923427664698e-06,
      "loss": 0.2079,
      "step": 4893
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.452520864938566,
      "learning_rate": 6.946730216951918e-06,
      "loss": 0.6906,
      "step": 4894
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4225880604839358,
      "learning_rate": 6.945536875549241e-06,
      "loss": 0.233,
      "step": 4895
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3161400077408332,
      "learning_rate": 6.944343403536777e-06,
      "loss": 0.2156,
      "step": 4896
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3357418031471076,
      "learning_rate": 6.94314980099465e-06,
      "loss": 0.1828,
      "step": 4897
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3265221410054067,
      "learning_rate": 6.941956068002991e-06,
      "loss": 0.1782,
      "step": 4898
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1662216431501797,
      "learning_rate": 6.940762204641935e-06,
      "loss": 0.1939,
      "step": 4899
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.395001291101406,
      "learning_rate": 6.939568210991633e-06,
      "loss": 0.2243,
      "step": 4900
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4352154910816537,
      "learning_rate": 6.9383740871322405e-06,
      "loss": 0.2332,
      "step": 4901
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4065655492432347,
      "learning_rate": 6.937179833143921e-06,
      "loss": 0.1957,
      "step": 4902
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3167961798580654,
      "learning_rate": 6.935985449106851e-06,
      "loss": 0.2061,
      "step": 4903
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2770916724380539,
      "learning_rate": 6.934790935101214e-06,
      "loss": 0.1932,
      "step": 4904
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.134553579940731,
      "learning_rate": 6.933596291207196e-06,
      "loss": 0.1788,
      "step": 4905
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3224940310551938,
      "learning_rate": 6.932401517505003e-06,
      "loss": 0.1939,
      "step": 4906
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.565108742040396,
      "learning_rate": 6.931206614074839e-06,
      "loss": 0.5762,
      "step": 4907
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.408024383876922,
      "learning_rate": 6.9300115809969245e-06,
      "loss": 0.2374,
      "step": 4908
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.199798691166958,
      "learning_rate": 6.928816418351485e-06,
      "loss": 0.5685,
      "step": 4909
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3526336714893068,
      "learning_rate": 6.927621126218756e-06,
      "loss": 0.1894,
      "step": 4910
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1935397918230348,
      "learning_rate": 6.9264257046789794e-06,
      "loss": 0.208,
      "step": 4911
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3047218043384239,
      "learning_rate": 6.9252301538124086e-06,
      "loss": 0.2082,
      "step": 4912
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4263092442658203,
      "learning_rate": 6.924034473699305e-06,
      "loss": 0.2068,
      "step": 4913
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.413327240446172,
      "learning_rate": 6.922838664419937e-06,
      "loss": 0.1767,
      "step": 4914
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.278508795441007,
      "learning_rate": 6.921642726054583e-06,
      "loss": 0.1972,
      "step": 4915
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4762990773159892,
      "learning_rate": 6.9204466586835305e-06,
      "loss": 0.2014,
      "step": 4916
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3690871783013314,
      "learning_rate": 6.919250462387075e-06,
      "loss": 0.2166,
      "step": 4917
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4212192863272215,
      "learning_rate": 6.9180541372455225e-06,
      "loss": 0.2302,
      "step": 4918
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4792930772918174,
      "learning_rate": 6.916857683339183e-06,
      "loss": 0.2076,
      "step": 4919
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.307779964272019,
      "learning_rate": 6.915661100748379e-06,
      "loss": 0.206,
      "step": 4920
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.602277155926983,
      "learning_rate": 6.914464389553443e-06,
      "loss": 0.7351,
      "step": 4921
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4759958696468491,
      "learning_rate": 6.91326754983471e-06,
      "loss": 0.1673,
      "step": 4922
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.319054898036481,
      "learning_rate": 6.91207058167253e-06,
      "loss": 0.1847,
      "step": 4923
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.216167784105681,
      "learning_rate": 6.910873485147258e-06,
      "loss": 0.1657,
      "step": 4924
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4408456792012903,
      "learning_rate": 6.9096762603392595e-06,
      "loss": 0.2279,
      "step": 4925
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2037805327361621,
      "learning_rate": 6.908478907328907e-06,
      "loss": 0.1711,
      "step": 4926
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5321231371973905,
      "learning_rate": 6.907281426196584e-06,
      "loss": 0.2339,
      "step": 4927
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.383333258786712,
      "learning_rate": 6.906083817022679e-06,
      "loss": 0.2172,
      "step": 4928
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.266122627689853,
      "learning_rate": 6.904886079887594e-06,
      "loss": 0.2348,
      "step": 4929
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5400489641294834,
      "learning_rate": 6.903688214871734e-06,
      "loss": 0.2304,
      "step": 4930
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5490376796386518,
      "learning_rate": 6.902490222055515e-06,
      "loss": 0.1923,
      "step": 4931
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.375380413865992,
      "learning_rate": 6.901292101519365e-06,
      "loss": 0.1847,
      "step": 4932
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.902342818607749,
      "learning_rate": 6.900093853343715e-06,
      "loss": 0.601,
      "step": 4933
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.307491509380475,
      "learning_rate": 6.898895477609007e-06,
      "loss": 0.2062,
      "step": 4934
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.36651356147818,
      "learning_rate": 6.897696974395691e-06,
      "loss": 0.21,
      "step": 4935
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3222386495184233,
      "learning_rate": 6.896498343784229e-06,
      "loss": 0.2319,
      "step": 4936
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5050735987420332,
      "learning_rate": 6.895299585855086e-06,
      "loss": 0.2525,
      "step": 4937
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.36016554479017,
      "learning_rate": 6.8941007006887405e-06,
      "loss": 0.5755,
      "step": 4938
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3893299537750121,
      "learning_rate": 6.892901688365677e-06,
      "loss": 0.2126,
      "step": 4939
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6459441379369277,
      "learning_rate": 6.891702548966386e-06,
      "loss": 0.2268,
      "step": 4940
    },
    {
      "epoch": 0.4,
      "grad_norm": 25.113430737548498,
      "learning_rate": 6.890503282571371e-06,
      "loss": 0.4396,
      "step": 4941
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.307314102051341,
      "learning_rate": 6.889303889261143e-06,
      "loss": 0.1999,
      "step": 4942
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3296348716315687,
      "learning_rate": 6.888104369116222e-06,
      "loss": 0.2017,
      "step": 4943
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3279160327902289,
      "learning_rate": 6.886904722217132e-06,
      "loss": 0.1911,
      "step": 4944
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.768981047396083,
      "learning_rate": 6.885704948644411e-06,
      "loss": 0.7529,
      "step": 4945
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2629903682824621,
      "learning_rate": 6.8845050484786045e-06,
      "loss": 0.2088,
      "step": 4946
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.334688790410686,
      "learning_rate": 6.883305021800264e-06,
      "loss": 0.2046,
      "step": 4947
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2984950009023064,
      "learning_rate": 6.8821048686899515e-06,
      "loss": 0.2026,
      "step": 4948
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2852530629635837,
      "learning_rate": 6.8809045892282365e-06,
      "loss": 0.1655,
      "step": 4949
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2376225065294297,
      "learning_rate": 6.8797041834956955e-06,
      "loss": 0.1771,
      "step": 4950
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.116744101955967,
      "learning_rate": 6.87850365157292e-06,
      "loss": 0.1486,
      "step": 4951
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5416820976558965,
      "learning_rate": 6.877302993540501e-06,
      "loss": 0.2004,
      "step": 4952
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4467629270827924,
      "learning_rate": 6.876102209479045e-06,
      "loss": 0.224,
      "step": 4953
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5139332612600673,
      "learning_rate": 6.874901299469164e-06,
      "loss": 0.2502,
      "step": 4954
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.421858902691484,
      "learning_rate": 6.873700263591476e-06,
      "loss": 0.2144,
      "step": 4955
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.321054108608638,
      "learning_rate": 6.872499101926615e-06,
      "loss": 0.1884,
      "step": 4956
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3692552611674051,
      "learning_rate": 6.871297814555214e-06,
      "loss": 0.2103,
      "step": 4957
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4155867024277298,
      "learning_rate": 6.870096401557921e-06,
      "loss": 0.2164,
      "step": 4958
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.797088952103575,
      "learning_rate": 6.86889486301539e-06,
      "loss": 0.7178,
      "step": 4959
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2949654574890377,
      "learning_rate": 6.867693199008285e-06,
      "loss": 0.2027,
      "step": 4960
    },
    {
      "epoch": 0.4,
      "grad_norm": 34.15486955323081,
      "learning_rate": 6.866491409617275e-06,
      "loss": 0.6208,
      "step": 4961
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3452684344912391,
      "learning_rate": 6.865289494923044e-06,
      "loss": 0.2018,
      "step": 4962
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4050764532151765,
      "learning_rate": 6.8640874550062765e-06,
      "loss": 0.1579,
      "step": 4963
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3179846356465217,
      "learning_rate": 6.862885289947669e-06,
      "loss": 0.2032,
      "step": 4964
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4399596586235432,
      "learning_rate": 6.8616829998279295e-06,
      "loss": 0.1943,
      "step": 4965
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.9626913697152992,
      "learning_rate": 6.86048058472777e-06,
      "loss": 0.5126,
      "step": 4966
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3146291322706247,
      "learning_rate": 6.8592780447279106e-06,
      "loss": 0.1914,
      "step": 4967
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4635886329776477,
      "learning_rate": 6.8580753799090825e-06,
      "loss": 0.2332,
      "step": 4968
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3034745213422039,
      "learning_rate": 6.856872590352025e-06,
      "loss": 0.2004,
      "step": 4969
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3443065348836594,
      "learning_rate": 6.8556696761374844e-06,
      "loss": 0.2166,
      "step": 4970
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3766822623297141,
      "learning_rate": 6.854466637346215e-06,
      "loss": 0.2212,
      "step": 4971
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3297164025013832,
      "learning_rate": 6.853263474058982e-06,
      "loss": 0.1739,
      "step": 4972
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3695836169183748,
      "learning_rate": 6.852060186356557e-06,
      "loss": 0.2255,
      "step": 4973
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.946767894543483,
      "learning_rate": 6.850856774319721e-06,
      "loss": 0.6077,
      "step": 4974
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3829320828718388,
      "learning_rate": 6.849653238029261e-06,
      "loss": 0.2182,
      "step": 4975
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4664919764851658,
      "learning_rate": 6.848449577565974e-06,
      "loss": 0.1779,
      "step": 4976
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.829624030885766,
      "learning_rate": 6.847245793010668e-06,
      "loss": 0.5379,
      "step": 4977
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1648620839624397,
      "learning_rate": 6.846041884444154e-06,
      "loss": 0.1277,
      "step": 4978
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3899190230686151,
      "learning_rate": 6.844837851947255e-06,
      "loss": 0.1926,
      "step": 4979
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3245950993499847,
      "learning_rate": 6.843633695600802e-06,
      "loss": 0.2225,
      "step": 4980
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.293372760045752,
      "learning_rate": 6.8424294154856315e-06,
      "loss": 0.1749,
      "step": 4981
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2931034185984664,
      "learning_rate": 6.841225011682594e-06,
      "loss": 0.1904,
      "step": 4982
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3950209648683762,
      "learning_rate": 6.840020484272543e-06,
      "loss": 0.2036,
      "step": 4983
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4741448372516144,
      "learning_rate": 6.83881583333634e-06,
      "loss": 0.1803,
      "step": 4984
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.956825256850357,
      "learning_rate": 6.837611058954858e-06,
      "loss": 0.5337,
      "step": 4985
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3085637376696329,
      "learning_rate": 6.836406161208979e-06,
      "loss": 0.2358,
      "step": 4986
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3999991063350994,
      "learning_rate": 6.835201140179592e-06,
      "loss": 0.2018,
      "step": 4987
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.248226332146478,
      "learning_rate": 6.83399599594759e-06,
      "loss": 0.1917,
      "step": 4988
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2962187144009307,
      "learning_rate": 6.83279072859388e-06,
      "loss": 0.1789,
      "step": 4989
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4934910549861211,
      "learning_rate": 6.831585338199375e-06,
      "loss": 0.1916,
      "step": 4990
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3457602355219391,
      "learning_rate": 6.830379824845e-06,
      "loss": 0.2226,
      "step": 4991
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3335046171770408,
      "learning_rate": 6.829174188611678e-06,
      "loss": 0.1897,
      "step": 4992
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.57326488350582,
      "learning_rate": 6.8279684295803525e-06,
      "loss": 0.2287,
      "step": 4993
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2669226364021462,
      "learning_rate": 6.826762547831967e-06,
      "loss": 0.1818,
      "step": 4994
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4424108482404092,
      "learning_rate": 6.825556543447476e-06,
      "loss": 0.211,
      "step": 4995
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4019277274493673,
      "learning_rate": 6.8243504165078445e-06,
      "loss": 0.1915,
      "step": 4996
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2147315170196327,
      "learning_rate": 6.823144167094042e-06,
      "loss": 0.2079,
      "step": 4997
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2701717115152467,
      "learning_rate": 6.821937795287046e-06,
      "loss": 0.1675,
      "step": 4998
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4425369966089794,
      "learning_rate": 6.820731301167845e-06,
      "loss": 0.2093,
      "step": 4999
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4492908897367405,
      "learning_rate": 6.819524684817439e-06,
      "loss": 0.206,
      "step": 5000
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.641069139104556,
      "learning_rate": 6.818317946316826e-06,
      "loss": 0.2191,
      "step": 5001
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.523562717886117,
      "learning_rate": 6.817111085747018e-06,
      "loss": 0.2309,
      "step": 5002
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6006521753071992,
      "learning_rate": 6.8159041031890395e-06,
      "loss": 0.2745,
      "step": 5003
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3417446947327318,
      "learning_rate": 6.8146969987239155e-06,
      "loss": 0.1765,
      "step": 5004
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2989670579840462,
      "learning_rate": 6.8134897724326846e-06,
      "loss": 0.1845,
      "step": 5005
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4715732549671763,
      "learning_rate": 6.812282424396389e-06,
      "loss": 0.2044,
      "step": 5006
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6852456234673696,
      "learning_rate": 6.811074954696084e-06,
      "loss": 0.2536,
      "step": 5007
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3991171018594237,
      "learning_rate": 6.809867363412832e-06,
      "loss": 0.1974,
      "step": 5008
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2239878316189168,
      "learning_rate": 6.808659650627698e-06,
      "loss": 0.1806,
      "step": 5009
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2660882630922488,
      "learning_rate": 6.807451816421762e-06,
      "loss": 0.1848,
      "step": 5010
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4374387985078176,
      "learning_rate": 6.8062438608761095e-06,
      "loss": 0.196,
      "step": 5011
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3944339961623857,
      "learning_rate": 6.805035784071833e-06,
      "loss": 0.2308,
      "step": 5012
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.411062033238887,
      "learning_rate": 6.803827586090036e-06,
      "loss": 0.2087,
      "step": 5013
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4193911557790597,
      "learning_rate": 6.802619267011828e-06,
      "loss": 0.2076,
      "step": 5014
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2961662675973553,
      "learning_rate": 6.801410826918327e-06,
      "loss": 0.208,
      "step": 5015
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2757708563696757,
      "learning_rate": 6.800202265890658e-06,
      "loss": 0.2044,
      "step": 5016
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4609176065044773,
      "learning_rate": 6.798993584009959e-06,
      "loss": 0.2118,
      "step": 5017
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.468038699485295,
      "learning_rate": 6.797784781357368e-06,
      "loss": 0.1872,
      "step": 5018
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2988052843294424,
      "learning_rate": 6.796575858014036e-06,
      "loss": 0.2065,
      "step": 5019
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5636886441328282,
      "learning_rate": 6.7953668140611264e-06,
      "loss": 0.2272,
      "step": 5020
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.806739700279936,
      "learning_rate": 6.794157649579801e-06,
      "loss": 0.5858,
      "step": 5021
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4620134666821374,
      "learning_rate": 6.792948364651237e-06,
      "loss": 0.2255,
      "step": 5022
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.30615466809214,
      "learning_rate": 6.791738959356616e-06,
      "loss": 0.1975,
      "step": 5023
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5922264549549825,
      "learning_rate": 6.790529433777131e-06,
      "loss": 0.2006,
      "step": 5024
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.372185829844158,
      "learning_rate": 6.78931978799398e-06,
      "loss": 0.1955,
      "step": 5025
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.150212198115859,
      "learning_rate": 6.78811002208837e-06,
      "loss": 0.4465,
      "step": 5026
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6188269041012755,
      "learning_rate": 6.786900136141516e-06,
      "loss": 0.2191,
      "step": 5027
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4740464036444292,
      "learning_rate": 6.785690130234643e-06,
      "loss": 0.1739,
      "step": 5028
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1255519897147408,
      "learning_rate": 6.78448000444898e-06,
      "loss": 0.2072,
      "step": 5029
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3151225369620891,
      "learning_rate": 6.783269758865768e-06,
      "loss": 0.1863,
      "step": 5030
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.428731589974488,
      "learning_rate": 6.782059393566254e-06,
      "loss": 0.2319,
      "step": 5031
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4195228746789723,
      "learning_rate": 6.780848908631694e-06,
      "loss": 0.2506,
      "step": 5032
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.379849430240999,
      "learning_rate": 6.779638304143349e-06,
      "loss": 0.2652,
      "step": 5033
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2133125559139302,
      "learning_rate": 6.778427580182494e-06,
      "loss": 0.1768,
      "step": 5034
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.437496250549501,
      "learning_rate": 6.777216736830409e-06,
      "loss": 0.2212,
      "step": 5035
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3274476220758487,
      "learning_rate": 6.7760057741683774e-06,
      "loss": 0.1995,
      "step": 5036
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3365914244953607,
      "learning_rate": 6.774794692277698e-06,
      "loss": 0.1629,
      "step": 5037
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.373296646580278,
      "learning_rate": 6.773583491239672e-06,
      "loss": 0.2184,
      "step": 5038
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.860886528026404,
      "learning_rate": 6.772372171135614e-06,
      "loss": 0.5923,
      "step": 5039
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3129858532952587,
      "learning_rate": 6.77116073204684e-06,
      "loss": 0.1918,
      "step": 5040
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.748459346325971,
      "learning_rate": 6.769949174054682e-06,
      "loss": 0.6236,
      "step": 5041
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4051356650142013,
      "learning_rate": 6.768737497240472e-06,
      "loss": 0.2026,
      "step": 5042
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2825537650348895,
      "learning_rate": 6.767525701685555e-06,
      "loss": 0.173,
      "step": 5043
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3237141716077738,
      "learning_rate": 6.766313787471283e-06,
      "loss": 0.2073,
      "step": 5044
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2711946889863033,
      "learning_rate": 6.765101754679015e-06,
      "loss": 0.1483,
      "step": 5045
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.72244253538842,
      "learning_rate": 6.7638896033901165e-06,
      "loss": 0.7456,
      "step": 5046
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2391592063988401,
      "learning_rate": 6.762677333685965e-06,
      "loss": 0.1533,
      "step": 5047
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.444001177481423,
      "learning_rate": 6.761464945647944e-06,
      "loss": 0.233,
      "step": 5048
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3584264534273747,
      "learning_rate": 6.760252439357444e-06,
      "loss": 0.233,
      "step": 5049
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3073027697573225,
      "learning_rate": 6.7590398148958625e-06,
      "loss": 0.184,
      "step": 5050
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4308227549997163,
      "learning_rate": 6.757827072344612e-06,
      "loss": 0.179,
      "step": 5051
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2201289525801633,
      "learning_rate": 6.756614211785103e-06,
      "loss": 0.1669,
      "step": 5052
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.930833694603219,
      "learning_rate": 6.755401233298758e-06,
      "loss": 0.6947,
      "step": 5053
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4120492897936616,
      "learning_rate": 6.754188136967011e-06,
      "loss": 0.2232,
      "step": 5054
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6083660082894295,
      "learning_rate": 6.7529749228712994e-06,
      "loss": 0.201,
      "step": 5055
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.30257800508728,
      "learning_rate": 6.751761591093069e-06,
      "loss": 0.761,
      "step": 5056
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1149760061769185,
      "learning_rate": 6.750548141713775e-06,
      "loss": 0.1221,
      "step": 5057
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.278570615557228,
      "learning_rate": 6.749334574814882e-06,
      "loss": 0.1432,
      "step": 5058
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3692256240549228,
      "learning_rate": 6.748120890477859e-06,
      "loss": 0.1613,
      "step": 5059
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4990085258271373,
      "learning_rate": 6.746907088784182e-06,
      "loss": 0.2077,
      "step": 5060
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1353209250819047,
      "learning_rate": 6.74569316981534e-06,
      "loss": 0.1389,
      "step": 5061
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4560256011956285,
      "learning_rate": 6.744479133652827e-06,
      "loss": 0.2174,
      "step": 5062
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.659779778638779,
      "learning_rate": 6.743264980378143e-06,
      "loss": 0.748,
      "step": 5063
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.367237654157424,
      "learning_rate": 6.7420507100727994e-06,
      "loss": 0.2146,
      "step": 5064
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1440305214043616,
      "learning_rate": 6.740836322818314e-06,
      "loss": 0.1362,
      "step": 5065
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2463456296363806,
      "learning_rate": 6.739621818696211e-06,
      "loss": 0.1552,
      "step": 5066
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2353888281433785,
      "learning_rate": 6.738407197788026e-06,
      "loss": 0.1788,
      "step": 5067
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.264158487750602,
      "learning_rate": 6.737192460175297e-06,
      "loss": 0.1852,
      "step": 5068
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.361099767610137,
      "learning_rate": 6.735977605939575e-06,
      "loss": 0.1891,
      "step": 5069
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4009815420599434,
      "learning_rate": 6.734762635162417e-06,
      "loss": 0.2236,
      "step": 5070
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4671571247351352,
      "learning_rate": 6.733547547925387e-06,
      "loss": 0.1999,
      "step": 5071
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.366800866590126,
      "learning_rate": 6.732332344310058e-06,
      "loss": 0.2418,
      "step": 5072
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4245584301024892,
      "learning_rate": 6.731117024398009e-06,
      "loss": 0.2009,
      "step": 5073
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3185407405524865,
      "learning_rate": 6.729901588270829e-06,
      "loss": 0.2007,
      "step": 5074
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4423637331372412,
      "learning_rate": 6.728686036010115e-06,
      "loss": 0.2207,
      "step": 5075
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4494530150560185,
      "learning_rate": 6.727470367697468e-06,
      "loss": 0.1914,
      "step": 5076
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.56831638037855,
      "learning_rate": 6.726254583414504e-06,
      "loss": 0.5959,
      "step": 5077
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3020232492479906,
      "learning_rate": 6.725038683242837e-06,
      "loss": 0.197,
      "step": 5078
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3582745578667978,
      "learning_rate": 6.723822667264098e-06,
      "loss": 0.2113,
      "step": 5079
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3703998711609113,
      "learning_rate": 6.7226065355599204e-06,
      "loss": 0.2328,
      "step": 5080
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4767342219225292,
      "learning_rate": 6.7213902882119455e-06,
      "loss": 0.2168,
      "step": 5081
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.465831578868476,
      "learning_rate": 6.720173925301825e-06,
      "loss": 0.6009,
      "step": 5082
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4052115706334376,
      "learning_rate": 6.718957446911218e-06,
      "loss": 0.1891,
      "step": 5083
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1180965430031633,
      "learning_rate": 6.717740853121789e-06,
      "loss": 0.1785,
      "step": 5084
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2194147213173732,
      "learning_rate": 6.716524144015212e-06,
      "loss": 0.2058,
      "step": 5085
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.138699990441886,
      "learning_rate": 6.7153073196731674e-06,
      "loss": 0.7238,
      "step": 5086
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4359326725586719,
      "learning_rate": 6.7140903801773484e-06,
      "loss": 0.1844,
      "step": 5087
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.296712383438515,
      "learning_rate": 6.712873325609445e-06,
      "loss": 0.1975,
      "step": 5088
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2789756214718724,
      "learning_rate": 6.711656156051167e-06,
      "loss": 0.1821,
      "step": 5089
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3187854076684844,
      "learning_rate": 6.710438871584225e-06,
      "loss": 0.1869,
      "step": 5090
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2994498665757666,
      "learning_rate": 6.709221472290339e-06,
      "loss": 0.1725,
      "step": 5091
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.360469534991078,
      "learning_rate": 6.708003958251237e-06,
      "loss": 0.2216,
      "step": 5092
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.570249553278296,
      "learning_rate": 6.706786329548654e-06,
      "loss": 0.2211,
      "step": 5093
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4245777614144628,
      "learning_rate": 6.7055685862643336e-06,
      "loss": 0.1907,
      "step": 5094
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3735958873485237,
      "learning_rate": 6.704350728480026e-06,
      "loss": 0.2185,
      "step": 5095
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2785229780748044,
      "learning_rate": 6.7031327562774914e-06,
      "loss": 0.1917,
      "step": 5096
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.278140072653446,
      "learning_rate": 6.701914669738494e-06,
      "loss": 0.2125,
      "step": 5097
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4434194683931856,
      "learning_rate": 6.700696468944806e-06,
      "loss": 0.2224,
      "step": 5098
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4729977219383583,
      "learning_rate": 6.699478153978214e-06,
      "loss": 0.2232,
      "step": 5099
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4187648921743439,
      "learning_rate": 6.698259724920503e-06,
      "loss": 0.2216,
      "step": 5100
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2021595713911357,
      "learning_rate": 6.697041181853472e-06,
      "loss": 0.1766,
      "step": 5101
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3605060827988127,
      "learning_rate": 6.695822524858922e-06,
      "loss": 0.1979,
      "step": 5102
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2135922307564213,
      "learning_rate": 6.69460375401867e-06,
      "loss": 0.1598,
      "step": 5103
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6471258015363548,
      "learning_rate": 6.693384869414534e-06,
      "loss": 0.2166,
      "step": 5104
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1291084871068817,
      "learning_rate": 6.69216587112834e-06,
      "loss": 0.1772,
      "step": 5105
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3507924587822886,
      "learning_rate": 6.690946759241925e-06,
      "loss": 0.1867,
      "step": 5106
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5264977503049746,
      "learning_rate": 6.689727533837129e-06,
      "loss": 0.1761,
      "step": 5107
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3464513142596253,
      "learning_rate": 6.688508194995806e-06,
      "loss": 0.1913,
      "step": 5108
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2005929596028337,
      "learning_rate": 6.68728874279981e-06,
      "loss": 0.2091,
      "step": 5109
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2836211981695371,
      "learning_rate": 6.686069177331009e-06,
      "loss": 0.2149,
      "step": 5110
    },
    {
      "epoch": 0.41,
      "grad_norm": 11.985673162085996,
      "learning_rate": 6.684849498671277e-06,
      "loss": 0.7009,
      "step": 5111
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3683650079021368,
      "learning_rate": 6.683629706902491e-06,
      "loss": 0.2422,
      "step": 5112
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2387814503990704,
      "learning_rate": 6.682409802106543e-06,
      "loss": 0.18,
      "step": 5113
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4213571056989656,
      "learning_rate": 6.681189784365327e-06,
      "loss": 0.1751,
      "step": 5114
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.337220979241151,
      "learning_rate": 6.679969653760747e-06,
      "loss": 0.2044,
      "step": 5115
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3924072353410815,
      "learning_rate": 6.678749410374714e-06,
      "loss": 0.2325,
      "step": 5116
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2504738047540074,
      "learning_rate": 6.677529054289147e-06,
      "loss": 0.1764,
      "step": 5117
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.2249976894632075,
      "learning_rate": 6.676308585585971e-06,
      "loss": 0.4245,
      "step": 5118
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5792309978259225,
      "learning_rate": 6.675088004347121e-06,
      "loss": 0.1946,
      "step": 5119
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.401352906677056,
      "learning_rate": 6.673867310654538e-06,
      "loss": 0.1719,
      "step": 5120
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4070388027380107,
      "learning_rate": 6.672646504590172e-06,
      "loss": 0.212,
      "step": 5121
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.510194873582569,
      "learning_rate": 6.671425586235978e-06,
      "loss": 0.258,
      "step": 5122
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4186702115404224,
      "learning_rate": 6.6702045556739195e-06,
      "loss": 0.206,
      "step": 5123
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3454907528411397,
      "learning_rate": 6.668983412985968e-06,
      "loss": 0.2013,
      "step": 5124
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.409654812960609,
      "learning_rate": 6.667762158254104e-06,
      "loss": 0.1984,
      "step": 5125
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3092907006306889,
      "learning_rate": 6.666540791560312e-06,
      "loss": 0.2202,
      "step": 5126
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3680391626889064,
      "learning_rate": 6.665319312986589e-06,
      "loss": 0.2169,
      "step": 5127
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.303102769323935,
      "learning_rate": 6.664097722614934e-06,
      "loss": 0.5969,
      "step": 5128
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3862753603771096,
      "learning_rate": 6.662876020527357e-06,
      "loss": 0.1817,
      "step": 5129
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2823484484426337,
      "learning_rate": 6.661654206805874e-06,
      "loss": 0.2128,
      "step": 5130
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4896645180184762,
      "learning_rate": 6.6604322815325105e-06,
      "loss": 0.169,
      "step": 5131
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4577324897450734,
      "learning_rate": 6.659210244789296e-06,
      "loss": 0.2058,
      "step": 5132
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2797823273685967,
      "learning_rate": 6.65798809665827e-06,
      "loss": 0.1528,
      "step": 5133
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4565319664506642,
      "learning_rate": 6.656765837221481e-06,
      "loss": 0.2789,
      "step": 5134
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4823194905770938,
      "learning_rate": 6.6555434665609806e-06,
      "loss": 0.1922,
      "step": 5135
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.217833003250719,
      "learning_rate": 6.654320984758832e-06,
      "loss": 0.5128,
      "step": 5136
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3322279771206507,
      "learning_rate": 6.653098391897102e-06,
      "loss": 0.2068,
      "step": 5137
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3271212350446058,
      "learning_rate": 6.65187568805787e-06,
      "loss": 0.2158,
      "step": 5138
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4117990795235238,
      "learning_rate": 6.650652873323218e-06,
      "loss": 0.1928,
      "step": 5139
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.380734059257349,
      "learning_rate": 6.6494299477752364e-06,
      "loss": 0.6327,
      "step": 5140
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.306915458809025,
      "learning_rate": 6.6482069114960245e-06,
      "loss": 0.2187,
      "step": 5141
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.408906183130293,
      "learning_rate": 6.646983764567689e-06,
      "loss": 0.1944,
      "step": 5142
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.36877857293987,
      "learning_rate": 6.645760507072343e-06,
      "loss": 0.1788,
      "step": 5143
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.538568546982342,
      "learning_rate": 6.644537139092109e-06,
      "loss": 0.2329,
      "step": 5144
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.414009116440467,
      "learning_rate": 6.643313660709114e-06,
      "loss": 0.1987,
      "step": 5145
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3831981680385181,
      "learning_rate": 6.642090072005493e-06,
      "loss": 0.2357,
      "step": 5146
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3173789491870171,
      "learning_rate": 6.6408663730633895e-06,
      "loss": 0.1492,
      "step": 5147
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2529265130886371,
      "learning_rate": 6.639642563964956e-06,
      "loss": 0.1347,
      "step": 5148
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.390799814663467,
      "learning_rate": 6.63841864479235e-06,
      "loss": 0.1955,
      "step": 5149
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3969541313429892,
      "learning_rate": 6.637194615627733e-06,
      "loss": 0.2217,
      "step": 5150
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.777349784117009,
      "learning_rate": 6.635970476553284e-06,
      "loss": 0.6616,
      "step": 5151
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2839660546476244,
      "learning_rate": 6.63474622765118e-06,
      "loss": 0.2314,
      "step": 5152
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2011964471013474,
      "learning_rate": 6.633521869003607e-06,
      "loss": 0.1615,
      "step": 5153
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.375710903732313,
      "learning_rate": 6.632297400692762e-06,
      "loss": 0.2183,
      "step": 5154
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5528831367273181,
      "learning_rate": 6.631072822800847e-06,
      "loss": 0.2024,
      "step": 5155
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2150661916156171,
      "learning_rate": 6.629848135410072e-06,
      "loss": 0.1668,
      "step": 5156
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2386011572181472,
      "learning_rate": 6.628623338602653e-06,
      "loss": 0.1629,
      "step": 5157
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6045531022216672,
      "learning_rate": 6.627398432460815e-06,
      "loss": 0.2398,
      "step": 5158
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.5666984658424905,
      "learning_rate": 6.626173417066789e-06,
      "loss": 0.6172,
      "step": 5159
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.445643071293786,
      "learning_rate": 6.624948292502814e-06,
      "loss": 0.2147,
      "step": 5160
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4017097066083113,
      "learning_rate": 6.623723058851137e-06,
      "loss": 0.2201,
      "step": 5161
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1477163864114104,
      "learning_rate": 6.6224977161940115e-06,
      "loss": 0.1419,
      "step": 5162
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.284885404375715,
      "learning_rate": 6.6212722646137e-06,
      "loss": 0.2063,
      "step": 5163
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3652835683167315,
      "learning_rate": 6.620046704192466e-06,
      "loss": 0.2406,
      "step": 5164
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.513823307266773,
      "learning_rate": 6.618821035012591e-06,
      "loss": 0.6655,
      "step": 5165
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.306438313565492,
      "learning_rate": 6.617595257156355e-06,
      "loss": 0.1788,
      "step": 5166
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4229374764507086,
      "learning_rate": 6.616369370706046e-06,
      "loss": 0.2123,
      "step": 5167
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6782941381860743,
      "learning_rate": 6.615143375743965e-06,
      "loss": 0.2354,
      "step": 5168
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3772041300385436,
      "learning_rate": 6.613917272352416e-06,
      "loss": 0.1655,
      "step": 5169
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3191228760552132,
      "learning_rate": 6.61269106061371e-06,
      "loss": 0.224,
      "step": 5170
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.365391664482848,
      "learning_rate": 6.611464740610165e-06,
      "loss": 0.206,
      "step": 5171
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4188461664109568,
      "learning_rate": 6.610238312424112e-06,
      "loss": 0.2089,
      "step": 5172
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2845131248319621,
      "learning_rate": 6.609011776137881e-06,
      "loss": 0.2282,
      "step": 5173
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.501063172050814,
      "learning_rate": 6.607785131833816e-06,
      "loss": 0.245,
      "step": 5174
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.181558809433394,
      "learning_rate": 6.6065583795942625e-06,
      "loss": 0.2312,
      "step": 5175
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.304446208118464,
      "learning_rate": 6.605331519501578e-06,
      "loss": 0.2161,
      "step": 5176
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2041310300143628,
      "learning_rate": 6.604104551638124e-06,
      "loss": 0.1983,
      "step": 5177
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.189824072740901,
      "learning_rate": 6.60287747608627e-06,
      "loss": 0.3953,
      "step": 5178
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3396686699741533,
      "learning_rate": 6.601650292928395e-06,
      "loss": 0.2344,
      "step": 5179
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3259471897185762,
      "learning_rate": 6.600423002246885e-06,
      "loss": 0.1865,
      "step": 5180
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.122777008167117,
      "learning_rate": 6.599195604124127e-06,
      "loss": 0.1651,
      "step": 5181
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6060983910820983,
      "learning_rate": 6.597968098642524e-06,
      "loss": 0.1945,
      "step": 5182
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3248521311946972,
      "learning_rate": 6.596740485884483e-06,
      "loss": 0.1862,
      "step": 5183
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2266571024389004,
      "learning_rate": 6.595512765932412e-06,
      "loss": 0.2094,
      "step": 5184
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.198211484396834,
      "learning_rate": 6.594284938868737e-06,
      "loss": 0.1677,
      "step": 5185
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.259374019543164,
      "learning_rate": 6.593057004775882e-06,
      "loss": 0.2022,
      "step": 5186
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4254052176614622,
      "learning_rate": 6.591828963736285e-06,
      "loss": 0.2259,
      "step": 5187
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3335011203915845,
      "learning_rate": 6.590600815832385e-06,
      "loss": 0.2365,
      "step": 5188
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.113398830976923,
      "learning_rate": 6.589372561146634e-06,
      "loss": 0.5034,
      "step": 5189
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.465590285250945,
      "learning_rate": 6.588144199761487e-06,
      "loss": 0.6081,
      "step": 5190
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2444079727787378,
      "learning_rate": 6.586915731759409e-06,
      "loss": 0.192,
      "step": 5191
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5219131647599278,
      "learning_rate": 6.585687157222869e-06,
      "loss": 0.2074,
      "step": 5192
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.235322431467677,
      "learning_rate": 6.584458476234345e-06,
      "loss": 0.4622,
      "step": 5193
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4100632439150653,
      "learning_rate": 6.583229688876323e-06,
      "loss": 0.1898,
      "step": 5194
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.833682386365358,
      "learning_rate": 6.582000795231296e-06,
      "loss": 0.6271,
      "step": 5195
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.279325582236379,
      "learning_rate": 6.58077179538176e-06,
      "loss": 0.1938,
      "step": 5196
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.137727543879439,
      "learning_rate": 6.579542689410227e-06,
      "loss": 0.1634,
      "step": 5197
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2817231274525154,
      "learning_rate": 6.578313477399206e-06,
      "loss": 0.1798,
      "step": 5198
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4140677225061073,
      "learning_rate": 6.577084159431218e-06,
      "loss": 0.2084,
      "step": 5199
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2156221993435592,
      "learning_rate": 6.5758547355887944e-06,
      "loss": 0.1945,
      "step": 5200
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3106859794344599,
      "learning_rate": 6.574625205954466e-06,
      "loss": 0.1553,
      "step": 5201
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.350128581282678,
      "learning_rate": 6.573395570610776e-06,
      "loss": 0.1935,
      "step": 5202
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.077204401761722,
      "learning_rate": 6.5721658296402756e-06,
      "loss": 0.3873,
      "step": 5203
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4068098275491667,
      "learning_rate": 6.570935983125519e-06,
      "loss": 0.2128,
      "step": 5204
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1937252631217878,
      "learning_rate": 6.5697060311490705e-06,
      "loss": 0.1877,
      "step": 5205
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3607506996251284,
      "learning_rate": 6.5684759737935e-06,
      "loss": 0.2213,
      "step": 5206
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.188717692893149,
      "learning_rate": 6.567245811141385e-06,
      "loss": 0.1705,
      "step": 5207
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.254447698085806,
      "learning_rate": 6.56601554327531e-06,
      "loss": 0.1866,
      "step": 5208
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3673909423473616,
      "learning_rate": 6.564785170277868e-06,
      "loss": 0.2116,
      "step": 5209
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8301995695810152,
      "learning_rate": 6.563554692231655e-06,
      "loss": 0.2007,
      "step": 5210
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3866670169513238,
      "learning_rate": 6.562324109219278e-06,
      "loss": 0.2417,
      "step": 5211
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4456057399847053,
      "learning_rate": 6.56109342132335e-06,
      "loss": 0.2426,
      "step": 5212
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.280624360632079,
      "learning_rate": 6.559862628626491e-06,
      "loss": 0.2202,
      "step": 5213
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1813613310739748,
      "learning_rate": 6.5586317312113265e-06,
      "loss": 0.159,
      "step": 5214
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.277997990418586,
      "learning_rate": 6.557400729160494e-06,
      "loss": 0.207,
      "step": 5215
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2871234247354348,
      "learning_rate": 6.556169622556629e-06,
      "loss": 0.1941,
      "step": 5216
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5459302285212213,
      "learning_rate": 6.554938411482383e-06,
      "loss": 0.2348,
      "step": 5217
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4305835030541343,
      "learning_rate": 6.553707096020412e-06,
      "loss": 0.2282,
      "step": 5218
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3662932242711432,
      "learning_rate": 6.552475676253374e-06,
      "loss": 0.1957,
      "step": 5219
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3056266276748318,
      "learning_rate": 6.5512441522639415e-06,
      "loss": 0.1935,
      "step": 5220
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3217043575378415,
      "learning_rate": 6.550012524134788e-06,
      "loss": 0.1814,
      "step": 5221
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3004994485496875,
      "learning_rate": 6.548780791948597e-06,
      "loss": 0.1876,
      "step": 5222
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2596767006464704,
      "learning_rate": 6.547548955788059e-06,
      "loss": 0.1866,
      "step": 5223
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2231860834910764,
      "learning_rate": 6.5463170157358725e-06,
      "loss": 0.1597,
      "step": 5224
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3802075244113663,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.2276,
      "step": 5225
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.539198423189714,
      "learning_rate": 6.543852824287369e-06,
      "loss": 0.615,
      "step": 5226
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.169818901508934,
      "learning_rate": 6.542620573056481e-06,
      "loss": 0.18,
      "step": 5227
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4855886860484384,
      "learning_rate": 6.5413882182648e-06,
      "loss": 0.207,
      "step": 5228
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2826828077468595,
      "learning_rate": 6.540155759995057e-06,
      "loss": 0.1702,
      "step": 5229
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3039588389191847,
      "learning_rate": 6.538923198329993e-06,
      "loss": 0.2288,
      "step": 5230
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5547514510576375,
      "learning_rate": 6.5376905333523525e-06,
      "loss": 0.2192,
      "step": 5231
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.629258780261147,
      "learning_rate": 6.536457765144886e-06,
      "loss": 0.3663,
      "step": 5232
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2238586296618559,
      "learning_rate": 6.535224893790354e-06,
      "loss": 0.1895,
      "step": 5233
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4729332120182608,
      "learning_rate": 6.533991919371524e-06,
      "loss": 0.2168,
      "step": 5234
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3246046912714695,
      "learning_rate": 6.5327588419711695e-06,
      "loss": 0.1917,
      "step": 5235
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5144653474749497,
      "learning_rate": 6.531525661672069e-06,
      "loss": 0.228,
      "step": 5236
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.506070311080983,
      "learning_rate": 6.530292378557011e-06,
      "loss": 0.5919,
      "step": 5237
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.39170617558987,
      "learning_rate": 6.529058992708788e-06,
      "loss": 0.206,
      "step": 5238
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3966972669628692,
      "learning_rate": 6.527825504210204e-06,
      "loss": 0.242,
      "step": 5239
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4078554800074747,
      "learning_rate": 6.526591913144062e-06,
      "loss": 0.1846,
      "step": 5240
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.387431627423938,
      "learning_rate": 6.525358219593181e-06,
      "loss": 0.68,
      "step": 5241
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3283502240473737,
      "learning_rate": 6.524124423640381e-06,
      "loss": 0.1948,
      "step": 5242
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.445215637872364,
      "learning_rate": 6.522890525368489e-06,
      "loss": 0.181,
      "step": 5243
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4414330232891739,
      "learning_rate": 6.521656524860345e-06,
      "loss": 0.2355,
      "step": 5244
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4504433642224108,
      "learning_rate": 6.5204224221987864e-06,
      "loss": 0.2079,
      "step": 5245
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6577254816797182,
      "learning_rate": 6.519188217466664e-06,
      "loss": 0.208,
      "step": 5246
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2590556838192246,
      "learning_rate": 6.517953910746832e-06,
      "loss": 0.1799,
      "step": 5247
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4374067504641521,
      "learning_rate": 6.516719502122158e-06,
      "loss": 0.2463,
      "step": 5248
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2851206548499534,
      "learning_rate": 6.515484991675506e-06,
      "loss": 0.2017,
      "step": 5249
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3290136415083715,
      "learning_rate": 6.514250379489754e-06,
      "loss": 0.2067,
      "step": 5250
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4519870786963927,
      "learning_rate": 6.513015665647787e-06,
      "loss": 0.2088,
      "step": 5251
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2461168106654088,
      "learning_rate": 6.511780850232495e-06,
      "loss": 0.1601,
      "step": 5252
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.030239900414695,
      "learning_rate": 6.510545933326774e-06,
      "loss": 0.7233,
      "step": 5253
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.383948713720009,
      "learning_rate": 6.509310915013527e-06,
      "loss": 0.2127,
      "step": 5254
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.254816776586974,
      "learning_rate": 6.508075795375666e-06,
      "loss": 0.1672,
      "step": 5255
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4958917702463408,
      "learning_rate": 6.506840574496107e-06,
      "loss": 0.204,
      "step": 5256
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2593127748738593,
      "learning_rate": 6.505605252457774e-06,
      "loss": 0.1784,
      "step": 5257
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1576071579002136,
      "learning_rate": 6.504369829343599e-06,
      "loss": 0.184,
      "step": 5258
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.31941450612502,
      "learning_rate": 6.50313430523652e-06,
      "loss": 0.5735,
      "step": 5259
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3870594752053418,
      "learning_rate": 6.5018986802194805e-06,
      "loss": 0.2062,
      "step": 5260
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3546006283955012,
      "learning_rate": 6.500662954375432e-06,
      "loss": 0.1911,
      "step": 5261
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.263165480690272,
      "learning_rate": 6.499427127787332e-06,
      "loss": 0.1839,
      "step": 5262
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4917809090089957,
      "learning_rate": 6.498191200538147e-06,
      "loss": 0.1859,
      "step": 5263
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3084704236254123,
      "learning_rate": 6.496955172710846e-06,
      "loss": 0.16,
      "step": 5264
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3970049996427865,
      "learning_rate": 6.495719044388409e-06,
      "loss": 0.2243,
      "step": 5265
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.592255641479745,
      "learning_rate": 6.494482815653822e-06,
      "loss": 0.2664,
      "step": 5266
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.272641813071551,
      "learning_rate": 6.493246486590074e-06,
      "loss": 0.1932,
      "step": 5267
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2638436118613734,
      "learning_rate": 6.492010057280165e-06,
      "loss": 0.2076,
      "step": 5268
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3169964189654806,
      "learning_rate": 6.4907735278071e-06,
      "loss": 0.201,
      "step": 5269
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.254891763412214,
      "learning_rate": 6.489536898253893e-06,
      "loss": 0.8521,
      "step": 5270
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.433486420115116,
      "learning_rate": 6.48830016870356e-06,
      "loss": 0.2136,
      "step": 5271
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2690419522221899,
      "learning_rate": 6.487063339239127e-06,
      "loss": 0.1677,
      "step": 5272
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3383836535511908,
      "learning_rate": 6.485826409943627e-06,
      "loss": 0.2125,
      "step": 5273
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1445528865312937,
      "learning_rate": 6.484589380900097e-06,
      "loss": 0.166,
      "step": 5274
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3504873031467828,
      "learning_rate": 6.483352252191585e-06,
      "loss": 0.2071,
      "step": 5275
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.395004743627818,
      "learning_rate": 6.482115023901141e-06,
      "loss": 0.7169,
      "step": 5276
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3680353562662346,
      "learning_rate": 6.480877696111826e-06,
      "loss": 0.1637,
      "step": 5277
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.933531513868063,
      "learning_rate": 6.479640268906703e-06,
      "loss": 0.6104,
      "step": 5278
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3290806034404452,
      "learning_rate": 6.478402742368847e-06,
      "loss": 0.205,
      "step": 5279
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4015827887361438,
      "learning_rate": 6.4771651165813345e-06,
      "loss": 0.2069,
      "step": 5280
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.533216254258825,
      "learning_rate": 6.4759273916272525e-06,
      "loss": 0.188,
      "step": 5281
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1712744974487044,
      "learning_rate": 6.4746895675896925e-06,
      "loss": 0.1824,
      "step": 5282
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2743347951300044,
      "learning_rate": 6.473451644551753e-06,
      "loss": 0.1988,
      "step": 5283
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3679096934570472,
      "learning_rate": 6.472213622596542e-06,
      "loss": 0.2122,
      "step": 5284
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.476808405349239,
      "learning_rate": 6.4709755018071685e-06,
      "loss": 0.2162,
      "step": 5285
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5483900357120055,
      "learning_rate": 6.469737282266752e-06,
      "loss": 0.2653,
      "step": 5286
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.350550264475312,
      "learning_rate": 6.468498964058421e-06,
      "loss": 0.2052,
      "step": 5287
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3848710410914897,
      "learning_rate": 6.4672605472653035e-06,
      "loss": 0.188,
      "step": 5288
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3128448240890085,
      "learning_rate": 6.466022031970541e-06,
      "loss": 0.1604,
      "step": 5289
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.313543420913374,
      "learning_rate": 6.464783418257278e-06,
      "loss": 0.2222,
      "step": 5290
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3533663807416145,
      "learning_rate": 6.463544706208663e-06,
      "loss": 0.2401,
      "step": 5291
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4547640740483785,
      "learning_rate": 6.46230589590786e-06,
      "loss": 0.1783,
      "step": 5292
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3873148679714888,
      "learning_rate": 6.461066987438032e-06,
      "loss": 0.2056,
      "step": 5293
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3308554909654229,
      "learning_rate": 6.45982798088235e-06,
      "loss": 0.2225,
      "step": 5294
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.161112617452865,
      "learning_rate": 6.45858887632399e-06,
      "loss": 0.2152,
      "step": 5295
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3387467898282273,
      "learning_rate": 6.457349673846143e-06,
      "loss": 0.1878,
      "step": 5296
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3113179596051425,
      "learning_rate": 6.4561103735319944e-06,
      "loss": 0.2051,
      "step": 5297
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4441212590945582,
      "learning_rate": 6.454870975464744e-06,
      "loss": 0.2193,
      "step": 5298
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.443603744365554,
      "learning_rate": 6.453631479727599e-06,
      "loss": 0.6446,
      "step": 5299
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3066552009333854,
      "learning_rate": 6.452391886403767e-06,
      "loss": 0.1882,
      "step": 5300
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.215473113162278,
      "learning_rate": 6.451152195576469e-06,
      "loss": 0.1817,
      "step": 5301
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5823306336168448,
      "learning_rate": 6.4499124073289246e-06,
      "loss": 0.2329,
      "step": 5302
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2454149933789684,
      "learning_rate": 6.448672521744369e-06,
      "loss": 0.2023,
      "step": 5303
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1938761025019873,
      "learning_rate": 6.447432538906038e-06,
      "loss": 0.184,
      "step": 5304
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.293354731936287,
      "learning_rate": 6.446192458897174e-06,
      "loss": 0.2189,
      "step": 5305
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2467637583450004,
      "learning_rate": 6.444952281801029e-06,
      "loss": 0.1986,
      "step": 5306
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3378535854635816,
      "learning_rate": 6.4437120077008595e-06,
      "loss": 0.1927,
      "step": 5307
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.392758857925226,
      "learning_rate": 6.4424716366799275e-06,
      "loss": 0.2207,
      "step": 5308
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2988957758820754,
      "learning_rate": 6.441231168821505e-06,
      "loss": 0.1931,
      "step": 5309
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3904771883192915,
      "learning_rate": 6.439990604208868e-06,
      "loss": 0.2337,
      "step": 5310
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.424339908342641,
      "learning_rate": 6.438749942925298e-06,
      "loss": 0.8492,
      "step": 5311
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1462883964418502,
      "learning_rate": 6.4375091850540834e-06,
      "loss": 0.1823,
      "step": 5312
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3111800378903464,
      "learning_rate": 6.436268330678523e-06,
      "loss": 0.1872,
      "step": 5313
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3095309408689453,
      "learning_rate": 6.435027379881921e-06,
      "loss": 0.1554,
      "step": 5314
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3144509835966167,
      "learning_rate": 6.433786332747578e-06,
      "loss": 0.1808,
      "step": 5315
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3222440176458186,
      "learning_rate": 6.432545189358818e-06,
      "loss": 0.203,
      "step": 5316
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6903296094257299,
      "learning_rate": 6.4313039497989575e-06,
      "loss": 0.2606,
      "step": 5317
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2779968845586214,
      "learning_rate": 6.4300626141513264e-06,
      "loss": 0.1547,
      "step": 5318
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3535903462433458,
      "learning_rate": 6.4288211824992575e-06,
      "loss": 0.186,
      "step": 5319
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.1276629223870405,
      "learning_rate": 6.427579654926095e-06,
      "loss": 0.5141,
      "step": 5320
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3643396483646841,
      "learning_rate": 6.426338031515184e-06,
      "loss": 0.2038,
      "step": 5321
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2272505186947735,
      "learning_rate": 6.425096312349881e-06,
      "loss": 0.1371,
      "step": 5322
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.309281920455454,
      "learning_rate": 6.423854497513544e-06,
      "loss": 0.2387,
      "step": 5323
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2030404204318856,
      "learning_rate": 6.422612587089541e-06,
      "loss": 0.1979,
      "step": 5324
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5750933372834501,
      "learning_rate": 6.421370581161244e-06,
      "loss": 0.2489,
      "step": 5325
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.568426883302416,
      "learning_rate": 6.420128479812032e-06,
      "loss": 0.756,
      "step": 5326
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2434292311130164,
      "learning_rate": 6.418886283125294e-06,
      "loss": 0.1695,
      "step": 5327
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3528966670120768,
      "learning_rate": 6.4176439911844205e-06,
      "loss": 0.1724,
      "step": 5328
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4541263609865582,
      "learning_rate": 6.41640160407281e-06,
      "loss": 0.2088,
      "step": 5329
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2989119295880054,
      "learning_rate": 6.415159121873868e-06,
      "loss": 0.2325,
      "step": 5330
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3221409393084043,
      "learning_rate": 6.413916544671008e-06,
      "loss": 0.1996,
      "step": 5331
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4618314050777363,
      "learning_rate": 6.4126738725476455e-06,
      "loss": 0.2076,
      "step": 5332
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.300968721589118,
      "learning_rate": 6.411431105587206e-06,
      "loss": 0.1937,
      "step": 5333
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3659040084487077,
      "learning_rate": 6.410188243873119e-06,
      "loss": 0.2315,
      "step": 5334
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.354034672312187,
      "learning_rate": 6.408945287488824e-06,
      "loss": 0.1986,
      "step": 5335
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.193026326031507,
      "learning_rate": 6.407702236517761e-06,
      "loss": 0.2098,
      "step": 5336
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3642984814235528,
      "learning_rate": 6.406459091043382e-06,
      "loss": 0.2169,
      "step": 5337
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1948336191487878,
      "learning_rate": 6.405215851149144e-06,
      "loss": 0.1563,
      "step": 5338
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.056444540088652,
      "learning_rate": 6.403972516918507e-06,
      "loss": 0.2201,
      "step": 5339
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3438577444630693,
      "learning_rate": 6.402729088434942e-06,
      "loss": 0.2193,
      "step": 5340
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2765018940213793,
      "learning_rate": 6.4014855657819246e-06,
      "loss": 0.2036,
      "step": 5341
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1858498448097516,
      "learning_rate": 6.400241949042933e-06,
      "loss": 0.2113,
      "step": 5342
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5088534030750504,
      "learning_rate": 6.398998238301456e-06,
      "loss": 0.2028,
      "step": 5343
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2420723623247931,
      "learning_rate": 6.397754433640991e-06,
      "loss": 0.1948,
      "step": 5344
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.506780832860764,
      "learning_rate": 6.396510535145033e-06,
      "loss": 0.2077,
      "step": 5345
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.309676888282968,
      "learning_rate": 6.395266542897093e-06,
      "loss": 0.189,
      "step": 5346
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2831909095321687,
      "learning_rate": 6.394022456980682e-06,
      "loss": 0.2126,
      "step": 5347
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3228370512165486,
      "learning_rate": 6.3927782774793214e-06,
      "loss": 0.1712,
      "step": 5348
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4232312872525599,
      "learning_rate": 6.391534004476534e-06,
      "loss": 0.1921,
      "step": 5349
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5566428415730347,
      "learning_rate": 6.390289638055851e-06,
      "loss": 0.2024,
      "step": 5350
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4302528070834486,
      "learning_rate": 6.389045178300815e-06,
      "loss": 0.2213,
      "step": 5351
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2036733420859314,
      "learning_rate": 6.387800625294966e-06,
      "loss": 0.2098,
      "step": 5352
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3568128290007944,
      "learning_rate": 6.386555979121855e-06,
      "loss": 0.1777,
      "step": 5353
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4208999324762037,
      "learning_rate": 6.385311239865042e-06,
      "loss": 0.2511,
      "step": 5354
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2495037713264239,
      "learning_rate": 6.384066407608087e-06,
      "loss": 0.1849,
      "step": 5355
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3389680201022816,
      "learning_rate": 6.382821482434562e-06,
      "loss": 0.1879,
      "step": 5356
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.318057254232226,
      "learning_rate": 6.38157646442804e-06,
      "loss": 0.2092,
      "step": 5357
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2895159134210021,
      "learning_rate": 6.380331353672105e-06,
      "loss": 0.1769,
      "step": 5358
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5525758983728335,
      "learning_rate": 6.379086150250342e-06,
      "loss": 0.22,
      "step": 5359
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.886839484155791,
      "learning_rate": 6.377840854246348e-06,
      "loss": 0.4984,
      "step": 5360
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.453405324390962,
      "learning_rate": 6.376595465743722e-06,
      "loss": 0.2087,
      "step": 5361
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.318995101929686,
      "learning_rate": 6.375349984826074e-06,
      "loss": 0.1759,
      "step": 5362
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4293502038668422,
      "learning_rate": 6.374104411577012e-06,
      "loss": 0.206,
      "step": 5363
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1850791397816645,
      "learning_rate": 6.372858746080159e-06,
      "loss": 0.147,
      "step": 5364
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3467871873601427,
      "learning_rate": 6.371612988419138e-06,
      "loss": 0.1741,
      "step": 5365
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3322382241356419,
      "learning_rate": 6.370367138677582e-06,
      "loss": 0.1792,
      "step": 5366
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.400308125292723,
      "learning_rate": 6.3691211969391266e-06,
      "loss": 0.2148,
      "step": 5367
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.821442781429495,
      "learning_rate": 6.367875163287418e-06,
      "loss": 0.5054,
      "step": 5368
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4544178573250153,
      "learning_rate": 6.366629037806105e-06,
      "loss": 0.2052,
      "step": 5369
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.957213947031811,
      "learning_rate": 6.3653828205788445e-06,
      "loss": 0.5777,
      "step": 5370
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3073903040647563,
      "learning_rate": 6.3641365116892965e-06,
      "loss": 0.1755,
      "step": 5371
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.253763341566155,
      "learning_rate": 6.362890111221133e-06,
      "loss": 0.1935,
      "step": 5372
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5841188123003762,
      "learning_rate": 6.361643619258027e-06,
      "loss": 0.2106,
      "step": 5373
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7548271730162313,
      "learning_rate": 6.3603970358836574e-06,
      "loss": 0.3025,
      "step": 5374
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3665032836119813,
      "learning_rate": 6.3591503611817155e-06,
      "loss": 0.176,
      "step": 5375
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.47572798582419,
      "learning_rate": 6.35790359523589e-06,
      "loss": 0.2004,
      "step": 5376
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3820995546220687,
      "learning_rate": 6.356656738129882e-06,
      "loss": 0.2215,
      "step": 5377
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2976019738382976,
      "learning_rate": 6.355409789947398e-06,
      "loss": 0.2344,
      "step": 5378
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2573809721318219,
      "learning_rate": 6.354162750772146e-06,
      "loss": 0.1921,
      "step": 5379
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.119552202024639,
      "learning_rate": 6.352915620687848e-06,
      "loss": 0.5219,
      "step": 5380
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.218835308112296,
      "learning_rate": 6.3516683997782225e-06,
      "loss": 0.1535,
      "step": 5381
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.857860334268619,
      "learning_rate": 6.350421088127004e-06,
      "loss": 0.4885,
      "step": 5382
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2227129432668795,
      "learning_rate": 6.349173685817927e-06,
      "loss": 0.1624,
      "step": 5383
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3669306171470308,
      "learning_rate": 6.3479261929347305e-06,
      "loss": 0.2026,
      "step": 5384
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6137258380058008,
      "learning_rate": 6.346678609561166e-06,
      "loss": 0.2468,
      "step": 5385
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3476505329111346,
      "learning_rate": 6.345430935780985e-06,
      "loss": 0.221,
      "step": 5386
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3344670018044502,
      "learning_rate": 6.344183171677949e-06,
      "loss": 0.2284,
      "step": 5387
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3140374570908187,
      "learning_rate": 6.342935317335823e-06,
      "loss": 0.1876,
      "step": 5388
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4001148862724935,
      "learning_rate": 6.341687372838382e-06,
      "loss": 0.1719,
      "step": 5389
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5609426715699228,
      "learning_rate": 6.340439338269402e-06,
      "loss": 0.2016,
      "step": 5390
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5743830365323122,
      "learning_rate": 6.339191213712668e-06,
      "loss": 0.2402,
      "step": 5391
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3446240618991603,
      "learning_rate": 6.33794299925197e-06,
      "loss": 0.1891,
      "step": 5392
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2907959163255103,
      "learning_rate": 6.336694694971106e-06,
      "loss": 0.2085,
      "step": 5393
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2530580264484246,
      "learning_rate": 6.3354463009538745e-06,
      "loss": 0.1805,
      "step": 5394
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1342455516422136,
      "learning_rate": 6.3341978172840875e-06,
      "loss": 0.1655,
      "step": 5395
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2450955306857623,
      "learning_rate": 6.33294924404556e-06,
      "loss": 0.2093,
      "step": 5396
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2800083447208719,
      "learning_rate": 6.33170058132211e-06,
      "loss": 0.2005,
      "step": 5397
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3700922904488833,
      "learning_rate": 6.330451829197564e-06,
      "loss": 0.191,
      "step": 5398
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3869866598564893,
      "learning_rate": 6.329202987755757e-06,
      "loss": 0.206,
      "step": 5399
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4549250310680832,
      "learning_rate": 6.3279540570805265e-06,
      "loss": 0.2407,
      "step": 5400
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4708987271829916,
      "learning_rate": 6.326705037255716e-06,
      "loss": 0.187,
      "step": 5401
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2708330988907208,
      "learning_rate": 6.325455928365176e-06,
      "loss": 0.1981,
      "step": 5402
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3278930084033125,
      "learning_rate": 6.324206730492765e-06,
      "loss": 0.1947,
      "step": 5403
    },
    {
      "epoch": 0.43,
      "grad_norm": 11.631395118333689,
      "learning_rate": 6.322957443722343e-06,
      "loss": 0.5051,
      "step": 5404
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3414880933294182,
      "learning_rate": 6.321708068137778e-06,
      "loss": 0.2147,
      "step": 5405
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.273271877234986,
      "learning_rate": 6.320458603822949e-06,
      "loss": 0.1756,
      "step": 5406
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4433138667020462,
      "learning_rate": 6.319209050861731e-06,
      "loss": 0.24,
      "step": 5407
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4369750645434807,
      "learning_rate": 6.317959409338013e-06,
      "loss": 0.2386,
      "step": 5408
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3846730456639411,
      "learning_rate": 6.316709679335686e-06,
      "loss": 0.1833,
      "step": 5409
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6771928145798876,
      "learning_rate": 6.315459860938649e-06,
      "loss": 0.2428,
      "step": 5410
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2703693105782188,
      "learning_rate": 6.314209954230806e-06,
      "loss": 0.1877,
      "step": 5411
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4845559789211467,
      "learning_rate": 6.3129599592960665e-06,
      "loss": 0.24,
      "step": 5412
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4596300977817065,
      "learning_rate": 6.311709876218347e-06,
      "loss": 0.2345,
      "step": 5413
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2514090257319725,
      "learning_rate": 6.31045970508157e-06,
      "loss": 0.1602,
      "step": 5414
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4402163982994178,
      "learning_rate": 6.30920944596966e-06,
      "loss": 0.2106,
      "step": 5415
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2112359101090187,
      "learning_rate": 6.307959098966556e-06,
      "loss": 0.1752,
      "step": 5416
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4604133026948893,
      "learning_rate": 6.306708664156194e-06,
      "loss": 0.2568,
      "step": 5417
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3759924422272316,
      "learning_rate": 6.305458141622521e-06,
      "loss": 0.2077,
      "step": 5418
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2561711478936384,
      "learning_rate": 6.304207531449486e-06,
      "loss": 0.2085,
      "step": 5419
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3966676902267345,
      "learning_rate": 6.302956833721048e-06,
      "loss": 0.2205,
      "step": 5420
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.223770500207937,
      "learning_rate": 6.30170604852117e-06,
      "loss": 0.7511,
      "step": 5421
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.460352601838733,
      "learning_rate": 6.3004551759338206e-06,
      "loss": 0.636,
      "step": 5422
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.384592431863059,
      "learning_rate": 6.299204216042976e-06,
      "loss": 0.2199,
      "step": 5423
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.592492932771925,
      "learning_rate": 6.2979531689326155e-06,
      "loss": 0.5394,
      "step": 5424
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4973907236482074,
      "learning_rate": 6.296702034686726e-06,
      "loss": 0.2183,
      "step": 5425
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3011599398966487,
      "learning_rate": 6.2954508133893e-06,
      "loss": 0.1847,
      "step": 5426
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2221327505109223,
      "learning_rate": 6.294199505124337e-06,
      "loss": 0.1838,
      "step": 5427
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.371463085781701,
      "learning_rate": 6.292948109975839e-06,
      "loss": 0.2079,
      "step": 5428
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.537249608860019,
      "learning_rate": 6.291696628027816e-06,
      "loss": 0.2235,
      "step": 5429
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3682102613975526,
      "learning_rate": 6.290445059364286e-06,
      "loss": 0.1925,
      "step": 5430
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.479780457950757,
      "learning_rate": 6.28919340406927e-06,
      "loss": 0.2284,
      "step": 5431
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3605682904442127,
      "learning_rate": 6.287941662226793e-06,
      "loss": 0.2141,
      "step": 5432
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3746007412737256,
      "learning_rate": 6.286689833920889e-06,
      "loss": 0.182,
      "step": 5433
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3005447096920353,
      "learning_rate": 6.2854379192356e-06,
      "loss": 0.1809,
      "step": 5434
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3283008000523862,
      "learning_rate": 6.284185918254968e-06,
      "loss": 0.2169,
      "step": 5435
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.115410286688982,
      "learning_rate": 6.282933831063045e-06,
      "loss": 0.642,
      "step": 5436
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.441000115481333,
      "learning_rate": 6.2816816577438866e-06,
      "loss": 0.569,
      "step": 5437
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.273076868154803,
      "learning_rate": 6.280429398381555e-06,
      "loss": 0.1692,
      "step": 5438
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.7827199444185995,
      "learning_rate": 6.279177053060117e-06,
      "loss": 0.6681,
      "step": 5439
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3271552793988675,
      "learning_rate": 6.277924621863649e-06,
      "loss": 0.1838,
      "step": 5440
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2939994277796243,
      "learning_rate": 6.276672104876229e-06,
      "loss": 0.1762,
      "step": 5441
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.370757268235068,
      "learning_rate": 6.275419502181943e-06,
      "loss": 0.1866,
      "step": 5442
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.400731535434586,
      "learning_rate": 6.2741668138648806e-06,
      "loss": 0.7704,
      "step": 5443
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.571225792974094,
      "learning_rate": 6.27291404000914e-06,
      "loss": 0.1926,
      "step": 5444
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1794177233995102,
      "learning_rate": 6.271661180698824e-06,
      "loss": 0.1855,
      "step": 5445
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3001735561132977,
      "learning_rate": 6.2704082360180375e-06,
      "loss": 0.2326,
      "step": 5446
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.349825326070215,
      "learning_rate": 6.269155206050899e-06,
      "loss": 0.5371,
      "step": 5447
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5092478634018724,
      "learning_rate": 6.2679020908815245e-06,
      "loss": 0.2429,
      "step": 5448
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.320858193169152,
      "learning_rate": 6.266648890594042e-06,
      "loss": 0.2128,
      "step": 5449
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3437308437382531,
      "learning_rate": 6.265395605272581e-06,
      "loss": 0.1734,
      "step": 5450
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2217923784710367,
      "learning_rate": 6.264142235001281e-06,
      "loss": 0.1794,
      "step": 5451
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.557230988065406,
      "learning_rate": 6.26288877986428e-06,
      "loss": 0.6333,
      "step": 5452
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.256231249048313,
      "learning_rate": 6.261635239945732e-06,
      "loss": 0.1857,
      "step": 5453
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4754016489852595,
      "learning_rate": 6.260381615329785e-06,
      "loss": 0.1922,
      "step": 5454
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.357227173467612,
      "learning_rate": 6.259127906100601e-06,
      "loss": 0.1997,
      "step": 5455
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3222266067958681,
      "learning_rate": 6.257874112342347e-06,
      "loss": 0.1818,
      "step": 5456
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2855673511165826,
      "learning_rate": 6.256620234139193e-06,
      "loss": 0.168,
      "step": 5457
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.262650856266188,
      "learning_rate": 6.255366271575315e-06,
      "loss": 0.126,
      "step": 5458
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3290899737063937,
      "learning_rate": 6.254112224734895e-06,
      "loss": 0.2125,
      "step": 5459
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.219599941419597,
      "learning_rate": 6.252858093702121e-06,
      "loss": 0.1952,
      "step": 5460
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4908047188101465,
      "learning_rate": 6.251603878561188e-06,
      "loss": 0.2506,
      "step": 5461
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4753254138787635,
      "learning_rate": 6.250349579396295e-06,
      "loss": 0.2436,
      "step": 5462
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.217883190537527,
      "learning_rate": 6.249095196291646e-06,
      "loss": 0.5653,
      "step": 5463
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1909966359856916,
      "learning_rate": 6.247840729331451e-06,
      "loss": 0.1837,
      "step": 5464
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2949619029483872,
      "learning_rate": 6.246586178599928e-06,
      "loss": 0.1783,
      "step": 5465
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.962156945258954,
      "learning_rate": 6.2453315441812975e-06,
      "loss": 0.621,
      "step": 5466
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.831423156152444,
      "learning_rate": 6.2440768261597865e-06,
      "loss": 0.8104,
      "step": 5467
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3580329664368471,
      "learning_rate": 6.24282202461963e-06,
      "loss": 0.2384,
      "step": 5468
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.209385288965838,
      "learning_rate": 6.241567139645065e-06,
      "loss": 0.4342,
      "step": 5469
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1649479546866952,
      "learning_rate": 6.240312171320336e-06,
      "loss": 0.1927,
      "step": 5470
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.394883141638777,
      "learning_rate": 6.2390571197296936e-06,
      "loss": 0.1745,
      "step": 5471
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.314870600579029,
      "learning_rate": 6.237801984957391e-06,
      "loss": 0.1933,
      "step": 5472
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3783853084859452,
      "learning_rate": 6.236546767087692e-06,
      "loss": 0.2187,
      "step": 5473
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2286470445799385,
      "learning_rate": 6.2352914662048604e-06,
      "loss": 0.1878,
      "step": 5474
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2371273765443371,
      "learning_rate": 6.234036082393171e-06,
      "loss": 0.1897,
      "step": 5475
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2505549747840392,
      "learning_rate": 6.232780615736901e-06,
      "loss": 0.1684,
      "step": 5476
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2000233428359381,
      "learning_rate": 6.231525066320332e-06,
      "loss": 0.1626,
      "step": 5477
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2480338738938548,
      "learning_rate": 6.230269434227755e-06,
      "loss": 0.1684,
      "step": 5478
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2854966498737794,
      "learning_rate": 6.229013719543464e-06,
      "loss": 0.2099,
      "step": 5479
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1592498686536488,
      "learning_rate": 6.227757922351756e-06,
      "loss": 0.1745,
      "step": 5480
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4122571271400273,
      "learning_rate": 6.226502042736939e-06,
      "loss": 0.1942,
      "step": 5481
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2314177812709297,
      "learning_rate": 6.225246080783325e-06,
      "loss": 0.1888,
      "step": 5482
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.347114574653997,
      "learning_rate": 6.223990036575229e-06,
      "loss": 0.1892,
      "step": 5483
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1726567275688713,
      "learning_rate": 6.222733910196972e-06,
      "loss": 0.1747,
      "step": 5484
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.553796451949075,
      "learning_rate": 6.221477701732884e-06,
      "loss": 0.2099,
      "step": 5485
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.380067424922898,
      "learning_rate": 6.220221411267297e-06,
      "loss": 0.231,
      "step": 5486
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5622996637742816,
      "learning_rate": 6.21896503888455e-06,
      "loss": 0.2885,
      "step": 5487
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.720143303713319,
      "learning_rate": 6.217708584668987e-06,
      "loss": 0.3691,
      "step": 5488
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2716801930960457,
      "learning_rate": 6.216452048704955e-06,
      "loss": 0.1786,
      "step": 5489
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4166012931967793,
      "learning_rate": 6.215195431076813e-06,
      "loss": 0.1664,
      "step": 5490
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4545816326470928,
      "learning_rate": 6.2139387318689185e-06,
      "loss": 0.2201,
      "step": 5491
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3578224879221072,
      "learning_rate": 6.21268195116564e-06,
      "loss": 0.1461,
      "step": 5492
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.539719633864262,
      "learning_rate": 6.211425089051347e-06,
      "loss": 0.2283,
      "step": 5493
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.173553867842234,
      "learning_rate": 6.210168145610418e-06,
      "loss": 0.4451,
      "step": 5494
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1312940147704096,
      "learning_rate": 6.208911120927233e-06,
      "loss": 0.1791,
      "step": 5495
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3928524463616982,
      "learning_rate": 6.207654015086184e-06,
      "loss": 0.1841,
      "step": 5496
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2744903443220463,
      "learning_rate": 6.206396828171663e-06,
      "loss": 0.176,
      "step": 5497
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4127866017101978,
      "learning_rate": 6.205139560268064e-06,
      "loss": 0.2147,
      "step": 5498
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3276172543322058,
      "learning_rate": 6.203882211459797e-06,
      "loss": 0.1681,
      "step": 5499
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3212396820311856,
      "learning_rate": 6.202624781831269e-06,
      "loss": 0.184,
      "step": 5500
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3900304031340762,
      "learning_rate": 6.201367271466895e-06,
      "loss": 0.2036,
      "step": 5501
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.441158686958361,
      "learning_rate": 6.200109680451096e-06,
      "loss": 0.2241,
      "step": 5502
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3511503213529095,
      "learning_rate": 6.1988520088683e-06,
      "loss": 0.2151,
      "step": 5503
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.312562500955458,
      "learning_rate": 6.197594256802935e-06,
      "loss": 0.2131,
      "step": 5504
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.654466215232279,
      "learning_rate": 6.1963364243394386e-06,
      "loss": 0.2394,
      "step": 5505
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3353902310632164,
      "learning_rate": 6.195078511562254e-06,
      "loss": 0.166,
      "step": 5506
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.670980632638383,
      "learning_rate": 6.193820518555827e-06,
      "loss": 0.5743,
      "step": 5507
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.755538847897823,
      "learning_rate": 6.192562445404612e-06,
      "loss": 0.6666,
      "step": 5508
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.393267037979869,
      "learning_rate": 6.191304292193068e-06,
      "loss": 0.5784,
      "step": 5509
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.279109141399603,
      "learning_rate": 6.190046059005655e-06,
      "loss": 0.548,
      "step": 5510
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.396254156480553,
      "learning_rate": 6.188787745926846e-06,
      "loss": 0.2257,
      "step": 5511
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3046528698882116,
      "learning_rate": 6.187529353041113e-06,
      "loss": 0.1685,
      "step": 5512
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3150714232156226,
      "learning_rate": 6.186270880432936e-06,
      "loss": 0.1599,
      "step": 5513
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1934054809583754,
      "learning_rate": 6.185012328186802e-06,
      "loss": 0.1737,
      "step": 5514
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4152977136108575,
      "learning_rate": 6.183753696387199e-06,
      "loss": 0.1942,
      "step": 5515
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4628005285769916,
      "learning_rate": 6.182494985118625e-06,
      "loss": 0.2282,
      "step": 5516
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4022228907273728,
      "learning_rate": 6.1812361944655784e-06,
      "loss": 0.2254,
      "step": 5517
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.43762550993024,
      "learning_rate": 6.179977324512567e-06,
      "loss": 0.2311,
      "step": 5518
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3557578697331547,
      "learning_rate": 6.178718375344103e-06,
      "loss": 0.2178,
      "step": 5519
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.453597665248679,
      "learning_rate": 6.177459347044703e-06,
      "loss": 0.2336,
      "step": 5520
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.416425650169366,
      "learning_rate": 6.176200239698889e-06,
      "loss": 0.2397,
      "step": 5521
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3710737589143889,
      "learning_rate": 6.174941053391189e-06,
      "loss": 0.1926,
      "step": 5522
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.671137593532443,
      "learning_rate": 6.1736817882061385e-06,
      "loss": 0.2503,
      "step": 5523
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3179961342657962,
      "learning_rate": 6.172422444228272e-06,
      "loss": 0.2128,
      "step": 5524
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.393337914080416,
      "learning_rate": 6.171163021542134e-06,
      "loss": 0.2236,
      "step": 5525
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3255149159865447,
      "learning_rate": 6.169903520232274e-06,
      "loss": 0.1917,
      "step": 5526
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3670054755188386,
      "learning_rate": 6.168643940383246e-06,
      "loss": 0.206,
      "step": 5527
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5495651959630052,
      "learning_rate": 6.16738428207961e-06,
      "loss": 0.1928,
      "step": 5528
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2811045955224534,
      "learning_rate": 6.166124545405929e-06,
      "loss": 0.199,
      "step": 5529
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2526082792701527,
      "learning_rate": 6.164864730446776e-06,
      "loss": 0.157,
      "step": 5530
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2815294247890403,
      "learning_rate": 6.163604837286724e-06,
      "loss": 0.1915,
      "step": 5531
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2067983948364358,
      "learning_rate": 6.162344866010353e-06,
      "loss": 0.1536,
      "step": 5532
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.429560085596385,
      "learning_rate": 6.161084816702249e-06,
      "loss": 0.1911,
      "step": 5533
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2802597179999669,
      "learning_rate": 6.159824689447003e-06,
      "loss": 0.2086,
      "step": 5534
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4367970778529962,
      "learning_rate": 6.158564484329212e-06,
      "loss": 0.1738,
      "step": 5535
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3509319373357598,
      "learning_rate": 6.157304201433476e-06,
      "loss": 0.2163,
      "step": 5536
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.368101384719906,
      "learning_rate": 6.1560438408444036e-06,
      "loss": 0.1918,
      "step": 5537
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4244294899129617,
      "learning_rate": 6.154783402646604e-06,
      "loss": 0.1898,
      "step": 5538
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3431657336329828,
      "learning_rate": 6.153522886924699e-06,
      "loss": 0.218,
      "step": 5539
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3338828083822012,
      "learning_rate": 6.1522622937633044e-06,
      "loss": 0.1971,
      "step": 5540
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2819599058283644,
      "learning_rate": 6.151001623247053e-06,
      "loss": 0.201,
      "step": 5541
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1556123682361188,
      "learning_rate": 6.149740875460572e-06,
      "loss": 0.1393,
      "step": 5542
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.298658837335167,
      "learning_rate": 6.148480050488506e-06,
      "loss": 0.1823,
      "step": 5543
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4024613622380637,
      "learning_rate": 6.147219148415493e-06,
      "loss": 0.2345,
      "step": 5544
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.135464295700088,
      "learning_rate": 6.1459581693261825e-06,
      "loss": 0.4547,
      "step": 5545
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.401314532990553,
      "learning_rate": 6.144697113305227e-06,
      "loss": 0.2165,
      "step": 5546
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2613489075282303,
      "learning_rate": 6.1434359804372875e-06,
      "loss": 0.199,
      "step": 5547
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2985802411800909,
      "learning_rate": 6.142174770807026e-06,
      "loss": 0.1904,
      "step": 5548
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2106002498194715,
      "learning_rate": 6.140913484499112e-06,
      "loss": 0.2399,
      "step": 5549
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1654411161583529,
      "learning_rate": 6.139652121598219e-06,
      "loss": 0.1629,
      "step": 5550
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2350272423782207,
      "learning_rate": 6.138390682189025e-06,
      "loss": 0.1917,
      "step": 5551
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2548833117756106,
      "learning_rate": 6.137129166356215e-06,
      "loss": 0.1542,
      "step": 5552
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.050054719861633,
      "learning_rate": 6.135867574184479e-06,
      "loss": 0.2429,
      "step": 5553
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4161378157279216,
      "learning_rate": 6.1346059057585115e-06,
      "loss": 0.2011,
      "step": 5554
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.410310996773235,
      "learning_rate": 6.133344161163012e-06,
      "loss": 0.2138,
      "step": 5555
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4564407259706964,
      "learning_rate": 6.132082340482684e-06,
      "loss": 0.1962,
      "step": 5556
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3792279705961341,
      "learning_rate": 6.130820443802239e-06,
      "loss": 0.1718,
      "step": 5557
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2004016298538722,
      "learning_rate": 6.1295584712063935e-06,
      "loss": 0.1877,
      "step": 5558
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2603421856261134,
      "learning_rate": 6.1282964227798634e-06,
      "loss": 0.1477,
      "step": 5559
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3101276987398394,
      "learning_rate": 6.127034298607375e-06,
      "loss": 0.1878,
      "step": 5560
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3121595124182959,
      "learning_rate": 6.125772098773661e-06,
      "loss": 0.1367,
      "step": 5561
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3514288366573788,
      "learning_rate": 6.124509823363455e-06,
      "loss": 0.2133,
      "step": 5562
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.8074282650061395,
      "learning_rate": 6.123247472461495e-06,
      "loss": 0.4858,
      "step": 5563
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4291102593567688,
      "learning_rate": 6.121985046152531e-06,
      "loss": 0.1591,
      "step": 5564
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.259740245412773,
      "learning_rate": 6.120722544521312e-06,
      "loss": 0.2019,
      "step": 5565
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.283692223561614,
      "learning_rate": 6.119459967652592e-06,
      "loss": 0.2043,
      "step": 5566
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2647097728298455,
      "learning_rate": 6.1181973156311325e-06,
      "loss": 0.1936,
      "step": 5567
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.329348476490451,
      "learning_rate": 6.1169345885417e-06,
      "loss": 0.1856,
      "step": 5568
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3812634944996203,
      "learning_rate": 6.115671786469063e-06,
      "loss": 0.2377,
      "step": 5569
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4205204920987597,
      "learning_rate": 6.114408909497999e-06,
      "loss": 0.2124,
      "step": 5570
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.776555534942818,
      "learning_rate": 6.113145957713289e-06,
      "loss": 0.5159,
      "step": 5571
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.597878293882241,
      "learning_rate": 6.1118829311997176e-06,
      "loss": 0.5918,
      "step": 5572
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3860500762253893,
      "learning_rate": 6.110619830042076e-06,
      "loss": 0.1832,
      "step": 5573
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.364695393745401,
      "learning_rate": 6.109356654325161e-06,
      "loss": 0.5517,
      "step": 5574
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4641869690735012,
      "learning_rate": 6.108093404133772e-06,
      "loss": 0.2184,
      "step": 5575
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5404193350813713,
      "learning_rate": 6.106830079552716e-06,
      "loss": 0.2411,
      "step": 5576
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.30869960881159,
      "learning_rate": 6.105566680666802e-06,
      "loss": 0.2159,
      "step": 5577
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2486705095388573,
      "learning_rate": 6.104303207560848e-06,
      "loss": 0.1822,
      "step": 5578
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.346870614426665,
      "learning_rate": 6.103039660319674e-06,
      "loss": 0.2016,
      "step": 5579
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3585629896264042,
      "learning_rate": 6.101776039028104e-06,
      "loss": 0.2191,
      "step": 5580
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.510382047584163,
      "learning_rate": 6.100512343770971e-06,
      "loss": 0.2168,
      "step": 5581
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3974176902369686,
      "learning_rate": 6.099248574633111e-06,
      "loss": 0.2057,
      "step": 5582
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5247654221894222,
      "learning_rate": 6.097984731699363e-06,
      "loss": 0.2291,
      "step": 5583
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3384639025850897,
      "learning_rate": 6.096720815054573e-06,
      "loss": 0.1921,
      "step": 5584
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4158085095343402,
      "learning_rate": 6.095456824783592e-06,
      "loss": 0.2042,
      "step": 5585
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.224681070365773,
      "learning_rate": 6.094192760971275e-06,
      "loss": 0.1642,
      "step": 5586
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.041993714675057,
      "learning_rate": 6.092928623702481e-06,
      "loss": 0.5225,
      "step": 5587
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3422488633802798,
      "learning_rate": 6.091664413062079e-06,
      "loss": 0.206,
      "step": 5588
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6114028105561287,
      "learning_rate": 6.0904001291349375e-06,
      "loss": 0.2429,
      "step": 5589
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.4517315506449195,
      "learning_rate": 6.089135772005932e-06,
      "loss": 0.4918,
      "step": 5590
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3024740869565015,
      "learning_rate": 6.08787134175994e-06,
      "loss": 0.2013,
      "step": 5591
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.432279796238654,
      "learning_rate": 6.086606838481851e-06,
      "loss": 0.176,
      "step": 5592
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.355272647117586,
      "learning_rate": 6.085342262256552e-06,
      "loss": 0.1718,
      "step": 5593
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.523873437848471,
      "learning_rate": 6.084077613168937e-06,
      "loss": 0.2202,
      "step": 5594
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.773408011584644,
      "learning_rate": 6.0828128913039085e-06,
      "loss": 0.5324,
      "step": 5595
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2508901584474192,
      "learning_rate": 6.08154809674637e-06,
      "loss": 0.166,
      "step": 5596
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.217697802148777,
      "learning_rate": 6.080283229581231e-06,
      "loss": 0.1801,
      "step": 5597
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.627104879007245,
      "learning_rate": 6.079018289893403e-06,
      "loss": 0.2267,
      "step": 5598
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.078833819597533,
      "learning_rate": 6.07775327776781e-06,
      "loss": 0.1552,
      "step": 5599
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3062748246185414,
      "learning_rate": 6.076488193289375e-06,
      "loss": 0.2017,
      "step": 5600
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2232917849384808,
      "learning_rate": 6.075223036543025e-06,
      "loss": 0.1602,
      "step": 5601
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2833413040414363,
      "learning_rate": 6.073957807613695e-06,
      "loss": 0.2157,
      "step": 5602
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1804234185114044,
      "learning_rate": 6.072692506586324e-06,
      "loss": 0.2038,
      "step": 5603
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4884602595204453,
      "learning_rate": 6.071427133545854e-06,
      "loss": 0.2258,
      "step": 5604
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2132038919084498,
      "learning_rate": 6.070161688577233e-06,
      "loss": 0.1806,
      "step": 5605
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4203566420277562,
      "learning_rate": 6.068896171765417e-06,
      "loss": 0.2288,
      "step": 5606
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.7014609136692567,
      "learning_rate": 6.067630583195362e-06,
      "loss": 0.6135,
      "step": 5607
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4300728423441043,
      "learning_rate": 6.066364922952031e-06,
      "loss": 0.2081,
      "step": 5608
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6948475514341557,
      "learning_rate": 6.065099191120391e-06,
      "loss": 0.2098,
      "step": 5609
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5598582050049703,
      "learning_rate": 6.0638333877854185e-06,
      "loss": 0.2217,
      "step": 5610
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.449739306948539,
      "learning_rate": 6.0625675130320835e-06,
      "loss": 0.2092,
      "step": 5611
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.284292180522819,
      "learning_rate": 6.061301566945373e-06,
      "loss": 0.1859,
      "step": 5612
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3588099833381357,
      "learning_rate": 6.060035549610275e-06,
      "loss": 0.189,
      "step": 5613
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3395562956315887,
      "learning_rate": 6.0587694611117766e-06,
      "loss": 0.1987,
      "step": 5614
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2979559564066512,
      "learning_rate": 6.057503301534875e-06,
      "loss": 0.2057,
      "step": 5615
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4481850777496268,
      "learning_rate": 6.056237070964575e-06,
      "loss": 0.2206,
      "step": 5616
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.938497053781902,
      "learning_rate": 6.054970769485879e-06,
      "loss": 0.4551,
      "step": 5617
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.253788182750065,
      "learning_rate": 6.053704397183799e-06,
      "loss": 0.1832,
      "step": 5618
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2058164707408878,
      "learning_rate": 6.052437954143351e-06,
      "loss": 0.1604,
      "step": 5619
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.547492703478182,
      "learning_rate": 6.051171440449555e-06,
      "loss": 0.5551,
      "step": 5620
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4518981116268568,
      "learning_rate": 6.049904856187434e-06,
      "loss": 0.1742,
      "step": 5621
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3981648420954715,
      "learning_rate": 6.048638201442017e-06,
      "loss": 0.2264,
      "step": 5622
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.286440402540848,
      "learning_rate": 6.047371476298344e-06,
      "loss": 0.2318,
      "step": 5623
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3787880114042406,
      "learning_rate": 6.046104680841448e-06,
      "loss": 0.22,
      "step": 5624
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4142469301141967,
      "learning_rate": 6.044837815156377e-06,
      "loss": 0.1858,
      "step": 5625
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2901149794355038,
      "learning_rate": 6.043570879328175e-06,
      "loss": 0.2025,
      "step": 5626
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2919481080843338,
      "learning_rate": 6.042303873441902e-06,
      "loss": 0.1706,
      "step": 5627
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.314075484074311,
      "learning_rate": 6.041036797582608e-06,
      "loss": 0.4891,
      "step": 5628
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4270442518471307,
      "learning_rate": 6.039769651835361e-06,
      "loss": 0.1971,
      "step": 5629
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3790628357555104,
      "learning_rate": 6.038502436285227e-06,
      "loss": 0.1523,
      "step": 5630
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.636564401301105,
      "learning_rate": 6.037235151017279e-06,
      "loss": 0.5862,
      "step": 5631
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2471536399905925,
      "learning_rate": 6.03596779611659e-06,
      "loss": 0.1565,
      "step": 5632
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4267873799868975,
      "learning_rate": 6.034700371668246e-06,
      "loss": 0.2647,
      "step": 5633
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2920682812630875,
      "learning_rate": 6.033432877757331e-06,
      "loss": 0.1677,
      "step": 5634
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3201117633822932,
      "learning_rate": 6.032165314468935e-06,
      "loss": 0.1918,
      "step": 5635
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.483827022879092,
      "learning_rate": 6.030897681888155e-06,
      "loss": 0.2374,
      "step": 5636
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3375296221624304,
      "learning_rate": 6.029629980100089e-06,
      "loss": 0.1888,
      "step": 5637
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3157550498962585,
      "learning_rate": 6.028362209189844e-06,
      "loss": 0.2101,
      "step": 5638
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.0071345452294755,
      "learning_rate": 6.027094369242526e-06,
      "loss": 0.4409,
      "step": 5639
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.750425550519562,
      "learning_rate": 6.025826460343252e-06,
      "loss": 0.5792,
      "step": 5640
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.987667271162686,
      "learning_rate": 6.024558482577141e-06,
      "loss": 0.6997,
      "step": 5641
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.344991530575316,
      "learning_rate": 6.023290436029314e-06,
      "loss": 0.1684,
      "step": 5642
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3584051200200764,
      "learning_rate": 6.022022320784899e-06,
      "loss": 0.1928,
      "step": 5643
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4604053560977028,
      "learning_rate": 6.020754136929029e-06,
      "loss": 0.228,
      "step": 5644
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2634227166406855,
      "learning_rate": 6.0194858845468425e-06,
      "loss": 0.1956,
      "step": 5645
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.16182366887117,
      "learning_rate": 6.018217563723478e-06,
      "loss": 0.1623,
      "step": 5646
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.265971671798676,
      "learning_rate": 6.016949174544085e-06,
      "loss": 0.1498,
      "step": 5647
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4929605025272874,
      "learning_rate": 6.015680717093811e-06,
      "loss": 0.1946,
      "step": 5648
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5007342859565824,
      "learning_rate": 6.014412191457812e-06,
      "loss": 0.2284,
      "step": 5649
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2035926870614377,
      "learning_rate": 6.013143597721252e-06,
      "loss": 0.1776,
      "step": 5650
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.584367991225246,
      "learning_rate": 6.0118749359692905e-06,
      "loss": 0.6809,
      "step": 5651
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1988240255697657,
      "learning_rate": 6.0106062062871e-06,
      "loss": 0.1706,
      "step": 5652
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3731706831798078,
      "learning_rate": 6.0093374087598525e-06,
      "loss": 0.1947,
      "step": 5653
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3617248174587089,
      "learning_rate": 6.0080685434727274e-06,
      "loss": 0.2135,
      "step": 5654
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3413421229530602,
      "learning_rate": 6.006799610510905e-06,
      "loss": 0.1823,
      "step": 5655
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2931279087310887,
      "learning_rate": 6.005530609959573e-06,
      "loss": 0.1805,
      "step": 5656
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4282597823033496,
      "learning_rate": 6.004261541903928e-06,
      "loss": 0.2102,
      "step": 5657
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2873044151900763,
      "learning_rate": 6.002992406429161e-06,
      "loss": 0.2033,
      "step": 5658
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2912729124684637,
      "learning_rate": 6.001723203620475e-06,
      "loss": 0.1871,
      "step": 5659
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3609049903957566,
      "learning_rate": 6.000453933563075e-06,
      "loss": 0.1977,
      "step": 5660
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4618183956745632,
      "learning_rate": 5.999184596342171e-06,
      "loss": 0.1833,
      "step": 5661
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.806190844357243,
      "learning_rate": 5.99791519204298e-06,
      "loss": 0.4771,
      "step": 5662
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1282979207018777,
      "learning_rate": 5.996645720750715e-06,
      "loss": 0.1703,
      "step": 5663
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.347838831437371,
      "learning_rate": 5.995376182550607e-06,
      "loss": 0.1921,
      "step": 5664
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4978255033837349,
      "learning_rate": 5.994106577527877e-06,
      "loss": 0.187,
      "step": 5665
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2253919196219925,
      "learning_rate": 5.992836905767762e-06,
      "loss": 0.1425,
      "step": 5666
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3414929880676703,
      "learning_rate": 5.991567167355495e-06,
      "loss": 0.2127,
      "step": 5667
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.282444006567151,
      "learning_rate": 5.990297362376322e-06,
      "loss": 0.2019,
      "step": 5668
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1778619482730026,
      "learning_rate": 5.989027490915485e-06,
      "loss": 0.1778,
      "step": 5669
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.257621079603397,
      "learning_rate": 5.987757553058236e-06,
      "loss": 0.2048,
      "step": 5670
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4901630198688518,
      "learning_rate": 5.986487548889832e-06,
      "loss": 0.2166,
      "step": 5671
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3094962804900228,
      "learning_rate": 5.9852174784955295e-06,
      "loss": 0.2678,
      "step": 5672
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.014818765928443,
      "learning_rate": 5.98394734196059e-06,
      "loss": 0.6236,
      "step": 5673
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.6065058986841345,
      "learning_rate": 5.982677139370287e-06,
      "loss": 0.6142,
      "step": 5674
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5048809190438395,
      "learning_rate": 5.981406870809889e-06,
      "loss": 0.2041,
      "step": 5675
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4422487109970714,
      "learning_rate": 5.980136536364673e-06,
      "loss": 0.2198,
      "step": 5676
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1994914231536837,
      "learning_rate": 5.9788661361199225e-06,
      "loss": 0.1967,
      "step": 5677
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3100926971962215,
      "learning_rate": 5.977595670160923e-06,
      "loss": 0.2053,
      "step": 5678
    },
    {
      "epoch": 0.45,
      "grad_norm": 13.02904922119633,
      "learning_rate": 5.976325138572964e-06,
      "loss": 0.7019,
      "step": 5679
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2956763540161398,
      "learning_rate": 5.9750545414413405e-06,
      "loss": 0.1794,
      "step": 5680
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.370151891092701,
      "learning_rate": 5.973783878851352e-06,
      "loss": 0.165,
      "step": 5681
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4211267576150197,
      "learning_rate": 5.972513150888299e-06,
      "loss": 0.231,
      "step": 5682
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.325627164389678,
      "learning_rate": 5.971242357637493e-06,
      "loss": 0.1591,
      "step": 5683
    },
    {
      "epoch": 0.45,
      "grad_norm": 14.933694127378622,
      "learning_rate": 5.969971499184244e-06,
      "loss": 0.6261,
      "step": 5684
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.093963592626661,
      "learning_rate": 5.96870057561387e-06,
      "loss": 0.1688,
      "step": 5685
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.258413838332183,
      "learning_rate": 5.967429587011691e-06,
      "loss": 0.6146,
      "step": 5686
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2926402999629654,
      "learning_rate": 5.966158533463033e-06,
      "loss": 0.1888,
      "step": 5687
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4847246449457343,
      "learning_rate": 5.964887415053227e-06,
      "loss": 0.1921,
      "step": 5688
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2726212771325793,
      "learning_rate": 5.963616231867602e-06,
      "loss": 0.1724,
      "step": 5689
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4581039557863034,
      "learning_rate": 5.962344983991503e-06,
      "loss": 0.1874,
      "step": 5690
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.645827002510589,
      "learning_rate": 5.9610736715102665e-06,
      "loss": 0.4644,
      "step": 5691
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3629035524712785,
      "learning_rate": 5.959802294509244e-06,
      "loss": 0.1968,
      "step": 5692
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2604163114797136,
      "learning_rate": 5.958530853073785e-06,
      "loss": 0.1611,
      "step": 5693
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.190489500479314,
      "learning_rate": 5.9572593472892445e-06,
      "loss": 0.5745,
      "step": 5694
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.294593535149013,
      "learning_rate": 5.955987777240985e-06,
      "loss": 0.2126,
      "step": 5695
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4208797127469808,
      "learning_rate": 5.95471614301437e-06,
      "loss": 0.1693,
      "step": 5696
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3067241364760294,
      "learning_rate": 5.953444444694767e-06,
      "loss": 0.2158,
      "step": 5697
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4441227441203894,
      "learning_rate": 5.95217268236755e-06,
      "loss": 0.1995,
      "step": 5698
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1814100998528794,
      "learning_rate": 5.950900856118096e-06,
      "loss": 0.1588,
      "step": 5699
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6285410725517602,
      "learning_rate": 5.949628966031785e-06,
      "loss": 0.2223,
      "step": 5700
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3199946041519168,
      "learning_rate": 5.948357012194005e-06,
      "loss": 0.1559,
      "step": 5701
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4954503441931697,
      "learning_rate": 5.947084994690145e-06,
      "loss": 0.2042,
      "step": 5702
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.356241854830979,
      "learning_rate": 5.945812913605601e-06,
      "loss": 0.188,
      "step": 5703
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3174144523348514,
      "learning_rate": 5.94454076902577e-06,
      "loss": 0.2139,
      "step": 5704
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1567834956873886,
      "learning_rate": 5.943268561036053e-06,
      "loss": 0.1483,
      "step": 5705
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5628398791804923,
      "learning_rate": 5.941996289721863e-06,
      "loss": 0.2171,
      "step": 5706
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3378853522278507,
      "learning_rate": 5.940723955168607e-06,
      "loss": 0.2095,
      "step": 5707
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3210256914346519,
      "learning_rate": 5.9394515574617e-06,
      "loss": 0.1842,
      "step": 5708
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6147589943459284,
      "learning_rate": 5.938179096686565e-06,
      "loss": 0.2171,
      "step": 5709
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.873899094459051,
      "learning_rate": 5.936906572928625e-06,
      "loss": 0.4317,
      "step": 5710
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2456435787960665,
      "learning_rate": 5.935633986273307e-06,
      "loss": 0.2166,
      "step": 5711
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.032914263274818,
      "learning_rate": 5.934361336806044e-06,
      "loss": 0.5615,
      "step": 5712
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2757499417853844,
      "learning_rate": 5.933088624612275e-06,
      "loss": 0.1554,
      "step": 5713
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4026906978634766,
      "learning_rate": 5.931815849777438e-06,
      "loss": 0.2516,
      "step": 5714
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3715539984439826,
      "learning_rate": 5.930543012386981e-06,
      "loss": 0.2451,
      "step": 5715
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.379447543188886,
      "learning_rate": 5.9292701125263515e-06,
      "loss": 0.2052,
      "step": 5716
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.285892812830182,
      "learning_rate": 5.927997150281004e-06,
      "loss": 0.1758,
      "step": 5717
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3339860754384354,
      "learning_rate": 5.926724125736396e-06,
      "loss": 0.2394,
      "step": 5718
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2272098958799753,
      "learning_rate": 5.925451038977989e-06,
      "loss": 0.1563,
      "step": 5719
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3270458506433893,
      "learning_rate": 5.924177890091251e-06,
      "loss": 0.189,
      "step": 5720
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3614365340955805,
      "learning_rate": 5.9229046791616505e-06,
      "loss": 0.2221,
      "step": 5721
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.092818923295385,
      "learning_rate": 5.921631406274661e-06,
      "loss": 0.4837,
      "step": 5722
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.413445133560478,
      "learning_rate": 5.920358071515766e-06,
      "loss": 0.1999,
      "step": 5723
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2682108646771542,
      "learning_rate": 5.919084674970443e-06,
      "loss": 0.1951,
      "step": 5724
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.420338640257742,
      "learning_rate": 5.9178112167241805e-06,
      "loss": 0.3494,
      "step": 5725
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.632667086552824,
      "learning_rate": 5.916537696862472e-06,
      "loss": 0.408,
      "step": 5726
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3209886840816925,
      "learning_rate": 5.91526411547081e-06,
      "loss": 0.1838,
      "step": 5727
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7830084521710565,
      "learning_rate": 5.913990472634695e-06,
      "loss": 0.228,
      "step": 5728
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3595171001940738,
      "learning_rate": 5.91271676843963e-06,
      "loss": 0.1775,
      "step": 5729
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2953320928459553,
      "learning_rate": 5.911443002971122e-06,
      "loss": 0.2041,
      "step": 5730
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2234037329831313,
      "learning_rate": 5.910169176314686e-06,
      "loss": 0.1534,
      "step": 5731
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2125689024486903,
      "learning_rate": 5.908895288555833e-06,
      "loss": 0.1843,
      "step": 5732
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1193579859698313,
      "learning_rate": 5.907621339780087e-06,
      "loss": 0.1814,
      "step": 5733
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1590477776574561,
      "learning_rate": 5.906347330072971e-06,
      "loss": 0.1606,
      "step": 5734
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.328192785851466,
      "learning_rate": 5.905073259520011e-06,
      "loss": 0.2261,
      "step": 5735
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1594818549014685,
      "learning_rate": 5.903799128206742e-06,
      "loss": 0.1606,
      "step": 5736
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.486957350480733,
      "learning_rate": 5.9025249362186985e-06,
      "loss": 0.2091,
      "step": 5737
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1073424003443189,
      "learning_rate": 5.901250683641422e-06,
      "loss": 0.1425,
      "step": 5738
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3545557141082347,
      "learning_rate": 5.899976370560455e-06,
      "loss": 0.1942,
      "step": 5739
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5788262608892951,
      "learning_rate": 5.898701997061349e-06,
      "loss": 0.1949,
      "step": 5740
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3268574334232077,
      "learning_rate": 5.897427563229655e-06,
      "loss": 0.2029,
      "step": 5741
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.414097255396974,
      "learning_rate": 5.896153069150928e-06,
      "loss": 0.2226,
      "step": 5742
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.853680107674938,
      "learning_rate": 5.894878514910731e-06,
      "loss": 0.6264,
      "step": 5743
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3845440847702832,
      "learning_rate": 5.893603900594629e-06,
      "loss": 0.1979,
      "step": 5744
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4010949960206476,
      "learning_rate": 5.89232922628819e-06,
      "loss": 0.1982,
      "step": 5745
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2942396864942263,
      "learning_rate": 5.891054492076984e-06,
      "loss": 0.191,
      "step": 5746
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.969964930270931,
      "learning_rate": 5.889779698046592e-06,
      "loss": 0.6958,
      "step": 5747
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3059372612615812,
      "learning_rate": 5.888504844282592e-06,
      "loss": 0.1786,
      "step": 5748
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.180403595993147,
      "learning_rate": 5.88722993087057e-06,
      "loss": 0.2091,
      "step": 5749
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3895937458109564,
      "learning_rate": 5.885954957896115e-06,
      "loss": 0.1827,
      "step": 5750
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2088660694052178,
      "learning_rate": 5.884679925444818e-06,
      "loss": 0.1576,
      "step": 5751
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3201010823692692,
      "learning_rate": 5.883404833602279e-06,
      "loss": 0.1649,
      "step": 5752
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.294232614173875,
      "learning_rate": 5.882129682454095e-06,
      "loss": 0.1718,
      "step": 5753
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.340744119901396,
      "learning_rate": 5.880854472085874e-06,
      "loss": 0.1718,
      "step": 5754
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3492284195686604,
      "learning_rate": 5.879579202583223e-06,
      "loss": 0.1923,
      "step": 5755
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3599677684101235,
      "learning_rate": 5.878303874031755e-06,
      "loss": 0.1989,
      "step": 5756
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3669908663394268,
      "learning_rate": 5.8770284865170865e-06,
      "loss": 0.2295,
      "step": 5757
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4500462428166774,
      "learning_rate": 5.87575304012484e-06,
      "loss": 0.18,
      "step": 5758
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4826992627403754,
      "learning_rate": 5.874477534940637e-06,
      "loss": 0.2172,
      "step": 5759
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4292364450017767,
      "learning_rate": 5.8732019710501075e-06,
      "loss": 0.2389,
      "step": 5760
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.187264464229376,
      "learning_rate": 5.871926348538885e-06,
      "loss": 0.4665,
      "step": 5761
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.386842895695305,
      "learning_rate": 5.870650667492604e-06,
      "loss": 0.2178,
      "step": 5762
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4285108558467636,
      "learning_rate": 5.869374927996905e-06,
      "loss": 0.2073,
      "step": 5763
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3935972831777168,
      "learning_rate": 5.868099130137434e-06,
      "loss": 0.215,
      "step": 5764
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.909830183041856,
      "learning_rate": 5.866823273999839e-06,
      "loss": 0.6254,
      "step": 5765
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3075461101427168,
      "learning_rate": 5.86554735966977e-06,
      "loss": 0.1809,
      "step": 5766
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2870174549240483,
      "learning_rate": 5.8642713872328845e-06,
      "loss": 0.1848,
      "step": 5767
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6485922931013042,
      "learning_rate": 5.862995356774843e-06,
      "loss": 0.1914,
      "step": 5768
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.243615690870402,
      "learning_rate": 5.861719268381307e-06,
      "loss": 0.1704,
      "step": 5769
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3619467382787693,
      "learning_rate": 5.860443122137946e-06,
      "loss": 0.1705,
      "step": 5770
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0968087815841516,
      "learning_rate": 5.859166918130432e-06,
      "loss": 0.1478,
      "step": 5771
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3551066786591732,
      "learning_rate": 5.857890656444439e-06,
      "loss": 0.1842,
      "step": 5772
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.337925809998819,
      "learning_rate": 5.856614337165647e-06,
      "loss": 0.1692,
      "step": 5773
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5104521816073047,
      "learning_rate": 5.8553379603797386e-06,
      "loss": 0.2455,
      "step": 5774
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.353913514979569,
      "learning_rate": 5.854061526172402e-06,
      "loss": 0.1958,
      "step": 5775
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4517306800177738,
      "learning_rate": 5.852785034629329e-06,
      "loss": 0.2464,
      "step": 5776
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3837920810602258,
      "learning_rate": 5.85150848583621e-06,
      "loss": 0.2338,
      "step": 5777
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.314685172807877,
      "learning_rate": 5.850231879878747e-06,
      "loss": 0.1781,
      "step": 5778
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2990389312399357,
      "learning_rate": 5.848955216842642e-06,
      "loss": 0.2097,
      "step": 5779
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2672165508534958,
      "learning_rate": 5.847678496813601e-06,
      "loss": 0.1729,
      "step": 5780
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.705583601276164,
      "learning_rate": 5.846401719877335e-06,
      "loss": 0.5704,
      "step": 5781
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2461234946688042,
      "learning_rate": 5.845124886119556e-06,
      "loss": 0.1703,
      "step": 5782
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.369370036417176,
      "learning_rate": 5.843847995625984e-06,
      "loss": 0.5121,
      "step": 5783
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2670824094483304,
      "learning_rate": 5.842571048482339e-06,
      "loss": 0.2214,
      "step": 5784
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.117239106485012,
      "learning_rate": 5.841294044774346e-06,
      "loss": 0.5197,
      "step": 5785
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3293867536367963,
      "learning_rate": 5.840016984587736e-06,
      "loss": 0.1961,
      "step": 5786
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.309808647074832,
      "learning_rate": 5.838739868008239e-06,
      "loss": 0.1565,
      "step": 5787
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4313639370345095,
      "learning_rate": 5.837462695121593e-06,
      "loss": 0.1925,
      "step": 5788
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3495533721692934,
      "learning_rate": 5.836185466013541e-06,
      "loss": 0.221,
      "step": 5789
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2619622877080199,
      "learning_rate": 5.834908180769824e-06,
      "loss": 0.1705,
      "step": 5790
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3666402266397282,
      "learning_rate": 5.83363083947619e-06,
      "loss": 0.228,
      "step": 5791
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.35107387972115,
      "learning_rate": 5.832353442218394e-06,
      "loss": 0.1915,
      "step": 5792
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2699050647588939,
      "learning_rate": 5.8310759890821884e-06,
      "loss": 0.1785,
      "step": 5793
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.236727980432842,
      "learning_rate": 5.829798480153333e-06,
      "loss": 0.1952,
      "step": 5794
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6037418210265613,
      "learning_rate": 5.828520915517593e-06,
      "loss": 0.2455,
      "step": 5795
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3022392744380422,
      "learning_rate": 5.827243295260733e-06,
      "loss": 0.1775,
      "step": 5796
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.388907440391088,
      "learning_rate": 5.825965619468524e-06,
      "loss": 0.2072,
      "step": 5797
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5415974744874508,
      "learning_rate": 5.82468788822674e-06,
      "loss": 0.2371,
      "step": 5798
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.242236943177887,
      "learning_rate": 5.82341010162116e-06,
      "loss": 0.1594,
      "step": 5799
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2637850756378195,
      "learning_rate": 5.822132259737565e-06,
      "loss": 0.1702,
      "step": 5800
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3314161899047547,
      "learning_rate": 5.8208543626617384e-06,
      "loss": 0.2131,
      "step": 5801
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.856706037233241,
      "learning_rate": 5.819576410479475e-06,
      "loss": 0.7953,
      "step": 5802
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4039522225306276,
      "learning_rate": 5.818298403276562e-06,
      "loss": 0.1534,
      "step": 5803
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2401778434788044,
      "learning_rate": 5.817020341138797e-06,
      "loss": 0.1792,
      "step": 5804
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3521341055963747,
      "learning_rate": 5.815742224151982e-06,
      "loss": 0.2126,
      "step": 5805
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.766174482820119,
      "learning_rate": 5.81446405240192e-06,
      "loss": 0.6264,
      "step": 5806
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4822014620439405,
      "learning_rate": 5.813185825974419e-06,
      "loss": 0.1925,
      "step": 5807
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2822804112503057,
      "learning_rate": 5.8119075449552885e-06,
      "loss": 0.1998,
      "step": 5808
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.637119501339548,
      "learning_rate": 5.8106292094303455e-06,
      "loss": 0.2073,
      "step": 5809
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4689000906776288,
      "learning_rate": 5.809350819485408e-06,
      "loss": 0.1762,
      "step": 5810
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4548121029579029,
      "learning_rate": 5.8080723752062955e-06,
      "loss": 0.2188,
      "step": 5811
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3463070708671028,
      "learning_rate": 5.806793876678838e-06,
      "loss": 0.195,
      "step": 5812
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4188589745907383,
      "learning_rate": 5.805515323988861e-06,
      "loss": 0.2223,
      "step": 5813
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2170357974782944,
      "learning_rate": 5.8042367172222025e-06,
      "loss": 0.2197,
      "step": 5814
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4724450339418798,
      "learning_rate": 5.802958056464694e-06,
      "loss": 0.2129,
      "step": 5815
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2905244170609236,
      "learning_rate": 5.80167934180218e-06,
      "loss": 0.1791,
      "step": 5816
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7781235668903357,
      "learning_rate": 5.800400573320502e-06,
      "loss": 0.2247,
      "step": 5817
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1915795460943785,
      "learning_rate": 5.799121751105507e-06,
      "loss": 0.1451,
      "step": 5818
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1996631586277335,
      "learning_rate": 5.7978428752430514e-06,
      "loss": 0.185,
      "step": 5819
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.290831565658375,
      "learning_rate": 5.796563945818984e-06,
      "loss": 0.2359,
      "step": 5820
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3212469880577664,
      "learning_rate": 5.795284962919164e-06,
      "loss": 0.1888,
      "step": 5821
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3889234367911285,
      "learning_rate": 5.794005926629457e-06,
      "loss": 0.1733,
      "step": 5822
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.337997496160111,
      "learning_rate": 5.792726837035726e-06,
      "loss": 0.2107,
      "step": 5823
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3234147492119162,
      "learning_rate": 5.791447694223841e-06,
      "loss": 0.2065,
      "step": 5824
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2316160911624536,
      "learning_rate": 5.7901684982796716e-06,
      "loss": 0.202,
      "step": 5825
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3519214390539434,
      "learning_rate": 5.788889249289099e-06,
      "loss": 0.2171,
      "step": 5826
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2540029835453093,
      "learning_rate": 5.787609947338001e-06,
      "loss": 0.1871,
      "step": 5827
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4241020949385594,
      "learning_rate": 5.78633059251226e-06,
      "loss": 0.212,
      "step": 5828
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3598279855358968,
      "learning_rate": 5.785051184897765e-06,
      "loss": 0.1728,
      "step": 5829
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.494814722974306,
      "learning_rate": 5.783771724580405e-06,
      "loss": 0.4617,
      "step": 5830
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1829142961754717,
      "learning_rate": 5.7824922116460735e-06,
      "loss": 0.1614,
      "step": 5831
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3843450315038315,
      "learning_rate": 5.78121264618067e-06,
      "loss": 0.1878,
      "step": 5832
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.334292893401392,
      "learning_rate": 5.779933028270093e-06,
      "loss": 0.166,
      "step": 5833
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.053860655002488,
      "learning_rate": 5.7786533580002506e-06,
      "loss": 0.4752,
      "step": 5834
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2691181839248344,
      "learning_rate": 5.777373635457049e-06,
      "loss": 0.1865,
      "step": 5835
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3650917512092926,
      "learning_rate": 5.7760938607263975e-06,
      "loss": 0.2186,
      "step": 5836
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.236867975576756,
      "learning_rate": 5.774814033894216e-06,
      "loss": 0.4932,
      "step": 5837
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3468026191871252,
      "learning_rate": 5.773534155046421e-06,
      "loss": 0.1712,
      "step": 5838
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4395188842457471,
      "learning_rate": 5.772254224268932e-06,
      "loss": 0.2044,
      "step": 5839
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5538177638515203,
      "learning_rate": 5.770974241647679e-06,
      "loss": 0.1908,
      "step": 5840
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2333528050865397,
      "learning_rate": 5.769694207268589e-06,
      "loss": 0.1652,
      "step": 5841
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4798651122671294,
      "learning_rate": 5.768414121217594e-06,
      "loss": 0.2026,
      "step": 5842
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3405146552385974,
      "learning_rate": 5.767133983580631e-06,
      "loss": 0.195,
      "step": 5843
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.361730241694057,
      "learning_rate": 5.765853794443641e-06,
      "loss": 0.6309,
      "step": 5844
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3086734966051299,
      "learning_rate": 5.764573553892564e-06,
      "loss": 0.1713,
      "step": 5845
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.409697746302821,
      "learning_rate": 5.763293262013347e-06,
      "loss": 0.169,
      "step": 5846
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4562134309812649,
      "learning_rate": 5.762012918891942e-06,
      "loss": 0.265,
      "step": 5847
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4400144460057551,
      "learning_rate": 5.760732524614301e-06,
      "loss": 0.1949,
      "step": 5848
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3043056604803445,
      "learning_rate": 5.75945207926638e-06,
      "loss": 0.2008,
      "step": 5849
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4397428459139365,
      "learning_rate": 5.75817158293414e-06,
      "loss": 0.2239,
      "step": 5850
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.211075218272331,
      "learning_rate": 5.756891035703544e-06,
      "loss": 0.5079,
      "step": 5851
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3321834904562107,
      "learning_rate": 5.755610437660561e-06,
      "loss": 0.1867,
      "step": 5852
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.47952047368519,
      "learning_rate": 5.754329788891158e-06,
      "loss": 0.1934,
      "step": 5853
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2760948316488285,
      "learning_rate": 5.753049089481313e-06,
      "loss": 0.1737,
      "step": 5854
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.310965478818646,
      "learning_rate": 5.751768339517e-06,
      "loss": 0.1757,
      "step": 5855
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3163542643126205,
      "learning_rate": 5.750487539084199e-06,
      "loss": 0.189,
      "step": 5856
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.420325486405605,
      "learning_rate": 5.749206688268897e-06,
      "loss": 0.3969,
      "step": 5857
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2956025285766046,
      "learning_rate": 5.747925787157081e-06,
      "loss": 0.191,
      "step": 5858
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4050429722620523,
      "learning_rate": 5.746644835834739e-06,
      "loss": 0.2031,
      "step": 5859
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3983575554920347,
      "learning_rate": 5.745363834387867e-06,
      "loss": 0.204,
      "step": 5860
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.017820838264491,
      "learning_rate": 5.744082782902463e-06,
      "loss": 0.5403,
      "step": 5861
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.284492374152771,
      "learning_rate": 5.742801681464527e-06,
      "loss": 0.1742,
      "step": 5862
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2886651121809594,
      "learning_rate": 5.741520530160064e-06,
      "loss": 0.1976,
      "step": 5863
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4065906291576153,
      "learning_rate": 5.740239329075081e-06,
      "loss": 0.2076,
      "step": 5864
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5034303781768974,
      "learning_rate": 5.73895807829559e-06,
      "loss": 0.1867,
      "step": 5865
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4697077068130926,
      "learning_rate": 5.737676777907602e-06,
      "loss": 0.1817,
      "step": 5866
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2086837213418151,
      "learning_rate": 5.736395427997138e-06,
      "loss": 0.166,
      "step": 5867
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4738967135035692,
      "learning_rate": 5.735114028650218e-06,
      "loss": 0.2564,
      "step": 5868
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2838124182342827,
      "learning_rate": 5.733832579952866e-06,
      "loss": 0.2041,
      "step": 5869
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2221743945868093,
      "learning_rate": 5.732551081991109e-06,
      "loss": 0.2034,
      "step": 5870
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3185088313696427,
      "learning_rate": 5.73126953485098e-06,
      "loss": 0.1796,
      "step": 5871
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.411420920784928,
      "learning_rate": 5.72998793861851e-06,
      "loss": 0.1986,
      "step": 5872
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7464490396659782,
      "learning_rate": 5.7287062933797365e-06,
      "loss": 0.2386,
      "step": 5873
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3327533738172825,
      "learning_rate": 5.727424599220703e-06,
      "loss": 0.2058,
      "step": 5874
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2560669788549168,
      "learning_rate": 5.726142856227453e-06,
      "loss": 0.1699,
      "step": 5875
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.035683554512246,
      "learning_rate": 5.724861064486031e-06,
      "loss": 0.5508,
      "step": 5876
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3009862270019332,
      "learning_rate": 5.723579224082491e-06,
      "loss": 0.1817,
      "step": 5877
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.257225330468414,
      "learning_rate": 5.7222973351028844e-06,
      "loss": 0.1518,
      "step": 5878
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3699903431022236,
      "learning_rate": 5.7210153976332696e-06,
      "loss": 0.1661,
      "step": 5879
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3002580818699006,
      "learning_rate": 5.719733411759707e-06,
      "loss": 0.1875,
      "step": 5880
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2086167438034907,
      "learning_rate": 5.7184513775682605e-06,
      "loss": 0.1636,
      "step": 5881
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2602920816050203,
      "learning_rate": 5.717169295144995e-06,
      "loss": 0.2114,
      "step": 5882
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3376660719533946,
      "learning_rate": 5.715887164575981e-06,
      "loss": 0.2312,
      "step": 5883
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2270800968932485,
      "learning_rate": 5.714604985947294e-06,
      "loss": 0.2333,
      "step": 5884
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4387223071593043,
      "learning_rate": 5.713322759345008e-06,
      "loss": 0.2358,
      "step": 5885
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3536225540000226,
      "learning_rate": 5.7120404848552056e-06,
      "loss": 0.1677,
      "step": 5886
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3537446329711529,
      "learning_rate": 5.710758162563965e-06,
      "loss": 0.1232,
      "step": 5887
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4819349973493916,
      "learning_rate": 5.709475792557379e-06,
      "loss": 0.2099,
      "step": 5888
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.243377464387006,
      "learning_rate": 5.708193374921533e-06,
      "loss": 0.1558,
      "step": 5889
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2313001480346943,
      "learning_rate": 5.7069109097425176e-06,
      "loss": 0.1999,
      "step": 5890
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3738288377187922,
      "learning_rate": 5.705628397106432e-06,
      "loss": 0.1989,
      "step": 5891
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.393464058116012,
      "learning_rate": 5.704345837099375e-06,
      "loss": 0.2253,
      "step": 5892
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3059650101949498,
      "learning_rate": 5.703063229807448e-06,
      "loss": 0.1969,
      "step": 5893
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.374530431275325,
      "learning_rate": 5.7017805753167545e-06,
      "loss": 0.2187,
      "step": 5894
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2885387102987482,
      "learning_rate": 5.700497873713405e-06,
      "loss": 0.1997,
      "step": 5895
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5326136160042831,
      "learning_rate": 5.699215125083512e-06,
      "loss": 0.2194,
      "step": 5896
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4584974092145213,
      "learning_rate": 5.6979323295131884e-06,
      "loss": 0.2084,
      "step": 5897
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4232275509842094,
      "learning_rate": 5.696649487088553e-06,
      "loss": 0.1867,
      "step": 5898
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6035620190930906,
      "learning_rate": 5.695366597895728e-06,
      "loss": 0.2144,
      "step": 5899
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5175640086121123,
      "learning_rate": 5.694083662020835e-06,
      "loss": 0.2135,
      "step": 5900
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1972701145487568,
      "learning_rate": 5.692800679550001e-06,
      "loss": 0.1597,
      "step": 5901
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.258488786213888,
      "learning_rate": 5.691517650569361e-06,
      "loss": 0.2291,
      "step": 5902
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0905339964775163,
      "learning_rate": 5.690234575165046e-06,
      "loss": 0.1782,
      "step": 5903
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.028483657076705,
      "learning_rate": 5.68895145342319e-06,
      "loss": 0.6673,
      "step": 5904
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5262428659966436,
      "learning_rate": 5.6876682854299385e-06,
      "loss": 0.182,
      "step": 5905
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4484051930873665,
      "learning_rate": 5.686385071271431e-06,
      "loss": 0.1834,
      "step": 5906
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3945717792931787,
      "learning_rate": 5.685101811033812e-06,
      "loss": 0.1878,
      "step": 5907
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3857285946671793,
      "learning_rate": 5.683818504803234e-06,
      "loss": 0.1653,
      "step": 5908
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3715151116159963,
      "learning_rate": 5.682535152665848e-06,
      "loss": 0.1985,
      "step": 5909
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2376143862653577,
      "learning_rate": 5.68125175470781e-06,
      "loss": 0.163,
      "step": 5910
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3063222372355119,
      "learning_rate": 5.679968311015275e-06,
      "loss": 0.2132,
      "step": 5911
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4300750249723542,
      "learning_rate": 5.678684821674409e-06,
      "loss": 0.2122,
      "step": 5912
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.765728847845143,
      "learning_rate": 5.677401286771373e-06,
      "loss": 0.2316,
      "step": 5913
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.366396651773216,
      "learning_rate": 5.676117706392339e-06,
      "loss": 0.2013,
      "step": 5914
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3375263478720958,
      "learning_rate": 5.674834080623472e-06,
      "loss": 0.2112,
      "step": 5915
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.538273385881684,
      "learning_rate": 5.673550409550951e-06,
      "loss": 0.2257,
      "step": 5916
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2713483112332928,
      "learning_rate": 5.67226669326095e-06,
      "loss": 0.1735,
      "step": 5917
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.415427280468103,
      "learning_rate": 5.6709829318396456e-06,
      "loss": 0.2196,
      "step": 5918
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4678369190488758,
      "learning_rate": 5.669699125373227e-06,
      "loss": 0.2155,
      "step": 5919
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.421014437684305,
      "learning_rate": 5.668415273947876e-06,
      "loss": 0.1763,
      "step": 5920
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3098766809686797,
      "learning_rate": 5.667131377649784e-06,
      "loss": 0.1829,
      "step": 5921
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.297367671952655,
      "learning_rate": 5.665847436565138e-06,
      "loss": 0.1827,
      "step": 5922
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1710379004700813,
      "learning_rate": 5.664563450780139e-06,
      "loss": 0.172,
      "step": 5923
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3894756830200443,
      "learning_rate": 5.663279420380983e-06,
      "loss": 0.2088,
      "step": 5924
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3302788824046827,
      "learning_rate": 5.661995345453867e-06,
      "loss": 0.2155,
      "step": 5925
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3791295597542517,
      "learning_rate": 5.660711226084999e-06,
      "loss": 0.1899,
      "step": 5926
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4059490903014586,
      "learning_rate": 5.659427062360584e-06,
      "loss": 0.2186,
      "step": 5927
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4752211888247597,
      "learning_rate": 5.658142854366831e-06,
      "loss": 0.2037,
      "step": 5928
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3130994824645255,
      "learning_rate": 5.656858602189956e-06,
      "loss": 0.1606,
      "step": 5929
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1639218661617001,
      "learning_rate": 5.655574305916173e-06,
      "loss": 0.1405,
      "step": 5930
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3979191580013663,
      "learning_rate": 5.6542899656317e-06,
      "loss": 0.1972,
      "step": 5931
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3898160957879981,
      "learning_rate": 5.653005581422759e-06,
      "loss": 0.1771,
      "step": 5932
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1920173219901289,
      "learning_rate": 5.651721153375576e-06,
      "loss": 0.1647,
      "step": 5933
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3017918184142756,
      "learning_rate": 5.650436681576377e-06,
      "loss": 0.1946,
      "step": 5934
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4025265945770609,
      "learning_rate": 5.64915216611139e-06,
      "loss": 0.2162,
      "step": 5935
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2808322477519614,
      "learning_rate": 5.647867607066855e-06,
      "loss": 0.1863,
      "step": 5936
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5272249065075783,
      "learning_rate": 5.646583004529003e-06,
      "loss": 0.2421,
      "step": 5937
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2115062389307953,
      "learning_rate": 5.645298358584075e-06,
      "loss": 0.2119,
      "step": 5938
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6579623764689173,
      "learning_rate": 5.644013669318311e-06,
      "loss": 0.2306,
      "step": 5939
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3569769686010955,
      "learning_rate": 5.64272893681796e-06,
      "loss": 0.1833,
      "step": 5940
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4221628216846194,
      "learning_rate": 5.64144416116927e-06,
      "loss": 0.2212,
      "step": 5941
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2463620184319113,
      "learning_rate": 5.640159342458487e-06,
      "loss": 0.2008,
      "step": 5942
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.575088956870121,
      "learning_rate": 5.638874480771869e-06,
      "loss": 0.2093,
      "step": 5943
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.157262309479802,
      "learning_rate": 5.637589576195671e-06,
      "loss": 0.168,
      "step": 5944
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2743218871551016,
      "learning_rate": 5.636304628816153e-06,
      "loss": 0.2018,
      "step": 5945
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.16500939068641,
      "learning_rate": 5.635019638719576e-06,
      "loss": 0.7054,
      "step": 5946
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1912768510895222,
      "learning_rate": 5.633734605992209e-06,
      "loss": 0.1937,
      "step": 5947
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4151370847374525,
      "learning_rate": 5.632449530720316e-06,
      "loss": 0.1904,
      "step": 5948
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.279416978014094,
      "learning_rate": 5.63116441299017e-06,
      "loss": 0.1693,
      "step": 5949
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3088873934527676,
      "learning_rate": 5.629879252888046e-06,
      "loss": 0.2056,
      "step": 5950
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.719504061327706,
      "learning_rate": 5.628594050500219e-06,
      "loss": 0.6353,
      "step": 5951
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3079975175667797,
      "learning_rate": 5.627308805912968e-06,
      "loss": 0.1778,
      "step": 5952
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4124124821782975,
      "learning_rate": 5.626023519212576e-06,
      "loss": 0.1655,
      "step": 5953
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2133313438798814,
      "learning_rate": 5.624738190485329e-06,
      "loss": 0.2016,
      "step": 5954
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1789270584811615,
      "learning_rate": 5.623452819817514e-06,
      "loss": 0.1295,
      "step": 5955
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3169481574395738,
      "learning_rate": 5.622167407295422e-06,
      "loss": 0.1836,
      "step": 5956
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.238564384781399,
      "learning_rate": 5.620881953005347e-06,
      "loss": 0.1798,
      "step": 5957
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3320269991842364,
      "learning_rate": 5.619596457033586e-06,
      "loss": 0.2275,
      "step": 5958
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.201729636212621,
      "learning_rate": 5.618310919466438e-06,
      "loss": 0.1929,
      "step": 5959
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4499074421537543,
      "learning_rate": 5.617025340390203e-06,
      "loss": 0.2656,
      "step": 5960
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.469263467714341,
      "learning_rate": 5.615739719891189e-06,
      "loss": 0.2073,
      "step": 5961
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.429833157384644,
      "learning_rate": 5.6144540580557e-06,
      "loss": 0.225,
      "step": 5962
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.392256258350648,
      "learning_rate": 5.6131683549700485e-06,
      "loss": 0.2064,
      "step": 5963
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.858762271755936,
      "learning_rate": 5.6118826107205496e-06,
      "loss": 0.6358,
      "step": 5964
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2196587850290672,
      "learning_rate": 5.610596825393516e-06,
      "loss": 0.1841,
      "step": 5965
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3803692772876153,
      "learning_rate": 5.609310999075268e-06,
      "loss": 0.1905,
      "step": 5966
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.232699013594476,
      "learning_rate": 5.608025131852126e-06,
      "loss": 0.1736,
      "step": 5967
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.485105631993959,
      "learning_rate": 5.6067392238104146e-06,
      "loss": 0.1898,
      "step": 5968
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.862350150569696,
      "learning_rate": 5.60545327503646e-06,
      "loss": 0.2372,
      "step": 5969
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3580474404830163,
      "learning_rate": 5.604167285616593e-06,
      "loss": 0.1976,
      "step": 5970
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.446761445578654,
      "learning_rate": 5.602881255637146e-06,
      "loss": 0.2183,
      "step": 5971
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1936653914287894,
      "learning_rate": 5.601595185184454e-06,
      "loss": 0.1638,
      "step": 5972
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4927953956870545,
      "learning_rate": 5.600309074344853e-06,
      "loss": 0.2181,
      "step": 5973
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.360328532071457,
      "learning_rate": 5.599022923204686e-06,
      "loss": 0.1994,
      "step": 5974
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4884517014081409,
      "learning_rate": 5.597736731850295e-06,
      "loss": 0.2026,
      "step": 5975
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.255441033656892,
      "learning_rate": 5.596450500368027e-06,
      "loss": 0.1966,
      "step": 5976
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3900141456820991,
      "learning_rate": 5.595164228844229e-06,
      "loss": 0.2173,
      "step": 5977
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2471511088921274,
      "learning_rate": 5.5938779173652535e-06,
      "loss": 0.1602,
      "step": 5978
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.158041466196222,
      "learning_rate": 5.592591566017454e-06,
      "loss": 0.1781,
      "step": 5979
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2836872740864629,
      "learning_rate": 5.591305174887185e-06,
      "loss": 0.1909,
      "step": 5980
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.946203578021819,
      "learning_rate": 5.59001874406081e-06,
      "loss": 0.4099,
      "step": 5981
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3868069436072263,
      "learning_rate": 5.588732273624689e-06,
      "loss": 0.1599,
      "step": 5982
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3702607578814197,
      "learning_rate": 5.587445763665187e-06,
      "loss": 0.2301,
      "step": 5983
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.457836279442077,
      "learning_rate": 5.586159214268669e-06,
      "loss": 0.2196,
      "step": 5984
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3278533013278107,
      "learning_rate": 5.58487262552151e-06,
      "loss": 0.1875,
      "step": 5985
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.516481069045735,
      "learning_rate": 5.583585997510078e-06,
      "loss": 0.2084,
      "step": 5986
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4088266528754982,
      "learning_rate": 5.582299330320749e-06,
      "loss": 0.1867,
      "step": 5987
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4743423439972172,
      "learning_rate": 5.581012624039903e-06,
      "loss": 0.249,
      "step": 5988
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3722060138197059,
      "learning_rate": 5.579725878753919e-06,
      "loss": 0.2041,
      "step": 5989
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5027671316126925,
      "learning_rate": 5.5784390945491784e-06,
      "loss": 0.2116,
      "step": 5990
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3441156452901943,
      "learning_rate": 5.577152271512071e-06,
      "loss": 0.2041,
      "step": 5991
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4045807658684673,
      "learning_rate": 5.5758654097289824e-06,
      "loss": 0.2105,
      "step": 5992
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.543958337108886,
      "learning_rate": 5.574578509286304e-06,
      "loss": 0.1992,
      "step": 5993
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.526558778526334,
      "learning_rate": 5.573291570270429e-06,
      "loss": 0.5413,
      "step": 5994
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5278364801345237,
      "learning_rate": 5.572004592767755e-06,
      "loss": 0.2295,
      "step": 5995
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3808539834434894,
      "learning_rate": 5.5707175768646784e-06,
      "loss": 0.174,
      "step": 5996
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.445568872702685,
      "learning_rate": 5.569430522647601e-06,
      "loss": 0.2244,
      "step": 5997
    },
    {
      "epoch": 0.48,
      "grad_norm": 30.60855702119436,
      "learning_rate": 5.568143430202928e-06,
      "loss": 0.484,
      "step": 5998
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5442187412434536,
      "learning_rate": 5.566856299617066e-06,
      "loss": 0.2039,
      "step": 5999
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3797616433037831,
      "learning_rate": 5.5655691309764225e-06,
      "loss": 0.2199,
      "step": 6000
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4139187635704322,
      "learning_rate": 5.5642819243674085e-06,
      "loss": 0.1939,
      "step": 6001
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4039016937741262,
      "learning_rate": 5.562994679876441e-06,
      "loss": 0.2376,
      "step": 6002
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.399714381052771,
      "learning_rate": 5.561707397589935e-06,
      "loss": 0.6284,
      "step": 6003
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4063465962241557,
      "learning_rate": 5.560420077594307e-06,
      "loss": 0.2326,
      "step": 6004
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3055414923882211,
      "learning_rate": 5.559132719975984e-06,
      "loss": 0.2075,
      "step": 6005
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2155928884225948,
      "learning_rate": 5.557845324821386e-06,
      "loss": 0.1754,
      "step": 6006
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2317597862453442,
      "learning_rate": 5.556557892216942e-06,
      "loss": 0.1806,
      "step": 6007
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.349210599719671,
      "learning_rate": 5.555270422249078e-06,
      "loss": 0.2299,
      "step": 6008
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4786096939073454,
      "learning_rate": 5.55398291500423e-06,
      "loss": 0.2549,
      "step": 6009
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7833786326948462,
      "learning_rate": 5.55269537056883e-06,
      "loss": 0.2124,
      "step": 6010
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4568108102214778,
      "learning_rate": 5.551407789029316e-06,
      "loss": 0.1815,
      "step": 6011
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3239751020441697,
      "learning_rate": 5.5501201704721266e-06,
      "loss": 0.19,
      "step": 6012
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3995322905941363,
      "learning_rate": 5.548832514983703e-06,
      "loss": 0.1975,
      "step": 6013
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.269746204973812,
      "learning_rate": 5.547544822650488e-06,
      "loss": 0.1727,
      "step": 6014
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4053687868927875,
      "learning_rate": 5.546257093558932e-06,
      "loss": 0.2032,
      "step": 6015
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.289738644628163,
      "learning_rate": 5.544969327795482e-06,
      "loss": 0.1843,
      "step": 6016
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3345296602743995,
      "learning_rate": 5.543681525446589e-06,
      "loss": 0.1714,
      "step": 6017
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2410913009582902,
      "learning_rate": 5.542393686598708e-06,
      "loss": 0.1347,
      "step": 6018
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3634781234050608,
      "learning_rate": 5.541105811338297e-06,
      "loss": 0.2115,
      "step": 6019
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4052460359533734,
      "learning_rate": 5.539817899751813e-06,
      "loss": 0.1802,
      "step": 6020
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4927289049843615,
      "learning_rate": 5.538529951925715e-06,
      "loss": 0.2104,
      "step": 6021
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2700050933356997,
      "learning_rate": 5.5372419679464725e-06,
      "loss": 0.1818,
      "step": 6022
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4948610127561053,
      "learning_rate": 5.5359539479005484e-06,
      "loss": 0.2066,
      "step": 6023
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1395970175907957,
      "learning_rate": 5.534665891874412e-06,
      "loss": 0.1411,
      "step": 6024
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3544384489584773,
      "learning_rate": 5.533377799954532e-06,
      "loss": 0.2128,
      "step": 6025
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4352742278683133,
      "learning_rate": 5.5320896722273874e-06,
      "loss": 0.2623,
      "step": 6026
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3174046615725108,
      "learning_rate": 5.5308015087794495e-06,
      "loss": 0.2173,
      "step": 6027
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.414602247754823,
      "learning_rate": 5.529513309697199e-06,
      "loss": 0.1897,
      "step": 6028
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.115487324121038,
      "learning_rate": 5.528225075067115e-06,
      "loss": 0.5867,
      "step": 6029
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5257386718151094,
      "learning_rate": 5.526936804975681e-06,
      "loss": 0.2314,
      "step": 6030
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.1303691661322866,
      "learning_rate": 5.525648499509383e-06,
      "loss": 0.6563,
      "step": 6031
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.503807382485949,
      "learning_rate": 5.524360158754708e-06,
      "loss": 0.2346,
      "step": 6032
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3938833340305543,
      "learning_rate": 5.523071782798147e-06,
      "loss": 0.2088,
      "step": 6033
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3300527781747324,
      "learning_rate": 5.521783371726193e-06,
      "loss": 0.1978,
      "step": 6034
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4444783328031934,
      "learning_rate": 5.520494925625339e-06,
      "loss": 0.1984,
      "step": 6035
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.280272864196463,
      "learning_rate": 5.519206444582085e-06,
      "loss": 0.1675,
      "step": 6036
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2930897811512372,
      "learning_rate": 5.51791792868293e-06,
      "loss": 0.2036,
      "step": 6037
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4211660092670169,
      "learning_rate": 5.516629378014373e-06,
      "loss": 0.2298,
      "step": 6038
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3377702612608382,
      "learning_rate": 5.515340792662922e-06,
      "loss": 0.2013,
      "step": 6039
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3960500478992524,
      "learning_rate": 5.5140521727150805e-06,
      "loss": 0.2534,
      "step": 6040
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4054634909028954,
      "learning_rate": 5.512763518257362e-06,
      "loss": 0.2258,
      "step": 6041
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5050815306855954,
      "learning_rate": 5.511474829376272e-06,
      "loss": 0.222,
      "step": 6042
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.565493234571033,
      "learning_rate": 5.510186106158329e-06,
      "loss": 0.4654,
      "step": 6043
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4397003624647202,
      "learning_rate": 5.5088973486900456e-06,
      "loss": 0.2364,
      "step": 6044
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.342030012261675,
      "learning_rate": 5.507608557057942e-06,
      "loss": 0.1906,
      "step": 6045
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.390965026917466,
      "learning_rate": 5.506319731348539e-06,
      "loss": 0.1829,
      "step": 6046
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2511271304861076,
      "learning_rate": 5.505030871648356e-06,
      "loss": 0.1746,
      "step": 6047
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.165958927056673,
      "learning_rate": 5.503741978043922e-06,
      "loss": 0.1879,
      "step": 6048
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4198084956443682,
      "learning_rate": 5.502453050621761e-06,
      "loss": 0.181,
      "step": 6049
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3648605611379505,
      "learning_rate": 5.501164089468406e-06,
      "loss": 0.2156,
      "step": 6050
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.411973970066241,
      "learning_rate": 5.499875094670387e-06,
      "loss": 0.1811,
      "step": 6051
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3006616472067594,
      "learning_rate": 5.4985860663142385e-06,
      "loss": 0.1661,
      "step": 6052
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5537058731460205,
      "learning_rate": 5.497297004486496e-06,
      "loss": 0.2095,
      "step": 6053
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4273018204910448,
      "learning_rate": 5.496007909273699e-06,
      "loss": 0.1773,
      "step": 6054
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3600341815047596,
      "learning_rate": 5.494718780762388e-06,
      "loss": 0.1932,
      "step": 6055
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3610735010958077,
      "learning_rate": 5.493429619039108e-06,
      "loss": 0.169,
      "step": 6056
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2882254305545053,
      "learning_rate": 5.492140424190401e-06,
      "loss": 0.1918,
      "step": 6057
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1444241060391807,
      "learning_rate": 5.490851196302817e-06,
      "loss": 0.1703,
      "step": 6058
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.213837916339109,
      "learning_rate": 5.489561935462904e-06,
      "loss": 0.1834,
      "step": 6059
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3141471728474285,
      "learning_rate": 5.488272641757215e-06,
      "loss": 0.1933,
      "step": 6060
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.261598207816413,
      "learning_rate": 5.486983315272305e-06,
      "loss": 0.6498,
      "step": 6061
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2221484568860745,
      "learning_rate": 5.485693956094728e-06,
      "loss": 0.1865,
      "step": 6062
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.539107532247561,
      "learning_rate": 5.484404564311047e-06,
      "loss": 0.2627,
      "step": 6063
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4492156246395256,
      "learning_rate": 5.483115140007818e-06,
      "loss": 0.2487,
      "step": 6064
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.140482901593522,
      "learning_rate": 5.481825683271607e-06,
      "loss": 0.1545,
      "step": 6065
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4175042877607387,
      "learning_rate": 5.480536194188976e-06,
      "loss": 0.2465,
      "step": 6066
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2329482824055489,
      "learning_rate": 5.479246672846496e-06,
      "loss": 0.1732,
      "step": 6067
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3903366091057725,
      "learning_rate": 5.477957119330733e-06,
      "loss": 0.224,
      "step": 6068
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4428585946416677,
      "learning_rate": 5.476667533728262e-06,
      "loss": 0.2187,
      "step": 6069
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1899966022709014,
      "learning_rate": 5.475377916125655e-06,
      "loss": 0.1638,
      "step": 6070
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2652472675313262,
      "learning_rate": 5.474088266609487e-06,
      "loss": 0.1775,
      "step": 6071
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2609118241822819,
      "learning_rate": 5.47279858526634e-06,
      "loss": 0.1528,
      "step": 6072
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.402843492300553,
      "learning_rate": 5.471508872182789e-06,
      "loss": 0.2316,
      "step": 6073
    },
    {
      "epoch": 0.49,
      "grad_norm": 11.304191971833744,
      "learning_rate": 5.47021912744542e-06,
      "loss": 0.5415,
      "step": 6074
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.322918411822587,
      "learning_rate": 5.4689293511408155e-06,
      "loss": 0.1855,
      "step": 6075
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2671456809530721,
      "learning_rate": 5.4676395433555615e-06,
      "loss": 0.1922,
      "step": 6076
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5184119140894168,
      "learning_rate": 5.46634970417625e-06,
      "loss": 0.1976,
      "step": 6077
    },
    {
      "epoch": 0.49,
      "grad_norm": 17.840925550591802,
      "learning_rate": 5.46505983368947e-06,
      "loss": 0.6233,
      "step": 6078
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5639266702624965,
      "learning_rate": 5.463769931981815e-06,
      "loss": 0.1949,
      "step": 6079
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4228140682918635,
      "learning_rate": 5.462479999139877e-06,
      "loss": 0.2233,
      "step": 6080
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.375151402801113,
      "learning_rate": 5.4611900352502574e-06,
      "loss": 0.1913,
      "step": 6081
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3851701151267617,
      "learning_rate": 5.459900040399552e-06,
      "loss": 0.1993,
      "step": 6082
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2932943105828254,
      "learning_rate": 5.458610014674363e-06,
      "loss": 0.1792,
      "step": 6083
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2596133349248404,
      "learning_rate": 5.457319958161295e-06,
      "loss": 0.1564,
      "step": 6084
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3313526912600322,
      "learning_rate": 5.456029870946954e-06,
      "loss": 0.1681,
      "step": 6085
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3035092700781619,
      "learning_rate": 5.454739753117944e-06,
      "loss": 0.176,
      "step": 6086
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2871509654360385,
      "learning_rate": 5.453449604760878e-06,
      "loss": 0.188,
      "step": 6087
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3753092076388826,
      "learning_rate": 5.4521594259623655e-06,
      "loss": 0.2036,
      "step": 6088
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.363233269141916,
      "learning_rate": 5.450869216809021e-06,
      "loss": 0.2049,
      "step": 6089
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2825721200453803,
      "learning_rate": 5.44957897738746e-06,
      "loss": 0.1387,
      "step": 6090
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.388168632814305,
      "learning_rate": 5.4482887077843e-06,
      "loss": 0.2289,
      "step": 6091
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3390687257731622,
      "learning_rate": 5.44699840808616e-06,
      "loss": 0.1927,
      "step": 6092
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4729524171846566,
      "learning_rate": 5.445708078379663e-06,
      "loss": 0.1899,
      "step": 6093
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.347943526888609,
      "learning_rate": 5.444417718751431e-06,
      "loss": 0.207,
      "step": 6094
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.335775746631166,
      "learning_rate": 5.443127329288093e-06,
      "loss": 0.1852,
      "step": 6095
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2708180955374653,
      "learning_rate": 5.441836910076275e-06,
      "loss": 0.197,
      "step": 6096
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2967646940082138,
      "learning_rate": 5.440546461202604e-06,
      "loss": 0.1898,
      "step": 6097
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3603966974977955,
      "learning_rate": 5.439255982753717e-06,
      "loss": 0.1728,
      "step": 6098
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3746809234063302,
      "learning_rate": 5.4379654748162435e-06,
      "loss": 0.1836,
      "step": 6099
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3264065959567641,
      "learning_rate": 5.43667493747682e-06,
      "loss": 0.1944,
      "step": 6100
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.578750562526942,
      "learning_rate": 5.4353843708220865e-06,
      "loss": 0.5186,
      "step": 6101
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4646458769165451,
      "learning_rate": 5.43409377493868e-06,
      "loss": 0.2153,
      "step": 6102
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3860030342104832,
      "learning_rate": 5.432803149913244e-06,
      "loss": 0.2266,
      "step": 6103
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4788283261826667,
      "learning_rate": 5.4315124958324205e-06,
      "loss": 0.246,
      "step": 6104
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.21950214440904,
      "learning_rate": 5.430221812782856e-06,
      "loss": 0.165,
      "step": 6105
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4704757530527062,
      "learning_rate": 5.428931100851198e-06,
      "loss": 0.2184,
      "step": 6106
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4050667092213878,
      "learning_rate": 5.427640360124095e-06,
      "loss": 0.2192,
      "step": 6107
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3206500519779876,
      "learning_rate": 5.426349590688199e-06,
      "loss": 0.1986,
      "step": 6108
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6990625831773556,
      "learning_rate": 5.425058792630163e-06,
      "loss": 0.1585,
      "step": 6109
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1943211238988904,
      "learning_rate": 5.423767966036644e-06,
      "loss": 0.198,
      "step": 6110
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4373194386978287,
      "learning_rate": 5.422477110994295e-06,
      "loss": 0.1889,
      "step": 6111
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3241384100961962,
      "learning_rate": 5.421186227589779e-06,
      "loss": 0.196,
      "step": 6112
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.575186805440506,
      "learning_rate": 5.419895315909755e-06,
      "loss": 0.539,
      "step": 6113
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.276069312805348,
      "learning_rate": 5.418604376040887e-06,
      "loss": 0.1621,
      "step": 6114
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.485312551419629,
      "learning_rate": 5.417313408069839e-06,
      "loss": 0.7423,
      "step": 6115
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3610200067186389,
      "learning_rate": 5.416022412083276e-06,
      "loss": 0.1678,
      "step": 6116
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2769873974180446,
      "learning_rate": 5.414731388167868e-06,
      "loss": 0.1781,
      "step": 6117
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4663467415968185,
      "learning_rate": 5.413440336410286e-06,
      "loss": 0.191,
      "step": 6118
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2838528189271403,
      "learning_rate": 5.412149256897201e-06,
      "loss": 0.2256,
      "step": 6119
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3625456885057272,
      "learning_rate": 5.410858149715289e-06,
      "loss": 0.1975,
      "step": 6120
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2569962244624773,
      "learning_rate": 5.409567014951223e-06,
      "loss": 0.2135,
      "step": 6121
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4678340695307412,
      "learning_rate": 5.408275852691683e-06,
      "loss": 0.1664,
      "step": 6122
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3579318190859855,
      "learning_rate": 5.406984663023348e-06,
      "loss": 0.1666,
      "step": 6123
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3600183234899443,
      "learning_rate": 5.4056934460329e-06,
      "loss": 0.1842,
      "step": 6124
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.208867291872801,
      "learning_rate": 5.404402201807022e-06,
      "loss": 0.7251,
      "step": 6125
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3840224244543287,
      "learning_rate": 5.403110930432398e-06,
      "loss": 0.19,
      "step": 6126
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6366906229699234,
      "learning_rate": 5.401819631995716e-06,
      "loss": 0.2274,
      "step": 6127
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.369461161231918,
      "learning_rate": 5.400528306583664e-06,
      "loss": 0.1861,
      "step": 6128
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3699924571907363,
      "learning_rate": 5.3992369542829335e-06,
      "loss": 0.2263,
      "step": 6129
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.28937723540913,
      "learning_rate": 5.3979455751802175e-06,
      "loss": 0.1754,
      "step": 6130
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3526611584679595,
      "learning_rate": 5.3966541693622095e-06,
      "loss": 0.1899,
      "step": 6131
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3562503652014586,
      "learning_rate": 5.395362736915604e-06,
      "loss": 0.1992,
      "step": 6132
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9131155075453725,
      "learning_rate": 5.394071277927102e-06,
      "loss": 0.201,
      "step": 6133
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2981926205240792,
      "learning_rate": 5.3927797924834e-06,
      "loss": 0.2092,
      "step": 6134
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.301547721791799,
      "learning_rate": 5.391488280671199e-06,
      "loss": 0.2314,
      "step": 6135
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.257158430939577,
      "learning_rate": 5.3901967425772054e-06,
      "loss": 0.1872,
      "step": 6136
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3260619773101494,
      "learning_rate": 5.388905178288122e-06,
      "loss": 0.2128,
      "step": 6137
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3568862506992065,
      "learning_rate": 5.387613587890655e-06,
      "loss": 0.2113,
      "step": 6138
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3278776492878688,
      "learning_rate": 5.386321971471513e-06,
      "loss": 0.2359,
      "step": 6139
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2872565173550587,
      "learning_rate": 5.3850303291174076e-06,
      "loss": 0.1757,
      "step": 6140
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.843111840020467,
      "learning_rate": 5.383738660915049e-06,
      "loss": 0.5242,
      "step": 6141
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3557938662274296,
      "learning_rate": 5.382446966951153e-06,
      "loss": 0.1909,
      "step": 6142
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.513328734571758,
      "learning_rate": 5.381155247312432e-06,
      "loss": 0.2315,
      "step": 6143
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4264195707805678,
      "learning_rate": 5.379863502085604e-06,
      "loss": 0.208,
      "step": 6144
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.482948563359553,
      "learning_rate": 5.378571731357388e-06,
      "loss": 0.2074,
      "step": 6145
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3249209265795843,
      "learning_rate": 5.377279935214506e-06,
      "loss": 0.1813,
      "step": 6146
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1861088988847668,
      "learning_rate": 5.375988113743677e-06,
      "loss": 0.1519,
      "step": 6147
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.92504338989327,
      "learning_rate": 5.374696267031628e-06,
      "loss": 0.4747,
      "step": 6148
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.625763186027714,
      "learning_rate": 5.3734043951650825e-06,
      "loss": 0.2403,
      "step": 6149
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.276998895442994,
      "learning_rate": 5.372112498230771e-06,
      "loss": 0.189,
      "step": 6150
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3248365580973975,
      "learning_rate": 5.3708205763154185e-06,
      "loss": 0.189,
      "step": 6151
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2857699143347978,
      "learning_rate": 5.369528629505756e-06,
      "loss": 0.1867,
      "step": 6152
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.52676283686413,
      "learning_rate": 5.368236657888518e-06,
      "loss": 0.1901,
      "step": 6153
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1932605242514978,
      "learning_rate": 5.366944661550437e-06,
      "loss": 0.1661,
      "step": 6154
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.381219198443967,
      "learning_rate": 5.365652640578249e-06,
      "loss": 0.2394,
      "step": 6155
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3485982997230657,
      "learning_rate": 5.364360595058692e-06,
      "loss": 0.2027,
      "step": 6156
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.370328321348115,
      "learning_rate": 5.363068525078504e-06,
      "loss": 0.2354,
      "step": 6157
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3582041242680245,
      "learning_rate": 5.361776430724426e-06,
      "loss": 0.1951,
      "step": 6158
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.751711131399174,
      "learning_rate": 5.3604843120832e-06,
      "loss": 0.6398,
      "step": 6159
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3910888823882277,
      "learning_rate": 5.3591921692415706e-06,
      "loss": 0.2199,
      "step": 6160
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3803874211146097,
      "learning_rate": 5.357900002286282e-06,
      "loss": 0.2063,
      "step": 6161
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.281008238473857,
      "learning_rate": 5.35660781130408e-06,
      "loss": 0.181,
      "step": 6162
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3549325706681161,
      "learning_rate": 5.3553155963817175e-06,
      "loss": 0.1646,
      "step": 6163
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4893205686623445,
      "learning_rate": 5.354023357605942e-06,
      "loss": 0.1891,
      "step": 6164
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.315072413842959,
      "learning_rate": 5.352731095063506e-06,
      "loss": 0.2207,
      "step": 6165
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1922754377449734,
      "learning_rate": 5.351438808841162e-06,
      "loss": 0.1673,
      "step": 6166
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.35114937941398,
      "learning_rate": 5.350146499025666e-06,
      "loss": 0.5019,
      "step": 6167
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.361136241527948,
      "learning_rate": 5.348854165703776e-06,
      "loss": 0.2094,
      "step": 6168
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.174433487579136,
      "learning_rate": 5.347561808962247e-06,
      "loss": 0.1781,
      "step": 6169
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.292939413961952,
      "learning_rate": 5.346269428887843e-06,
      "loss": 0.6418,
      "step": 6170
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2857080483477152,
      "learning_rate": 5.34497702556732e-06,
      "loss": 0.1833,
      "step": 6171
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5339134688011482,
      "learning_rate": 5.343684599087448e-06,
      "loss": 0.1782,
      "step": 6172
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2416299723464432,
      "learning_rate": 5.342392149534983e-06,
      "loss": 0.1996,
      "step": 6173
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1471377388798076,
      "learning_rate": 5.341099676996698e-06,
      "loss": 0.1638,
      "step": 6174
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.165250223323812,
      "learning_rate": 5.339807181559359e-06,
      "loss": 0.1578,
      "step": 6175
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3456868721814605,
      "learning_rate": 5.338514663309733e-06,
      "loss": 0.1992,
      "step": 6176
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5029129229630251,
      "learning_rate": 5.337222122334593e-06,
      "loss": 0.1897,
      "step": 6177
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3350535291981411,
      "learning_rate": 5.33592955872071e-06,
      "loss": 0.2192,
      "step": 6178
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4363070923361838,
      "learning_rate": 5.334636972554858e-06,
      "loss": 0.2279,
      "step": 6179
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2691040738835486,
      "learning_rate": 5.33334436392381e-06,
      "loss": 0.1832,
      "step": 6180
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5669046984464379,
      "learning_rate": 5.3320517329143475e-06,
      "loss": 0.2177,
      "step": 6181
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2540121364576495,
      "learning_rate": 5.330759079613244e-06,
      "loss": 0.1639,
      "step": 6182
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3067742642281885,
      "learning_rate": 5.329466404107282e-06,
      "loss": 0.2078,
      "step": 6183
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.367268499801735,
      "learning_rate": 5.328173706483243e-06,
      "loss": 0.2286,
      "step": 6184
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.147007254075189,
      "learning_rate": 5.3268809868279095e-06,
      "loss": 0.1761,
      "step": 6185
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.416699859637547,
      "learning_rate": 5.325588245228063e-06,
      "loss": 0.2066,
      "step": 6186
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.335338703606333,
      "learning_rate": 5.324295481770492e-06,
      "loss": 0.2273,
      "step": 6187
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3001132257848957,
      "learning_rate": 5.323002696541983e-06,
      "loss": 0.174,
      "step": 6188
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3527813319619386,
      "learning_rate": 5.321709889629323e-06,
      "loss": 0.2322,
      "step": 6189
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.458390890121769,
      "learning_rate": 5.320417061119303e-06,
      "loss": 0.5533,
      "step": 6190
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2107505972568007,
      "learning_rate": 5.319124211098716e-06,
      "loss": 0.1678,
      "step": 6191
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.25474148136366,
      "learning_rate": 5.3178313396543546e-06,
      "loss": 0.7318,
      "step": 6192
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.218351143906593,
      "learning_rate": 5.316538446873012e-06,
      "loss": 0.6997,
      "step": 6193
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.356771075062425,
      "learning_rate": 5.315245532841483e-06,
      "loss": 0.1885,
      "step": 6194
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.556763924259674,
      "learning_rate": 5.3139525976465675e-06,
      "loss": 0.2088,
      "step": 6195
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.57119935181716,
      "learning_rate": 5.312659641375063e-06,
      "loss": 0.6587,
      "step": 6196
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4878667827648613,
      "learning_rate": 5.311366664113768e-06,
      "loss": 0.2061,
      "step": 6197
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4454518241148875,
      "learning_rate": 5.3100736659494865e-06,
      "loss": 0.188,
      "step": 6198
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4128697764717462,
      "learning_rate": 5.30878064696902e-06,
      "loss": 0.2365,
      "step": 6199
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.464325736623637,
      "learning_rate": 5.307487607259175e-06,
      "loss": 0.1907,
      "step": 6200
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2769964390926019,
      "learning_rate": 5.306194546906752e-06,
      "loss": 0.1686,
      "step": 6201
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2938500521694163,
      "learning_rate": 5.304901465998563e-06,
      "loss": 0.181,
      "step": 6202
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2944682676645638,
      "learning_rate": 5.303608364621416e-06,
      "loss": 0.1791,
      "step": 6203
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3152122083320263,
      "learning_rate": 5.302315242862119e-06,
      "loss": 0.1948,
      "step": 6204
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.326311443135585,
      "learning_rate": 5.301022100807482e-06,
      "loss": 0.1663,
      "step": 6205
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1644837629925227,
      "learning_rate": 5.29972893854432e-06,
      "loss": 0.1898,
      "step": 6206
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.353841133564367,
      "learning_rate": 5.298435756159447e-06,
      "loss": 0.1395,
      "step": 6207
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3412561345227323,
      "learning_rate": 5.297142553739677e-06,
      "loss": 0.2038,
      "step": 6208
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4056916873216339,
      "learning_rate": 5.295849331371828e-06,
      "loss": 0.2388,
      "step": 6209
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2172610738102185,
      "learning_rate": 5.294556089142716e-06,
      "loss": 0.1714,
      "step": 6210
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.30391090406548,
      "learning_rate": 5.293262827139161e-06,
      "loss": 0.5142,
      "step": 6211
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2855616429579746,
      "learning_rate": 5.291969545447985e-06,
      "loss": 0.1879,
      "step": 6212
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3520245848231778,
      "learning_rate": 5.290676244156009e-06,
      "loss": 0.174,
      "step": 6213
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.34374424559405,
      "learning_rate": 5.289382923350054e-06,
      "loss": 0.1864,
      "step": 6214
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4250860385520456,
      "learning_rate": 5.2880895831169476e-06,
      "loss": 0.1698,
      "step": 6215
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.8148650798024475,
      "learning_rate": 5.286796223543514e-06,
      "loss": 0.635,
      "step": 6216
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3711275170569162,
      "learning_rate": 5.2855028447165824e-06,
      "loss": 0.1885,
      "step": 6217
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.48435436364416,
      "learning_rate": 5.284209446722979e-06,
      "loss": 0.4296,
      "step": 6218
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3686758686712193,
      "learning_rate": 5.282916029649534e-06,
      "loss": 0.231,
      "step": 6219
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5031396702649311,
      "learning_rate": 5.28162259358308e-06,
      "loss": 0.203,
      "step": 6220
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.87619495820924,
      "learning_rate": 5.280329138610446e-06,
      "loss": 0.5631,
      "step": 6221
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3295513414410984,
      "learning_rate": 5.27903566481847e-06,
      "loss": 0.2018,
      "step": 6222
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2935896253719033,
      "learning_rate": 5.277742172293983e-06,
      "loss": 0.1523,
      "step": 6223
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.255733188177954,
      "learning_rate": 5.276448661123822e-06,
      "loss": 0.1514,
      "step": 6224
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4982246059469662,
      "learning_rate": 5.275155131394825e-06,
      "loss": 0.1922,
      "step": 6225
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.338161997524272,
      "learning_rate": 5.27386158319383e-06,
      "loss": 0.1732,
      "step": 6226
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.502402498985422,
      "learning_rate": 5.272568016607678e-06,
      "loss": 0.4369,
      "step": 6227
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.644401590868851,
      "learning_rate": 5.271274431723208e-06,
      "loss": 0.6347,
      "step": 6228
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4177339788861734,
      "learning_rate": 5.269980828627266e-06,
      "loss": 0.2073,
      "step": 6229
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1885238581679605,
      "learning_rate": 5.268687207406692e-06,
      "loss": 0.1907,
      "step": 6230
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3500017580095913,
      "learning_rate": 5.267393568148329e-06,
      "loss": 0.2119,
      "step": 6231
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4093337071058287,
      "learning_rate": 5.2660999109390285e-06,
      "loss": 0.1687,
      "step": 6232
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.154269169118788,
      "learning_rate": 5.264806235865634e-06,
      "loss": 0.1676,
      "step": 6233
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5550816011762203,
      "learning_rate": 5.263512543014995e-06,
      "loss": 0.2188,
      "step": 6234
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3124130687453872,
      "learning_rate": 5.262218832473959e-06,
      "loss": 0.1678,
      "step": 6235
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6378629819936357,
      "learning_rate": 5.260925104329379e-06,
      "loss": 0.2204,
      "step": 6236
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2443336396054396,
      "learning_rate": 5.259631358668107e-06,
      "loss": 0.183,
      "step": 6237
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4490440991160176,
      "learning_rate": 5.258337595576994e-06,
      "loss": 0.1593,
      "step": 6238
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2619955416133604,
      "learning_rate": 5.257043815142897e-06,
      "loss": 0.1771,
      "step": 6239
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.559735367267125,
      "learning_rate": 5.25575001745267e-06,
      "loss": 0.7524,
      "step": 6240
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5424281167939606,
      "learning_rate": 5.254456202593167e-06,
      "loss": 0.2393,
      "step": 6241
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4033943700063696,
      "learning_rate": 5.25316237065125e-06,
      "loss": 0.2091,
      "step": 6242
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3166563250082306,
      "learning_rate": 5.251868521713775e-06,
      "loss": 0.171,
      "step": 6243
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3677143418444035,
      "learning_rate": 5.250574655867604e-06,
      "loss": 0.2333,
      "step": 6244
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.424957758269985,
      "learning_rate": 5.249280773199597e-06,
      "loss": 0.2008,
      "step": 6245
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.548882291278945,
      "learning_rate": 5.247986873796616e-06,
      "loss": 0.6412,
      "step": 6246
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.254563942971284,
      "learning_rate": 5.246692957745526e-06,
      "loss": 0.1997,
      "step": 6247
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.31680943829475,
      "learning_rate": 5.245399025133189e-06,
      "loss": 0.1846,
      "step": 6248
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.430612759942501,
      "learning_rate": 5.244105076046472e-06,
      "loss": 0.2216,
      "step": 6249
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5505943413042584,
      "learning_rate": 5.242811110572243e-06,
      "loss": 0.2069,
      "step": 6250
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4182960011606025,
      "learning_rate": 5.241517128797366e-06,
      "loss": 0.182,
      "step": 6251
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.290315328036719,
      "learning_rate": 5.240223130808714e-06,
      "loss": 0.4789,
      "step": 6252
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3044802578652985,
      "learning_rate": 5.238929116693154e-06,
      "loss": 0.2195,
      "step": 6253
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2368560200831011,
      "learning_rate": 5.23763508653756e-06,
      "loss": 0.164,
      "step": 6254
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2172317642959076,
      "learning_rate": 5.236341040428803e-06,
      "loss": 0.2064,
      "step": 6255
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2327472522110754,
      "learning_rate": 5.235046978453755e-06,
      "loss": 0.1855,
      "step": 6256
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3262363626507576,
      "learning_rate": 5.233752900699291e-06,
      "loss": 0.1757,
      "step": 6257
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1879333437098638,
      "learning_rate": 5.2324588072522875e-06,
      "loss": 0.1665,
      "step": 6258
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3185553752647492,
      "learning_rate": 5.231164698199618e-06,
      "loss": 0.2106,
      "step": 6259
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3250489434983634,
      "learning_rate": 5.229870573628163e-06,
      "loss": 0.1951,
      "step": 6260
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3271923135117893,
      "learning_rate": 5.2285764336248e-06,
      "loss": 0.1911,
      "step": 6261
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3267738810883845,
      "learning_rate": 5.227282278276409e-06,
      "loss": 0.1724,
      "step": 6262
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.388770080187803,
      "learning_rate": 5.225988107669868e-06,
      "loss": 0.2124,
      "step": 6263
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4358798842707865,
      "learning_rate": 5.224693921892064e-06,
      "loss": 0.2383,
      "step": 6264
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3787768392862565,
      "learning_rate": 5.223399721029875e-06,
      "loss": 0.2015,
      "step": 6265
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.630419813457188,
      "learning_rate": 5.2221055051701845e-06,
      "loss": 0.4685,
      "step": 6266
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4566655880541437,
      "learning_rate": 5.220811274399878e-06,
      "loss": 0.1532,
      "step": 6267
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1617384552249408,
      "learning_rate": 5.2195170288058435e-06,
      "loss": 0.13,
      "step": 6268
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3629098046170762,
      "learning_rate": 5.218222768474964e-06,
      "loss": 0.2369,
      "step": 6269
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3486723099597446,
      "learning_rate": 5.21692849349413e-06,
      "loss": 0.2029,
      "step": 6270
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.445537848619636,
      "learning_rate": 5.21563420395023e-06,
      "loss": 0.2137,
      "step": 6271
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5945006796204164,
      "learning_rate": 5.214339899930151e-06,
      "loss": 0.2251,
      "step": 6272
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.712302116973072,
      "learning_rate": 5.2130455815207855e-06,
      "loss": 0.6219,
      "step": 6273
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3763313083280944,
      "learning_rate": 5.211751248809025e-06,
      "loss": 0.1615,
      "step": 6274
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.446830603750063,
      "learning_rate": 5.210456901881761e-06,
      "loss": 0.2396,
      "step": 6275
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4296338724637219,
      "learning_rate": 5.209162540825887e-06,
      "loss": 0.2016,
      "step": 6276
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.15501297483856,
      "learning_rate": 5.2078681657282994e-06,
      "loss": 0.5962,
      "step": 6277
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3644696065676662,
      "learning_rate": 5.206573776675891e-06,
      "loss": 0.1593,
      "step": 6278
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.359840783281693,
      "learning_rate": 5.20527937375556e-06,
      "loss": 0.2155,
      "step": 6279
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1531130367251454,
      "learning_rate": 5.2039849570542e-06,
      "loss": 0.1559,
      "step": 6280
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2974861367052535,
      "learning_rate": 5.202690526658715e-06,
      "loss": 0.195,
      "step": 6281
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3096591769738086,
      "learning_rate": 5.201396082656001e-06,
      "loss": 0.2015,
      "step": 6282
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3015592076096667,
      "learning_rate": 5.200101625132955e-06,
      "loss": 0.1818,
      "step": 6283
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4243204668795564,
      "learning_rate": 5.1988071541764814e-06,
      "loss": 0.2127,
      "step": 6284
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.212059500878854,
      "learning_rate": 5.197512669873482e-06,
      "loss": 0.1518,
      "step": 6285
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.430545895006179,
      "learning_rate": 5.196218172310858e-06,
      "loss": 0.736,
      "step": 6286
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2254113833640334,
      "learning_rate": 5.194923661575514e-06,
      "loss": 0.1911,
      "step": 6287
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2697928522573918,
      "learning_rate": 5.193629137754355e-06,
      "loss": 0.19,
      "step": 6288
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3851992415996301,
      "learning_rate": 5.192334600934284e-06,
      "loss": 0.1757,
      "step": 6289
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3363734321951881,
      "learning_rate": 5.1910400512022084e-06,
      "loss": 0.1633,
      "step": 6290
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.185884664214091,
      "learning_rate": 5.189745488645036e-06,
      "loss": 0.5321,
      "step": 6291
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2282431502504434,
      "learning_rate": 5.188450913349674e-06,
      "loss": 0.1837,
      "step": 6292
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4183703530101455,
      "learning_rate": 5.18715632540303e-06,
      "loss": 0.2132,
      "step": 6293
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3282381526127776,
      "learning_rate": 5.185861724892017e-06,
      "loss": 0.2209,
      "step": 6294
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3496458825344297,
      "learning_rate": 5.184567111903541e-06,
      "loss": 0.2595,
      "step": 6295
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4482981453686798,
      "learning_rate": 5.183272486524516e-06,
      "loss": 0.1942,
      "step": 6296
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3767640755892694,
      "learning_rate": 5.181977848841854e-06,
      "loss": 0.1838,
      "step": 6297
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.661692935557831,
      "learning_rate": 5.180683198942468e-06,
      "loss": 0.478,
      "step": 6298
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2692566319290866,
      "learning_rate": 5.1793885369132724e-06,
      "loss": 0.2078,
      "step": 6299
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3128242909347663,
      "learning_rate": 5.1780938628411795e-06,
      "loss": 0.1804,
      "step": 6300
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.380746640626029,
      "learning_rate": 5.1767991768131064e-06,
      "loss": 0.193,
      "step": 6301
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3376665180678404,
      "learning_rate": 5.175504478915968e-06,
      "loss": 0.2053,
      "step": 6302
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5282774562113082,
      "learning_rate": 5.1742097692366836e-06,
      "loss": 0.2521,
      "step": 6303
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4647091623871322,
      "learning_rate": 5.172915047862168e-06,
      "loss": 0.2197,
      "step": 6304
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3730826972308972,
      "learning_rate": 5.171620314879342e-06,
      "loss": 0.1796,
      "step": 6305
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4539461470657398,
      "learning_rate": 5.170325570375124e-06,
      "loss": 0.2101,
      "step": 6306
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.4313619719408885,
      "learning_rate": 5.169030814436436e-06,
      "loss": 0.5814,
      "step": 6307
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2695290912945862,
      "learning_rate": 5.1677360471501955e-06,
      "loss": 0.1737,
      "step": 6308
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.932141893561712,
      "learning_rate": 5.1664412686033276e-06,
      "loss": 0.6466,
      "step": 6309
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2910082903113071,
      "learning_rate": 5.165146478882751e-06,
      "loss": 0.1613,
      "step": 6310
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3272140675626496,
      "learning_rate": 5.163851678075391e-06,
      "loss": 0.1853,
      "step": 6311
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.622696097831712,
      "learning_rate": 5.1625568662681735e-06,
      "loss": 0.7612,
      "step": 6312
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4497650320558,
      "learning_rate": 5.16126204354802e-06,
      "loss": 0.1977,
      "step": 6313
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3717893413295899,
      "learning_rate": 5.159967210001855e-06,
      "loss": 0.1919,
      "step": 6314
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2203759944908106,
      "learning_rate": 5.158672365716608e-06,
      "loss": 0.1722,
      "step": 6315
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4638437345441884,
      "learning_rate": 5.157377510779207e-06,
      "loss": 0.2059,
      "step": 6316
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3953522484032583,
      "learning_rate": 5.156082645276573e-06,
      "loss": 0.2096,
      "step": 6317
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.35800286829192,
      "learning_rate": 5.154787769295639e-06,
      "loss": 0.1861,
      "step": 6318
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2655597434076797,
      "learning_rate": 5.153492882923335e-06,
      "loss": 0.1898,
      "step": 6319
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2904288729258366,
      "learning_rate": 5.152197986246586e-06,
      "loss": 0.1889,
      "step": 6320
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3636888587163787,
      "learning_rate": 5.150903079352326e-06,
      "loss": 0.2248,
      "step": 6321
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3673063628543014,
      "learning_rate": 5.149608162327486e-06,
      "loss": 0.2037,
      "step": 6322
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3141754862891193,
      "learning_rate": 5.148313235258996e-06,
      "loss": 0.1782,
      "step": 6323
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2190904449100082,
      "learning_rate": 5.1470182982337905e-06,
      "loss": 0.1759,
      "step": 6324
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3029187254268115,
      "learning_rate": 5.145723351338799e-06,
      "loss": 0.23,
      "step": 6325
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.392547922698511,
      "learning_rate": 5.144428394660958e-06,
      "loss": 0.1616,
      "step": 6326
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.26866731840062,
      "learning_rate": 5.143133428287202e-06,
      "loss": 0.1501,
      "step": 6327
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2483544462435243,
      "learning_rate": 5.141838452304465e-06,
      "loss": 0.1791,
      "step": 6328
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.383080049451087,
      "learning_rate": 5.140543466799683e-06,
      "loss": 0.1979,
      "step": 6329
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3636725407881545,
      "learning_rate": 5.139248471859793e-06,
      "loss": 0.2232,
      "step": 6330
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.388481945447857,
      "learning_rate": 5.137953467571729e-06,
      "loss": 0.6247,
      "step": 6331
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1772141983436382,
      "learning_rate": 5.136658454022432e-06,
      "loss": 0.148,
      "step": 6332
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3430838162519119,
      "learning_rate": 5.13536343129884e-06,
      "loss": 0.182,
      "step": 6333
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.534180365224956,
      "learning_rate": 5.134068399487889e-06,
      "loss": 0.583,
      "step": 6334
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4833657678020846,
      "learning_rate": 5.1327733586765205e-06,
      "loss": 0.1895,
      "step": 6335
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3442009876980543,
      "learning_rate": 5.131478308951675e-06,
      "loss": 0.1879,
      "step": 6336
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.225065962486156,
      "learning_rate": 5.130183250400292e-06,
      "loss": 0.1879,
      "step": 6337
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4489579094893268,
      "learning_rate": 5.1288881831093115e-06,
      "loss": 0.1943,
      "step": 6338
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2248707013790785,
      "learning_rate": 5.127593107165678e-06,
      "loss": 0.1734,
      "step": 6339
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4666455425776792,
      "learning_rate": 5.126298022656333e-06,
      "loss": 0.2064,
      "step": 6340
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.407926718043965,
      "learning_rate": 5.125002929668219e-06,
      "loss": 0.1918,
      "step": 6341
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3436786397388014,
      "learning_rate": 5.123707828288279e-06,
      "loss": 0.2204,
      "step": 6342
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3532685411349283,
      "learning_rate": 5.122412718603458e-06,
      "loss": 0.2055,
      "step": 6343
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4988625299318015,
      "learning_rate": 5.1211176007007e-06,
      "loss": 0.1987,
      "step": 6344
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2780826241443235,
      "learning_rate": 5.11982247466695e-06,
      "loss": 0.1485,
      "step": 6345
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3495948559642899,
      "learning_rate": 5.118527340589155e-06,
      "loss": 0.185,
      "step": 6346
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.4506138426105055,
      "learning_rate": 5.117232198554262e-06,
      "loss": 0.5483,
      "step": 6347
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.009250781544445,
      "learning_rate": 5.1159370486492145e-06,
      "loss": 0.6166,
      "step": 6348
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2568757837198627,
      "learning_rate": 5.114641890960961e-06,
      "loss": 0.1607,
      "step": 6349
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.432461779300337,
      "learning_rate": 5.11334672557645e-06,
      "loss": 0.2268,
      "step": 6350
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2308839028714296,
      "learning_rate": 5.1120515525826335e-06,
      "loss": 0.1856,
      "step": 6351
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3534130155833344,
      "learning_rate": 5.110756372066452e-06,
      "loss": 0.1636,
      "step": 6352
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4008718746468871,
      "learning_rate": 5.1094611841148625e-06,
      "loss": 0.2087,
      "step": 6353
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3810654127721416,
      "learning_rate": 5.108165988814811e-06,
      "loss": 0.1883,
      "step": 6354
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4254427437994086,
      "learning_rate": 5.106870786253248e-06,
      "loss": 0.2242,
      "step": 6355
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3495622555173048,
      "learning_rate": 5.1055755765171265e-06,
      "loss": 0.1735,
      "step": 6356
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2798924917983223,
      "learning_rate": 5.104280359693397e-06,
      "loss": 0.1706,
      "step": 6357
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.491052737624853,
      "learning_rate": 5.102985135869011e-06,
      "loss": 0.245,
      "step": 6358
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1461273940637602,
      "learning_rate": 5.10168990513092e-06,
      "loss": 0.1486,
      "step": 6359
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.859547458948115,
      "learning_rate": 5.100394667566079e-06,
      "loss": 0.5333,
      "step": 6360
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.232011364879343,
      "learning_rate": 5.0990994232614386e-06,
      "loss": 0.5242,
      "step": 6361
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3055087151280933,
      "learning_rate": 5.0978041723039536e-06,
      "loss": 0.2017,
      "step": 6362
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4619621926700757,
      "learning_rate": 5.096508914780579e-06,
      "loss": 0.2103,
      "step": 6363
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.30166151703843,
      "learning_rate": 5.0952136507782675e-06,
      "loss": 0.1621,
      "step": 6364
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.434144424870652,
      "learning_rate": 5.093918380383977e-06,
      "loss": 0.6433,
      "step": 6365
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3664909365480735,
      "learning_rate": 5.092623103684657e-06,
      "loss": 0.1856,
      "step": 6366
    },
    {
      "epoch": 0.51,
      "grad_norm": 22.373376418414317,
      "learning_rate": 5.091327820767271e-06,
      "loss": 0.5479,
      "step": 6367
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3433504680403467,
      "learning_rate": 5.090032531718773e-06,
      "loss": 0.2207,
      "step": 6368
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2947627924842384,
      "learning_rate": 5.088737236626114e-06,
      "loss": 0.1707,
      "step": 6369
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.308123110750913,
      "learning_rate": 5.0874419355762585e-06,
      "loss": 0.1982,
      "step": 6370
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1800048212229322,
      "learning_rate": 5.08614662865616e-06,
      "loss": 0.1784,
      "step": 6371
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1064509431006522,
      "learning_rate": 5.084851315952778e-06,
      "loss": 0.1515,
      "step": 6372
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1949298119046323,
      "learning_rate": 5.083555997553068e-06,
      "loss": 0.1432,
      "step": 6373
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4413531240848336,
      "learning_rate": 5.082260673543992e-06,
      "loss": 0.1986,
      "step": 6374
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2200435922631385,
      "learning_rate": 5.080965344012509e-06,
      "loss": 0.2025,
      "step": 6375
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.222024129003145,
      "learning_rate": 5.079670009045574e-06,
      "loss": 0.1871,
      "step": 6376
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4566266278238087,
      "learning_rate": 5.078374668730154e-06,
      "loss": 0.191,
      "step": 6377
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2745843149390028,
      "learning_rate": 5.0770793231532025e-06,
      "loss": 0.184,
      "step": 6378
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3323997465696902,
      "learning_rate": 5.075783972401682e-06,
      "loss": 0.1848,
      "step": 6379
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5434798376949976,
      "learning_rate": 5.074488616562555e-06,
      "loss": 0.2132,
      "step": 6380
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1753449813561117,
      "learning_rate": 5.0731932557227814e-06,
      "loss": 0.1774,
      "step": 6381
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2471094037111388,
      "learning_rate": 5.0718978899693224e-06,
      "loss": 0.1726,
      "step": 6382
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4454714705841416,
      "learning_rate": 5.070602519389139e-06,
      "loss": 0.2298,
      "step": 6383
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2802330352406381,
      "learning_rate": 5.069307144069196e-06,
      "loss": 0.1659,
      "step": 6384
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1227732605890215,
      "learning_rate": 5.068011764096455e-06,
      "loss": 0.1498,
      "step": 6385
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3386868239234615,
      "learning_rate": 5.066716379557879e-06,
      "loss": 0.2309,
      "step": 6386
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3431133499369152,
      "learning_rate": 5.065420990540428e-06,
      "loss": 0.2133,
      "step": 6387
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3016129798515141,
      "learning_rate": 5.064125597131069e-06,
      "loss": 0.1949,
      "step": 6388
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.960514266339366,
      "learning_rate": 5.062830199416764e-06,
      "loss": 0.7845,
      "step": 6389
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3866127992472959,
      "learning_rate": 5.061534797484476e-06,
      "loss": 0.2475,
      "step": 6390
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.8681522122684235,
      "learning_rate": 5.0602393914211724e-06,
      "loss": 0.5361,
      "step": 6391
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2843313985431033,
      "learning_rate": 5.0589439813138165e-06,
      "loss": 0.1751,
      "step": 6392
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.0784505366617285,
      "learning_rate": 5.057648567249372e-06,
      "loss": 0.4926,
      "step": 6393
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.254534413096096,
      "learning_rate": 5.056353149314803e-06,
      "loss": 0.4868,
      "step": 6394
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5730514321996294,
      "learning_rate": 5.055057727597078e-06,
      "loss": 0.1815,
      "step": 6395
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.366804493328478,
      "learning_rate": 5.0537623021831595e-06,
      "loss": 0.1857,
      "step": 6396
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1974273885225233,
      "learning_rate": 5.0524668731600154e-06,
      "loss": 0.1434,
      "step": 6397
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.290462268700418,
      "learning_rate": 5.051171440614612e-06,
      "loss": 0.1682,
      "step": 6398
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3408080319511082,
      "learning_rate": 5.049876004633914e-06,
      "loss": 0.1726,
      "step": 6399
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3301174211156808,
      "learning_rate": 5.048580565304887e-06,
      "loss": 0.2001,
      "step": 6400
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1295720778513296,
      "learning_rate": 5.047285122714501e-06,
      "loss": 0.1356,
      "step": 6401
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.307628524028128,
      "learning_rate": 5.045989676949722e-06,
      "loss": 0.6551,
      "step": 6402
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.30717220964506,
      "learning_rate": 5.044694228097516e-06,
      "loss": 0.1868,
      "step": 6403
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.308617542381283,
      "learning_rate": 5.043398776244852e-06,
      "loss": 0.1999,
      "step": 6404
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3594059347674798,
      "learning_rate": 5.0421033214786965e-06,
      "loss": 0.168,
      "step": 6405
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2978934715585597,
      "learning_rate": 5.040807863886017e-06,
      "loss": 0.1998,
      "step": 6406
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2274527728906517,
      "learning_rate": 5.039512403553782e-06,
      "loss": 0.1575,
      "step": 6407
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3690895326544734,
      "learning_rate": 5.038216940568958e-06,
      "loss": 0.1922,
      "step": 6408
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2294123798712022,
      "learning_rate": 5.036921475018517e-06,
      "loss": 0.166,
      "step": 6409
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3792524669883612,
      "learning_rate": 5.035626006989426e-06,
      "loss": 0.2063,
      "step": 6410
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.382789409496291,
      "learning_rate": 5.0343305365686505e-06,
      "loss": 0.2011,
      "step": 6411
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3569064722631767,
      "learning_rate": 5.033035063843164e-06,
      "loss": 0.1964,
      "step": 6412
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2371414487308672,
      "learning_rate": 5.031739588899934e-06,
      "loss": 0.1674,
      "step": 6413
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.245679788522383,
      "learning_rate": 5.030444111825928e-06,
      "loss": 0.2054,
      "step": 6414
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3043462745913819,
      "learning_rate": 5.029148632708117e-06,
      "loss": 0.1809,
      "step": 6415
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6185587609100252,
      "learning_rate": 5.02785315163347e-06,
      "loss": 0.2793,
      "step": 6416
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3674335989262343,
      "learning_rate": 5.026557668688956e-06,
      "loss": 0.1736,
      "step": 6417
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3205410181330746,
      "learning_rate": 5.025262183961546e-06,
      "loss": 0.2165,
      "step": 6418
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2722612485187108,
      "learning_rate": 5.023966697538209e-06,
      "loss": 0.1688,
      "step": 6419
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.302922757379929,
      "learning_rate": 5.022671209505916e-06,
      "loss": 0.179,
      "step": 6420
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4126428658380186,
      "learning_rate": 5.021375719951634e-06,
      "loss": 0.2164,
      "step": 6421
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4134471536090285,
      "learning_rate": 5.020080228962337e-06,
      "loss": 0.1885,
      "step": 6422
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2672185260207796,
      "learning_rate": 5.018784736624993e-06,
      "loss": 0.1808,
      "step": 6423
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.372429659648729,
      "learning_rate": 5.017489243026571e-06,
      "loss": 0.2286,
      "step": 6424
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2675801486466107,
      "learning_rate": 5.016193748254045e-06,
      "loss": 0.1684,
      "step": 6425
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.531537854993101,
      "learning_rate": 5.014898252394382e-06,
      "loss": 0.5572,
      "step": 6426
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1972047922719196,
      "learning_rate": 5.013602755534555e-06,
      "loss": 0.1712,
      "step": 6427
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0860373471353395,
      "learning_rate": 5.012307257761534e-06,
      "loss": 0.1385,
      "step": 6428
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.24897348644714,
      "learning_rate": 5.01101175916229e-06,
      "loss": 0.1729,
      "step": 6429
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5392819274344394,
      "learning_rate": 5.009716259823792e-06,
      "loss": 0.191,
      "step": 6430
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3156815177679642,
      "learning_rate": 5.008420759833013e-06,
      "loss": 0.1654,
      "step": 6431
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4624911622140855,
      "learning_rate": 5.007125259276922e-06,
      "loss": 0.2059,
      "step": 6432
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3450597396063462,
      "learning_rate": 5.005829758242492e-06,
      "loss": 0.1888,
      "step": 6433
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4196980970344129,
      "learning_rate": 5.004534256816692e-06,
      "loss": 0.2108,
      "step": 6434
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3615912207164587,
      "learning_rate": 5.003238755086492e-06,
      "loss": 0.1791,
      "step": 6435
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2695951334060156,
      "learning_rate": 5.001943253138866e-06,
      "loss": 0.1936,
      "step": 6436
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4216257341166583,
      "learning_rate": 5.000647751060784e-06,
      "loss": 0.2276,
      "step": 6437
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.478867227948118,
      "learning_rate": 4.999352248939216e-06,
      "loss": 0.1894,
      "step": 6438
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.40206226767812,
      "learning_rate": 4.998056746861134e-06,
      "loss": 0.1821,
      "step": 6439
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.065017659028959,
      "learning_rate": 4.996761244913508e-06,
      "loss": 0.6221,
      "step": 6440
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.267014851979614,
      "learning_rate": 4.99546574318331e-06,
      "loss": 0.171,
      "step": 6441
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3887548877704219,
      "learning_rate": 4.994170241757511e-06,
      "loss": 0.205,
      "step": 6442
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4505973721594676,
      "learning_rate": 4.992874740723079e-06,
      "loss": 0.2222,
      "step": 6443
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2742638484813196,
      "learning_rate": 4.991579240166989e-06,
      "loss": 0.1612,
      "step": 6444
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5338614536974189,
      "learning_rate": 4.9902837401762085e-06,
      "loss": 0.2202,
      "step": 6445
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.307843015493231,
      "learning_rate": 4.988988240837713e-06,
      "loss": 0.2246,
      "step": 6446
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2438444874820898,
      "learning_rate": 4.987692742238467e-06,
      "loss": 0.1623,
      "step": 6447
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.786782519616948,
      "learning_rate": 4.9863972444654454e-06,
      "loss": 0.6145,
      "step": 6448
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3497394214361547,
      "learning_rate": 4.985101747605618e-06,
      "loss": 0.1605,
      "step": 6449
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2277053799918165,
      "learning_rate": 4.983806251745958e-06,
      "loss": 0.1741,
      "step": 6450
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4964492040567676,
      "learning_rate": 4.98251075697343e-06,
      "loss": 0.2321,
      "step": 6451
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3098906850210155,
      "learning_rate": 4.9812152633750084e-06,
      "loss": 0.2036,
      "step": 6452
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3346761739871242,
      "learning_rate": 4.979919771037666e-06,
      "loss": 0.1904,
      "step": 6453
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.354355557286669,
      "learning_rate": 4.978624280048367e-06,
      "loss": 0.205,
      "step": 6454
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1932310185295678,
      "learning_rate": 4.9773287904940856e-06,
      "loss": 0.1656,
      "step": 6455
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.256375970899375,
      "learning_rate": 4.976033302461791e-06,
      "loss": 0.5707,
      "step": 6456
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.954872842855911,
      "learning_rate": 4.974737816038454e-06,
      "loss": 0.6552,
      "step": 6457
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4617277256974235,
      "learning_rate": 4.973442331311043e-06,
      "loss": 0.1907,
      "step": 6458
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4977121736074688,
      "learning_rate": 4.9721468483665305e-06,
      "loss": 0.2165,
      "step": 6459
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.232628270215334,
      "learning_rate": 4.9708513672918854e-06,
      "loss": 0.179,
      "step": 6460
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1183375655846886,
      "learning_rate": 4.969555888174073e-06,
      "loss": 0.186,
      "step": 6461
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.387930690600105,
      "learning_rate": 4.968260411100068e-06,
      "loss": 0.176,
      "step": 6462
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.339045318017827,
      "learning_rate": 4.9669649361568365e-06,
      "loss": 0.2372,
      "step": 6463
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2680504752869848,
      "learning_rate": 4.96566946343135e-06,
      "loss": 0.1885,
      "step": 6464
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2817023328445856,
      "learning_rate": 4.964373993010576e-06,
      "loss": 0.1767,
      "step": 6465
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4428992505542968,
      "learning_rate": 4.963078524981485e-06,
      "loss": 0.1899,
      "step": 6466
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.236973634061518,
      "learning_rate": 4.961783059431042e-06,
      "loss": 0.1597,
      "step": 6467
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.292994240421266,
      "learning_rate": 4.960487596446221e-06,
      "loss": 0.1876,
      "step": 6468
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2664576054590746,
      "learning_rate": 4.959192136113985e-06,
      "loss": 0.1722,
      "step": 6469
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.781154130795651,
      "learning_rate": 4.957896678521305e-06,
      "loss": 0.736,
      "step": 6470
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2718884440393683,
      "learning_rate": 4.95660122375515e-06,
      "loss": 0.2094,
      "step": 6471
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2211291498781898,
      "learning_rate": 4.9553057719024845e-06,
      "loss": 0.1923,
      "step": 6472
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1897030103496622,
      "learning_rate": 4.954010323050279e-06,
      "loss": 0.1459,
      "step": 6473
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3364748722559496,
      "learning_rate": 4.952714877285499e-06,
      "loss": 0.1873,
      "step": 6474
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3843335320933292,
      "learning_rate": 4.951419434695115e-06,
      "loss": 0.2159,
      "step": 6475
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2825141205513542,
      "learning_rate": 4.950123995366088e-06,
      "loss": 0.1934,
      "step": 6476
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.768346422506801,
      "learning_rate": 4.9488285593853906e-06,
      "loss": 0.3353,
      "step": 6477
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.85330666587967,
      "learning_rate": 4.947533126839987e-06,
      "loss": 0.4567,
      "step": 6478
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.473994232305882,
      "learning_rate": 4.946237697816842e-06,
      "loss": 0.2068,
      "step": 6479
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3247575744265563,
      "learning_rate": 4.944942272402925e-06,
      "loss": 0.1594,
      "step": 6480
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2617809596681895,
      "learning_rate": 4.943646850685198e-06,
      "loss": 0.1896,
      "step": 6481
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0979430708527276,
      "learning_rate": 4.942351432750631e-06,
      "loss": 0.1659,
      "step": 6482
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3092274348911537,
      "learning_rate": 4.941056018686184e-06,
      "loss": 0.1451,
      "step": 6483
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3828935306361319,
      "learning_rate": 4.939760608578828e-06,
      "loss": 0.2128,
      "step": 6484
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.364895657879168,
      "learning_rate": 4.938465202515524e-06,
      "loss": 0.1507,
      "step": 6485
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1063316112840047,
      "learning_rate": 4.937169800583237e-06,
      "loss": 0.1439,
      "step": 6486
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.557868565247235,
      "learning_rate": 4.9358744028689336e-06,
      "loss": 0.6502,
      "step": 6487
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2198421968036965,
      "learning_rate": 4.934579009459573e-06,
      "loss": 0.1753,
      "step": 6488
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4636363883426746,
      "learning_rate": 4.933283620442123e-06,
      "loss": 0.2249,
      "step": 6489
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3388299236756833,
      "learning_rate": 4.931988235903545e-06,
      "loss": 0.2105,
      "step": 6490
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3802305389526213,
      "learning_rate": 4.930692855930805e-06,
      "loss": 0.2114,
      "step": 6491
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.320776490708098,
      "learning_rate": 4.929397480610861e-06,
      "loss": 0.1963,
      "step": 6492
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1974184321539358,
      "learning_rate": 4.928102110030678e-06,
      "loss": 0.1808,
      "step": 6493
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1702978295024231,
      "learning_rate": 4.9268067442772185e-06,
      "loss": 0.1377,
      "step": 6494
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.675271783210693,
      "learning_rate": 4.925511383437446e-06,
      "loss": 0.7834,
      "step": 6495
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.694876623033931,
      "learning_rate": 4.9242160275983195e-06,
      "loss": 0.729,
      "step": 6496
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4140300028028574,
      "learning_rate": 4.922920676846798e-06,
      "loss": 0.2222,
      "step": 6497
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1759189966700123,
      "learning_rate": 4.9216253312698496e-06,
      "loss": 0.2047,
      "step": 6498
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.308159060446919,
      "learning_rate": 4.9203299909544264e-06,
      "loss": 0.2135,
      "step": 6499
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3963517342358278,
      "learning_rate": 4.919034655987493e-06,
      "loss": 0.1931,
      "step": 6500
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3231052685470979,
      "learning_rate": 4.917739326456008e-06,
      "loss": 0.1968,
      "step": 6501
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3380638222983487,
      "learning_rate": 4.916444002446934e-06,
      "loss": 0.2155,
      "step": 6502
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3618715141522857,
      "learning_rate": 4.915148684047225e-06,
      "loss": 0.2015,
      "step": 6503
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2871346959458516,
      "learning_rate": 4.913853371343841e-06,
      "loss": 0.1516,
      "step": 6504
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.315687767609004,
      "learning_rate": 4.912558064423744e-06,
      "loss": 0.2039,
      "step": 6505
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5768749703894354,
      "learning_rate": 4.911262763373887e-06,
      "loss": 0.218,
      "step": 6506
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2817894268522665,
      "learning_rate": 4.909967468281231e-06,
      "loss": 0.1461,
      "step": 6507
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1916827267320635,
      "learning_rate": 4.908672179232729e-06,
      "loss": 0.1895,
      "step": 6508
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3134336602352426,
      "learning_rate": 4.9073768963153434e-06,
      "loss": 0.1973,
      "step": 6509
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2125077345570685,
      "learning_rate": 4.906081619616026e-06,
      "loss": 0.1838,
      "step": 6510
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5232579603192709,
      "learning_rate": 4.904786349221733e-06,
      "loss": 0.2077,
      "step": 6511
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.200847275730482,
      "learning_rate": 4.903491085219424e-06,
      "loss": 0.7319,
      "step": 6512
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.68495153331719,
      "learning_rate": 4.902195827696048e-06,
      "loss": 0.4831,
      "step": 6513
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.704743142056712,
      "learning_rate": 4.900900576738563e-06,
      "loss": 0.2297,
      "step": 6514
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3430704691585156,
      "learning_rate": 4.899605332433922e-06,
      "loss": 0.1869,
      "step": 6515
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4240321833279908,
      "learning_rate": 4.898310094869082e-06,
      "loss": 0.2126,
      "step": 6516
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2345239307897238,
      "learning_rate": 4.89701486413099e-06,
      "loss": 0.1512,
      "step": 6517
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1919226427385405,
      "learning_rate": 4.8957196403066036e-06,
      "loss": 0.1506,
      "step": 6518
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.619942926220242,
      "learning_rate": 4.8944244234828734e-06,
      "loss": 0.6776,
      "step": 6519
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.9474411120294155,
      "learning_rate": 4.8931292137467525e-06,
      "loss": 0.62,
      "step": 6520
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2935708106312878,
      "learning_rate": 4.891834011185191e-06,
      "loss": 0.204,
      "step": 6521
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.281559968102252,
      "learning_rate": 4.890538815885139e-06,
      "loss": 0.1876,
      "step": 6522
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.32306481120636,
      "learning_rate": 4.88924362793355e-06,
      "loss": 0.1685,
      "step": 6523
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3601948506417618,
      "learning_rate": 4.887948447417369e-06,
      "loss": 0.1971,
      "step": 6524
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.970631538545379,
      "learning_rate": 4.886653274423551e-06,
      "loss": 0.2331,
      "step": 6525
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.286643234670694,
      "learning_rate": 4.88535810903904e-06,
      "loss": 0.1372,
      "step": 6526
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4579908451993773,
      "learning_rate": 4.884062951350787e-06,
      "loss": 0.2111,
      "step": 6527
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.48490233924593,
      "learning_rate": 4.882767801445739e-06,
      "loss": 0.2039,
      "step": 6528
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3382582884559513,
      "learning_rate": 4.8814726594108455e-06,
      "loss": 0.2321,
      "step": 6529
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2096611266947055,
      "learning_rate": 4.880177525333051e-06,
      "loss": 0.148,
      "step": 6530
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.330107188156349,
      "learning_rate": 4.878882399299301e-06,
      "loss": 0.2161,
      "step": 6531
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.129105008008197,
      "learning_rate": 4.877587281396544e-06,
      "loss": 0.1628,
      "step": 6532
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.240770920588671,
      "learning_rate": 4.876292171711723e-06,
      "loss": 0.2072,
      "step": 6533
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2656581851091837,
      "learning_rate": 4.8749970703317825e-06,
      "loss": 0.1601,
      "step": 6534
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5426382569629271,
      "learning_rate": 4.873701977343667e-06,
      "loss": 0.2213,
      "step": 6535
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.319723210346589,
      "learning_rate": 4.872406892834323e-06,
      "loss": 0.6139,
      "step": 6536
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2412588837549297,
      "learning_rate": 4.8711118168906885e-06,
      "loss": 0.1932,
      "step": 6537
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1903991722047078,
      "learning_rate": 4.8698167495997095e-06,
      "loss": 0.1565,
      "step": 6538
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.344746043596786,
      "learning_rate": 4.8685216910483275e-06,
      "loss": 0.213,
      "step": 6539
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2403383165534796,
      "learning_rate": 4.867226641323481e-06,
      "loss": 0.1966,
      "step": 6540
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.3370077629390495,
      "learning_rate": 4.8659316005121125e-06,
      "loss": 0.5838,
      "step": 6541
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5358547601022514,
      "learning_rate": 4.864636568701162e-06,
      "loss": 0.2513,
      "step": 6542
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4304870506413077,
      "learning_rate": 4.86334154597757e-06,
      "loss": 0.1888,
      "step": 6543
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.40764727334343,
      "learning_rate": 4.862046532428272e-06,
      "loss": 0.216,
      "step": 6544
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.352628833835352,
      "learning_rate": 4.860751528140209e-06,
      "loss": 0.2036,
      "step": 6545
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1290700925638177,
      "learning_rate": 4.859456533200318e-06,
      "loss": 0.1305,
      "step": 6546
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.164014569001093,
      "learning_rate": 4.858161547695537e-06,
      "loss": 0.1523,
      "step": 6547
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.299323689716936,
      "learning_rate": 4.8568665717128e-06,
      "loss": 0.1931,
      "step": 6548
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3619883406877493,
      "learning_rate": 4.855571605339043e-06,
      "loss": 0.1787,
      "step": 6549
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4368002238089048,
      "learning_rate": 4.8542766486612035e-06,
      "loss": 0.2307,
      "step": 6550
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4073693146032153,
      "learning_rate": 4.852981701766212e-06,
      "loss": 0.1963,
      "step": 6551
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.413877235349347,
      "learning_rate": 4.851686764741005e-06,
      "loss": 0.2066,
      "step": 6552
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.243283087319646,
      "learning_rate": 4.850391837672515e-06,
      "loss": 0.2017,
      "step": 6553
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4244229440760505,
      "learning_rate": 4.849096920647675e-06,
      "loss": 0.2555,
      "step": 6554
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2860802610125572,
      "learning_rate": 4.847802013753415e-06,
      "loss": 0.1962,
      "step": 6555
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.36055728292811,
      "learning_rate": 4.846507117076667e-06,
      "loss": 0.1921,
      "step": 6556
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.326818838117966,
      "learning_rate": 4.845212230704363e-06,
      "loss": 0.2104,
      "step": 6557
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4904421650460549,
      "learning_rate": 4.843917354723428e-06,
      "loss": 0.1812,
      "step": 6558
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3284220479526134,
      "learning_rate": 4.842622489220796e-06,
      "loss": 0.2138,
      "step": 6559
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4775484909757695,
      "learning_rate": 4.841327634283392e-06,
      "loss": 0.212,
      "step": 6560
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3869806558083997,
      "learning_rate": 4.840032789998147e-06,
      "loss": 0.2169,
      "step": 6561
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4400567605583703,
      "learning_rate": 4.838737956451982e-06,
      "loss": 0.1811,
      "step": 6562
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.398036744317056,
      "learning_rate": 4.837443133731828e-06,
      "loss": 0.1748,
      "step": 6563
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3478880238171476,
      "learning_rate": 4.836148321924611e-06,
      "loss": 0.1825,
      "step": 6564
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4267354229943439,
      "learning_rate": 4.834853521117251e-06,
      "loss": 0.1925,
      "step": 6565
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3057799642802572,
      "learning_rate": 4.833558731396675e-06,
      "loss": 0.1918,
      "step": 6566
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3220262574264439,
      "learning_rate": 4.832263952849805e-06,
      "loss": 0.2005,
      "step": 6567
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2677563871001374,
      "learning_rate": 4.830969185563566e-06,
      "loss": 0.1809,
      "step": 6568
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5501322312444912,
      "learning_rate": 4.829674429624876e-06,
      "loss": 0.2258,
      "step": 6569
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.32090832886925,
      "learning_rate": 4.828379685120659e-06,
      "loss": 0.1399,
      "step": 6570
    },
    {
      "epoch": 0.53,
      "grad_norm": 16.768903833901742,
      "learning_rate": 4.827084952137833e-06,
      "loss": 0.4583,
      "step": 6571
    },
    {
      "epoch": 0.53,
      "grad_norm": 12.574780789212554,
      "learning_rate": 4.825790230763318e-06,
      "loss": 0.558,
      "step": 6572
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.361489051072294,
      "learning_rate": 4.824495521084034e-06,
      "loss": 0.1653,
      "step": 6573
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4523552664448192,
      "learning_rate": 4.823200823186895e-06,
      "loss": 0.2397,
      "step": 6574
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3485866956157013,
      "learning_rate": 4.821906137158822e-06,
      "loss": 0.1804,
      "step": 6575
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4429614801960882,
      "learning_rate": 4.820611463086728e-06,
      "loss": 0.148,
      "step": 6576
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.332727271255758,
      "learning_rate": 4.819316801057533e-06,
      "loss": 0.2024,
      "step": 6577
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.350745128049199,
      "learning_rate": 4.818022151158146e-06,
      "loss": 0.5582,
      "step": 6578
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1734769646762175,
      "learning_rate": 4.816727513475485e-06,
      "loss": 0.1613,
      "step": 6579
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4472294667921561,
      "learning_rate": 4.815432888096459e-06,
      "loss": 0.2135,
      "step": 6580
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3287841250929464,
      "learning_rate": 4.814138275107986e-06,
      "loss": 0.1529,
      "step": 6581
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2545308503158996,
      "learning_rate": 4.812843674596972e-06,
      "loss": 0.2039,
      "step": 6582
    },
    {
      "epoch": 0.53,
      "grad_norm": 12.660946926334574,
      "learning_rate": 4.811549086650327e-06,
      "loss": 0.7344,
      "step": 6583
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.298850196854735,
      "learning_rate": 4.810254511354966e-06,
      "loss": 0.2003,
      "step": 6584
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1489309815756354,
      "learning_rate": 4.808959948797793e-06,
      "loss": 0.1579,
      "step": 6585
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2166805966225172,
      "learning_rate": 4.807665399065718e-06,
      "loss": 0.206,
      "step": 6586
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.195503116143096,
      "learning_rate": 4.806370862245646e-06,
      "loss": 0.1833,
      "step": 6587
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4104544362476874,
      "learning_rate": 4.8050763384244875e-06,
      "loss": 0.1837,
      "step": 6588
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2951669589209311,
      "learning_rate": 4.8037818276891425e-06,
      "loss": 0.2196,
      "step": 6589
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.448977612935867,
      "learning_rate": 4.802487330126519e-06,
      "loss": 0.2073,
      "step": 6590
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.291150336912249,
      "learning_rate": 4.801192845823521e-06,
      "loss": 0.188,
      "step": 6591
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5970412398479783,
      "learning_rate": 4.799898374867047e-06,
      "loss": 0.2277,
      "step": 6592
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5365158972020774,
      "learning_rate": 4.798603917344003e-06,
      "loss": 0.2269,
      "step": 6593
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4209885368414394,
      "learning_rate": 4.797309473341286e-06,
      "loss": 0.2294,
      "step": 6594
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4341493811388837,
      "learning_rate": 4.796015042945801e-06,
      "loss": 0.1677,
      "step": 6595
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1379282760771579,
      "learning_rate": 4.794720626244442e-06,
      "loss": 0.1542,
      "step": 6596
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.448270841428079,
      "learning_rate": 4.79342622332411e-06,
      "loss": 0.1933,
      "step": 6597
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6272632472609554,
      "learning_rate": 4.792131834271701e-06,
      "loss": 0.2313,
      "step": 6598
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2302965393176406,
      "learning_rate": 4.790837459174114e-06,
      "loss": 0.1841,
      "step": 6599
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1708301466977171,
      "learning_rate": 4.7895430981182415e-06,
      "loss": 0.1622,
      "step": 6600
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3263735959669816,
      "learning_rate": 4.788248751190977e-06,
      "loss": 0.1992,
      "step": 6601
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.493462081051432,
      "learning_rate": 4.786954418479217e-06,
      "loss": 0.1796,
      "step": 6602
    },
    {
      "epoch": 0.53,
      "grad_norm": 19.994128076040145,
      "learning_rate": 4.785660100069851e-06,
      "loss": 0.5566,
      "step": 6603
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.42106058863406,
      "learning_rate": 4.784365796049772e-06,
      "loss": 0.2132,
      "step": 6604
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1495963360561818,
      "learning_rate": 4.78307150650587e-06,
      "loss": 0.1552,
      "step": 6605
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4525780583231667,
      "learning_rate": 4.781777231525037e-06,
      "loss": 0.2092,
      "step": 6606
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4602496248873231,
      "learning_rate": 4.780482971194157e-06,
      "loss": 0.2046,
      "step": 6607
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2674157587172645,
      "learning_rate": 4.779188725600123e-06,
      "loss": 0.1717,
      "step": 6608
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.280533476688851,
      "learning_rate": 4.777894494829819e-06,
      "loss": 0.2241,
      "step": 6609
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.306963050104407,
      "learning_rate": 4.776600278970127e-06,
      "loss": 0.1839,
      "step": 6610
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1984225559989583,
      "learning_rate": 4.775306078107938e-06,
      "loss": 0.1741,
      "step": 6611
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5708246597589846,
      "learning_rate": 4.774011892330132e-06,
      "loss": 0.1984,
      "step": 6612
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.671244243284006,
      "learning_rate": 4.772717721723593e-06,
      "loss": 0.4876,
      "step": 6613
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3646108395698546,
      "learning_rate": 4.7714235663752e-06,
      "loss": 0.2456,
      "step": 6614
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3201079504129871,
      "learning_rate": 4.770129426371838e-06,
      "loss": 0.1929,
      "step": 6615
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4239607468807445,
      "learning_rate": 4.768835301800383e-06,
      "loss": 0.1901,
      "step": 6616
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3873728676656973,
      "learning_rate": 4.767541192747714e-06,
      "loss": 0.1796,
      "step": 6617
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3079091238882445,
      "learning_rate": 4.766247099300711e-06,
      "loss": 0.1902,
      "step": 6618
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4915154985268286,
      "learning_rate": 4.764953021546247e-06,
      "loss": 0.244,
      "step": 6619
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2341851321558228,
      "learning_rate": 4.763658959571199e-06,
      "loss": 0.146,
      "step": 6620
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2547974964996214,
      "learning_rate": 4.76236491346244e-06,
      "loss": 0.172,
      "step": 6621
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4681611132235795,
      "learning_rate": 4.7610708833068466e-06,
      "loss": 0.1947,
      "step": 6622
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2618533945433152,
      "learning_rate": 4.759776869191287e-06,
      "loss": 0.1914,
      "step": 6623
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3502634055977463,
      "learning_rate": 4.7584828712026345e-06,
      "loss": 0.1879,
      "step": 6624
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2971771694383252,
      "learning_rate": 4.757188889427761e-06,
      "loss": 0.1754,
      "step": 6625
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.567244467715232,
      "learning_rate": 4.7558949239535295e-06,
      "loss": 0.6236,
      "step": 6626
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2869608603762317,
      "learning_rate": 4.754600974866813e-06,
      "loss": 0.1906,
      "step": 6627
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2692170245999832,
      "learning_rate": 4.753307042254476e-06,
      "loss": 0.2024,
      "step": 6628
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3213774859705014,
      "learning_rate": 4.752013126203386e-06,
      "loss": 0.1849,
      "step": 6629
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.115208416368678,
      "learning_rate": 4.750719226800404e-06,
      "loss": 0.1663,
      "step": 6630
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.392426964838036,
      "learning_rate": 4.749425344132397e-06,
      "loss": 0.2175,
      "step": 6631
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3450686686800324,
      "learning_rate": 4.748131478286225e-06,
      "loss": 0.1782,
      "step": 6632
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.356508495742206,
      "learning_rate": 4.7468376293487515e-06,
      "loss": 0.216,
      "step": 6633
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4635077573282622,
      "learning_rate": 4.745543797406835e-06,
      "loss": 0.1606,
      "step": 6634
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3847681783984467,
      "learning_rate": 4.744249982547332e-06,
      "loss": 0.1944,
      "step": 6635
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5316911420192632,
      "learning_rate": 4.7429561848571055e-06,
      "loss": 0.195,
      "step": 6636
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.020441401890124,
      "learning_rate": 4.741662404423007e-06,
      "loss": 0.5007,
      "step": 6637
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.416073094247277,
      "learning_rate": 4.740368641331894e-06,
      "loss": 0.5204,
      "step": 6638
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1765016169106137,
      "learning_rate": 4.7390748956706216e-06,
      "loss": 0.1628,
      "step": 6639
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3786089061978328,
      "learning_rate": 4.737781167526043e-06,
      "loss": 0.1735,
      "step": 6640
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4602935960213266,
      "learning_rate": 4.736487456985007e-06,
      "loss": 0.2048,
      "step": 6641
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3581421998607388,
      "learning_rate": 4.735193764134367e-06,
      "loss": 0.1906,
      "step": 6642
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4038250599105053,
      "learning_rate": 4.733900089060974e-06,
      "loss": 0.1943,
      "step": 6643
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.4694231968500855,
      "learning_rate": 4.732606431851672e-06,
      "loss": 0.6254,
      "step": 6644
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5271908450546945,
      "learning_rate": 4.731312792593311e-06,
      "loss": 0.2146,
      "step": 6645
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.292783833906374,
      "learning_rate": 4.730019171372736e-06,
      "loss": 0.1778,
      "step": 6646
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.150040173649783,
      "learning_rate": 4.728725568276793e-06,
      "loss": 0.5066,
      "step": 6647
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2625397301628674,
      "learning_rate": 4.727431983392323e-06,
      "loss": 0.166,
      "step": 6648
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1938061483824554,
      "learning_rate": 4.726138416806171e-06,
      "loss": 0.1538,
      "step": 6649
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3148058608095305,
      "learning_rate": 4.724844868605176e-06,
      "loss": 0.1851,
      "step": 6650
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.563592166279795,
      "learning_rate": 4.72355133887618e-06,
      "loss": 0.2397,
      "step": 6651
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.649686829916074,
      "learning_rate": 4.72225782770602e-06,
      "loss": 0.6216,
      "step": 6652
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.28005366089802,
      "learning_rate": 4.720964335181532e-06,
      "loss": 0.2087,
      "step": 6653
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4425657997833392,
      "learning_rate": 4.7196708613895555e-06,
      "loss": 0.202,
      "step": 6654
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5114304971400478,
      "learning_rate": 4.7183774064169215e-06,
      "loss": 0.2341,
      "step": 6655
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2185997645296378,
      "learning_rate": 4.717083970350468e-06,
      "loss": 0.1609,
      "step": 6656
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2764221484305232,
      "learning_rate": 4.715790553277022e-06,
      "loss": 0.1982,
      "step": 6657
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1754715359095889,
      "learning_rate": 4.714497155283419e-06,
      "loss": 0.1671,
      "step": 6658
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2241228217885318,
      "learning_rate": 4.713203776456486e-06,
      "loss": 0.1716,
      "step": 6659
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3776858559092595,
      "learning_rate": 4.711910416883054e-06,
      "loss": 0.1867,
      "step": 6660
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2032887300114448,
      "learning_rate": 4.710617076649948e-06,
      "loss": 0.1806,
      "step": 6661
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1698300582628904,
      "learning_rate": 4.709323755843993e-06,
      "loss": 0.1826,
      "step": 6662
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4432394440958993,
      "learning_rate": 4.7080304545520175e-06,
      "loss": 0.2008,
      "step": 6663
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1829300783352419,
      "learning_rate": 4.7067371728608395e-06,
      "loss": 0.1873,
      "step": 6664
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.45153836950273,
      "learning_rate": 4.7054439108572856e-06,
      "loss": 0.2458,
      "step": 6665
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3260024828157815,
      "learning_rate": 4.704150668628173e-06,
      "loss": 0.2096,
      "step": 6666
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4735533273079124,
      "learning_rate": 4.702857446260324e-06,
      "loss": 0.2051,
      "step": 6667
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4589662077916596,
      "learning_rate": 4.701564243840553e-06,
      "loss": 0.1531,
      "step": 6668
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6493672384073172,
      "learning_rate": 4.7002710614556805e-06,
      "loss": 0.2205,
      "step": 6669
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4681277141064315,
      "learning_rate": 4.69897789919252e-06,
      "loss": 0.1786,
      "step": 6670
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5257840952115231,
      "learning_rate": 4.697684757137883e-06,
      "loss": 0.2151,
      "step": 6671
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2791329601220198,
      "learning_rate": 4.696391635378586e-06,
      "loss": 0.1613,
      "step": 6672
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2818441955948678,
      "learning_rate": 4.695098534001437e-06,
      "loss": 0.1845,
      "step": 6673
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3830763697195854,
      "learning_rate": 4.693805453093249e-06,
      "loss": 0.1969,
      "step": 6674
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3291137688769834,
      "learning_rate": 4.6925123927408265e-06,
      "loss": 0.1763,
      "step": 6675
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3309494797501018,
      "learning_rate": 4.6912193530309805e-06,
      "loss": 0.1995,
      "step": 6676
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3087947815823218,
      "learning_rate": 4.689926334050513e-06,
      "loss": 0.1614,
      "step": 6677
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2510757329777877,
      "learning_rate": 4.6886333358862326e-06,
      "loss": 0.1615,
      "step": 6678
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3248074725593237,
      "learning_rate": 4.687340358624939e-06,
      "loss": 0.1776,
      "step": 6679
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.051690357740911,
      "learning_rate": 4.686047402353433e-06,
      "loss": 0.5547,
      "step": 6680
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4333452627641259,
      "learning_rate": 4.6847544671585185e-06,
      "loss": 0.1705,
      "step": 6681
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3700514680388276,
      "learning_rate": 4.68346155312699e-06,
      "loss": 0.1811,
      "step": 6682
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5123805389442506,
      "learning_rate": 4.682168660345647e-06,
      "loss": 0.2417,
      "step": 6683
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.237346360285923,
      "learning_rate": 4.680875788901284e-06,
      "loss": 0.1632,
      "step": 6684
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3940856177077163,
      "learning_rate": 4.679582938880698e-06,
      "loss": 0.1963,
      "step": 6685
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3460362806248896,
      "learning_rate": 4.67829011037068e-06,
      "loss": 0.1876,
      "step": 6686
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4680524905144217,
      "learning_rate": 4.67699730345802e-06,
      "loss": 0.2129,
      "step": 6687
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2990172369197281,
      "learning_rate": 4.675704518229512e-06,
      "loss": 0.1671,
      "step": 6688
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3201759114050347,
      "learning_rate": 4.674411754771939e-06,
      "loss": 0.1898,
      "step": 6689
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2850800343236037,
      "learning_rate": 4.673119013172093e-06,
      "loss": 0.1819,
      "step": 6690
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4471974965475833,
      "learning_rate": 4.671826293516758e-06,
      "loss": 0.2097,
      "step": 6691
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3200560564800181,
      "learning_rate": 4.67053359589272e-06,
      "loss": 0.1577,
      "step": 6692
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3889660411751608,
      "learning_rate": 4.669240920386756e-06,
      "loss": 0.1903,
      "step": 6693
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3565185059812563,
      "learning_rate": 4.667948267085655e-06,
      "loss": 0.2231,
      "step": 6694
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.336468069301615,
      "learning_rate": 4.6666556360761925e-06,
      "loss": 0.2062,
      "step": 6695
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2524329255250914,
      "learning_rate": 4.665363027445144e-06,
      "loss": 0.1547,
      "step": 6696
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.237215631747086,
      "learning_rate": 4.664070441279293e-06,
      "loss": 0.1884,
      "step": 6697
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2084125654754692,
      "learning_rate": 4.6627778776654085e-06,
      "loss": 0.1546,
      "step": 6698
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2999805282299985,
      "learning_rate": 4.661485336690268e-06,
      "loss": 0.1593,
      "step": 6699
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1575440435029185,
      "learning_rate": 4.660192818440642e-06,
      "loss": 0.1354,
      "step": 6700
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.264280740098368,
      "learning_rate": 4.658900323003303e-06,
      "loss": 0.2101,
      "step": 6701
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3298447557650057,
      "learning_rate": 4.657607850465017e-06,
      "loss": 0.1855,
      "step": 6702
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3083684128468835,
      "learning_rate": 4.656315400912554e-06,
      "loss": 0.1718,
      "step": 6703
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2592558368512883,
      "learning_rate": 4.655022974432681e-06,
      "loss": 0.2094,
      "step": 6704
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4013461092096426,
      "learning_rate": 4.653730571112159e-06,
      "loss": 0.1955,
      "step": 6705
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.7943585929237,
      "learning_rate": 4.652438191037754e-06,
      "loss": 0.5767,
      "step": 6706
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4174380779838232,
      "learning_rate": 4.6511458342962255e-06,
      "loss": 0.1847,
      "step": 6707
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4264623097995257,
      "learning_rate": 4.649853500974336e-06,
      "loss": 0.1986,
      "step": 6708
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3778893323818147,
      "learning_rate": 4.648561191158839e-06,
      "loss": 0.2025,
      "step": 6709
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4797639419767474,
      "learning_rate": 4.647268904936495e-06,
      "loss": 0.1947,
      "step": 6710
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2872928782716377,
      "learning_rate": 4.645976642394058e-06,
      "loss": 0.1729,
      "step": 6711
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3683633335942158,
      "learning_rate": 4.644684403618284e-06,
      "loss": 0.1759,
      "step": 6712
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3546114391091568,
      "learning_rate": 4.643392188695921e-06,
      "loss": 0.1784,
      "step": 6713
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2771567235567156,
      "learning_rate": 4.64209999771372e-06,
      "loss": 0.1771,
      "step": 6714
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3676320339308632,
      "learning_rate": 4.640807830758433e-06,
      "loss": 0.1817,
      "step": 6715
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.292661147361725,
      "learning_rate": 4.639515687916801e-06,
      "loss": 0.1881,
      "step": 6716
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.044023872874119,
      "learning_rate": 4.6382235692755756e-06,
      "loss": 0.5193,
      "step": 6717
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3572757170486722,
      "learning_rate": 4.6369314749214965e-06,
      "loss": 0.2144,
      "step": 6718
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3236558914891,
      "learning_rate": 4.63563940494131e-06,
      "loss": 0.2093,
      "step": 6719
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2950585456047743,
      "learning_rate": 4.6343473594217515e-06,
      "loss": 0.1617,
      "step": 6720
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3796416644826845,
      "learning_rate": 4.633055338449564e-06,
      "loss": 0.1796,
      "step": 6721
    },
    {
      "epoch": 0.54,
      "grad_norm": 16.65793525007263,
      "learning_rate": 4.631763342111485e-06,
      "loss": 0.7074,
      "step": 6722
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3174321036395324,
      "learning_rate": 4.630471370494246e-06,
      "loss": 0.1992,
      "step": 6723
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.392529332795741,
      "learning_rate": 4.629179423684585e-06,
      "loss": 0.2088,
      "step": 6724
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2651433582092273,
      "learning_rate": 4.627887501769231e-06,
      "loss": 0.2112,
      "step": 6725
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5069683250385393,
      "learning_rate": 4.626595604834918e-06,
      "loss": 0.2257,
      "step": 6726
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.307569893996264,
      "learning_rate": 4.625303732968373e-06,
      "loss": 0.1594,
      "step": 6727
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.365099838652735,
      "learning_rate": 4.624011886256323e-06,
      "loss": 0.1871,
      "step": 6728
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3808385984584466,
      "learning_rate": 4.622720064785495e-06,
      "loss": 0.1917,
      "step": 6729
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3902979454676903,
      "learning_rate": 4.621428268642613e-06,
      "loss": 0.1733,
      "step": 6730
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3121360568472442,
      "learning_rate": 4.620136497914399e-06,
      "loss": 0.2159,
      "step": 6731
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.483295275526807,
      "learning_rate": 4.61884475268757e-06,
      "loss": 0.2096,
      "step": 6732
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3756638486987285,
      "learning_rate": 4.617553033048851e-06,
      "loss": 0.2176,
      "step": 6733
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3572507463226735,
      "learning_rate": 4.616261339084952e-06,
      "loss": 0.1812,
      "step": 6734
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2292532494166757,
      "learning_rate": 4.614969670882594e-06,
      "loss": 0.1828,
      "step": 6735
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.388551462670128,
      "learning_rate": 4.6136780285284875e-06,
      "loss": 0.6536,
      "step": 6736
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.306036422348256,
      "learning_rate": 4.612386412109346e-06,
      "loss": 0.178,
      "step": 6737
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.668020671128,
      "learning_rate": 4.611094821711879e-06,
      "loss": 0.5992,
      "step": 6738
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2274674376363142,
      "learning_rate": 4.609803257422796e-06,
      "loss": 0.178,
      "step": 6739
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.39855725603969,
      "learning_rate": 4.608511719328803e-06,
      "loss": 0.1871,
      "step": 6740
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4803677707756495,
      "learning_rate": 4.607220207516602e-06,
      "loss": 0.2165,
      "step": 6741
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3838949620759733,
      "learning_rate": 4.605928722072901e-06,
      "loss": 0.1747,
      "step": 6742
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2106907767590576,
      "learning_rate": 4.604637263084397e-06,
      "loss": 0.133,
      "step": 6743
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.5326695306776905,
      "learning_rate": 4.603345830637792e-06,
      "loss": 0.6715,
      "step": 6744
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2970835298083734,
      "learning_rate": 4.602054424819782e-06,
      "loss": 0.1831,
      "step": 6745
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4159173533431881,
      "learning_rate": 4.600763045717067e-06,
      "loss": 0.1396,
      "step": 6746
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.458435106898632,
      "learning_rate": 4.599471693416337e-06,
      "loss": 0.2519,
      "step": 6747
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.748821595226993,
      "learning_rate": 4.598180368004285e-06,
      "loss": 0.5151,
      "step": 6748
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4275414204827768,
      "learning_rate": 4.5968890695676044e-06,
      "loss": 0.2001,
      "step": 6749
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.324307154241692,
      "learning_rate": 4.59559779819298e-06,
      "loss": 0.1582,
      "step": 6750
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4471224506599358,
      "learning_rate": 4.594306553967101e-06,
      "loss": 0.2059,
      "step": 6751
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3110883587353426,
      "learning_rate": 4.5930153369766515e-06,
      "loss": 0.1577,
      "step": 6752
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.474741907873556,
      "learning_rate": 4.591724147308318e-06,
      "loss": 0.1817,
      "step": 6753
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.250814597500748,
      "learning_rate": 4.590432985048777e-06,
      "loss": 0.6239,
      "step": 6754
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2688258885221322,
      "learning_rate": 4.589141850284712e-06,
      "loss": 0.1812,
      "step": 6755
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4355226756937578,
      "learning_rate": 4.5878507431028005e-06,
      "loss": 0.2043,
      "step": 6756
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3545989128067748,
      "learning_rate": 4.5865596635897155e-06,
      "loss": 0.151,
      "step": 6757
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3843928848876577,
      "learning_rate": 4.585268611832134e-06,
      "loss": 0.1955,
      "step": 6758
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.246665766955404,
      "learning_rate": 4.583977587916725e-06,
      "loss": 0.1499,
      "step": 6759
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3654541255562844,
      "learning_rate": 4.5826865919301645e-06,
      "loss": 0.1985,
      "step": 6760
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4636712890094812,
      "learning_rate": 4.581395623959115e-06,
      "loss": 0.2374,
      "step": 6761
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8189095309876577,
      "learning_rate": 4.580104684090246e-06,
      "loss": 0.2052,
      "step": 6762
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.282203251845499,
      "learning_rate": 4.578813772410221e-06,
      "loss": 0.1994,
      "step": 6763
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5291326164997934,
      "learning_rate": 4.577522889005706e-06,
      "loss": 0.2213,
      "step": 6764
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4112570720414033,
      "learning_rate": 4.5762320339633585e-06,
      "loss": 0.2364,
      "step": 6765
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4494647196970396,
      "learning_rate": 4.5749412073698376e-06,
      "loss": 0.2213,
      "step": 6766
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.391002288696143,
      "learning_rate": 4.573650409311803e-06,
      "loss": 0.1769,
      "step": 6767
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4422199926436379,
      "learning_rate": 4.572359639875906e-06,
      "loss": 0.2052,
      "step": 6768
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2539810914171403,
      "learning_rate": 4.571068899148804e-06,
      "loss": 0.2116,
      "step": 6769
    },
    {
      "epoch": 0.54,
      "grad_norm": 21.730678253932666,
      "learning_rate": 4.569778187217144e-06,
      "loss": 0.5484,
      "step": 6770
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.372478911989215,
      "learning_rate": 4.568487504167581e-06,
      "loss": 0.1715,
      "step": 6771
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3829501142972014,
      "learning_rate": 4.567196850086757e-06,
      "loss": 0.1625,
      "step": 6772
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4281710141016053,
      "learning_rate": 4.565906225061321e-06,
      "loss": 0.1642,
      "step": 6773
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.419482849634811,
      "learning_rate": 4.564615629177916e-06,
      "loss": 0.195,
      "step": 6774
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.296173511883808,
      "learning_rate": 4.5633250625231806e-06,
      "loss": 0.1844,
      "step": 6775
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4392133476221018,
      "learning_rate": 4.562034525183758e-06,
      "loss": 0.2245,
      "step": 6776
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3359981875258926,
      "learning_rate": 4.560744017246284e-06,
      "loss": 0.1947,
      "step": 6777
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9189177146640684,
      "learning_rate": 4.559453538797398e-06,
      "loss": 0.151,
      "step": 6778
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3112614638625646,
      "learning_rate": 4.558163089923726e-06,
      "loss": 0.1754,
      "step": 6779
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3247573946077167,
      "learning_rate": 4.556872670711908e-06,
      "loss": 0.1868,
      "step": 6780
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6118133676327768,
      "learning_rate": 4.555582281248569e-06,
      "loss": 0.2036,
      "step": 6781
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2877771635078612,
      "learning_rate": 4.554291921620338e-06,
      "loss": 0.1689,
      "step": 6782
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.206120581386168,
      "learning_rate": 4.5530015919138425e-06,
      "loss": 0.665,
      "step": 6783
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4884691110504413,
      "learning_rate": 4.551711292215702e-06,
      "loss": 0.1476,
      "step": 6784
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3523883070191174,
      "learning_rate": 4.550421022612542e-06,
      "loss": 0.1904,
      "step": 6785
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3039106296455183,
      "learning_rate": 4.549130783190979e-06,
      "loss": 0.1995,
      "step": 6786
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2735095733703758,
      "learning_rate": 4.547840574037636e-06,
      "loss": 0.1899,
      "step": 6787
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2882889695637363,
      "learning_rate": 4.546550395239123e-06,
      "loss": 0.1637,
      "step": 6788
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.476598782247935,
      "learning_rate": 4.545260246882056e-06,
      "loss": 0.2163,
      "step": 6789
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3469773257509285,
      "learning_rate": 4.543970129053047e-06,
      "loss": 0.1845,
      "step": 6790
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.341372296139701,
      "learning_rate": 4.542680041838705e-06,
      "loss": 0.2192,
      "step": 6791
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5037484943636448,
      "learning_rate": 4.541389985325637e-06,
      "loss": 0.202,
      "step": 6792
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.263852218207307,
      "learning_rate": 4.5400999596004484e-06,
      "loss": 0.1762,
      "step": 6793
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5119791726063663,
      "learning_rate": 4.538809964749745e-06,
      "loss": 0.2354,
      "step": 6794
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.275755030366224,
      "learning_rate": 4.537520000860124e-06,
      "loss": 0.533,
      "step": 6795
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3754948908209557,
      "learning_rate": 4.536230068018187e-06,
      "loss": 0.2322,
      "step": 6796
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2910823051613778,
      "learning_rate": 4.53494016631053e-06,
      "loss": 0.1942,
      "step": 6797
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.407064863655799,
      "learning_rate": 4.533650295823751e-06,
      "loss": 0.2049,
      "step": 6798
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.314626772977091,
      "learning_rate": 4.532360456644438e-06,
      "loss": 0.2379,
      "step": 6799
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5363117560579744,
      "learning_rate": 4.531070648859186e-06,
      "loss": 0.1625,
      "step": 6800
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2941640012380784,
      "learning_rate": 4.529780872554582e-06,
      "loss": 0.1993,
      "step": 6801
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.345887212749787,
      "learning_rate": 4.528491127817213e-06,
      "loss": 0.1971,
      "step": 6802
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.199838695557625,
      "learning_rate": 4.5272014147336625e-06,
      "loss": 0.1809,
      "step": 6803
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3335644289691695,
      "learning_rate": 4.525911733390513e-06,
      "loss": 0.185,
      "step": 6804
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.283270776937424,
      "learning_rate": 4.524622083874347e-06,
      "loss": 0.5506,
      "step": 6805
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.819464192253964,
      "learning_rate": 4.523332466271739e-06,
      "loss": 0.5976,
      "step": 6806
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3239690413298923,
      "learning_rate": 4.522042880669268e-06,
      "loss": 0.1849,
      "step": 6807
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4368972244893568,
      "learning_rate": 4.5207533271535075e-06,
      "loss": 0.2026,
      "step": 6808
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4140409112760415,
      "learning_rate": 4.519463805811026e-06,
      "loss": 0.191,
      "step": 6809
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.321285354453828,
      "learning_rate": 4.518174316728396e-06,
      "loss": 0.1913,
      "step": 6810
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4098685212310236,
      "learning_rate": 4.516884859992183e-06,
      "loss": 0.1836,
      "step": 6811
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.453116455546688,
      "learning_rate": 4.5155954356889555e-06,
      "loss": 0.2032,
      "step": 6812
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.076907258154822,
      "learning_rate": 4.5143060439052725e-06,
      "loss": 0.1391,
      "step": 6813
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.581716515987097,
      "learning_rate": 4.513016684727697e-06,
      "loss": 0.7686,
      "step": 6814
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.17948294903175,
      "learning_rate": 4.511727358242786e-06,
      "loss": 0.6994,
      "step": 6815
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.97366186282323,
      "learning_rate": 4.5104380645370986e-06,
      "loss": 0.5059,
      "step": 6816
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4225888087599794,
      "learning_rate": 4.509148803697186e-06,
      "loss": 0.2253,
      "step": 6817
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4557908129285921,
      "learning_rate": 4.507859575809601e-06,
      "loss": 0.1578,
      "step": 6818
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3421951143343749,
      "learning_rate": 4.506570380960895e-06,
      "loss": 0.216,
      "step": 6819
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3360203227973844,
      "learning_rate": 4.505281219237613e-06,
      "loss": 0.1912,
      "step": 6820
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5284278674803466,
      "learning_rate": 4.503992090726302e-06,
      "loss": 0.2369,
      "step": 6821
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3458386937740916,
      "learning_rate": 4.5027029955135045e-06,
      "loss": 0.1861,
      "step": 6822
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.430955980185571,
      "learning_rate": 4.501413933685763e-06,
      "loss": 0.2296,
      "step": 6823
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3130955042187646,
      "learning_rate": 4.500124905329613e-06,
      "loss": 0.1872,
      "step": 6824
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3574178417303149,
      "learning_rate": 4.498835910531595e-06,
      "loss": 0.1816,
      "step": 6825
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4886990548509318,
      "learning_rate": 4.4975469493782405e-06,
      "loss": 0.2169,
      "step": 6826
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4922240717432786,
      "learning_rate": 4.496258021956079e-06,
      "loss": 0.2192,
      "step": 6827
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2060400949021521,
      "learning_rate": 4.494969128351646e-06,
      "loss": 0.1756,
      "step": 6828
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.473464837586222,
      "learning_rate": 4.493680268651464e-06,
      "loss": 0.1987,
      "step": 6829
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.719087545658891,
      "learning_rate": 4.4923914429420595e-06,
      "loss": 0.6771,
      "step": 6830
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.350113007890623,
      "learning_rate": 4.491102651309954e-06,
      "loss": 0.2071,
      "step": 6831
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4119349846512868,
      "learning_rate": 4.489813893841673e-06,
      "loss": 0.2046,
      "step": 6832
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3773206686993182,
      "learning_rate": 4.488525170623729e-06,
      "loss": 0.19,
      "step": 6833
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2684699818610259,
      "learning_rate": 4.48723648174264e-06,
      "loss": 0.2011,
      "step": 6834
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3613182146830667,
      "learning_rate": 4.485947827284921e-06,
      "loss": 0.216,
      "step": 6835
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.344441486208463,
      "learning_rate": 4.48465920733708e-06,
      "loss": 0.1688,
      "step": 6836
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.269048211686111,
      "learning_rate": 4.483370621985629e-06,
      "loss": 0.2608,
      "step": 6837
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3238833773034413,
      "learning_rate": 4.482082071317071e-06,
      "loss": 0.1863,
      "step": 6838
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4585108576039179,
      "learning_rate": 4.480793555417917e-06,
      "loss": 0.2283,
      "step": 6839
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3794390147776732,
      "learning_rate": 4.479505074374662e-06,
      "loss": 0.179,
      "step": 6840
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.242688706855339,
      "learning_rate": 4.478216628273809e-06,
      "loss": 0.1832,
      "step": 6841
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4551189133191802,
      "learning_rate": 4.4769282172018535e-06,
      "loss": 0.2051,
      "step": 6842
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3252719160204203,
      "learning_rate": 4.475639841245294e-06,
      "loss": 0.188,
      "step": 6843
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.321046666609145,
      "learning_rate": 4.47435150049062e-06,
      "loss": 0.144,
      "step": 6844
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2666294160922735,
      "learning_rate": 4.47306319502432e-06,
      "loss": 0.1668,
      "step": 6845
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2520355778465115,
      "learning_rate": 4.4717749249328876e-06,
      "loss": 0.1753,
      "step": 6846
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3341770310064853,
      "learning_rate": 4.470486690302803e-06,
      "loss": 0.2079,
      "step": 6847
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3143972729257676,
      "learning_rate": 4.469198491220551e-06,
      "loss": 0.2113,
      "step": 6848
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2747024485157306,
      "learning_rate": 4.467910327772613e-06,
      "loss": 0.1636,
      "step": 6849
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3692371705987472,
      "learning_rate": 4.4666222000454685e-06,
      "loss": 0.1638,
      "step": 6850
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.282137349925222,
      "learning_rate": 4.4653341081255895e-06,
      "loss": 0.6384,
      "step": 6851
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.277097758981777,
      "learning_rate": 4.464046052099453e-06,
      "loss": 0.2047,
      "step": 6852
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3148740244438768,
      "learning_rate": 4.46275803205353e-06,
      "loss": 0.1945,
      "step": 6853
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1592012204460431,
      "learning_rate": 4.461470048074286e-06,
      "loss": 0.1322,
      "step": 6854
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.591474962860619,
      "learning_rate": 4.46018210024819e-06,
      "loss": 0.5113,
      "step": 6855
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.116987230832049,
      "learning_rate": 4.458894188661704e-06,
      "loss": 0.621,
      "step": 6856
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4075600658815688,
      "learning_rate": 4.4576063134012935e-06,
      "loss": 0.1896,
      "step": 6857
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.193468523185192,
      "learning_rate": 4.456318474553412e-06,
      "loss": 0.1836,
      "step": 6858
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.399412090114669,
      "learning_rate": 4.45503067220452e-06,
      "loss": 0.1941,
      "step": 6859
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2848275985331417,
      "learning_rate": 4.4537429064410685e-06,
      "loss": 0.1609,
      "step": 6860
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.282004781173103,
      "learning_rate": 4.452455177349513e-06,
      "loss": 0.1974,
      "step": 6861
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.41569947839218,
      "learning_rate": 4.4511674850163e-06,
      "loss": 0.202,
      "step": 6862
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4823514214157687,
      "learning_rate": 4.449879829527875e-06,
      "loss": 0.2178,
      "step": 6863
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2950654121797143,
      "learning_rate": 4.448592210970687e-06,
      "loss": 0.1638,
      "step": 6864
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4244018868375872,
      "learning_rate": 4.44730462943117e-06,
      "loss": 0.2138,
      "step": 6865
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.558880671815828,
      "learning_rate": 4.446017084995771e-06,
      "loss": 0.2455,
      "step": 6866
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1399705800233373,
      "learning_rate": 4.444729577750922e-06,
      "loss": 0.1811,
      "step": 6867
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.410044543661609,
      "learning_rate": 4.44344210778306e-06,
      "loss": 0.1747,
      "step": 6868
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.277621234477029,
      "learning_rate": 4.442154675178617e-06,
      "loss": 0.1688,
      "step": 6869
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.820579125163696,
      "learning_rate": 4.4408672800240185e-06,
      "loss": 0.3982,
      "step": 6870
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2871127706426828,
      "learning_rate": 4.439579922405694e-06,
      "loss": 0.1496,
      "step": 6871
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3362104227675218,
      "learning_rate": 4.438292602410067e-06,
      "loss": 0.1793,
      "step": 6872
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.291759069437098,
      "learning_rate": 4.437005320123561e-06,
      "loss": 0.17,
      "step": 6873
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5110594109568594,
      "learning_rate": 4.4357180756325915e-06,
      "loss": 0.2699,
      "step": 6874
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4701123236645992,
      "learning_rate": 4.434430869023579e-06,
      "loss": 0.2441,
      "step": 6875
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.160343572115143,
      "learning_rate": 4.433143700382935e-06,
      "loss": 0.1671,
      "step": 6876
    },
    {
      "epoch": 0.55,
      "grad_norm": 12.434658475586604,
      "learning_rate": 4.4318565697970734e-06,
      "loss": 0.5358,
      "step": 6877
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.625572663278618,
      "learning_rate": 4.4305694773524e-06,
      "loss": 0.2088,
      "step": 6878
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.440715595174671,
      "learning_rate": 4.429282423135323e-06,
      "loss": 0.2129,
      "step": 6879
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6063175666066074,
      "learning_rate": 4.4279954072322486e-06,
      "loss": 0.2345,
      "step": 6880
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1151437160679623,
      "learning_rate": 4.426708429729573e-06,
      "loss": 0.1838,
      "step": 6881
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2969865591012761,
      "learning_rate": 4.425421490713698e-06,
      "loss": 0.1628,
      "step": 6882
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2689328201483292,
      "learning_rate": 4.424134590271018e-06,
      "loss": 0.1839,
      "step": 6883
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1777786371775967,
      "learning_rate": 4.4228477284879305e-06,
      "loss": 0.1797,
      "step": 6884
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2768149001123978,
      "learning_rate": 4.4215609054508215e-06,
      "loss": 0.2222,
      "step": 6885
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.41245343953405,
      "learning_rate": 4.420274121246083e-06,
      "loss": 0.4898,
      "step": 6886
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2982857678427775,
      "learning_rate": 4.418987375960099e-06,
      "loss": 0.1878,
      "step": 6887
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3818504444036372,
      "learning_rate": 4.417700669679252e-06,
      "loss": 0.2157,
      "step": 6888
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2954598200406084,
      "learning_rate": 4.416414002489924e-06,
      "loss": 0.2078,
      "step": 6889
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3541970334816915,
      "learning_rate": 4.415127374478491e-06,
      "loss": 0.1508,
      "step": 6890
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.298558014680946,
      "learning_rate": 4.413840785731332e-06,
      "loss": 0.1592,
      "step": 6891
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2252981819302706,
      "learning_rate": 4.412554236334815e-06,
      "loss": 0.2092,
      "step": 6892
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5705488227601931,
      "learning_rate": 4.411267726375312e-06,
      "loss": 0.2424,
      "step": 6893
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2593899933175228,
      "learning_rate": 4.4099812559391906e-06,
      "loss": 0.174,
      "step": 6894
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.220553384140846,
      "learning_rate": 4.4086948251128155e-06,
      "loss": 0.6292,
      "step": 6895
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4870955066854525,
      "learning_rate": 4.407408433982549e-06,
      "loss": 0.1898,
      "step": 6896
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.223490542937882,
      "learning_rate": 4.406122082634748e-06,
      "loss": 0.1466,
      "step": 6897
    },
    {
      "epoch": 0.55,
      "grad_norm": 16.646636375550507,
      "learning_rate": 4.404835771155774e-06,
      "loss": 0.5402,
      "step": 6898
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4078963759997407,
      "learning_rate": 4.403549499631974e-06,
      "loss": 0.1839,
      "step": 6899
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.404806455728005,
      "learning_rate": 4.402263268149707e-06,
      "loss": 0.6505,
      "step": 6900
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4117806496624512,
      "learning_rate": 4.400977076795314e-06,
      "loss": 0.2174,
      "step": 6901
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3217136664515383,
      "learning_rate": 4.399690925655148e-06,
      "loss": 0.1952,
      "step": 6902
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3545036325548305,
      "learning_rate": 4.398404814815548e-06,
      "loss": 0.194,
      "step": 6903
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4617570792078414,
      "learning_rate": 4.397118744362855e-06,
      "loss": 0.1657,
      "step": 6904
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1795052885377475,
      "learning_rate": 4.39583271438341e-06,
      "loss": 0.1618,
      "step": 6905
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3258257555662984,
      "learning_rate": 4.394546724963542e-06,
      "loss": 0.1983,
      "step": 6906
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.282195119884134,
      "learning_rate": 4.393260776189589e-06,
      "loss": 0.1733,
      "step": 6907
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2744893457013906,
      "learning_rate": 4.391974868147875e-06,
      "loss": 0.167,
      "step": 6908
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.318358811017309,
      "learning_rate": 4.390689000924734e-06,
      "loss": 0.2041,
      "step": 6909
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.746314917075147,
      "learning_rate": 4.389403174606484e-06,
      "loss": 0.6059,
      "step": 6910
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4109131453650452,
      "learning_rate": 4.388117389279452e-06,
      "loss": 0.2037,
      "step": 6911
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.684213385405756,
      "learning_rate": 4.386831645029951e-06,
      "loss": 0.5197,
      "step": 6912
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4488463711988058,
      "learning_rate": 4.385545941944301e-06,
      "loss": 0.22,
      "step": 6913
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4166541641350296,
      "learning_rate": 4.3842602801088145e-06,
      "loss": 0.1896,
      "step": 6914
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2281973131274015,
      "learning_rate": 4.3829746596097975e-06,
      "loss": 0.2029,
      "step": 6915
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2917018253838513,
      "learning_rate": 4.381689080533564e-06,
      "loss": 0.1982,
      "step": 6916
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3502571550689517,
      "learning_rate": 4.380403542966414e-06,
      "loss": 0.1692,
      "step": 6917
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5262757647622938,
      "learning_rate": 4.379118046994654e-06,
      "loss": 0.2155,
      "step": 6918
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2411322665755142,
      "learning_rate": 4.377832592704578e-06,
      "loss": 0.1634,
      "step": 6919
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.292136317930179,
      "learning_rate": 4.3765471801824865e-06,
      "loss": 0.1866,
      "step": 6920
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.252074488100489,
      "learning_rate": 4.375261809514671e-06,
      "loss": 0.1468,
      "step": 6921
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.862687447389899,
      "learning_rate": 4.373976480787425e-06,
      "loss": 0.5212,
      "step": 6922
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.344089182351659,
      "learning_rate": 4.372691194087034e-06,
      "loss": 0.2248,
      "step": 6923
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.467271247957386,
      "learning_rate": 4.371405949499782e-06,
      "loss": 0.2118,
      "step": 6924
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4654762448921959,
      "learning_rate": 4.370120747111956e-06,
      "loss": 0.1774,
      "step": 6925
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1342282856647932,
      "learning_rate": 4.36883558700983e-06,
      "loss": 0.1314,
      "step": 6926
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.196690057521876,
      "learning_rate": 4.3675504692796845e-06,
      "loss": 0.1498,
      "step": 6927
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1382294653395975,
      "learning_rate": 4.3662653940077915e-06,
      "loss": 0.1502,
      "step": 6928
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3348484263262625,
      "learning_rate": 4.364980361280425e-06,
      "loss": 0.179,
      "step": 6929
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1444186481565346,
      "learning_rate": 4.363695371183849e-06,
      "loss": 0.1401,
      "step": 6930
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2163865394392357,
      "learning_rate": 4.362410423804331e-06,
      "loss": 0.1526,
      "step": 6931
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.5688700201532395,
      "learning_rate": 4.3611255192281335e-06,
      "loss": 0.5703,
      "step": 6932
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.550668132794035,
      "learning_rate": 4.359840657541515e-06,
      "loss": 0.215,
      "step": 6933
    },
    {
      "epoch": 0.55,
      "grad_norm": 17.169716410449166,
      "learning_rate": 4.358555838830733e-06,
      "loss": 0.6075,
      "step": 6934
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2975465806460014,
      "learning_rate": 4.35727106318204e-06,
      "loss": 0.1593,
      "step": 6935
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2756947230864846,
      "learning_rate": 4.35598633068169e-06,
      "loss": 0.1676,
      "step": 6936
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4927521500964291,
      "learning_rate": 4.354701641415927e-06,
      "loss": 0.2429,
      "step": 6937
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2187304943784556,
      "learning_rate": 4.353416995470999e-06,
      "loss": 0.1419,
      "step": 6938
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2939802364891733,
      "learning_rate": 4.352132392933148e-06,
      "loss": 0.1838,
      "step": 6939
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3744035356063795,
      "learning_rate": 4.3508478338886105e-06,
      "loss": 0.2148,
      "step": 6940
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4078221056672584,
      "learning_rate": 4.3495633184236265e-06,
      "loss": 0.2053,
      "step": 6941
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3336891451497794,
      "learning_rate": 4.348278846624426e-06,
      "loss": 0.1849,
      "step": 6942
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.265314235694341,
      "learning_rate": 4.346994418577243e-06,
      "loss": 0.1739,
      "step": 6943
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.251636127096702,
      "learning_rate": 4.345710034368301e-06,
      "loss": 0.7474,
      "step": 6944
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3544297690245544,
      "learning_rate": 4.344425694083829e-06,
      "loss": 0.2082,
      "step": 6945
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5084209740631263,
      "learning_rate": 4.343141397810044e-06,
      "loss": 0.189,
      "step": 6946
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.165570371887376,
      "learning_rate": 4.34185714563317e-06,
      "loss": 0.1777,
      "step": 6947
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2745033339331835,
      "learning_rate": 4.340572937639419e-06,
      "loss": 0.205,
      "step": 6948
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.312434279134441,
      "learning_rate": 4.339288773915003e-06,
      "loss": 0.5459,
      "step": 6949
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.840235480049677,
      "learning_rate": 4.338004654546136e-06,
      "loss": 0.5943,
      "step": 6950
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4353667086587494,
      "learning_rate": 4.336720579619019e-06,
      "loss": 0.2088,
      "step": 6951
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5458032003734916,
      "learning_rate": 4.335436549219862e-06,
      "loss": 0.2337,
      "step": 6952
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.228362293302541,
      "learning_rate": 4.3341525634348615e-06,
      "loss": 0.1868,
      "step": 6953
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2961893278710508,
      "learning_rate": 4.332868622350218e-06,
      "loss": 0.1395,
      "step": 6954
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3210471595097677,
      "learning_rate": 4.331584726052124e-06,
      "loss": 0.1764,
      "step": 6955
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2363000973947313,
      "learning_rate": 4.330300874626774e-06,
      "loss": 0.2007,
      "step": 6956
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3783389122480192,
      "learning_rate": 4.329017068160355e-06,
      "loss": 0.1761,
      "step": 6957
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2709263564353028,
      "learning_rate": 4.327733306739053e-06,
      "loss": 0.1892,
      "step": 6958
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4770586036175555,
      "learning_rate": 4.3264495904490514e-06,
      "loss": 0.1886,
      "step": 6959
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2471612185615193,
      "learning_rate": 4.325165919376528e-06,
      "loss": 0.1552,
      "step": 6960
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3455348326533412,
      "learning_rate": 4.323882293607663e-06,
      "loss": 0.1782,
      "step": 6961
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4291009403387405,
      "learning_rate": 4.322598713228626e-06,
      "loss": 0.2572,
      "step": 6962
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2909725625655353,
      "learning_rate": 4.321315178325593e-06,
      "loss": 0.1698,
      "step": 6963
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.409756755830917,
      "learning_rate": 4.320031688984726e-06,
      "loss": 0.2246,
      "step": 6964
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.26250424352819,
      "learning_rate": 4.318748245292193e-06,
      "loss": 0.1939,
      "step": 6965
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4627545534851207,
      "learning_rate": 4.317464847334154e-06,
      "loss": 0.2048,
      "step": 6966
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.297654323781324,
      "learning_rate": 4.316181495196767e-06,
      "loss": 0.1873,
      "step": 6967
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.150156674525904,
      "learning_rate": 4.31489818896619e-06,
      "loss": 0.1614,
      "step": 6968
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3586783765911885,
      "learning_rate": 4.313614928728571e-06,
      "loss": 0.2232,
      "step": 6969
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.351133736023884,
      "learning_rate": 4.312331714570064e-06,
      "loss": 0.1849,
      "step": 6970
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.8321359627386,
      "learning_rate": 4.31104854657681e-06,
      "loss": 0.5887,
      "step": 6971
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5096575919780169,
      "learning_rate": 4.3097654248349565e-06,
      "loss": 0.2179,
      "step": 6972
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.439767616226352,
      "learning_rate": 4.30848234943064e-06,
      "loss": 0.5017,
      "step": 6973
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.369051248796137,
      "learning_rate": 4.30719932045e-06,
      "loss": 0.2071,
      "step": 6974
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3652275023393845,
      "learning_rate": 4.3059163379791676e-06,
      "loss": 0.1726,
      "step": 6975
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2776663218437052,
      "learning_rate": 4.304633402104274e-06,
      "loss": 0.1833,
      "step": 6976
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4201243365422023,
      "learning_rate": 4.303350512911449e-06,
      "loss": 0.1919,
      "step": 6977
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9669754494502172,
      "learning_rate": 4.302067670486813e-06,
      "loss": 0.2488,
      "step": 6978
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2792280518222674,
      "learning_rate": 4.300784874916489e-06,
      "loss": 0.1897,
      "step": 6979
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6070159266529367,
      "learning_rate": 4.299502126286596e-06,
      "loss": 0.23,
      "step": 6980
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1430570673456661,
      "learning_rate": 4.298219424683247e-06,
      "loss": 0.1629,
      "step": 6981
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2649039861967415,
      "learning_rate": 4.296936770192554e-06,
      "loss": 0.206,
      "step": 6982
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4334455219775208,
      "learning_rate": 4.295654162900626e-06,
      "loss": 0.1932,
      "step": 6983
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.174591000577947,
      "learning_rate": 4.29437160289357e-06,
      "loss": 0.1828,
      "step": 6984
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4484149257549188,
      "learning_rate": 4.293089090257484e-06,
      "loss": 0.1616,
      "step": 6985
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4359006713643834,
      "learning_rate": 4.29180662507847e-06,
      "loss": 0.1716,
      "step": 6986
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5546392124346744,
      "learning_rate": 4.290524207442621e-06,
      "loss": 0.1598,
      "step": 6987
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2801803014038105,
      "learning_rate": 4.289241837436036e-06,
      "loss": 0.1824,
      "step": 6988
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1898885197234932,
      "learning_rate": 4.287959515144796e-06,
      "loss": 0.1636,
      "step": 6989
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3472893795377605,
      "learning_rate": 4.286677240654993e-06,
      "loss": 0.1757,
      "step": 6990
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4512722993445515,
      "learning_rate": 4.285395014052707e-06,
      "loss": 0.1798,
      "step": 6991
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4854113372085438,
      "learning_rate": 4.28411283542402e-06,
      "loss": 0.2449,
      "step": 6992
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1713536800234112,
      "learning_rate": 4.282830704855008e-06,
      "loss": 0.1826,
      "step": 6993
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2877775016406923,
      "learning_rate": 4.281548622431741e-06,
      "loss": 0.167,
      "step": 6994
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2880159931247408,
      "learning_rate": 4.280266588240294e-06,
      "loss": 0.1991,
      "step": 6995
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2399922881753271,
      "learning_rate": 4.27898460236673e-06,
      "loss": 0.2155,
      "step": 6996
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3932948467073165,
      "learning_rate": 4.277702664897117e-06,
      "loss": 0.1929,
      "step": 6997
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.272301501424295,
      "learning_rate": 4.27642077591751e-06,
      "loss": 0.1816,
      "step": 6998
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4366076533987884,
      "learning_rate": 4.27513893551397e-06,
      "loss": 0.1995,
      "step": 6999
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3474693807999927,
      "learning_rate": 4.27385714377255e-06,
      "loss": 0.192,
      "step": 7000
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3257592617600902,
      "learning_rate": 4.272575400779298e-06,
      "loss": 0.2066,
      "step": 7001
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1077017797789337,
      "learning_rate": 4.271293706620265e-06,
      "loss": 0.1563,
      "step": 7002
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.719836949545226,
      "learning_rate": 4.270012061381492e-06,
      "loss": 0.5142,
      "step": 7003
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3454663428912361,
      "learning_rate": 4.268730465149024e-06,
      "loss": 0.1582,
      "step": 7004
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.0936098196911415,
      "learning_rate": 4.267448918008892e-06,
      "loss": 0.4865,
      "step": 7005
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4557771354901219,
      "learning_rate": 4.266167420047136e-06,
      "loss": 0.202,
      "step": 7006
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3924029285186255,
      "learning_rate": 4.264885971349782e-06,
      "loss": 0.1816,
      "step": 7007
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.332664448003593,
      "learning_rate": 4.263604572002863e-06,
      "loss": 0.1891,
      "step": 7008
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3502064968333063,
      "learning_rate": 4.262323222092399e-06,
      "loss": 0.1712,
      "step": 7009
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.97225876758747,
      "learning_rate": 4.2610419217044115e-06,
      "loss": 0.4607,
      "step": 7010
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2914339009004334,
      "learning_rate": 4.25976067092492e-06,
      "loss": 0.1592,
      "step": 7011
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2548329272902767,
      "learning_rate": 4.2584794698399364e-06,
      "loss": 0.1751,
      "step": 7012
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4242878698417076,
      "learning_rate": 4.257198318535474e-06,
      "loss": 0.2037,
      "step": 7013
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3964214133138348,
      "learning_rate": 4.255917217097537e-06,
      "loss": 0.2058,
      "step": 7014
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2803677529839728,
      "learning_rate": 4.254636165612135e-06,
      "loss": 0.1515,
      "step": 7015
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4076245415062942,
      "learning_rate": 4.253355164165262e-06,
      "loss": 0.1716,
      "step": 7016
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2401166292813786,
      "learning_rate": 4.252074212842922e-06,
      "loss": 0.1662,
      "step": 7017
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2208282596730105,
      "learning_rate": 4.2507933117311055e-06,
      "loss": 0.1341,
      "step": 7018
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4909854122146164,
      "learning_rate": 4.249512460915802e-06,
      "loss": 0.1604,
      "step": 7019
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.419998057773177,
      "learning_rate": 4.248231660483002e-06,
      "loss": 0.2213,
      "step": 7020
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3084865725366324,
      "learning_rate": 4.2469509105186884e-06,
      "loss": 0.1862,
      "step": 7021
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.218506054439855,
      "learning_rate": 4.245670211108843e-06,
      "loss": 0.1776,
      "step": 7022
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2450991116548429,
      "learning_rate": 4.24438956233944e-06,
      "loss": 0.1815,
      "step": 7023
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.516621515048516,
      "learning_rate": 4.2431089642964564e-06,
      "loss": 0.2309,
      "step": 7024
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2602599177120537,
      "learning_rate": 4.24182841706586e-06,
      "loss": 0.1947,
      "step": 7025
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.364875354269461,
      "learning_rate": 4.240547920733622e-06,
      "loss": 0.152,
      "step": 7026
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2943959212217564,
      "learning_rate": 4.239267475385701e-06,
      "loss": 0.1955,
      "step": 7027
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.375323779324813,
      "learning_rate": 4.2379870811080585e-06,
      "loss": 0.4726,
      "step": 7028
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2400978945827188,
      "learning_rate": 4.236706737986654e-06,
      "loss": 0.1652,
      "step": 7029
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.419501598502113,
      "learning_rate": 4.235426446107437e-06,
      "loss": 0.1746,
      "step": 7030
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3570713962999794,
      "learning_rate": 4.2341462055563605e-06,
      "loss": 0.234,
      "step": 7031
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4746259743296237,
      "learning_rate": 4.2328660164193695e-06,
      "loss": 0.2509,
      "step": 7032
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2357181973315485,
      "learning_rate": 4.231585878782408e-06,
      "loss": 0.1859,
      "step": 7033
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.006620903034838,
      "learning_rate": 4.230305792731411e-06,
      "loss": 0.4002,
      "step": 7034
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2816006095719912,
      "learning_rate": 4.229025758352322e-06,
      "loss": 0.2076,
      "step": 7035
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3689260346436294,
      "learning_rate": 4.227745775731071e-06,
      "loss": 0.2045,
      "step": 7036
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.88186483104484,
      "learning_rate": 4.226465844953581e-06,
      "loss": 0.6176,
      "step": 7037
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3170562453134165,
      "learning_rate": 4.225185966105786e-06,
      "loss": 0.1458,
      "step": 7038
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4164123237210053,
      "learning_rate": 4.223906139273603e-06,
      "loss": 0.19,
      "step": 7039
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2660214247633335,
      "learning_rate": 4.2226263645429536e-06,
      "loss": 0.1776,
      "step": 7040
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3488838044225833,
      "learning_rate": 4.22134664199975e-06,
      "loss": 0.1777,
      "step": 7041
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2477248586884493,
      "learning_rate": 4.220066971729908e-06,
      "loss": 0.1672,
      "step": 7042
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.210233556837474,
      "learning_rate": 4.218787353819331e-06,
      "loss": 0.1928,
      "step": 7043
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5574468566890245,
      "learning_rate": 4.217507788353927e-06,
      "loss": 0.2522,
      "step": 7044
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4767334596986261,
      "learning_rate": 4.216228275419598e-06,
      "loss": 0.1757,
      "step": 7045
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0810576555502336,
      "learning_rate": 4.214948815102237e-06,
      "loss": 0.1637,
      "step": 7046
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.865566914567689,
      "learning_rate": 4.213669407487741e-06,
      "loss": 0.4762,
      "step": 7047
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2631110663073912,
      "learning_rate": 4.212390052661999e-06,
      "loss": 0.1766,
      "step": 7048
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2462824563629258,
      "learning_rate": 4.2111107507109025e-06,
      "loss": 0.1685,
      "step": 7049
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.495245549531543,
      "learning_rate": 4.209831501720328e-06,
      "loss": 0.6785,
      "step": 7050
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1862416029389309,
      "learning_rate": 4.208552305776162e-06,
      "loss": 0.1693,
      "step": 7051
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.410656178955726,
      "learning_rate": 4.207273162964274e-06,
      "loss": 0.1863,
      "step": 7052
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1858643107273845,
      "learning_rate": 4.205994073370545e-06,
      "loss": 0.1609,
      "step": 7053
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4130544533361011,
      "learning_rate": 4.2047150370808366e-06,
      "loss": 0.2121,
      "step": 7054
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2534465938727777,
      "learning_rate": 4.203436054181017e-06,
      "loss": 0.1495,
      "step": 7055
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4031607436177427,
      "learning_rate": 4.202157124756951e-06,
      "loss": 0.2121,
      "step": 7056
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3183160861657968,
      "learning_rate": 4.200878248894493e-06,
      "loss": 0.2265,
      "step": 7057
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4880532670727866,
      "learning_rate": 4.199599426679499e-06,
      "loss": 0.213,
      "step": 7058
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4103992586051797,
      "learning_rate": 4.198320658197821e-06,
      "loss": 0.212,
      "step": 7059
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4274076082759222,
      "learning_rate": 4.197041943535307e-06,
      "loss": 0.1871,
      "step": 7060
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4104829129971896,
      "learning_rate": 4.1957632827778e-06,
      "loss": 0.2057,
      "step": 7061
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2693657613305573,
      "learning_rate": 4.1944846760111395e-06,
      "loss": 0.2021,
      "step": 7062
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.41727602523015,
      "learning_rate": 4.1932061233211655e-06,
      "loss": 0.1916,
      "step": 7063
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4407651063867084,
      "learning_rate": 4.191927624793705e-06,
      "loss": 0.2206,
      "step": 7064
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4052147815000997,
      "learning_rate": 4.190649180514595e-06,
      "loss": 0.2218,
      "step": 7065
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.533223574273064,
      "learning_rate": 4.189370790569655e-06,
      "loss": 0.6424,
      "step": 7066
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2892464479098074,
      "learning_rate": 4.188092455044713e-06,
      "loss": 0.1586,
      "step": 7067
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.448601912956667,
      "learning_rate": 4.186814174025582e-06,
      "loss": 0.1728,
      "step": 7068
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4520893250336793,
      "learning_rate": 4.185535947598081e-06,
      "loss": 0.1886,
      "step": 7069
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2475585712879984,
      "learning_rate": 4.18425777584802e-06,
      "loss": 0.1564,
      "step": 7070
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.7682966964681,
      "learning_rate": 4.182979658861204e-06,
      "loss": 0.6638,
      "step": 7071
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4924274224976382,
      "learning_rate": 4.18170159672344e-06,
      "loss": 0.169,
      "step": 7072
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4082240650346431,
      "learning_rate": 4.180423589520526e-06,
      "loss": 0.1756,
      "step": 7073
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3158201895127204,
      "learning_rate": 4.179145637338262e-06,
      "loss": 0.1723,
      "step": 7074
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4436627337464247,
      "learning_rate": 4.177867740262437e-06,
      "loss": 0.2328,
      "step": 7075
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3763500700279003,
      "learning_rate": 4.176589898378843e-06,
      "loss": 0.1465,
      "step": 7076
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5612885961219398,
      "learning_rate": 4.175312111773261e-06,
      "loss": 0.2105,
      "step": 7077
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3852514398412912,
      "learning_rate": 4.1740343805314776e-06,
      "loss": 0.1951,
      "step": 7078
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.353553475531793,
      "learning_rate": 4.17275670473927e-06,
      "loss": 0.1679,
      "step": 7079
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.409360648745272,
      "learning_rate": 4.171479084482408e-06,
      "loss": 0.2005,
      "step": 7080
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1911695971613676,
      "learning_rate": 4.1702015198466675e-06,
      "loss": 0.1786,
      "step": 7081
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.9542908327631725,
      "learning_rate": 4.168924010917812e-06,
      "loss": 0.571,
      "step": 7082
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1497839165226071,
      "learning_rate": 4.167646557781608e-06,
      "loss": 0.1231,
      "step": 7083
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3351486249153444,
      "learning_rate": 4.166369160523811e-06,
      "loss": 0.1958,
      "step": 7084
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3848677872586626,
      "learning_rate": 4.165091819230178e-06,
      "loss": 0.2093,
      "step": 7085
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4968563556603347,
      "learning_rate": 4.16381453398646e-06,
      "loss": 0.2019,
      "step": 7086
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4847521307670821,
      "learning_rate": 4.162537304878408e-06,
      "loss": 0.185,
      "step": 7087
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3418719908094174,
      "learning_rate": 4.1612601319917635e-06,
      "loss": 0.1816,
      "step": 7088
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2478757878901254,
      "learning_rate": 4.159983015412266e-06,
      "loss": 0.1644,
      "step": 7089
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5087309612078421,
      "learning_rate": 4.1587059552256566e-06,
      "loss": 0.1914,
      "step": 7090
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3795125220343416,
      "learning_rate": 4.157428951517662e-06,
      "loss": 0.1999,
      "step": 7091
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3783564072433927,
      "learning_rate": 4.156152004374018e-06,
      "loss": 0.163,
      "step": 7092
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3975036601694872,
      "learning_rate": 4.154875113880444e-06,
      "loss": 0.1996,
      "step": 7093
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.226472394631826,
      "learning_rate": 4.1535982801226665e-06,
      "loss": 0.5928,
      "step": 7094
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.390939995390869,
      "learning_rate": 4.152321503186399e-06,
      "loss": 0.4105,
      "step": 7095
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.183662954790662,
      "learning_rate": 4.1510447831573585e-06,
      "loss": 0.1795,
      "step": 7096
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.269219912067945,
      "learning_rate": 4.149768120121255e-06,
      "loss": 0.1534,
      "step": 7097
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.244767844828059,
      "learning_rate": 4.148491514163791e-06,
      "loss": 0.1985,
      "step": 7098
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.430119963557811,
      "learning_rate": 4.147214965370674e-06,
      "loss": 0.1861,
      "step": 7099
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4142096833252733,
      "learning_rate": 4.145938473827598e-06,
      "loss": 0.1983,
      "step": 7100
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2768083412801121,
      "learning_rate": 4.144662039620263e-06,
      "loss": 0.173,
      "step": 7101
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3002099585512168,
      "learning_rate": 4.143385662834354e-06,
      "loss": 0.1818,
      "step": 7102
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2994922290986646,
      "learning_rate": 4.142109343555562e-06,
      "loss": 0.1636,
      "step": 7103
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2745952938425549,
      "learning_rate": 4.1408330818695685e-06,
      "loss": 0.1895,
      "step": 7104
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5745945557139085,
      "learning_rate": 4.139556877862055e-06,
      "loss": 0.24,
      "step": 7105
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5083353817236569,
      "learning_rate": 4.138280731618694e-06,
      "loss": 0.2418,
      "step": 7106
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3244041117760628,
      "learning_rate": 4.137004643225158e-06,
      "loss": 0.1869,
      "step": 7107
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4102245231239432,
      "learning_rate": 4.135728612767117e-06,
      "loss": 0.1776,
      "step": 7108
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2312525690469984,
      "learning_rate": 4.134452640330231e-06,
      "loss": 0.1826,
      "step": 7109
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3731248695167848,
      "learning_rate": 4.133176726000163e-06,
      "loss": 0.1952,
      "step": 7110
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2711694566215608,
      "learning_rate": 4.131900869862566e-06,
      "loss": 0.1777,
      "step": 7111
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3142604154496367,
      "learning_rate": 4.130625072003096e-06,
      "loss": 0.1797,
      "step": 7112
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.532066756912865,
      "learning_rate": 4.1293493325073975e-06,
      "loss": 0.6115,
      "step": 7113
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2076710368340655,
      "learning_rate": 4.128073651461116e-06,
      "loss": 0.1663,
      "step": 7114
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1908681142964177,
      "learning_rate": 4.126798028949894e-06,
      "loss": 0.1651,
      "step": 7115
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3945835314930477,
      "learning_rate": 4.1255224650593645e-06,
      "loss": 0.2007,
      "step": 7116
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4869429437352748,
      "learning_rate": 4.124246959875162e-06,
      "loss": 0.2196,
      "step": 7117
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3246297788794583,
      "learning_rate": 4.1229715134829135e-06,
      "loss": 0.1942,
      "step": 7118
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.141957665213491,
      "learning_rate": 4.121696125968247e-06,
      "loss": 0.1692,
      "step": 7119
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4959309780197632,
      "learning_rate": 4.120420797416777e-06,
      "loss": 0.1967,
      "step": 7120
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1784984704479655,
      "learning_rate": 4.119145527914127e-06,
      "loss": 0.1731,
      "step": 7121
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3004888118672955,
      "learning_rate": 4.1178703175459074e-06,
      "loss": 0.1843,
      "step": 7122
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1552866776476427,
      "learning_rate": 4.116595166397722e-06,
      "loss": 0.1719,
      "step": 7123
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2806797992837116,
      "learning_rate": 4.1153200745551835e-06,
      "loss": 0.1725,
      "step": 7124
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4054384204147619,
      "learning_rate": 4.1140450421038865e-06,
      "loss": 0.219,
      "step": 7125
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2787984463052422,
      "learning_rate": 4.112770069129431e-06,
      "loss": 0.183,
      "step": 7126
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.306066311441043,
      "learning_rate": 4.111495155717409e-06,
      "loss": 0.1734,
      "step": 7127
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2590250408511354,
      "learning_rate": 4.11022030195341e-06,
      "loss": 0.177,
      "step": 7128
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4417645248348994,
      "learning_rate": 4.108945507923017e-06,
      "loss": 0.1989,
      "step": 7129
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1198982616013609,
      "learning_rate": 4.107670773711812e-06,
      "loss": 0.1543,
      "step": 7130
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3014226841351553,
      "learning_rate": 4.106396099405373e-06,
      "loss": 0.1667,
      "step": 7131
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5295199405581843,
      "learning_rate": 4.1051214850892694e-06,
      "loss": 0.1713,
      "step": 7132
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.494564081114908,
      "learning_rate": 4.103846930849073e-06,
      "loss": 0.1962,
      "step": 7133
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4465061374527886,
      "learning_rate": 4.102572436770346e-06,
      "loss": 0.2053,
      "step": 7134
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3721712143392322,
      "learning_rate": 4.101298002938653e-06,
      "loss": 0.2523,
      "step": 7135
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2682682117900668,
      "learning_rate": 4.1000236294395455e-06,
      "loss": 0.2146,
      "step": 7136
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.270931477161842,
      "learning_rate": 4.0987493163585795e-06,
      "loss": 0.2101,
      "step": 7137
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2647219112463088,
      "learning_rate": 4.097475063781302e-06,
      "loss": 0.1659,
      "step": 7138
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.706494840582202,
      "learning_rate": 4.09620087179326e-06,
      "loss": 0.1803,
      "step": 7139
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2066537738883378,
      "learning_rate": 4.094926740479991e-06,
      "loss": 0.1763,
      "step": 7140
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3781641249871743,
      "learning_rate": 4.09365266992703e-06,
      "loss": 0.1967,
      "step": 7141
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.439782858536694,
      "learning_rate": 4.092378660219914e-06,
      "loss": 0.1703,
      "step": 7142
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.687432764947525,
      "learning_rate": 4.0911047114441675e-06,
      "loss": 0.5043,
      "step": 7143
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3210832283737193,
      "learning_rate": 4.089830823685316e-06,
      "loss": 0.2099,
      "step": 7144
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2916430816375275,
      "learning_rate": 4.088556997028878e-06,
      "loss": 0.1538,
      "step": 7145
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.343492408637802,
      "learning_rate": 4.087283231560371e-06,
      "loss": 0.1671,
      "step": 7146
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3766599620367632,
      "learning_rate": 4.086009527365306e-06,
      "loss": 0.1826,
      "step": 7147
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.40180586715717,
      "learning_rate": 4.084735884529191e-06,
      "loss": 0.1984,
      "step": 7148
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3478697507644612,
      "learning_rate": 4.08346230313753e-06,
      "loss": 0.1944,
      "step": 7149
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3804928364443272,
      "learning_rate": 4.08218878327582e-06,
      "loss": 0.1809,
      "step": 7150
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2019422840099405,
      "learning_rate": 4.080915325029559e-06,
      "loss": 0.177,
      "step": 7151
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3947795096219517,
      "learning_rate": 4.0796419284842355e-06,
      "loss": 0.1914,
      "step": 7152
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.759258089829613,
      "learning_rate": 4.078368593725339e-06,
      "loss": 0.574,
      "step": 7153
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5091467172058228,
      "learning_rate": 4.077095320838351e-06,
      "loss": 0.1784,
      "step": 7154
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7024569264346199,
      "learning_rate": 4.07582210990875e-06,
      "loss": 0.2042,
      "step": 7155
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5937428405681713,
      "learning_rate": 4.074548961022012e-06,
      "loss": 0.1941,
      "step": 7156
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2509277229114883,
      "learning_rate": 4.073275874263606e-06,
      "loss": 0.1531,
      "step": 7157
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3856597485476003,
      "learning_rate": 4.072002849718998e-06,
      "loss": 0.2308,
      "step": 7158
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3277721935626268,
      "learning_rate": 4.07072988747365e-06,
      "loss": 0.1697,
      "step": 7159
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.9030577571300284,
      "learning_rate": 4.069456987613022e-06,
      "loss": 0.5279,
      "step": 7160
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2070186202854654,
      "learning_rate": 4.068184150222564e-06,
      "loss": 0.1834,
      "step": 7161
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3225249978663933,
      "learning_rate": 4.066911375387728e-06,
      "loss": 0.1868,
      "step": 7162
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4078303787036004,
      "learning_rate": 4.065638663193957e-06,
      "loss": 0.1971,
      "step": 7163
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.472866445248734,
      "learning_rate": 4.064366013726695e-06,
      "loss": 0.1926,
      "step": 7164
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4199919415901312,
      "learning_rate": 4.063093427071376e-06,
      "loss": 0.2183,
      "step": 7165
    },
    {
      "epoch": 0.57,
      "grad_norm": 36.08454216215092,
      "learning_rate": 4.061820903313437e-06,
      "loss": 0.595,
      "step": 7166
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4274507379174581,
      "learning_rate": 4.060548442538301e-06,
      "loss": 0.2156,
      "step": 7167
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3184223209217552,
      "learning_rate": 4.059276044831394e-06,
      "loss": 0.2018,
      "step": 7168
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5204527125321803,
      "learning_rate": 4.0580037102781386e-06,
      "loss": 0.2179,
      "step": 7169
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3466677998238636,
      "learning_rate": 4.056731438963947e-06,
      "loss": 0.194,
      "step": 7170
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5423211441548683,
      "learning_rate": 4.055459230974232e-06,
      "loss": 0.2367,
      "step": 7171
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2106235090400705,
      "learning_rate": 4.0541870863944e-06,
      "loss": 0.1827,
      "step": 7172
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5400957963639583,
      "learning_rate": 4.0529150053098555e-06,
      "loss": 0.2008,
      "step": 7173
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2964735070603133,
      "learning_rate": 4.0516429878059955e-06,
      "loss": 0.2112,
      "step": 7174
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.230011040644138,
      "learning_rate": 4.050371033968216e-06,
      "loss": 0.1728,
      "step": 7175
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4800379876456038,
      "learning_rate": 4.049099143881907e-06,
      "loss": 0.1993,
      "step": 7176
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1961287070366364,
      "learning_rate": 4.047827317632452e-06,
      "loss": 0.2237,
      "step": 7177
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2225442732621181,
      "learning_rate": 4.0465555553052344e-06,
      "loss": 0.1735,
      "step": 7178
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.974183862850482,
      "learning_rate": 4.04528385698563e-06,
      "loss": 0.5976,
      "step": 7179
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4847414068981177,
      "learning_rate": 4.044012222759016e-06,
      "loss": 0.1917,
      "step": 7180
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2039028716992002,
      "learning_rate": 4.0427406527107554e-06,
      "loss": 0.1461,
      "step": 7181
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4164210169001747,
      "learning_rate": 4.041469146926216e-06,
      "loss": 0.1942,
      "step": 7182
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.314132299011415,
      "learning_rate": 4.0401977054907585e-06,
      "loss": 0.1785,
      "step": 7183
    },
    {
      "epoch": 0.57,
      "grad_norm": 16.712991586046304,
      "learning_rate": 4.038926328489735e-06,
      "loss": 0.5571,
      "step": 7184
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2727397348308978,
      "learning_rate": 4.0376550160085e-06,
      "loss": 0.1724,
      "step": 7185
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4778044335377742,
      "learning_rate": 4.0363837681323984e-06,
      "loss": 0.1785,
      "step": 7186
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5306149605175132,
      "learning_rate": 4.035112584946776e-06,
      "loss": 0.194,
      "step": 7187
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3353237263199496,
      "learning_rate": 4.033841466536968e-06,
      "loss": 0.2045,
      "step": 7188
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4915665948339425,
      "learning_rate": 4.0325704129883095e-06,
      "loss": 0.1889,
      "step": 7189
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3231406288773528,
      "learning_rate": 4.0312994243861304e-06,
      "loss": 0.179,
      "step": 7190
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2859822603240152,
      "learning_rate": 4.030028500815757e-06,
      "loss": 0.1545,
      "step": 7191
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1822528162203252,
      "learning_rate": 4.0287576423625095e-06,
      "loss": 0.1932,
      "step": 7192
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1917582617478444,
      "learning_rate": 4.027486849111702e-06,
      "loss": 0.1699,
      "step": 7193
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5079727696296428,
      "learning_rate": 4.026216121148652e-06,
      "loss": 0.2038,
      "step": 7194
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3085550863427062,
      "learning_rate": 4.024945458558661e-06,
      "loss": 0.1808,
      "step": 7195
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.470159106747156,
      "learning_rate": 4.023674861427038e-06,
      "loss": 0.2059,
      "step": 7196
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.367049874024757,
      "learning_rate": 4.022404329839077e-06,
      "loss": 0.209,
      "step": 7197
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4336436737776328,
      "learning_rate": 4.021133863880079e-06,
      "loss": 0.2264,
      "step": 7198
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5652711500456202,
      "learning_rate": 4.019863463635328e-06,
      "loss": 0.226,
      "step": 7199
    },
    {
      "epoch": 0.58,
      "grad_norm": 13.266973465291466,
      "learning_rate": 4.018593129190113e-06,
      "loss": 0.4666,
      "step": 7200
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.211890274683438,
      "learning_rate": 4.0173228606297165e-06,
      "loss": 0.4664,
      "step": 7201
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.29203987956892,
      "learning_rate": 4.016052658039411e-06,
      "loss": 0.1973,
      "step": 7202
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.412956216007654,
      "learning_rate": 4.014782521504473e-06,
      "loss": 0.172,
      "step": 7203
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1547722489529157,
      "learning_rate": 4.013512451110169e-06,
      "loss": 0.1448,
      "step": 7204
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3855161037256063,
      "learning_rate": 4.012242446941765e-06,
      "loss": 0.219,
      "step": 7205
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3583219849609947,
      "learning_rate": 4.010972509084514e-06,
      "loss": 0.1698,
      "step": 7206
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.701926580429876,
      "learning_rate": 4.00970263762368e-06,
      "loss": 0.6528,
      "step": 7207
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.257386964436387,
      "learning_rate": 4.008432832644505e-06,
      "loss": 0.754,
      "step": 7208
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2217689457375813,
      "learning_rate": 4.00716309423224e-06,
      "loss": 0.1953,
      "step": 7209
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3542655307330598,
      "learning_rate": 4.005893422472125e-06,
      "loss": 0.1663,
      "step": 7210
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3815808893498447,
      "learning_rate": 4.004623817449395e-06,
      "loss": 0.1818,
      "step": 7211
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.351319682743735,
      "learning_rate": 4.0033542792492855e-06,
      "loss": 0.2107,
      "step": 7212
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2566727843314711,
      "learning_rate": 4.002084807957021e-06,
      "loss": 0.1937,
      "step": 7213
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2135643715306499,
      "learning_rate": 4.0008154036578295e-06,
      "loss": 0.1648,
      "step": 7214
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2706846039476762,
      "learning_rate": 3.9995460664369254e-06,
      "loss": 0.1617,
      "step": 7215
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4484983551075723,
      "learning_rate": 3.998276796379526e-06,
      "loss": 0.2387,
      "step": 7216
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3907165236871228,
      "learning_rate": 3.997007593570839e-06,
      "loss": 0.2108,
      "step": 7217
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.494295581784815,
      "learning_rate": 3.995738458096074e-06,
      "loss": 0.5456,
      "step": 7218
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2814808353939966,
      "learning_rate": 3.994469390040428e-06,
      "loss": 0.2003,
      "step": 7219
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4836780479030092,
      "learning_rate": 3.993200389489096e-06,
      "loss": 0.2324,
      "step": 7220
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2133254981524315,
      "learning_rate": 3.991931456527276e-06,
      "loss": 0.1787,
      "step": 7221
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2024494787929823,
      "learning_rate": 3.990662591240148e-06,
      "loss": 0.1815,
      "step": 7222
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3462095973212893,
      "learning_rate": 3.989393793712901e-06,
      "loss": 0.2104,
      "step": 7223
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.245980330866405,
      "learning_rate": 3.9881250640307095e-06,
      "loss": 0.5851,
      "step": 7224
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.70998629330086,
      "learning_rate": 3.98685640227875e-06,
      "loss": 0.5983,
      "step": 7225
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3110779427085042,
      "learning_rate": 3.985587808542188e-06,
      "loss": 0.1839,
      "step": 7226
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3044678919267945,
      "learning_rate": 3.98431928290619e-06,
      "loss": 0.1689,
      "step": 7227
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.115639521634886,
      "learning_rate": 3.983050825455919e-06,
      "loss": 0.45,
      "step": 7228
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2136124632940601,
      "learning_rate": 3.981782436276523e-06,
      "loss": 0.1416,
      "step": 7229
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1596483271668667,
      "learning_rate": 3.98051411545316e-06,
      "loss": 0.1614,
      "step": 7230
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2485450499440605,
      "learning_rate": 3.979245863070971e-06,
      "loss": 0.1639,
      "step": 7231
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2677789447725334,
      "learning_rate": 3.977977679215103e-06,
      "loss": 0.1812,
      "step": 7232
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2780789794305005,
      "learning_rate": 3.9767095639706875e-06,
      "loss": 0.1652,
      "step": 7233
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.369279145262741,
      "learning_rate": 3.975441517422861e-06,
      "loss": 0.1715,
      "step": 7234
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.314966166574355,
      "learning_rate": 3.974173539656747e-06,
      "loss": 0.1867,
      "step": 7235
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.329478524343141,
      "learning_rate": 3.972905630757475e-06,
      "loss": 0.2423,
      "step": 7236
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5870198651267564,
      "learning_rate": 3.9716377908101585e-06,
      "loss": 0.2551,
      "step": 7237
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4593353090742065,
      "learning_rate": 3.9703700198999114e-06,
      "loss": 0.1982,
      "step": 7238
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1610521933801066,
      "learning_rate": 3.969102318111848e-06,
      "loss": 0.1674,
      "step": 7239
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3747922249590092,
      "learning_rate": 3.967834685531066e-06,
      "loss": 0.1909,
      "step": 7240
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2550429461544248,
      "learning_rate": 3.966567122242671e-06,
      "loss": 0.1551,
      "step": 7241
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.38659900582652,
      "learning_rate": 3.965299628331755e-06,
      "loss": 0.1892,
      "step": 7242
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3723044564517897,
      "learning_rate": 3.964032203883411e-06,
      "loss": 0.2046,
      "step": 7243
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1091939809357614,
      "learning_rate": 3.962764848982724e-06,
      "loss": 0.1493,
      "step": 7244
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2813879385188385,
      "learning_rate": 3.961497563714774e-06,
      "loss": 0.2185,
      "step": 7245
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3824616647924133,
      "learning_rate": 3.960230348164641e-06,
      "loss": 0.198,
      "step": 7246
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2879046942682892,
      "learning_rate": 3.958963202417393e-06,
      "loss": 0.1956,
      "step": 7247
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.211005352708064,
      "learning_rate": 3.9576961265581015e-06,
      "loss": 0.1974,
      "step": 7248
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3470318334291898,
      "learning_rate": 3.9564291206718256e-06,
      "loss": 0.1899,
      "step": 7249
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4477895981839048,
      "learning_rate": 3.955162184843625e-06,
      "loss": 0.1957,
      "step": 7250
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4987139836942087,
      "learning_rate": 3.953895319158552e-06,
      "loss": 0.1997,
      "step": 7251
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4833305313701641,
      "learning_rate": 3.952628523701658e-06,
      "loss": 0.2362,
      "step": 7252
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4129279648189654,
      "learning_rate": 3.9513617985579835e-06,
      "loss": 0.1699,
      "step": 7253
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5007501820886517,
      "learning_rate": 3.9500951438125675e-06,
      "loss": 0.2394,
      "step": 7254
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4333423660478632,
      "learning_rate": 3.948828559550448e-06,
      "loss": 0.1958,
      "step": 7255
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3374767923072521,
      "learning_rate": 3.94756204585665e-06,
      "loss": 0.1834,
      "step": 7256
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3712144727403748,
      "learning_rate": 3.9462956028162015e-06,
      "loss": 0.2029,
      "step": 7257
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3796075899174916,
      "learning_rate": 3.945029230514121e-06,
      "loss": 0.1987,
      "step": 7258
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3833419033480725,
      "learning_rate": 3.943762929035427e-06,
      "loss": 0.1609,
      "step": 7259
    },
    {
      "epoch": 0.58,
      "grad_norm": 16.861947351501957,
      "learning_rate": 3.942496698465125e-06,
      "loss": 0.4959,
      "step": 7260
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1904953967903564,
      "learning_rate": 3.941230538888225e-06,
      "loss": 0.2039,
      "step": 7261
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5791500606135864,
      "learning_rate": 3.939964450389728e-06,
      "loss": 0.1495,
      "step": 7262
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2502932460711642,
      "learning_rate": 3.9386984330546275e-06,
      "loss": 0.1989,
      "step": 7263
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3030336865069205,
      "learning_rate": 3.937432486967917e-06,
      "loss": 0.2172,
      "step": 7264
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.305623944447648,
      "learning_rate": 3.936166612214583e-06,
      "loss": 0.1938,
      "step": 7265
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5583041322604052,
      "learning_rate": 3.93490080887961e-06,
      "loss": 0.1913,
      "step": 7266
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.408221927955319,
      "learning_rate": 3.93363507704797e-06,
      "loss": 0.2101,
      "step": 7267
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.756777031566555,
      "learning_rate": 3.932369416804639e-06,
      "loss": 0.6343,
      "step": 7268
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2643221031309184,
      "learning_rate": 3.9311038282345835e-06,
      "loss": 0.1828,
      "step": 7269
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3947445559094458,
      "learning_rate": 3.9298383114227675e-06,
      "loss": 0.1716,
      "step": 7270
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.366569760486067,
      "learning_rate": 3.928572866454149e-06,
      "loss": 0.216,
      "step": 7271
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2084717731202264,
      "learning_rate": 3.927307493413678e-06,
      "loss": 0.1623,
      "step": 7272
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3309071308044624,
      "learning_rate": 3.926042192386307e-06,
      "loss": 0.1766,
      "step": 7273
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.96154928890973,
      "learning_rate": 3.924776963456976e-06,
      "loss": 0.4559,
      "step": 7274
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3408256639531484,
      "learning_rate": 3.9235118067106255e-06,
      "loss": 0.2199,
      "step": 7275
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2296060003888072,
      "learning_rate": 3.922246722232189e-06,
      "loss": 0.1989,
      "step": 7276
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3739498120185585,
      "learning_rate": 3.9209817101065976e-06,
      "loss": 0.1819,
      "step": 7277
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3116633482285325,
      "learning_rate": 3.919716770418771e-06,
      "loss": 0.2217,
      "step": 7278
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.264149681101779,
      "learning_rate": 3.918451903253632e-06,
      "loss": 0.1912,
      "step": 7279
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.505841838659166,
      "learning_rate": 3.917187108696094e-06,
      "loss": 0.5852,
      "step": 7280
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3175286290627164,
      "learning_rate": 3.9159223868310644e-06,
      "loss": 0.1959,
      "step": 7281
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.394973453483789,
      "learning_rate": 3.9146577377434504e-06,
      "loss": 0.2123,
      "step": 7282
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.737236978795701,
      "learning_rate": 3.91339316151815e-06,
      "loss": 0.7347,
      "step": 7283
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3913340592525023,
      "learning_rate": 3.912128658240062e-06,
      "loss": 0.1944,
      "step": 7284
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3486856455863978,
      "learning_rate": 3.91086422799407e-06,
      "loss": 0.1855,
      "step": 7285
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.332341132058153,
      "learning_rate": 3.909599870865064e-06,
      "loss": 0.2036,
      "step": 7286
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.408748693035325,
      "learning_rate": 3.908335586937921e-06,
      "loss": 0.1891,
      "step": 7287
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2564039424723448,
      "learning_rate": 3.90707137629752e-06,
      "loss": 0.2182,
      "step": 7288
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4108145844456303,
      "learning_rate": 3.905807239028728e-06,
      "loss": 0.197,
      "step": 7289
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2652573889140213,
      "learning_rate": 3.90454317521641e-06,
      "loss": 0.1799,
      "step": 7290
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5325672993235506,
      "learning_rate": 3.90327918494543e-06,
      "loss": 0.2133,
      "step": 7291
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2664930846929467,
      "learning_rate": 3.902015268300638e-06,
      "loss": 0.1434,
      "step": 7292
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.294479181218753,
      "learning_rate": 3.900751425366891e-06,
      "loss": 0.16,
      "step": 7293
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.014476079817677,
      "learning_rate": 3.899487656229029e-06,
      "loss": 0.5761,
      "step": 7294
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.190055259488617,
      "learning_rate": 3.8982239609718965e-06,
      "loss": 0.1642,
      "step": 7295
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4242098772805492,
      "learning_rate": 3.896960339680327e-06,
      "loss": 0.1639,
      "step": 7296
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2903622985436611,
      "learning_rate": 3.895696792439154e-06,
      "loss": 0.1747,
      "step": 7297
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2411493020242073,
      "learning_rate": 3.8944333193332e-06,
      "loss": 0.203,
      "step": 7298
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3927088272991854,
      "learning_rate": 3.893169920447285e-06,
      "loss": 0.1657,
      "step": 7299
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.689409650056911,
      "learning_rate": 3.89190659586623e-06,
      "loss": 0.5041,
      "step": 7300
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4352267375140668,
      "learning_rate": 3.89064334567484e-06,
      "loss": 0.1475,
      "step": 7301
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.44691062335425,
      "learning_rate": 3.889380169957925e-06,
      "loss": 0.2112,
      "step": 7302
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4681919586999757,
      "learning_rate": 3.888117068800283e-06,
      "loss": 0.2034,
      "step": 7303
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4238332640880818,
      "learning_rate": 3.886854042286713e-06,
      "loss": 0.2178,
      "step": 7304
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3541971856001938,
      "learning_rate": 3.885591090502003e-06,
      "loss": 0.2056,
      "step": 7305
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2478681334904405,
      "learning_rate": 3.884328213530938e-06,
      "loss": 0.1475,
      "step": 7306
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2430071161457696,
      "learning_rate": 3.883065411458303e-06,
      "loss": 0.1686,
      "step": 7307
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3380785198151155,
      "learning_rate": 3.881802684368869e-06,
      "loss": 0.1855,
      "step": 7308
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4204962082399126,
      "learning_rate": 3.88054003234741e-06,
      "loss": 0.2054,
      "step": 7309
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7171196870596477,
      "learning_rate": 3.879277455478689e-06,
      "loss": 0.1483,
      "step": 7310
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3653533886676135,
      "learning_rate": 3.8780149538474695e-06,
      "loss": 0.2143,
      "step": 7311
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2245971956851003,
      "learning_rate": 3.876752527538505e-06,
      "loss": 0.1511,
      "step": 7312
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.577989930108368,
      "learning_rate": 3.875490176636547e-06,
      "loss": 0.1944,
      "step": 7313
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.290433229405294,
      "learning_rate": 3.874227901226342e-06,
      "loss": 0.1689,
      "step": 7314
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2561466549410572,
      "learning_rate": 3.872965701392626e-06,
      "loss": 0.1672,
      "step": 7315
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.202917230697213,
      "learning_rate": 3.871703577220139e-06,
      "loss": 0.1635,
      "step": 7316
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.348856074094093,
      "learning_rate": 3.870441528793608e-06,
      "loss": 0.6248,
      "step": 7317
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5216629122170828,
      "learning_rate": 3.869179556197762e-06,
      "loss": 0.2223,
      "step": 7318
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1523428611296114,
      "learning_rate": 3.8679176595173165e-06,
      "loss": 0.1492,
      "step": 7319
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3462948490019122,
      "learning_rate": 3.8666558388369895e-06,
      "loss": 0.1923,
      "step": 7320
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5065131342707745,
      "learning_rate": 3.865394094241489e-06,
      "loss": 0.19,
      "step": 7321
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.722886375756484,
      "learning_rate": 3.864132425815523e-06,
      "loss": 0.5388,
      "step": 7322
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3577759439674237,
      "learning_rate": 3.862870833643787e-06,
      "loss": 0.1642,
      "step": 7323
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3350309504315643,
      "learning_rate": 3.861609317810977e-06,
      "loss": 0.1742,
      "step": 7324
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.499224511840807,
      "learning_rate": 3.8603478784017845e-06,
      "loss": 0.1968,
      "step": 7325
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2979463969605345,
      "learning_rate": 3.859086515500889e-06,
      "loss": 0.1676,
      "step": 7326
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2978126268074157,
      "learning_rate": 3.857825229192975e-06,
      "loss": 0.2114,
      "step": 7327
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2623104915215173,
      "learning_rate": 3.8565640195627124e-06,
      "loss": 0.1651,
      "step": 7328
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3179280639468434,
      "learning_rate": 3.8553028866947735e-06,
      "loss": 0.1801,
      "step": 7329
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5416835110817393,
      "learning_rate": 3.854041830673818e-06,
      "loss": 0.1971,
      "step": 7330
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.369397764878767,
      "learning_rate": 3.8527808515845085e-06,
      "loss": 0.1945,
      "step": 7331
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3705099918842498,
      "learning_rate": 3.851519949511497e-06,
      "loss": 0.1758,
      "step": 7332
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.20368987412326,
      "learning_rate": 3.8502591245394284e-06,
      "loss": 0.1402,
      "step": 7333
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.902551554561779,
      "learning_rate": 3.8489983767529496e-06,
      "loss": 0.6011,
      "step": 7334
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3342795731762285,
      "learning_rate": 3.847737706236696e-06,
      "loss": 0.1653,
      "step": 7335
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3971889778707622,
      "learning_rate": 3.846477113075303e-06,
      "loss": 0.2057,
      "step": 7336
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5417989066358895,
      "learning_rate": 3.845216597353395e-06,
      "loss": 0.1849,
      "step": 7337
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.239566196087453,
      "learning_rate": 3.843956159155598e-06,
      "loss": 0.184,
      "step": 7338
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.676868403445291,
      "learning_rate": 3.842695798566524e-06,
      "loss": 0.6097,
      "step": 7339
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1238695914318746,
      "learning_rate": 3.8414355156707895e-06,
      "loss": 0.1914,
      "step": 7340
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2619884640776078,
      "learning_rate": 3.8401753105529995e-06,
      "loss": 0.1773,
      "step": 7341
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3834536670447832,
      "learning_rate": 3.838915183297754e-06,
      "loss": 0.187,
      "step": 7342
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3599772940636474,
      "learning_rate": 3.83765513398965e-06,
      "loss": 0.1738,
      "step": 7343
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3168460736640262,
      "learning_rate": 3.836395162713277e-06,
      "loss": 0.1547,
      "step": 7344
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.174196725447432,
      "learning_rate": 3.835135269553226e-06,
      "loss": 0.1662,
      "step": 7345
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3547812151742071,
      "learning_rate": 3.833875454594071e-06,
      "loss": 0.2106,
      "step": 7346
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.319521086767589,
      "learning_rate": 3.832615717920391e-06,
      "loss": 0.2394,
      "step": 7347
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4349408488912931,
      "learning_rate": 3.831356059616754e-06,
      "loss": 0.1837,
      "step": 7348
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3923479895558382,
      "learning_rate": 3.830096479767727e-06,
      "loss": 0.2017,
      "step": 7349
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4188967395044612,
      "learning_rate": 3.828836978457868e-06,
      "loss": 0.1957,
      "step": 7350
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3807581340849508,
      "learning_rate": 3.82757755577173e-06,
      "loss": 0.197,
      "step": 7351
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4309835614507902,
      "learning_rate": 3.826318211793863e-06,
      "loss": 0.1679,
      "step": 7352
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3533229613754219,
      "learning_rate": 3.825058946608811e-06,
      "loss": 0.1831,
      "step": 7353
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1144253943891669,
      "learning_rate": 3.8237997603011114e-06,
      "loss": 0.168,
      "step": 7354
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3355510851959091,
      "learning_rate": 3.822540652955298e-06,
      "loss": 0.2068,
      "step": 7355
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2854360436540522,
      "learning_rate": 3.821281624655899e-06,
      "loss": 0.1712,
      "step": 7356
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3075091330380737,
      "learning_rate": 3.820022675487434e-06,
      "loss": 0.167,
      "step": 7357
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.438058917774604,
      "learning_rate": 3.818763805534423e-06,
      "loss": 0.2079,
      "step": 7358
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.344845578199535,
      "learning_rate": 3.817505014881378e-06,
      "loss": 0.1969,
      "step": 7359
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.685035028985734,
      "learning_rate": 3.816246303612802e-06,
      "loss": 0.4843,
      "step": 7360
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1494345867212967,
      "learning_rate": 3.8149876718132e-06,
      "loss": 0.1422,
      "step": 7361
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.1339793517121945,
      "learning_rate": 3.813729119567064e-06,
      "loss": 0.5433,
      "step": 7362
    },
    {
      "epoch": 0.59,
      "grad_norm": 23.71054043634761,
      "learning_rate": 3.812470646958889e-06,
      "loss": 0.6882,
      "step": 7363
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1904680559619152,
      "learning_rate": 3.8112122540731555e-06,
      "loss": 0.155,
      "step": 7364
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3882303923335082,
      "learning_rate": 3.809953940994346e-06,
      "loss": 0.1964,
      "step": 7365
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3482520761208174,
      "learning_rate": 3.8086957078069355e-06,
      "loss": 0.1459,
      "step": 7366
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.413423983941598,
      "learning_rate": 3.807437554595389e-06,
      "loss": 0.1841,
      "step": 7367
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.407147553834495,
      "learning_rate": 3.8061794814441744e-06,
      "loss": 0.2143,
      "step": 7368
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4219859576684297,
      "learning_rate": 3.8049214884377473e-06,
      "loss": 0.2003,
      "step": 7369
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2524907926269377,
      "learning_rate": 3.8036635756605635e-06,
      "loss": 0.1714,
      "step": 7370
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5840792263172339,
      "learning_rate": 3.8024057431970664e-06,
      "loss": 0.2093,
      "step": 7371
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4949971648879672,
      "learning_rate": 3.801147991131702e-06,
      "loss": 0.2439,
      "step": 7372
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3924912339448179,
      "learning_rate": 3.7998903195489036e-06,
      "loss": 0.1723,
      "step": 7373
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2453052012600203,
      "learning_rate": 3.7986327285331065e-06,
      "loss": 0.1556,
      "step": 7374
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3140227569702692,
      "learning_rate": 3.7973752181687336e-06,
      "loss": 0.1842,
      "step": 7375
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5165240450501503,
      "learning_rate": 3.7961177885402047e-06,
      "loss": 0.2034,
      "step": 7376
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4845952230114245,
      "learning_rate": 3.794860439731938e-06,
      "loss": 0.2301,
      "step": 7377
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4541924883692308,
      "learning_rate": 3.793603171828339e-06,
      "loss": 0.2159,
      "step": 7378
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3678090368279336,
      "learning_rate": 3.7923459849138173e-06,
      "loss": 0.1569,
      "step": 7379
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.344868895910062,
      "learning_rate": 3.791088879072766e-06,
      "loss": 0.1805,
      "step": 7380
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.252466757603435,
      "learning_rate": 3.789831854389583e-06,
      "loss": 0.1429,
      "step": 7381
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5925025446285608,
      "learning_rate": 3.788574910948653e-06,
      "loss": 0.1623,
      "step": 7382
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1187667842454694,
      "learning_rate": 3.7873180488343617e-06,
      "loss": 0.1623,
      "step": 7383
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.355446057563517,
      "learning_rate": 3.786061268131083e-06,
      "loss": 0.7082,
      "step": 7384
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.208466107096428,
      "learning_rate": 3.784804568923188e-06,
      "loss": 0.1551,
      "step": 7385
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3243283590524273,
      "learning_rate": 3.783547951295047e-06,
      "loss": 0.1916,
      "step": 7386
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.359905349145487,
      "learning_rate": 3.7822914153310155e-06,
      "loss": 0.1769,
      "step": 7387
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1345870616449818,
      "learning_rate": 3.781034961115452e-06,
      "loss": 0.1667,
      "step": 7388
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4260018473807656,
      "learning_rate": 3.779778588732703e-06,
      "loss": 0.2153,
      "step": 7389
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3788630135488815,
      "learning_rate": 3.778522298267117e-06,
      "loss": 0.2208,
      "step": 7390
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3096845616975608,
      "learning_rate": 3.777266089803028e-06,
      "loss": 0.1575,
      "step": 7391
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5467397213947347,
      "learning_rate": 3.776009963424772e-06,
      "loss": 0.2403,
      "step": 7392
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3938433913520791,
      "learning_rate": 3.774753919216677e-06,
      "loss": 0.1802,
      "step": 7393
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5138906224724487,
      "learning_rate": 3.7734979572630614e-06,
      "loss": 0.1883,
      "step": 7394
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3757047289707907,
      "learning_rate": 3.772242077648246e-06,
      "loss": 0.1912,
      "step": 7395
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4464650100877543,
      "learning_rate": 3.770986280456538e-06,
      "loss": 0.2076,
      "step": 7396
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2061700703240137,
      "learning_rate": 3.769730565772247e-06,
      "loss": 0.1624,
      "step": 7397
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.654950339576268,
      "learning_rate": 3.768474933679668e-06,
      "loss": 0.668,
      "step": 7398
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1532735947493509,
      "learning_rate": 3.7672193842631e-06,
      "loss": 0.1382,
      "step": 7399
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.074590478714253,
      "learning_rate": 3.7659639176068287e-06,
      "loss": 0.6437,
      "step": 7400
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2989898142938618,
      "learning_rate": 3.7647085337951404e-06,
      "loss": 0.1666,
      "step": 7401
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.69321618047491,
      "learning_rate": 3.76345323291231e-06,
      "loss": 0.5837,
      "step": 7402
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2830949680657486,
      "learning_rate": 3.7621980150426103e-06,
      "loss": 0.1912,
      "step": 7403
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3350723415202501,
      "learning_rate": 3.7609428802703098e-06,
      "loss": 0.1945,
      "step": 7404
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4555069450911204,
      "learning_rate": 3.7596878286796657e-06,
      "loss": 0.2106,
      "step": 7405
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.195476203237622,
      "learning_rate": 3.758432860354937e-06,
      "loss": 0.4393,
      "step": 7406
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3189697340255864,
      "learning_rate": 3.7571779753803704e-06,
      "loss": 0.1553,
      "step": 7407
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.306025697135167,
      "learning_rate": 3.7559231738402148e-06,
      "loss": 0.5819,
      "step": 7408
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.041457264706336,
      "learning_rate": 3.7546684558187033e-06,
      "loss": 0.624,
      "step": 7409
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.142283640186236,
      "learning_rate": 3.753413821400073e-06,
      "loss": 0.5308,
      "step": 7410
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3305192291163936,
      "learning_rate": 3.752159270668551e-06,
      "loss": 0.1655,
      "step": 7411
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.222602004316321,
      "learning_rate": 3.7509048037083556e-06,
      "loss": 0.522,
      "step": 7412
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3644494747539337,
      "learning_rate": 3.7496504206037066e-06,
      "loss": 0.2146,
      "step": 7413
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3802943046770164,
      "learning_rate": 3.748396121438812e-06,
      "loss": 0.1972,
      "step": 7414
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3198448301625334,
      "learning_rate": 3.74714190629788e-06,
      "loss": 0.1576,
      "step": 7415
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3464307843040537,
      "learning_rate": 3.7458877752651063e-06,
      "loss": 0.1763,
      "step": 7416
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5295505680178565,
      "learning_rate": 3.744633728424687e-06,
      "loss": 0.1921,
      "step": 7417
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4924802414110787,
      "learning_rate": 3.743379765860807e-06,
      "loss": 0.2009,
      "step": 7418
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.711246508724797,
      "learning_rate": 3.7421258876576543e-06,
      "loss": 0.5653,
      "step": 7419
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3744690951190388,
      "learning_rate": 3.7408720938994003e-06,
      "loss": 0.1861,
      "step": 7420
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4289181650334266,
      "learning_rate": 3.7396183846702164e-06,
      "loss": 0.2224,
      "step": 7421
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2537257287751429,
      "learning_rate": 3.7383647600542704e-06,
      "loss": 0.1683,
      "step": 7422
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6447995244516012,
      "learning_rate": 3.7371112201357195e-06,
      "loss": 0.2004,
      "step": 7423
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4164785605311527,
      "learning_rate": 3.735857764998721e-06,
      "loss": 0.2419,
      "step": 7424
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3062254200506616,
      "learning_rate": 3.734604394727419e-06,
      "loss": 0.1567,
      "step": 7425
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2216825637783326,
      "learning_rate": 3.733351109405959e-06,
      "loss": 0.1904,
      "step": 7426
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4614287469241358,
      "learning_rate": 3.732097909118477e-06,
      "loss": 0.1818,
      "step": 7427
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3939627781385393,
      "learning_rate": 3.730844793949103e-06,
      "loss": 0.2065,
      "step": 7428
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3809734065067316,
      "learning_rate": 3.729591763981964e-06,
      "loss": 0.1821,
      "step": 7429
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.386682358518999,
      "learning_rate": 3.7283388193011776e-06,
      "loss": 0.2017,
      "step": 7430
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.650003644111188,
      "learning_rate": 3.727085959990862e-06,
      "loss": 0.572,
      "step": 7431
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.439480014225976,
      "learning_rate": 3.7258331861351203e-06,
      "loss": 0.1981,
      "step": 7432
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3096835472161665,
      "learning_rate": 3.7245804978180582e-06,
      "loss": 0.146,
      "step": 7433
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.238783004698527,
      "learning_rate": 3.7233278951237706e-06,
      "loss": 0.5511,
      "step": 7434
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4667865170044283,
      "learning_rate": 3.722075378136352e-06,
      "loss": 0.2056,
      "step": 7435
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2024470881132698,
      "learning_rate": 3.7208229469398843e-06,
      "loss": 0.1463,
      "step": 7436
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5640815794616751,
      "learning_rate": 3.719570601618446e-06,
      "loss": 0.1833,
      "step": 7437
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.341529092575993,
      "learning_rate": 3.7183183422561164e-06,
      "loss": 0.182,
      "step": 7438
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4003543930946474,
      "learning_rate": 3.717066168936957e-06,
      "loss": 0.223,
      "step": 7439
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5641772603681234,
      "learning_rate": 3.7158140817450335e-06,
      "loss": 0.1789,
      "step": 7440
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3208516158449901,
      "learning_rate": 3.7145620807644007e-06,
      "loss": 0.2067,
      "step": 7441
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3728452818160795,
      "learning_rate": 3.713310166079112e-06,
      "loss": 0.1856,
      "step": 7442
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.367049377524347,
      "learning_rate": 3.712058337773209e-06,
      "loss": 0.189,
      "step": 7443
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.27522340932222,
      "learning_rate": 3.7108065959307325e-06,
      "loss": 0.1835,
      "step": 7444
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3062382342742045,
      "learning_rate": 3.709554940635717e-06,
      "loss": 0.1396,
      "step": 7445
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4265092881339705,
      "learning_rate": 3.7083033719721852e-06,
      "loss": 0.2247,
      "step": 7446
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3336420016664583,
      "learning_rate": 3.7070518900241635e-06,
      "loss": 0.1604,
      "step": 7447
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2980302367673076,
      "learning_rate": 3.7058004948756644e-06,
      "loss": 0.1848,
      "step": 7448
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.471112017561321,
      "learning_rate": 3.704549186610702e-06,
      "loss": 0.1712,
      "step": 7449
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5335352648919394,
      "learning_rate": 3.703297965313275e-06,
      "loss": 0.1892,
      "step": 7450
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3888886623159546,
      "learning_rate": 3.702046831067386e-06,
      "loss": 0.1799,
      "step": 7451
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3461049545194068,
      "learning_rate": 3.7007957839570247e-06,
      "loss": 0.2082,
      "step": 7452
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3408958982430474,
      "learning_rate": 3.6995448240661807e-06,
      "loss": 0.1557,
      "step": 7453
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1371965313493715,
      "learning_rate": 3.698293951478832e-06,
      "loss": 0.1522,
      "step": 7454
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3508594150597606,
      "learning_rate": 3.6970431662789534e-06,
      "loss": 0.1603,
      "step": 7455
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.316035061808714,
      "learning_rate": 3.695792468550517e-06,
      "loss": 0.1589,
      "step": 7456
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.393082655964855,
      "learning_rate": 3.694541858377482e-06,
      "loss": 0.2224,
      "step": 7457
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2701291626823512,
      "learning_rate": 3.6932913358438073e-06,
      "loss": 0.1548,
      "step": 7458
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.247004888546274,
      "learning_rate": 3.6920409010334447e-06,
      "loss": 0.162,
      "step": 7459
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3799562516138077,
      "learning_rate": 3.69079055403034e-06,
      "loss": 0.193,
      "step": 7460
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1650803040967792,
      "learning_rate": 3.68954029491843e-06,
      "loss": 0.1743,
      "step": 7461
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3596883535007305,
      "learning_rate": 3.6882901237816536e-06,
      "loss": 0.1979,
      "step": 7462
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3471113064670701,
      "learning_rate": 3.687040040703936e-06,
      "loss": 0.1731,
      "step": 7463
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.119143016073538,
      "learning_rate": 3.6857900457691946e-06,
      "loss": 0.5809,
      "step": 7464
    },
    {
      "epoch": 0.6,
      "grad_norm": 11.528103008599109,
      "learning_rate": 3.6845401390613528e-06,
      "loss": 0.5254,
      "step": 7465
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.178810788826177,
      "learning_rate": 3.6832903206643152e-06,
      "loss": 0.1495,
      "step": 7466
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.438829844479788,
      "learning_rate": 3.682040590661988e-06,
      "loss": 0.5242,
      "step": 7467
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.444508332531294,
      "learning_rate": 3.6807909491382697e-06,
      "loss": 0.1668,
      "step": 7468
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.228405633222191,
      "learning_rate": 3.679541396177053e-06,
      "loss": 0.168,
      "step": 7469
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3985970841179658,
      "learning_rate": 3.678291931862221e-06,
      "loss": 0.2054,
      "step": 7470
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3779728182027506,
      "learning_rate": 3.6770425562776577e-06,
      "loss": 0.1752,
      "step": 7471
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2267918382770493,
      "learning_rate": 3.6757932695072373e-06,
      "loss": 0.189,
      "step": 7472
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3988149699461732,
      "learning_rate": 3.6745440716348245e-06,
      "loss": 0.1723,
      "step": 7473
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2327734782981703,
      "learning_rate": 3.6732949627442855e-06,
      "loss": 0.1426,
      "step": 7474
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2593554544320047,
      "learning_rate": 3.6720459429194743e-06,
      "loss": 0.1903,
      "step": 7475
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.498535192534434,
      "learning_rate": 3.670797012244244e-06,
      "loss": 0.1702,
      "step": 7476
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5066249917509211,
      "learning_rate": 3.669548170802436e-06,
      "loss": 0.1875,
      "step": 7477
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5060216568663658,
      "learning_rate": 3.6682994186778914e-06,
      "loss": 0.2372,
      "step": 7478
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2209663220825024,
      "learning_rate": 3.6670507559544403e-06,
      "loss": 0.2133,
      "step": 7479
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.225800617361338,
      "learning_rate": 3.665802182715913e-06,
      "loss": 0.1656,
      "step": 7480
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1811440149872499,
      "learning_rate": 3.664553699046127e-06,
      "loss": 0.1341,
      "step": 7481
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.159204200656736,
      "learning_rate": 3.663305305028896e-06,
      "loss": 0.1612,
      "step": 7482
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1319149802103186,
      "learning_rate": 3.6620570007480317e-06,
      "loss": 0.1545,
      "step": 7483
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5053936472723846,
      "learning_rate": 3.660808786287333e-06,
      "loss": 0.1985,
      "step": 7484
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.716754270459973,
      "learning_rate": 3.659560661730599e-06,
      "loss": 0.5565,
      "step": 7485
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2617823518684466,
      "learning_rate": 3.658312627161618e-06,
      "loss": 0.1621,
      "step": 7486
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.3171838514049075,
      "learning_rate": 3.6570646826641777e-06,
      "loss": 0.5938,
      "step": 7487
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4781055033004342,
      "learning_rate": 3.6558168283220536e-06,
      "loss": 0.1859,
      "step": 7488
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1345368734666488,
      "learning_rate": 3.6545690642190166e-06,
      "loss": 0.1717,
      "step": 7489
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2550985424396772,
      "learning_rate": 3.6533213904388377e-06,
      "loss": 0.207,
      "step": 7490
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.165873514046666,
      "learning_rate": 3.6520738070652716e-06,
      "loss": 0.5318,
      "step": 7491
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2936108174242225,
      "learning_rate": 3.650826314182076e-06,
      "loss": 0.2,
      "step": 7492
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3209209801809625,
      "learning_rate": 3.6495789118729966e-06,
      "loss": 0.1961,
      "step": 7493
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.182927659862257,
      "learning_rate": 3.6483316002217788e-06,
      "loss": 0.1313,
      "step": 7494
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.055882311002078,
      "learning_rate": 3.6470843793121536e-06,
      "loss": 0.544,
      "step": 7495
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.24321505104082,
      "learning_rate": 3.645837249227855e-06,
      "loss": 0.1649,
      "step": 7496
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3793996212009143,
      "learning_rate": 3.6445902100526055e-06,
      "loss": 0.1942,
      "step": 7497
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5282592808581068,
      "learning_rate": 3.6433432618701193e-06,
      "loss": 0.2232,
      "step": 7498
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3766589611361668,
      "learning_rate": 3.6420964047641117e-06,
      "loss": 0.1794,
      "step": 7499
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.9885020315649236,
      "learning_rate": 3.6408496388182857e-06,
      "loss": 0.5112,
      "step": 7500
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4073188779139203,
      "learning_rate": 3.6396029641163442e-06,
      "loss": 0.1957,
      "step": 7501
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3374305074074107,
      "learning_rate": 3.6383563807419754e-06,
      "loss": 0.1633,
      "step": 7502
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3508366168139432,
      "learning_rate": 3.6371098887788686e-06,
      "loss": 0.2046,
      "step": 7503
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.27426675678328,
      "learning_rate": 3.6358634883107034e-06,
      "loss": 0.192,
      "step": 7504
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.393696764505489,
      "learning_rate": 3.634617179421157e-06,
      "loss": 0.1785,
      "step": 7505
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4690214137560773,
      "learning_rate": 3.6333709621938974e-06,
      "loss": 0.2088,
      "step": 7506
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.187533976740001,
      "learning_rate": 3.632124836712583e-06,
      "loss": 0.5973,
      "step": 7507
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4211628709665354,
      "learning_rate": 3.6308788030608743e-06,
      "loss": 0.164,
      "step": 7508
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2812885848728743,
      "learning_rate": 3.629632861322419e-06,
      "loss": 0.1788,
      "step": 7509
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3052675154983016,
      "learning_rate": 3.628387011580864e-06,
      "loss": 0.1882,
      "step": 7510
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4031165308666953,
      "learning_rate": 3.6271412539198426e-06,
      "loss": 0.2241,
      "step": 7511
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.185697503275086,
      "learning_rate": 3.6258955884229883e-06,
      "loss": 0.159,
      "step": 7512
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4927268336185662,
      "learning_rate": 3.624650015173927e-06,
      "loss": 0.2578,
      "step": 7513
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5207832070740166,
      "learning_rate": 3.6234045342562783e-06,
      "loss": 0.17,
      "step": 7514
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.878797628230829,
      "learning_rate": 3.622159145753654e-06,
      "loss": 0.4862,
      "step": 7515
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.566785270264143,
      "learning_rate": 3.6209138497496594e-06,
      "loss": 0.556,
      "step": 7516
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3556702769874027,
      "learning_rate": 3.6196686463278985e-06,
      "loss": 0.1486,
      "step": 7517
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2834144008357922,
      "learning_rate": 3.6184235355719623e-06,
      "loss": 0.1589,
      "step": 7518
    },
    {
      "epoch": 0.6,
      "grad_norm": 11.995637246160584,
      "learning_rate": 3.61717851756544e-06,
      "loss": 0.6514,
      "step": 7519
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4698049443010799,
      "learning_rate": 3.615933592391913e-06,
      "loss": 0.1884,
      "step": 7520
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.337639983835457,
      "learning_rate": 3.61468876013496e-06,
      "loss": 0.6289,
      "step": 7521
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1915264872383433,
      "learning_rate": 3.6134440208781453e-06,
      "loss": 0.1589,
      "step": 7522
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4176072280633316,
      "learning_rate": 3.612199374705036e-06,
      "loss": 0.1904,
      "step": 7523
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2780457789704345,
      "learning_rate": 3.6109548216991886e-06,
      "loss": 0.1827,
      "step": 7524
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3048692375746669,
      "learning_rate": 3.6097103619441505e-06,
      "loss": 0.1895,
      "step": 7525
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2157856552469801,
      "learning_rate": 3.608465995523469e-06,
      "loss": 0.1515,
      "step": 7526
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2854898774742747,
      "learning_rate": 3.6072217225206807e-06,
      "loss": 0.1454,
      "step": 7527
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5673726047714536,
      "learning_rate": 3.6059775430193202e-06,
      "loss": 0.2069,
      "step": 7528
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4160033912894239,
      "learning_rate": 3.604733457102908e-06,
      "loss": 0.2027,
      "step": 7529
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.210170009294503,
      "learning_rate": 3.6034894648549677e-06,
      "loss": 0.1859,
      "step": 7530
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3530422447115213,
      "learning_rate": 3.6022455663590106e-06,
      "loss": 0.1748,
      "step": 7531
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3276246144575294,
      "learning_rate": 3.6010017616985454e-06,
      "loss": 0.1605,
      "step": 7532
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.290852611777824,
      "learning_rate": 3.5997580509570696e-06,
      "loss": 0.1975,
      "step": 7533
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2203583241435583,
      "learning_rate": 3.598514434218078e-06,
      "loss": 0.1772,
      "step": 7534
    },
    {
      "epoch": 0.6,
      "grad_norm": 14.385052275967093,
      "learning_rate": 3.5972709115650594e-06,
      "loss": 0.6447,
      "step": 7535
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.132578278136999,
      "learning_rate": 3.5960274830814933e-06,
      "loss": 0.6043,
      "step": 7536
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3429669488483478,
      "learning_rate": 3.5947841488508573e-06,
      "loss": 0.1597,
      "step": 7537
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.586767056217181,
      "learning_rate": 3.593540908956618e-06,
      "loss": 0.2388,
      "step": 7538
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4365577172555777,
      "learning_rate": 3.5922977634822408e-06,
      "loss": 0.221,
      "step": 7539
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5278347668007461,
      "learning_rate": 3.5910547125111783e-06,
      "loss": 0.2072,
      "step": 7540
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4105822989562757,
      "learning_rate": 3.589811756126882e-06,
      "loss": 0.222,
      "step": 7541
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4175948042221154,
      "learning_rate": 3.588568894412797e-06,
      "loss": 0.194,
      "step": 7542
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.481420494634635,
      "learning_rate": 3.587326127452356e-06,
      "loss": 0.5376,
      "step": 7543
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.400464888990005,
      "learning_rate": 3.5860834553289936e-06,
      "loss": 0.1888,
      "step": 7544
    },
    {
      "epoch": 0.6,
      "grad_norm": 26.48533235470663,
      "learning_rate": 3.5848408781261323e-06,
      "loss": 0.6311,
      "step": 7545
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.245566176923873,
      "learning_rate": 3.583598395927192e-06,
      "loss": 0.1525,
      "step": 7546
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5548970456730904,
      "learning_rate": 3.5823560088155795e-06,
      "loss": 0.2024,
      "step": 7547
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7874123140806166,
      "learning_rate": 3.5811137168747077e-06,
      "loss": 0.2004,
      "step": 7548
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2231374061988407,
      "learning_rate": 3.5798715201879704e-06,
      "loss": 0.2022,
      "step": 7549
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.276897209985682,
      "learning_rate": 3.578629418838757e-06,
      "loss": 0.1991,
      "step": 7550
    },
    {
      "epoch": 0.6,
      "grad_norm": 15.977103607319526,
      "learning_rate": 3.5773874129104612e-06,
      "loss": 0.5714,
      "step": 7551
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3115728722161808,
      "learning_rate": 3.5761455024864566e-06,
      "loss": 0.1794,
      "step": 7552
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.352057917511193,
      "learning_rate": 3.5749036876501196e-06,
      "loss": 0.4839,
      "step": 7553
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3934247326858136,
      "learning_rate": 3.573661968484815e-06,
      "loss": 0.1907,
      "step": 7554
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2854035859297976,
      "learning_rate": 3.572420345073906e-06,
      "loss": 0.2049,
      "step": 7555
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4767036692536906,
      "learning_rate": 3.571178817500742e-06,
      "loss": 0.21,
      "step": 7556
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.326700707158939,
      "learning_rate": 3.5699373858486752e-06,
      "loss": 0.1847,
      "step": 7557
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3569676489826743,
      "learning_rate": 3.5686960502010455e-06,
      "loss": 0.2228,
      "step": 7558
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2272979979778909,
      "learning_rate": 3.5674548106411844e-06,
      "loss": 0.1724,
      "step": 7559
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4133170140252964,
      "learning_rate": 3.566213667252423e-06,
      "loss": 0.217,
      "step": 7560
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3659650901587805,
      "learning_rate": 3.564972620118081e-06,
      "loss": 0.1844,
      "step": 7561
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3379080507717929,
      "learning_rate": 3.5637316693214772e-06,
      "loss": 0.1761,
      "step": 7562
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2400290425687925,
      "learning_rate": 3.562490814945917e-06,
      "loss": 0.1847,
      "step": 7563
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3477721264104463,
      "learning_rate": 3.5612500570747037e-06,
      "loss": 0.1973,
      "step": 7564
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3367138246389696,
      "learning_rate": 3.560009395791133e-06,
      "loss": 0.1787,
      "step": 7565
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2800948325263146,
      "learning_rate": 3.5587688311784963e-06,
      "loss": 0.1529,
      "step": 7566
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3257861510925042,
      "learning_rate": 3.557528363320074e-06,
      "loss": 0.1444,
      "step": 7567
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4869956882389639,
      "learning_rate": 3.556287992299142e-06,
      "loss": 0.1905,
      "step": 7568
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3946942671802685,
      "learning_rate": 3.5550477181989735e-06,
      "loss": 0.167,
      "step": 7569
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5693927004355817,
      "learning_rate": 3.553807541102827e-06,
      "loss": 0.1942,
      "step": 7570
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2585721562428815,
      "learning_rate": 3.552567461093964e-06,
      "loss": 0.1708,
      "step": 7571
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.649281630139912,
      "learning_rate": 3.5513274782556317e-06,
      "loss": 0.6914,
      "step": 7572
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3754912565916209,
      "learning_rate": 3.5500875926710767e-06,
      "loss": 0.1753,
      "step": 7573
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3172062692412108,
      "learning_rate": 3.548847804423533e-06,
      "loss": 0.2079,
      "step": 7574
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.289101228007214,
      "learning_rate": 3.5476081135962335e-06,
      "loss": 0.1724,
      "step": 7575
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3324086064607157,
      "learning_rate": 3.546368520272404e-06,
      "loss": 0.1579,
      "step": 7576
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.338356358261095,
      "learning_rate": 3.545129024535257e-06,
      "loss": 0.165,
      "step": 7577
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4686927822429694,
      "learning_rate": 3.5438896264680076e-06,
      "loss": 0.2159,
      "step": 7578
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3205479928749009,
      "learning_rate": 3.5426503261538593e-06,
      "loss": 0.179,
      "step": 7579
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2585725435464314,
      "learning_rate": 3.541411123676012e-06,
      "loss": 0.1817,
      "step": 7580
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3819962298236277,
      "learning_rate": 3.5401720191176527e-06,
      "loss": 0.1539,
      "step": 7581
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3453113246727373,
      "learning_rate": 3.5389330125619704e-06,
      "loss": 0.195,
      "step": 7582
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5013823222163911,
      "learning_rate": 3.5376941040921405e-06,
      "loss": 0.2354,
      "step": 7583
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.166024823687403,
      "learning_rate": 3.536455293791338e-06,
      "loss": 0.5595,
      "step": 7584
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.423666160410486,
      "learning_rate": 3.5352165817427255e-06,
      "loss": 0.1924,
      "step": 7585
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.262063551825697,
      "learning_rate": 3.5339779680294607e-06,
      "loss": 0.1449,
      "step": 7586
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4286578508004075,
      "learning_rate": 3.532739452734698e-06,
      "loss": 0.1837,
      "step": 7587
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6105203598287376,
      "learning_rate": 3.53150103594158e-06,
      "loss": 0.1903,
      "step": 7588
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.324396296583796,
      "learning_rate": 3.530262717733248e-06,
      "loss": 0.19,
      "step": 7589
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1603638911680982,
      "learning_rate": 3.5290244981928323e-06,
      "loss": 0.1544,
      "step": 7590
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.245711260333263,
      "learning_rate": 3.527786377403459e-06,
      "loss": 0.1593,
      "step": 7591
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5162537477604785,
      "learning_rate": 3.5265483554482467e-06,
      "loss": 0.181,
      "step": 7592
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3329470001493515,
      "learning_rate": 3.525310432410309e-06,
      "loss": 0.1617,
      "step": 7593
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4462918183988256,
      "learning_rate": 3.5240726083727496e-06,
      "loss": 0.1754,
      "step": 7594
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2569823386119239,
      "learning_rate": 3.5228348834186663e-06,
      "loss": 0.1661,
      "step": 7595
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4367291315088795,
      "learning_rate": 3.521597257631155e-06,
      "loss": 0.2202,
      "step": 7596
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3726751461783724,
      "learning_rate": 3.520359731093298e-06,
      "loss": 0.1998,
      "step": 7597
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3912977858240747,
      "learning_rate": 3.519122303888176e-06,
      "loss": 0.2047,
      "step": 7598
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4015214203826476,
      "learning_rate": 3.5178849760988593e-06,
      "loss": 0.1694,
      "step": 7599
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3991201624490694,
      "learning_rate": 3.516647747808417e-06,
      "loss": 0.2459,
      "step": 7600
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2774890632509242,
      "learning_rate": 3.5154106190999036e-06,
      "loss": 0.1465,
      "step": 7601
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3324939861498035,
      "learning_rate": 3.5141735900563746e-06,
      "loss": 0.1719,
      "step": 7602
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3542774066196974,
      "learning_rate": 3.512936660760875e-06,
      "loss": 0.1955,
      "step": 7603
    },
    {
      "epoch": 0.61,
      "grad_norm": 11.307975947363177,
      "learning_rate": 3.5116998312964413e-06,
      "loss": 0.659,
      "step": 7604
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2935988722347063,
      "learning_rate": 3.510463101746109e-06,
      "loss": 0.2074,
      "step": 7605
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.266181534546647,
      "learning_rate": 3.509226472192899e-06,
      "loss": 0.1909,
      "step": 7606
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.2111987523200485,
      "learning_rate": 3.507989942719837e-06,
      "loss": 0.6301,
      "step": 7607
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.178419033978148,
      "learning_rate": 3.5067535134099274e-06,
      "loss": 0.6075,
      "step": 7608
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3089199409041692,
      "learning_rate": 3.50551718434618e-06,
      "loss": 0.175,
      "step": 7609
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.833102505061196,
      "learning_rate": 3.504280955611593e-06,
      "loss": 0.54,
      "step": 7610
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.363730638232593,
      "learning_rate": 3.5030448272891553e-06,
      "loss": 0.1575,
      "step": 7611
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4370250673821596,
      "learning_rate": 3.5018087994618556e-06,
      "loss": 0.2055,
      "step": 7612
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4068002437763976,
      "learning_rate": 3.5005728722126686e-06,
      "loss": 0.1842,
      "step": 7613
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3511516319963588,
      "learning_rate": 3.49933704562457e-06,
      "loss": 0.2344,
      "step": 7614
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.962411198775959,
      "learning_rate": 3.4981013197805208e-06,
      "loss": 0.3796,
      "step": 7615
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5168025411515509,
      "learning_rate": 3.4968656947634814e-06,
      "loss": 0.1945,
      "step": 7616
    },
    {
      "epoch": 0.61,
      "grad_norm": 13.672894166039077,
      "learning_rate": 3.495630170656401e-06,
      "loss": 0.6423,
      "step": 7617
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4326468776112502,
      "learning_rate": 3.4943947475422278e-06,
      "loss": 0.1993,
      "step": 7618
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3149488398469633,
      "learning_rate": 3.493159425503896e-06,
      "loss": 0.2069,
      "step": 7619
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4174205893387537,
      "learning_rate": 3.491924204624336e-06,
      "loss": 0.1712,
      "step": 7620
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3653752078130281,
      "learning_rate": 3.490689084986475e-06,
      "loss": 0.1848,
      "step": 7621
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.93548743011966,
      "learning_rate": 3.489454066673228e-06,
      "loss": 0.5816,
      "step": 7622
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.159499819188324,
      "learning_rate": 3.4882191497675065e-06,
      "loss": 0.1582,
      "step": 7623
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2767827020845832,
      "learning_rate": 3.4869843343522123e-06,
      "loss": 0.1686,
      "step": 7624
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.37048136204285,
      "learning_rate": 3.4857496205102475e-06,
      "loss": 0.209,
      "step": 7625
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3891180972239856,
      "learning_rate": 3.4845150083244963e-06,
      "loss": 0.2104,
      "step": 7626
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2515538765774992,
      "learning_rate": 3.4832804978778447e-06,
      "loss": 0.183,
      "step": 7627
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3235646726261303,
      "learning_rate": 3.4820460892531706e-06,
      "loss": 0.1473,
      "step": 7628
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3708491025148604,
      "learning_rate": 3.480811782533339e-06,
      "loss": 0.2017,
      "step": 7629
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.273397435985446,
      "learning_rate": 3.4795775778012165e-06,
      "loss": 0.1494,
      "step": 7630
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.921175869396291,
      "learning_rate": 3.4783434751396567e-06,
      "loss": 0.5103,
      "step": 7631
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2883991178882805,
      "learning_rate": 3.4771094746315126e-06,
      "loss": 0.1798,
      "step": 7632
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3906837367068652,
      "learning_rate": 3.4758755763596197e-06,
      "loss": 0.1867,
      "step": 7633
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5226721301884454,
      "learning_rate": 3.474641780406821e-06,
      "loss": 0.1836,
      "step": 7634
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2205031281529022,
      "learning_rate": 3.473408086855939e-06,
      "loss": 0.1892,
      "step": 7635
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3446611788610965,
      "learning_rate": 3.4721744957897984e-06,
      "loss": 0.2019,
      "step": 7636
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3964940878261913,
      "learning_rate": 3.470941007291214e-06,
      "loss": 0.1809,
      "step": 7637
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.427049912357395,
      "learning_rate": 3.469707621442991e-06,
      "loss": 0.1649,
      "step": 7638
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4225924599064432,
      "learning_rate": 3.468474338327933e-06,
      "loss": 0.2261,
      "step": 7639
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2929281936854873,
      "learning_rate": 3.4672411580288313e-06,
      "loss": 0.1748,
      "step": 7640
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.66907069235532,
      "learning_rate": 3.466008080628478e-06,
      "loss": 0.6368,
      "step": 7641
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.312003020561713,
      "learning_rate": 3.464775106209647e-06,
      "loss": 0.1764,
      "step": 7642
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.868388268494398,
      "learning_rate": 3.463542234855116e-06,
      "loss": 0.5721,
      "step": 7643
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.321817824598701,
      "learning_rate": 3.4623094666476487e-06,
      "loss": 0.1845,
      "step": 7644
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.129538034585738,
      "learning_rate": 3.461076801670008e-06,
      "loss": 0.1349,
      "step": 7645
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2563461149416881,
      "learning_rate": 3.4598442400049436e-06,
      "loss": 0.2118,
      "step": 7646
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2342879626180465,
      "learning_rate": 3.458611781735201e-06,
      "loss": 0.1692,
      "step": 7647
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1630122494249018,
      "learning_rate": 3.4573794269435213e-06,
      "loss": 0.16,
      "step": 7648
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.461939952654303,
      "learning_rate": 3.456147175712633e-06,
      "loss": 0.2348,
      "step": 7649
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4328513030543346,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.193,
      "step": 7650
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3540324226204616,
      "learning_rate": 3.4536829842641283e-06,
      "loss": 0.2145,
      "step": 7651
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.611858139845058,
      "learning_rate": 3.452451044211942e-06,
      "loss": 0.5789,
      "step": 7652
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.235855693672821,
      "learning_rate": 3.4512192080514034e-06,
      "loss": 0.5535,
      "step": 7653
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3627137730320358,
      "learning_rate": 3.449987475865213e-06,
      "loss": 0.1726,
      "step": 7654
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3406922542141064,
      "learning_rate": 3.448755847736062e-06,
      "loss": 0.1482,
      "step": 7655
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4493548693777634,
      "learning_rate": 3.447524323746627e-06,
      "loss": 0.1662,
      "step": 7656
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3488234622088493,
      "learning_rate": 3.44629290397959e-06,
      "loss": 0.1826,
      "step": 7657
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3300626906485307,
      "learning_rate": 3.4450615885176166e-06,
      "loss": 0.2215,
      "step": 7658
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.605805248608052,
      "learning_rate": 3.4438303774433726e-06,
      "loss": 0.2204,
      "step": 7659
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.13786421682388,
      "learning_rate": 3.442599270839508e-06,
      "loss": 0.1601,
      "step": 7660
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4071404716846476,
      "learning_rate": 3.441368268788674e-06,
      "loss": 0.1972,
      "step": 7661
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2391649788741828,
      "learning_rate": 3.44013737137351e-06,
      "loss": 0.2007,
      "step": 7662
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2652779391049336,
      "learning_rate": 3.4389065786766516e-06,
      "loss": 0.1821,
      "step": 7663
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3730957516895934,
      "learning_rate": 3.437675890780724e-06,
      "loss": 0.176,
      "step": 7664
    },
    {
      "epoch": 0.61,
      "grad_norm": 14.907094636467663,
      "learning_rate": 3.436445307768347e-06,
      "loss": 0.5789,
      "step": 7665
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4595203884817498,
      "learning_rate": 3.4352148297221355e-06,
      "loss": 0.2077,
      "step": 7666
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3858247353886801,
      "learning_rate": 3.4339844567246918e-06,
      "loss": 0.1401,
      "step": 7667
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.331750757497658,
      "learning_rate": 3.4327541888586168e-06,
      "loss": 0.1857,
      "step": 7668
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4545698433165115,
      "learning_rate": 3.431524026206501e-06,
      "loss": 0.1835,
      "step": 7669
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4580999643554522,
      "learning_rate": 3.430293968850931e-06,
      "loss": 0.1555,
      "step": 7670
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3779169160989622,
      "learning_rate": 3.4290640168744826e-06,
      "loss": 0.2026,
      "step": 7671
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.938520070855153,
      "learning_rate": 3.4278341703597253e-06,
      "loss": 0.6658,
      "step": 7672
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5223876222983108,
      "learning_rate": 3.4266044293892254e-06,
      "loss": 0.2085,
      "step": 7673
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.312150468697119,
      "learning_rate": 3.425374794045536e-06,
      "loss": 0.1794,
      "step": 7674
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1899787047623551,
      "learning_rate": 3.4241452644112085e-06,
      "loss": 0.1765,
      "step": 7675
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3446414265389965,
      "learning_rate": 3.4229158405687824e-06,
      "loss": 0.2066,
      "step": 7676
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4684917839439975,
      "learning_rate": 3.4216865226007956e-06,
      "loss": 0.1669,
      "step": 7677
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4924335066369387,
      "learning_rate": 3.420457310589774e-06,
      "loss": 0.2005,
      "step": 7678
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.224297504529469,
      "learning_rate": 3.41922820461824e-06,
      "loss": 0.1585,
      "step": 7679
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3101762361023874,
      "learning_rate": 3.417999204768706e-06,
      "loss": 0.1978,
      "step": 7680
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4138713086705448,
      "learning_rate": 3.416770311123677e-06,
      "loss": 0.2163,
      "step": 7681
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.34388490539271,
      "learning_rate": 3.4155415237656573e-06,
      "loss": 0.1603,
      "step": 7682
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.33357589681583,
      "learning_rate": 3.4143128427771327e-06,
      "loss": 0.1781,
      "step": 7683
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3801021835872518,
      "learning_rate": 3.413084268240593e-06,
      "loss": 0.1795,
      "step": 7684
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5027114688289296,
      "learning_rate": 3.4118558002385127e-06,
      "loss": 0.1846,
      "step": 7685
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4440702629590831,
      "learning_rate": 3.4106274388533677e-06,
      "loss": 0.176,
      "step": 7686
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3353566814125521,
      "learning_rate": 3.4093991841676155e-06,
      "loss": 0.1743,
      "step": 7687
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2163258164733546,
      "learning_rate": 3.4081710362637167e-06,
      "loss": 0.1359,
      "step": 7688
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2251374318458983,
      "learning_rate": 3.4069429952241196e-06,
      "loss": 0.1703,
      "step": 7689
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4283081048042512,
      "learning_rate": 3.4057150611312644e-06,
      "loss": 0.182,
      "step": 7690
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4760255211680222,
      "learning_rate": 3.404487234067589e-06,
      "loss": 0.2401,
      "step": 7691
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2494232795423872,
      "learning_rate": 3.4032595141155183e-06,
      "loss": 0.175,
      "step": 7692
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3604982892370947,
      "learning_rate": 3.4020319013574765e-06,
      "loss": 0.1725,
      "step": 7693
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.341118165255663,
      "learning_rate": 3.400804395875873e-06,
      "loss": 0.1509,
      "step": 7694
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.492104364674769,
      "learning_rate": 3.399576997753117e-06,
      "loss": 0.2353,
      "step": 7695
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2745162264618315,
      "learning_rate": 3.398349707071604e-06,
      "loss": 0.208,
      "step": 7696
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2720977657707615,
      "learning_rate": 3.397122523913732e-06,
      "loss": 0.1508,
      "step": 7697
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3838446990349442,
      "learning_rate": 3.3958954483618787e-06,
      "loss": 0.1731,
      "step": 7698
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3683112633740533,
      "learning_rate": 3.394668480498424e-06,
      "loss": 0.1798,
      "step": 7699
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3225059137619923,
      "learning_rate": 3.3934416204057396e-06,
      "loss": 0.1735,
      "step": 7700
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4857631690953512,
      "learning_rate": 3.392214868166186e-06,
      "loss": 0.1877,
      "step": 7701
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4729268937410531,
      "learning_rate": 3.3909882238621194e-06,
      "loss": 0.1771,
      "step": 7702
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4064951237762462,
      "learning_rate": 3.389761687575888e-06,
      "loss": 0.1787,
      "step": 7703
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.415383611665988,
      "learning_rate": 3.388535259389836e-06,
      "loss": 0.5764,
      "step": 7704
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.352028177375627,
      "learning_rate": 3.387308939386291e-06,
      "loss": 0.1423,
      "step": 7705
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1279000827610544,
      "learning_rate": 3.386082727647586e-06,
      "loss": 0.1291,
      "step": 7706
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.536461377366811,
      "learning_rate": 3.3848566242560375e-06,
      "loss": 0.6163,
      "step": 7707
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.501787021073541,
      "learning_rate": 3.3836306292939557e-06,
      "loss": 0.1833,
      "step": 7708
    },
    {
      "epoch": 0.62,
      "grad_norm": 12.615404589841992,
      "learning_rate": 3.382404742843648e-06,
      "loss": 0.4738,
      "step": 7709
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.192450974711765,
      "learning_rate": 3.38117896498741e-06,
      "loss": 0.6507,
      "step": 7710
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3267504834250634,
      "learning_rate": 3.3799532958075353e-06,
      "loss": 0.1815,
      "step": 7711
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3311650267560347,
      "learning_rate": 3.378727735386302e-06,
      "loss": 0.1678,
      "step": 7712
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5202942941280007,
      "learning_rate": 3.377502283805989e-06,
      "loss": 0.1754,
      "step": 7713
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2640521743395985,
      "learning_rate": 3.376276941148863e-06,
      "loss": 0.1567,
      "step": 7714
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.461272207425104,
      "learning_rate": 3.375051707497187e-06,
      "loss": 0.4622,
      "step": 7715
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4444112970232195,
      "learning_rate": 3.373826582933213e-06,
      "loss": 0.1896,
      "step": 7716
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3759555178134941,
      "learning_rate": 3.3726015675391864e-06,
      "loss": 0.1558,
      "step": 7717
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6195306744055968,
      "learning_rate": 3.371376661397349e-06,
      "loss": 0.1983,
      "step": 7718
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.637523838995361,
      "learning_rate": 3.370151864589928e-06,
      "loss": 0.2058,
      "step": 7719
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3481068474569262,
      "learning_rate": 3.368927177199154e-06,
      "loss": 0.1615,
      "step": 7720
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4378463529080978,
      "learning_rate": 3.367702599307239e-06,
      "loss": 0.1949,
      "step": 7721
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4169852129334608,
      "learning_rate": 3.366478130996394e-06,
      "loss": 0.2023,
      "step": 7722
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.269424226763755,
      "learning_rate": 3.365253772348821e-06,
      "loss": 0.1729,
      "step": 7723
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3385725195025544,
      "learning_rate": 3.3640295234467167e-06,
      "loss": 0.182,
      "step": 7724
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.242560393136787,
      "learning_rate": 3.3628053843722674e-06,
      "loss": 0.1487,
      "step": 7725
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3922206540484354,
      "learning_rate": 3.3615813552076513e-06,
      "loss": 0.1884,
      "step": 7726
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4277988751449666,
      "learning_rate": 3.3603574360350454e-06,
      "loss": 0.1825,
      "step": 7727
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3209581766965965,
      "learning_rate": 3.359133626936611e-06,
      "loss": 0.182,
      "step": 7728
    },
    {
      "epoch": 0.62,
      "grad_norm": 12.484666441966619,
      "learning_rate": 3.3579099279945083e-06,
      "loss": 0.7144,
      "step": 7729
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3261838838735356,
      "learning_rate": 3.3566863392908864e-06,
      "loss": 0.1851,
      "step": 7730
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.502429668614384,
      "learning_rate": 3.355462860907892e-06,
      "loss": 0.2346,
      "step": 7731
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3985378017109573,
      "learning_rate": 3.354239492927658e-06,
      "loss": 0.1883,
      "step": 7732
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5973159535836983,
      "learning_rate": 3.3530162354323113e-06,
      "loss": 0.1749,
      "step": 7733
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.3514475249712,
      "learning_rate": 3.351793088503977e-06,
      "loss": 0.4768,
      "step": 7734
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.7001347967437965,
      "learning_rate": 3.3505700522247652e-06,
      "loss": 0.5179,
      "step": 7735
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3680083301584514,
      "learning_rate": 3.349347126676784e-06,
      "loss": 0.1973,
      "step": 7736
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3763767039578554,
      "learning_rate": 3.34812431194213e-06,
      "loss": 0.2082,
      "step": 7737
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2190872441625846,
      "learning_rate": 3.346901608102899e-06,
      "loss": 0.1366,
      "step": 7738
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3634348471065028,
      "learning_rate": 3.3456790152411687e-06,
      "loss": 0.2048,
      "step": 7739
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.345356500562698,
      "learning_rate": 3.34445653343902e-06,
      "loss": 0.1634,
      "step": 7740
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4082196968512268,
      "learning_rate": 3.3432341627785216e-06,
      "loss": 0.2276,
      "step": 7741
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.515889426489952,
      "learning_rate": 3.342011903341731e-06,
      "loss": 0.1796,
      "step": 7742
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.226804421674503,
      "learning_rate": 3.340789755210706e-06,
      "loss": 0.1494,
      "step": 7743
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3031227965647956,
      "learning_rate": 3.339567718467491e-06,
      "loss": 0.1708,
      "step": 7744
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3784855369885873,
      "learning_rate": 3.3383457931941275e-06,
      "loss": 0.1652,
      "step": 7745
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.384856629314098,
      "learning_rate": 3.337123979472644e-06,
      "loss": 0.1563,
      "step": 7746
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.22030422619057,
      "learning_rate": 3.3359022773850673e-06,
      "loss": 0.1685,
      "step": 7747
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.531771514871892,
      "learning_rate": 3.334680687013412e-06,
      "loss": 0.245,
      "step": 7748
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2510684197019721,
      "learning_rate": 3.3334592084396886e-06,
      "loss": 0.1488,
      "step": 7749
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3728412774049734,
      "learning_rate": 3.3322378417458985e-06,
      "loss": 0.1792,
      "step": 7750
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5609031167471188,
      "learning_rate": 3.3310165870140333e-06,
      "loss": 0.2277,
      "step": 7751
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.46860824450199,
      "learning_rate": 3.329795444326084e-06,
      "loss": 0.1731,
      "step": 7752
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1880995044353495,
      "learning_rate": 3.328574413764024e-06,
      "loss": 0.1669,
      "step": 7753
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3035460095575666,
      "learning_rate": 3.3273534954098298e-06,
      "loss": 0.172,
      "step": 7754
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4116346485182145,
      "learning_rate": 3.3261326893454617e-06,
      "loss": 0.2273,
      "step": 7755
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.323658549401525,
      "learning_rate": 3.3249119956528804e-06,
      "loss": 0.1861,
      "step": 7756
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5599893933014164,
      "learning_rate": 3.3236914144140297e-06,
      "loss": 0.2074,
      "step": 7757
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2284783631430836,
      "learning_rate": 3.3224709457108545e-06,
      "loss": 0.1788,
      "step": 7758
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2170113560535942,
      "learning_rate": 3.3212505896252887e-06,
      "loss": 0.1938,
      "step": 7759
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3725682407425117,
      "learning_rate": 3.3200303462392548e-06,
      "loss": 0.1853,
      "step": 7760
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3996863564165063,
      "learning_rate": 3.318810215634675e-06,
      "loss": 0.1887,
      "step": 7761
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4758313997933659,
      "learning_rate": 3.317590197893458e-06,
      "loss": 0.1805,
      "step": 7762
    },
    {
      "epoch": 0.62,
      "grad_norm": 22.01601098908984,
      "learning_rate": 3.31637029309751e-06,
      "loss": 0.5699,
      "step": 7763
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4469828343136375,
      "learning_rate": 3.3151505013287245e-06,
      "loss": 0.1887,
      "step": 7764
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.421835163108374,
      "learning_rate": 3.313930822668992e-06,
      "loss": 0.1942,
      "step": 7765
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5270922145613641,
      "learning_rate": 3.312711257200191e-06,
      "loss": 0.1919,
      "step": 7766
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3394326876416576,
      "learning_rate": 3.3114918050041956e-06,
      "loss": 0.1793,
      "step": 7767
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3330306915356975,
      "learning_rate": 3.3102724661628727e-06,
      "loss": 0.2077,
      "step": 7768
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4173312056534677,
      "learning_rate": 3.3090532407580767e-06,
      "loss": 0.1844,
      "step": 7769
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3595523899259356,
      "learning_rate": 3.307834128871661e-06,
      "loss": 0.1966,
      "step": 7770
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.124063701109471,
      "learning_rate": 3.3066151305854665e-06,
      "loss": 0.5644,
      "step": 7771
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.375166845040154,
      "learning_rate": 3.305396245981331e-06,
      "loss": 0.2014,
      "step": 7772
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2475913257964744,
      "learning_rate": 3.304177475141078e-06,
      "loss": 0.1362,
      "step": 7773
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4014181548955384,
      "learning_rate": 3.30295881814653e-06,
      "loss": 0.1846,
      "step": 7774
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3482336749955173,
      "learning_rate": 3.3017402750794976e-06,
      "loss": 0.1955,
      "step": 7775
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1905663697529167,
      "learning_rate": 3.300521846021788e-06,
      "loss": 0.1731,
      "step": 7776
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4089279381462456,
      "learning_rate": 3.2993035310551945e-06,
      "loss": 0.2009,
      "step": 7777
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2138497644044441,
      "learning_rate": 3.298085330261508e-06,
      "loss": 0.1436,
      "step": 7778
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4767676287058094,
      "learning_rate": 3.2968672437225115e-06,
      "loss": 0.1858,
      "step": 7779
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3205353358612255,
      "learning_rate": 3.2956492715199744e-06,
      "loss": 0.1824,
      "step": 7780
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2895950792824262,
      "learning_rate": 3.294431413735667e-06,
      "loss": 0.1989,
      "step": 7781
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.443016719972543,
      "learning_rate": 3.293213670451346e-06,
      "loss": 0.2092,
      "step": 7782
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4186463782525403,
      "learning_rate": 3.2919960417487643e-06,
      "loss": 0.2003,
      "step": 7783
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3304624205199254,
      "learning_rate": 3.2907785277096616e-06,
      "loss": 0.1587,
      "step": 7784
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1785842702036096,
      "learning_rate": 3.2895611284157757e-06,
      "loss": 0.1338,
      "step": 7785
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1763513998631734,
      "learning_rate": 3.288343843948835e-06,
      "loss": 0.1818,
      "step": 7786
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4085880984414645,
      "learning_rate": 3.287126674390556e-06,
      "loss": 0.1879,
      "step": 7787
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4839166910308192,
      "learning_rate": 3.285909619822655e-06,
      "loss": 0.1811,
      "step": 7788
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2913595765556065,
      "learning_rate": 3.2846926803268325e-06,
      "loss": 0.1584,
      "step": 7789
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3162434867545292,
      "learning_rate": 3.2834758559847903e-06,
      "loss": 0.2202,
      "step": 7790
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.493748746273879,
      "learning_rate": 3.2822591468782115e-06,
      "loss": 0.6513,
      "step": 7791
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.656035272661837,
      "learning_rate": 3.2810425530887836e-06,
      "loss": 0.217,
      "step": 7792
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3094612398965175,
      "learning_rate": 3.279826074698177e-06,
      "loss": 0.2148,
      "step": 7793
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4038220647668633,
      "learning_rate": 3.2786097117880566e-06,
      "loss": 0.1677,
      "step": 7794
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4030254942476899,
      "learning_rate": 3.2773934644400825e-06,
      "loss": 0.1965,
      "step": 7795
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.328350866961088,
      "learning_rate": 3.276177332735903e-06,
      "loss": 0.1945,
      "step": 7796
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3654663453642728,
      "learning_rate": 3.2749613167571647e-06,
      "loss": 0.1848,
      "step": 7797
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3191245026600775,
      "learning_rate": 3.2737454165854976e-06,
      "loss": 0.1872,
      "step": 7798
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.462070637370341,
      "learning_rate": 3.2725296323025326e-06,
      "loss": 0.2282,
      "step": 7799
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4968293239609527,
      "learning_rate": 3.271313963989886e-06,
      "loss": 0.2511,
      "step": 7800
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2311298454863742,
      "learning_rate": 3.2700984117291722e-06,
      "loss": 0.2021,
      "step": 7801
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3079270525733746,
      "learning_rate": 3.2688829756019934e-06,
      "loss": 0.1938,
      "step": 7802
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.205557791342672,
      "learning_rate": 3.267667655689944e-06,
      "loss": 0.5393,
      "step": 7803
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.416263705880692,
      "learning_rate": 3.2664524520746154e-06,
      "loss": 0.1645,
      "step": 7804
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3744361023492,
      "learning_rate": 3.2652373648375836e-06,
      "loss": 0.1935,
      "step": 7805
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.319110613616149,
      "learning_rate": 3.2640223940604266e-06,
      "loss": 0.1962,
      "step": 7806
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2476514942331776,
      "learning_rate": 3.2628075398247037e-06,
      "loss": 0.1504,
      "step": 7807
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.316200756428,
      "learning_rate": 3.2615928022119757e-06,
      "loss": 0.1839,
      "step": 7808
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.371596941761974,
      "learning_rate": 3.260378181303788e-06,
      "loss": 0.1492,
      "step": 7809
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2817443162043318,
      "learning_rate": 3.259163677181687e-06,
      "loss": 0.1679,
      "step": 7810
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1158090261561235,
      "learning_rate": 3.2579492899272014e-06,
      "loss": 0.1249,
      "step": 7811
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.583192836582663,
      "learning_rate": 3.256735019621857e-06,
      "loss": 0.454,
      "step": 7812
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3117492656924534,
      "learning_rate": 3.2555208663471748e-06,
      "loss": 0.1576,
      "step": 7813
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3093809825661626,
      "learning_rate": 3.2543068301846607e-06,
      "loss": 0.1974,
      "step": 7814
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.076801601266826,
      "learning_rate": 3.2530929112158194e-06,
      "loss": 0.5858,
      "step": 7815
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4272500568790647,
      "learning_rate": 3.2518791095221425e-06,
      "loss": 0.1625,
      "step": 7816
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.391157088025085,
      "learning_rate": 3.2506654251851194e-06,
      "loss": 0.2138,
      "step": 7817
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4036951424243067,
      "learning_rate": 3.249451858286225e-06,
      "loss": 0.225,
      "step": 7818
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6447430793369968,
      "learning_rate": 3.248238408906932e-06,
      "loss": 0.172,
      "step": 7819
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4464535407033532,
      "learning_rate": 3.247025077128704e-06,
      "loss": 0.1967,
      "step": 7820
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3394565615373315,
      "learning_rate": 3.2458118630329905e-06,
      "loss": 0.1889,
      "step": 7821
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3283642111956842,
      "learning_rate": 3.2445987667012435e-06,
      "loss": 0.1775,
      "step": 7822
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5320664837364364,
      "learning_rate": 3.243385788214899e-06,
      "loss": 0.2146,
      "step": 7823
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2869894925539433,
      "learning_rate": 3.2421729276553903e-06,
      "loss": 0.1629,
      "step": 7824
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6769634201741217,
      "learning_rate": 3.240960185104137e-06,
      "loss": 0.2091,
      "step": 7825
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3527731199345756,
      "learning_rate": 3.2397475606425577e-06,
      "loss": 0.1864,
      "step": 7826
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3256152073090754,
      "learning_rate": 3.2385350543520564e-06,
      "loss": 0.1744,
      "step": 7827
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4980477140988482,
      "learning_rate": 3.237322666314037e-06,
      "loss": 0.1979,
      "step": 7828
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2977104675572346,
      "learning_rate": 3.2361103966098856e-06,
      "loss": 0.1751,
      "step": 7829
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1910525961959622,
      "learning_rate": 3.234898245320987e-06,
      "loss": 0.1441,
      "step": 7830
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4361172315887394,
      "learning_rate": 3.2336862125287193e-06,
      "loss": 0.193,
      "step": 7831
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.730218691043036,
      "learning_rate": 3.2324742983144463e-06,
      "loss": 0.585,
      "step": 7832
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.351932024338407,
      "learning_rate": 3.231262502759529e-06,
      "loss": 0.1879,
      "step": 7833
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3642289781127743,
      "learning_rate": 3.230050825945319e-06,
      "loss": 0.2124,
      "step": 7834
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.340660273633863,
      "learning_rate": 3.2288392679531612e-06,
      "loss": 0.1688,
      "step": 7835
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3850406138284692,
      "learning_rate": 3.227627828864388e-06,
      "loss": 0.1662,
      "step": 7836
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.397425834733644,
      "learning_rate": 3.2264165087603293e-06,
      "loss": 0.1761,
      "step": 7837
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3937390422650353,
      "learning_rate": 3.2252053077223056e-06,
      "loss": 0.1889,
      "step": 7838
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2166233662075885,
      "learning_rate": 3.2239942258316247e-06,
      "loss": 0.1465,
      "step": 7839
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.420200911465506,
      "learning_rate": 3.2227832631695936e-06,
      "loss": 0.1882,
      "step": 7840
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3424886021171207,
      "learning_rate": 3.2215724198175056e-06,
      "loss": 0.202,
      "step": 7841
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6242838269570241,
      "learning_rate": 3.220361695856652e-06,
      "loss": 0.1786,
      "step": 7842
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2401779048018964,
      "learning_rate": 3.219151091368308e-06,
      "loss": 0.1675,
      "step": 7843
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4277772454644777,
      "learning_rate": 3.217940606433747e-06,
      "loss": 0.2226,
      "step": 7844
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2406971436276348,
      "learning_rate": 3.2167302411342322e-06,
      "loss": 0.1784,
      "step": 7845
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2155890972610872,
      "learning_rate": 3.2155199955510214e-06,
      "loss": 0.1533,
      "step": 7846
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.285633685865284,
      "learning_rate": 3.2143098697653586e-06,
      "loss": 0.1747,
      "step": 7847
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3749982491877086,
      "learning_rate": 3.2130998638584842e-06,
      "loss": 0.201,
      "step": 7848
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5525914416059414,
      "learning_rate": 3.2118899779116307e-06,
      "loss": 0.2492,
      "step": 7849
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2461663559532454,
      "learning_rate": 3.2106802120060197e-06,
      "loss": 0.1727,
      "step": 7850
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5308293375182658,
      "learning_rate": 3.2094705662228697e-06,
      "loss": 0.2049,
      "step": 7851
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1509735500015763,
      "learning_rate": 3.2082610406433838e-06,
      "loss": 0.1432,
      "step": 7852
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.438699278371207,
      "learning_rate": 3.2070516353487634e-06,
      "loss": 0.7067,
      "step": 7853
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1952676313598005,
      "learning_rate": 3.2058423504202007e-06,
      "loss": 0.1743,
      "step": 7854
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.136077856187923,
      "learning_rate": 3.2046331859388757e-06,
      "loss": 0.5843,
      "step": 7855
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2460555574057655,
      "learning_rate": 3.2034241419859645e-06,
      "loss": 0.166,
      "step": 7856
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.622753746961374,
      "learning_rate": 3.2022152186426337e-06,
      "loss": 0.5816,
      "step": 7857
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.3953046945051435,
      "learning_rate": 3.201006415990044e-06,
      "loss": 0.4799,
      "step": 7858
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4801755667147358,
      "learning_rate": 3.199797734109342e-06,
      "loss": 0.1765,
      "step": 7859
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.508675854023696,
      "learning_rate": 3.198589173081674e-06,
      "loss": 0.1513,
      "step": 7860
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2101298017208768,
      "learning_rate": 3.1973807329881724e-06,
      "loss": 0.152,
      "step": 7861
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4796869854528971,
      "learning_rate": 3.196172413909965e-06,
      "loss": 0.179,
      "step": 7862
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3940533757989904,
      "learning_rate": 3.1949642159281683e-06,
      "loss": 0.1805,
      "step": 7863
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.724050690088736,
      "learning_rate": 3.1937561391238918e-06,
      "loss": 0.4812,
      "step": 7864
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2971570030071555,
      "learning_rate": 3.19254818357824e-06,
      "loss": 0.1542,
      "step": 7865
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1366325056736128,
      "learning_rate": 3.191340349372304e-06,
      "loss": 0.1251,
      "step": 7866
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3500883088843927,
      "learning_rate": 3.19013263658717e-06,
      "loss": 0.1559,
      "step": 7867
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2692844803483487,
      "learning_rate": 3.1889250453039156e-06,
      "loss": 0.177,
      "step": 7868
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.293474769002433,
      "learning_rate": 3.187717575603612e-06,
      "loss": 0.1718,
      "step": 7869
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4909055622686322,
      "learning_rate": 3.1865102275673167e-06,
      "loss": 0.1882,
      "step": 7870
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1882880789279795,
      "learning_rate": 3.1853030012760853e-06,
      "loss": 0.1579,
      "step": 7871
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.228916074395396,
      "learning_rate": 3.184095896810963e-06,
      "loss": 0.1559,
      "step": 7872
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.673248565033386,
      "learning_rate": 3.1828889142529828e-06,
      "loss": 0.504,
      "step": 7873
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.441706467128113,
      "learning_rate": 3.1816820536831774e-06,
      "loss": 0.2082,
      "step": 7874
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3555205795658416,
      "learning_rate": 3.180475315182563e-06,
      "loss": 0.1974,
      "step": 7875
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.351168669725019,
      "learning_rate": 3.179268698832155e-06,
      "loss": 0.1864,
      "step": 7876
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3693316013945411,
      "learning_rate": 3.178062204712955e-06,
      "loss": 0.1546,
      "step": 7877
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3564658330938701,
      "learning_rate": 3.1768558329059605e-06,
      "loss": 0.2183,
      "step": 7878
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3597714213269423,
      "learning_rate": 3.1756495834921563e-06,
      "loss": 0.1803,
      "step": 7879
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4989735460200675,
      "learning_rate": 3.1744434565525252e-06,
      "loss": 0.2109,
      "step": 7880
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3761517824161937,
      "learning_rate": 3.1732374521680355e-06,
      "loss": 0.1527,
      "step": 7881
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.98422907051574,
      "learning_rate": 3.172031570419649e-06,
      "loss": 0.6473,
      "step": 7882
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7494956667675856,
      "learning_rate": 3.170825811388324e-06,
      "loss": 0.1552,
      "step": 7883
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.536085026547384,
      "learning_rate": 3.169620175155002e-06,
      "loss": 0.2053,
      "step": 7884
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3288692813015224,
      "learning_rate": 3.168414661800625e-06,
      "loss": 0.1478,
      "step": 7885
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4485212484820296,
      "learning_rate": 3.16720927140612e-06,
      "loss": 0.2265,
      "step": 7886
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.155445287583118,
      "learning_rate": 3.1660040040524122e-06,
      "loss": 0.4887,
      "step": 7887
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2289088055151425,
      "learning_rate": 3.164798859820408e-06,
      "loss": 0.184,
      "step": 7888
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3349459061874227,
      "learning_rate": 3.1635938387910216e-06,
      "loss": 0.1835,
      "step": 7889
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3655323058446471,
      "learning_rate": 3.1623889410451435e-06,
      "loss": 0.1718,
      "step": 7890
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5648082694013254,
      "learning_rate": 3.161184166663661e-06,
      "loss": 0.1794,
      "step": 7891
    },
    {
      "epoch": 0.63,
      "grad_norm": 11.219835770567885,
      "learning_rate": 3.1599795157274604e-06,
      "loss": 0.6247,
      "step": 7892
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3802950873605506,
      "learning_rate": 3.1587749883174074e-06,
      "loss": 0.2174,
      "step": 7893
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.867841308632259,
      "learning_rate": 3.157570584514369e-06,
      "loss": 0.4619,
      "step": 7894
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.546799712327459,
      "learning_rate": 3.1563663043991987e-06,
      "loss": 0.2003,
      "step": 7895
    },
    {
      "epoch": 0.63,
      "grad_norm": 26.439785755216093,
      "learning_rate": 3.1551621480527466e-06,
      "loss": 0.7484,
      "step": 7896
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.242211273511225,
      "learning_rate": 3.1539581155558467e-06,
      "loss": 0.1469,
      "step": 7897
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5403378121062878,
      "learning_rate": 3.1527542069893334e-06,
      "loss": 0.233,
      "step": 7898
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.336525950504161,
      "learning_rate": 3.1515504224340277e-06,
      "loss": 0.1391,
      "step": 7899
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.558556340081549,
      "learning_rate": 3.1503467619707407e-06,
      "loss": 0.2176,
      "step": 7900
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.457297069163006,
      "learning_rate": 3.149143225680281e-06,
      "loss": 0.2003,
      "step": 7901
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5429458923521044,
      "learning_rate": 3.1479398136434438e-06,
      "loss": 0.1997,
      "step": 7902
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2540375996622262,
      "learning_rate": 3.1467365259410194e-06,
      "loss": 0.2265,
      "step": 7903
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.038069062383274,
      "learning_rate": 3.1455333626537852e-06,
      "loss": 0.6149,
      "step": 7904
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2581951508839013,
      "learning_rate": 3.1443303238625172e-06,
      "loss": 0.1698,
      "step": 7905
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.336066610476519,
      "learning_rate": 3.143127409647976e-06,
      "loss": 0.1822,
      "step": 7906
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2870040444365258,
      "learning_rate": 3.141924620090919e-06,
      "loss": 0.149,
      "step": 7907
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.626078311840451,
      "learning_rate": 3.1407219552720915e-06,
      "loss": 0.2213,
      "step": 7908
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4622793258906668,
      "learning_rate": 3.139519415272231e-06,
      "loss": 0.2009,
      "step": 7909
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4444155569498092,
      "learning_rate": 3.138317000172072e-06,
      "loss": 0.2155,
      "step": 7910
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3096067830056106,
      "learning_rate": 3.1371147100523308e-06,
      "loss": 0.199,
      "step": 7911
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.464231902271683,
      "learning_rate": 3.1359125449937243e-06,
      "loss": 0.2315,
      "step": 7912
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.9267884060269225,
      "learning_rate": 3.1347105050769565e-06,
      "loss": 0.6076,
      "step": 7913
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2765696927219845,
      "learning_rate": 3.1335085903827257e-06,
      "loss": 0.1666,
      "step": 7914
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3095728531874564,
      "learning_rate": 3.1323068009917174e-06,
      "loss": 0.1678,
      "step": 7915
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4075681281647845,
      "learning_rate": 3.1311051369846114e-06,
      "loss": 0.1688,
      "step": 7916
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.497622412455353,
      "learning_rate": 3.129903598442082e-06,
      "loss": 0.3725,
      "step": 7917
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2897731598339204,
      "learning_rate": 3.1287021854447884e-06,
      "loss": 0.1541,
      "step": 7918
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.054746550074537,
      "learning_rate": 3.127500898073388e-06,
      "loss": 0.5173,
      "step": 7919
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3866776097092954,
      "learning_rate": 3.1262997364085248e-06,
      "loss": 0.2185,
      "step": 7920
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2419760877543193,
      "learning_rate": 3.125098700530839e-06,
      "loss": 0.1753,
      "step": 7921
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3698172289136854,
      "learning_rate": 3.1238977905209554e-06,
      "loss": 0.1551,
      "step": 7922
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.225018726144095,
      "learning_rate": 3.1226970064595e-06,
      "loss": 0.1676,
      "step": 7923
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2920389546834399,
      "learning_rate": 3.121496348427083e-06,
      "loss": 0.2073,
      "step": 7924
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5462429014092947,
      "learning_rate": 3.1202958165043053e-06,
      "loss": 0.1737,
      "step": 7925
    },
    {
      "epoch": 0.63,
      "grad_norm": 14.685746266884983,
      "learning_rate": 3.1190954107717664e-06,
      "loss": 0.631,
      "step": 7926
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.218413056739077,
      "learning_rate": 3.11789513131005e-06,
      "loss": 0.1368,
      "step": 7927
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4745941208586935,
      "learning_rate": 3.116694978199738e-06,
      "loss": 0.2253,
      "step": 7928
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3242431806499277,
      "learning_rate": 3.115494951521397e-06,
      "loss": 0.1991,
      "step": 7929
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2534074074076318,
      "learning_rate": 3.1142950513555903e-06,
      "loss": 0.1852,
      "step": 7930
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3691204058955233,
      "learning_rate": 3.1130952777828694e-06,
      "loss": 0.207,
      "step": 7931
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.058957440022268,
      "learning_rate": 3.11189563088378e-06,
      "loss": 0.6252,
      "step": 7932
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.409825452184253,
      "learning_rate": 3.110696110738859e-06,
      "loss": 0.2144,
      "step": 7933
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.277516986678831,
      "learning_rate": 3.1094967174286307e-06,
      "loss": 0.1613,
      "step": 7934
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4242585071223857,
      "learning_rate": 3.1082974510336163e-06,
      "loss": 0.2122,
      "step": 7935
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1411302891522008,
      "learning_rate": 3.1070983116343245e-06,
      "loss": 0.1467,
      "step": 7936
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3535875281186098,
      "learning_rate": 3.105899299311261e-06,
      "loss": 0.1639,
      "step": 7937
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.356475780365062,
      "learning_rate": 3.1047004141449143e-06,
      "loss": 0.1883,
      "step": 7938
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4580495081435039,
      "learning_rate": 3.1035016562157716e-06,
      "loss": 0.1804,
      "step": 7939
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4036354169981848,
      "learning_rate": 3.1023030256043087e-06,
      "loss": 0.19,
      "step": 7940
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4519009655631587,
      "learning_rate": 3.1011045223909954e-06,
      "loss": 0.1669,
      "step": 7941
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2164440799792453,
      "learning_rate": 3.0999061466562873e-06,
      "loss": 0.1642,
      "step": 7942
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1890463668990345,
      "learning_rate": 3.0987078984806363e-06,
      "loss": 0.173,
      "step": 7943
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2026975896469725,
      "learning_rate": 3.0975097779444864e-06,
      "loss": 0.1992,
      "step": 7944
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3387900797436894,
      "learning_rate": 3.0963117851282677e-06,
      "loss": 0.19,
      "step": 7945
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2187033439473896,
      "learning_rate": 3.0951139201124075e-06,
      "loss": 0.1553,
      "step": 7946
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.299014180477559,
      "learning_rate": 3.0939161829773208e-06,
      "loss": 0.1965,
      "step": 7947
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3636786086259918,
      "learning_rate": 3.0927185738034175e-06,
      "loss": 0.22,
      "step": 7948
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3841461213661053,
      "learning_rate": 3.091521092671094e-06,
      "loss": 0.1616,
      "step": 7949
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1866380015993783,
      "learning_rate": 3.090323739660742e-06,
      "loss": 0.1506,
      "step": 7950
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3102686206110632,
      "learning_rate": 3.089126514852745e-06,
      "loss": 0.1965,
      "step": 7951
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.245647558296622,
      "learning_rate": 3.087929418327472e-06,
      "loss": 0.1584,
      "step": 7952
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1954272025574186,
      "learning_rate": 3.0867324501652923e-06,
      "loss": 0.16,
      "step": 7953
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2078550701523807,
      "learning_rate": 3.0855356104465593e-06,
      "loss": 0.1825,
      "step": 7954
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.213482010562243,
      "learning_rate": 3.084338899251623e-06,
      "loss": 0.1552,
      "step": 7955
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4544353136242343,
      "learning_rate": 3.0831423166608186e-06,
      "loss": 0.1773,
      "step": 7956
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.226658074350585,
      "learning_rate": 3.0819458627544796e-06,
      "loss": 0.5129,
      "step": 7957
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4342336246594842,
      "learning_rate": 3.080749537612925e-06,
      "loss": 0.2105,
      "step": 7958
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2493829783418786,
      "learning_rate": 3.079553341316471e-06,
      "loss": 0.161,
      "step": 7959
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2888975380250858,
      "learning_rate": 3.078357273945419e-06,
      "loss": 0.1542,
      "step": 7960
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2491285654903272,
      "learning_rate": 3.0771613355800647e-06,
      "loss": 0.1838,
      "step": 7961
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3019263927742257,
      "learning_rate": 3.0759655263006983e-06,
      "loss": 0.1773,
      "step": 7962
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4576355004291175,
      "learning_rate": 3.074769846187593e-06,
      "loss": 0.2294,
      "step": 7963
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1938537892672316,
      "learning_rate": 3.0735742953210222e-06,
      "loss": 0.1472,
      "step": 7964
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0785292728841107,
      "learning_rate": 3.072378873781245e-06,
      "loss": 0.1483,
      "step": 7965
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5566486230054253,
      "learning_rate": 3.0711835816485163e-06,
      "loss": 0.1867,
      "step": 7966
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1769989579443552,
      "learning_rate": 3.069988419003076e-06,
      "loss": 0.1665,
      "step": 7967
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6003297593823855,
      "learning_rate": 3.0687933859251618e-06,
      "loss": 0.2034,
      "step": 7968
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3034821184664414,
      "learning_rate": 3.067598482495e-06,
      "loss": 0.1403,
      "step": 7969
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.17609484037814,
      "learning_rate": 3.066403708792805e-06,
      "loss": 0.5155,
      "step": 7970
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.339265253243504,
      "learning_rate": 3.0652090648987885e-06,
      "loss": 0.196,
      "step": 7971
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.386467679292854,
      "learning_rate": 3.0640145508931487e-06,
      "loss": 0.1867,
      "step": 7972
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3769171864786163,
      "learning_rate": 3.06282016685608e-06,
      "loss": 0.1639,
      "step": 7973
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.246791308558666,
      "learning_rate": 3.0616259128677594e-06,
      "loss": 0.1659,
      "step": 7974
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2241482141651574,
      "learning_rate": 3.060431789008368e-06,
      "loss": 0.1673,
      "step": 7975
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4337045249299292,
      "learning_rate": 3.0592377953580652e-06,
      "loss": 0.2072,
      "step": 7976
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.36498907173208,
      "learning_rate": 3.0580439319970103e-06,
      "loss": 0.1924,
      "step": 7977
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.352640548386208,
      "learning_rate": 3.0568501990053513e-06,
      "loss": 0.1781,
      "step": 7978
    },
    {
      "epoch": 0.64,
      "grad_norm": 13.881520506931842,
      "learning_rate": 3.0556565964632236e-06,
      "loss": 0.645,
      "step": 7979
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.075007046271416,
      "learning_rate": 3.0544631244507607e-06,
      "loss": 0.1296,
      "step": 7980
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.851927678669725,
      "learning_rate": 3.0532697830480816e-06,
      "loss": 0.6398,
      "step": 7981
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3839623298678445,
      "learning_rate": 3.0520765723353036e-06,
      "loss": 0.1745,
      "step": 7982
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.316093053741538,
      "learning_rate": 3.0508834923925245e-06,
      "loss": 0.1902,
      "step": 7983
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.583808859957966,
      "learning_rate": 3.049690543299843e-06,
      "loss": 0.2251,
      "step": 7984
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.260040471732798,
      "learning_rate": 3.0484977251373458e-06,
      "loss": 0.1664,
      "step": 7985
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3872466776814742,
      "learning_rate": 3.047305037985107e-06,
      "loss": 0.1842,
      "step": 7986
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.472579694627104,
      "learning_rate": 3.0461124819231984e-06,
      "loss": 0.1951,
      "step": 7987
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4264483729058504,
      "learning_rate": 3.0449200570316783e-06,
      "loss": 0.2034,
      "step": 7988
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2885910413014916,
      "learning_rate": 3.0437277633906006e-06,
      "loss": 0.1921,
      "step": 7989
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.147170209707625,
      "learning_rate": 3.0425356010800022e-06,
      "loss": 0.5489,
      "step": 7990
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8506478967544588,
      "learning_rate": 3.041343570179922e-06,
      "loss": 0.202,
      "step": 7991
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4062582976594764,
      "learning_rate": 3.0401516707703802e-06,
      "loss": 0.2036,
      "step": 7992
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.461011709558017,
      "learning_rate": 3.0389599029313983e-06,
      "loss": 0.5216,
      "step": 7993
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.365670652398302,
      "learning_rate": 3.037768266742978e-06,
      "loss": 0.1933,
      "step": 7994
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2913508927494168,
      "learning_rate": 3.036576762285118e-06,
      "loss": 0.1956,
      "step": 7995
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4455988966274207,
      "learning_rate": 3.0353853896378116e-06,
      "loss": 0.1993,
      "step": 7996
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2546018677158817,
      "learning_rate": 3.0341941488810346e-06,
      "loss": 0.178,
      "step": 7997
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.897786235426149,
      "learning_rate": 3.0330030400947606e-06,
      "loss": 0.4207,
      "step": 7998
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3039938817751673,
      "learning_rate": 3.031812063358952e-06,
      "loss": 0.1504,
      "step": 7999
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4734481915874804,
      "learning_rate": 3.0306212187535653e-06,
      "loss": 0.181,
      "step": 8000
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.416902362371618,
      "learning_rate": 3.0294305063585407e-06,
      "loss": 0.2144,
      "step": 8001
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4232988731374037,
      "learning_rate": 3.0282399262538175e-06,
      "loss": 0.1856,
      "step": 8002
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3555897918422068,
      "learning_rate": 3.027049478519324e-06,
      "loss": 0.2003,
      "step": 8003
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2731707083265038,
      "learning_rate": 3.0258591632349745e-06,
      "loss": 0.1585,
      "step": 8004
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3192928700010218,
      "learning_rate": 3.024668980480681e-06,
      "loss": 0.195,
      "step": 8005
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.456842749092586,
      "learning_rate": 3.023478930336343e-06,
      "loss": 0.5053,
      "step": 8006
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2783613966556062,
      "learning_rate": 3.022289012881856e-06,
      "loss": 0.1729,
      "step": 8007
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3271185074953322,
      "learning_rate": 3.0210992281970973e-06,
      "loss": 0.1863,
      "step": 8008
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2119478069963066,
      "learning_rate": 3.0199095763619445e-06,
      "loss": 0.1634,
      "step": 8009
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5163075324953244,
      "learning_rate": 3.0187200574562605e-06,
      "loss": 0.2108,
      "step": 8010
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2282426128983028,
      "learning_rate": 3.017530671559903e-06,
      "loss": 0.1582,
      "step": 8011
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.287490365919821,
      "learning_rate": 3.0163414187527174e-06,
      "loss": 0.183,
      "step": 8012
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4927378130290627,
      "learning_rate": 3.015152299114542e-06,
      "loss": 0.1944,
      "step": 8013
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4342657760444664,
      "learning_rate": 3.0139633127252086e-06,
      "loss": 0.2024,
      "step": 8014
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5665119269965095,
      "learning_rate": 3.0127744596645337e-06,
      "loss": 0.2572,
      "step": 8015
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2640150226807745,
      "learning_rate": 3.011585740012331e-06,
      "loss": 0.188,
      "step": 8016
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2337965305511025,
      "learning_rate": 3.010397153848401e-06,
      "loss": 0.1568,
      "step": 8017
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2444202683856274,
      "learning_rate": 3.00920870125254e-06,
      "loss": 0.1755,
      "step": 8018
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3839843708869037,
      "learning_rate": 3.0080203823045294e-06,
      "loss": 0.1605,
      "step": 8019
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0052052154318911,
      "learning_rate": 3.0068321970841484e-06,
      "loss": 0.1175,
      "step": 8020
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3013046054221236,
      "learning_rate": 3.0056441456711593e-06,
      "loss": 0.1732,
      "step": 8021
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.524499557127757,
      "learning_rate": 3.004456228145321e-06,
      "loss": 0.1715,
      "step": 8022
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2546895391428723,
      "learning_rate": 3.003268444586384e-06,
      "loss": 0.1836,
      "step": 8023
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1781988702479556,
      "learning_rate": 3.0020807950740844e-06,
      "loss": 0.1254,
      "step": 8024
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5944806254334936,
      "learning_rate": 3.000893279688155e-06,
      "loss": 0.2104,
      "step": 8025
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3500123985762384,
      "learning_rate": 2.999705898508316e-06,
      "loss": 0.1877,
      "step": 8026
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2135488547060997,
      "learning_rate": 2.998518651614283e-06,
      "loss": 0.1864,
      "step": 8027
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4062597794056038,
      "learning_rate": 2.9973315390857547e-06,
      "loss": 0.1886,
      "step": 8028
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3934617473424993,
      "learning_rate": 2.99614456100243e-06,
      "loss": 0.1721,
      "step": 8029
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.906684028431125,
      "learning_rate": 2.9949577174439926e-06,
      "loss": 0.5709,
      "step": 8030
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3390328777048515,
      "learning_rate": 2.993771008490117e-06,
      "loss": 0.186,
      "step": 8031
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.226946132336187,
      "learning_rate": 2.9925844342204725e-06,
      "loss": 0.146,
      "step": 8032
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.351549217559366,
      "learning_rate": 2.9913979947147166e-06,
      "loss": 0.1784,
      "step": 8033
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.259593173882677,
      "learning_rate": 2.9902116900525012e-06,
      "loss": 0.1622,
      "step": 8034
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4140991981829438,
      "learning_rate": 2.9890255203134622e-06,
      "loss": 0.1991,
      "step": 8035
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2024614312362996,
      "learning_rate": 2.9878394855772343e-06,
      "loss": 0.1787,
      "step": 8036
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4269016367493343,
      "learning_rate": 2.9866535859234377e-06,
      "loss": 0.1427,
      "step": 8037
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3466872664107317,
      "learning_rate": 2.9854678214316875e-06,
      "loss": 0.1844,
      "step": 8038
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2845359808395693,
      "learning_rate": 2.984282192181586e-06,
      "loss": 0.1622,
      "step": 8039
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5712207498775685,
      "learning_rate": 2.983096698252726e-06,
      "loss": 0.2158,
      "step": 8040
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1640605388889678,
      "learning_rate": 2.9819113397246985e-06,
      "loss": 0.1705,
      "step": 8041
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2466452164168533,
      "learning_rate": 2.980726116677075e-06,
      "loss": 0.1823,
      "step": 8042
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.376129183381196,
      "learning_rate": 2.9795410291894267e-06,
      "loss": 0.1786,
      "step": 8043
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3002618051340231,
      "learning_rate": 2.97835607734131e-06,
      "loss": 0.1786,
      "step": 8044
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3385294258781577,
      "learning_rate": 2.9771712612122765e-06,
      "loss": 0.1619,
      "step": 8045
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.324770729323476,
      "learning_rate": 2.975986580881865e-06,
      "loss": 0.1803,
      "step": 8046
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1993692767461797,
      "learning_rate": 2.9748020364296054e-06,
      "loss": 0.1486,
      "step": 8047
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3557773116067577,
      "learning_rate": 2.973617627935023e-06,
      "loss": 0.1899,
      "step": 8048
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2914036560602702,
      "learning_rate": 2.9724333554776276e-06,
      "loss": 0.1857,
      "step": 8049
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3742001099535555,
      "learning_rate": 2.9712492191369245e-06,
      "loss": 0.1654,
      "step": 8050
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.387449563862524,
      "learning_rate": 2.9700652189924075e-06,
      "loss": 0.1845,
      "step": 8051
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4989405709507886,
      "learning_rate": 2.968881355123565e-06,
      "loss": 0.1684,
      "step": 8052
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4434483826255395,
      "learning_rate": 2.967697627609869e-06,
      "loss": 0.2085,
      "step": 8053
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3887033350360027,
      "learning_rate": 2.96651403653079e-06,
      "loss": 0.1721,
      "step": 8054
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2300670224017218,
      "learning_rate": 2.965330581965786e-06,
      "loss": 0.1848,
      "step": 8055
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.852549892498079,
      "learning_rate": 2.9641472639943025e-06,
      "loss": 0.6236,
      "step": 8056
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.992211664224177,
      "learning_rate": 2.9629640826957827e-06,
      "loss": 0.6342,
      "step": 8057
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3452607159548737,
      "learning_rate": 2.961781038149656e-06,
      "loss": 0.1326,
      "step": 8058
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3716831958929907,
      "learning_rate": 2.9605981304353443e-06,
      "loss": 0.1663,
      "step": 8059
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3581235994240486,
      "learning_rate": 2.959415359632257e-06,
      "loss": 0.1741,
      "step": 8060
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1883025077721634,
      "learning_rate": 2.9582327258198016e-06,
      "loss": 0.1623,
      "step": 8061
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2389728084338965,
      "learning_rate": 2.957050229077368e-06,
      "loss": 0.1459,
      "step": 8062
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3688091674913991,
      "learning_rate": 2.9558678694843446e-06,
      "loss": 0.2218,
      "step": 8063
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3007065479293345,
      "learning_rate": 2.9546856471201046e-06,
      "loss": 0.2152,
      "step": 8064
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.406598720099492,
      "learning_rate": 2.9535035620640117e-06,
      "loss": 0.2147,
      "step": 8065
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4860810740630899,
      "learning_rate": 2.952321614395427e-06,
      "loss": 0.2069,
      "step": 8066
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3962717353760248,
      "learning_rate": 2.9511398041936952e-06,
      "loss": 0.1952,
      "step": 8067
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3240819203873126,
      "learning_rate": 2.949958131538158e-06,
      "loss": 0.1792,
      "step": 8068
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3077646152931324,
      "learning_rate": 2.9487765965081417e-06,
      "loss": 0.1752,
      "step": 8069
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3380654715786018,
      "learning_rate": 2.9475951991829676e-06,
      "loss": 0.2152,
      "step": 8070
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4176775022314827,
      "learning_rate": 2.9464139396419457e-06,
      "loss": 0.1885,
      "step": 8071
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.800748759644525,
      "learning_rate": 2.9452328179643797e-06,
      "loss": 0.4841,
      "step": 8072
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3836659358370496,
      "learning_rate": 2.9440518342295598e-06,
      "loss": 0.1716,
      "step": 8073
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4362636113371143,
      "learning_rate": 2.9428709885167685e-06,
      "loss": 0.1945,
      "step": 8074
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3339571131382026,
      "learning_rate": 2.9416902809052817e-06,
      "loss": 0.2269,
      "step": 8075
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2823320417334052,
      "learning_rate": 2.940509711474362e-06,
      "loss": 0.1585,
      "step": 8076
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4026145473441565,
      "learning_rate": 2.9393292803032656e-06,
      "loss": 0.1876,
      "step": 8077
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3952444802487651,
      "learning_rate": 2.9381489874712377e-06,
      "loss": 0.2413,
      "step": 8078
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4398183982341632,
      "learning_rate": 2.936968833057517e-06,
      "loss": 0.2085,
      "step": 8079
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3420214819739829,
      "learning_rate": 2.9357888171413273e-06,
      "loss": 0.1935,
      "step": 8080
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.146966903763939,
      "learning_rate": 2.93460893980189e-06,
      "loss": 0.1621,
      "step": 8081
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.242604526530489,
      "learning_rate": 2.9334292011184128e-06,
      "loss": 0.187,
      "step": 8082
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2556549049230001,
      "learning_rate": 2.932249601170094e-06,
      "loss": 0.1481,
      "step": 8083
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3982109169877928,
      "learning_rate": 2.931070140036124e-06,
      "loss": 0.2428,
      "step": 8084
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.433095304380012,
      "learning_rate": 2.9298908177956843e-06,
      "loss": 0.1899,
      "step": 8085
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3907870019876247,
      "learning_rate": 2.9287116345279477e-06,
      "loss": 0.158,
      "step": 8086
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.493594805341812,
      "learning_rate": 2.9275325903120734e-06,
      "loss": 0.6959,
      "step": 8087
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.164188341112721,
      "learning_rate": 2.926353685227216e-06,
      "loss": 0.633,
      "step": 8088
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.307978765465989,
      "learning_rate": 2.925174919352517e-06,
      "loss": 0.1716,
      "step": 8089
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3631015633830275,
      "learning_rate": 2.923996292767115e-06,
      "loss": 0.1801,
      "step": 8090
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2153432465393395,
      "learning_rate": 2.9228178055501313e-06,
      "loss": 0.1339,
      "step": 8091
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.260359282636512,
      "learning_rate": 2.92163945778068e-06,
      "loss": 0.1572,
      "step": 8092
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.218848672577065,
      "learning_rate": 2.9204612495378716e-06,
      "loss": 0.1929,
      "step": 8093
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.377511796936783,
      "learning_rate": 2.919283180900798e-06,
      "loss": 0.1922,
      "step": 8094
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3042255967785155,
      "learning_rate": 2.9181052519485496e-06,
      "loss": 0.1649,
      "step": 8095
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4951667941057978,
      "learning_rate": 2.916927462760204e-06,
      "loss": 0.1738,
      "step": 8096
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.370409239861674,
      "learning_rate": 2.9157498134148294e-06,
      "loss": 0.2086,
      "step": 8097
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.652994563352957,
      "learning_rate": 2.9145723039914826e-06,
      "loss": 0.6323,
      "step": 8098
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3343953595466327,
      "learning_rate": 2.9133949345692192e-06,
      "loss": 0.1752,
      "step": 8099
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3668372892918503,
      "learning_rate": 2.912217705227075e-06,
      "loss": 0.1715,
      "step": 8100
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.331016076162309,
      "learning_rate": 2.911040616044079e-06,
      "loss": 0.1725,
      "step": 8101
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2166038551640277,
      "learning_rate": 2.9098636670992603e-06,
      "loss": 0.1275,
      "step": 8102
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3553323170045553,
      "learning_rate": 2.908686858471622e-06,
      "loss": 0.1456,
      "step": 8103
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3762522583748829,
      "learning_rate": 2.9075101902401737e-06,
      "loss": 0.1327,
      "step": 8104
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.455919710095317,
      "learning_rate": 2.9063336624839065e-06,
      "loss": 0.1996,
      "step": 8105
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.559671321365754,
      "learning_rate": 2.9051572752818035e-06,
      "loss": 0.1798,
      "step": 8106
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3350050039751369,
      "learning_rate": 2.9039810287128408e-06,
      "loss": 0.1717,
      "step": 8107
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5037140018234578,
      "learning_rate": 2.9028049228559813e-06,
      "loss": 0.1948,
      "step": 8108
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3383606511889683,
      "learning_rate": 2.901628957790182e-06,
      "loss": 0.2034,
      "step": 8109
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5419290032589072,
      "learning_rate": 2.9004531335943865e-06,
      "loss": 0.1888,
      "step": 8110
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0135988291228506,
      "learning_rate": 2.8992774503475373e-06,
      "loss": 0.1167,
      "step": 8111
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.22977531257775,
      "learning_rate": 2.8981019081285546e-06,
      "loss": 0.1707,
      "step": 8112
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4768854382936154,
      "learning_rate": 2.89692650701636e-06,
      "loss": 0.2437,
      "step": 8113
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2324231092387152,
      "learning_rate": 2.895751247089862e-06,
      "loss": 0.1577,
      "step": 8114
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3448048546626383,
      "learning_rate": 2.8945761284279583e-06,
      "loss": 0.2127,
      "step": 8115
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3054498520217608,
      "learning_rate": 2.8934011511095376e-06,
      "loss": 0.1833,
      "step": 8116
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1734087151844261,
      "learning_rate": 2.892226315213481e-06,
      "loss": 0.1632,
      "step": 8117
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4091496751575994,
      "learning_rate": 2.8910516208186578e-06,
      "loss": 0.2047,
      "step": 8118
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4823502947624343,
      "learning_rate": 2.88987706800393e-06,
      "loss": 0.24,
      "step": 8119
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.821066701982557,
      "learning_rate": 2.888702656848147e-06,
      "loss": 0.5307,
      "step": 8120
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4653192839245515,
      "learning_rate": 2.88752838743015e-06,
      "loss": 0.2012,
      "step": 8121
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6209104197072142,
      "learning_rate": 2.8863542598287775e-06,
      "loss": 0.1911,
      "step": 8122
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4629892858105946,
      "learning_rate": 2.885180274122843e-06,
      "loss": 0.2104,
      "step": 8123
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5341898147874118,
      "learning_rate": 2.8840064303911662e-06,
      "loss": 0.2034,
      "step": 8124
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3724469970331425,
      "learning_rate": 2.882832728712551e-06,
      "loss": 0.1604,
      "step": 8125
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.337702243188611,
      "learning_rate": 2.881659169165786e-06,
      "loss": 0.1573,
      "step": 8126
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.26593516140244,
      "learning_rate": 2.88048575182966e-06,
      "loss": 0.134,
      "step": 8127
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1783578134638426,
      "learning_rate": 2.879312476782949e-06,
      "loss": 0.1731,
      "step": 8128
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.216165217306844,
      "learning_rate": 2.878139344104416e-06,
      "loss": 0.1544,
      "step": 8129
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4986083017097345,
      "learning_rate": 2.8769663538728174e-06,
      "loss": 0.2236,
      "step": 8130
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6044003516611425,
      "learning_rate": 2.875793506166901e-06,
      "loss": 0.2208,
      "step": 8131
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.266545184443366,
      "learning_rate": 2.8746208010654e-06,
      "loss": 0.1649,
      "step": 8132
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2636217835007633,
      "learning_rate": 2.8734482386470486e-06,
      "loss": 0.1585,
      "step": 8133
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3715053948714575,
      "learning_rate": 2.872275818990558e-06,
      "loss": 0.1613,
      "step": 8134
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2228003457887464,
      "learning_rate": 2.871103542174637e-06,
      "loss": 0.1515,
      "step": 8135
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.612261722423232,
      "learning_rate": 2.8699314082779884e-06,
      "loss": 0.185,
      "step": 8136
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2198719025746336,
      "learning_rate": 2.868759417379295e-06,
      "loss": 0.1533,
      "step": 8137
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1465738644507792,
      "learning_rate": 2.867587569557241e-06,
      "loss": 0.1436,
      "step": 8138
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2981937212751469,
      "learning_rate": 2.866415864890494e-06,
      "loss": 0.1858,
      "step": 8139
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3482742108232029,
      "learning_rate": 2.865244303457715e-06,
      "loss": 0.1747,
      "step": 8140
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.60813444855071,
      "learning_rate": 2.8640728853375545e-06,
      "loss": 0.1933,
      "step": 8141
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3532145849527302,
      "learning_rate": 2.8629016106086515e-06,
      "loss": 0.1921,
      "step": 8142
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4498115384857877,
      "learning_rate": 2.8617304793496393e-06,
      "loss": 0.1859,
      "step": 8143
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3016485857741171,
      "learning_rate": 2.8605594916391388e-06,
      "loss": 0.1865,
      "step": 8144
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1928587714221008,
      "learning_rate": 2.859388647555762e-06,
      "loss": 0.1497,
      "step": 8145
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.82991062665454,
      "learning_rate": 2.8582179471781086e-06,
      "loss": 0.5546,
      "step": 8146
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2247550776926104,
      "learning_rate": 2.8570473905847764e-06,
      "loss": 0.1908,
      "step": 8147
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1224005157470753,
      "learning_rate": 2.855876977854345e-06,
      "loss": 0.1476,
      "step": 8148
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2108798789993558,
      "learning_rate": 2.854706709065389e-06,
      "loss": 0.1385,
      "step": 8149
    },
    {
      "epoch": 0.65,
      "grad_norm": 16.25821285418179,
      "learning_rate": 2.8535365842964713e-06,
      "loss": 0.4215,
      "step": 8150
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3856642732435283,
      "learning_rate": 2.8523666036261467e-06,
      "loss": 0.1813,
      "step": 8151
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3401070247680038,
      "learning_rate": 2.851196767132958e-06,
      "loss": 0.1782,
      "step": 8152
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.508413431160601,
      "learning_rate": 2.8500270748954417e-06,
      "loss": 0.1966,
      "step": 8153
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.72845395549302,
      "learning_rate": 2.8488575269921227e-06,
      "loss": 0.5491,
      "step": 8154
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4148164142055286,
      "learning_rate": 2.8476881235015126e-06,
      "loss": 0.1645,
      "step": 8155
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.425671154122458,
      "learning_rate": 2.846518864502124e-06,
      "loss": 0.1921,
      "step": 8156
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.278035632164817,
      "learning_rate": 2.8453497500724454e-06,
      "loss": 0.1646,
      "step": 8157
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3269113152611829,
      "learning_rate": 2.844180780290968e-06,
      "loss": 0.1378,
      "step": 8158
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.401812305703745,
      "learning_rate": 2.8430119552361668e-06,
      "loss": 0.2111,
      "step": 8159
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.38742239160579,
      "learning_rate": 2.841843274986509e-06,
      "loss": 0.1788,
      "step": 8160
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.372275527640766,
      "learning_rate": 2.8406747396204505e-06,
      "loss": 0.6411,
      "step": 8161
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4888021117244161,
      "learning_rate": 2.8395063492164387e-06,
      "loss": 0.1985,
      "step": 8162
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3641315841153145,
      "learning_rate": 2.8383381038529125e-06,
      "loss": 0.1613,
      "step": 8163
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3375085842299907,
      "learning_rate": 2.8371700036082987e-06,
      "loss": 0.1795,
      "step": 8164
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.312796766957599,
      "learning_rate": 2.8360020485610164e-06,
      "loss": 0.1685,
      "step": 8165
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.362974841797472,
      "learning_rate": 2.83483423878947e-06,
      "loss": 0.1694,
      "step": 8166
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3134154835935457,
      "learning_rate": 2.8336665743720657e-06,
      "loss": 0.1918,
      "step": 8167
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4466690825315403,
      "learning_rate": 2.832499055387187e-06,
      "loss": 0.162,
      "step": 8168
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2437395164720058,
      "learning_rate": 2.8313316819132107e-06,
      "loss": 0.1911,
      "step": 8169
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3252796866560008,
      "learning_rate": 2.8301644540285137e-06,
      "loss": 0.1846,
      "step": 8170
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4107240030359327,
      "learning_rate": 2.8289973718114476e-06,
      "loss": 0.1698,
      "step": 8171
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.471574696148342,
      "learning_rate": 2.8278304353403673e-06,
      "loss": 0.2125,
      "step": 8172
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3324876747937011,
      "learning_rate": 2.8266636446936126e-06,
      "loss": 0.2325,
      "step": 8173
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0812886995315816,
      "learning_rate": 2.825496999949511e-06,
      "loss": 0.1245,
      "step": 8174
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.420777595369157,
      "learning_rate": 2.8243305011863843e-06,
      "loss": 0.1818,
      "step": 8175
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4177489202985598,
      "learning_rate": 2.8231641484825435e-06,
      "loss": 0.1661,
      "step": 8176
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5638281700479613,
      "learning_rate": 2.821997941916289e-06,
      "loss": 0.2321,
      "step": 8177
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3840143267847742,
      "learning_rate": 2.820831881565912e-06,
      "loss": 0.2133,
      "step": 8178
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.255900802878477,
      "learning_rate": 2.8196659675096925e-06,
      "loss": 0.154,
      "step": 8179
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.413599801881487,
      "learning_rate": 2.818500199825902e-06,
      "loss": 0.1688,
      "step": 8180
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2990797834020438,
      "learning_rate": 2.8173345785928057e-06,
      "loss": 0.1799,
      "step": 8181
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3613083596592144,
      "learning_rate": 2.8161691038886486e-06,
      "loss": 0.1862,
      "step": 8182
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.307526131543236,
      "learning_rate": 2.815003775791679e-06,
      "loss": 0.567,
      "step": 8183
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.836551021659636,
      "learning_rate": 2.8138385943801248e-06,
      "loss": 0.4816,
      "step": 8184
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2042362627583352,
      "learning_rate": 2.812673559732211e-06,
      "loss": 0.1902,
      "step": 8185
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4214243612726447,
      "learning_rate": 2.8115086719261474e-06,
      "loss": 0.183,
      "step": 8186
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2489295891226895,
      "learning_rate": 2.810343931040138e-06,
      "loss": 0.1757,
      "step": 8187
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.219105045203035,
      "learning_rate": 2.8091793371523758e-06,
      "loss": 0.1803,
      "step": 8188
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.90324681193005,
      "learning_rate": 2.8080148903410392e-06,
      "loss": 0.4731,
      "step": 8189
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3884067642615159,
      "learning_rate": 2.806850590684309e-06,
      "loss": 0.2018,
      "step": 8190
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.08733715561001,
      "learning_rate": 2.80568643826034e-06,
      "loss": 0.6197,
      "step": 8191
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2197716814846384,
      "learning_rate": 2.8045224331472905e-06,
      "loss": 0.2167,
      "step": 8192
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3637476702772617,
      "learning_rate": 2.8033585754233016e-06,
      "loss": 0.1425,
      "step": 8193
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3124378630396867,
      "learning_rate": 2.8021948651665076e-06,
      "loss": 0.1562,
      "step": 8194
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4023071019221858,
      "learning_rate": 2.801031302455032e-06,
      "loss": 0.2013,
      "step": 8195
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4764818449105874,
      "learning_rate": 2.7998678873669865e-06,
      "loss": 0.2047,
      "step": 8196
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2809537032740461,
      "learning_rate": 2.798704619980477e-06,
      "loss": 0.1484,
      "step": 8197
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4262301916846456,
      "learning_rate": 2.797541500373595e-06,
      "loss": 0.1735,
      "step": 8198
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.660828530205685,
      "learning_rate": 2.796378528624426e-06,
      "loss": 0.1954,
      "step": 8199
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5164652248087216,
      "learning_rate": 2.7952157048110406e-06,
      "loss": 0.1801,
      "step": 8200
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3383935882724372,
      "learning_rate": 2.7940530290115093e-06,
      "loss": 0.1811,
      "step": 8201
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4938821995861717,
      "learning_rate": 2.7928905013038774e-06,
      "loss": 0.2521,
      "step": 8202
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3692213093388257,
      "learning_rate": 2.791728121766196e-06,
      "loss": 0.1803,
      "step": 8203
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.447852193673213,
      "learning_rate": 2.7905658904764994e-06,
      "loss": 0.1818,
      "step": 8204
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.326692920546855,
      "learning_rate": 2.7894038075128038e-06,
      "loss": 0.1506,
      "step": 8205
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.347939193197956,
      "learning_rate": 2.7882418729531307e-06,
      "loss": 0.1646,
      "step": 8206
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.811364101410414,
      "learning_rate": 2.787080086875482e-06,
      "loss": 0.3927,
      "step": 8207
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3363433683173385,
      "learning_rate": 2.7859184493578517e-06,
      "loss": 0.1526,
      "step": 8208
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.880797734402494,
      "learning_rate": 2.784756960478225e-06,
      "loss": 0.6027,
      "step": 8209
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4408869393439778,
      "learning_rate": 2.7835956203145754e-06,
      "loss": 0.2133,
      "step": 8210
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.207624716735158,
      "learning_rate": 2.7824344289448646e-06,
      "loss": 0.1658,
      "step": 8211
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3730893811206233,
      "learning_rate": 2.7812733864470536e-06,
      "loss": 0.1733,
      "step": 8212
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.743111497454227,
      "learning_rate": 2.780112492899081e-06,
      "loss": 0.4282,
      "step": 8213
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3203663741074512,
      "learning_rate": 2.7789517483788807e-06,
      "loss": 0.2028,
      "step": 8214
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2203683058740593,
      "learning_rate": 2.777791152964383e-06,
      "loss": 0.171,
      "step": 8215
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4042304145668758,
      "learning_rate": 2.776630706733494e-06,
      "loss": 0.1841,
      "step": 8216
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2677535200978691,
      "learning_rate": 2.7754704097641246e-06,
      "loss": 0.1756,
      "step": 8217
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4118838062054786,
      "learning_rate": 2.7743102621341666e-06,
      "loss": 0.2069,
      "step": 8218
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3425218361546978,
      "learning_rate": 2.7731502639215037e-06,
      "loss": 0.1897,
      "step": 8219
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.272412169417104,
      "learning_rate": 2.7719904152040112e-06,
      "loss": 0.1637,
      "step": 8220
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.491814432811376,
      "learning_rate": 2.7708307160595526e-06,
      "loss": 0.1903,
      "step": 8221
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3821696297160513,
      "learning_rate": 2.7696711665659825e-06,
      "loss": 0.1684,
      "step": 8222
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5353597017467464,
      "learning_rate": 2.768511766801144e-06,
      "loss": 0.1978,
      "step": 8223
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4580019435728244,
      "learning_rate": 2.7673525168428716e-06,
      "loss": 0.1835,
      "step": 8224
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2614078371469561,
      "learning_rate": 2.7661934167689887e-06,
      "loss": 0.1713,
      "step": 8225
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3411207250056796,
      "learning_rate": 2.7650344666573125e-06,
      "loss": 0.1797,
      "step": 8226
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2989201034624076,
      "learning_rate": 2.7638756665856402e-06,
      "loss": 0.1708,
      "step": 8227
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4212189933161887,
      "learning_rate": 2.762717016631773e-06,
      "loss": 0.1769,
      "step": 8228
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.394863677016402,
      "learning_rate": 2.7615585168734915e-06,
      "loss": 0.5518,
      "step": 8229
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4444900450669798,
      "learning_rate": 2.760400167388566e-06,
      "loss": 0.2489,
      "step": 8230
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4244509237057872,
      "learning_rate": 2.759241968254765e-06,
      "loss": 0.1775,
      "step": 8231
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.386791454415684,
      "learning_rate": 2.7580839195498397e-06,
      "loss": 0.6115,
      "step": 8232
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3149351229111497,
      "learning_rate": 2.7569260213515336e-06,
      "loss": 0.1761,
      "step": 8233
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.442269624319686,
      "learning_rate": 2.755768273737578e-06,
      "loss": 0.2038,
      "step": 8234
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3286050596293064,
      "learning_rate": 2.754610676785702e-06,
      "loss": 0.1709,
      "step": 8235
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.297555810338083,
      "learning_rate": 2.7534532305736094e-06,
      "loss": 0.1648,
      "step": 8236
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4306340058496405,
      "learning_rate": 2.752295935179011e-06,
      "loss": 0.2198,
      "step": 8237
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4975989650519839,
      "learning_rate": 2.7511387906795996e-06,
      "loss": 0.195,
      "step": 8238
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.344218308312577,
      "learning_rate": 2.7499817971530502e-06,
      "loss": 0.1502,
      "step": 8239
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3996292929523575,
      "learning_rate": 2.748824954677042e-06,
      "loss": 0.2092,
      "step": 8240
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5792798855200623,
      "learning_rate": 2.747668263329235e-06,
      "loss": 0.2384,
      "step": 8241
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4674512480717126,
      "learning_rate": 2.746511723187283e-06,
      "loss": 0.1591,
      "step": 8242
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.325400646463198,
      "learning_rate": 2.745355334328826e-06,
      "loss": 0.1643,
      "step": 8243
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2124832975218465,
      "learning_rate": 2.7441990968314967e-06,
      "loss": 0.1519,
      "step": 8244
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.384503602650137,
      "learning_rate": 2.7430430107729144e-06,
      "loss": 0.1704,
      "step": 8245
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3893304988749504,
      "learning_rate": 2.741887076230698e-06,
      "loss": 0.2218,
      "step": 8246
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4415206907102687,
      "learning_rate": 2.7407312932824416e-06,
      "loss": 0.1896,
      "step": 8247
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4453486737555465,
      "learning_rate": 2.739575662005737e-06,
      "loss": 0.1939,
      "step": 8248
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5658119240728812,
      "learning_rate": 2.738420182478171e-06,
      "loss": 0.1918,
      "step": 8249
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3169004691334731,
      "learning_rate": 2.7372648547773063e-06,
      "loss": 0.1835,
      "step": 8250
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4347636850761758,
      "learning_rate": 2.7361096789807097e-06,
      "loss": 0.1774,
      "step": 8251
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.453222150367343,
      "learning_rate": 2.734954655165929e-06,
      "loss": 0.6196,
      "step": 8252
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.246336446218398,
      "learning_rate": 2.7337997834105058e-06,
      "loss": 0.1807,
      "step": 8253
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5502133061418724,
      "learning_rate": 2.7326450637919694e-06,
      "loss": 0.189,
      "step": 8254
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3116290106849469,
      "learning_rate": 2.7314904963878397e-06,
      "loss": 0.1722,
      "step": 8255
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3606258011633294,
      "learning_rate": 2.7303360812756265e-06,
      "loss": 0.1673,
      "step": 8256
    },
    {
      "epoch": 0.66,
      "grad_norm": 15.366059486873533,
      "learning_rate": 2.729181818532829e-06,
      "loss": 0.6163,
      "step": 8257
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2952209935395895,
      "learning_rate": 2.7280277082369356e-06,
      "loss": 0.1254,
      "step": 8258
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2057614639652108,
      "learning_rate": 2.7268737504654242e-06,
      "loss": 0.1557,
      "step": 8259
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3392405405592644,
      "learning_rate": 2.7257199452957693e-06,
      "loss": 0.1514,
      "step": 8260
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.687759815561765,
      "learning_rate": 2.724566292805422e-06,
      "loss": 0.6903,
      "step": 8261
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.295979724309028,
      "learning_rate": 2.7234127930718354e-06,
      "loss": 0.5088,
      "step": 8262
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4013456438350602,
      "learning_rate": 2.722259446172445e-06,
      "loss": 0.1751,
      "step": 8263
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.662885107094121,
      "learning_rate": 2.7211062521846798e-06,
      "loss": 0.3866,
      "step": 8264
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4904595686156994,
      "learning_rate": 2.719953211185957e-06,
      "loss": 0.1731,
      "step": 8265
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5217263606407998,
      "learning_rate": 2.718800323253683e-06,
      "loss": 0.1701,
      "step": 8266
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3857532103780656,
      "learning_rate": 2.717647588465255e-06,
      "loss": 0.1918,
      "step": 8267
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4207325461457037,
      "learning_rate": 2.71649500689806e-06,
      "loss": 0.2283,
      "step": 8268
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4075779665781691,
      "learning_rate": 2.715342578629474e-06,
      "loss": 0.2127,
      "step": 8269
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.337208250402908,
      "learning_rate": 2.71419030373686e-06,
      "loss": 0.165,
      "step": 8270
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4014631208363169,
      "learning_rate": 2.7130381822975816e-06,
      "loss": 0.182,
      "step": 8271
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3454458999667605,
      "learning_rate": 2.7118862143889745e-06,
      "loss": 0.1751,
      "step": 8272
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5390851863572763,
      "learning_rate": 2.7107344000883815e-06,
      "loss": 0.204,
      "step": 8273
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5298681987468916,
      "learning_rate": 2.7095827394731255e-06,
      "loss": 0.1912,
      "step": 8274
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2659084111637122,
      "learning_rate": 2.7084312326205164e-06,
      "loss": 0.1502,
      "step": 8275
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.949398461314595,
      "learning_rate": 2.7072798796078637e-06,
      "loss": 0.5114,
      "step": 8276
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.434917861837227,
      "learning_rate": 2.7061286805124597e-06,
      "loss": 0.1865,
      "step": 8277
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2556630682747942,
      "learning_rate": 2.704977635411587e-06,
      "loss": 0.1752,
      "step": 8278
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2628487014630823,
      "learning_rate": 2.7038267443825174e-06,
      "loss": 0.1558,
      "step": 8279
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.388196570661171,
      "learning_rate": 2.7026760075025195e-06,
      "loss": 0.1524,
      "step": 8280
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3276699165992805,
      "learning_rate": 2.701525424848838e-06,
      "loss": 0.1766,
      "step": 8281
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2410157063968865,
      "learning_rate": 2.7003749964987215e-06,
      "loss": 0.161,
      "step": 8282
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2036379745242725,
      "learning_rate": 2.6992247225294e-06,
      "loss": 0.1738,
      "step": 8283
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.663086822592764,
      "learning_rate": 2.6980746030180904e-06,
      "loss": 0.5636,
      "step": 8284
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9928667480651018,
      "learning_rate": 2.6969246380420088e-06,
      "loss": 0.1242,
      "step": 8285
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3942696023362178,
      "learning_rate": 2.6957748276783547e-06,
      "loss": 0.1748,
      "step": 8286
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.534515750358792,
      "learning_rate": 2.694625172004318e-06,
      "loss": 0.1972,
      "step": 8287
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2741706467113365,
      "learning_rate": 2.693475671097079e-06,
      "loss": 0.1876,
      "step": 8288
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1634624021256645,
      "learning_rate": 2.6923263250338056e-06,
      "loss": 0.1495,
      "step": 8289
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.739318096948371,
      "learning_rate": 2.691177133891658e-06,
      "loss": 0.5646,
      "step": 8290
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4530826292526324,
      "learning_rate": 2.6900280977477855e-06,
      "loss": 0.1933,
      "step": 8291
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2264394114662693,
      "learning_rate": 2.688879216679326e-06,
      "loss": 0.1251,
      "step": 8292
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4737147977735179,
      "learning_rate": 2.6877304907634045e-06,
      "loss": 0.2399,
      "step": 8293
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2986615672670123,
      "learning_rate": 2.6865819200771455e-06,
      "loss": 0.1724,
      "step": 8294
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2335084138572132,
      "learning_rate": 2.685433504697647e-06,
      "loss": 0.1578,
      "step": 8295
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3642247260447673,
      "learning_rate": 2.684285244702013e-06,
      "loss": 0.2141,
      "step": 8296
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.561878618907496,
      "learning_rate": 2.6831371401673257e-06,
      "loss": 0.1974,
      "step": 8297
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6222764211564789,
      "learning_rate": 2.681989191170663e-06,
      "loss": 0.1931,
      "step": 8298
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1706686783812512,
      "learning_rate": 2.680841397789089e-06,
      "loss": 0.156,
      "step": 8299
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2836917250516335,
      "learning_rate": 2.6796937600996587e-06,
      "loss": 0.2042,
      "step": 8300
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2446628252941623,
      "learning_rate": 2.6785462781794158e-06,
      "loss": 0.1666,
      "step": 8301
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.363875117129697,
      "learning_rate": 2.6773989521053955e-06,
      "loss": 0.1765,
      "step": 8302
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4197133687724275,
      "learning_rate": 2.676251781954621e-06,
      "loss": 0.2067,
      "step": 8303
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4393984741123935,
      "learning_rate": 2.6751047678041033e-06,
      "loss": 0.2207,
      "step": 8304
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2378351179754488,
      "learning_rate": 2.67395790973085e-06,
      "loss": 0.169,
      "step": 8305
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2334454725461579,
      "learning_rate": 2.672811207811846e-06,
      "loss": 0.1583,
      "step": 8306
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3576199813228453,
      "learning_rate": 2.6716646621240794e-06,
      "loss": 0.1632,
      "step": 8307
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.344384291012976,
      "learning_rate": 2.6705182727445194e-06,
      "loss": 0.1721,
      "step": 8308
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4233836859669649,
      "learning_rate": 2.6693720397501228e-06,
      "loss": 0.1999,
      "step": 8309
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3743066738517795,
      "learning_rate": 2.668225963217844e-06,
      "loss": 0.1947,
      "step": 8310
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6570564427600556,
      "learning_rate": 2.6670800432246217e-06,
      "loss": 0.2118,
      "step": 8311
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3590532377170166,
      "learning_rate": 2.6659342798473846e-06,
      "loss": 0.1893,
      "step": 8312
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.201603673134636,
      "learning_rate": 2.664788673163051e-06,
      "loss": 0.1878,
      "step": 8313
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6083432237818807,
      "learning_rate": 2.66364322324853e-06,
      "loss": 0.2138,
      "step": 8314
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5048847965041539,
      "learning_rate": 2.662497930180715e-06,
      "loss": 0.213,
      "step": 8315
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1058786738734272,
      "learning_rate": 2.661352794036501e-06,
      "loss": 0.1404,
      "step": 8316
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.69329960554202,
      "learning_rate": 2.660207814892758e-06,
      "loss": 0.5211,
      "step": 8317
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4223825603456628,
      "learning_rate": 2.659062992826351e-06,
      "loss": 0.1947,
      "step": 8318
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7434518129438166,
      "learning_rate": 2.657918327914141e-06,
      "loss": 0.2038,
      "step": 8319
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3918002521049717,
      "learning_rate": 2.6567738202329684e-06,
      "loss": 0.1935,
      "step": 8320
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2687108962909985,
      "learning_rate": 2.6556294698596695e-06,
      "loss": 0.1636,
      "step": 8321
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2172700294627605,
      "learning_rate": 2.6544852768710685e-06,
      "loss": 0.1513,
      "step": 8322
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.385810546657486,
      "learning_rate": 2.6533412413439763e-06,
      "loss": 0.1997,
      "step": 8323
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3369828981952543,
      "learning_rate": 2.6521973633551957e-06,
      "loss": 0.1925,
      "step": 8324
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3363980843922303,
      "learning_rate": 2.6510536429815224e-06,
      "loss": 0.1887,
      "step": 8325
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.369907471551549,
      "learning_rate": 2.6499100802997336e-06,
      "loss": 0.1881,
      "step": 8326
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3650807653077166,
      "learning_rate": 2.6487666753866005e-06,
      "loss": 0.1633,
      "step": 8327
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.507369354758025,
      "learning_rate": 2.6476234283188873e-06,
      "loss": 0.221,
      "step": 8328
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.277527879111273,
      "learning_rate": 2.646480339173337e-06,
      "loss": 0.1689,
      "step": 8329
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3825803415055011,
      "learning_rate": 2.6453374080266947e-06,
      "loss": 0.2296,
      "step": 8330
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4198192025347276,
      "learning_rate": 2.6441946349556864e-06,
      "loss": 0.166,
      "step": 8331
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5289788123477608,
      "learning_rate": 2.6430520200370303e-06,
      "loss": 0.2076,
      "step": 8332
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.358464394048185,
      "learning_rate": 2.6419095633474323e-06,
      "loss": 0.1848,
      "step": 8333
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.437009421412895,
      "learning_rate": 2.640767264963591e-06,
      "loss": 0.1874,
      "step": 8334
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4057292957038556,
      "learning_rate": 2.639625124962192e-06,
      "loss": 0.1916,
      "step": 8335
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.308352056010595,
      "learning_rate": 2.6384831434199087e-06,
      "loss": 0.1742,
      "step": 8336
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.3940799343030434,
      "learning_rate": 2.637341320413408e-06,
      "loss": 0.5314,
      "step": 8337
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.424853640928415,
      "learning_rate": 2.63619965601934e-06,
      "loss": 0.1846,
      "step": 8338
    },
    {
      "epoch": 0.67,
      "grad_norm": 11.397831585139151,
      "learning_rate": 2.6350581503143564e-06,
      "loss": 0.5994,
      "step": 8339
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7567645369929097,
      "learning_rate": 2.63391680337508e-06,
      "loss": 0.1903,
      "step": 8340
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3748505507214188,
      "learning_rate": 2.632775615278139e-06,
      "loss": 0.1772,
      "step": 8341
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4305998248238738,
      "learning_rate": 2.6316345861001445e-06,
      "loss": 0.226,
      "step": 8342
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.862078336036875,
      "learning_rate": 2.6304937159176956e-06,
      "loss": 0.6725,
      "step": 8343
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.224074036050416,
      "learning_rate": 2.6293530048073824e-06,
      "loss": 0.1759,
      "step": 8344
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2793527014775796,
      "learning_rate": 2.6282124528457852e-06,
      "loss": 0.1644,
      "step": 8345
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4620300834513775,
      "learning_rate": 2.6270720601094723e-06,
      "loss": 0.1828,
      "step": 8346
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3350169251144521,
      "learning_rate": 2.6259318266750024e-06,
      "loss": 0.1594,
      "step": 8347
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6145892461887903,
      "learning_rate": 2.624791752618921e-06,
      "loss": 0.1765,
      "step": 8348
    },
    {
      "epoch": 0.67,
      "grad_norm": 17.10406048172628,
      "learning_rate": 2.6236518380177644e-06,
      "loss": 0.6177,
      "step": 8349
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2703090655014542,
      "learning_rate": 2.622512082948063e-06,
      "loss": 0.1867,
      "step": 8350
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4865630127432794,
      "learning_rate": 2.6213724874863276e-06,
      "loss": 0.1572,
      "step": 8351
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5149217946291809,
      "learning_rate": 2.6202330517090624e-06,
      "loss": 0.2068,
      "step": 8352
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3783866237694922,
      "learning_rate": 2.619093775692767e-06,
      "loss": 0.1633,
      "step": 8353
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.310155360866443,
      "learning_rate": 2.617954659513916e-06,
      "loss": 0.2063,
      "step": 8354
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3282874736984818,
      "learning_rate": 2.6168157032489883e-06,
      "loss": 0.2007,
      "step": 8355
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4036790059330233,
      "learning_rate": 2.6156769069744425e-06,
      "loss": 0.2154,
      "step": 8356
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3221212302242065,
      "learning_rate": 2.6145382707667307e-06,
      "loss": 0.1876,
      "step": 8357
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4435399543818757,
      "learning_rate": 2.6133997947022904e-06,
      "loss": 0.1774,
      "step": 8358
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.635382291907966,
      "learning_rate": 2.6122614788575563e-06,
      "loss": 0.1842,
      "step": 8359
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2805976058806852,
      "learning_rate": 2.611123323308943e-06,
      "loss": 0.1928,
      "step": 8360
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3746477688795287,
      "learning_rate": 2.6099853281328556e-06,
      "loss": 0.1732,
      "step": 8361
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3703191102077759,
      "learning_rate": 2.6088474934056997e-06,
      "loss": 0.187,
      "step": 8362
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1730626745601922,
      "learning_rate": 2.607709819203852e-06,
      "loss": 0.153,
      "step": 8363
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.012358728915169,
      "learning_rate": 2.606572305603694e-06,
      "loss": 0.5336,
      "step": 8364
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2827423625753318,
      "learning_rate": 2.605434952681589e-06,
      "loss": 0.1529,
      "step": 8365
    },
    {
      "epoch": 0.67,
      "grad_norm": 12.371205268757304,
      "learning_rate": 2.604297760513892e-06,
      "loss": 0.5584,
      "step": 8366
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.483951994621688,
      "learning_rate": 2.603160729176944e-06,
      "loss": 0.1862,
      "step": 8367
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5264475948639284,
      "learning_rate": 2.6020238587470786e-06,
      "loss": 0.2238,
      "step": 8368
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4742128626067894,
      "learning_rate": 2.6008871493006173e-06,
      "loss": 0.2212,
      "step": 8369
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.389661631413501,
      "learning_rate": 2.5997506009138707e-06,
      "loss": 0.6088,
      "step": 8370
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3923283647070297,
      "learning_rate": 2.598614213663139e-06,
      "loss": 0.2079,
      "step": 8371
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3840309042318228,
      "learning_rate": 2.597477987624708e-06,
      "loss": 0.1836,
      "step": 8372
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3775238701544763,
      "learning_rate": 2.596341922874864e-06,
      "loss": 0.1763,
      "step": 8373
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3548391936560993,
      "learning_rate": 2.5952060194898655e-06,
      "loss": 0.152,
      "step": 8374
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2664372152024646,
      "learning_rate": 2.594070277545975e-06,
      "loss": 0.1671,
      "step": 8375
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3257243520647308,
      "learning_rate": 2.5929346971194356e-06,
      "loss": 0.1401,
      "step": 8376
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.643772094137073,
      "learning_rate": 2.5917992782864844e-06,
      "loss": 0.5311,
      "step": 8377
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.537068809858651,
      "learning_rate": 2.5906640211233435e-06,
      "loss": 0.1946,
      "step": 8378
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2784505933322527,
      "learning_rate": 2.5895289257062274e-06,
      "loss": 0.1377,
      "step": 8379
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3036414254508844,
      "learning_rate": 2.5883939921113373e-06,
      "loss": 0.1609,
      "step": 8380
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.983427887120907,
      "learning_rate": 2.5872592204148654e-06,
      "loss": 0.6134,
      "step": 8381
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.300533596061534,
      "learning_rate": 2.5861246106929926e-06,
      "loss": 0.1336,
      "step": 8382
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4279479926840726,
      "learning_rate": 2.5849901630218867e-06,
      "loss": 0.1665,
      "step": 8383
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.293795834205817,
      "learning_rate": 2.5838558774777125e-06,
      "loss": 0.1961,
      "step": 8384
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2929536969861828,
      "learning_rate": 2.582721754136609e-06,
      "loss": 0.1649,
      "step": 8385
    },
    {
      "epoch": 0.67,
      "grad_norm": 13.214554928506967,
      "learning_rate": 2.5815877930747207e-06,
      "loss": 0.6791,
      "step": 8386
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.727042803026171,
      "learning_rate": 2.580453994368173e-06,
      "loss": 0.4667,
      "step": 8387
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.270006033304784,
      "learning_rate": 2.5793203580930758e-06,
      "loss": 0.16,
      "step": 8388
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3575929154072648,
      "learning_rate": 2.5781868843255398e-06,
      "loss": 0.1784,
      "step": 8389
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3802315973198427,
      "learning_rate": 2.5770535731416556e-06,
      "loss": 0.1667,
      "step": 8390
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5928041467690033,
      "learning_rate": 2.5759204246175062e-06,
      "loss": 0.2109,
      "step": 8391
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2908215368502198,
      "learning_rate": 2.574787438829164e-06,
      "loss": 0.1889,
      "step": 8392
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3239839347610995,
      "learning_rate": 2.573654615852688e-06,
      "loss": 0.1905,
      "step": 8393
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.2647769965038815,
      "learning_rate": 2.572521955764128e-06,
      "loss": 0.536,
      "step": 8394
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.5582608544364,
      "learning_rate": 2.5713894586395282e-06,
      "loss": 0.5374,
      "step": 8395
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2040471192119289,
      "learning_rate": 2.57025712455491e-06,
      "loss": 0.1779,
      "step": 8396
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.69235247895066,
      "learning_rate": 2.5691249535862907e-06,
      "loss": 0.4164,
      "step": 8397
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4703416391993587,
      "learning_rate": 2.567992945809683e-06,
      "loss": 0.1958,
      "step": 8398
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.382781051457089,
      "learning_rate": 2.5668611013010734e-06,
      "loss": 0.1752,
      "step": 8399
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.287516538348383,
      "learning_rate": 2.5657294201364526e-06,
      "loss": 0.4783,
      "step": 8400
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7374085538513262,
      "learning_rate": 2.5645979023917906e-06,
      "loss": 0.1619,
      "step": 8401
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2865812796166598,
      "learning_rate": 2.56346654814305e-06,
      "loss": 0.1843,
      "step": 8402
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3202424365720382,
      "learning_rate": 2.5623353574661814e-06,
      "loss": 0.1907,
      "step": 8403
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5169410652836222,
      "learning_rate": 2.5612043304371304e-06,
      "loss": 0.1658,
      "step": 8404
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3246159512880782,
      "learning_rate": 2.560073467131819e-06,
      "loss": 0.1641,
      "step": 8405
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.450833532937457,
      "learning_rate": 2.558942767626167e-06,
      "loss": 0.1739,
      "step": 8406
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4894617836154485,
      "learning_rate": 2.5578122319960865e-06,
      "loss": 0.1637,
      "step": 8407
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3930469140727428,
      "learning_rate": 2.556681860317467e-06,
      "loss": 0.2119,
      "step": 8408
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2092638310034658,
      "learning_rate": 2.5555516526661996e-06,
      "loss": 0.1608,
      "step": 8409
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.439167988330913,
      "learning_rate": 2.554421609118155e-06,
      "loss": 0.2016,
      "step": 8410
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3370190346669266,
      "learning_rate": 2.553291729749198e-06,
      "loss": 0.1882,
      "step": 8411
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3923730892049928,
      "learning_rate": 2.5521620146351806e-06,
      "loss": 0.1318,
      "step": 8412
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5155836519904236,
      "learning_rate": 2.5510324638519435e-06,
      "loss": 0.1867,
      "step": 8413
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4016559992356261,
      "learning_rate": 2.5499030774753176e-06,
      "loss": 0.1916,
      "step": 8414
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.261624161534118,
      "learning_rate": 2.5487738555811215e-06,
      "loss": 0.1456,
      "step": 8415
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.088674480367247,
      "learning_rate": 2.5476447982451636e-06,
      "loss": 0.1399,
      "step": 8416
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.918340357854911,
      "learning_rate": 2.5465159055432383e-06,
      "loss": 0.5938,
      "step": 8417
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3824811142382467,
      "learning_rate": 2.545387177551138e-06,
      "loss": 0.222,
      "step": 8418
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3849157281916746,
      "learning_rate": 2.54425861434463e-06,
      "loss": 0.1821,
      "step": 8419
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3411100337093516,
      "learning_rate": 2.5431302159994835e-06,
      "loss": 0.2,
      "step": 8420
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.7547661560311525,
      "learning_rate": 2.542001982591452e-06,
      "loss": 0.5695,
      "step": 8421
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2408585504338,
      "learning_rate": 2.5408739141962704e-06,
      "loss": 0.1459,
      "step": 8422
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.338497924203863,
      "learning_rate": 2.539746010889676e-06,
      "loss": 0.1728,
      "step": 8423
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3126759660134741,
      "learning_rate": 2.5386182727473862e-06,
      "loss": 0.1942,
      "step": 8424
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2820197881024753,
      "learning_rate": 2.5374906998451094e-06,
      "loss": 0.1822,
      "step": 8425
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4291055005477529,
      "learning_rate": 2.536363292258543e-06,
      "loss": 0.1978,
      "step": 8426
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4594811843319682,
      "learning_rate": 2.5352360500633733e-06,
      "loss": 0.1982,
      "step": 8427
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4581060922940934,
      "learning_rate": 2.5341089733352737e-06,
      "loss": 0.1899,
      "step": 8428
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.340334290982926,
      "learning_rate": 2.5329820621499135e-06,
      "loss": 0.1769,
      "step": 8429
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.351724274707295,
      "learning_rate": 2.5318553165829407e-06,
      "loss": 0.2084,
      "step": 8430
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.431960827401473,
      "learning_rate": 2.530728736709997e-06,
      "loss": 0.6228,
      "step": 8431
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3914139408565374,
      "learning_rate": 2.5296023226067183e-06,
      "loss": 0.196,
      "step": 8432
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.389280058175246,
      "learning_rate": 2.528476074348717e-06,
      "loss": 0.2093,
      "step": 8433
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4076540112760858,
      "learning_rate": 2.5273499920116073e-06,
      "loss": 0.1778,
      "step": 8434
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3230740434138704,
      "learning_rate": 2.5262240756709838e-06,
      "loss": 0.1811,
      "step": 8435
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2443327889627611,
      "learning_rate": 2.525098325402434e-06,
      "loss": 0.1666,
      "step": 8436
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2163024197581889,
      "learning_rate": 2.5239727412815317e-06,
      "loss": 0.1502,
      "step": 8437
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.572666608384148,
      "learning_rate": 2.522847323383842e-06,
      "loss": 0.3843,
      "step": 8438
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.295502398616411,
      "learning_rate": 2.5217220717849167e-06,
      "loss": 0.1554,
      "step": 8439
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.979633956432414,
      "learning_rate": 2.5205969865602974e-06,
      "loss": 0.3847,
      "step": 8440
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3255602677274645,
      "learning_rate": 2.519472067785515e-06,
      "loss": 0.1904,
      "step": 8441
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.329054071426642,
      "learning_rate": 2.518347315536086e-06,
      "loss": 0.1864,
      "step": 8442
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.466200607498436,
      "learning_rate": 2.5172227298875253e-06,
      "loss": 0.6387,
      "step": 8443
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3573221609395365,
      "learning_rate": 2.5160983109153207e-06,
      "loss": 0.1757,
      "step": 8444
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3763026235568596,
      "learning_rate": 2.514974058694965e-06,
      "loss": 0.1839,
      "step": 8445
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3155024991664774,
      "learning_rate": 2.5138499733019294e-06,
      "loss": 0.1522,
      "step": 8446
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2972188218378276,
      "learning_rate": 2.512726054811678e-06,
      "loss": 0.152,
      "step": 8447
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.375938653985904,
      "learning_rate": 2.5116023032996635e-06,
      "loss": 0.1799,
      "step": 8448
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.365649460624493,
      "learning_rate": 2.510478718841325e-06,
      "loss": 0.179,
      "step": 8449
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5045482140220925,
      "learning_rate": 2.5093553015120937e-06,
      "loss": 0.2217,
      "step": 8450
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.468083324936883,
      "learning_rate": 2.508232051387385e-06,
      "loss": 0.2069,
      "step": 8451
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3331638487191784,
      "learning_rate": 2.5071089685426127e-06,
      "loss": 0.1576,
      "step": 8452
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2545180613070355,
      "learning_rate": 2.5059860530531644e-06,
      "loss": 0.1818,
      "step": 8453
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.336553781113412,
      "learning_rate": 2.5048633049944305e-06,
      "loss": 0.1693,
      "step": 8454
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4923150824370275,
      "learning_rate": 2.5037407244417834e-06,
      "loss": 0.1943,
      "step": 8455
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4859046782697194,
      "learning_rate": 2.5026183114705844e-06,
      "loss": 0.1779,
      "step": 8456
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4170496787062508,
      "learning_rate": 2.501496066156185e-06,
      "loss": 0.1691,
      "step": 8457
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.696476482692896,
      "learning_rate": 2.5003739885739253e-06,
      "loss": 0.587,
      "step": 8458
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5324499197982238,
      "learning_rate": 2.4992520787991328e-06,
      "loss": 0.2156,
      "step": 8459
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3942672615624903,
      "learning_rate": 2.498130336907125e-06,
      "loss": 0.1787,
      "step": 8460
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3526214712013132,
      "learning_rate": 2.497008762973209e-06,
      "loss": 0.1915,
      "step": 8461
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4260154575269086,
      "learning_rate": 2.495887357072676e-06,
      "loss": 0.1763,
      "step": 8462
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2826193221392896,
      "learning_rate": 2.494766119280816e-06,
      "loss": 0.1677,
      "step": 8463
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5360239255235244,
      "learning_rate": 2.4936450496728927e-06,
      "loss": 0.1727,
      "step": 8464
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1597837442557963,
      "learning_rate": 2.492524148324173e-06,
      "loss": 0.1325,
      "step": 8465
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.293857847534176,
      "learning_rate": 2.491403415309907e-06,
      "loss": 0.1734,
      "step": 8466
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3485591114822921,
      "learning_rate": 2.490282850705326e-06,
      "loss": 0.1697,
      "step": 8467
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2351140589428389,
      "learning_rate": 2.489162454585664e-06,
      "loss": 0.1731,
      "step": 8468
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3010663457449723,
      "learning_rate": 2.488042227026134e-06,
      "loss": 0.2016,
      "step": 8469
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4550253190322415,
      "learning_rate": 2.4869221681019394e-06,
      "loss": 0.1684,
      "step": 8470
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.412379849386844,
      "learning_rate": 2.4858022778882746e-06,
      "loss": 0.5663,
      "step": 8471
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4494060978272503,
      "learning_rate": 2.4846825564603204e-06,
      "loss": 0.2182,
      "step": 8472
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4083347938367576,
      "learning_rate": 2.4835630038932475e-06,
      "loss": 0.1949,
      "step": 8473
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9480023195484835,
      "learning_rate": 2.4824436202622147e-06,
      "loss": 0.1383,
      "step": 8474
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3617399392324856,
      "learning_rate": 2.4813244056423692e-06,
      "loss": 0.1653,
      "step": 8475
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8211578636303276,
      "learning_rate": 2.4802053601088462e-06,
      "loss": 0.248,
      "step": 8476
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.26647954245297,
      "learning_rate": 2.479086483736775e-06,
      "loss": 0.4828,
      "step": 8477
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3162224897957153,
      "learning_rate": 2.477967776601263e-06,
      "loss": 0.1739,
      "step": 8478
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0629831172172495,
      "learning_rate": 2.476849238777418e-06,
      "loss": 0.1252,
      "step": 8479
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5453624777820747,
      "learning_rate": 2.4757308703403275e-06,
      "loss": 0.1896,
      "step": 8480
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4116068290697115,
      "learning_rate": 2.4746126713650724e-06,
      "loss": 0.1993,
      "step": 8481
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3818204282674493,
      "learning_rate": 2.47349464192672e-06,
      "loss": 0.1652,
      "step": 8482
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2422014901595066,
      "learning_rate": 2.472376782100328e-06,
      "loss": 0.1694,
      "step": 8483
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3860698153181032,
      "learning_rate": 2.47125909196094e-06,
      "loss": 0.1825,
      "step": 8484
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2190229467045905,
      "learning_rate": 2.4701415715835917e-06,
      "loss": 0.1966,
      "step": 8485
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.556590601919081,
      "learning_rate": 2.4690242210433046e-06,
      "loss": 0.2251,
      "step": 8486
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2690971971154463,
      "learning_rate": 2.4679070404150883e-06,
      "loss": 0.1828,
      "step": 8487
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4563588891784556,
      "learning_rate": 2.4667900297739463e-06,
      "loss": 0.1862,
      "step": 8488
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.261680432088371,
      "learning_rate": 2.465673189194865e-06,
      "loss": 0.1732,
      "step": 8489
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4187220805158927,
      "learning_rate": 2.464556518752821e-06,
      "loss": 0.155,
      "step": 8490
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4621249492381339,
      "learning_rate": 2.4634400185227795e-06,
      "loss": 0.2164,
      "step": 8491
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.713201434067235,
      "learning_rate": 2.462323688579696e-06,
      "loss": 0.6061,
      "step": 8492
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3308084600874768,
      "learning_rate": 2.461207528998511e-06,
      "loss": 0.1909,
      "step": 8493
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2456786050921536,
      "learning_rate": 2.4600915398541576e-06,
      "loss": 0.1283,
      "step": 8494
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.378428536750042,
      "learning_rate": 2.458975721221555e-06,
      "loss": 0.6686,
      "step": 8495
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.870328674334021,
      "learning_rate": 2.4578600731756087e-06,
      "loss": 0.6198,
      "step": 8496
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4557764756101716,
      "learning_rate": 2.4567445957912217e-06,
      "loss": 0.1839,
      "step": 8497
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.486948935891597,
      "learning_rate": 2.4556292891432728e-06,
      "loss": 0.5416,
      "step": 8498
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.372545367371641,
      "learning_rate": 2.4545141533066403e-06,
      "loss": 0.6967,
      "step": 8499
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.284953442519931,
      "learning_rate": 2.4533991883561868e-06,
      "loss": 0.1499,
      "step": 8500
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3539728771278943,
      "learning_rate": 2.4522843943667586e-06,
      "loss": 0.1767,
      "step": 8501
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.150123276735659,
      "learning_rate": 2.4511697714131993e-06,
      "loss": 0.59,
      "step": 8502
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.479145826437882,
      "learning_rate": 2.4500553195703364e-06,
      "loss": 0.5093,
      "step": 8503
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3562506376801557,
      "learning_rate": 2.448941038912986e-06,
      "loss": 0.1855,
      "step": 8504
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.858679188655305,
      "learning_rate": 2.447826929515953e-06,
      "loss": 0.5704,
      "step": 8505
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3647684528844777,
      "learning_rate": 2.4467129914540304e-06,
      "loss": 0.2007,
      "step": 8506
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.312033575876438,
      "learning_rate": 2.445599224802e-06,
      "loss": 0.1598,
      "step": 8507
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3416506917789155,
      "learning_rate": 2.444485629634636e-06,
      "loss": 0.1505,
      "step": 8508
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2490840818307274,
      "learning_rate": 2.4433722060266933e-06,
      "loss": 0.1703,
      "step": 8509
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2175259768823494,
      "learning_rate": 2.4422589540529187e-06,
      "loss": 0.1585,
      "step": 8510
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.208344891110285,
      "learning_rate": 2.4411458737880537e-06,
      "loss": 0.1699,
      "step": 8511
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.362061170228578,
      "learning_rate": 2.4400329653068156e-06,
      "loss": 0.1587,
      "step": 8512
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.658947845851851,
      "learning_rate": 2.4389202286839225e-06,
      "loss": 0.2246,
      "step": 8513
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.441581689359193,
      "learning_rate": 2.4378076639940746e-06,
      "loss": 0.1782,
      "step": 8514
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2868330244084218,
      "learning_rate": 2.436695271311961e-06,
      "loss": 0.1636,
      "step": 8515
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4014748678064373,
      "learning_rate": 2.435583050712261e-06,
      "loss": 0.1958,
      "step": 8516
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.441058278933007,
      "learning_rate": 2.43447100226964e-06,
      "loss": 0.1861,
      "step": 8517
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4708312812227287,
      "learning_rate": 2.433359126058754e-06,
      "loss": 0.1832,
      "step": 8518
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3846367761401979,
      "learning_rate": 2.432247422154247e-06,
      "loss": 0.189,
      "step": 8519
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4974904045135156,
      "learning_rate": 2.43113589063075e-06,
      "loss": 0.1851,
      "step": 8520
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.423384314606963,
      "learning_rate": 2.4300245315628824e-06,
      "loss": 0.1887,
      "step": 8521
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3007616293554927,
      "learning_rate": 2.4289133450252583e-06,
      "loss": 0.1507,
      "step": 8522
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.459632629691166,
      "learning_rate": 2.4278023310924676e-06,
      "loss": 0.6831,
      "step": 8523
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.391848402516739,
      "learning_rate": 2.426691489839102e-06,
      "loss": 0.1795,
      "step": 8524
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2901079400731768,
      "learning_rate": 2.425580821339733e-06,
      "loss": 0.1483,
      "step": 8525
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1845993696400745,
      "learning_rate": 2.4244703256689244e-06,
      "loss": 0.1346,
      "step": 8526
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.908571437799625,
      "learning_rate": 2.423360002901226e-06,
      "loss": 0.5316,
      "step": 8527
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4592188552388006,
      "learning_rate": 2.422249853111177e-06,
      "loss": 0.1785,
      "step": 8528
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.856562566303992,
      "learning_rate": 2.4211398763733063e-06,
      "loss": 0.5992,
      "step": 8529
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2643571241493792,
      "learning_rate": 2.420030072762127e-06,
      "loss": 0.1405,
      "step": 8530
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.217700546821981,
      "learning_rate": 2.4189204423521506e-06,
      "loss": 0.1378,
      "step": 8531
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3077190110334977,
      "learning_rate": 2.4178109852178606e-06,
      "loss": 0.1731,
      "step": 8532
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3426155906656483,
      "learning_rate": 2.416701701433745e-06,
      "loss": 0.1646,
      "step": 8533
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3765645013620476,
      "learning_rate": 2.4155925910742735e-06,
      "loss": 0.2029,
      "step": 8534
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4235353698933184,
      "learning_rate": 2.4144836542138975e-06,
      "loss": 0.1924,
      "step": 8535
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3779149974869869,
      "learning_rate": 2.4133748909270697e-06,
      "loss": 0.1615,
      "step": 8536
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3497584861664549,
      "learning_rate": 2.4122663012882227e-06,
      "loss": 0.1727,
      "step": 8537
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3858682469275923,
      "learning_rate": 2.411157885371779e-06,
      "loss": 0.197,
      "step": 8538
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.243929244430079,
      "learning_rate": 2.4100496432521507e-06,
      "loss": 0.1492,
      "step": 8539
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3831255144952614,
      "learning_rate": 2.408941575003737e-06,
      "loss": 0.1866,
      "step": 8540
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3449284565220094,
      "learning_rate": 2.4078336807009247e-06,
      "loss": 0.1917,
      "step": 8541
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3677498202982803,
      "learning_rate": 2.406725960418094e-06,
      "loss": 0.1763,
      "step": 8542
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3847629074884495,
      "learning_rate": 2.4056184142296056e-06,
      "loss": 0.2275,
      "step": 8543
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3418862308949862,
      "learning_rate": 2.4045110422098122e-06,
      "loss": 0.1677,
      "step": 8544
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.304021088813838,
      "learning_rate": 2.4034038444330597e-06,
      "loss": 0.17,
      "step": 8545
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2660214074555907,
      "learning_rate": 2.4022968209736713e-06,
      "loss": 0.1713,
      "step": 8546
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3000719692074154,
      "learning_rate": 2.4011899719059696e-06,
      "loss": 0.1437,
      "step": 8547
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5150766519191672,
      "learning_rate": 2.400083297304259e-06,
      "loss": 0.1792,
      "step": 8548
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3977149142442993,
      "learning_rate": 2.398976797242834e-06,
      "loss": 0.2329,
      "step": 8549
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4984021145073811,
      "learning_rate": 2.3978704717959777e-06,
      "loss": 0.1754,
      "step": 8550
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1591378510486372,
      "learning_rate": 2.3967643210379608e-06,
      "loss": 0.1536,
      "step": 8551
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.360829072861916,
      "learning_rate": 2.3956583450430426e-06,
      "loss": 0.1603,
      "step": 8552
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5410472497864467,
      "learning_rate": 2.39455254388547e-06,
      "loss": 0.2128,
      "step": 8553
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4776471399141238,
      "learning_rate": 2.39344691763948e-06,
      "loss": 0.1702,
      "step": 8554
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3010736893991974,
      "learning_rate": 2.392341466379294e-06,
      "loss": 0.178,
      "step": 8555
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2545142007384433,
      "learning_rate": 2.391236190179129e-06,
      "loss": 0.1782,
      "step": 8556
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3078817829096632,
      "learning_rate": 2.3901310891131803e-06,
      "loss": 0.1413,
      "step": 8557
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2566846988619564,
      "learning_rate": 2.3890261632556405e-06,
      "loss": 0.1677,
      "step": 8558
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4187241728697129,
      "learning_rate": 2.3879214126806856e-06,
      "loss": 0.1842,
      "step": 8559
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2248998847088113,
      "learning_rate": 2.38681683746248e-06,
      "loss": 0.1707,
      "step": 8560
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3837954012065063,
      "learning_rate": 2.385712437675178e-06,
      "loss": 0.1679,
      "step": 8561
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.326986572992212,
      "learning_rate": 2.384608213392921e-06,
      "loss": 0.21,
      "step": 8562
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1122187033578725,
      "learning_rate": 2.383504164689839e-06,
      "loss": 0.1539,
      "step": 8563
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2486025623598727,
      "learning_rate": 2.38240029164005e-06,
      "loss": 0.1451,
      "step": 8564
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2364221172478997,
      "learning_rate": 2.3812965943176608e-06,
      "loss": 0.1398,
      "step": 8565
    },
    {
      "epoch": 0.69,
      "grad_norm": 25.063600812477958,
      "learning_rate": 2.380193072796763e-06,
      "loss": 0.5526,
      "step": 8566
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4038204898179123,
      "learning_rate": 2.3790897271514464e-06,
      "loss": 0.1913,
      "step": 8567
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.996757008323058,
      "learning_rate": 2.377986557455774e-06,
      "loss": 0.4078,
      "step": 8568
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.073085636818596,
      "learning_rate": 2.376883563783809e-06,
      "loss": 0.544,
      "step": 8569
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.226665789625632,
      "learning_rate": 2.3757807462096013e-06,
      "loss": 0.1226,
      "step": 8570
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3547237112169002,
      "learning_rate": 2.374678104807178e-06,
      "loss": 0.15,
      "step": 8571
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5254561677903267,
      "learning_rate": 2.373575639650571e-06,
      "loss": 0.2089,
      "step": 8572
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4258364748371088,
      "learning_rate": 2.372473350813789e-06,
      "loss": 0.1585,
      "step": 8573
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4384587384880048,
      "learning_rate": 2.371371238370831e-06,
      "loss": 0.2186,
      "step": 8574
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.453410433920425,
      "learning_rate": 2.3702693023956853e-06,
      "loss": 0.1898,
      "step": 8575
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.401553812305528,
      "learning_rate": 2.3691675429623322e-06,
      "loss": 0.1819,
      "step": 8576
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.664635466718382,
      "learning_rate": 2.36806596014473e-06,
      "loss": 0.5996,
      "step": 8577
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2046505080563843,
      "learning_rate": 2.3669645540168355e-06,
      "loss": 0.1729,
      "step": 8578
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.678740454617855,
      "learning_rate": 2.3658633246525906e-06,
      "loss": 0.2152,
      "step": 8579
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2753931285971467,
      "learning_rate": 2.3647622721259184e-06,
      "loss": 0.1483,
      "step": 8580
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3614560587276388,
      "learning_rate": 2.3636613965107406e-06,
      "loss": 0.1627,
      "step": 8581
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3647480392753688,
      "learning_rate": 2.3625606978809614e-06,
      "loss": 0.1706,
      "step": 8582
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.30355084569227,
      "learning_rate": 2.3614601763104733e-06,
      "loss": 0.1725,
      "step": 8583
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5795291484128728,
      "learning_rate": 2.3603598318731584e-06,
      "loss": 0.2356,
      "step": 8584
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3890008674163354,
      "learning_rate": 2.3592596646428855e-06,
      "loss": 0.1623,
      "step": 8585
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6400365006859408,
      "learning_rate": 2.358159674693511e-06,
      "loss": 0.1672,
      "step": 8586
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.353642454611162,
      "learning_rate": 2.357059862098885e-06,
      "loss": 0.1956,
      "step": 8587
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4017000427419182,
      "learning_rate": 2.355960226932837e-06,
      "loss": 0.1924,
      "step": 8588
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2809217862880078,
      "learning_rate": 2.354860769269188e-06,
      "loss": 0.1455,
      "step": 8589
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2906602762927917,
      "learning_rate": 2.353761489181754e-06,
      "loss": 0.1714,
      "step": 8590
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.282015781676501,
      "learning_rate": 2.3526623867443248e-06,
      "loss": 0.1726,
      "step": 8591
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3792755320530645,
      "learning_rate": 2.3515634620306925e-06,
      "loss": 0.1928,
      "step": 8592
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3595467302432158,
      "learning_rate": 2.35046471511463e-06,
      "loss": 0.1711,
      "step": 8593
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6470457636587517,
      "learning_rate": 2.3493661460698986e-06,
      "loss": 0.2254,
      "step": 8594
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3172558940297976,
      "learning_rate": 2.3482677549702493e-06,
      "loss": 0.1705,
      "step": 8595
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4354833300218077,
      "learning_rate": 2.34716954188942e-06,
      "loss": 0.1668,
      "step": 8596
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1950440706486667,
      "learning_rate": 2.3460715069011364e-06,
      "loss": 0.1774,
      "step": 8597
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3152357756919197,
      "learning_rate": 2.344973650079114e-06,
      "loss": 0.2087,
      "step": 8598
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2153941066514582,
      "learning_rate": 2.343875971497055e-06,
      "loss": 0.1571,
      "step": 8599
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.357844507461579,
      "learning_rate": 2.342778471228648e-06,
      "loss": 0.179,
      "step": 8600
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4036411008739,
      "learning_rate": 2.3416811493475767e-06,
      "loss": 0.1687,
      "step": 8601
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.442257632305718,
      "learning_rate": 2.340584005927501e-06,
      "loss": 0.1801,
      "step": 8602
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4020140667254148,
      "learning_rate": 2.33948704104208e-06,
      "loss": 0.2135,
      "step": 8603
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2231075488783294,
      "learning_rate": 2.3383902547649567e-06,
      "loss": 0.1546,
      "step": 8604
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3729470096320693,
      "learning_rate": 2.3372936471697564e-06,
      "loss": 0.1705,
      "step": 8605
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3213457258767163,
      "learning_rate": 2.336197218330104e-06,
      "loss": 0.1569,
      "step": 8606
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.377234671918558,
      "learning_rate": 2.3351009683196025e-06,
      "loss": 0.5272,
      "step": 8607
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2533129565825858,
      "learning_rate": 2.3340048972118475e-06,
      "loss": 0.1523,
      "step": 8608
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4437289383505454,
      "learning_rate": 2.332909005080421e-06,
      "loss": 0.1751,
      "step": 8609
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1818405728761945,
      "learning_rate": 2.3318132919988944e-06,
      "loss": 0.1662,
      "step": 8610
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.237474968897468,
      "learning_rate": 2.3307177580408243e-06,
      "loss": 0.5038,
      "step": 8611
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5583317725987094,
      "learning_rate": 2.3296224032797616e-06,
      "loss": 0.2323,
      "step": 8612
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6800599716369158,
      "learning_rate": 2.328527227789236e-06,
      "loss": 0.1912,
      "step": 8613
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1616636644168354,
      "learning_rate": 2.3274322316427706e-06,
      "loss": 0.174,
      "step": 8614
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3998644498786548,
      "learning_rate": 2.326337414913881e-06,
      "loss": 0.17,
      "step": 8615
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.292234164920583,
      "learning_rate": 2.325242777676057e-06,
      "loss": 0.5858,
      "step": 8616
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.754097612815069,
      "learning_rate": 2.324148320002792e-06,
      "loss": 0.4215,
      "step": 8617
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3801948094943512,
      "learning_rate": 2.323054041967557e-06,
      "loss": 0.141,
      "step": 8618
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1804844524325584,
      "learning_rate": 2.3219599436438162e-06,
      "loss": 0.1644,
      "step": 8619
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4615781525710068,
      "learning_rate": 2.320866025105016e-06,
      "loss": 0.2289,
      "step": 8620
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0737896593414822,
      "learning_rate": 2.319772286424601e-06,
      "loss": 0.1197,
      "step": 8621
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.958260260142388,
      "learning_rate": 2.318678727675991e-06,
      "loss": 0.5602,
      "step": 8622
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.309184401051319,
      "learning_rate": 2.3175853489326e-06,
      "loss": 0.1315,
      "step": 8623
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2973334380537973,
      "learning_rate": 2.3164921502678363e-06,
      "loss": 0.1802,
      "step": 8624
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2102528402280974,
      "learning_rate": 2.315399131755081e-06,
      "loss": 0.1697,
      "step": 8625
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3414110501478933,
      "learning_rate": 2.314306293467718e-06,
      "loss": 0.1557,
      "step": 8626
    },
    {
      "epoch": 0.69,
      "grad_norm": 14.041071212461024,
      "learning_rate": 2.313213635479111e-06,
      "loss": 0.6111,
      "step": 8627
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5257508939651871,
      "learning_rate": 2.312121157862613e-06,
      "loss": 0.1897,
      "step": 8628
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3856273625845739,
      "learning_rate": 2.311028860691566e-06,
      "loss": 0.1505,
      "step": 8629
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3685700282297448,
      "learning_rate": 2.3099367440392985e-06,
      "loss": 0.218,
      "step": 8630
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.856678168937938,
      "learning_rate": 2.308844807979128e-06,
      "loss": 0.4455,
      "step": 8631
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2522641174410185,
      "learning_rate": 2.3077530525843593e-06,
      "loss": 0.1914,
      "step": 8632
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5594677146749716,
      "learning_rate": 2.306661477928286e-06,
      "loss": 0.1833,
      "step": 8633
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3523388746249685,
      "learning_rate": 2.3055700840841857e-06,
      "loss": 0.1917,
      "step": 8634
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.494809927899121,
      "learning_rate": 2.3044788711253336e-06,
      "loss": 0.1775,
      "step": 8635
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1440631106618129,
      "learning_rate": 2.3033878391249775e-06,
      "loss": 0.1361,
      "step": 8636
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3077896538758447,
      "learning_rate": 2.302296988156369e-06,
      "loss": 0.1699,
      "step": 8637
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3165458622636665,
      "learning_rate": 2.301206318292737e-06,
      "loss": 0.1864,
      "step": 8638
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4201494771515406,
      "learning_rate": 2.300115829607302e-06,
      "loss": 0.1656,
      "step": 8639
    },
    {
      "epoch": 0.69,
      "grad_norm": 14.588831789297222,
      "learning_rate": 2.299025522173271e-06,
      "loss": 0.5851,
      "step": 8640
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4072551172991863,
      "learning_rate": 2.297935396063841e-06,
      "loss": 0.153,
      "step": 8641
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1859591460674632,
      "learning_rate": 2.296845451352195e-06,
      "loss": 0.1174,
      "step": 8642
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2535864752171626,
      "learning_rate": 2.2957556881115033e-06,
      "loss": 0.1188,
      "step": 8643
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.411947794916505,
      "learning_rate": 2.294666106414926e-06,
      "loss": 0.2099,
      "step": 8644
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4879657084185662,
      "learning_rate": 2.2935767063356084e-06,
      "loss": 0.1846,
      "step": 8645
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4057771121186284,
      "learning_rate": 2.2924874879466903e-06,
      "loss": 0.1666,
      "step": 8646
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6089047311051063,
      "learning_rate": 2.2913984513212866e-06,
      "loss": 0.2122,
      "step": 8647
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3227789332390976,
      "learning_rate": 2.2903095965325135e-06,
      "loss": 0.1384,
      "step": 8648
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2453708652530837,
      "learning_rate": 2.289220923653469e-06,
      "loss": 0.1607,
      "step": 8649
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.113529211518167,
      "learning_rate": 2.2881324327572336e-06,
      "loss": 0.1256,
      "step": 8650
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.291072188271209,
      "learning_rate": 2.287044123916887e-06,
      "loss": 0.1633,
      "step": 8651
    },
    {
      "epoch": 0.69,
      "grad_norm": 15.141908833234195,
      "learning_rate": 2.2859559972054874e-06,
      "loss": 0.561,
      "step": 8652
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3701657754788936,
      "learning_rate": 2.2848680526960854e-06,
      "loss": 0.2,
      "step": 8653
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3821793473968764,
      "learning_rate": 2.283780290461717e-06,
      "loss": 0.1936,
      "step": 8654
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4330477840192175,
      "learning_rate": 2.2826927105754083e-06,
      "loss": 0.1665,
      "step": 8655
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.59051277537362,
      "learning_rate": 2.2816053131101706e-06,
      "loss": 0.1976,
      "step": 8656
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3372352117982291,
      "learning_rate": 2.2805180981390047e-06,
      "loss": 0.1659,
      "step": 8657
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3881366115859393,
      "learning_rate": 2.279431065734899e-06,
      "loss": 0.1809,
      "step": 8658
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2384241350736567,
      "learning_rate": 2.2783442159708262e-06,
      "loss": 0.1781,
      "step": 8659
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.645180923109874,
      "learning_rate": 2.2772575489197553e-06,
      "loss": 0.4778,
      "step": 8660
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4698639921262209,
      "learning_rate": 2.2761710646546336e-06,
      "loss": 0.2238,
      "step": 8661
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2257661877328063,
      "learning_rate": 2.275084763248402e-06,
      "loss": 0.1447,
      "step": 8662
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2082783756899877,
      "learning_rate": 2.2739986447739864e-06,
      "loss": 0.1382,
      "step": 8663
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3665834296831005,
      "learning_rate": 2.272912709304301e-06,
      "loss": 0.1364,
      "step": 8664
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3139959335967346,
      "learning_rate": 2.271826956912248e-06,
      "loss": 0.1757,
      "step": 8665
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.283557105834495,
      "learning_rate": 2.2707413876707176e-06,
      "loss": 0.1428,
      "step": 8666
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2684148495355902,
      "learning_rate": 2.2696560016525872e-06,
      "loss": 0.1531,
      "step": 8667
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1756378583122005,
      "learning_rate": 2.2685707989307206e-06,
      "loss": 0.1514,
      "step": 8668
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5057887378935848,
      "learning_rate": 2.2674857795779747e-06,
      "loss": 0.1902,
      "step": 8669
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4528078861420395,
      "learning_rate": 2.266400943667185e-06,
      "loss": 0.1927,
      "step": 8670
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4810948757119218,
      "learning_rate": 2.265316291271184e-06,
      "loss": 0.1702,
      "step": 8671
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2615013220150761,
      "learning_rate": 2.2642318224627856e-06,
      "loss": 0.1509,
      "step": 8672
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.316791946086133,
      "learning_rate": 2.2631475373147947e-06,
      "loss": 0.1695,
      "step": 8673
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2520488729890291,
      "learning_rate": 2.2620634359000017e-06,
      "loss": 0.1702,
      "step": 8674
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4645323598742592,
      "learning_rate": 2.260979518291186e-06,
      "loss": 0.2026,
      "step": 8675
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3225285487978233,
      "learning_rate": 2.259895784561114e-06,
      "loss": 0.2131,
      "step": 8676
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5041488606363822,
      "learning_rate": 2.2588122347825403e-06,
      "loss": 0.1702,
      "step": 8677
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1408835350586002,
      "learning_rate": 2.257728869028206e-06,
      "loss": 0.124,
      "step": 8678
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3526120289817962,
      "learning_rate": 2.2566456873708407e-06,
      "loss": 0.1905,
      "step": 8679
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.285780156576686,
      "learning_rate": 2.255562689883166e-06,
      "loss": 0.152,
      "step": 8680
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.71512727478539,
      "learning_rate": 2.254479876637879e-06,
      "loss": 0.6604,
      "step": 8681
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.165899180043969,
      "learning_rate": 2.253397247707679e-06,
      "loss": 0.1529,
      "step": 8682
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3454976114941755,
      "learning_rate": 2.2523148031652447e-06,
      "loss": 0.1216,
      "step": 8683
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.365999298343403,
      "learning_rate": 2.2512325430832393e-06,
      "loss": 0.1645,
      "step": 8684
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3170762927418018,
      "learning_rate": 2.2501504675343237e-06,
      "loss": 0.1943,
      "step": 8685
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4784706699991588,
      "learning_rate": 2.249068576591139e-06,
      "loss": 0.228,
      "step": 8686
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.296589181648553,
      "learning_rate": 2.247986870326316e-06,
      "loss": 0.1738,
      "step": 8687
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5356612640489036,
      "learning_rate": 2.246905348812472e-06,
      "loss": 0.2028,
      "step": 8688
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.393529379541585,
      "learning_rate": 2.245824012122214e-06,
      "loss": 0.7075,
      "step": 8689
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3127502337152592,
      "learning_rate": 2.244742860328133e-06,
      "loss": 0.1737,
      "step": 8690
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3317927276905537,
      "learning_rate": 2.2436618935028152e-06,
      "loss": 0.1857,
      "step": 8691
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.408709377149671,
      "learning_rate": 2.2425811117188245e-06,
      "loss": 0.2097,
      "step": 8692
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5359756895497416,
      "learning_rate": 2.241500515048716e-06,
      "loss": 0.1925,
      "step": 8693
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.57717298846275,
      "learning_rate": 2.2404201035650404e-06,
      "loss": 0.5537,
      "step": 8694
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.20115842475598,
      "learning_rate": 2.2393398773403196e-06,
      "loss": 0.1354,
      "step": 8695
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1815064544567289,
      "learning_rate": 2.23825983644708e-06,
      "loss": 0.1499,
      "step": 8696
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4232128296982534,
      "learning_rate": 2.2371799809578254e-06,
      "loss": 0.1868,
      "step": 8697
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2135499758565902,
      "learning_rate": 2.236100310945049e-06,
      "loss": 0.1431,
      "step": 8698
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.337339011347196,
      "learning_rate": 2.2350208264812318e-06,
      "loss": 0.186,
      "step": 8699
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5417855191289902,
      "learning_rate": 2.233941527638848e-06,
      "loss": 0.1835,
      "step": 8700
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5400669998025067,
      "learning_rate": 2.232862414490347e-06,
      "loss": 0.1764,
      "step": 8701
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.85876936960936,
      "learning_rate": 2.2317834871081757e-06,
      "loss": 0.6021,
      "step": 8702
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2261340481247933,
      "learning_rate": 2.2307047455647696e-06,
      "loss": 0.1643,
      "step": 8703
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3717648273852179,
      "learning_rate": 2.2296261899325407e-06,
      "loss": 0.1678,
      "step": 8704
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.31532119975373,
      "learning_rate": 2.228547820283902e-06,
      "loss": 0.1753,
      "step": 8705
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.399501526894658,
      "learning_rate": 2.2274696366912448e-06,
      "loss": 0.1747,
      "step": 8706
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5062145517272492,
      "learning_rate": 2.226391639226951e-06,
      "loss": 0.2285,
      "step": 8707
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3491365043466192,
      "learning_rate": 2.2253138279633907e-06,
      "loss": 0.1623,
      "step": 8708
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.394595576009113,
      "learning_rate": 2.22423620297292e-06,
      "loss": 0.165,
      "step": 8709
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.462069929474237,
      "learning_rate": 2.2231587643278827e-06,
      "loss": 0.1799,
      "step": 8710
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4534174807138118,
      "learning_rate": 2.2220815121006116e-06,
      "loss": 0.1495,
      "step": 8711
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0298562362298234,
      "learning_rate": 2.2210044463634252e-06,
      "loss": 0.1717,
      "step": 8712
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.262769019496148,
      "learning_rate": 2.219927567188629e-06,
      "loss": 0.136,
      "step": 8713
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3324393388167715,
      "learning_rate": 2.2188508746485214e-06,
      "loss": 0.1984,
      "step": 8714
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.48438789746588,
      "learning_rate": 2.217774368815378e-06,
      "loss": 0.2132,
      "step": 8715
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.246894482827469,
      "learning_rate": 2.2166980497614728e-06,
      "loss": 0.1403,
      "step": 8716
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4094312823907584,
      "learning_rate": 2.2156219175590623e-06,
      "loss": 0.1451,
      "step": 8717
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5048823293038702,
      "learning_rate": 2.214545972280385e-06,
      "loss": 0.2391,
      "step": 8718
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.588603464127296,
      "learning_rate": 2.2134702139976777e-06,
      "loss": 0.6963,
      "step": 8719
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.403886494343395,
      "learning_rate": 2.2123946427831582e-06,
      "loss": 0.2175,
      "step": 8720
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2626957641134713,
      "learning_rate": 2.2113192587090327e-06,
      "loss": 0.1679,
      "step": 8721
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6784919996542202,
      "learning_rate": 2.2102440618474942e-06,
      "loss": 0.1646,
      "step": 8722
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2948792986247344,
      "learning_rate": 2.2091690522707248e-06,
      "loss": 0.168,
      "step": 8723
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.249720254991113,
      "learning_rate": 2.2080942300508907e-06,
      "loss": 0.5663,
      "step": 8724
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4229058812396012,
      "learning_rate": 2.207019595260154e-06,
      "loss": 0.2145,
      "step": 8725
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4630662476865703,
      "learning_rate": 2.2059451479706522e-06,
      "loss": 0.2301,
      "step": 8726
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3195765864235,
      "learning_rate": 2.2048708882545166e-06,
      "loss": 0.1634,
      "step": 8727
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6388587411312872,
      "learning_rate": 2.203796816183871e-06,
      "loss": 0.2031,
      "step": 8728
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3542096925167477,
      "learning_rate": 2.2027229318308135e-06,
      "loss": 0.1702,
      "step": 8729
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4778953744058572,
      "learning_rate": 2.2016492352674424e-06,
      "loss": 0.1868,
      "step": 8730
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4677367710726397,
      "learning_rate": 2.2005757265658375e-06,
      "loss": 0.1966,
      "step": 8731
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2931955724901292,
      "learning_rate": 2.1995024057980656e-06,
      "loss": 0.1652,
      "step": 8732
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2514041651540069,
      "learning_rate": 2.198429273036182e-06,
      "loss": 0.2037,
      "step": 8733
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4305431824315464,
      "learning_rate": 2.1973563283522305e-06,
      "loss": 0.1515,
      "step": 8734
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2808339942607438,
      "learning_rate": 2.19628357181824e-06,
      "loss": 0.1745,
      "step": 8735
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3083878047924136,
      "learning_rate": 2.1952110035062284e-06,
      "loss": 0.1881,
      "step": 8736
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4242874062845257,
      "learning_rate": 2.1941386234882e-06,
      "loss": 0.1868,
      "step": 8737
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3482473959845853,
      "learning_rate": 2.1930664318361457e-06,
      "loss": 0.1663,
      "step": 8738
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3831857066941897,
      "learning_rate": 2.19199442862205e-06,
      "loss": 0.2388,
      "step": 8739
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.952667737017903,
      "learning_rate": 2.1909226139178723e-06,
      "loss": 0.549,
      "step": 8740
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.448249893474245,
      "learning_rate": 2.1898509877955733e-06,
      "loss": 0.1887,
      "step": 8741
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.448377709262811,
      "learning_rate": 2.1887795503270914e-06,
      "loss": 0.4552,
      "step": 8742
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.082956309024548,
      "learning_rate": 2.187708301584356e-06,
      "loss": 0.5641,
      "step": 8743
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.318625735997687,
      "learning_rate": 2.1866372416392827e-06,
      "loss": 0.1976,
      "step": 8744
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.187578344779676,
      "learning_rate": 2.1855663705637763e-06,
      "loss": 0.1798,
      "step": 8745
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4578983494895568,
      "learning_rate": 2.1844956884297265e-06,
      "loss": 0.1934,
      "step": 8746
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3237372610631304,
      "learning_rate": 2.1834251953090096e-06,
      "loss": 0.2015,
      "step": 8747
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.528414428453423,
      "learning_rate": 2.1823548912734976e-06,
      "loss": 0.1796,
      "step": 8748
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3324822343648361,
      "learning_rate": 2.181284776395035e-06,
      "loss": 0.1774,
      "step": 8749
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.391034644324073,
      "learning_rate": 2.1802148507454675e-06,
      "loss": 0.1518,
      "step": 8750
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4596720917481374,
      "learning_rate": 2.17914511439662e-06,
      "loss": 0.1886,
      "step": 8751
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4719213776194484,
      "learning_rate": 2.1780755674203087e-06,
      "loss": 0.1463,
      "step": 8752
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2633251822427136,
      "learning_rate": 2.1770062098883343e-06,
      "loss": 0.1563,
      "step": 8753
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.354272484944648,
      "learning_rate": 2.175937041872487e-06,
      "loss": 0.1637,
      "step": 8754
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3420536308280704,
      "learning_rate": 2.174868063444542e-06,
      "loss": 0.1884,
      "step": 8755
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5413568596475649,
      "learning_rate": 2.173799274676264e-06,
      "loss": 0.2399,
      "step": 8756
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4809524084303805,
      "learning_rate": 2.1727306756394027e-06,
      "loss": 0.1893,
      "step": 8757
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3050136822491063,
      "learning_rate": 2.171662266405697e-06,
      "loss": 0.1806,
      "step": 8758
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3846842094315535,
      "learning_rate": 2.1705940470468758e-06,
      "loss": 0.1807,
      "step": 8759
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4035795574727916,
      "learning_rate": 2.1695260176346453e-06,
      "loss": 0.203,
      "step": 8760
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2896832358138453,
      "learning_rate": 2.1684581782407117e-06,
      "loss": 0.1725,
      "step": 8761
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5857872105033077,
      "learning_rate": 2.167390528936761e-06,
      "loss": 0.2258,
      "step": 8762
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5072083789835178,
      "learning_rate": 2.166323069794463e-06,
      "loss": 0.2018,
      "step": 8763
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2004578291724939,
      "learning_rate": 2.1652558008854853e-06,
      "loss": 0.156,
      "step": 8764
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4177180195213406,
      "learning_rate": 2.164188722281474e-06,
      "loss": 0.1839,
      "step": 8765
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.405902876673419,
      "learning_rate": 2.163121834054066e-06,
      "loss": 0.1893,
      "step": 8766
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5589266291088337,
      "learning_rate": 2.1620551362748854e-06,
      "loss": 0.1912,
      "step": 8767
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.638788751173921,
      "learning_rate": 2.1609886290155412e-06,
      "loss": 0.4962,
      "step": 8768
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3779042004614497,
      "learning_rate": 2.159922312347631e-06,
      "loss": 0.2079,
      "step": 8769
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2521203725918555,
      "learning_rate": 2.158856186342745e-06,
      "loss": 0.1556,
      "step": 8770
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4078683413357185,
      "learning_rate": 2.1577902510724486e-06,
      "loss": 0.1892,
      "step": 8771
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2253088175861557,
      "learning_rate": 2.1567245066083033e-06,
      "loss": 0.1866,
      "step": 8772
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3814699745572565,
      "learning_rate": 2.15565895302186e-06,
      "loss": 0.1877,
      "step": 8773
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5342635587114835,
      "learning_rate": 2.1545935903846456e-06,
      "loss": 0.1952,
      "step": 8774
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5421558089029654,
      "learning_rate": 2.1535284187681866e-06,
      "loss": 0.221,
      "step": 8775
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3670858207147938,
      "learning_rate": 2.152463438243989e-06,
      "loss": 0.186,
      "step": 8776
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3266752703845073,
      "learning_rate": 2.1513986488835486e-06,
      "loss": 0.2035,
      "step": 8777
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1514188028614187,
      "learning_rate": 2.1503340507583482e-06,
      "loss": 0.1672,
      "step": 8778
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1590982308808748,
      "learning_rate": 2.1492696439398563e-06,
      "loss": 0.1748,
      "step": 8779
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5924349186199698,
      "learning_rate": 2.148205428499531e-06,
      "loss": 0.2184,
      "step": 8780
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7672018950883812,
      "learning_rate": 2.147141404508816e-06,
      "loss": 0.1977,
      "step": 8781
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4514485893671827,
      "learning_rate": 2.146077572039143e-06,
      "loss": 0.2069,
      "step": 8782
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4475352734861087,
      "learning_rate": 2.145013931161927e-06,
      "loss": 0.1871,
      "step": 8783
    },
    {
      "epoch": 0.7,
      "grad_norm": 17.056245322540924,
      "learning_rate": 2.1439504819485795e-06,
      "loss": 0.6315,
      "step": 8784
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5210892120096347,
      "learning_rate": 2.1428872244704862e-06,
      "loss": 0.1925,
      "step": 8785
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1989500402999689,
      "learning_rate": 2.1418241587990326e-06,
      "loss": 0.1647,
      "step": 8786
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.622820220211232,
      "learning_rate": 2.140761285005584e-06,
      "loss": 0.627,
      "step": 8787
    },
    {
      "epoch": 0.7,
      "grad_norm": 47.784501888223176,
      "learning_rate": 2.139698603161491e-06,
      "loss": 0.6089,
      "step": 8788
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2652650592349,
      "learning_rate": 2.1386361133380985e-06,
      "loss": 0.1798,
      "step": 8789
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3280266661705573,
      "learning_rate": 2.1375738156067327e-06,
      "loss": 0.1981,
      "step": 8790
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2822586657578385,
      "learning_rate": 2.13651171003871e-06,
      "loss": 0.1593,
      "step": 8791
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2566453274980987,
      "learning_rate": 2.1354497967053304e-06,
      "loss": 0.1666,
      "step": 8792
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2849340099753168,
      "learning_rate": 2.1343880756778884e-06,
      "loss": 0.1712,
      "step": 8793
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3134977391382963,
      "learning_rate": 2.133326547027654e-06,
      "loss": 0.1613,
      "step": 8794
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6583229142328548,
      "learning_rate": 2.132265210825896e-06,
      "loss": 0.2101,
      "step": 8795
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2836326828938867,
      "learning_rate": 2.1312040671438656e-06,
      "loss": 0.1474,
      "step": 8796
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3714070436635597,
      "learning_rate": 2.130143116052794e-06,
      "loss": 0.1853,
      "step": 8797
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3802971835800655,
      "learning_rate": 2.129082357623912e-06,
      "loss": 0.1656,
      "step": 8798
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2090062461686482,
      "learning_rate": 2.128021791928431e-06,
      "loss": 0.1497,
      "step": 8799
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.480064634597544,
      "learning_rate": 2.1269614190375477e-06,
      "loss": 0.1853,
      "step": 8800
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4062062432419404,
      "learning_rate": 2.12590123902245e-06,
      "loss": 0.1997,
      "step": 8801
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4300533792985821,
      "learning_rate": 2.12484125195431e-06,
      "loss": 0.1735,
      "step": 8802
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.384543903209185,
      "learning_rate": 2.123781457904286e-06,
      "loss": 0.1824,
      "step": 8803
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5132930671969893,
      "learning_rate": 2.122721856943531e-06,
      "loss": 0.1796,
      "step": 8804
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4532391516521237,
      "learning_rate": 2.1216624491431744e-06,
      "loss": 0.1927,
      "step": 8805
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.369314650979274,
      "learning_rate": 2.120603234574336e-06,
      "loss": 0.1997,
      "step": 8806
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3858502990956465,
      "learning_rate": 2.1195442133081307e-06,
      "loss": 0.1721,
      "step": 8807
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4622561652982093,
      "learning_rate": 2.1184853854156462e-06,
      "loss": 0.1745,
      "step": 8808
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4447131585698556,
      "learning_rate": 2.1174267509679703e-06,
      "loss": 0.1717,
      "step": 8809
    },
    {
      "epoch": 0.7,
      "grad_norm": 13.866334365437547,
      "learning_rate": 2.1163683100361702e-06,
      "loss": 0.5617,
      "step": 8810
    },
    {
      "epoch": 0.7,
      "grad_norm": 12.041647197855086,
      "learning_rate": 2.1153100626913027e-06,
      "loss": 0.3806,
      "step": 8811
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.550844057590258,
      "learning_rate": 2.114252009004411e-06,
      "loss": 0.6173,
      "step": 8812
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2360118798995223,
      "learning_rate": 2.113194149046525e-06,
      "loss": 0.1541,
      "step": 8813
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4922707384325054,
      "learning_rate": 2.112136482888663e-06,
      "loss": 0.2145,
      "step": 8814
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.71839473296996,
      "learning_rate": 2.1110790106018286e-06,
      "loss": 0.3803,
      "step": 8815
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.413837554148325,
      "learning_rate": 2.1100217322570133e-06,
      "loss": 0.182,
      "step": 8816
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.475626157536165,
      "learning_rate": 2.1089646479251936e-06,
      "loss": 0.5344,
      "step": 8817
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2816078715522858,
      "learning_rate": 2.10790775767734e-06,
      "loss": 0.168,
      "step": 8818
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3580897520767927,
      "learning_rate": 2.106851061584398e-06,
      "loss": 0.1941,
      "step": 8819
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5446506490868177,
      "learning_rate": 2.105794559717311e-06,
      "loss": 0.1941,
      "step": 8820
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4381451154685287,
      "learning_rate": 2.1047382521470057e-06,
      "loss": 0.2112,
      "step": 8821
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.304268565476533,
      "learning_rate": 2.1036821389443927e-06,
      "loss": 0.162,
      "step": 8822
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.376732000216103,
      "learning_rate": 2.1026262201803738e-06,
      "loss": 0.1981,
      "step": 8823
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4773961092945924,
      "learning_rate": 2.1015704959258355e-06,
      "loss": 0.2161,
      "step": 8824
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4320544687799417,
      "learning_rate": 2.1005149662516517e-06,
      "loss": 0.1975,
      "step": 8825
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3237543295352794,
      "learning_rate": 2.099459631228683e-06,
      "loss": 0.2086,
      "step": 8826
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3892191534369167,
      "learning_rate": 2.0984044909277777e-06,
      "loss": 0.2306,
      "step": 8827
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2142666790063947,
      "learning_rate": 2.0973495454197686e-06,
      "loss": 0.1563,
      "step": 8828
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5062731682482111,
      "learning_rate": 2.096294794775481e-06,
      "loss": 0.2151,
      "step": 8829
    },
    {
      "epoch": 0.71,
      "grad_norm": 17.263257745407792,
      "learning_rate": 2.0952402390657215e-06,
      "loss": 0.5625,
      "step": 8830
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.275642617991941,
      "learning_rate": 2.0941858783612863e-06,
      "loss": 0.1386,
      "step": 8831
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3848665576330261,
      "learning_rate": 2.093131712732957e-06,
      "loss": 0.2391,
      "step": 8832
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2702835364104217,
      "learning_rate": 2.0920777422515033e-06,
      "loss": 0.1578,
      "step": 8833
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.893659356488322,
      "learning_rate": 2.0910239669876814e-06,
      "loss": 0.6647,
      "step": 8834
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5017451947179192,
      "learning_rate": 2.0899703870122347e-06,
      "loss": 0.1868,
      "step": 8835
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.213723079554798,
      "learning_rate": 2.0889170023958927e-06,
      "loss": 0.4584,
      "step": 8836
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4903907522906628,
      "learning_rate": 2.0878638132093703e-06,
      "loss": 0.2304,
      "step": 8837
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3420402720303217,
      "learning_rate": 2.0868108195233783e-06,
      "loss": 0.1876,
      "step": 8838
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5246173645413683,
      "learning_rate": 2.0857580214086e-06,
      "loss": 0.2007,
      "step": 8839
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3505345935443007,
      "learning_rate": 2.0847054189357136e-06,
      "loss": 0.1856,
      "step": 8840
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.328069713639952,
      "learning_rate": 2.0836530121753895e-06,
      "loss": 0.4315,
      "step": 8841
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2589652011839645,
      "learning_rate": 2.0826008011982714e-06,
      "loss": 0.196,
      "step": 8842
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5765691334292593,
      "learning_rate": 2.0815487860750023e-06,
      "loss": 0.2135,
      "step": 8843
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4865032367365123,
      "learning_rate": 2.0804969668762052e-06,
      "loss": 0.1885,
      "step": 8844
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2294452274400172,
      "learning_rate": 2.079445343672493e-06,
      "loss": 0.144,
      "step": 8845
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1103684526624893,
      "learning_rate": 2.078393916534464e-06,
      "loss": 0.1069,
      "step": 8846
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3410438379703813,
      "learning_rate": 2.0773426855327034e-06,
      "loss": 0.1708,
      "step": 8847
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6506223626781895,
      "learning_rate": 2.076291650737784e-06,
      "loss": 0.251,
      "step": 8848
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2074409122689882,
      "learning_rate": 2.075240812220264e-06,
      "loss": 0.1644,
      "step": 8849
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.258498636978523,
      "learning_rate": 2.07419017005069e-06,
      "loss": 0.1767,
      "step": 8850
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.389742377300548,
      "learning_rate": 2.0731397242995932e-06,
      "loss": 0.172,
      "step": 8851
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.243204367665582,
      "learning_rate": 2.0720894750374982e-06,
      "loss": 0.1629,
      "step": 8852
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2684072703099805,
      "learning_rate": 2.071039422334904e-06,
      "loss": 0.1891,
      "step": 8853
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.208393405917905,
      "learning_rate": 2.0699895662623097e-06,
      "loss": 0.1739,
      "step": 8854
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.401908456132836,
      "learning_rate": 2.068939906890194e-06,
      "loss": 0.1795,
      "step": 8855
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.712359394773343,
      "learning_rate": 2.0678904442890222e-06,
      "loss": 0.6128,
      "step": 8856
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4498698601565287,
      "learning_rate": 2.06684117852925e-06,
      "loss": 0.171,
      "step": 8857
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4905052209777583,
      "learning_rate": 2.0657921096813154e-06,
      "loss": 0.205,
      "step": 8858
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.37779395509775,
      "learning_rate": 2.0647432378156473e-06,
      "loss": 0.7501,
      "step": 8859
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.777066165302513,
      "learning_rate": 2.0636945630026594e-06,
      "loss": 0.5364,
      "step": 8860
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3493024116593981,
      "learning_rate": 2.062646085312752e-06,
      "loss": 0.1744,
      "step": 8861
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4172313940747723,
      "learning_rate": 2.0615978048163106e-06,
      "loss": 0.1897,
      "step": 8862
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.258228085163617,
      "learning_rate": 2.0605497215837157e-06,
      "loss": 0.1969,
      "step": 8863
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.414241917380858,
      "learning_rate": 2.05950183568532e-06,
      "loss": 0.1947,
      "step": 8864
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3681719346891756,
      "learning_rate": 2.058454147191478e-06,
      "loss": 0.1462,
      "step": 8865
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4210413124718841,
      "learning_rate": 2.057406656172522e-06,
      "loss": 0.2069,
      "step": 8866
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.129885131379842,
      "learning_rate": 2.0563593626987705e-06,
      "loss": 0.7347,
      "step": 8867
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3353185414284146,
      "learning_rate": 2.0553122668405352e-06,
      "loss": 0.188,
      "step": 8868
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.478511375119496,
      "learning_rate": 2.0542653686681096e-06,
      "loss": 0.6067,
      "step": 8869
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2807392110065132,
      "learning_rate": 2.053218668251775e-06,
      "loss": 0.1913,
      "step": 8870
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.820035919517078,
      "learning_rate": 2.0521721656617976e-06,
      "loss": 0.624,
      "step": 8871
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.367497406936669,
      "learning_rate": 2.051125860968438e-06,
      "loss": 0.1809,
      "step": 8872
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.849808753412626,
      "learning_rate": 2.050079754241931e-06,
      "loss": 0.58,
      "step": 8873
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4757352890299462,
      "learning_rate": 2.0490338455525095e-06,
      "loss": 0.2111,
      "step": 8874
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3113034833813948,
      "learning_rate": 2.0479881349703885e-06,
      "loss": 0.1857,
      "step": 8875
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4037693468083008,
      "learning_rate": 2.0469426225657647e-06,
      "loss": 0.1662,
      "step": 8876
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3466721617272521,
      "learning_rate": 2.0458973084088316e-06,
      "loss": 0.1955,
      "step": 8877
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5872954258490564,
      "learning_rate": 2.0448521925697634e-06,
      "loss": 0.1887,
      "step": 8878
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.72511532302446,
      "learning_rate": 2.043807275118721e-06,
      "loss": 0.6621,
      "step": 8879
    },
    {
      "epoch": 0.71,
      "grad_norm": 15.598762438867025,
      "learning_rate": 2.042762556125853e-06,
      "loss": 0.6081,
      "step": 8880
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.480833664347717,
      "learning_rate": 2.041718035661295e-06,
      "loss": 0.4832,
      "step": 8881
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3889063549569087,
      "learning_rate": 2.0406737137951666e-06,
      "loss": 0.1722,
      "step": 8882
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4433172288199752,
      "learning_rate": 2.0396295905975816e-06,
      "loss": 0.1836,
      "step": 8883
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.432354872475748,
      "learning_rate": 2.03858566613863e-06,
      "loss": 0.1659,
      "step": 8884
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.223139700753808,
      "learning_rate": 2.0375419404883938e-06,
      "loss": 0.6951,
      "step": 8885
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4733168650617896,
      "learning_rate": 2.0364984137169464e-06,
      "loss": 0.2128,
      "step": 8886
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2806633412295598,
      "learning_rate": 2.035455085894336e-06,
      "loss": 0.149,
      "step": 8887
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3235700119958946,
      "learning_rate": 2.0344119570906097e-06,
      "loss": 0.1682,
      "step": 8888
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2938884332148652,
      "learning_rate": 2.033369027375794e-06,
      "loss": 0.1999,
      "step": 8889
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3949791198469503,
      "learning_rate": 2.0323262968199043e-06,
      "loss": 0.1474,
      "step": 8890
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.459419889213352,
      "learning_rate": 2.0312837654929412e-06,
      "loss": 0.1595,
      "step": 8891
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.451162222231715,
      "learning_rate": 2.0302414334648944e-06,
      "loss": 0.2459,
      "step": 8892
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1609369611756744,
      "learning_rate": 2.029199300805739e-06,
      "loss": 0.1336,
      "step": 8893
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1017142913133107,
      "learning_rate": 2.0281573675854344e-06,
      "loss": 0.1037,
      "step": 8894
    },
    {
      "epoch": 0.71,
      "grad_norm": 73.31681932343245,
      "learning_rate": 2.027115633873931e-06,
      "loss": 0.5081,
      "step": 8895
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4475739857474033,
      "learning_rate": 2.02607409974116e-06,
      "loss": 0.1844,
      "step": 8896
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.374014870378563,
      "learning_rate": 2.025032765257049e-06,
      "loss": 0.1574,
      "step": 8897
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5335452122548567,
      "learning_rate": 2.0239916304914984e-06,
      "loss": 0.2,
      "step": 8898
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3057293089429303,
      "learning_rate": 2.022950695514409e-06,
      "loss": 0.1797,
      "step": 8899
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2563082709326832,
      "learning_rate": 2.021909960395661e-06,
      "loss": 0.161,
      "step": 8900
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4537795638776398,
      "learning_rate": 2.0208694252051174e-06,
      "loss": 0.2429,
      "step": 8901
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3175652381616214,
      "learning_rate": 2.0198290900126373e-06,
      "loss": 0.1537,
      "step": 8902
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4615986199852011,
      "learning_rate": 2.0187889548880594e-06,
      "loss": 0.1703,
      "step": 8903
    },
    {
      "epoch": 0.71,
      "grad_norm": 22.194715048393576,
      "learning_rate": 2.017749019901213e-06,
      "loss": 0.4577,
      "step": 8904
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.174107113559744,
      "learning_rate": 2.0167092851219094e-06,
      "loss": 0.1644,
      "step": 8905
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.578276450839011,
      "learning_rate": 2.0156697506199512e-06,
      "loss": 0.545,
      "step": 8906
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1836351038222372,
      "learning_rate": 2.0146304164651227e-06,
      "loss": 0.174,
      "step": 8907
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3432759545366408,
      "learning_rate": 2.0135912827272036e-06,
      "loss": 0.1759,
      "step": 8908
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.664527552885824,
      "learning_rate": 2.012552349475948e-06,
      "loss": 0.4168,
      "step": 8909
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.292971414037827,
      "learning_rate": 2.0115136167811033e-06,
      "loss": 0.1603,
      "step": 8910
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2405040693404323,
      "learning_rate": 2.0104750847124075e-06,
      "loss": 0.1556,
      "step": 8911
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5065017470853874,
      "learning_rate": 2.009436753339574e-06,
      "loss": 0.1861,
      "step": 8912
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3557858429895708,
      "learning_rate": 2.0083986227323138e-06,
      "loss": 0.1824,
      "step": 8913
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3434318012983424,
      "learning_rate": 2.007360692960318e-06,
      "loss": 0.1647,
      "step": 8914
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5414041357145596,
      "learning_rate": 2.0063229640932664e-06,
      "loss": 0.2005,
      "step": 8915
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4355962043303128,
      "learning_rate": 2.0052854362008232e-06,
      "loss": 0.1981,
      "step": 8916
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5044123223932846,
      "learning_rate": 2.0042481093526454e-06,
      "loss": 0.1865,
      "step": 8917
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3832212558379213,
      "learning_rate": 2.003210983618368e-06,
      "loss": 0.1786,
      "step": 8918
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3732643237452844,
      "learning_rate": 2.002174059067614e-06,
      "loss": 0.1458,
      "step": 8919
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.751721697102618,
      "learning_rate": 2.001137335770003e-06,
      "loss": 0.4929,
      "step": 8920
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3124551098178798,
      "learning_rate": 2.000100813795125e-06,
      "loss": 0.1718,
      "step": 8921
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.455805796739455,
      "learning_rate": 1.9990644932125704e-06,
      "loss": 0.1802,
      "step": 8922
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3375243772516927,
      "learning_rate": 1.9980283740919086e-06,
      "loss": 0.2127,
      "step": 8923
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2237679361904445,
      "learning_rate": 1.9969924565026982e-06,
      "loss": 0.1442,
      "step": 8924
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4206898912657797,
      "learning_rate": 1.9959567405144825e-06,
      "loss": 0.1586,
      "step": 8925
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.387873481284477,
      "learning_rate": 1.994921226196793e-06,
      "loss": 0.2045,
      "step": 8926
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.503667368614702,
      "learning_rate": 1.993885913619147e-06,
      "loss": 0.204,
      "step": 8927
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.398104153294139,
      "learning_rate": 1.9928508028510467e-06,
      "loss": 0.1778,
      "step": 8928
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.280793073788728,
      "learning_rate": 1.9918158939619843e-06,
      "loss": 0.1581,
      "step": 8929
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3960895174296555,
      "learning_rate": 1.9907811870214334e-06,
      "loss": 0.1912,
      "step": 8930
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3269657227466958,
      "learning_rate": 1.9897466820988622e-06,
      "loss": 0.1712,
      "step": 8931
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.076675910648911,
      "learning_rate": 1.988712379263713e-06,
      "loss": 0.5833,
      "step": 8932
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4895155877351163,
      "learning_rate": 1.9876782785854283e-06,
      "loss": 0.1869,
      "step": 8933
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3819105899604691,
      "learning_rate": 1.9866443801334274e-06,
      "loss": 0.1687,
      "step": 8934
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1871228628252093,
      "learning_rate": 1.9856106839771194e-06,
      "loss": 0.1372,
      "step": 8935
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3653868366131665,
      "learning_rate": 1.984577190185899e-06,
      "loss": 0.1708,
      "step": 8936
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1849947991323169,
      "learning_rate": 1.983543898829149e-06,
      "loss": 0.1634,
      "step": 8937
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1534516532065513,
      "learning_rate": 1.9825108099762358e-06,
      "loss": 0.1406,
      "step": 8938
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2652940800700563,
      "learning_rate": 1.981477923696515e-06,
      "loss": 0.1485,
      "step": 8939
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3976634087237663,
      "learning_rate": 1.9804452400593265e-06,
      "loss": 0.1801,
      "step": 8940
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.490153783334788,
      "learning_rate": 1.9794127591339967e-06,
      "loss": 0.1817,
      "step": 8941
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3693676471876894,
      "learning_rate": 1.978380480989844e-06,
      "loss": 0.1564,
      "step": 8942
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5448776724041051,
      "learning_rate": 1.9773484056961605e-06,
      "loss": 0.2015,
      "step": 8943
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3041591000333528,
      "learning_rate": 1.976316533322239e-06,
      "loss": 0.1871,
      "step": 8944
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.266478614858927,
      "learning_rate": 1.975284863937352e-06,
      "loss": 0.1654,
      "step": 8945
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4347066422425818,
      "learning_rate": 1.9742533976107524e-06,
      "loss": 0.1781,
      "step": 8946
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2832198876254115,
      "learning_rate": 1.973222134411692e-06,
      "loss": 0.1938,
      "step": 8947
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4391846452093422,
      "learning_rate": 1.9721910744094006e-06,
      "loss": 0.1963,
      "step": 8948
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2209384930910225,
      "learning_rate": 1.9711602176730967e-06,
      "loss": 0.1871,
      "step": 8949
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3893943483714364,
      "learning_rate": 1.9701295642719836e-06,
      "loss": 0.1568,
      "step": 8950
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4592579135093358,
      "learning_rate": 1.969099114275253e-06,
      "loss": 0.216,
      "step": 8951
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.20634802262108,
      "learning_rate": 1.968068867752081e-06,
      "loss": 0.1709,
      "step": 8952
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2887015641935737,
      "learning_rate": 1.9670388247716354e-06,
      "loss": 0.1752,
      "step": 8953
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6014548586869528,
      "learning_rate": 1.9660089854030612e-06,
      "loss": 0.2275,
      "step": 8954
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2307201572158966,
      "learning_rate": 1.9649793497154953e-06,
      "loss": 0.1373,
      "step": 8955
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.608759894031024,
      "learning_rate": 1.9639499177780642e-06,
      "loss": 0.2286,
      "step": 8956
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5872218847628707,
      "learning_rate": 1.9629206896598707e-06,
      "loss": 0.2,
      "step": 8957
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4387261174009294,
      "learning_rate": 1.9618916654300144e-06,
      "loss": 0.2176,
      "step": 8958
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3472256930456061,
      "learning_rate": 1.960862845157577e-06,
      "loss": 0.1758,
      "step": 8959
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4559821839385172,
      "learning_rate": 1.959834228911624e-06,
      "loss": 0.1455,
      "step": 8960
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4175853859582273,
      "learning_rate": 1.9588058167612113e-06,
      "loss": 0.2027,
      "step": 8961
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5182053199409054,
      "learning_rate": 1.957777608775378e-06,
      "loss": 0.2195,
      "step": 8962
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4210613805210779,
      "learning_rate": 1.9567496050231516e-06,
      "loss": 0.1708,
      "step": 8963
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.992888157424812,
      "learning_rate": 1.955721805573543e-06,
      "loss": 0.5966,
      "step": 8964
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1769085169195812,
      "learning_rate": 1.9546942104955567e-06,
      "loss": 0.1625,
      "step": 8965
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2709786771199052,
      "learning_rate": 1.953666819858172e-06,
      "loss": 0.1666,
      "step": 8966
    },
    {
      "epoch": 0.72,
      "grad_norm": 11.871684728700824,
      "learning_rate": 1.9526396337303654e-06,
      "loss": 0.472,
      "step": 8967
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3024000046297861,
      "learning_rate": 1.951612652181093e-06,
      "loss": 0.1625,
      "step": 8968
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3745739019797572,
      "learning_rate": 1.9505858752792995e-06,
      "loss": 0.163,
      "step": 8969
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.66430553107358,
      "learning_rate": 1.949559303093916e-06,
      "loss": 0.5885,
      "step": 8970
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0866830098427749,
      "learning_rate": 1.9485329356938587e-06,
      "loss": 0.1443,
      "step": 8971
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3714377355503682,
      "learning_rate": 1.9475067731480308e-06,
      "loss": 0.1481,
      "step": 8972
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.47456644670411,
      "learning_rate": 1.9464808155253225e-06,
      "loss": 0.1648,
      "step": 8973
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.17560716864444,
      "learning_rate": 1.945455062894608e-06,
      "loss": 0.5775,
      "step": 8974
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4322579342718345,
      "learning_rate": 1.944429515324749e-06,
      "loss": 0.1656,
      "step": 8975
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3149312056892075,
      "learning_rate": 1.943404172884598e-06,
      "loss": 0.1842,
      "step": 8976
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3631057634103574,
      "learning_rate": 1.9423790356429815e-06,
      "loss": 0.1854,
      "step": 8977
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2898927589982514,
      "learning_rate": 1.9413541036687272e-06,
      "loss": 0.1927,
      "step": 8978
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5952432585175471,
      "learning_rate": 1.9403293770306414e-06,
      "loss": 0.1702,
      "step": 8979
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.507514038451187,
      "learning_rate": 1.939304855797511e-06,
      "loss": 0.5915,
      "step": 8980
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4655712159404735,
      "learning_rate": 1.9382805400381206e-06,
      "loss": 0.1734,
      "step": 8981
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.157475564032177,
      "learning_rate": 1.9372564298212343e-06,
      "loss": 0.1617,
      "step": 8982
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.707654693270186,
      "learning_rate": 1.936232525215604e-06,
      "loss": 0.657,
      "step": 8983
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2320959475797324,
      "learning_rate": 1.935208826289967e-06,
      "loss": 0.1324,
      "step": 8984
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4570816928890984,
      "learning_rate": 1.9341853331130472e-06,
      "loss": 0.1779,
      "step": 8985
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8486861670890355,
      "learning_rate": 1.933162045753554e-06,
      "loss": 0.1765,
      "step": 8986
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2213251735617248,
      "learning_rate": 1.9321389642801875e-06,
      "loss": 0.1598,
      "step": 8987
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.928762655034381,
      "learning_rate": 1.931116088761626e-06,
      "loss": 0.5843,
      "step": 8988
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6031241876257238,
      "learning_rate": 1.9300934192665383e-06,
      "loss": 0.2282,
      "step": 8989
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3370554296302009,
      "learning_rate": 1.929070955863584e-06,
      "loss": 0.1954,
      "step": 8990
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2593402690332172,
      "learning_rate": 1.9280486986213976e-06,
      "loss": 0.1537,
      "step": 8991
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2433238086882437,
      "learning_rate": 1.9270266476086115e-06,
      "loss": 0.1173,
      "step": 8992
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3582152451748397,
      "learning_rate": 1.926004802893837e-06,
      "loss": 0.1753,
      "step": 8993
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.13069566882756,
      "learning_rate": 1.9249831645456744e-06,
      "loss": 0.4784,
      "step": 8994
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.444084739450104,
      "learning_rate": 1.923961732632709e-06,
      "loss": 0.4935,
      "step": 8995
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.454621925803563,
      "learning_rate": 1.922940507223512e-06,
      "loss": 0.1627,
      "step": 8996
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3768881871308623,
      "learning_rate": 1.9219194883866423e-06,
      "loss": 0.1553,
      "step": 8997
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.416366284882426,
      "learning_rate": 1.9208986761906434e-06,
      "loss": 0.1783,
      "step": 8998
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3985872053516994,
      "learning_rate": 1.919878070704045e-06,
      "loss": 0.2194,
      "step": 8999
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.47871055597183,
      "learning_rate": 1.9188576719953635e-06,
      "loss": 0.236,
      "step": 9000
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4647047385880896,
      "learning_rate": 1.917837480133103e-06,
      "loss": 0.1785,
      "step": 9001
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6027860735149817,
      "learning_rate": 1.9168174951857514e-06,
      "loss": 0.1887,
      "step": 9002
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3959321590021587,
      "learning_rate": 1.915797717221783e-06,
      "loss": 0.2174,
      "step": 9003
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.787307661870827,
      "learning_rate": 1.9147781463096583e-06,
      "loss": 0.5173,
      "step": 9004
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.219231121041604,
      "learning_rate": 1.9137587825178243e-06,
      "loss": 0.4628,
      "step": 9005
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.58597382904973,
      "learning_rate": 1.9127396259147147e-06,
      "loss": 0.5857,
      "step": 9006
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.53125390728812,
      "learning_rate": 1.9117206765687478e-06,
      "loss": 0.1922,
      "step": 9007
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2925076690762765,
      "learning_rate": 1.910701934548329e-06,
      "loss": 0.1527,
      "step": 9008
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.986660858718342,
      "learning_rate": 1.9096833999218485e-06,
      "loss": 0.5722,
      "step": 9009
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5049045595525858,
      "learning_rate": 1.908665072757687e-06,
      "loss": 0.1701,
      "step": 9010
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3013312296424395,
      "learning_rate": 1.9076469531242027e-06,
      "loss": 0.1739,
      "step": 9011
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.3234320645364,
      "learning_rate": 1.9066290410897492e-06,
      "loss": 0.57,
      "step": 9012
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4188351085388629,
      "learning_rate": 1.9056113367226615e-06,
      "loss": 0.1867,
      "step": 9013
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5266610457055567,
      "learning_rate": 1.9045938400912594e-06,
      "loss": 0.1906,
      "step": 9014
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5998476745666153,
      "learning_rate": 1.903576551263852e-06,
      "loss": 0.2458,
      "step": 9015
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1725142056660862,
      "learning_rate": 1.9025594703087324e-06,
      "loss": 0.1578,
      "step": 9016
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.346640033083446,
      "learning_rate": 1.9015425972941809e-06,
      "loss": 0.188,
      "step": 9017
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5046373584086468,
      "learning_rate": 1.9005259322884623e-06,
      "loss": 0.184,
      "step": 9018
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4630580127232968,
      "learning_rate": 1.8995094753598293e-06,
      "loss": 0.2007,
      "step": 9019
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2287556456925455,
      "learning_rate": 1.8984932265765172e-06,
      "loss": 0.1206,
      "step": 9020
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1752296313228991,
      "learning_rate": 1.8974771860067554e-06,
      "loss": 0.1728,
      "step": 9021
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2871000491879698,
      "learning_rate": 1.8964613537187471e-06,
      "loss": 0.2006,
      "step": 9022
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.91631462333189,
      "learning_rate": 1.8954457297806934e-06,
      "loss": 0.5449,
      "step": 9023
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3739381624627762,
      "learning_rate": 1.8944303142607757e-06,
      "loss": 0.1614,
      "step": 9024
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3703448006553145,
      "learning_rate": 1.8934151072271573e-06,
      "loss": 0.1778,
      "step": 9025
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4427572437740512,
      "learning_rate": 1.8924001087479964e-06,
      "loss": 0.1888,
      "step": 9026
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4269691327156102,
      "learning_rate": 1.8913853188914328e-06,
      "loss": 0.2005,
      "step": 9027
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3619245569886933,
      "learning_rate": 1.8903707377255908e-06,
      "loss": 0.1646,
      "step": 9028
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3632806771743786,
      "learning_rate": 1.8893563653185831e-06,
      "loss": 0.164,
      "step": 9029
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.891872794522661,
      "learning_rate": 1.8883422017385078e-06,
      "loss": 0.6036,
      "step": 9030
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2510058092993064,
      "learning_rate": 1.8873282470534482e-06,
      "loss": 0.1655,
      "step": 9031
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.343872727473007,
      "learning_rate": 1.8863145013314743e-06,
      "loss": 0.537,
      "step": 9032
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3497207121481087,
      "learning_rate": 1.8853009646406422e-06,
      "loss": 0.1832,
      "step": 9033
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.106506353877256,
      "learning_rate": 1.8842876370489916e-06,
      "loss": 0.61,
      "step": 9034
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5031398259556004,
      "learning_rate": 1.883274518624556e-06,
      "loss": 0.1875,
      "step": 9035
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.527589239881181,
      "learning_rate": 1.8822616094353414e-06,
      "loss": 0.2137,
      "step": 9036
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2357193565509907,
      "learning_rate": 1.8812489095493531e-06,
      "loss": 0.1461,
      "step": 9037
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.460796920353672,
      "learning_rate": 1.8802364190345751e-06,
      "loss": 0.1995,
      "step": 9038
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.377444800361589,
      "learning_rate": 1.8792241379589787e-06,
      "loss": 0.1724,
      "step": 9039
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3899413562607357,
      "learning_rate": 1.8782120663905218e-06,
      "loss": 0.1825,
      "step": 9040
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5718998315126693,
      "learning_rate": 1.8772002043971472e-06,
      "loss": 0.2286,
      "step": 9041
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3161028412589368,
      "learning_rate": 1.8761885520467843e-06,
      "loss": 0.1617,
      "step": 9042
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4837234358946882,
      "learning_rate": 1.8751771094073474e-06,
      "loss": 0.1907,
      "step": 9043
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.434352862307001,
      "learning_rate": 1.8741658765467425e-06,
      "loss": 0.1706,
      "step": 9044
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2339523146681706,
      "learning_rate": 1.8731548535328497e-06,
      "loss": 0.1349,
      "step": 9045
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2312461382027569,
      "learning_rate": 1.8721440404335472e-06,
      "loss": 0.195,
      "step": 9046
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4088767929434909,
      "learning_rate": 1.8711334373166923e-06,
      "loss": 0.1845,
      "step": 9047
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.852212003456255,
      "learning_rate": 1.8701230442501295e-06,
      "loss": 0.4429,
      "step": 9048
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3596457027927031,
      "learning_rate": 1.86911286130169e-06,
      "loss": 0.1894,
      "step": 9049
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.267271473852765,
      "learning_rate": 1.8681028885391905e-06,
      "loss": 0.1597,
      "step": 9050
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.86910531694426,
      "learning_rate": 1.8670931260304336e-06,
      "loss": 0.5523,
      "step": 9051
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2256981376416658,
      "learning_rate": 1.8660835738432071e-06,
      "loss": 0.1831,
      "step": 9052
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.346391409208626,
      "learning_rate": 1.8650742320452858e-06,
      "loss": 0.1651,
      "step": 9053
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.163021428506186,
      "learning_rate": 1.864065100704428e-06,
      "loss": 0.161,
      "step": 9054
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.319772788648218,
      "learning_rate": 1.863056179888385e-06,
      "loss": 0.1565,
      "step": 9055
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3685326345418296,
      "learning_rate": 1.8620474696648822e-06,
      "loss": 0.1907,
      "step": 9056
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.171267534366849,
      "learning_rate": 1.8610389701016412e-06,
      "loss": 0.1356,
      "step": 9057
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4527260649860825,
      "learning_rate": 1.860030681266367e-06,
      "loss": 0.1935,
      "step": 9058
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.562844654247753,
      "learning_rate": 1.8590226032267438e-06,
      "loss": 0.2084,
      "step": 9059
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.243419185090541,
      "learning_rate": 1.8580147360504513e-06,
      "loss": 0.182,
      "step": 9060
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.178463100952765,
      "learning_rate": 1.8570070798051492e-06,
      "loss": 0.1597,
      "step": 9061
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.401016803160156,
      "learning_rate": 1.8559996345584853e-06,
      "loss": 0.2113,
      "step": 9062
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2701134727499686,
      "learning_rate": 1.8549924003780918e-06,
      "loss": 0.1782,
      "step": 9063
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8300056128021036,
      "learning_rate": 1.8539853773315869e-06,
      "loss": 0.1309,
      "step": 9064
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4755951844743371,
      "learning_rate": 1.8529785654865744e-06,
      "loss": 0.2041,
      "step": 9065
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2414245381253806,
      "learning_rate": 1.8519719649106493e-06,
      "loss": 0.1611,
      "step": 9066
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.157672311174175,
      "learning_rate": 1.8509655756713823e-06,
      "loss": 0.1657,
      "step": 9067
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2777072949644641,
      "learning_rate": 1.8499593978363363e-06,
      "loss": 0.1424,
      "step": 9068
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5135583124810161,
      "learning_rate": 1.8489534314730634e-06,
      "loss": 0.2181,
      "step": 9069
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.46751220400743,
      "learning_rate": 1.84794767664909e-06,
      "loss": 0.2033,
      "step": 9070
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.68212170102838,
      "learning_rate": 1.8469421334319416e-06,
      "loss": 0.5123,
      "step": 9071
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3363714638406035,
      "learning_rate": 1.8459368018891211e-06,
      "loss": 0.1731,
      "step": 9072
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2602180715452724,
      "learning_rate": 1.8449316820881192e-06,
      "loss": 0.1834,
      "step": 9073
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2009255504542025,
      "learning_rate": 1.8439267740964135e-06,
      "loss": 0.1493,
      "step": 9074
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.316314397381077,
      "learning_rate": 1.8429220779814654e-06,
      "loss": 0.1512,
      "step": 9075
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.356786016465121,
      "learning_rate": 1.8419175938107241e-06,
      "loss": 0.204,
      "step": 9076
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3761389580037218,
      "learning_rate": 1.8409133216516235e-06,
      "loss": 0.1431,
      "step": 9077
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.349315800941496,
      "learning_rate": 1.8399092615715831e-06,
      "loss": 0.3529,
      "step": 9078
    },
    {
      "epoch": 0.73,
      "grad_norm": 11.839520506734722,
      "learning_rate": 1.838905413638007e-06,
      "loss": 0.627,
      "step": 9079
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.492488805802242,
      "learning_rate": 1.837901777918291e-06,
      "loss": 0.2119,
      "step": 9080
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6033427980714965,
      "learning_rate": 1.836898354479807e-06,
      "loss": 0.2181,
      "step": 9081
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2334260100206724,
      "learning_rate": 1.8358951433899214e-06,
      "loss": 0.1452,
      "step": 9082
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2876735053287762,
      "learning_rate": 1.8348921447159818e-06,
      "loss": 0.1637,
      "step": 9083
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.44960032521332,
      "learning_rate": 1.8338893585253225e-06,
      "loss": 0.1969,
      "step": 9084
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.217997157541277,
      "learning_rate": 1.8328867848852633e-06,
      "loss": 0.1794,
      "step": 9085
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3901485362419175,
      "learning_rate": 1.8318844238631106e-06,
      "loss": 0.1835,
      "step": 9086
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4451519436859581,
      "learning_rate": 1.8308822755261551e-06,
      "loss": 0.1957,
      "step": 9087
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2626627471763259,
      "learning_rate": 1.8298803399416732e-06,
      "loss": 0.1849,
      "step": 9088
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.59624752127837,
      "learning_rate": 1.8288786171769323e-06,
      "loss": 0.6307,
      "step": 9089
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2955364080876017,
      "learning_rate": 1.8278771072991748e-06,
      "loss": 0.1623,
      "step": 9090
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.381177841983565,
      "learning_rate": 1.8268758103756396e-06,
      "loss": 0.1815,
      "step": 9091
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4769305224715712,
      "learning_rate": 1.825874726473547e-06,
      "loss": 0.1741,
      "step": 9092
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2338641845515417,
      "learning_rate": 1.824873855660098e-06,
      "loss": 0.156,
      "step": 9093
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4439577592167092,
      "learning_rate": 1.8238731980024892e-06,
      "loss": 0.2064,
      "step": 9094
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4243003326926151,
      "learning_rate": 1.8228727535678959e-06,
      "loss": 0.1782,
      "step": 9095
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.391349945278128,
      "learning_rate": 1.8218725224234806e-06,
      "loss": 0.1903,
      "step": 9096
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.377665989122746,
      "learning_rate": 1.8208725046363924e-06,
      "loss": 0.1813,
      "step": 9097
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4098146012303618,
      "learning_rate": 1.8198727002737653e-06,
      "loss": 0.1904,
      "step": 9098
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.773847920528601,
      "learning_rate": 1.8188731094027178e-06,
      "loss": 0.6139,
      "step": 9099
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1951072495858535,
      "learning_rate": 1.81787373209036e-06,
      "loss": 0.1538,
      "step": 9100
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3115695788184494,
      "learning_rate": 1.8168745684037787e-06,
      "loss": 0.1656,
      "step": 9101
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.549892380551993,
      "learning_rate": 1.81587561841005e-06,
      "loss": 0.1716,
      "step": 9102
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.859074154748873,
      "learning_rate": 1.8148768821762425e-06,
      "loss": 0.513,
      "step": 9103
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.07762320570059,
      "learning_rate": 1.8138783597693965e-06,
      "loss": 0.397,
      "step": 9104
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.666020303578491,
      "learning_rate": 1.8128800512565514e-06,
      "loss": 0.1835,
      "step": 9105
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.204458757872615,
      "learning_rate": 1.8118819567047252e-06,
      "loss": 0.5086,
      "step": 9106
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.300713819667264,
      "learning_rate": 1.8108840761809232e-06,
      "loss": 0.1784,
      "step": 9107
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4458389900729802,
      "learning_rate": 1.8098864097521358e-06,
      "loss": 0.145,
      "step": 9108
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4692793185726798,
      "learning_rate": 1.808888957485339e-06,
      "loss": 0.1571,
      "step": 9109
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.387262291171904,
      "learning_rate": 1.8078917194474954e-06,
      "loss": 0.5367,
      "step": 9110
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2841130242550987,
      "learning_rate": 1.8068946957055521e-06,
      "loss": 0.1753,
      "step": 9111
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3658696655829443,
      "learning_rate": 1.8058978863264426e-06,
      "loss": 0.1834,
      "step": 9112
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1193693847613664,
      "learning_rate": 1.8049012913770842e-06,
      "loss": 0.1257,
      "step": 9113
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3748410745207416,
      "learning_rate": 1.8039049109243861e-06,
      "loss": 0.1362,
      "step": 9114
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3949136419365638,
      "learning_rate": 1.8029087450352323e-06,
      "loss": 0.1784,
      "step": 9115
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6000969253971802,
      "learning_rate": 1.8019127937765024e-06,
      "loss": 0.1968,
      "step": 9116
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2878760047334998,
      "learning_rate": 1.8009170572150563e-06,
      "loss": 0.1498,
      "step": 9117
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.354325901892797,
      "learning_rate": 1.7999215354177412e-06,
      "loss": 0.6027,
      "step": 9118
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3800090408292782,
      "learning_rate": 1.7989262284513897e-06,
      "loss": 0.1886,
      "step": 9119
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1320182271889063,
      "learning_rate": 1.797931136382819e-06,
      "loss": 0.169,
      "step": 9120
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1833916862645706,
      "learning_rate": 1.7969362592788331e-06,
      "loss": 0.13,
      "step": 9121
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2661554648471567,
      "learning_rate": 1.795941597206221e-06,
      "loss": 0.1504,
      "step": 9122
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4274754951410569,
      "learning_rate": 1.7949471502317572e-06,
      "loss": 0.1881,
      "step": 9123
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3987655227780484,
      "learning_rate": 1.7939529184222004e-06,
      "loss": 0.2016,
      "step": 9124
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5070902294649455,
      "learning_rate": 1.7929589018443016e-06,
      "loss": 0.2186,
      "step": 9125
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.673346831008698,
      "learning_rate": 1.7919651005647852e-06,
      "loss": 0.546,
      "step": 9126
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.411344556144927,
      "learning_rate": 1.7909715146503736e-06,
      "loss": 0.1995,
      "step": 9127
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2296788567845378,
      "learning_rate": 1.7899781441677688e-06,
      "loss": 0.1382,
      "step": 9128
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.723067033845056,
      "learning_rate": 1.7889849891836537e-06,
      "loss": 0.4328,
      "step": 9129
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2377433870944272,
      "learning_rate": 1.7879920497647068e-06,
      "loss": 0.1657,
      "step": 9130
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3664850067175895,
      "learning_rate": 1.786999325977586e-06,
      "loss": 0.1894,
      "step": 9131
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2488048557385354,
      "learning_rate": 1.786006817888935e-06,
      "loss": 0.1476,
      "step": 9132
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.415161664205637,
      "learning_rate": 1.7850145255653828e-06,
      "loss": 0.1517,
      "step": 9133
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3116741753313503,
      "learning_rate": 1.7840224490735498e-06,
      "loss": 0.1601,
      "step": 9134
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2280865705221102,
      "learning_rate": 1.7830305884800302e-06,
      "loss": 0.1658,
      "step": 9135
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3332937180170799,
      "learning_rate": 1.782038943851417e-06,
      "loss": 0.1792,
      "step": 9136
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1655021043449045,
      "learning_rate": 1.78104751525428e-06,
      "loss": 0.1308,
      "step": 9137
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4056584339153708,
      "learning_rate": 1.7800563027551737e-06,
      "loss": 0.1573,
      "step": 9138
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3676531582355078,
      "learning_rate": 1.7790653064206454e-06,
      "loss": 0.173,
      "step": 9139
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.914572675069034,
      "learning_rate": 1.7780745263172216e-06,
      "loss": 0.5155,
      "step": 9140
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4437519724786982,
      "learning_rate": 1.7770839625114173e-06,
      "loss": 0.2257,
      "step": 9141
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3175726896556128,
      "learning_rate": 1.7760936150697316e-06,
      "loss": 0.1661,
      "step": 9142
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.8282546548058,
      "learning_rate": 1.7751034840586495e-06,
      "loss": 0.4521,
      "step": 9143
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2015361374381475,
      "learning_rate": 1.77411356954464e-06,
      "loss": 0.1311,
      "step": 9144
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3093938978987165,
      "learning_rate": 1.773123871594164e-06,
      "loss": 0.1902,
      "step": 9145
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.236017886291195,
      "learning_rate": 1.7721343902736577e-06,
      "loss": 0.1671,
      "step": 9146
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4171283963075427,
      "learning_rate": 1.7711451256495482e-06,
      "loss": 0.197,
      "step": 9147
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2078737330068696,
      "learning_rate": 1.7701560777882531e-06,
      "loss": 0.1359,
      "step": 9148
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3276099376275687,
      "learning_rate": 1.7691672467561627e-06,
      "loss": 0.1679,
      "step": 9149
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1288656214939443,
      "learning_rate": 1.7681786326196665e-06,
      "loss": 0.1292,
      "step": 9150
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.461758550108675,
      "learning_rate": 1.7671902354451298e-06,
      "loss": 0.5478,
      "step": 9151
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3937972455820422,
      "learning_rate": 1.7662020552989085e-06,
      "loss": 0.1785,
      "step": 9152
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2031226628548526,
      "learning_rate": 1.7652140922473403e-06,
      "loss": 0.1758,
      "step": 9153
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4401764649614757,
      "learning_rate": 1.7642263463567517e-06,
      "loss": 0.2074,
      "step": 9154
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4707547211423908,
      "learning_rate": 1.7632388176934523e-06,
      "loss": 0.1971,
      "step": 9155
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4703205804579635,
      "learning_rate": 1.7622515063237382e-06,
      "loss": 0.2004,
      "step": 9156
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6052850321146184,
      "learning_rate": 1.7612644123138906e-06,
      "loss": 0.1982,
      "step": 9157
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.012456061399446,
      "learning_rate": 1.7602775357301738e-06,
      "loss": 0.5874,
      "step": 9158
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3071494090716804,
      "learning_rate": 1.7592908766388456e-06,
      "loss": 0.199,
      "step": 9159
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3281227183022555,
      "learning_rate": 1.7583044351061369e-06,
      "loss": 0.1382,
      "step": 9160
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3780077078245516,
      "learning_rate": 1.7573182111982745e-06,
      "loss": 0.1526,
      "step": 9161
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6010940253210684,
      "learning_rate": 1.7563322049814674e-06,
      "loss": 0.2028,
      "step": 9162
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3511678098730722,
      "learning_rate": 1.7553464165219036e-06,
      "loss": 0.2116,
      "step": 9163
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4354990201199087,
      "learning_rate": 1.754360845885768e-06,
      "loss": 0.1761,
      "step": 9164
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3811452135719158,
      "learning_rate": 1.7533754931392227e-06,
      "loss": 0.1645,
      "step": 9165
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3561983613114927,
      "learning_rate": 1.752390358348417e-06,
      "loss": 0.1735,
      "step": 9166
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.444992386053985,
      "learning_rate": 1.7514054415794868e-06,
      "loss": 0.1877,
      "step": 9167
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6217837162963722,
      "learning_rate": 1.750420742898552e-06,
      "loss": 0.1834,
      "step": 9168
    },
    {
      "epoch": 0.73,
      "grad_norm": 11.361423833064933,
      "learning_rate": 1.7494362623717166e-06,
      "loss": 0.5676,
      "step": 9169
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3386263027220062,
      "learning_rate": 1.7484520000650757e-06,
      "loss": 0.1323,
      "step": 9170
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2904389289987186,
      "learning_rate": 1.7474679560447056e-06,
      "loss": 0.1672,
      "step": 9171
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.66356773113316,
      "learning_rate": 1.7464841303766628e-06,
      "loss": 0.5819,
      "step": 9172
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3389462043080522,
      "learning_rate": 1.745500523127e-06,
      "loss": 0.1728,
      "step": 9173
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.249835068796126,
      "learning_rate": 1.7445171343617473e-06,
      "loss": 0.5887,
      "step": 9174
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4742614594215442,
      "learning_rate": 1.743533964146924e-06,
      "loss": 0.192,
      "step": 9175
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6508322006594116,
      "learning_rate": 1.7425510125485318e-06,
      "loss": 0.1721,
      "step": 9176
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2952475706868518,
      "learning_rate": 1.7415682796325595e-06,
      "loss": 0.1765,
      "step": 9177
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4100633240113714,
      "learning_rate": 1.7405857654649798e-06,
      "loss": 0.1742,
      "step": 9178
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.417327823333738,
      "learning_rate": 1.739603470111757e-06,
      "loss": 0.1913,
      "step": 9179
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2475244160830607,
      "learning_rate": 1.7386213936388303e-06,
      "loss": 0.1828,
      "step": 9180
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.684409191312877,
      "learning_rate": 1.737639536112129e-06,
      "loss": 0.5451,
      "step": 9181
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3792925771356725,
      "learning_rate": 1.7366578975975734e-06,
      "loss": 0.2137,
      "step": 9182
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4230665283851671,
      "learning_rate": 1.7356764781610574e-06,
      "loss": 0.1814,
      "step": 9183
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4349885751563647,
      "learning_rate": 1.7346952778684718e-06,
      "loss": 0.2046,
      "step": 9184
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3007414196960718,
      "learning_rate": 1.7337142967856857e-06,
      "loss": 0.1754,
      "step": 9185
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5080921896753834,
      "learning_rate": 1.7327335349785552e-06,
      "loss": 0.1955,
      "step": 9186
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.23886854470887,
      "learning_rate": 1.7317529925129217e-06,
      "loss": 0.1718,
      "step": 9187
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1889302492512888,
      "learning_rate": 1.7307726694546128e-06,
      "loss": 0.1582,
      "step": 9188
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.350155097170726,
      "learning_rate": 1.7297925658694393e-06,
      "loss": 0.2031,
      "step": 9189
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5503632463867265,
      "learning_rate": 1.7288126818231998e-06,
      "loss": 0.2062,
      "step": 9190
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.642558768113016,
      "learning_rate": 1.727833017381676e-06,
      "loss": 0.4599,
      "step": 9191
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4082854346195635,
      "learning_rate": 1.726853572610634e-06,
      "loss": 0.1664,
      "step": 9192
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.35170824266716,
      "learning_rate": 1.7258743475758328e-06,
      "loss": 0.1956,
      "step": 9193
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4355182429581022,
      "learning_rate": 1.724895342343003e-06,
      "loss": 0.2212,
      "step": 9194
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3171630102653067,
      "learning_rate": 1.7239165569778738e-06,
      "loss": 0.1831,
      "step": 9195
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.390088773667043,
      "learning_rate": 1.7229379915461526e-06,
      "loss": 0.1783,
      "step": 9196
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.321707622942941,
      "learning_rate": 1.721959646113533e-06,
      "loss": 0.1883,
      "step": 9197
    },
    {
      "epoch": 0.74,
      "grad_norm": 14.605022186230592,
      "learning_rate": 1.7209815207456941e-06,
      "loss": 0.7236,
      "step": 9198
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4367012060529436,
      "learning_rate": 1.720003615508301e-06,
      "loss": 0.2162,
      "step": 9199
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.480257189356589,
      "learning_rate": 1.7190259304670038e-06,
      "loss": 0.5246,
      "step": 9200
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.319189640947264,
      "learning_rate": 1.7180484656874357e-06,
      "loss": 0.1557,
      "step": 9201
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2914883112552646,
      "learning_rate": 1.7170712212352187e-06,
      "loss": 0.1975,
      "step": 9202
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4158668217907435,
      "learning_rate": 1.7160941971759558e-06,
      "loss": 0.1862,
      "step": 9203
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3265497105863124,
      "learning_rate": 1.7151173935752425e-06,
      "loss": 0.1581,
      "step": 9204
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.578711914848806,
      "learning_rate": 1.714140810498648e-06,
      "loss": 0.1838,
      "step": 9205
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.20358599873575,
      "learning_rate": 1.7131644480117381e-06,
      "loss": 0.14,
      "step": 9206
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1304955742676857,
      "learning_rate": 1.7121883061800598e-06,
      "loss": 0.1137,
      "step": 9207
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.313040639447187,
      "learning_rate": 1.7112123850691386e-06,
      "loss": 0.1767,
      "step": 9208
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.411500047521928,
      "learning_rate": 1.7102366847444963e-06,
      "loss": 0.155,
      "step": 9209
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3040926799210366,
      "learning_rate": 1.709261205271633e-06,
      "loss": 0.1775,
      "step": 9210
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3588452174313836,
      "learning_rate": 1.7082859467160351e-06,
      "loss": 0.1573,
      "step": 9211
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3604806158128333,
      "learning_rate": 1.7073109091431734e-06,
      "loss": 0.1633,
      "step": 9212
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.280809109485256,
      "learning_rate": 1.7063360926185108e-06,
      "loss": 0.5909,
      "step": 9213
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2762231048290287,
      "learning_rate": 1.7053614972074833e-06,
      "loss": 0.1428,
      "step": 9214
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4394083421690795,
      "learning_rate": 1.7043871229755198e-06,
      "loss": 0.1511,
      "step": 9215
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4694081397116232,
      "learning_rate": 1.703412969988037e-06,
      "loss": 0.1854,
      "step": 9216
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2221345946599484,
      "learning_rate": 1.7024390383104267e-06,
      "loss": 0.1489,
      "step": 9217
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3768626461435447,
      "learning_rate": 1.7014653280080768e-06,
      "loss": 0.1993,
      "step": 9218
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4508423976464029,
      "learning_rate": 1.7004918391463531e-06,
      "loss": 0.1832,
      "step": 9219
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.586727272575284,
      "learning_rate": 1.6995185717906092e-06,
      "loss": 0.1718,
      "step": 9220
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.322618490027282,
      "learning_rate": 1.6985455260061845e-06,
      "loss": 0.1733,
      "step": 9221
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.147784180702518,
      "learning_rate": 1.697572701858401e-06,
      "loss": 0.1611,
      "step": 9222
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.4411954352452305,
      "learning_rate": 1.6966000994125682e-06,
      "loss": 0.5393,
      "step": 9223
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.243911175939272,
      "learning_rate": 1.6956277187339798e-06,
      "loss": 0.1726,
      "step": 9224
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5517853781787916,
      "learning_rate": 1.6946555598879138e-06,
      "loss": 0.2118,
      "step": 9225
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2757837179388372,
      "learning_rate": 1.6936836229396336e-06,
      "loss": 0.1813,
      "step": 9226
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.355245094856216,
      "learning_rate": 1.692711907954393e-06,
      "loss": 0.1849,
      "step": 9227
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.873139660448752,
      "learning_rate": 1.6917404149974193e-06,
      "loss": 0.5636,
      "step": 9228
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2488597610842453,
      "learning_rate": 1.690769144133937e-06,
      "loss": 0.153,
      "step": 9229
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4749167905824174,
      "learning_rate": 1.6897980954291483e-06,
      "loss": 0.1895,
      "step": 9230
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3212231119976772,
      "learning_rate": 1.6888272689482433e-06,
      "loss": 0.1511,
      "step": 9231
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.562784536519942,
      "learning_rate": 1.687856664756396e-06,
      "loss": 0.623,
      "step": 9232
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.409321025546184,
      "learning_rate": 1.686886282918766e-06,
      "loss": 0.2025,
      "step": 9233
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.261353951473355,
      "learning_rate": 1.6859161235004984e-06,
      "loss": 0.1403,
      "step": 9234
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3172663320919502,
      "learning_rate": 1.6849461865667226e-06,
      "loss": 0.1504,
      "step": 9235
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.473388485476268,
      "learning_rate": 1.6839764721825535e-06,
      "loss": 0.2052,
      "step": 9236
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.416037488662575,
      "learning_rate": 1.6830069804130894e-06,
      "loss": 0.1912,
      "step": 9237
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1759164300042428,
      "learning_rate": 1.6820377113234203e-06,
      "loss": 0.1624,
      "step": 9238
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2422302204738789,
      "learning_rate": 1.681068664978609e-06,
      "loss": 0.1398,
      "step": 9239
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4192031276831634,
      "learning_rate": 1.6800998414437165e-06,
      "loss": 0.2122,
      "step": 9240
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.305504551274611,
      "learning_rate": 1.6791312407837812e-06,
      "loss": 0.186,
      "step": 9241
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3942829190287833,
      "learning_rate": 1.6781628630638247e-06,
      "loss": 0.1688,
      "step": 9242
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4623816778595087,
      "learning_rate": 1.677194708348862e-06,
      "loss": 0.1932,
      "step": 9243
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4062078042403683,
      "learning_rate": 1.676226776703886e-06,
      "loss": 0.1986,
      "step": 9244
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3661368554371822,
      "learning_rate": 1.6752590681938768e-06,
      "loss": 0.1898,
      "step": 9245
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.26341414816587,
      "learning_rate": 1.6742915828838003e-06,
      "loss": 0.1672,
      "step": 9246
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2678653255602448,
      "learning_rate": 1.6733243208386057e-06,
      "loss": 0.1477,
      "step": 9247
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2497552078537733,
      "learning_rate": 1.6723572821232275e-06,
      "loss": 0.1494,
      "step": 9248
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2592193533709861,
      "learning_rate": 1.671390466802591e-06,
      "loss": 0.1675,
      "step": 9249
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2422435095041522,
      "learning_rate": 1.6704238749415958e-06,
      "loss": 0.1604,
      "step": 9250
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3542371304434624,
      "learning_rate": 1.6694575066051327e-06,
      "loss": 0.1489,
      "step": 9251
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.37494241514877,
      "learning_rate": 1.6684913618580811e-06,
      "loss": 0.3888,
      "step": 9252
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4749907282645558,
      "learning_rate": 1.6675254407652958e-06,
      "loss": 0.1498,
      "step": 9253
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3655219194332666,
      "learning_rate": 1.666559743391626e-06,
      "loss": 0.1495,
      "step": 9254
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2238659357015582,
      "learning_rate": 1.6655942698019001e-06,
      "loss": 0.158,
      "step": 9255
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2447053038839422,
      "learning_rate": 1.6646290200609344e-06,
      "loss": 0.1825,
      "step": 9256
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3157849727553947,
      "learning_rate": 1.6636639942335264e-06,
      "loss": 0.1912,
      "step": 9257
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.532190861284905,
      "learning_rate": 1.6626991923844666e-06,
      "loss": 0.2227,
      "step": 9258
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3017882347859238,
      "learning_rate": 1.6617346145785196e-06,
      "loss": 0.1873,
      "step": 9259
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2399111343791365,
      "learning_rate": 1.6607702608804416e-06,
      "loss": 0.1581,
      "step": 9260
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2322812395962557,
      "learning_rate": 1.6598061313549763e-06,
      "loss": 0.1223,
      "step": 9261
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3202594536486618,
      "learning_rate": 1.6588422260668425e-06,
      "loss": 0.1901,
      "step": 9262
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.887564970257272,
      "learning_rate": 1.6578785450807545e-06,
      "loss": 0.6083,
      "step": 9263
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3644309077112091,
      "learning_rate": 1.6569150884614067e-06,
      "loss": 0.1712,
      "step": 9264
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.236849750581803,
      "learning_rate": 1.6559518562734777e-06,
      "loss": 0.1842,
      "step": 9265
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3231989287190236,
      "learning_rate": 1.654988848581633e-06,
      "loss": 0.2158,
      "step": 9266
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4568027585555277,
      "learning_rate": 1.654026065450522e-06,
      "loss": 0.1919,
      "step": 9267
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2997956067430405,
      "learning_rate": 1.6530635069447787e-06,
      "loss": 0.1588,
      "step": 9268
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2583025044293814,
      "learning_rate": 1.6521011731290232e-06,
      "loss": 0.137,
      "step": 9269
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4121831615704283,
      "learning_rate": 1.6511390640678592e-06,
      "loss": 0.1802,
      "step": 9270
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1690021025018584,
      "learning_rate": 1.6501771798258753e-06,
      "loss": 0.1436,
      "step": 9271
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3477639530979049,
      "learning_rate": 1.6492155204676503e-06,
      "loss": 0.1503,
      "step": 9272
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2437246983368497,
      "learning_rate": 1.6482540860577368e-06,
      "loss": 0.1789,
      "step": 9273
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3164176830262688,
      "learning_rate": 1.6472928766606828e-06,
      "loss": 0.1981,
      "step": 9274
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.853506558062393,
      "learning_rate": 1.6463318923410183e-06,
      "loss": 0.5632,
      "step": 9275
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.554141924999955,
      "learning_rate": 1.6453711331632516e-06,
      "loss": 0.1712,
      "step": 9276
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4165565642352327,
      "learning_rate": 1.6444105991918867e-06,
      "loss": 0.1983,
      "step": 9277
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.395133690830671,
      "learning_rate": 1.6434502904914056e-06,
      "loss": 0.1803,
      "step": 9278
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.765559196909282,
      "learning_rate": 1.642490207126276e-06,
      "loss": 0.4761,
      "step": 9279
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1008247602290997,
      "learning_rate": 1.6415303491609519e-06,
      "loss": 0.1566,
      "step": 9280
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4044265421306688,
      "learning_rate": 1.6405707166598712e-06,
      "loss": 0.1525,
      "step": 9281
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.201148571553906,
      "learning_rate": 1.6396113096874549e-06,
      "loss": 0.1411,
      "step": 9282
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5347791889838944,
      "learning_rate": 1.6386521283081163e-06,
      "loss": 0.1762,
      "step": 9283
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.397035950735275,
      "learning_rate": 1.6376931725862438e-06,
      "loss": 0.1386,
      "step": 9284
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3947534737814915,
      "learning_rate": 1.6367344425862136e-06,
      "loss": 0.1844,
      "step": 9285
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2779211471534706,
      "learning_rate": 1.6357759383723943e-06,
      "loss": 0.1584,
      "step": 9286
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2374402719010764,
      "learning_rate": 1.6348176600091265e-06,
      "loss": 0.1521,
      "step": 9287
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.511379302818712,
      "learning_rate": 1.633859607560746e-06,
      "loss": 0.2106,
      "step": 9288
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.363485819857421,
      "learning_rate": 1.6329017810915698e-06,
      "loss": 0.2151,
      "step": 9289
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2560928067368509,
      "learning_rate": 1.6319441806658987e-06,
      "loss": 0.1809,
      "step": 9290
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2658214573477322,
      "learning_rate": 1.6309868063480195e-06,
      "loss": 0.1526,
      "step": 9291
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.460191505969931,
      "learning_rate": 1.630029658202204e-06,
      "loss": 0.4932,
      "step": 9292
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2676719893803832,
      "learning_rate": 1.6290727362927079e-06,
      "loss": 0.1395,
      "step": 9293
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3244213324953213,
      "learning_rate": 1.628116040683772e-06,
      "loss": 0.1776,
      "step": 9294
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5994883085491038,
      "learning_rate": 1.6271595714396233e-06,
      "loss": 0.2004,
      "step": 9295
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3469249171089657,
      "learning_rate": 1.6262033286244706e-06,
      "loss": 0.1822,
      "step": 9296
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.489173908576333,
      "learning_rate": 1.6252473123025132e-06,
      "loss": 0.1601,
      "step": 9297
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3050814138525575,
      "learning_rate": 1.6242915225379259e-06,
      "loss": 0.1503,
      "step": 9298
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5750673609743637,
      "learning_rate": 1.6233359593948777e-06,
      "loss": 0.185,
      "step": 9299
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.201749909127806,
      "learning_rate": 1.6223806229375182e-06,
      "loss": 0.1784,
      "step": 9300
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.4536117859407325,
      "learning_rate": 1.621425513229981e-06,
      "loss": 0.6193,
      "step": 9301
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5494388422491012,
      "learning_rate": 1.620470630336386e-06,
      "loss": 0.2129,
      "step": 9302
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.4522675276715,
      "learning_rate": 1.6195159743208366e-06,
      "loss": 0.4635,
      "step": 9303
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1955909192405902,
      "learning_rate": 1.618561545247423e-06,
      "loss": 0.1658,
      "step": 9304
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.317684304811304,
      "learning_rate": 1.6176073431802158e-06,
      "loss": 0.2214,
      "step": 9305
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.810840328400755,
      "learning_rate": 1.6166533681832797e-06,
      "loss": 0.1784,
      "step": 9306
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4051241753665276,
      "learning_rate": 1.615699620320651e-06,
      "loss": 0.2164,
      "step": 9307
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.350655311438031,
      "learning_rate": 1.6147460996563618e-06,
      "loss": 0.1896,
      "step": 9308
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3227261744109176,
      "learning_rate": 1.6137928062544244e-06,
      "loss": 0.1685,
      "step": 9309
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3608382948963378,
      "learning_rate": 1.6128397401788353e-06,
      "loss": 0.1683,
      "step": 9310
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4523858568971029,
      "learning_rate": 1.6118869014935773e-06,
      "loss": 0.1881,
      "step": 9311
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3265709584342793,
      "learning_rate": 1.6109342902626175e-06,
      "loss": 0.1845,
      "step": 9312
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3045222093432671,
      "learning_rate": 1.6099819065499068e-06,
      "loss": 0.1985,
      "step": 9313
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4297427325254977,
      "learning_rate": 1.6090297504193824e-06,
      "loss": 0.1674,
      "step": 9314
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2857805506530455,
      "learning_rate": 1.6080778219349652e-06,
      "loss": 0.1635,
      "step": 9315
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3513704585722028,
      "learning_rate": 1.6071261211605587e-06,
      "loss": 0.1762,
      "step": 9316
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3606228279863979,
      "learning_rate": 1.6061746481600593e-06,
      "loss": 0.1976,
      "step": 9317
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1964455532862093,
      "learning_rate": 1.6052234029973356e-06,
      "loss": 0.1674,
      "step": 9318
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4047082067376941,
      "learning_rate": 1.604272385736252e-06,
      "loss": 0.1986,
      "step": 9319
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.796150518058644,
      "learning_rate": 1.6033215964406534e-06,
      "loss": 0.5328,
      "step": 9320
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4014825614928708,
      "learning_rate": 1.6023710351743642e-06,
      "loss": 0.1824,
      "step": 9321
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1599802589763089,
      "learning_rate": 1.6014207020012034e-06,
      "loss": 0.1168,
      "step": 9322
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2601789289680414,
      "learning_rate": 1.6004705969849687e-06,
      "loss": 0.1912,
      "step": 9323
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5343333138502464,
      "learning_rate": 1.5995207201894424e-06,
      "loss": 0.1814,
      "step": 9324
    },
    {
      "epoch": 0.75,
      "grad_norm": 18.747394343373486,
      "learning_rate": 1.5985710716783936e-06,
      "loss": 0.627,
      "step": 9325
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.300032923152989,
      "learning_rate": 1.5976216515155746e-06,
      "loss": 0.139,
      "step": 9326
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3334952657094212,
      "learning_rate": 1.5966724597647209e-06,
      "loss": 0.2097,
      "step": 9327
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2723720253733435,
      "learning_rate": 1.5957234964895602e-06,
      "loss": 0.1852,
      "step": 9328
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.339249180252441,
      "learning_rate": 1.5947747617537945e-06,
      "loss": 0.1387,
      "step": 9329
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5230290255531902,
      "learning_rate": 1.5938262556211142e-06,
      "loss": 0.1761,
      "step": 9330
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.35805051320388,
      "learning_rate": 1.5928779781552012e-06,
      "loss": 0.1695,
      "step": 9331
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.258997654406913,
      "learning_rate": 1.5919299294197093e-06,
      "loss": 0.1689,
      "step": 9332
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3070742762293202,
      "learning_rate": 1.5909821094782891e-06,
      "loss": 0.1585,
      "step": 9333
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2073520879048403,
      "learning_rate": 1.5900345183945688e-06,
      "loss": 0.1735,
      "step": 9334
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3524267854993093,
      "learning_rate": 1.589087156232163e-06,
      "loss": 0.1709,
      "step": 9335
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.208146166698898,
      "learning_rate": 1.5881400230546712e-06,
      "loss": 0.1572,
      "step": 9336
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.617003209428942,
      "learning_rate": 1.5871931189256768e-06,
      "loss": 0.6768,
      "step": 9337
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2680820431940585,
      "learning_rate": 1.5862464439087488e-06,
      "loss": 0.1639,
      "step": 9338
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3076368488237644,
      "learning_rate": 1.5852999980674404e-06,
      "loss": 0.1898,
      "step": 9339
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4637877634606185,
      "learning_rate": 1.5843537814652894e-06,
      "loss": 0.2175,
      "step": 9340
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.401853692267398,
      "learning_rate": 1.5834077941658165e-06,
      "loss": 0.2033,
      "step": 9341
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.396712624745321,
      "learning_rate": 1.5824620362325315e-06,
      "loss": 0.2002,
      "step": 9342
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.648326557893304,
      "learning_rate": 1.5815165077289247e-06,
      "loss": 0.7711,
      "step": 9343
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.407157150159328,
      "learning_rate": 1.5805712087184727e-06,
      "loss": 0.1772,
      "step": 9344
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4307108385827259,
      "learning_rate": 1.5796261392646357e-06,
      "loss": 0.1886,
      "step": 9345
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.206561352200287,
      "learning_rate": 1.5786812994308592e-06,
      "loss": 0.182,
      "step": 9346
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.345536603614382,
      "learning_rate": 1.5777366892805735e-06,
      "loss": 0.1773,
      "step": 9347
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3621866577813369,
      "learning_rate": 1.576792308877193e-06,
      "loss": 0.2023,
      "step": 9348
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2919590550630158,
      "learning_rate": 1.5758481582841162e-06,
      "loss": 0.184,
      "step": 9349
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2355816848368182,
      "learning_rate": 1.5749042375647261e-06,
      "loss": 0.1471,
      "step": 9350
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3459914724858888,
      "learning_rate": 1.5739605467823953e-06,
      "loss": 0.1543,
      "step": 9351
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3708064910140392,
      "learning_rate": 1.57301708600047e-06,
      "loss": 0.1811,
      "step": 9352
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1994024936767052,
      "learning_rate": 1.5720738552822929e-06,
      "loss": 0.1586,
      "step": 9353
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3388958208750605,
      "learning_rate": 1.5711308546911859e-06,
      "loss": 0.1761,
      "step": 9354
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.368201819488567,
      "learning_rate": 1.5701880842904503e-06,
      "loss": 0.1747,
      "step": 9355
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.245913086368507,
      "learning_rate": 1.569245544143382e-06,
      "loss": 0.1692,
      "step": 9356
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.874973148266775,
      "learning_rate": 1.5683032343132554e-06,
      "loss": 0.586,
      "step": 9357
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2867405194831163,
      "learning_rate": 1.56736115486333e-06,
      "loss": 0.1414,
      "step": 9358
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4718809798163417,
      "learning_rate": 1.5664193058568505e-06,
      "loss": 0.1908,
      "step": 9359
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.420506295551474,
      "learning_rate": 1.565477687357047e-06,
      "loss": 0.1877,
      "step": 9360
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4261483500855614,
      "learning_rate": 1.5645362994271306e-06,
      "loss": 0.1591,
      "step": 9361
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4713040744058778,
      "learning_rate": 1.5635951421303047e-06,
      "loss": 0.1949,
      "step": 9362
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4247911789022698,
      "learning_rate": 1.562654215529747e-06,
      "loss": 0.1857,
      "step": 9363
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.376379764522003,
      "learning_rate": 1.5617135196886251e-06,
      "loss": 0.1728,
      "step": 9364
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.158168869944392,
      "learning_rate": 1.5607730546700956e-06,
      "loss": 0.5003,
      "step": 9365
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3831778491072337,
      "learning_rate": 1.5598328205372882e-06,
      "loss": 0.1821,
      "step": 9366
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4609672337910595,
      "learning_rate": 1.558892817353328e-06,
      "loss": 0.1994,
      "step": 9367
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2934071496884783,
      "learning_rate": 1.5579530451813197e-06,
      "loss": 0.1532,
      "step": 9368
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3797129027006125,
      "learning_rate": 1.557013504084352e-06,
      "loss": 0.1894,
      "step": 9369
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3038280298391,
      "learning_rate": 1.5560741941254998e-06,
      "loss": 0.1371,
      "step": 9370
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.244811957969455,
      "learning_rate": 1.5551351153678219e-06,
      "loss": 0.6788,
      "step": 9371
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3645078033817268,
      "learning_rate": 1.5541962678743606e-06,
      "loss": 0.1672,
      "step": 9372
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2510419882515846,
      "learning_rate": 1.553257651708145e-06,
      "loss": 0.1507,
      "step": 9373
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3938383786711575,
      "learning_rate": 1.5523192669321858e-06,
      "loss": 0.1551,
      "step": 9374
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.355667226661782,
      "learning_rate": 1.5513811136094786e-06,
      "loss": 0.1524,
      "step": 9375
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5274339936161818,
      "learning_rate": 1.55044319180301e-06,
      "loss": 0.2166,
      "step": 9376
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4408613814447968,
      "learning_rate": 1.5495055015757377e-06,
      "loss": 0.1901,
      "step": 9377
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3228786210345276,
      "learning_rate": 1.5485680429906175e-06,
      "loss": 0.1515,
      "step": 9378
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.326616846397789,
      "learning_rate": 1.5476308161105825e-06,
      "loss": 0.1613,
      "step": 9379
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2268189457206167,
      "learning_rate": 1.5466938209985504e-06,
      "loss": 0.1622,
      "step": 9380
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4602018070776943,
      "learning_rate": 1.5457570577174257e-06,
      "loss": 0.1645,
      "step": 9381
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2905483115486185,
      "learning_rate": 1.5448205263300952e-06,
      "loss": 0.1719,
      "step": 9382
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5059262715438797,
      "learning_rate": 1.543884226899432e-06,
      "loss": 0.214,
      "step": 9383
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4258629349772773,
      "learning_rate": 1.5429481594882905e-06,
      "loss": 0.2125,
      "step": 9384
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.325015343329243,
      "learning_rate": 1.5420123241595169e-06,
      "loss": 0.1846,
      "step": 9385
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.518488103145518,
      "learning_rate": 1.5410767209759298e-06,
      "loss": 0.1792,
      "step": 9386
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4142280390012976,
      "learning_rate": 1.5401413500003443e-06,
      "loss": 0.1797,
      "step": 9387
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.422754865308121,
      "learning_rate": 1.5392062112955524e-06,
      "loss": 0.1551,
      "step": 9388
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2653878894750799,
      "learning_rate": 1.5382713049243336e-06,
      "loss": 0.1589,
      "step": 9389
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.308876400514231,
      "learning_rate": 1.5373366309494515e-06,
      "loss": 0.1953,
      "step": 9390
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3247587030491197,
      "learning_rate": 1.5364021894336517e-06,
      "loss": 0.183,
      "step": 9391
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4539961124735112,
      "learning_rate": 1.5354679804396678e-06,
      "loss": 0.1942,
      "step": 9392
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4486015459406543,
      "learning_rate": 1.5345340040302153e-06,
      "loss": 0.1938,
      "step": 9393
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5192065569005047,
      "learning_rate": 1.533600260267995e-06,
      "loss": 0.2159,
      "step": 9394
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.753101693366677,
      "learning_rate": 1.5326667492156905e-06,
      "loss": 0.6111,
      "step": 9395
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4420065680208252,
      "learning_rate": 1.531733470935976e-06,
      "loss": 0.178,
      "step": 9396
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2974879604420024,
      "learning_rate": 1.5308004254915004e-06,
      "loss": 0.1673,
      "step": 9397
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2610943344961074,
      "learning_rate": 1.5298676129449019e-06,
      "loss": 0.1675,
      "step": 9398
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4652424221987772,
      "learning_rate": 1.5289350333588076e-06,
      "loss": 0.1764,
      "step": 9399
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.353272512092655,
      "learning_rate": 1.5280026867958186e-06,
      "loss": 0.1754,
      "step": 9400
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2256948729947745,
      "learning_rate": 1.5270705733185304e-06,
      "loss": 0.1194,
      "step": 9401
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.23078983120021,
      "learning_rate": 1.5261386929895173e-06,
      "loss": 0.1305,
      "step": 9402
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3413357405182518,
      "learning_rate": 1.5252070458713393e-06,
      "loss": 0.2305,
      "step": 9403
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5429711768944137,
      "learning_rate": 1.52427563202654e-06,
      "loss": 0.159,
      "step": 9404
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3853649450490844,
      "learning_rate": 1.5233444515176488e-06,
      "loss": 0.213,
      "step": 9405
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.075035658825802,
      "learning_rate": 1.5224135044071782e-06,
      "loss": 0.7079,
      "step": 9406
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4105589901166866,
      "learning_rate": 1.5214827907576257e-06,
      "loss": 0.1658,
      "step": 9407
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2749283408560548,
      "learning_rate": 1.5205523106314736e-06,
      "loss": 0.1348,
      "step": 9408
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.24389986257992,
      "learning_rate": 1.5196220640911846e-06,
      "loss": 0.1868,
      "step": 9409
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.39323421657642,
      "learning_rate": 1.5186920511992154e-06,
      "loss": 0.1798,
      "step": 9410
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4008239742941895,
      "learning_rate": 1.5177622720179936e-06,
      "loss": 0.1709,
      "step": 9411
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4552369695828133,
      "learning_rate": 1.5168327266099424e-06,
      "loss": 0.1741,
      "step": 9412
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.406483843143933,
      "learning_rate": 1.515903415037464e-06,
      "loss": 0.195,
      "step": 9413
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6819854428902734,
      "learning_rate": 1.514974337362946e-06,
      "loss": 0.1556,
      "step": 9414
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5179444271714662,
      "learning_rate": 1.5140454936487597e-06,
      "loss": 0.2041,
      "step": 9415
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.972367202012451,
      "learning_rate": 1.513116883957262e-06,
      "loss": 0.5602,
      "step": 9416
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6198569338249735,
      "learning_rate": 1.512188508350792e-06,
      "loss": 0.2139,
      "step": 9417
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2829441028382433,
      "learning_rate": 1.5112603668916752e-06,
      "loss": 0.1969,
      "step": 9418
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.459096298611839,
      "learning_rate": 1.5103324596422209e-06,
      "loss": 0.1471,
      "step": 9419
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3704958231998041,
      "learning_rate": 1.5094047866647194e-06,
      "loss": 0.1648,
      "step": 9420
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2939810664602074,
      "learning_rate": 1.5084773480214543e-06,
      "loss": 0.1577,
      "step": 9421
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1733310321007817,
      "learning_rate": 1.507550143774681e-06,
      "loss": 0.1512,
      "step": 9422
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3289051790554052,
      "learning_rate": 1.5066231739866494e-06,
      "loss": 0.1905,
      "step": 9423
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4622796242195417,
      "learning_rate": 1.5056964387195899e-06,
      "loss": 0.1979,
      "step": 9424
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5330226359248196,
      "learning_rate": 1.5047699380357134e-06,
      "loss": 0.1564,
      "step": 9425
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.317815827195486,
      "learning_rate": 1.503843671997222e-06,
      "loss": 0.146,
      "step": 9426
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.187941949202876,
      "learning_rate": 1.5029176406662982e-06,
      "loss": 0.1511,
      "step": 9427
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6317481333486337,
      "learning_rate": 1.5019918441051084e-06,
      "loss": 0.1874,
      "step": 9428
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4309893462426606,
      "learning_rate": 1.5010662823758037e-06,
      "loss": 0.1736,
      "step": 9429
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.27630107118488,
      "learning_rate": 1.5001409555405238e-06,
      "loss": 0.1742,
      "step": 9430
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4565101545326837,
      "learning_rate": 1.4992158636613824e-06,
      "loss": 0.2006,
      "step": 9431
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.442116411631485,
      "learning_rate": 1.4982910068004886e-06,
      "loss": 0.1847,
      "step": 9432
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3457790366624383,
      "learning_rate": 1.497366385019931e-06,
      "loss": 0.1788,
      "step": 9433
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.234378309465253,
      "learning_rate": 1.4964419983817773e-06,
      "loss": 0.4991,
      "step": 9434
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.383115835112116,
      "learning_rate": 1.4955178469480891e-06,
      "loss": 0.1761,
      "step": 9435
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5689932363071797,
      "learning_rate": 1.4945939307809066e-06,
      "loss": 0.1964,
      "step": 9436
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2459260633590892,
      "learning_rate": 1.4936702499422534e-06,
      "loss": 0.1229,
      "step": 9437
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3299548849751863,
      "learning_rate": 1.4927468044941407e-06,
      "loss": 0.1777,
      "step": 9438
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.382942554996044,
      "learning_rate": 1.4918235944985616e-06,
      "loss": 0.2114,
      "step": 9439
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.203411238077802,
      "learning_rate": 1.4909006200174924e-06,
      "loss": 0.5328,
      "step": 9440
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3732862362328369,
      "learning_rate": 1.4899778811128996e-06,
      "loss": 0.1793,
      "step": 9441
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.454908155860332,
      "learning_rate": 1.4890553778467253e-06,
      "loss": 0.1627,
      "step": 9442
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2432888231024148,
      "learning_rate": 1.4881331102808994e-06,
      "loss": 0.1768,
      "step": 9443
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2234125004691994,
      "learning_rate": 1.487211078477342e-06,
      "loss": 0.1503,
      "step": 9444
    },
    {
      "epoch": 0.76,
      "grad_norm": 13.029984790168864,
      "learning_rate": 1.4862892824979448e-06,
      "loss": 0.6348,
      "step": 9445
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4085605201468197,
      "learning_rate": 1.4853677224045955e-06,
      "loss": 0.1938,
      "step": 9446
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3470430131286248,
      "learning_rate": 1.48444639825916e-06,
      "loss": 0.179,
      "step": 9447
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4529842541068745,
      "learning_rate": 1.4835253101234891e-06,
      "loss": 0.2109,
      "step": 9448
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3601666403547403,
      "learning_rate": 1.4826044580594197e-06,
      "loss": 0.2092,
      "step": 9449
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2523536514003595,
      "learning_rate": 1.4816838421287693e-06,
      "loss": 0.1718,
      "step": 9450
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3290787140527853,
      "learning_rate": 1.4807634623933431e-06,
      "loss": 0.1559,
      "step": 9451
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5961875526630813,
      "learning_rate": 1.4798433189149286e-06,
      "loss": 0.1615,
      "step": 9452
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.320723290932673,
      "learning_rate": 1.4789234117552974e-06,
      "loss": 0.5378,
      "step": 9453
    },
    {
      "epoch": 0.76,
      "grad_norm": 14.269721066179851,
      "learning_rate": 1.478003740976205e-06,
      "loss": 0.3311,
      "step": 9454
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.960558457189034,
      "learning_rate": 1.4770843066393954e-06,
      "loss": 0.5803,
      "step": 9455
    },
    {
      "epoch": 0.76,
      "grad_norm": 23.920351685010257,
      "learning_rate": 1.4761651088065871e-06,
      "loss": 0.5061,
      "step": 9456
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2857335728530066,
      "learning_rate": 1.4752461475394936e-06,
      "loss": 0.1549,
      "step": 9457
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2179194459186946,
      "learning_rate": 1.4743274228998073e-06,
      "loss": 0.156,
      "step": 9458
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4717195560907594,
      "learning_rate": 1.4734089349492003e-06,
      "loss": 0.2067,
      "step": 9459
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1810822913162073,
      "learning_rate": 1.4724906837493386e-06,
      "loss": 0.1897,
      "step": 9460
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3460237229914176,
      "learning_rate": 1.4715726693618648e-06,
      "loss": 0.1867,
      "step": 9461
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6483617258778553,
      "learning_rate": 1.4706548918484086e-06,
      "loss": 0.2299,
      "step": 9462
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.467713366860506,
      "learning_rate": 1.4697373512705825e-06,
      "loss": 0.1827,
      "step": 9463
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5593949516406573,
      "learning_rate": 1.468820047689985e-06,
      "loss": 0.1881,
      "step": 9464
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2559534204729685,
      "learning_rate": 1.467902981168195e-06,
      "loss": 0.1515,
      "step": 9465
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6295212064514297,
      "learning_rate": 1.4669861517667827e-06,
      "loss": 0.2045,
      "step": 9466
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3799786900028381,
      "learning_rate": 1.466069559547293e-06,
      "loss": 0.1896,
      "step": 9467
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3109390977073188,
      "learning_rate": 1.4651532045712596e-06,
      "loss": 0.1905,
      "step": 9468
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.189816932410113,
      "learning_rate": 1.4642370869002048e-06,
      "loss": 0.128,
      "step": 9469
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.294885055021585,
      "learning_rate": 1.4633212065956248e-06,
      "loss": 0.1799,
      "step": 9470
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.414707058226305,
      "learning_rate": 1.4624055637190087e-06,
      "loss": 0.1865,
      "step": 9471
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3791342531933861,
      "learning_rate": 1.4614901583318258e-06,
      "loss": 0.1349,
      "step": 9472
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3363566031400298,
      "learning_rate": 1.4605749904955301e-06,
      "loss": 0.1665,
      "step": 9473
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.669581603404619,
      "learning_rate": 1.4596600602715576e-06,
      "loss": 0.5442,
      "step": 9474
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.301003993697531,
      "learning_rate": 1.4587453677213348e-06,
      "loss": 0.1716,
      "step": 9475
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.394557039397332,
      "learning_rate": 1.457830912906264e-06,
      "loss": 0.1912,
      "step": 9476
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.424606987188368,
      "learning_rate": 1.4569166958877344e-06,
      "loss": 0.1671,
      "step": 9477
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6733430037287949,
      "learning_rate": 1.4560027167271257e-06,
      "loss": 0.2028,
      "step": 9478
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.324640204933129,
      "learning_rate": 1.4550889754857895e-06,
      "loss": 0.1393,
      "step": 9479
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3455280875323246,
      "learning_rate": 1.4541754722250716e-06,
      "loss": 0.153,
      "step": 9480
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.665598052227999,
      "learning_rate": 1.4532622070062985e-06,
      "loss": 0.4305,
      "step": 9481
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4761610330466073,
      "learning_rate": 1.4523491798907797e-06,
      "loss": 0.1839,
      "step": 9482
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4106730429956367,
      "learning_rate": 1.4514363909398087e-06,
      "loss": 0.2174,
      "step": 9483
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3602914294857062,
      "learning_rate": 1.4505238402146655e-06,
      "loss": 0.1709,
      "step": 9484
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3280141635271685,
      "learning_rate": 1.4496115277766105e-06,
      "loss": 0.1112,
      "step": 9485
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2936525507891432,
      "learning_rate": 1.4486994536868914e-06,
      "loss": 0.1666,
      "step": 9486
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4751379848237982,
      "learning_rate": 1.447787618006738e-06,
      "loss": 0.1817,
      "step": 9487
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3204226270972617,
      "learning_rate": 1.4468760207973626e-06,
      "loss": 0.1759,
      "step": 9488
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.490590648041613,
      "learning_rate": 1.4459646621199686e-06,
      "loss": 0.1698,
      "step": 9489
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4824303748700929,
      "learning_rate": 1.4450535420357325e-06,
      "loss": 0.1814,
      "step": 9490
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.325868276688208,
      "learning_rate": 1.4441426606058245e-06,
      "loss": 0.1578,
      "step": 9491
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4634316077127678,
      "learning_rate": 1.4432320178913933e-06,
      "loss": 0.165,
      "step": 9492
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.964189516750006,
      "learning_rate": 1.4423216139535735e-06,
      "loss": 0.6755,
      "step": 9493
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3944858190029523,
      "learning_rate": 1.441411448853483e-06,
      "loss": 0.1877,
      "step": 9494
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.650562732169268,
      "learning_rate": 1.440501522652224e-06,
      "loss": 0.6248,
      "step": 9495
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5755838456817048,
      "learning_rate": 1.4395918354108828e-06,
      "loss": 0.1954,
      "step": 9496
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4080230874275284,
      "learning_rate": 1.4386823871905293e-06,
      "loss": 0.1699,
      "step": 9497
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2773079929378146,
      "learning_rate": 1.4377731780522176e-06,
      "loss": 0.1933,
      "step": 9498
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.342458829294653,
      "learning_rate": 1.4368642080569839e-06,
      "loss": 0.1717,
      "step": 9499
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.469860056379433,
      "learning_rate": 1.4359554772658551e-06,
      "loss": 0.1864,
      "step": 9500
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3246879485208942,
      "learning_rate": 1.4350469857398302e-06,
      "loss": 0.1395,
      "step": 9501
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2934110728182642,
      "learning_rate": 1.4341387335399043e-06,
      "loss": 0.195,
      "step": 9502
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3401304518663661,
      "learning_rate": 1.4332307207270507e-06,
      "loss": 0.2016,
      "step": 9503
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1972405696058863,
      "learning_rate": 1.432322947362223e-06,
      "loss": 0.1792,
      "step": 9504
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3033608810952786,
      "learning_rate": 1.4314154135063668e-06,
      "loss": 0.1839,
      "step": 9505
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3170656636689433,
      "learning_rate": 1.4305081192204063e-06,
      "loss": 0.1501,
      "step": 9506
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.331578605320142,
      "learning_rate": 1.4296010645652508e-06,
      "loss": 0.5321,
      "step": 9507
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.233407614184908,
      "learning_rate": 1.4286942496017936e-06,
      "loss": 0.1604,
      "step": 9508
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.241474233955057,
      "learning_rate": 1.4277876743909126e-06,
      "loss": 0.6613,
      "step": 9509
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2304216286611123,
      "learning_rate": 1.426881338993466e-06,
      "loss": 0.1756,
      "step": 9510
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.644379302117223,
      "learning_rate": 1.425975243470305e-06,
      "loss": 0.4953,
      "step": 9511
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.738090691030682,
      "learning_rate": 1.425069387882253e-06,
      "loss": 0.4325,
      "step": 9512
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3724100130778067,
      "learning_rate": 1.4241637722901231e-06,
      "loss": 0.1504,
      "step": 9513
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2401692803813313,
      "learning_rate": 1.4232583967547153e-06,
      "loss": 0.1648,
      "step": 9514
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4101990466290486,
      "learning_rate": 1.422353261336808e-06,
      "loss": 0.1616,
      "step": 9515
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.490879904374797,
      "learning_rate": 1.4214483660971663e-06,
      "loss": 0.532,
      "step": 9516
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1882894761098333,
      "learning_rate": 1.420543711096538e-06,
      "loss": 0.1763,
      "step": 9517
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3590309846896866,
      "learning_rate": 1.4196392963956562e-06,
      "loss": 0.1633,
      "step": 9518
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6747899669887836,
      "learning_rate": 1.4187351220552365e-06,
      "loss": 0.2204,
      "step": 9519
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3801906646257749,
      "learning_rate": 1.4178311881359785e-06,
      "loss": 0.1583,
      "step": 9520
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3418794197945618,
      "learning_rate": 1.4169274946985666e-06,
      "loss": 0.1798,
      "step": 9521
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.324226482857568,
      "learning_rate": 1.416024041803667e-06,
      "loss": 0.1677,
      "step": 9522
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.702841458168953,
      "learning_rate": 1.4151208295119351e-06,
      "loss": 0.5708,
      "step": 9523
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4673549968322426,
      "learning_rate": 1.4142178578840005e-06,
      "loss": 0.1721,
      "step": 9524
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4989579406884164,
      "learning_rate": 1.4133151269804873e-06,
      "loss": 0.1627,
      "step": 9525
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1826466321446618,
      "learning_rate": 1.4124126368619972e-06,
      "loss": 0.1834,
      "step": 9526
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.435829386157774,
      "learning_rate": 1.4115103875891161e-06,
      "loss": 0.1607,
      "step": 9527
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2813394459905327,
      "learning_rate": 1.4106083792224161e-06,
      "loss": 0.1768,
      "step": 9528
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3482329560110613,
      "learning_rate": 1.4097066118224505e-06,
      "loss": 0.2215,
      "step": 9529
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4349688413644328,
      "learning_rate": 1.4088050854497587e-06,
      "loss": 0.1605,
      "step": 9530
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3434978840786715,
      "learning_rate": 1.407903800164862e-06,
      "loss": 0.1776,
      "step": 9531
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.16202447088763,
      "learning_rate": 1.4070027560282668e-06,
      "loss": 0.4032,
      "step": 9532
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1604662711088827,
      "learning_rate": 1.4061019531004617e-06,
      "loss": 0.1627,
      "step": 9533
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.388689996160774,
      "learning_rate": 1.4052013914419244e-06,
      "loss": 0.1911,
      "step": 9534
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.179192419260396,
      "learning_rate": 1.404301071113106e-06,
      "loss": 0.1339,
      "step": 9535
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.867951917332689,
      "learning_rate": 1.4034009921744534e-06,
      "loss": 0.4695,
      "step": 9536
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.654643297565091,
      "learning_rate": 1.402501154686391e-06,
      "loss": 0.2051,
      "step": 9537
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.967300197272294,
      "learning_rate": 1.4016015587093223e-06,
      "loss": 0.7145,
      "step": 9538
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.30620331114358,
      "learning_rate": 1.4007022043036462e-06,
      "loss": 0.675,
      "step": 9539
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3238881189565663,
      "learning_rate": 1.3998030915297357e-06,
      "loss": 0.1653,
      "step": 9540
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3828199078702719,
      "learning_rate": 1.3989042204479524e-06,
      "loss": 0.1395,
      "step": 9541
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5020418118996313,
      "learning_rate": 1.3980055911186397e-06,
      "loss": 0.1489,
      "step": 9542
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2890011781073085,
      "learning_rate": 1.397107203602125e-06,
      "loss": 0.1746,
      "step": 9543
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4484678371927784,
      "learning_rate": 1.3962090579587185e-06,
      "loss": 0.1761,
      "step": 9544
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.354060512818933,
      "learning_rate": 1.3953111542487202e-06,
      "loss": 0.1693,
      "step": 9545
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.504551825013934,
      "learning_rate": 1.3944134925324042e-06,
      "loss": 0.1499,
      "step": 9546
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2296212074264161,
      "learning_rate": 1.393516072870033e-06,
      "loss": 0.1563,
      "step": 9547
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.796733755882371,
      "learning_rate": 1.3926188953218588e-06,
      "loss": 0.4771,
      "step": 9548
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3670592063763816,
      "learning_rate": 1.3917219599481052e-06,
      "loss": 0.1639,
      "step": 9549
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.292539834765507,
      "learning_rate": 1.39082526680899e-06,
      "loss": 0.1626,
      "step": 9550
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2594938858716314,
      "learning_rate": 1.38992881596471e-06,
      "loss": 0.149,
      "step": 9551
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.544498350106149,
      "learning_rate": 1.3890326074754474e-06,
      "loss": 0.1728,
      "step": 9552
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.330752902030664,
      "learning_rate": 1.3881366414013642e-06,
      "loss": 0.1527,
      "step": 9553
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.435867482970378,
      "learning_rate": 1.3872409178026153e-06,
      "loss": 0.1653,
      "step": 9554
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.392670813415072,
      "learning_rate": 1.386345436739328e-06,
      "loss": 0.1826,
      "step": 9555
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3935418335271457,
      "learning_rate": 1.3854501982716196e-06,
      "loss": 0.1946,
      "step": 9556
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.949126955465833,
      "learning_rate": 1.3845552024595937e-06,
      "loss": 0.6535,
      "step": 9557
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.635963701515663,
      "learning_rate": 1.3836604493633282e-06,
      "loss": 0.2358,
      "step": 9558
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4534131931170744,
      "learning_rate": 1.3827659390428954e-06,
      "loss": 0.2097,
      "step": 9559
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.236820020529011,
      "learning_rate": 1.3818716715583452e-06,
      "loss": 0.1728,
      "step": 9560
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.599596893025295,
      "learning_rate": 1.3809776469697117e-06,
      "loss": 0.6319,
      "step": 9561
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4500552908205495,
      "learning_rate": 1.380083865337014e-06,
      "loss": 0.1852,
      "step": 9562
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4283826583397312,
      "learning_rate": 1.379190326720255e-06,
      "loss": 0.1835,
      "step": 9563
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.296572454013597,
      "learning_rate": 1.378297031179419e-06,
      "loss": 0.1424,
      "step": 9564
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5683992555739819,
      "learning_rate": 1.3774039787744776e-06,
      "loss": 0.1953,
      "step": 9565
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.305474895223133,
      "learning_rate": 1.3765111695653827e-06,
      "loss": 0.1842,
      "step": 9566
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.183909120786941,
      "learning_rate": 1.37561860361207e-06,
      "loss": 0.1534,
      "step": 9567
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5633094592245167,
      "learning_rate": 1.374726280974465e-06,
      "loss": 0.183,
      "step": 9568
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3819098969708152,
      "learning_rate": 1.3738342017124663e-06,
      "loss": 0.1498,
      "step": 9569
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3013755691246405,
      "learning_rate": 1.3729423658859654e-06,
      "loss": 0.1619,
      "step": 9570
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4299547381396105,
      "learning_rate": 1.3720507735548333e-06,
      "loss": 0.1668,
      "step": 9571
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.383860121985393,
      "learning_rate": 1.371159424778925e-06,
      "loss": 0.1745,
      "step": 9572
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5092454863802427,
      "learning_rate": 1.3702683196180793e-06,
      "loss": 0.1581,
      "step": 9573
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.244768949441808,
      "learning_rate": 1.3693774581321195e-06,
      "loss": 0.157,
      "step": 9574
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0406806728956774,
      "learning_rate": 1.368486840380851e-06,
      "loss": 0.1824,
      "step": 9575
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3862038264360002,
      "learning_rate": 1.3675964664240638e-06,
      "loss": 0.1584,
      "step": 9576
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.580882455541988,
      "learning_rate": 1.3667063363215317e-06,
      "loss": 0.6504,
      "step": 9577
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5972889989994792,
      "learning_rate": 1.36581645013301e-06,
      "loss": 0.1904,
      "step": 9578
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3081239334781531,
      "learning_rate": 1.3649268079182443e-06,
      "loss": 0.1762,
      "step": 9579
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.022901863252361,
      "learning_rate": 1.364037409736954e-06,
      "loss": 0.7905,
      "step": 9580
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4132776820530764,
      "learning_rate": 1.363148255648848e-06,
      "loss": 0.1875,
      "step": 9581
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3697041909350678,
      "learning_rate": 1.3622593457136213e-06,
      "loss": 0.2101,
      "step": 9582
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.338149560813029,
      "learning_rate": 1.3613706799909437e-06,
      "loss": 0.1978,
      "step": 9583
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2971856484520121,
      "learning_rate": 1.3604822585404788e-06,
      "loss": 0.1923,
      "step": 9584
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3585148024154676,
      "learning_rate": 1.3595940814218668e-06,
      "loss": 0.1986,
      "step": 9585
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7303474405411523,
      "learning_rate": 1.3587061486947346e-06,
      "loss": 0.1748,
      "step": 9586
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.489074911175372,
      "learning_rate": 1.3578184604186905e-06,
      "loss": 0.5244,
      "step": 9587
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3111334344209886,
      "learning_rate": 1.3569310166533296e-06,
      "loss": 0.1661,
      "step": 9588
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5057802914941214,
      "learning_rate": 1.3560438174582269e-06,
      "loss": 0.211,
      "step": 9589
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0799185766573356,
      "learning_rate": 1.3551568628929434e-06,
      "loss": 0.0989,
      "step": 9590
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.407109771371685,
      "learning_rate": 1.3542701530170232e-06,
      "loss": 0.5072,
      "step": 9591
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2780794928594057,
      "learning_rate": 1.3533836878899926e-06,
      "loss": 0.1735,
      "step": 9592
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2807341333372013,
      "learning_rate": 1.3524974675713665e-06,
      "loss": 0.2034,
      "step": 9593
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5352565879390407,
      "learning_rate": 1.351611492120634e-06,
      "loss": 0.2073,
      "step": 9594
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3406718189532885,
      "learning_rate": 1.3507257615972779e-06,
      "loss": 0.1956,
      "step": 9595
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3768010480101185,
      "learning_rate": 1.3498402760607583e-06,
      "loss": 0.1795,
      "step": 9596
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3753312696708002,
      "learning_rate": 1.34895503557052e-06,
      "loss": 0.2164,
      "step": 9597
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3285053397933568,
      "learning_rate": 1.348070040185993e-06,
      "loss": 0.1853,
      "step": 9598
    },
    {
      "epoch": 0.77,
      "grad_norm": 18.219939954146714,
      "learning_rate": 1.3471852899665888e-06,
      "loss": 0.4674,
      "step": 9599
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5750489160314116,
      "learning_rate": 1.3463007849717035e-06,
      "loss": 0.1866,
      "step": 9600
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.293972076259141,
      "learning_rate": 1.3454165252607155e-06,
      "loss": 0.1728,
      "step": 9601
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.326216574248781,
      "learning_rate": 1.3445325108929925e-06,
      "loss": 0.1854,
      "step": 9602
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2888354948646372,
      "learning_rate": 1.3436487419278738e-06,
      "loss": 0.1601,
      "step": 9603
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.221012335123318,
      "learning_rate": 1.3427652184246952e-06,
      "loss": 0.1541,
      "step": 9604
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2524904239281136,
      "learning_rate": 1.341881940442769e-06,
      "loss": 0.1479,
      "step": 9605
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4512552393785314,
      "learning_rate": 1.340998908041391e-06,
      "loss": 0.1569,
      "step": 9606
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3239944720160173,
      "learning_rate": 1.340116121279843e-06,
      "loss": 0.1569,
      "step": 9607
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3476387513839778,
      "learning_rate": 1.339233580217389e-06,
      "loss": 0.1866,
      "step": 9608
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3471928892707412,
      "learning_rate": 1.3383512849132763e-06,
      "loss": 0.1597,
      "step": 9609
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.291681048339552,
      "learning_rate": 1.337469235426736e-06,
      "loss": 0.6777,
      "step": 9610
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.375597595402478,
      "learning_rate": 1.336587431816983e-06,
      "loss": 0.1757,
      "step": 9611
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4699522996503478,
      "learning_rate": 1.3357058741432143e-06,
      "loss": 0.1737,
      "step": 9612
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.335584692191746,
      "learning_rate": 1.334824562464615e-06,
      "loss": 0.1521,
      "step": 9613
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4958505138259426,
      "learning_rate": 1.3339434968403453e-06,
      "loss": 0.1985,
      "step": 9614
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3676360208857195,
      "learning_rate": 1.3330626773295579e-06,
      "loss": 0.1691,
      "step": 9615
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3925730442750044,
      "learning_rate": 1.3321821039913846e-06,
      "loss": 0.1521,
      "step": 9616
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6120207165975067,
      "learning_rate": 1.3313017768849368e-06,
      "loss": 0.2055,
      "step": 9617
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.419546355501113,
      "learning_rate": 1.3304216960693184e-06,
      "loss": 0.1777,
      "step": 9618
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.353822303456167,
      "learning_rate": 1.3295418616036093e-06,
      "loss": 0.159,
      "step": 9619
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0496577997395307,
      "learning_rate": 1.3286622735468764e-06,
      "loss": 0.1365,
      "step": 9620
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.538599804232529,
      "learning_rate": 1.3277829319581682e-06,
      "loss": 0.465,
      "step": 9621
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6189600171400274,
      "learning_rate": 1.3269038368965186e-06,
      "loss": 0.2215,
      "step": 9622
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2736620650765598,
      "learning_rate": 1.326024988420942e-06,
      "loss": 0.1729,
      "step": 9623
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3582058906548065,
      "learning_rate": 1.325146386590443e-06,
      "loss": 0.1742,
      "step": 9624
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3683837501519065,
      "learning_rate": 1.3242680314639995e-06,
      "loss": 0.1732,
      "step": 9625
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3809632047167297,
      "learning_rate": 1.323389923100578e-06,
      "loss": 0.1954,
      "step": 9626
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.272012411658825,
      "learning_rate": 1.3225120615591348e-06,
      "loss": 0.145,
      "step": 9627
    },
    {
      "epoch": 0.77,
      "grad_norm": 15.67671607290729,
      "learning_rate": 1.3216344468985952e-06,
      "loss": 0.6937,
      "step": 9628
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.113232189526291,
      "learning_rate": 1.3207570791778824e-06,
      "loss": 0.4933,
      "step": 9629
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.189196847889087,
      "learning_rate": 1.319879958455894e-06,
      "loss": 0.1366,
      "step": 9630
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3498062707465581,
      "learning_rate": 1.3190030847915148e-06,
      "loss": 0.1858,
      "step": 9631
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4802678123197617,
      "learning_rate": 1.3181264582436115e-06,
      "loss": 0.2463,
      "step": 9632
    },
    {
      "epoch": 0.77,
      "grad_norm": 19.39872033281013,
      "learning_rate": 1.3172500788710347e-06,
      "loss": 0.6008,
      "step": 9633
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5084644539602123,
      "learning_rate": 1.316373946732618e-06,
      "loss": 0.2255,
      "step": 9634
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3364142835123023,
      "learning_rate": 1.3154980618871793e-06,
      "loss": 0.175,
      "step": 9635
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.435541684885667,
      "learning_rate": 1.3146224243935195e-06,
      "loss": 0.5252,
      "step": 9636
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.336266437398066,
      "learning_rate": 1.3137470343104208e-06,
      "loss": 0.1925,
      "step": 9637
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3194209815748466,
      "learning_rate": 1.312871891696656e-06,
      "loss": 0.1894,
      "step": 9638
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3433903912664904,
      "learning_rate": 1.311996996610969e-06,
      "loss": 0.1727,
      "step": 9639
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3381210493808817,
      "learning_rate": 1.3111223491121e-06,
      "loss": 0.1484,
      "step": 9640
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4304298538470888,
      "learning_rate": 1.3102479492587662e-06,
      "loss": 0.1599,
      "step": 9641
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3436631237523706,
      "learning_rate": 1.3093737971096637e-06,
      "loss": 0.1715,
      "step": 9642
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3384039013836901,
      "learning_rate": 1.3084998927234816e-06,
      "loss": 0.1922,
      "step": 9643
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7619959023130334,
      "learning_rate": 1.3076262361588871e-06,
      "loss": 0.2137,
      "step": 9644
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4271910533463508,
      "learning_rate": 1.30675282747453e-06,
      "loss": 0.1542,
      "step": 9645
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7225465626074374,
      "learning_rate": 1.3058796667290446e-06,
      "loss": 0.1976,
      "step": 9646
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1662364861741215,
      "learning_rate": 1.3050067539810529e-06,
      "loss": 0.1558,
      "step": 9647
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3803226612548904,
      "learning_rate": 1.3041340892891508e-06,
      "loss": 0.1955,
      "step": 9648
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4167938145683001,
      "learning_rate": 1.3032616727119263e-06,
      "loss": 0.1851,
      "step": 9649
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3559721378665677,
      "learning_rate": 1.3023895043079476e-06,
      "loss": 0.1606,
      "step": 9650
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4106247093248743,
      "learning_rate": 1.3015175841357618e-06,
      "loss": 0.1945,
      "step": 9651
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3915104658500859,
      "learning_rate": 1.3006459122539083e-06,
      "loss": 0.1773,
      "step": 9652
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3949293004165577,
      "learning_rate": 1.2997744887209034e-06,
      "loss": 0.1424,
      "step": 9653
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.498613016747376,
      "learning_rate": 1.2989033135952484e-06,
      "loss": 0.1665,
      "step": 9654
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3817191561852455,
      "learning_rate": 1.2980323869354277e-06,
      "loss": 0.1787,
      "step": 9655
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.33106650852278,
      "learning_rate": 1.297161708799909e-06,
      "loss": 0.1853,
      "step": 9656
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.326393210649177,
      "learning_rate": 1.2962912792471426e-06,
      "loss": 0.1527,
      "step": 9657
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2609976685904807,
      "learning_rate": 1.295421098335568e-06,
      "loss": 0.1604,
      "step": 9658
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3997698941181176,
      "learning_rate": 1.2945511661235977e-06,
      "loss": 0.1663,
      "step": 9659
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2754168893415605,
      "learning_rate": 1.2936814826696326e-06,
      "loss": 0.1487,
      "step": 9660
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2369092001936701,
      "learning_rate": 1.2928120480320627e-06,
      "loss": 0.1597,
      "step": 9661
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.252003589635802,
      "learning_rate": 1.2919428622692499e-06,
      "loss": 0.1559,
      "step": 9662
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.165678739915954,
      "learning_rate": 1.2910739254395482e-06,
      "loss": 0.5545,
      "step": 9663
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4924209069746268,
      "learning_rate": 1.290205237601292e-06,
      "loss": 0.2011,
      "step": 9664
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.403136813357024,
      "learning_rate": 1.2893367988127986e-06,
      "loss": 0.177,
      "step": 9665
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3795896465290092,
      "learning_rate": 1.2884686091323684e-06,
      "loss": 0.1831,
      "step": 9666
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2436644503950356,
      "learning_rate": 1.2876006686182863e-06,
      "loss": 0.1508,
      "step": 9667
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3099909187940835,
      "learning_rate": 1.286732977328819e-06,
      "loss": 0.1717,
      "step": 9668
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.293489586165324,
      "learning_rate": 1.2858655353222182e-06,
      "loss": 0.4487,
      "step": 9669
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4364614103042515,
      "learning_rate": 1.284998342656717e-06,
      "loss": 0.1828,
      "step": 9670
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3990358228027717,
      "learning_rate": 1.284131399390532e-06,
      "loss": 0.1775,
      "step": 9671
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.48244109101915,
      "learning_rate": 1.283264705581867e-06,
      "loss": 0.2129,
      "step": 9672
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2635632079279266,
      "learning_rate": 1.2823982612889013e-06,
      "loss": 0.1806,
      "step": 9673
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5154657192724275,
      "learning_rate": 1.2815320665698056e-06,
      "loss": 0.2039,
      "step": 9674
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.432365553539627,
      "learning_rate": 1.2806661214827286e-06,
      "loss": 0.1984,
      "step": 9675
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4054911432786705,
      "learning_rate": 1.2798004260858039e-06,
      "loss": 0.1841,
      "step": 9676
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4472745361627866,
      "learning_rate": 1.2789349804371486e-06,
      "loss": 0.2116,
      "step": 9677
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3531110065039962,
      "learning_rate": 1.2780697845948625e-06,
      "loss": 0.1603,
      "step": 9678
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3939290198745053,
      "learning_rate": 1.2772048386170284e-06,
      "loss": 0.1897,
      "step": 9679
    },
    {
      "epoch": 0.77,
      "grad_norm": 17.94111117270653,
      "learning_rate": 1.2763401425617134e-06,
      "loss": 0.7047,
      "step": 9680
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.209768005446248,
      "learning_rate": 1.275475696486967e-06,
      "loss": 0.164,
      "step": 9681
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.491920262071602,
      "learning_rate": 1.2746115004508198e-06,
      "loss": 0.4811,
      "step": 9682
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3630257227242328,
      "learning_rate": 1.273747554511292e-06,
      "loss": 0.1761,
      "step": 9683
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.277766373987406,
      "learning_rate": 1.272883858726381e-06,
      "loss": 0.1563,
      "step": 9684
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1902572638906392,
      "learning_rate": 1.2720204131540693e-06,
      "loss": 0.1559,
      "step": 9685
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.342748353602773,
      "learning_rate": 1.2711572178523224e-06,
      "loss": 0.188,
      "step": 9686
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2036780390289268,
      "learning_rate": 1.2702942728790897e-06,
      "loss": 0.1509,
      "step": 9687
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3068500263494836,
      "learning_rate": 1.2694315782923029e-06,
      "loss": 0.178,
      "step": 9688
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2930488902783386,
      "learning_rate": 1.2685691341498775e-06,
      "loss": 0.1767,
      "step": 9689
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.143972226164976,
      "learning_rate": 1.2677069405097115e-06,
      "loss": 0.1349,
      "step": 9690
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2186527908945055,
      "learning_rate": 1.266844997429686e-06,
      "loss": 0.1859,
      "step": 9691
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4151349414493946,
      "learning_rate": 1.2659833049676695e-06,
      "loss": 0.1692,
      "step": 9692
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.381076332409303,
      "learning_rate": 1.2651218631815043e-06,
      "loss": 0.1805,
      "step": 9693
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.406152524967701,
      "learning_rate": 1.2642606721290262e-06,
      "loss": 0.1578,
      "step": 9694
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2651239677311,
      "learning_rate": 1.2633997318680496e-06,
      "loss": 0.1383,
      "step": 9695
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3895068288622148,
      "learning_rate": 1.2625390424563677e-06,
      "loss": 0.1769,
      "step": 9696
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3513493743654383,
      "learning_rate": 1.2616786039517653e-06,
      "loss": 0.1734,
      "step": 9697
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2204624576270215,
      "learning_rate": 1.2608184164120046e-06,
      "loss": 0.1546,
      "step": 9698
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.462617742241187,
      "learning_rate": 1.2599584798948334e-06,
      "loss": 0.2129,
      "step": 9699
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2789824326871861,
      "learning_rate": 1.2590987944579808e-06,
      "loss": 0.1517,
      "step": 9700
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3958115021693471,
      "learning_rate": 1.2582393601591608e-06,
      "loss": 0.1753,
      "step": 9701
    },
    {
      "epoch": 0.78,
      "grad_norm": 10.14100832097093,
      "learning_rate": 1.2573801770560696e-06,
      "loss": 0.4916,
      "step": 9702
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.38204250596385,
      "learning_rate": 1.2565212452063868e-06,
      "loss": 0.1638,
      "step": 9703
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1958446392121997,
      "learning_rate": 1.255662564667775e-06,
      "loss": 0.1915,
      "step": 9704
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3256701764725474,
      "learning_rate": 1.254804135497879e-06,
      "loss": 0.1576,
      "step": 9705
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.326200893413067,
      "learning_rate": 1.2539459577543317e-06,
      "loss": 0.6434,
      "step": 9706
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3307907023325738,
      "learning_rate": 1.253088031494739e-06,
      "loss": 0.1665,
      "step": 9707
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1027352884085972,
      "learning_rate": 1.2522303567767013e-06,
      "loss": 0.1602,
      "step": 9708
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.089395155964005,
      "learning_rate": 1.2513729336577952e-06,
      "loss": 0.5007,
      "step": 9709
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3625978555690292,
      "learning_rate": 1.2505157621955815e-06,
      "loss": 0.1835,
      "step": 9710
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.794887340986408,
      "learning_rate": 1.2496588424476053e-06,
      "loss": 0.4645,
      "step": 9711
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3745107318293512,
      "learning_rate": 1.248802174471394e-06,
      "loss": 0.1886,
      "step": 9712
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4450027980655469,
      "learning_rate": 1.2479457583244586e-06,
      "loss": 0.1705,
      "step": 9713
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2559512424854085,
      "learning_rate": 1.247089594064293e-06,
      "loss": 0.1635,
      "step": 9714
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2897402455614253,
      "learning_rate": 1.2462336817483734e-06,
      "loss": 0.1601,
      "step": 9715
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.159294875052488,
      "learning_rate": 1.2453780214341588e-06,
      "loss": 0.6042,
      "step": 9716
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1571272965770274,
      "learning_rate": 1.2445226131790965e-06,
      "loss": 0.1301,
      "step": 9717
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2995103041134548,
      "learning_rate": 1.243667457040607e-06,
      "loss": 0.1562,
      "step": 9718
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4025446666148669,
      "learning_rate": 1.2428125530761043e-06,
      "loss": 0.1692,
      "step": 9719
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2906886628993186,
      "learning_rate": 1.2419579013429795e-06,
      "loss": 0.163,
      "step": 9720
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.279015857952301,
      "learning_rate": 1.2411035018986045e-06,
      "loss": 0.1545,
      "step": 9721
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3208093120753732,
      "learning_rate": 1.2402493548003418e-06,
      "loss": 0.1467,
      "step": 9722
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3797915765546418,
      "learning_rate": 1.2393954601055318e-06,
      "loss": 0.1683,
      "step": 9723
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.334191449613719,
      "learning_rate": 1.2385418178714987e-06,
      "loss": 0.181,
      "step": 9724
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1788189523618704,
      "learning_rate": 1.2376884281555485e-06,
      "loss": 0.1477,
      "step": 9725
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3170985118720009,
      "learning_rate": 1.2368352910149761e-06,
      "loss": 0.1309,
      "step": 9726
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.516135494720798,
      "learning_rate": 1.2359824065070496e-06,
      "loss": 0.5503,
      "step": 9727
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.374627311459824,
      "learning_rate": 1.2351297746890306e-06,
      "loss": 0.1896,
      "step": 9728
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2960278679368211,
      "learning_rate": 1.2342773956181585e-06,
      "loss": 0.1693,
      "step": 9729
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4729670515988442,
      "learning_rate": 1.2334252693516512e-06,
      "loss": 0.1745,
      "step": 9730
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.535430541928961,
      "learning_rate": 1.2325733959467196e-06,
      "loss": 0.2111,
      "step": 9731
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.718949537900114,
      "learning_rate": 1.2317217754605503e-06,
      "loss": 0.5044,
      "step": 9732
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3396746144986362,
      "learning_rate": 1.2308704079503164e-06,
      "loss": 0.1516,
      "step": 9733
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1854988857431092,
      "learning_rate": 1.2300192934731713e-06,
      "loss": 0.1581,
      "step": 9734
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2666466863429922,
      "learning_rate": 1.229168432086254e-06,
      "loss": 0.1492,
      "step": 9735
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4410346665142189,
      "learning_rate": 1.2283178238466837e-06,
      "loss": 0.1903,
      "step": 9736
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3333431762939871,
      "learning_rate": 1.2274674688115684e-06,
      "loss": 0.1834,
      "step": 9737
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5080363678671707,
      "learning_rate": 1.2266173670379905e-06,
      "loss": 0.1497,
      "step": 9738
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4000910908927513,
      "learning_rate": 1.2257675185830203e-06,
      "loss": 0.1783,
      "step": 9739
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1189839561871457,
      "learning_rate": 1.224917923503715e-06,
      "loss": 0.1413,
      "step": 9740
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2571164045497523,
      "learning_rate": 1.2240685818571047e-06,
      "loss": 0.1747,
      "step": 9741
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.82598303779388,
      "learning_rate": 1.2232194937002124e-06,
      "loss": 0.5132,
      "step": 9742
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4111242016095809,
      "learning_rate": 1.222370659090039e-06,
      "loss": 0.1898,
      "step": 9743
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5109913798176613,
      "learning_rate": 1.221522078083569e-06,
      "loss": 0.1949,
      "step": 9744
    },
    {
      "epoch": 0.78,
      "grad_norm": 10.931788468827273,
      "learning_rate": 1.2206737507377698e-06,
      "loss": 0.5002,
      "step": 9745
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2957869521268661,
      "learning_rate": 1.219825677109593e-06,
      "loss": 0.1992,
      "step": 9746
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4035660794505376,
      "learning_rate": 1.2189778572559718e-06,
      "loss": 0.1778,
      "step": 9747
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3973899103952812,
      "learning_rate": 1.2181302912338234e-06,
      "loss": 0.1895,
      "step": 9748
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3981419509518755,
      "learning_rate": 1.2172829791000468e-06,
      "loss": 0.1673,
      "step": 9749
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1885338700672288,
      "learning_rate": 1.2164359209115235e-06,
      "loss": 0.1487,
      "step": 9750
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3879392677242115,
      "learning_rate": 1.215589116725124e-06,
      "loss": 0.1629,
      "step": 9751
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.210056448384112,
      "learning_rate": 1.2147425665976898e-06,
      "loss": 0.1561,
      "step": 9752
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3404829400958798,
      "learning_rate": 1.2138962705860574e-06,
      "loss": 0.1737,
      "step": 9753
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3718911425512998,
      "learning_rate": 1.2130502287470402e-06,
      "loss": 0.1721,
      "step": 9754
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3170201905848402,
      "learning_rate": 1.212204441137435e-06,
      "loss": 0.1674,
      "step": 9755
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.596004767264219,
      "learning_rate": 1.2113589078140225e-06,
      "loss": 0.1427,
      "step": 9756
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.781898949815977,
      "learning_rate": 1.2105136288335655e-06,
      "loss": 0.1551,
      "step": 9757
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3831569328733846,
      "learning_rate": 1.20966860425281e-06,
      "loss": 0.1462,
      "step": 9758
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.364261201260296,
      "learning_rate": 1.2088238341284858e-06,
      "loss": 0.1862,
      "step": 9759
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6029419981000008,
      "learning_rate": 1.2079793185173045e-06,
      "loss": 0.201,
      "step": 9760
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3130531112834292,
      "learning_rate": 1.2071350574759589e-06,
      "loss": 0.1738,
      "step": 9761
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3129234346522478,
      "learning_rate": 1.2062910510611319e-06,
      "loss": 0.1852,
      "step": 9762
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4665037289324505,
      "learning_rate": 1.2054472993294796e-06,
      "loss": 0.1865,
      "step": 9763
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.125969839204317,
      "learning_rate": 1.204603802337646e-06,
      "loss": 0.1419,
      "step": 9764
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2998347139224293,
      "learning_rate": 1.2037605601422614e-06,
      "loss": 0.1253,
      "step": 9765
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.25177259854596,
      "learning_rate": 1.2029175727999298e-06,
      "loss": 0.4844,
      "step": 9766
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4175279570114816,
      "learning_rate": 1.2020748403672478e-06,
      "loss": 0.1926,
      "step": 9767
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.690616535051294,
      "learning_rate": 1.2012323629007887e-06,
      "loss": 0.1847,
      "step": 9768
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6055183078216466,
      "learning_rate": 1.200390140457111e-06,
      "loss": 0.2277,
      "step": 9769
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3128020818922936,
      "learning_rate": 1.1995481730927538e-06,
      "loss": 0.1735,
      "step": 9770
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3757135722992135,
      "learning_rate": 1.1987064608642463e-06,
      "loss": 0.21,
      "step": 9771
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4778586265268383,
      "learning_rate": 1.1978650038280893e-06,
      "loss": 0.1625,
      "step": 9772
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.500562524060788,
      "learning_rate": 1.1970238020407737e-06,
      "loss": 0.1979,
      "step": 9773
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2042070084911467,
      "learning_rate": 1.1961828555587756e-06,
      "loss": 0.1746,
      "step": 9774
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3520633651449392,
      "learning_rate": 1.1953421644385444e-06,
      "loss": 0.2057,
      "step": 9775
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5652978519860317,
      "learning_rate": 1.194501728736523e-06,
      "loss": 0.1698,
      "step": 9776
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3108565773116279,
      "learning_rate": 1.1936615485091307e-06,
      "loss": 0.1494,
      "step": 9777
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6544316856254455,
      "learning_rate": 1.1928216238127715e-06,
      "loss": 0.2057,
      "step": 9778
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.9969959673169155,
      "learning_rate": 1.191981954703832e-06,
      "loss": 0.5536,
      "step": 9779
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3191856511542102,
      "learning_rate": 1.1911425412386811e-06,
      "loss": 0.1903,
      "step": 9780
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.364585765194263,
      "learning_rate": 1.1903033834736726e-06,
      "loss": 0.1774,
      "step": 9781
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2470652865156833,
      "learning_rate": 1.1894644814651407e-06,
      "loss": 0.1394,
      "step": 9782
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.441906404573823,
      "learning_rate": 1.188625835269403e-06,
      "loss": 0.1959,
      "step": 9783
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.317836894345022,
      "learning_rate": 1.18778744494276e-06,
      "loss": 0.201,
      "step": 9784
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.67963698826465,
      "learning_rate": 1.1869493105414999e-06,
      "loss": 0.5676,
      "step": 9785
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3632883445348702,
      "learning_rate": 1.186111432121882e-06,
      "loss": 0.1735,
      "step": 9786
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.035546708199437,
      "learning_rate": 1.1852738097401605e-06,
      "loss": 0.5192,
      "step": 9787
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.245154129589304,
      "learning_rate": 1.1844364434525667e-06,
      "loss": 0.1492,
      "step": 9788
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.573357832333633,
      "learning_rate": 1.1835993333153156e-06,
      "loss": 0.1968,
      "step": 9789
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.320261224224351,
      "learning_rate": 1.1827624793846037e-06,
      "loss": 0.1914,
      "step": 9790
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3701948222707918,
      "learning_rate": 1.181925881716613e-06,
      "loss": 0.1699,
      "step": 9791
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3611585384831602,
      "learning_rate": 1.1810895403675055e-06,
      "loss": 0.1646,
      "step": 9792
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5001972070458662,
      "learning_rate": 1.1802534553934286e-06,
      "loss": 0.1932,
      "step": 9793
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3276122673765458,
      "learning_rate": 1.1794176268505103e-06,
      "loss": 0.1761,
      "step": 9794
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.309131040558233,
      "learning_rate": 1.1785820547948612e-06,
      "loss": 0.1415,
      "step": 9795
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3833571927434074,
      "learning_rate": 1.1777467392825797e-06,
      "loss": 0.1824,
      "step": 9796
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3789760450518598,
      "learning_rate": 1.1769116803697383e-06,
      "loss": 0.1957,
      "step": 9797
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5158437345390863,
      "learning_rate": 1.1760768781124005e-06,
      "loss": 0.1886,
      "step": 9798
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.005585897188372,
      "learning_rate": 1.1752423325666096e-06,
      "loss": 0.5743,
      "step": 9799
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4842342645797546,
      "learning_rate": 1.1744080437883859e-06,
      "loss": 0.168,
      "step": 9800
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5296894005019765,
      "learning_rate": 1.1735740118337436e-06,
      "loss": 0.1935,
      "step": 9801
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3541812302831,
      "learning_rate": 1.172740236758671e-06,
      "loss": 0.1644,
      "step": 9802
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.440122790678042,
      "learning_rate": 1.1719067186191423e-06,
      "loss": 0.1695,
      "step": 9803
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.583215023448982,
      "learning_rate": 1.1710734574711147e-06,
      "loss": 0.7161,
      "step": 9804
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.367271088658042,
      "learning_rate": 1.1702404533705264e-06,
      "loss": 0.1604,
      "step": 9805
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3359624408024706,
      "learning_rate": 1.1694077063732995e-06,
      "loss": 0.1684,
      "step": 9806
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.46502154979226,
      "learning_rate": 1.168575216535342e-06,
      "loss": 0.1753,
      "step": 9807
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.148531340240961,
      "learning_rate": 1.1677429839125376e-06,
      "loss": 0.6221,
      "step": 9808
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2114163514730203,
      "learning_rate": 1.1669110085607565e-06,
      "loss": 0.1579,
      "step": 9809
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.351153092094994,
      "learning_rate": 1.166079290535856e-06,
      "loss": 0.1849,
      "step": 9810
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2988569914155577,
      "learning_rate": 1.1652478298936664e-06,
      "loss": 0.1487,
      "step": 9811
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.467092991046771,
      "learning_rate": 1.1644166266900098e-06,
      "loss": 0.1576,
      "step": 9812
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4390196816106162,
      "learning_rate": 1.1635856809806866e-06,
      "loss": 0.1637,
      "step": 9813
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.30118363739276,
      "learning_rate": 1.162754992821481e-06,
      "loss": 0.1408,
      "step": 9814
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2726260879024283,
      "learning_rate": 1.1619245622681575e-06,
      "loss": 0.1799,
      "step": 9815
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3693734420967312,
      "learning_rate": 1.1610943893764692e-06,
      "loss": 0.2109,
      "step": 9816
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.38509827532211,
      "learning_rate": 1.160264474202145e-06,
      "loss": 0.1703,
      "step": 9817
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4405518033055595,
      "learning_rate": 1.1594348168008995e-06,
      "loss": 0.1805,
      "step": 9818
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3693944669054656,
      "learning_rate": 1.1586054172284338e-06,
      "loss": 0.1863,
      "step": 9819
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4491888048478812,
      "learning_rate": 1.1577762755404227e-06,
      "loss": 0.1608,
      "step": 9820
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.506491004958339,
      "learning_rate": 1.156947391792533e-06,
      "loss": 0.2121,
      "step": 9821
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2687628177018353,
      "learning_rate": 1.1561187660404083e-06,
      "loss": 0.1733,
      "step": 9822
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.286893964367683,
      "learning_rate": 1.1552903983396779e-06,
      "loss": 0.1729,
      "step": 9823
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.364489297235388,
      "learning_rate": 1.1544622887459522e-06,
      "loss": 0.1923,
      "step": 9824
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4330355666731789,
      "learning_rate": 1.1536344373148245e-06,
      "loss": 0.1677,
      "step": 9825
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6660367364474653,
      "learning_rate": 1.1528068441018708e-06,
      "loss": 0.1853,
      "step": 9826
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.313874155436797,
      "learning_rate": 1.15197950916265e-06,
      "loss": 0.1395,
      "step": 9827
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4798038758361267,
      "learning_rate": 1.151152432552704e-06,
      "loss": 0.1614,
      "step": 9828
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5400662740034599,
      "learning_rate": 1.1503256143275559e-06,
      "loss": 0.1863,
      "step": 9829
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4065515040691345,
      "learning_rate": 1.1494990545427153e-06,
      "loss": 0.2081,
      "step": 9830
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4361973341200527,
      "learning_rate": 1.1486727532536674e-06,
      "loss": 0.1514,
      "step": 9831
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4330070199656062,
      "learning_rate": 1.147846710515888e-06,
      "loss": 0.2242,
      "step": 9832
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.405907560108396,
      "learning_rate": 1.147020926384832e-06,
      "loss": 0.1651,
      "step": 9833
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.414990176268025,
      "learning_rate": 1.146195400915932e-06,
      "loss": 0.6261,
      "step": 9834
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3914045403464677,
      "learning_rate": 1.1453701341646134e-06,
      "loss": 0.1667,
      "step": 9835
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2805940094851456,
      "learning_rate": 1.1445451261862761e-06,
      "loss": 0.1473,
      "step": 9836
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4342410011455013,
      "learning_rate": 1.1437203770363064e-06,
      "loss": 0.2031,
      "step": 9837
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4119699327648056,
      "learning_rate": 1.1428958867700717e-06,
      "loss": 0.1668,
      "step": 9838
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.996793088071932,
      "learning_rate": 1.1420716554429229e-06,
      "loss": 0.5711,
      "step": 9839
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1840580020234677,
      "learning_rate": 1.1412476831101916e-06,
      "loss": 0.1478,
      "step": 9840
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2660520782487996,
      "learning_rate": 1.1404239698271975e-06,
      "loss": 0.1792,
      "step": 9841
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4008899204196315,
      "learning_rate": 1.1396005156492346e-06,
      "loss": 0.1969,
      "step": 9842
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3411630258498726,
      "learning_rate": 1.1387773206315845e-06,
      "loss": 0.1985,
      "step": 9843
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3996694457080312,
      "learning_rate": 1.1379543848295149e-06,
      "loss": 0.1776,
      "step": 9844
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3719295176852393,
      "learning_rate": 1.1371317082982658e-06,
      "loss": 0.1841,
      "step": 9845
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3841330521808748,
      "learning_rate": 1.1363092910930707e-06,
      "loss": 0.1502,
      "step": 9846
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.019650446786424,
      "learning_rate": 1.1354871332691396e-06,
      "loss": 0.5841,
      "step": 9847
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.10280298369289,
      "learning_rate": 1.1346652348816662e-06,
      "loss": 0.467,
      "step": 9848
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.241730860964395,
      "learning_rate": 1.1338435959858268e-06,
      "loss": 0.6076,
      "step": 9849
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.50279421666206,
      "learning_rate": 1.133022216636781e-06,
      "loss": 0.4449,
      "step": 9850
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.535252145004815,
      "learning_rate": 1.1322010968896708e-06,
      "loss": 0.1841,
      "step": 9851
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4591776923716469,
      "learning_rate": 1.131380236799619e-06,
      "loss": 0.2065,
      "step": 9852
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2376492482486345,
      "learning_rate": 1.1305596364217336e-06,
      "loss": 0.1808,
      "step": 9853
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4192122429504153,
      "learning_rate": 1.1297392958111025e-06,
      "loss": 0.1653,
      "step": 9854
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.271300616045784,
      "learning_rate": 1.1289192150228007e-06,
      "loss": 0.152,
      "step": 9855
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4472716395399354,
      "learning_rate": 1.1280993941118805e-06,
      "loss": 0.1686,
      "step": 9856
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2862849271318963,
      "learning_rate": 1.1272798331333789e-06,
      "loss": 0.1412,
      "step": 9857
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.9446450623625875,
      "learning_rate": 1.1264605321423167e-06,
      "loss": 0.5053,
      "step": 9858
    },
    {
      "epoch": 0.79,
      "grad_norm": 15.141549849846838,
      "learning_rate": 1.1256414911936952e-06,
      "loss": 0.5147,
      "step": 9859
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5087631801234787,
      "learning_rate": 1.124822710342499e-06,
      "loss": 0.1856,
      "step": 9860
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.1730253870854,
      "learning_rate": 1.124004189643696e-06,
      "loss": 0.644,
      "step": 9861
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4861562021172667,
      "learning_rate": 1.123185929152235e-06,
      "loss": 0.1987,
      "step": 9862
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6112319318027872,
      "learning_rate": 1.1223679289230478e-06,
      "loss": 0.2063,
      "step": 9863
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1987742143599338,
      "learning_rate": 1.121550189011053e-06,
      "loss": 0.1346,
      "step": 9864
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.40178869813193,
      "learning_rate": 1.1207327094711423e-06,
      "loss": 0.1582,
      "step": 9865
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.533594816996713,
      "learning_rate": 1.1199154903582e-06,
      "loss": 0.5875,
      "step": 9866
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2835925162805852,
      "learning_rate": 1.119098531727087e-06,
      "loss": 0.1587,
      "step": 9867
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2994861647513538,
      "learning_rate": 1.1182818336326485e-06,
      "loss": 0.1718,
      "step": 9868
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.557713741927612,
      "learning_rate": 1.1174653961297116e-06,
      "loss": 0.5139,
      "step": 9869
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3583825787438852,
      "learning_rate": 1.116649219273086e-06,
      "loss": 0.1791,
      "step": 9870
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4391856689933962,
      "learning_rate": 1.1158333031175643e-06,
      "loss": 0.1716,
      "step": 9871
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2922499001840557,
      "learning_rate": 1.115017647717922e-06,
      "loss": 0.1787,
      "step": 9872
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.027893253493273,
      "learning_rate": 1.114202253128916e-06,
      "loss": 0.7926,
      "step": 9873
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2184271608164499,
      "learning_rate": 1.1133871194052843e-06,
      "loss": 0.1335,
      "step": 9874
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4356445500207249,
      "learning_rate": 1.1125722466017547e-06,
      "loss": 0.1895,
      "step": 9875
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4031336807382615,
      "learning_rate": 1.1117576347730253e-06,
      "loss": 0.208,
      "step": 9876
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3037425473228466,
      "learning_rate": 1.1109432839737883e-06,
      "loss": 0.1657,
      "step": 9877
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.51971929251121,
      "learning_rate": 1.1101291942587133e-06,
      "loss": 0.2174,
      "step": 9878
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.373883784449355,
      "learning_rate": 1.109315365682449e-06,
      "loss": 0.1868,
      "step": 9879
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.41286035815426,
      "learning_rate": 1.1085017982996337e-06,
      "loss": 0.1658,
      "step": 9880
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1771025471141001,
      "learning_rate": 1.1076884921648834e-06,
      "loss": 0.1359,
      "step": 9881
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.327708055429888,
      "learning_rate": 1.1068754473327975e-06,
      "loss": 0.1667,
      "step": 9882
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.29914385240458,
      "learning_rate": 1.1060626638579591e-06,
      "loss": 0.1487,
      "step": 9883
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4481201333699947,
      "learning_rate": 1.1052501417949319e-06,
      "loss": 0.1784,
      "step": 9884
    },
    {
      "epoch": 0.79,
      "grad_norm": 20.053654196460826,
      "learning_rate": 1.1044378811982631e-06,
      "loss": 0.52,
      "step": 9885
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6351945014187972,
      "learning_rate": 1.1036258821224833e-06,
      "loss": 0.2039,
      "step": 9886
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.290663199134415,
      "learning_rate": 1.1028141446221024e-06,
      "loss": 0.1377,
      "step": 9887
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.390442926306491,
      "learning_rate": 1.1020026687516155e-06,
      "loss": 0.1798,
      "step": 9888
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3756236616506705,
      "learning_rate": 1.1011914545655023e-06,
      "loss": 0.1757,
      "step": 9889
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4165927348687577,
      "learning_rate": 1.1003805021182169e-06,
      "loss": 0.1818,
      "step": 9890
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.794723577350158,
      "learning_rate": 1.099569811464205e-06,
      "loss": 0.5262,
      "step": 9891
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.787289761416629,
      "learning_rate": 1.09875938265789e-06,
      "loss": 0.4017,
      "step": 9892
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1931108154160246,
      "learning_rate": 1.0979492157536774e-06,
      "loss": 0.1426,
      "step": 9893
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2358440075177703,
      "learning_rate": 1.0971393108059569e-06,
      "loss": 0.1707,
      "step": 9894
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2917390844122292,
      "learning_rate": 1.0963296678691e-06,
      "loss": 0.1548,
      "step": 9895
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5793335636325065,
      "learning_rate": 1.0955202869974597e-06,
      "loss": 0.2102,
      "step": 9896
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3941711111515733,
      "learning_rate": 1.0947111682453714e-06,
      "loss": 0.1696,
      "step": 9897
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5026409551203748,
      "learning_rate": 1.0939023116671577e-06,
      "loss": 0.1793,
      "step": 9898
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2197598431153427,
      "learning_rate": 1.093093717317114e-06,
      "loss": 0.1622,
      "step": 9899
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4246894183172512,
      "learning_rate": 1.092285385249528e-06,
      "loss": 0.1996,
      "step": 9900
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3001198652226,
      "learning_rate": 1.0914773155186641e-06,
      "loss": 0.1906,
      "step": 9901
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7084001017920598,
      "learning_rate": 1.09066950817877e-06,
      "loss": 0.1689,
      "step": 9902
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4909422922896476,
      "learning_rate": 1.0898619632840768e-06,
      "loss": 0.1887,
      "step": 9903
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5946068088880412,
      "learning_rate": 1.0890546808887976e-06,
      "loss": 0.1896,
      "step": 9904
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4047885579053343,
      "learning_rate": 1.088247661047127e-06,
      "loss": 0.1434,
      "step": 9905
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4935789187329487,
      "learning_rate": 1.0874409038132427e-06,
      "loss": 0.1525,
      "step": 9906
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1826451456761589,
      "learning_rate": 1.0866344092413056e-06,
      "loss": 0.1561,
      "step": 9907
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3058903123930754,
      "learning_rate": 1.0858281773854562e-06,
      "loss": 0.1762,
      "step": 9908
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.358816848690121,
      "learning_rate": 1.0850222082998235e-06,
      "loss": 0.1791,
      "step": 9909
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.6416147911387515,
      "learning_rate": 1.0842165020385092e-06,
      "loss": 0.5557,
      "step": 9910
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3240333403954518,
      "learning_rate": 1.0834110586556062e-06,
      "loss": 0.1414,
      "step": 9911
    },
    {
      "epoch": 0.79,
      "grad_norm": 17.731540954490463,
      "learning_rate": 1.082605878205188e-06,
      "loss": 0.5485,
      "step": 9912
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4424079952966151,
      "learning_rate": 1.0818009607413027e-06,
      "loss": 0.1795,
      "step": 9913
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1425857698632826,
      "learning_rate": 1.0809963063179923e-06,
      "loss": 0.1335,
      "step": 9914
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2819325096079859,
      "learning_rate": 1.0801919149892743e-06,
      "loss": 0.1564,
      "step": 9915
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3553206959507982,
      "learning_rate": 1.0793877868091496e-06,
      "loss": 0.1759,
      "step": 9916
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.372575918424012,
      "learning_rate": 1.0785839218316012e-06,
      "loss": 0.1819,
      "step": 9917
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2436419482167722,
      "learning_rate": 1.0777803201105963e-06,
      "loss": 0.1467,
      "step": 9918
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1845850604167574,
      "learning_rate": 1.0769769817000802e-06,
      "loss": 0.1672,
      "step": 9919
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2066982636060177,
      "learning_rate": 1.0761739066539888e-06,
      "loss": 0.1707,
      "step": 9920
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4449870183064657,
      "learning_rate": 1.0753710950262303e-06,
      "loss": 0.2513,
      "step": 9921
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.527841818275548,
      "learning_rate": 1.0745685468707002e-06,
      "loss": 0.1872,
      "step": 9922
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3175448272322143,
      "learning_rate": 1.0737662622412792e-06,
      "loss": 0.1823,
      "step": 9923
    },
    {
      "epoch": 0.79,
      "grad_norm": 16.282928289820504,
      "learning_rate": 1.0729642411918228e-06,
      "loss": 0.6064,
      "step": 9924
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.365857240688093,
      "learning_rate": 1.0721624837761768e-06,
      "loss": 0.634,
      "step": 9925
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3304839967444,
      "learning_rate": 1.0713609900481642e-06,
      "loss": 0.1692,
      "step": 9926
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1987163480870806,
      "learning_rate": 1.0705597600615913e-06,
      "loss": 0.1473,
      "step": 9927
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3411165268945053,
      "learning_rate": 1.0697587938702474e-06,
      "loss": 0.1735,
      "step": 9928
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1573819944727495,
      "learning_rate": 1.0689580915279041e-06,
      "loss": 0.1238,
      "step": 9929
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.328414404905445,
      "learning_rate": 1.0681576530883148e-06,
      "loss": 0.147,
      "step": 9930
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4592088754315973,
      "learning_rate": 1.0673574786052148e-06,
      "loss": 0.154,
      "step": 9931
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3097856303510884,
      "learning_rate": 1.0665575681323232e-06,
      "loss": 0.1732,
      "step": 9932
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1193675688775948,
      "learning_rate": 1.0657579217233384e-06,
      "loss": 0.1403,
      "step": 9933
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3504398478891557,
      "learning_rate": 1.0649585394319468e-06,
      "loss": 0.178,
      "step": 9934
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3841293253155458,
      "learning_rate": 1.064159421311809e-06,
      "loss": 0.1711,
      "step": 9935
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1729021763892495,
      "learning_rate": 1.0633605674165753e-06,
      "loss": 0.1535,
      "step": 9936
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.944858378451434,
      "learning_rate": 1.0625619777998746e-06,
      "loss": 0.509,
      "step": 9937
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3850922834110508,
      "learning_rate": 1.061763652515318e-06,
      "loss": 0.1873,
      "step": 9938
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.974479260329819,
      "learning_rate": 1.0609655916165001e-06,
      "loss": 0.5892,
      "step": 9939
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4145160287732046,
      "learning_rate": 1.0601677951569967e-06,
      "loss": 0.2047,
      "step": 9940
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2707893181561614,
      "learning_rate": 1.0593702631903668e-06,
      "loss": 0.1833,
      "step": 9941
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.424826215870271,
      "learning_rate": 1.0585729957701491e-06,
      "loss": 0.5179,
      "step": 9942
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4135029807660044,
      "learning_rate": 1.0577759929498704e-06,
      "loss": 0.1705,
      "step": 9943
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1061780767478582,
      "learning_rate": 1.0569792547830321e-06,
      "loss": 0.1171,
      "step": 9944
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2590441478411794,
      "learning_rate": 1.056182781323124e-06,
      "loss": 0.1276,
      "step": 9945
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.300748557697865,
      "learning_rate": 1.0553865726236162e-06,
      "loss": 0.1767,
      "step": 9946
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3675357273246527,
      "learning_rate": 1.0545906287379577e-06,
      "loss": 0.1714,
      "step": 9947
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.791488917962926,
      "learning_rate": 1.053794949719585e-06,
      "loss": 0.5518,
      "step": 9948
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.162905316454507,
      "learning_rate": 1.0529995356219136e-06,
      "loss": 0.1368,
      "step": 9949
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2995644358985834,
      "learning_rate": 1.0522043864983428e-06,
      "loss": 0.1695,
      "step": 9950
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3957546600893522,
      "learning_rate": 1.051409502402253e-06,
      "loss": 0.1885,
      "step": 9951
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2705138969544578,
      "learning_rate": 1.0506148833870066e-06,
      "loss": 0.1648,
      "step": 9952
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.603592887565018,
      "learning_rate": 1.0498205295059482e-06,
      "loss": 0.4356,
      "step": 9953
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3285689063144732,
      "learning_rate": 1.049026440812409e-06,
      "loss": 0.1791,
      "step": 9954
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2284098677397233,
      "learning_rate": 1.0482326173596947e-06,
      "loss": 0.1476,
      "step": 9955
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1601105981174051,
      "learning_rate": 1.0474390592010963e-06,
      "loss": 0.138,
      "step": 9956
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3594060912168493,
      "learning_rate": 1.0466457663898927e-06,
      "loss": 0.195,
      "step": 9957
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2448690300506786,
      "learning_rate": 1.0458527389793339e-06,
      "loss": 0.157,
      "step": 9958
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3323220869448942,
      "learning_rate": 1.0450599770226627e-06,
      "loss": 0.1833,
      "step": 9959
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.415787210757145,
      "learning_rate": 1.0442674805730986e-06,
      "loss": 0.1755,
      "step": 9960
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4543179219423301,
      "learning_rate": 1.0434752496838435e-06,
      "loss": 0.1736,
      "step": 9961
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4665996593384516,
      "learning_rate": 1.042683284408083e-06,
      "loss": 0.163,
      "step": 9962
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.648706940443995,
      "learning_rate": 1.041891584798984e-06,
      "loss": 0.5727,
      "step": 9963
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.394710071179995,
      "learning_rate": 1.0411001509096952e-06,
      "loss": 0.1741,
      "step": 9964
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4257600857208834,
      "learning_rate": 1.0403089827933482e-06,
      "loss": 0.1802,
      "step": 9965
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5760279842274405,
      "learning_rate": 1.039518080503057e-06,
      "loss": 0.1908,
      "step": 9966
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4366350849790221,
      "learning_rate": 1.0387274440919149e-06,
      "loss": 0.1747,
      "step": 9967
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6765152122595948,
      "learning_rate": 1.0379370736130045e-06,
      "loss": 0.1922,
      "step": 9968
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3085727482768688,
      "learning_rate": 1.0371469691193802e-06,
      "loss": 0.1601,
      "step": 9969
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2297706640886659,
      "learning_rate": 1.0363571306640885e-06,
      "loss": 0.1799,
      "step": 9970
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4477931536776054,
      "learning_rate": 1.0355675583001518e-06,
      "loss": 0.2076,
      "step": 9971
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5680284072089725,
      "learning_rate": 1.0347782520805767e-06,
      "loss": 0.1653,
      "step": 9972
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2855100083732538,
      "learning_rate": 1.0339892120583517e-06,
      "loss": 0.202,
      "step": 9973
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.350768838295333,
      "learning_rate": 1.0332004382864475e-06,
      "loss": 0.1596,
      "step": 9974
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5135083309328234,
      "learning_rate": 1.0324119308178166e-06,
      "loss": 0.152,
      "step": 9975
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4230355166752526,
      "learning_rate": 1.0316236897053939e-06,
      "loss": 0.1704,
      "step": 9976
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3684835133805795,
      "learning_rate": 1.030835715002097e-06,
      "loss": 0.2232,
      "step": 9977
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3194609257669547,
      "learning_rate": 1.0300480067608232e-06,
      "loss": 0.1562,
      "step": 9978
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.758528060597476,
      "learning_rate": 1.0292605650344578e-06,
      "loss": 0.4557,
      "step": 9979
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.194294415865535,
      "learning_rate": 1.0284733898758587e-06,
      "loss": 0.1433,
      "step": 9980
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3540332969031237,
      "learning_rate": 1.0276864813378756e-06,
      "loss": 0.147,
      "step": 9981
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.39919577735848,
      "learning_rate": 1.0268998394733364e-06,
      "loss": 0.1761,
      "step": 9982
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3030668727110852,
      "learning_rate": 1.0261134643350457e-06,
      "loss": 0.1695,
      "step": 9983
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.485700178341435,
      "learning_rate": 1.0253273559758003e-06,
      "loss": 0.2108,
      "step": 9984
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.61418828446539,
      "learning_rate": 1.0245415144483722e-06,
      "loss": 0.2075,
      "step": 9985
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3760858704228203,
      "learning_rate": 1.0237559398055175e-06,
      "loss": 0.1632,
      "step": 9986
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.520507919029659,
      "learning_rate": 1.022970632099972e-06,
      "loss": 0.1939,
      "step": 9987
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3807956913923043,
      "learning_rate": 1.0221855913844613e-06,
      "loss": 0.1679,
      "step": 9988
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3203257908824977,
      "learning_rate": 1.021400817711682e-06,
      "loss": 0.1759,
      "step": 9989
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3288954024080573,
      "learning_rate": 1.020616311134321e-06,
      "loss": 0.1715,
      "step": 9990
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2957800879574517,
      "learning_rate": 1.0198320717050452e-06,
      "loss": 0.1428,
      "step": 9991
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5339538411519418,
      "learning_rate": 1.0190480994764994e-06,
      "loss": 0.1868,
      "step": 9992
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.509352197383225,
      "learning_rate": 1.0182643945013182e-06,
      "loss": 0.1709,
      "step": 9993
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4521454106620841,
      "learning_rate": 1.0174809568321126e-06,
      "loss": 0.2399,
      "step": 9994
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.408399137157275,
      "learning_rate": 1.016697786521476e-06,
      "loss": 0.1406,
      "step": 9995
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3154743899784178,
      "learning_rate": 1.0159148836219869e-06,
      "loss": 0.1396,
      "step": 9996
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.258736788933737,
      "learning_rate": 1.0151322481862019e-06,
      "loss": 0.5629,
      "step": 9997
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.510595066973942,
      "learning_rate": 1.0143498802666623e-06,
      "loss": 0.6855,
      "step": 9998
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2668513919324005,
      "learning_rate": 1.0135677799158938e-06,
      "loss": 0.1696,
      "step": 9999
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2965564196795851,
      "learning_rate": 1.012785947186397e-06,
      "loss": 0.1463,
      "step": 10000
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5113976463799976,
      "learning_rate": 1.012004382130659e-06,
      "loss": 0.1437,
      "step": 10001
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1422974617878314,
      "learning_rate": 1.0112230848011528e-06,
      "loss": 0.1376,
      "step": 10002
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.431943433380309,
      "learning_rate": 1.0104420552503235e-06,
      "loss": 0.1876,
      "step": 10003
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2763291212059635,
      "learning_rate": 1.009661293530609e-06,
      "loss": 0.1762,
      "step": 10004
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4575440803545459,
      "learning_rate": 1.008880799694421e-06,
      "loss": 0.1702,
      "step": 10005
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2582510640421622,
      "learning_rate": 1.0081005737941584e-06,
      "loss": 0.1682,
      "step": 10006
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2908233987972584,
      "learning_rate": 1.007320615882199e-06,
      "loss": 0.172,
      "step": 10007
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2629518894701661,
      "learning_rate": 1.0065409260109039e-06,
      "loss": 0.1244,
      "step": 10008
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5314323875631555,
      "learning_rate": 1.0057615042326163e-06,
      "loss": 0.171,
      "step": 10009
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2880729335275378,
      "learning_rate": 1.0049823505996608e-06,
      "loss": 0.1936,
      "step": 10010
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5795944641399233,
      "learning_rate": 1.0042034651643445e-06,
      "loss": 0.1765,
      "step": 10011
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.455117094432103,
      "learning_rate": 1.0034248479789554e-06,
      "loss": 0.6211,
      "step": 10012
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3365878827926418,
      "learning_rate": 1.002646499095768e-06,
      "loss": 0.1736,
      "step": 10013
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2650396275193536,
      "learning_rate": 1.0018684185670296e-06,
      "loss": 0.1282,
      "step": 10014
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6322753331810513,
      "learning_rate": 1.00109060644498e-06,
      "loss": 0.2102,
      "step": 10015
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4607266041436615,
      "learning_rate": 1.0003130627818346e-06,
      "loss": 0.1899,
      "step": 10016
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.823361724112483,
      "learning_rate": 9.9953578762979e-07,
      "loss": 0.4014,
      "step": 10017
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4584880952681074,
      "learning_rate": 9.987587810410298e-07,
      "loss": 0.1789,
      "step": 10018
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.346574596436633,
      "learning_rate": 9.979820430677163e-07,
      "loss": 0.5732,
      "step": 10019
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2625369222092133,
      "learning_rate": 9.972055737619935e-07,
      "loss": 0.1906,
      "step": 10020
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4600617952469055,
      "learning_rate": 9.96429373175989e-07,
      "loss": 0.195,
      "step": 10021
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3992841356793502,
      "learning_rate": 9.956534413618108e-07,
      "loss": 0.1985,
      "step": 10022
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.546058854916641,
      "learning_rate": 9.948777783715479e-07,
      "loss": 0.1801,
      "step": 10023
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5179555295351908,
      "learning_rate": 9.941023842572766e-07,
      "loss": 0.1879,
      "step": 10024
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4755108028697184,
      "learning_rate": 9.933272590710508e-07,
      "loss": 0.2336,
      "step": 10025
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.201938120011718,
      "learning_rate": 9.925524028649036e-07,
      "loss": 0.1374,
      "step": 10026
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3021580616389516,
      "learning_rate": 9.917778156908564e-07,
      "loss": 0.1742,
      "step": 10027
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3863431901738839,
      "learning_rate": 9.91003497600909e-07,
      "loss": 0.1751,
      "step": 10028
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2301073836730858,
      "learning_rate": 9.902294486470442e-07,
      "loss": 0.1635,
      "step": 10029
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.373950681062246,
      "learning_rate": 9.89455668881225e-07,
      "loss": 0.1491,
      "step": 10030
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4945135817804187,
      "learning_rate": 9.886821583553985e-07,
      "loss": 0.1625,
      "step": 10031
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.149791924314525,
      "learning_rate": 9.879089171214912e-07,
      "loss": 0.5265,
      "step": 10032
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5076205177859383,
      "learning_rate": 9.871359452314167e-07,
      "loss": 0.186,
      "step": 10033
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6314450456455643,
      "learning_rate": 9.86363242737064e-07,
      "loss": 0.2116,
      "step": 10034
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3856036412719313,
      "learning_rate": 9.855908096903055e-07,
      "loss": 0.1793,
      "step": 10035
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.312065376778226,
      "learning_rate": 9.848186461430026e-07,
      "loss": 0.4693,
      "step": 10036
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2612520084481689,
      "learning_rate": 9.840467521469865e-07,
      "loss": 0.1321,
      "step": 10037
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4247847936515623,
      "learning_rate": 9.832751277540809e-07,
      "loss": 0.1658,
      "step": 10038
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4647215319895794,
      "learning_rate": 9.825037730160864e-07,
      "loss": 0.2035,
      "step": 10039
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.301271780810483,
      "learning_rate": 9.81732687984786e-07,
      "loss": 0.1704,
      "step": 10040
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.354442230069663,
      "learning_rate": 9.809618727119451e-07,
      "loss": 0.1937,
      "step": 10041
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.37472021069647,
      "learning_rate": 9.801913272493114e-07,
      "loss": 0.1615,
      "step": 10042
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3594527216471728,
      "learning_rate": 9.79421051648613e-07,
      "loss": 0.193,
      "step": 10043
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3056440212367473,
      "learning_rate": 9.786510459615611e-07,
      "loss": 0.1443,
      "step": 10044
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3322938711358383,
      "learning_rate": 9.778813102398494e-07,
      "loss": 0.1767,
      "step": 10045
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.368881752894228,
      "learning_rate": 9.771118445351496e-07,
      "loss": 0.6692,
      "step": 10046
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2688116061534769,
      "learning_rate": 9.76342648899124e-07,
      "loss": 0.1864,
      "step": 10047
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4286197167591534,
      "learning_rate": 9.755737233834045e-07,
      "loss": 0.1648,
      "step": 10048
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3696945301502303,
      "learning_rate": 9.748050680396159e-07,
      "loss": 0.1843,
      "step": 10049
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.572372266645736,
      "learning_rate": 9.740366829193587e-07,
      "loss": 0.2255,
      "step": 10050
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.348929924351992,
      "learning_rate": 9.732685680742176e-07,
      "loss": 0.1616,
      "step": 10051
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2850472358091638,
      "learning_rate": 9.725007235557576e-07,
      "loss": 0.1603,
      "step": 10052
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.308562313068008,
      "learning_rate": 9.717331494155275e-07,
      "loss": 0.1769,
      "step": 10053
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1715487025446583,
      "learning_rate": 9.709658457050558e-07,
      "loss": 0.1374,
      "step": 10054
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4853189358539207,
      "learning_rate": 9.701988124758544e-07,
      "loss": 0.1848,
      "step": 10055
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4012689182217613,
      "learning_rate": 9.694320497794163e-07,
      "loss": 0.1625,
      "step": 10056
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.204404334360558,
      "learning_rate": 9.686655576672155e-07,
      "loss": 0.1355,
      "step": 10057
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.536848817875134,
      "learning_rate": 9.678993361907135e-07,
      "loss": 0.6657,
      "step": 10058
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.446758772543423,
      "learning_rate": 9.671333854013425e-07,
      "loss": 0.1859,
      "step": 10059
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2761416967895671,
      "learning_rate": 9.663677053505283e-07,
      "loss": 0.1261,
      "step": 10060
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5326020254158912,
      "learning_rate": 9.656022960896726e-07,
      "loss": 0.1909,
      "step": 10061
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4577038129245006,
      "learning_rate": 9.64837157670156e-07,
      "loss": 0.1804,
      "step": 10062
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3703161050977117,
      "learning_rate": 9.640722901433481e-07,
      "loss": 0.1931,
      "step": 10063
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3940980140958328,
      "learning_rate": 9.633076935605956e-07,
      "loss": 0.1637,
      "step": 10064
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.193801389969481,
      "learning_rate": 9.625433679732288e-07,
      "loss": 0.1409,
      "step": 10065
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4505685167036253,
      "learning_rate": 9.617793134325571e-07,
      "loss": 0.137,
      "step": 10066
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.229713337947451,
      "learning_rate": 9.610155299898782e-07,
      "loss": 0.1791,
      "step": 10067
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3515869156981164,
      "learning_rate": 9.602520176964619e-07,
      "loss": 0.1714,
      "step": 10068
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.360944203376149,
      "learning_rate": 9.594887766035688e-07,
      "loss": 0.1686,
      "step": 10069
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5328530078835172,
      "learning_rate": 9.587258067624373e-07,
      "loss": 0.2233,
      "step": 10070
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4136293820562986,
      "learning_rate": 9.579631082242846e-07,
      "loss": 0.1607,
      "step": 10071
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3307677882007,
      "learning_rate": 9.572006810403168e-07,
      "loss": 0.1765,
      "step": 10072
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.388010675280717,
      "learning_rate": 9.564385252617164e-07,
      "loss": 0.1936,
      "step": 10073
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3439623839947026,
      "learning_rate": 9.556766409396494e-07,
      "loss": 0.1417,
      "step": 10074
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5410267554431705,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.1902,
      "step": 10075
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.402278580065271,
      "learning_rate": 9.54153686869687e-07,
      "loss": 0.4114,
      "step": 10076
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.942463613217658,
      "learning_rate": 9.533926172240327e-07,
      "loss": 0.5871,
      "step": 10077
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3959807496627432,
      "learning_rate": 9.526318192393924e-07,
      "loss": 0.1872,
      "step": 10078
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.441370761589804,
      "learning_rate": 9.518712929668416e-07,
      "loss": 0.1716,
      "step": 10079
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3495230536321061,
      "learning_rate": 9.511110384574345e-07,
      "loss": 0.1727,
      "step": 10080
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2211091018447036,
      "learning_rate": 9.503510557622137e-07,
      "loss": 0.1621,
      "step": 10081
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1767406562933385,
      "learning_rate": 9.495913449321942e-07,
      "loss": 0.1516,
      "step": 10082
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2368803547707719,
      "learning_rate": 9.48831906018381e-07,
      "loss": 0.1582,
      "step": 10083
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2957876988295747,
      "learning_rate": 9.480727390717565e-07,
      "loss": 0.189,
      "step": 10084
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.5991153851941915,
      "learning_rate": 9.473138441432855e-07,
      "loss": 0.5869,
      "step": 10085
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3022780964024205,
      "learning_rate": 9.465552212839158e-07,
      "loss": 0.1621,
      "step": 10086
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3377884186304294,
      "learning_rate": 9.457968705445753e-07,
      "loss": 0.1826,
      "step": 10087
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5249628537459352,
      "learning_rate": 9.45038791976175e-07,
      "loss": 0.2333,
      "step": 10088
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2496405454674895,
      "learning_rate": 9.442809856296059e-07,
      "loss": 0.1538,
      "step": 10089
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3712781912859877,
      "learning_rate": 9.435234515557434e-07,
      "loss": 0.187,
      "step": 10090
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.712126387153397,
      "learning_rate": 9.427661898054402e-07,
      "loss": 0.5242,
      "step": 10091
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2821361773532347,
      "learning_rate": 9.420092004295389e-07,
      "loss": 0.1736,
      "step": 10092
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.429774165106857,
      "learning_rate": 9.412524834788522e-07,
      "loss": 0.1767,
      "step": 10093
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3546507984119363,
      "learning_rate": 9.40496039004185e-07,
      "loss": 0.1891,
      "step": 10094
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3985843222252938,
      "learning_rate": 9.397398670563201e-07,
      "loss": 0.1662,
      "step": 10095
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3557829102545365,
      "learning_rate": 9.389839676860174e-07,
      "loss": 0.1807,
      "step": 10096
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2329748714711344,
      "learning_rate": 9.382283409440262e-07,
      "loss": 0.1717,
      "step": 10097
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.179315096265657,
      "learning_rate": 9.374729868810739e-07,
      "loss": 0.1241,
      "step": 10098
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1497764550843674,
      "learning_rate": 9.367179055478681e-07,
      "loss": 0.1582,
      "step": 10099
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.986894478010182,
      "learning_rate": 9.359630969951012e-07,
      "loss": 0.5949,
      "step": 10100
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2197150578858686,
      "learning_rate": 9.352085612734452e-07,
      "loss": 0.1889,
      "step": 10101
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3800968861366738,
      "learning_rate": 9.344542984335525e-07,
      "loss": 0.1823,
      "step": 10102
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.4281243335426685,
      "learning_rate": 9.337003085260638e-07,
      "loss": 0.4373,
      "step": 10103
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.526783397251375,
      "learning_rate": 9.329465916015923e-07,
      "loss": 0.482,
      "step": 10104
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4055324414121964,
      "learning_rate": 9.321931477107377e-07,
      "loss": 0.1495,
      "step": 10105
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4401525537341084,
      "learning_rate": 9.314399769040844e-07,
      "loss": 0.186,
      "step": 10106
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4846350846867904,
      "learning_rate": 9.306870792321903e-07,
      "loss": 0.1693,
      "step": 10107
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6982012014069987,
      "learning_rate": 9.299344547456035e-07,
      "loss": 0.1918,
      "step": 10108
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4553762991102994,
      "learning_rate": 9.291821034948484e-07,
      "loss": 0.2124,
      "step": 10109
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3083853785930153,
      "learning_rate": 9.284300255304329e-07,
      "loss": 0.1375,
      "step": 10110
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1610137892233543,
      "learning_rate": 9.276782209028446e-07,
      "loss": 0.1431,
      "step": 10111
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3412238319325342,
      "learning_rate": 9.269266896625584e-07,
      "loss": 0.1798,
      "step": 10112
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.363693221338907,
      "learning_rate": 9.261754318600235e-07,
      "loss": 0.185,
      "step": 10113
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2262690995616388,
      "learning_rate": 9.254244475456742e-07,
      "loss": 0.1498,
      "step": 10114
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3229844619457494,
      "learning_rate": 9.246737367699287e-07,
      "loss": 0.1555,
      "step": 10115
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3770174025288464,
      "learning_rate": 9.239232995831815e-07,
      "loss": 0.1969,
      "step": 10116
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4644121366983032,
      "learning_rate": 9.231731360358137e-07,
      "loss": 0.2008,
      "step": 10117
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.9286313986876635,
      "learning_rate": 9.224232461781863e-07,
      "loss": 0.6244,
      "step": 10118
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3478104728444238,
      "learning_rate": 9.216736300606405e-07,
      "loss": 0.1453,
      "step": 10119
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.062679257046323,
      "learning_rate": 9.209242877335006e-07,
      "loss": 0.5292,
      "step": 10120
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3631776764650816,
      "learning_rate": 9.201752192470725e-07,
      "loss": 0.1752,
      "step": 10121
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4125811182196126,
      "learning_rate": 9.194264246516438e-07,
      "loss": 0.1655,
      "step": 10122
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2692478886240817,
      "learning_rate": 9.186779039974825e-07,
      "loss": 0.1667,
      "step": 10123
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1785626696055125,
      "learning_rate": 9.179296573348389e-07,
      "loss": 0.1505,
      "step": 10124
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1580569792355342,
      "learning_rate": 9.171816847139447e-07,
      "loss": 0.1788,
      "step": 10125
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2877269136183882,
      "learning_rate": 9.164339861850175e-07,
      "loss": 0.1397,
      "step": 10126
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4043346446213893,
      "learning_rate": 9.156865617982463e-07,
      "loss": 0.1709,
      "step": 10127
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3162552386435677,
      "learning_rate": 9.14939411603813e-07,
      "loss": 0.1602,
      "step": 10128
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4356209036912964,
      "learning_rate": 9.141925356518738e-07,
      "loss": 0.1797,
      "step": 10129
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.405150612804565,
      "learning_rate": 9.134459339925694e-07,
      "loss": 0.5443,
      "step": 10130
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1413983824023846,
      "learning_rate": 9.126996066760213e-07,
      "loss": 0.1306,
      "step": 10131
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2461612744569732,
      "learning_rate": 9.119535537523322e-07,
      "loss": 0.1513,
      "step": 10132
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2036110814627825,
      "learning_rate": 9.112077752715876e-07,
      "loss": 0.1673,
      "step": 10133
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3496340584246618,
      "learning_rate": 9.104622712838535e-07,
      "loss": 0.1901,
      "step": 10134
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2153105576874768,
      "learning_rate": 9.097170418391782e-07,
      "loss": 0.1486,
      "step": 10135
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.29087728148871,
      "learning_rate": 9.089720869875896e-07,
      "loss": 0.1762,
      "step": 10136
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1990201679849206,
      "learning_rate": 9.082274067791025e-07,
      "loss": 0.1723,
      "step": 10137
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4442979123706765,
      "learning_rate": 9.074830012637059e-07,
      "loss": 0.179,
      "step": 10138
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3194447894321615,
      "learning_rate": 9.067388704913743e-07,
      "loss": 0.179,
      "step": 10139
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3310475818763872,
      "learning_rate": 9.059950145120666e-07,
      "loss": 0.151,
      "step": 10140
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4155127187803922,
      "learning_rate": 9.052514333757156e-07,
      "loss": 0.1987,
      "step": 10141
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4405919859424205,
      "learning_rate": 9.045081271322437e-07,
      "loss": 0.1662,
      "step": 10142
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.9126493216228955,
      "learning_rate": 9.037650958315497e-07,
      "loss": 0.551,
      "step": 10143
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5951300933305457,
      "learning_rate": 9.03022339523516e-07,
      "loss": 0.1929,
      "step": 10144
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1274138164519414,
      "learning_rate": 9.022798582580067e-07,
      "loss": 0.1467,
      "step": 10145
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5903364046697648,
      "learning_rate": 9.015376520848651e-07,
      "loss": 0.1447,
      "step": 10146
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3020114522763622,
      "learning_rate": 9.007957210539197e-07,
      "loss": 0.188,
      "step": 10147
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.565293678684902,
      "learning_rate": 9.00054065214977e-07,
      "loss": 0.1723,
      "step": 10148
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.389178217662915,
      "learning_rate": 8.993126846178274e-07,
      "loss": 0.1911,
      "step": 10149
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3146572208404759,
      "learning_rate": 8.985715793122407e-07,
      "loss": 0.1579,
      "step": 10150
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3014947288220622,
      "learning_rate": 8.978307493479732e-07,
      "loss": 0.1649,
      "step": 10151
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5167848691754688,
      "learning_rate": 8.97090194774754e-07,
      "loss": 0.1892,
      "step": 10152
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3089259110386358,
      "learning_rate": 8.963499156423028e-07,
      "loss": 0.191,
      "step": 10153
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5133426054415562,
      "learning_rate": 8.956099120003153e-07,
      "loss": 0.2266,
      "step": 10154
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6346932047428013,
      "learning_rate": 8.948701838984702e-07,
      "loss": 0.1675,
      "step": 10155
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2817442343581882,
      "learning_rate": 8.941307313864278e-07,
      "loss": 0.15,
      "step": 10156
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.232264931871029,
      "learning_rate": 8.933915545138294e-07,
      "loss": 0.1648,
      "step": 10157
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5466996781683333,
      "learning_rate": 8.926526533302987e-07,
      "loss": 0.2042,
      "step": 10158
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.343454333888892,
      "learning_rate": 8.919140278854388e-07,
      "loss": 0.1724,
      "step": 10159
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.11117693645915,
      "learning_rate": 8.911756782288394e-07,
      "loss": 0.4436,
      "step": 10160
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3681415721076284,
      "learning_rate": 8.904376044100638e-07,
      "loss": 0.1708,
      "step": 10161
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2694094600991526,
      "learning_rate": 8.896998064786644e-07,
      "loss": 0.1422,
      "step": 10162
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5185317545157733,
      "learning_rate": 8.889622844841711e-07,
      "loss": 0.19,
      "step": 10163
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4643123021737832,
      "learning_rate": 8.882250384760949e-07,
      "loss": 0.2178,
      "step": 10164
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2730700245486566,
      "learning_rate": 8.874880685039305e-07,
      "loss": 0.1494,
      "step": 10165
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3224266243528155,
      "learning_rate": 8.867513746171525e-07,
      "loss": 0.168,
      "step": 10166
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3743402053642682,
      "learning_rate": 8.860149568652177e-07,
      "loss": 0.1879,
      "step": 10167
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3521331636649152,
      "learning_rate": 8.852788152975627e-07,
      "loss": 0.1714,
      "step": 10168
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3785988865360714,
      "learning_rate": 8.845429499636088e-07,
      "loss": 0.1745,
      "step": 10169
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2955353682641584,
      "learning_rate": 8.838073609127546e-07,
      "loss": 0.1757,
      "step": 10170
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5032068541606938,
      "learning_rate": 8.830720481943861e-07,
      "loss": 0.1771,
      "step": 10171
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4066557829522546,
      "learning_rate": 8.823370118578628e-07,
      "loss": 0.1836,
      "step": 10172
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.832307198834741,
      "learning_rate": 8.816022519525325e-07,
      "loss": 0.6178,
      "step": 10173
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3504139353560576,
      "learning_rate": 8.808677685277228e-07,
      "loss": 0.1527,
      "step": 10174
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4863088404845417,
      "learning_rate": 8.801335616327378e-07,
      "loss": 0.1578,
      "step": 10175
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3123930284380871,
      "learning_rate": 8.793996313168713e-07,
      "loss": 0.1767,
      "step": 10176
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4173868503259681,
      "learning_rate": 8.786659776293915e-07,
      "loss": 0.191,
      "step": 10177
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3300662074215983,
      "learning_rate": 8.779326006195521e-07,
      "loss": 0.1803,
      "step": 10178
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.432763562212586,
      "learning_rate": 8.771995003365869e-07,
      "loss": 0.1822,
      "step": 10179
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.581227949084694,
      "learning_rate": 8.764666768297108e-07,
      "loss": 0.1851,
      "step": 10180
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1725873961372244,
      "learning_rate": 8.757341301481187e-07,
      "loss": 0.1403,
      "step": 10181
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.314622327606304,
      "learning_rate": 8.750018603409937e-07,
      "loss": 0.1239,
      "step": 10182
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3602771488132364,
      "learning_rate": 8.742698674574901e-07,
      "loss": 0.1369,
      "step": 10183
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2930471104708752,
      "learning_rate": 8.7353815154675e-07,
      "loss": 0.1685,
      "step": 10184
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.320414040092015,
      "learning_rate": 8.728067126578988e-07,
      "loss": 0.5274,
      "step": 10185
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.566157736565579,
      "learning_rate": 8.720755508400358e-07,
      "loss": 0.1887,
      "step": 10186
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2926427349554652,
      "learning_rate": 8.713446661422487e-07,
      "loss": 0.1705,
      "step": 10187
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3268460563198965,
      "learning_rate": 8.706140586136042e-07,
      "loss": 0.1732,
      "step": 10188
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2833558655714281,
      "learning_rate": 8.698837283031486e-07,
      "loss": 0.1883,
      "step": 10189
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4524148890269948,
      "learning_rate": 8.691536752599128e-07,
      "loss": 0.2212,
      "step": 10190
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2532004466690323,
      "learning_rate": 8.684238995329064e-07,
      "loss": 0.1541,
      "step": 10191
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4253423445528368,
      "learning_rate": 8.676944011711219e-07,
      "loss": 0.146,
      "step": 10192
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.350281275742027,
      "learning_rate": 8.66965180223533e-07,
      "loss": 0.5016,
      "step": 10193
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4374510606643063,
      "learning_rate": 8.662362367390936e-07,
      "loss": 0.1757,
      "step": 10194
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.380235153998684,
      "learning_rate": 8.655075707667399e-07,
      "loss": 0.179,
      "step": 10195
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4426369974105007,
      "learning_rate": 8.647791823553903e-07,
      "loss": 0.2063,
      "step": 10196
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2096274356416226,
      "learning_rate": 8.640510715539441e-07,
      "loss": 0.1143,
      "step": 10197
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3078016981872718,
      "learning_rate": 8.633232384112805e-07,
      "loss": 0.188,
      "step": 10198
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1304389196037985,
      "learning_rate": 8.625956829762616e-07,
      "loss": 0.1402,
      "step": 10199
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.295331212149966,
      "learning_rate": 8.618684052977305e-07,
      "loss": 0.6101,
      "step": 10200
    },
    {
      "epoch": 0.82,
      "grad_norm": 14.217922861346684,
      "learning_rate": 8.611414054245115e-07,
      "loss": 0.8411,
      "step": 10201
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3834796284681978,
      "learning_rate": 8.6041468340541e-07,
      "loss": 0.1693,
      "step": 10202
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5542061501098994,
      "learning_rate": 8.596882392892131e-07,
      "loss": 0.1859,
      "step": 10203
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.126338155061678,
      "learning_rate": 8.589620731246883e-07,
      "loss": 0.5776,
      "step": 10204
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.476175690080442,
      "learning_rate": 8.582361849605891e-07,
      "loss": 0.209,
      "step": 10205
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2925080028539806,
      "learning_rate": 8.575105748456408e-07,
      "loss": 0.1711,
      "step": 10206
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4374310240949202,
      "learning_rate": 8.567852428285606e-07,
      "loss": 0.1916,
      "step": 10207
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3711472107200957,
      "learning_rate": 8.560601889580416e-07,
      "loss": 0.1379,
      "step": 10208
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4185831402537163,
      "learning_rate": 8.553354132827552e-07,
      "loss": 0.1811,
      "step": 10209
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.284024720252546,
      "learning_rate": 8.546109158513615e-07,
      "loss": 0.1266,
      "step": 10210
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3846411167175223,
      "learning_rate": 8.53886696712497e-07,
      "loss": 0.1641,
      "step": 10211
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3589856101357376,
      "learning_rate": 8.531627559147809e-07,
      "loss": 0.159,
      "step": 10212
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2942546956250405,
      "learning_rate": 8.524390935068133e-07,
      "loss": 0.1649,
      "step": 10213
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4209052169030074,
      "learning_rate": 8.517157095371764e-07,
      "loss": 0.1829,
      "step": 10214
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.528574988377464,
      "learning_rate": 8.509926040544308e-07,
      "loss": 0.1897,
      "step": 10215
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4277434223581493,
      "learning_rate": 8.502697771071255e-07,
      "loss": 0.1623,
      "step": 10216
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3526510814419397,
      "learning_rate": 8.495472287437823e-07,
      "loss": 0.1592,
      "step": 10217
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.201078939685123,
      "learning_rate": 8.488249590129078e-07,
      "loss": 0.5186,
      "step": 10218
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5419635103865734,
      "learning_rate": 8.481029679629932e-07,
      "loss": 0.1829,
      "step": 10219
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3076809294870706,
      "learning_rate": 8.473812556425037e-07,
      "loss": 0.1656,
      "step": 10220
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4475514325305514,
      "learning_rate": 8.466598220998939e-07,
      "loss": 0.2103,
      "step": 10221
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4635748686452354,
      "learning_rate": 8.459386673835945e-07,
      "loss": 0.1576,
      "step": 10222
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5854066548269123,
      "learning_rate": 8.45217791542019e-07,
      "loss": 0.1956,
      "step": 10223
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3487467933644017,
      "learning_rate": 8.444971946235608e-07,
      "loss": 0.1756,
      "step": 10224
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2799984585236162,
      "learning_rate": 8.437768766765975e-07,
      "loss": 0.1437,
      "step": 10225
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3299686338232437,
      "learning_rate": 8.430568377494847e-07,
      "loss": 0.1995,
      "step": 10226
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.780738622550268,
      "learning_rate": 8.423370778905615e-07,
      "loss": 0.4675,
      "step": 10227
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1932586788249298,
      "learning_rate": 8.416175971481471e-07,
      "loss": 0.1586,
      "step": 10228
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.045581922329502,
      "learning_rate": 8.408983955705424e-07,
      "loss": 0.5504,
      "step": 10229
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.177603483615978,
      "learning_rate": 8.40179473206032e-07,
      "loss": 0.1296,
      "step": 10230
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.24634579885937,
      "learning_rate": 8.394608301028751e-07,
      "loss": 0.1384,
      "step": 10231
    },
    {
      "epoch": 0.82,
      "grad_norm": 34.178454368488396,
      "learning_rate": 8.387424663093197e-07,
      "loss": 0.4803,
      "step": 10232
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4796242159355029,
      "learning_rate": 8.380243818735901e-07,
      "loss": 0.213,
      "step": 10233
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.182760868179684,
      "learning_rate": 8.373065768438948e-07,
      "loss": 0.1776,
      "step": 10234
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3875879422596897,
      "learning_rate": 8.365890512684211e-07,
      "loss": 0.1753,
      "step": 10235
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3055608926558224,
      "learning_rate": 8.35871805195339e-07,
      "loss": 0.1725,
      "step": 10236
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3900499999286327,
      "learning_rate": 8.351548386727998e-07,
      "loss": 0.1658,
      "step": 10237
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4287482035934234,
      "learning_rate": 8.344381517489336e-07,
      "loss": 0.1837,
      "step": 10238
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2754125935991587,
      "learning_rate": 8.337217444718582e-07,
      "loss": 0.1793,
      "step": 10239
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2165289290680796,
      "learning_rate": 8.330056168896628e-07,
      "loss": 0.1288,
      "step": 10240
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.240056194016386,
      "learning_rate": 8.322897690504267e-07,
      "loss": 0.1645,
      "step": 10241
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4599209036711218,
      "learning_rate": 8.315742010022065e-07,
      "loss": 0.2121,
      "step": 10242
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3473861084448016,
      "learning_rate": 8.3085891279304e-07,
      "loss": 0.1593,
      "step": 10243
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.068711740818255,
      "learning_rate": 8.301439044709464e-07,
      "loss": 0.1337,
      "step": 10244
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3978541147115797,
      "learning_rate": 8.294291760839268e-07,
      "loss": 0.1906,
      "step": 10245
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3591865112054997,
      "learning_rate": 8.287147276799629e-07,
      "loss": 0.1744,
      "step": 10246
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3809659197413373,
      "learning_rate": 8.280005593070184e-07,
      "loss": 0.1895,
      "step": 10247
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3398066610758947,
      "learning_rate": 8.272866710130362e-07,
      "loss": 0.1357,
      "step": 10248
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3317007447203062,
      "learning_rate": 8.265730628459417e-07,
      "loss": 0.167,
      "step": 10249
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.943742627975102,
      "learning_rate": 8.258597348536452e-07,
      "loss": 0.6698,
      "step": 10250
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.27753281613557,
      "learning_rate": 8.251466870840292e-07,
      "loss": 0.1658,
      "step": 10251
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.435594771040495,
      "learning_rate": 8.24433919584966e-07,
      "loss": 0.1877,
      "step": 10252
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3489143008296707,
      "learning_rate": 8.237214324043069e-07,
      "loss": 0.1665,
      "step": 10253
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3345924308029946,
      "learning_rate": 8.230092255898797e-07,
      "loss": 0.1512,
      "step": 10254
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4335615320364972,
      "learning_rate": 8.222972991894995e-07,
      "loss": 0.1603,
      "step": 10255
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.296630469785851,
      "learning_rate": 8.215856532509597e-07,
      "loss": 0.1567,
      "step": 10256
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3172950175589975,
      "learning_rate": 8.208742878220355e-07,
      "loss": 0.1599,
      "step": 10257
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4205039970725006,
      "learning_rate": 8.201632029504825e-07,
      "loss": 0.2046,
      "step": 10258
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.669082247410548,
      "learning_rate": 8.194523986840375e-07,
      "loss": 0.496,
      "step": 10259
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3981452525884392,
      "learning_rate": 8.187418750704202e-07,
      "loss": 0.1513,
      "step": 10260
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4767843656938935,
      "learning_rate": 8.180316321573295e-07,
      "loss": 0.1753,
      "step": 10261
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.55169400769144,
      "learning_rate": 8.173216699924458e-07,
      "loss": 0.6634,
      "step": 10262
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3578846611849653,
      "learning_rate": 8.166119886234302e-07,
      "loss": 0.145,
      "step": 10263
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2586239644009136,
      "learning_rate": 8.159025880979299e-07,
      "loss": 0.1532,
      "step": 10264
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4690218720524801,
      "learning_rate": 8.151934684635632e-07,
      "loss": 0.1487,
      "step": 10265
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3058316899303153,
      "learning_rate": 8.144846297679398e-07,
      "loss": 0.1768,
      "step": 10266
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4187275367977383,
      "learning_rate": 8.137760720586446e-07,
      "loss": 0.1879,
      "step": 10267
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.174436859255855,
      "learning_rate": 8.13067795383246e-07,
      "loss": 0.1162,
      "step": 10268
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4840360799713945,
      "learning_rate": 8.123597997892918e-07,
      "loss": 0.1914,
      "step": 10269
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.449240062470318,
      "learning_rate": 8.116520853243126e-07,
      "loss": 0.1648,
      "step": 10270
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.441033624486293,
      "learning_rate": 8.109446520358188e-07,
      "loss": 0.1894,
      "step": 10271
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4038540803967303,
      "learning_rate": 8.102374999713025e-07,
      "loss": 0.1669,
      "step": 10272
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.403092525328924,
      "learning_rate": 8.095306291782378e-07,
      "loss": 0.1981,
      "step": 10273
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.997909340123656,
      "learning_rate": 8.088240397040776e-07,
      "loss": 0.4697,
      "step": 10274
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3799519560359428,
      "learning_rate": 8.081177315962601e-07,
      "loss": 0.1654,
      "step": 10275
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4284416808049756,
      "learning_rate": 8.074117049021985e-07,
      "loss": 0.1944,
      "step": 10276
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3285882879490167,
      "learning_rate": 8.067059596692928e-07,
      "loss": 0.1938,
      "step": 10277
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.456921684426916,
      "learning_rate": 8.060004959449225e-07,
      "loss": 0.1929,
      "step": 10278
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2809114171272378,
      "learning_rate": 8.052953137764441e-07,
      "loss": 0.1611,
      "step": 10279
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4048132551457462,
      "learning_rate": 8.04590413211202e-07,
      "loss": 0.1646,
      "step": 10280
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.896066147142604,
      "learning_rate": 8.038857942965167e-07,
      "loss": 0.6755,
      "step": 10281
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3090141934836064,
      "learning_rate": 8.03181457079692e-07,
      "loss": 0.1735,
      "step": 10282
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3708271984127252,
      "learning_rate": 8.024774016080105e-07,
      "loss": 0.1804,
      "step": 10283
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.196900333388832,
      "learning_rate": 8.017736279287414e-07,
      "loss": 0.1324,
      "step": 10284
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3169728460582562,
      "learning_rate": 8.010701360891265e-07,
      "loss": 0.1667,
      "step": 10285
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.637443000044592,
      "learning_rate": 8.003669261363972e-07,
      "loss": 0.1959,
      "step": 10286
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2216941627208409,
      "learning_rate": 7.996639981177612e-07,
      "loss": 0.149,
      "step": 10287
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3581970271787118,
      "learning_rate": 7.989613520804051e-07,
      "loss": 0.1927,
      "step": 10288
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2676407083427539,
      "learning_rate": 7.982589880715036e-07,
      "loss": 0.1688,
      "step": 10289
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4461391514232231,
      "learning_rate": 7.975569061382066e-07,
      "loss": 0.1895,
      "step": 10290
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3737987257525235,
      "learning_rate": 7.96855106327648e-07,
      "loss": 0.1688,
      "step": 10291
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.883446170259811,
      "learning_rate": 7.96153588686941e-07,
      "loss": 0.5334,
      "step": 10292
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2515985009100468,
      "learning_rate": 7.954523532631802e-07,
      "loss": 0.1396,
      "step": 10293
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.424250936452341,
      "learning_rate": 7.947514001034412e-07,
      "loss": 0.1481,
      "step": 10294
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1785945464249235,
      "learning_rate": 7.94050729254785e-07,
      "loss": 0.1648,
      "step": 10295
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1600100051842837,
      "learning_rate": 7.933503407642451e-07,
      "loss": 0.1715,
      "step": 10296
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.528021105323564,
      "learning_rate": 7.926502346788412e-07,
      "loss": 0.1688,
      "step": 10297
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3333077728031326,
      "learning_rate": 7.919504110455778e-07,
      "loss": 0.1578,
      "step": 10298
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2619531300973006,
      "learning_rate": 7.912508699114302e-07,
      "loss": 0.1516,
      "step": 10299
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.9652040621247835,
      "learning_rate": 7.905516113233652e-07,
      "loss": 0.555,
      "step": 10300
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2526973218103923,
      "learning_rate": 7.898526353283248e-07,
      "loss": 0.1413,
      "step": 10301
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3284422797678117,
      "learning_rate": 7.89153941973233e-07,
      "loss": 0.1441,
      "step": 10302
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5250098737111806,
      "learning_rate": 7.884555313049952e-07,
      "loss": 0.2019,
      "step": 10303
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4564496313900295,
      "learning_rate": 7.877574033704982e-07,
      "loss": 0.1629,
      "step": 10304
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.587573692658112,
      "learning_rate": 7.870595582166096e-07,
      "loss": 0.1985,
      "step": 10305
    },
    {
      "epoch": 0.82,
      "grad_norm": 10.106469808086027,
      "learning_rate": 7.863619958901775e-07,
      "loss": 0.527,
      "step": 10306
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3792532837625195,
      "learning_rate": 7.856647164380315e-07,
      "loss": 0.1633,
      "step": 10307
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3403237328401127,
      "learning_rate": 7.849677199069805e-07,
      "loss": 0.1993,
      "step": 10308
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2895561568780798,
      "learning_rate": 7.842710063438202e-07,
      "loss": 0.1781,
      "step": 10309
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3278200557695121,
      "learning_rate": 7.835745757953178e-07,
      "loss": 0.192,
      "step": 10310
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4079285895297915,
      "learning_rate": 7.828784283082303e-07,
      "loss": 0.1669,
      "step": 10311
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2861926768039775,
      "learning_rate": 7.821825639292918e-07,
      "loss": 0.16,
      "step": 10312
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4617563603954875,
      "learning_rate": 7.814869827052168e-07,
      "loss": 0.1566,
      "step": 10313
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.497629502190096,
      "learning_rate": 7.807916846827024e-07,
      "loss": 0.7361,
      "step": 10314
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2581980042741252,
      "learning_rate": 7.800966699084262e-07,
      "loss": 0.1838,
      "step": 10315
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.391378832734386,
      "learning_rate": 7.794019384290464e-07,
      "loss": 0.1863,
      "step": 10316
    },
    {
      "epoch": 0.83,
      "grad_norm": 14.500750199627793,
      "learning_rate": 7.787074902912018e-07,
      "loss": 0.6652,
      "step": 10317
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1593065025799003,
      "learning_rate": 7.780133255415139e-07,
      "loss": 0.1048,
      "step": 10318
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4932383857417515,
      "learning_rate": 7.773194442265819e-07,
      "loss": 0.1585,
      "step": 10319
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3300243050752893,
      "learning_rate": 7.766258463929926e-07,
      "loss": 0.1692,
      "step": 10320
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.293679057996431,
      "learning_rate": 7.759325320873051e-07,
      "loss": 0.1861,
      "step": 10321
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.094428191613522,
      "learning_rate": 7.752395013560643e-07,
      "loss": 0.5467,
      "step": 10322
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.279826462186724,
      "learning_rate": 7.745467542457985e-07,
      "loss": 0.1898,
      "step": 10323
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3608804146744042,
      "learning_rate": 7.738542908030095e-07,
      "loss": 0.1712,
      "step": 10324
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.237587981488496,
      "learning_rate": 7.731621110741871e-07,
      "loss": 0.125,
      "step": 10325
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4840417345202028,
      "learning_rate": 7.724702151058e-07,
      "loss": 0.1595,
      "step": 10326
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.36070465111008,
      "learning_rate": 7.717786029442953e-07,
      "loss": 0.1905,
      "step": 10327
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.522324888305765,
      "learning_rate": 7.710872746361037e-07,
      "loss": 0.1705,
      "step": 10328
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.595643286629416,
      "learning_rate": 7.70396230227638e-07,
      "loss": 0.5053,
      "step": 10329
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6167632138969317,
      "learning_rate": 7.697054697652879e-07,
      "loss": 0.1571,
      "step": 10330
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4735380195479784,
      "learning_rate": 7.690149932954255e-07,
      "loss": 0.2098,
      "step": 10331
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2823875882351807,
      "learning_rate": 7.683248008644084e-07,
      "loss": 0.1645,
      "step": 10332
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.97413401718711,
      "learning_rate": 7.676348925185667e-07,
      "loss": 0.6059,
      "step": 10333
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3642457433622908,
      "learning_rate": 7.669452683042194e-07,
      "loss": 0.161,
      "step": 10334
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3132387397590377,
      "learning_rate": 7.662559282676619e-07,
      "loss": 0.1692,
      "step": 10335
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3607245235273113,
      "learning_rate": 7.655668724551713e-07,
      "loss": 0.1509,
      "step": 10336
    },
    {
      "epoch": 0.83,
      "grad_norm": 11.348280907923808,
      "learning_rate": 7.648781009130068e-07,
      "loss": 0.5216,
      "step": 10337
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6543635617578722,
      "learning_rate": 7.641896136874067e-07,
      "loss": 0.199,
      "step": 10338
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3340460227706135,
      "learning_rate": 7.635014108245925e-07,
      "loss": 0.1758,
      "step": 10339
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3498039307992271,
      "learning_rate": 7.628134923707642e-07,
      "loss": 0.1837,
      "step": 10340
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4413871961685307,
      "learning_rate": 7.621258583721047e-07,
      "loss": 0.2095,
      "step": 10341
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.408815771379749,
      "learning_rate": 7.61438508874775e-07,
      "loss": 0.1964,
      "step": 10342
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3254240878451835,
      "learning_rate": 7.60751443924923e-07,
      "loss": 0.1799,
      "step": 10343
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4716355231314966,
      "learning_rate": 7.600646635686687e-07,
      "loss": 0.2109,
      "step": 10344
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4283905441000049,
      "learning_rate": 7.593781678521212e-07,
      "loss": 0.1834,
      "step": 10345
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2364220957612122,
      "learning_rate": 7.586919568213663e-07,
      "loss": 0.1343,
      "step": 10346
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1992997613432193,
      "learning_rate": 7.580060305224701e-07,
      "loss": 0.1309,
      "step": 10347
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5024655394058375,
      "learning_rate": 7.573203890014824e-07,
      "loss": 0.1522,
      "step": 10348
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.477035873887483,
      "learning_rate": 7.566350323044319e-07,
      "loss": 0.4559,
      "step": 10349
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4865055360236739,
      "learning_rate": 7.55949960477328e-07,
      "loss": 0.1815,
      "step": 10350
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1905828713122224,
      "learning_rate": 7.552651735661626e-07,
      "loss": 0.1606,
      "step": 10351
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3156678256939212,
      "learning_rate": 7.545806716169074e-07,
      "loss": 0.1691,
      "step": 10352
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.392991131927554,
      "learning_rate": 7.538964546755128e-07,
      "loss": 0.1779,
      "step": 10353
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5783029536339719,
      "learning_rate": 7.532125227879178e-07,
      "loss": 0.1874,
      "step": 10354
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.204889617914643,
      "learning_rate": 7.525288760000304e-07,
      "loss": 0.1535,
      "step": 10355
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1389205384642376,
      "learning_rate": 7.518455143577502e-07,
      "loss": 0.1439,
      "step": 10356
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1364661112191676,
      "learning_rate": 7.511624379069532e-07,
      "loss": 0.1532,
      "step": 10357
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3695818613221862,
      "learning_rate": 7.504796466934927e-07,
      "loss": 0.1752,
      "step": 10358
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2894407255180065,
      "learning_rate": 7.497971407632105e-07,
      "loss": 0.1544,
      "step": 10359
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.393065319647897,
      "learning_rate": 7.491149201619236e-07,
      "loss": 0.1538,
      "step": 10360
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2348200093865822,
      "learning_rate": 7.484329849354316e-07,
      "loss": 0.1568,
      "step": 10361
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.689310826589822,
      "learning_rate": 7.477513351295152e-07,
      "loss": 0.6105,
      "step": 10362
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.287140174619003,
      "learning_rate": 7.47069970789936e-07,
      "loss": 0.1709,
      "step": 10363
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3229519000931695,
      "learning_rate": 7.46388891962434e-07,
      "loss": 0.1567,
      "step": 10364
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4493219300391396,
      "learning_rate": 7.457080986927357e-07,
      "loss": 0.1875,
      "step": 10365
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.792194172152708,
      "learning_rate": 7.450275910265415e-07,
      "loss": 0.6442,
      "step": 10366
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3707211756003133,
      "learning_rate": 7.443473690095365e-07,
      "loss": 0.1763,
      "step": 10367
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.560652280672448,
      "learning_rate": 7.436674326873878e-07,
      "loss": 0.1588,
      "step": 10368
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.62717761181012,
      "learning_rate": 7.429877821057402e-07,
      "loss": 0.4933,
      "step": 10369
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.388677768337239,
      "learning_rate": 7.423084173102213e-07,
      "loss": 0.2074,
      "step": 10370
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3741211327882001,
      "learning_rate": 7.416293383464384e-07,
      "loss": 0.1724,
      "step": 10371
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3881543362742537,
      "learning_rate": 7.409505452599808e-07,
      "loss": 0.1982,
      "step": 10372
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4340467799022172,
      "learning_rate": 7.402720380964157e-07,
      "loss": 0.2199,
      "step": 10373
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.407404142006903,
      "learning_rate": 7.395938169012978e-07,
      "loss": 0.1596,
      "step": 10374
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3696161465291723,
      "learning_rate": 7.389158817201541e-07,
      "loss": 0.1881,
      "step": 10375
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.336794376240014,
      "learning_rate": 7.382382325984971e-07,
      "loss": 0.1835,
      "step": 10376
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3986486910743217,
      "learning_rate": 7.375608695818226e-07,
      "loss": 0.171,
      "step": 10377
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3028313584057856,
      "learning_rate": 7.368837927155986e-07,
      "loss": 0.1515,
      "step": 10378
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.433895737476724,
      "learning_rate": 7.362070020452838e-07,
      "loss": 0.1946,
      "step": 10379
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.137831613535778,
      "learning_rate": 7.355304976163119e-07,
      "loss": 0.1266,
      "step": 10380
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.114095810093628,
      "learning_rate": 7.348542794740987e-07,
      "loss": 0.62,
      "step": 10381
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4945798395123349,
      "learning_rate": 7.341783476640402e-07,
      "loss": 0.1671,
      "step": 10382
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2706722717863457,
      "learning_rate": 7.335027022315144e-07,
      "loss": 0.1622,
      "step": 10383
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3211014343548666,
      "learning_rate": 7.328273432218791e-07,
      "loss": 0.1761,
      "step": 10384
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3340180668525505,
      "learning_rate": 7.32152270680473e-07,
      "loss": 0.1751,
      "step": 10385
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4613555819734572,
      "learning_rate": 7.314774846526163e-07,
      "loss": 0.2022,
      "step": 10386
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5240540572586232,
      "learning_rate": 7.308029851836079e-07,
      "loss": 0.185,
      "step": 10387
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.408532372791918,
      "learning_rate": 7.301287723187328e-07,
      "loss": 0.1824,
      "step": 10388
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3939470377133603,
      "learning_rate": 7.294548461032475e-07,
      "loss": 0.2046,
      "step": 10389
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4707705338869925,
      "learning_rate": 7.287812065823996e-07,
      "loss": 0.2012,
      "step": 10390
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4205622566664506,
      "learning_rate": 7.281078538014108e-07,
      "loss": 0.1613,
      "step": 10391
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5883635998644619,
      "learning_rate": 7.274347878054839e-07,
      "loss": 0.218,
      "step": 10392
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3895197472350735,
      "learning_rate": 7.267620086398053e-07,
      "loss": 0.1505,
      "step": 10393
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3440498097504125,
      "learning_rate": 7.260895163495407e-07,
      "loss": 0.1501,
      "step": 10394
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2859237779745636,
      "learning_rate": 7.254173109798363e-07,
      "loss": 0.1747,
      "step": 10395
    },
    {
      "epoch": 0.83,
      "grad_norm": 44.79975121297489,
      "learning_rate": 7.247453925758197e-07,
      "loss": 0.5969,
      "step": 10396
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4802334572852698,
      "learning_rate": 7.240737611825976e-07,
      "loss": 0.2012,
      "step": 10397
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.363300433590267,
      "learning_rate": 7.234024168452592e-07,
      "loss": 0.6022,
      "step": 10398
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2232651354608663,
      "learning_rate": 7.227313596088759e-07,
      "loss": 0.1434,
      "step": 10399
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5870548395422266,
      "learning_rate": 7.220605895184946e-07,
      "loss": 0.2197,
      "step": 10400
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.141045740445732,
      "learning_rate": 7.21390106619147e-07,
      "loss": 0.1225,
      "step": 10401
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.871453980027333,
      "learning_rate": 7.207199109558472e-07,
      "loss": 0.5547,
      "step": 10402
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0299418025395037,
      "learning_rate": 7.200500025735835e-07,
      "loss": 0.1956,
      "step": 10403
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.284970320221558,
      "learning_rate": 7.193803815173317e-07,
      "loss": 0.1501,
      "step": 10404
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3116599532908575,
      "learning_rate": 7.187110478320447e-07,
      "loss": 0.1706,
      "step": 10405
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5792095019405397,
      "learning_rate": 7.180420015626571e-07,
      "loss": 0.203,
      "step": 10406
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.088503193312967,
      "learning_rate": 7.173732427540824e-07,
      "loss": 0.6149,
      "step": 10407
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2227501452513352,
      "learning_rate": 7.167047714512199e-07,
      "loss": 0.1577,
      "step": 10408
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2543306642982393,
      "learning_rate": 7.160365876989428e-07,
      "loss": 0.1799,
      "step": 10409
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2909038878245038,
      "learning_rate": 7.153686915421087e-07,
      "loss": 0.1964,
      "step": 10410
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5155298423600947,
      "learning_rate": 7.147010830255585e-07,
      "loss": 0.2085,
      "step": 10411
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4026772316442193,
      "learning_rate": 7.140337621941057e-07,
      "loss": 0.1928,
      "step": 10412
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.058285708043789,
      "learning_rate": 7.133667290925538e-07,
      "loss": 0.6879,
      "step": 10413
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4252336191440804,
      "learning_rate": 7.126999837656817e-07,
      "loss": 0.1749,
      "step": 10414
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.308836359682954,
      "learning_rate": 7.120335262582495e-07,
      "loss": 0.156,
      "step": 10415
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2792537675442102,
      "learning_rate": 7.113673566149987e-07,
      "loss": 0.169,
      "step": 10416
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3615722536345924,
      "learning_rate": 7.107014748806512e-07,
      "loss": 0.1697,
      "step": 10417
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3088875539799676,
      "learning_rate": 7.100358810999098e-07,
      "loss": 0.1658,
      "step": 10418
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5571371698779042,
      "learning_rate": 7.093705753174579e-07,
      "loss": 0.196,
      "step": 10419
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3941635948160789,
      "learning_rate": 7.087055575779594e-07,
      "loss": 0.19,
      "step": 10420
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3056844359898923,
      "learning_rate": 7.080408279260575e-07,
      "loss": 0.1697,
      "step": 10421
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.040287885088047,
      "learning_rate": 7.073763864063815e-07,
      "loss": 0.567,
      "step": 10422
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4141785875279016,
      "learning_rate": 7.067122330635323e-07,
      "loss": 0.1741,
      "step": 10423
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7525062077173894,
      "learning_rate": 7.060483679421004e-07,
      "loss": 0.1879,
      "step": 10424
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.252254370329039,
      "learning_rate": 7.053847910866513e-07,
      "loss": 0.1472,
      "step": 10425
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4161467081144168,
      "learning_rate": 7.047215025417337e-07,
      "loss": 0.1707,
      "step": 10426
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2001568021114077,
      "learning_rate": 7.040585023518759e-07,
      "loss": 0.1288,
      "step": 10427
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.264703695645213,
      "learning_rate": 7.033957905615867e-07,
      "loss": 0.4106,
      "step": 10428
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3494966062795628,
      "learning_rate": 7.027333672153563e-07,
      "loss": 0.1778,
      "step": 10429
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4139999939526524,
      "learning_rate": 7.020712323576556e-07,
      "loss": 0.1779,
      "step": 10430
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3768242262370547,
      "learning_rate": 7.014093860329346e-07,
      "loss": 0.1624,
      "step": 10431
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4684058975777654,
      "learning_rate": 7.00747828285625e-07,
      "loss": 0.1823,
      "step": 10432
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3020954810503225,
      "learning_rate": 7.000865591601424e-07,
      "loss": 0.156,
      "step": 10433
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.8281829169847,
      "learning_rate": 6.994255787008747e-07,
      "loss": 0.5676,
      "step": 10434
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4333836977903172,
      "learning_rate": 6.987648869521996e-07,
      "loss": 0.1648,
      "step": 10435
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2199258430677205,
      "learning_rate": 6.981044839584705e-07,
      "loss": 0.168,
      "step": 10436
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.339178820289547,
      "learning_rate": 6.974443697640193e-07,
      "loss": 0.1826,
      "step": 10437
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3367819388389524,
      "learning_rate": 6.967845444131654e-07,
      "loss": 0.1834,
      "step": 10438
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4540219629407722,
      "learning_rate": 6.96125007950203e-07,
      "loss": 0.1962,
      "step": 10439
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3467724801680916,
      "learning_rate": 6.954657604194093e-07,
      "loss": 0.1467,
      "step": 10440
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.575143644477294,
      "learning_rate": 6.948068018650411e-07,
      "loss": 0.1707,
      "step": 10441
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3082776299698478,
      "learning_rate": 6.941481323313365e-07,
      "loss": 0.1782,
      "step": 10442
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1948581144121222,
      "learning_rate": 6.934897518625138e-07,
      "loss": 0.1328,
      "step": 10443
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.610104160435083,
      "learning_rate": 6.928316605027724e-07,
      "loss": 0.207,
      "step": 10444
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2510590343014063,
      "learning_rate": 6.921738582962923e-07,
      "loss": 0.1578,
      "step": 10445
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4065339460194437,
      "learning_rate": 6.915163452872315e-07,
      "loss": 0.1875,
      "step": 10446
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3183170444399515,
      "learning_rate": 6.908591215197353e-07,
      "loss": 0.1481,
      "step": 10447
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.281114293935818,
      "learning_rate": 6.902021870379199e-07,
      "loss": 0.5272,
      "step": 10448
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3945064991214662,
      "learning_rate": 6.89545541885891e-07,
      "loss": 0.1556,
      "step": 10449
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4610791153539384,
      "learning_rate": 6.888891861077301e-07,
      "loss": 0.2009,
      "step": 10450
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4592205866589194,
      "learning_rate": 6.882331197475006e-07,
      "loss": 0.1665,
      "step": 10451
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2465658339537615,
      "learning_rate": 6.875773428492455e-07,
      "loss": 0.1299,
      "step": 10452
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.243520508754784,
      "learning_rate": 6.869218554569895e-07,
      "loss": 0.1703,
      "step": 10453
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3044428334714238,
      "learning_rate": 6.862666576147375e-07,
      "loss": 0.1697,
      "step": 10454
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3348660441537628,
      "learning_rate": 6.856117493664743e-07,
      "loss": 0.1515,
      "step": 10455
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2668288952144335,
      "learning_rate": 6.84957130756168e-07,
      "loss": 0.181,
      "step": 10456
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3595987251388104,
      "learning_rate": 6.843028018277614e-07,
      "loss": 0.1998,
      "step": 10457
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5413024697497035,
      "learning_rate": 6.836487626251853e-07,
      "loss": 0.1509,
      "step": 10458
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.331531227951978,
      "learning_rate": 6.829950131923452e-07,
      "loss": 0.1348,
      "step": 10459
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3619631378209511,
      "learning_rate": 6.823415535731303e-07,
      "loss": 0.1562,
      "step": 10460
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.273094679328015,
      "learning_rate": 6.816883838114085e-07,
      "loss": 0.1461,
      "step": 10461
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5072539063605697,
      "learning_rate": 6.810355039510302e-07,
      "loss": 0.2246,
      "step": 10462
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4653968230754655,
      "learning_rate": 6.803829140358237e-07,
      "loss": 0.2092,
      "step": 10463
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.414135191644308,
      "learning_rate": 6.797306141096005e-07,
      "loss": 0.1677,
      "step": 10464
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.448125786890402,
      "learning_rate": 6.790786042161507e-07,
      "loss": 0.1774,
      "step": 10465
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2059194481116273,
      "learning_rate": 6.784268843992453e-07,
      "loss": 0.1376,
      "step": 10466
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2271038939011716,
      "learning_rate": 6.777754547026393e-07,
      "loss": 0.1532,
      "step": 10467
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4440136151008223,
      "learning_rate": 6.771243151700608e-07,
      "loss": 0.197,
      "step": 10468
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.530559934225353,
      "learning_rate": 6.764734658452255e-07,
      "loss": 0.5969,
      "step": 10469
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2924688849985508,
      "learning_rate": 6.758229067718269e-07,
      "loss": 0.1594,
      "step": 10470
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2520174244144135,
      "learning_rate": 6.751726379935369e-07,
      "loss": 0.1384,
      "step": 10471
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.03350384176968,
      "learning_rate": 6.745226595540122e-07,
      "loss": 0.6921,
      "step": 10472
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.408125780349087,
      "learning_rate": 6.738729714968872e-07,
      "loss": 0.5453,
      "step": 10473
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.096457808805035,
      "learning_rate": 6.732235738657767e-07,
      "loss": 0.6267,
      "step": 10474
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3671381284147623,
      "learning_rate": 6.725744667042778e-07,
      "loss": 0.2228,
      "step": 10475
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2564026402349633,
      "learning_rate": 6.719256500559668e-07,
      "loss": 0.1704,
      "step": 10476
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4735287577873937,
      "learning_rate": 6.71277123964399e-07,
      "loss": 0.1943,
      "step": 10477
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2508352843864312,
      "learning_rate": 6.706288884731155e-07,
      "loss": 0.1597,
      "step": 10478
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3805284620481453,
      "learning_rate": 6.699809436256311e-07,
      "loss": 0.1336,
      "step": 10479
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5672606853586122,
      "learning_rate": 6.693332894654442e-07,
      "loss": 0.1746,
      "step": 10480
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2830717694871543,
      "learning_rate": 6.686859260360374e-07,
      "loss": 0.1516,
      "step": 10481
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4604620027618846,
      "learning_rate": 6.680388533808652e-07,
      "loss": 0.186,
      "step": 10482
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2006048237971076,
      "learning_rate": 6.673920715433718e-07,
      "loss": 0.1545,
      "step": 10483
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3034675254005228,
      "learning_rate": 6.667455805669753e-07,
      "loss": 0.1684,
      "step": 10484
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4093061470312744,
      "learning_rate": 6.660993804950777e-07,
      "loss": 0.1415,
      "step": 10485
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4586209518755469,
      "learning_rate": 6.654534713710597e-07,
      "loss": 0.1809,
      "step": 10486
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.514294809139086,
      "learning_rate": 6.648078532382835e-07,
      "loss": 0.1888,
      "step": 10487
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3807195729913626,
      "learning_rate": 6.641625261400908e-07,
      "loss": 0.1832,
      "step": 10488
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3911693397736682,
      "learning_rate": 6.63517490119805e-07,
      "loss": 0.1636,
      "step": 10489
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.482956961084758,
      "learning_rate": 6.62872745220729e-07,
      "loss": 0.183,
      "step": 10490
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4257052717413892,
      "learning_rate": 6.622282914861456e-07,
      "loss": 0.154,
      "step": 10491
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.168065860879332,
      "learning_rate": 6.615841289593223e-07,
      "loss": 0.1366,
      "step": 10492
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5306795326379492,
      "learning_rate": 6.609402576834989e-07,
      "loss": 0.1758,
      "step": 10493
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.31979508552305,
      "learning_rate": 6.602966777019038e-07,
      "loss": 0.41,
      "step": 10494
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.403948128906313,
      "learning_rate": 6.596533890577417e-07,
      "loss": 0.1339,
      "step": 10495
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3866326970952532,
      "learning_rate": 6.590103917941982e-07,
      "loss": 0.1883,
      "step": 10496
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.440204418410186,
      "learning_rate": 6.583676859544402e-07,
      "loss": 0.1697,
      "step": 10497
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5487438483192193,
      "learning_rate": 6.577252715816135e-07,
      "loss": 0.1611,
      "step": 10498
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3817312824363903,
      "learning_rate": 6.570831487188461e-07,
      "loss": 0.1691,
      "step": 10499
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3098629824634276,
      "learning_rate": 6.564413174092443e-07,
      "loss": 0.1547,
      "step": 10500
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4483042601938303,
      "learning_rate": 6.557997776958997e-07,
      "loss": 0.2156,
      "step": 10501
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2732973811825736,
      "learning_rate": 6.55158529621876e-07,
      "loss": 0.1255,
      "step": 10502
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3997232956852186,
      "learning_rate": 6.545175732302256e-07,
      "loss": 0.1847,
      "step": 10503
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9112349805222262,
      "learning_rate": 6.538769085639779e-07,
      "loss": 0.252,
      "step": 10504
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4570916189976346,
      "learning_rate": 6.532365356661397e-07,
      "loss": 0.1734,
      "step": 10505
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3840406879774412,
      "learning_rate": 6.525964545797042e-07,
      "loss": 0.1917,
      "step": 10506
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.274855629113032,
      "learning_rate": 6.519566653476405e-07,
      "loss": 0.6124,
      "step": 10507
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3036791104548606,
      "learning_rate": 6.513171680128999e-07,
      "loss": 0.1952,
      "step": 10508
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4431391582727755,
      "learning_rate": 6.506779626184134e-07,
      "loss": 0.1663,
      "step": 10509
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.528983189047326,
      "learning_rate": 6.500390492070941e-07,
      "loss": 0.14,
      "step": 10510
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3403699270904939,
      "learning_rate": 6.494004278218313e-07,
      "loss": 0.171,
      "step": 10511
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5072303195863255,
      "learning_rate": 6.487620985055027e-07,
      "loss": 0.1954,
      "step": 10512
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.378166066652208,
      "learning_rate": 6.481240613009565e-07,
      "loss": 0.188,
      "step": 10513
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.316787863168727,
      "learning_rate": 6.474863162510264e-07,
      "loss": 0.1485,
      "step": 10514
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3513723797005748,
      "learning_rate": 6.468488633985299e-07,
      "loss": 0.1739,
      "step": 10515
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4173785156978218,
      "learning_rate": 6.462117027862569e-07,
      "loss": 0.1672,
      "step": 10516
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.486898221715606,
      "learning_rate": 6.45574834456984e-07,
      "loss": 0.1952,
      "step": 10517
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.366305949854242,
      "learning_rate": 6.449382584534663e-07,
      "loss": 0.5917,
      "step": 10518
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.33096666804939,
      "learning_rate": 6.443019748184392e-07,
      "loss": 0.5946,
      "step": 10519
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5176979379713582,
      "learning_rate": 6.436659835946168e-07,
      "loss": 0.1979,
      "step": 10520
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3500045910213623,
      "learning_rate": 6.430302848246967e-07,
      "loss": 0.1503,
      "step": 10521
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.395576689272778,
      "learning_rate": 6.423948785513545e-07,
      "loss": 0.1524,
      "step": 10522
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5893826037061731,
      "learning_rate": 6.417597648172474e-07,
      "loss": 0.1872,
      "step": 10523
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4268238123990546,
      "learning_rate": 6.411249436650119e-07,
      "loss": 0.1868,
      "step": 10524
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3014838688902484,
      "learning_rate": 6.404904151372649e-07,
      "loss": 0.1324,
      "step": 10525
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2429703358514754,
      "learning_rate": 6.398561792766067e-07,
      "loss": 0.1406,
      "step": 10526
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.605700248752528,
      "learning_rate": 6.392222361256123e-07,
      "loss": 0.4614,
      "step": 10527
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4760903581195748,
      "learning_rate": 6.385885857268431e-07,
      "loss": 0.1495,
      "step": 10528
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3130970035006684,
      "learning_rate": 6.379552281228363e-07,
      "loss": 0.1679,
      "step": 10529
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2695191090714528,
      "learning_rate": 6.373221633561117e-07,
      "loss": 0.1435,
      "step": 10530
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3669620982376145,
      "learning_rate": 6.366893914691685e-07,
      "loss": 0.19,
      "step": 10531
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3939842218772,
      "learning_rate": 6.360569125044869e-07,
      "loss": 0.1902,
      "step": 10532
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5050198271685364,
      "learning_rate": 6.354247265045266e-07,
      "loss": 0.1543,
      "step": 10533
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4002381812318612,
      "learning_rate": 6.347928335117287e-07,
      "loss": 0.1931,
      "step": 10534
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3764659487344357,
      "learning_rate": 6.341612335685143e-07,
      "loss": 0.1705,
      "step": 10535
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2294240787670405,
      "learning_rate": 6.335299267172828e-07,
      "loss": 0.1341,
      "step": 10536
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4846665612558694,
      "learning_rate": 6.328989130004187e-07,
      "loss": 0.182,
      "step": 10537
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3084862438377158,
      "learning_rate": 6.322681924602819e-07,
      "loss": 0.1699,
      "step": 10538
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3900223974804318,
      "learning_rate": 6.316377651392153e-07,
      "loss": 0.1501,
      "step": 10539
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2956409310994568,
      "learning_rate": 6.31007631079541e-07,
      "loss": 0.174,
      "step": 10540
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.244418695242432,
      "learning_rate": 6.30377790323562e-07,
      "loss": 0.1663,
      "step": 10541
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.401353893196756,
      "learning_rate": 6.297482429135615e-07,
      "loss": 0.1799,
      "step": 10542
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.511367684092856,
      "learning_rate": 6.291189888918025e-07,
      "loss": 0.1962,
      "step": 10543
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2924466348207186,
      "learning_rate": 6.284900283005297e-07,
      "loss": 0.1592,
      "step": 10544
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0751484817266148,
      "learning_rate": 6.278613611819645e-07,
      "loss": 0.1377,
      "step": 10545
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2400031467123105,
      "learning_rate": 6.272329875783157e-07,
      "loss": 0.1617,
      "step": 10546
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.444685289841513,
      "learning_rate": 6.266049075317631e-07,
      "loss": 0.2082,
      "step": 10547
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3473433970249304,
      "learning_rate": 6.259771210844751e-07,
      "loss": 0.1719,
      "step": 10548
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3282156815852242,
      "learning_rate": 6.25349628278597e-07,
      "loss": 0.154,
      "step": 10549
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4409660171190206,
      "learning_rate": 6.24722429156251e-07,
      "loss": 0.1743,
      "step": 10550
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3190231016451954,
      "learning_rate": 6.240955237595453e-07,
      "loss": 0.1451,
      "step": 10551
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.16478661642939,
      "learning_rate": 6.234689121305659e-07,
      "loss": 0.105,
      "step": 10552
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2688257054702625,
      "learning_rate": 6.228425943113792e-07,
      "loss": 0.1897,
      "step": 10553
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.539205650900151,
      "learning_rate": 6.222165703440308e-07,
      "loss": 0.1638,
      "step": 10554
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5565763836371498,
      "learning_rate": 6.215908402705484e-07,
      "loss": 0.227,
      "step": 10555
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.999370863016364,
      "learning_rate": 6.209654041329377e-07,
      "loss": 0.5521,
      "step": 10556
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.761255762303853,
      "learning_rate": 6.203402619731897e-07,
      "loss": 0.5766,
      "step": 10557
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.394167469202081,
      "learning_rate": 6.19715413833269e-07,
      "loss": 0.1718,
      "step": 10558
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.24601338639437,
      "learning_rate": 6.190908597551226e-07,
      "loss": 0.6815,
      "step": 10559
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.425094184879317,
      "learning_rate": 6.184665997806832e-07,
      "loss": 0.1733,
      "step": 10560
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4626667873287005,
      "learning_rate": 6.178426339518539e-07,
      "loss": 0.2131,
      "step": 10561
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.82423807772496,
      "learning_rate": 6.172189623105268e-07,
      "loss": 0.5531,
      "step": 10562
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3049301440550842,
      "learning_rate": 6.165955848985705e-07,
      "loss": 0.1763,
      "step": 10563
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.340973208507046,
      "learning_rate": 6.159725017578338e-07,
      "loss": 0.1686,
      "step": 10564
    },
    {
      "epoch": 0.85,
      "grad_norm": 11.119707414439866,
      "learning_rate": 6.153497129301461e-07,
      "loss": 0.7096,
      "step": 10565
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3365318394036592,
      "learning_rate": 6.147272184573172e-07,
      "loss": 0.1779,
      "step": 10566
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5019321988976653,
      "learning_rate": 6.141050183811364e-07,
      "loss": 0.1671,
      "step": 10567
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3240951436671438,
      "learning_rate": 6.134831127433749e-07,
      "loss": 0.1821,
      "step": 10568
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.23077180632172,
      "learning_rate": 6.128615015857831e-07,
      "loss": 0.1432,
      "step": 10569
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3886829029373806,
      "learning_rate": 6.122401849500892e-07,
      "loss": 0.1721,
      "step": 10570
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.331990608184947,
      "learning_rate": 6.116191628780088e-07,
      "loss": 0.1961,
      "step": 10571
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4671516418360893,
      "learning_rate": 6.10998435411228e-07,
      "loss": 0.1741,
      "step": 10572
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.224131425694233,
      "learning_rate": 6.103780025914213e-07,
      "loss": 0.1606,
      "step": 10573
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.311785059706187,
      "learning_rate": 6.097578644602409e-07,
      "loss": 0.1671,
      "step": 10574
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3475492983218158,
      "learning_rate": 6.091380210593145e-07,
      "loss": 0.1629,
      "step": 10575
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4146782064920833,
      "learning_rate": 6.085184724302573e-07,
      "loss": 0.1815,
      "step": 10576
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4794577218910716,
      "learning_rate": 6.07899218614661e-07,
      "loss": 0.1718,
      "step": 10577
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.373963825721224,
      "learning_rate": 6.072802596540978e-07,
      "loss": 0.5084,
      "step": 10578
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2086793299923775,
      "learning_rate": 6.066615955901189e-07,
      "loss": 0.1424,
      "step": 10579
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2952511224664978,
      "learning_rate": 6.060432264642601e-07,
      "loss": 0.1509,
      "step": 10580
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4484589906310303,
      "learning_rate": 6.054251523180316e-07,
      "loss": 0.1896,
      "step": 10581
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.274939998354236,
      "learning_rate": 6.048073731929277e-07,
      "loss": 0.1508,
      "step": 10582
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3822241428217132,
      "learning_rate": 6.041898891304232e-07,
      "loss": 0.1841,
      "step": 10583
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2742229129461664,
      "learning_rate": 6.035727001719682e-07,
      "loss": 0.1957,
      "step": 10584
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.493077854071611,
      "learning_rate": 6.029558063589996e-07,
      "loss": 0.213,
      "step": 10585
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3062933900875577,
      "learning_rate": 6.023392077329298e-07,
      "loss": 0.1351,
      "step": 10586
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.483754633905308,
      "learning_rate": 6.017229043351535e-07,
      "loss": 0.5208,
      "step": 10587
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3166882406360825,
      "learning_rate": 6.011068962070454e-07,
      "loss": 0.1834,
      "step": 10588
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4238748839588746,
      "learning_rate": 6.00491183389959e-07,
      "loss": 0.163,
      "step": 10589
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5839068604098732,
      "learning_rate": 5.998757659252285e-07,
      "loss": 0.2023,
      "step": 10590
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2846631601546437,
      "learning_rate": 5.99260643854172e-07,
      "loss": 0.1294,
      "step": 10591
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4439416956071258,
      "learning_rate": 5.986458172180809e-07,
      "loss": 0.1641,
      "step": 10592
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.46594038114184,
      "learning_rate": 5.980312860582305e-07,
      "loss": 0.1696,
      "step": 10593
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5274450308424767,
      "learning_rate": 5.974170504158794e-07,
      "loss": 0.2228,
      "step": 10594
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.42685855904396,
      "learning_rate": 5.968031103322592e-07,
      "loss": 0.4682,
      "step": 10595
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.192632531109942,
      "learning_rate": 5.961894658485884e-07,
      "loss": 0.149,
      "step": 10596
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.378053807274557,
      "learning_rate": 5.955761170060614e-07,
      "loss": 0.5758,
      "step": 10597
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7145537328717946,
      "learning_rate": 5.949630638458548e-07,
      "loss": 0.2066,
      "step": 10598
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5403380476082977,
      "learning_rate": 5.943503064091243e-07,
      "loss": 0.2104,
      "step": 10599
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.710931006526699,
      "learning_rate": 5.937378447370068e-07,
      "loss": 0.5195,
      "step": 10600
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2420568628435795,
      "learning_rate": 5.931256788706175e-07,
      "loss": 0.1753,
      "step": 10601
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2585527638284753,
      "learning_rate": 5.925138088510541e-07,
      "loss": 0.1762,
      "step": 10602
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4047215880490322,
      "learning_rate": 5.919022347193926e-07,
      "loss": 0.1579,
      "step": 10603
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3449013104744025,
      "learning_rate": 5.912909565166896e-07,
      "loss": 0.1937,
      "step": 10604
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.503198894031086,
      "learning_rate": 5.906799742839842e-07,
      "loss": 0.1948,
      "step": 10605
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.280119902006962,
      "learning_rate": 5.900692880622899e-07,
      "loss": 0.157,
      "step": 10606
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.387600039260772,
      "learning_rate": 5.894588978926069e-07,
      "loss": 0.1813,
      "step": 10607
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4244714486925334,
      "learning_rate": 5.888488038159112e-07,
      "loss": 0.1703,
      "step": 10608
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4603517493620952,
      "learning_rate": 5.882390058731607e-07,
      "loss": 0.1891,
      "step": 10609
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4153686565982728,
      "learning_rate": 5.876295041052932e-07,
      "loss": 0.1939,
      "step": 10610
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3708018147958438,
      "learning_rate": 5.870202985532253e-07,
      "loss": 0.1764,
      "step": 10611
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4293280795843777,
      "learning_rate": 5.86411389257856e-07,
      "loss": 0.1859,
      "step": 10612
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4520204756596706,
      "learning_rate": 5.858027762600632e-07,
      "loss": 0.1685,
      "step": 10613
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.399259892432451,
      "learning_rate": 5.851944596007036e-07,
      "loss": 0.1463,
      "step": 10614
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4839897916603628,
      "learning_rate": 5.845864393206158e-07,
      "loss": 0.1371,
      "step": 10615
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3670438706079497,
      "learning_rate": 5.839787154606208e-07,
      "loss": 0.1513,
      "step": 10616
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4792979586309143,
      "learning_rate": 5.833712880615122e-07,
      "loss": 0.1984,
      "step": 10617
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3932551885664783,
      "learning_rate": 5.827641571640718e-07,
      "loss": 0.1232,
      "step": 10618
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.535431722590136,
      "learning_rate": 5.821573228090582e-07,
      "loss": 0.1622,
      "step": 10619
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.807338066248929,
      "learning_rate": 5.815507850372077e-07,
      "loss": 0.6901,
      "step": 10620
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3897973085916457,
      "learning_rate": 5.809445438892408e-07,
      "loss": 0.1677,
      "step": 10621
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4116097718477323,
      "learning_rate": 5.803385994058558e-07,
      "loss": 0.1614,
      "step": 10622
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.098898525817662,
      "learning_rate": 5.797329516277317e-07,
      "loss": 0.6058,
      "step": 10623
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4215288717268604,
      "learning_rate": 5.791276005955266e-07,
      "loss": 0.189,
      "step": 10624
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3057289802591212,
      "learning_rate": 5.785225463498828e-07,
      "loss": 0.1751,
      "step": 10625
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2724361094558154,
      "learning_rate": 5.779177889314153e-07,
      "loss": 0.1511,
      "step": 10626
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.317483034883664,
      "learning_rate": 5.773133283807242e-07,
      "loss": 0.1921,
      "step": 10627
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6579604976603532,
      "learning_rate": 5.767091647383916e-07,
      "loss": 0.1638,
      "step": 10628
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.361385894563717,
      "learning_rate": 5.761052980449732e-07,
      "loss": 0.1702,
      "step": 10629
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4987005539239486,
      "learning_rate": 5.755017283410102e-07,
      "loss": 0.1954,
      "step": 10630
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2134378179620855,
      "learning_rate": 5.748984556670229e-07,
      "loss": 0.1689,
      "step": 10631
    },
    {
      "epoch": 0.85,
      "grad_norm": 12.173148674523707,
      "learning_rate": 5.742954800635092e-07,
      "loss": 0.4498,
      "step": 10632
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2999402968366802,
      "learning_rate": 5.736928015709492e-07,
      "loss": 0.1632,
      "step": 10633
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.730892312351319,
      "learning_rate": 5.730904202298026e-07,
      "loss": 0.1724,
      "step": 10634
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3053735325753895,
      "learning_rate": 5.724883360805095e-07,
      "loss": 0.1725,
      "step": 10635
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5242452414354464,
      "learning_rate": 5.71886549163489e-07,
      "loss": 0.2201,
      "step": 10636
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4089103139522323,
      "learning_rate": 5.71285059519141e-07,
      "loss": 0.2015,
      "step": 10637
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.340136113570122,
      "learning_rate": 5.70683867187844e-07,
      "loss": 0.1539,
      "step": 10638
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2574858794164834,
      "learning_rate": 5.700829722099621e-07,
      "loss": 0.1654,
      "step": 10639
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2545192468012578,
      "learning_rate": 5.694823746258299e-07,
      "loss": 0.1714,
      "step": 10640
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5946377909347815,
      "learning_rate": 5.688820744757712e-07,
      "loss": 0.1479,
      "step": 10641
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4158074518883674,
      "learning_rate": 5.682820718000842e-07,
      "loss": 0.1772,
      "step": 10642
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2227927017458096,
      "learning_rate": 5.676823666390496e-07,
      "loss": 0.1723,
      "step": 10643
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3296695819257498,
      "learning_rate": 5.670829590329269e-07,
      "loss": 0.1726,
      "step": 10644
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0793434212961626,
      "learning_rate": 5.664838490219565e-07,
      "loss": 0.1197,
      "step": 10645
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.303235710964803,
      "learning_rate": 5.658850366463581e-07,
      "loss": 0.6706,
      "step": 10646
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3196752122690611,
      "learning_rate": 5.652865219463321e-07,
      "loss": 0.1539,
      "step": 10647
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.723741662394361,
      "learning_rate": 5.646883049620589e-07,
      "loss": 0.5579,
      "step": 10648
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.366949607399701,
      "learning_rate": 5.640903857336972e-07,
      "loss": 0.1671,
      "step": 10649
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3584449760063195,
      "learning_rate": 5.634927643013899e-07,
      "loss": 0.1716,
      "step": 10650
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2110692632425746,
      "learning_rate": 5.628954407052534e-07,
      "loss": 0.1502,
      "step": 10651
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3576052282926303,
      "learning_rate": 5.622984149853906e-07,
      "loss": 0.1512,
      "step": 10652
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.55089964573737,
      "learning_rate": 5.617016871818826e-07,
      "loss": 0.7438,
      "step": 10653
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3276497540068664,
      "learning_rate": 5.611052573347852e-07,
      "loss": 0.184,
      "step": 10654
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2540148809381224,
      "learning_rate": 5.605091254841427e-07,
      "loss": 0.1204,
      "step": 10655
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4085778618371645,
      "learning_rate": 5.599132916699735e-07,
      "loss": 0.1504,
      "step": 10656
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3583813078344065,
      "learning_rate": 5.593177559322776e-07,
      "loss": 0.1611,
      "step": 10657
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2822691300659912,
      "learning_rate": 5.587225183110362e-07,
      "loss": 0.1725,
      "step": 10658
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.267253713850386,
      "learning_rate": 5.581275788462081e-07,
      "loss": 0.1907,
      "step": 10659
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5352307580405857,
      "learning_rate": 5.575329375777333e-07,
      "loss": 0.1933,
      "step": 10660
    },
    {
      "epoch": 0.85,
      "grad_norm": 11.157385536059648,
      "learning_rate": 5.569385945455342e-07,
      "loss": 0.665,
      "step": 10661
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.1209071136158135,
      "learning_rate": 5.563445497895087e-07,
      "loss": 0.5758,
      "step": 10662
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5510846194446903,
      "learning_rate": 5.557508033495356e-07,
      "loss": 0.1976,
      "step": 10663
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.56975586978172,
      "learning_rate": 5.551573552654793e-07,
      "loss": 0.1634,
      "step": 10664
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.429251854362159,
      "learning_rate": 5.545642055771749e-07,
      "loss": 0.1736,
      "step": 10665
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4202109241285006,
      "learning_rate": 5.539713543244451e-07,
      "loss": 0.1574,
      "step": 10666
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.844452706750555,
      "learning_rate": 5.533788015470898e-07,
      "loss": 0.6601,
      "step": 10667
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4023188170926022,
      "learning_rate": 5.527865472848881e-07,
      "loss": 0.1577,
      "step": 10668
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3946165271918907,
      "learning_rate": 5.521945915775989e-07,
      "loss": 0.1496,
      "step": 10669
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2168763992378715,
      "learning_rate": 5.516029344649649e-07,
      "loss": 0.1583,
      "step": 10670
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.317322727025469,
      "learning_rate": 5.510115759867035e-07,
      "loss": 0.5213,
      "step": 10671
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4299673563390836,
      "learning_rate": 5.504205161825132e-07,
      "loss": 0.1527,
      "step": 10672
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3929594209464373,
      "learning_rate": 5.498297550920778e-07,
      "loss": 0.1767,
      "step": 10673
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.250503706121312,
      "learning_rate": 5.492392927550527e-07,
      "loss": 0.2001,
      "step": 10674
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5179472152032525,
      "learning_rate": 5.486491292110796e-07,
      "loss": 0.234,
      "step": 10675
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5420003597440808,
      "learning_rate": 5.480592644997778e-07,
      "loss": 0.1893,
      "step": 10676
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5262011594782208,
      "learning_rate": 5.474696986607464e-07,
      "loss": 0.1441,
      "step": 10677
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1783701722251738,
      "learning_rate": 5.468804317335647e-07,
      "loss": 0.1579,
      "step": 10678
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3843769445665046,
      "learning_rate": 5.462914637577921e-07,
      "loss": 0.1807,
      "step": 10679
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4048882880311022,
      "learning_rate": 5.457027947729676e-07,
      "loss": 0.2243,
      "step": 10680
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3927134109780797,
      "learning_rate": 5.451144248186108e-07,
      "loss": 0.1618,
      "step": 10681
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3669259670055096,
      "learning_rate": 5.4452635393422e-07,
      "loss": 0.1763,
      "step": 10682
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2763733629170713,
      "learning_rate": 5.43938582159274e-07,
      "loss": 0.1548,
      "step": 10683
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3822167284937537,
      "learning_rate": 5.433511095332339e-07,
      "loss": 0.1722,
      "step": 10684
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.548060214449014,
      "learning_rate": 5.42763936095535e-07,
      "loss": 0.4781,
      "step": 10685
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.288812261654522,
      "learning_rate": 5.421770618855993e-07,
      "loss": 0.1371,
      "step": 10686
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3876410932880585,
      "learning_rate": 5.415904869428251e-07,
      "loss": 0.1955,
      "step": 10687
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.465154381259502,
      "learning_rate": 5.410042113065877e-07,
      "loss": 0.4814,
      "step": 10688
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3162355070592981,
      "learning_rate": 5.40418235016249e-07,
      "loss": 0.1601,
      "step": 10689
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.161538755306522,
      "learning_rate": 5.398325581111458e-07,
      "loss": 0.534,
      "step": 10690
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2962372775697666,
      "learning_rate": 5.392471806305971e-07,
      "loss": 0.1569,
      "step": 10691
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.855524527558247,
      "learning_rate": 5.386621026139006e-07,
      "loss": 0.5027,
      "step": 10692
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2636364632270607,
      "learning_rate": 5.380773241003345e-07,
      "loss": 0.1601,
      "step": 10693
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2566763585641445,
      "learning_rate": 5.374928451291556e-07,
      "loss": 0.1677,
      "step": 10694
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.497679311190612,
      "learning_rate": 5.36908665739605e-07,
      "loss": 0.2121,
      "step": 10695
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.317051581535928,
      "learning_rate": 5.36324785970897e-07,
      "loss": 0.1797,
      "step": 10696
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3029451211369458,
      "learning_rate": 5.357412058622297e-07,
      "loss": 0.1532,
      "step": 10697
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4293064237889748,
      "learning_rate": 5.351579254527839e-07,
      "loss": 0.184,
      "step": 10698
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2470858045215023,
      "learning_rate": 5.345749447817117e-07,
      "loss": 0.1662,
      "step": 10699
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3635457116660699,
      "learning_rate": 5.339922638881545e-07,
      "loss": 0.1503,
      "step": 10700
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4821456405705027,
      "learning_rate": 5.334098828112283e-07,
      "loss": 0.1714,
      "step": 10701
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3707107659771045,
      "learning_rate": 5.328278015900296e-07,
      "loss": 0.1801,
      "step": 10702
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3654871727784446,
      "learning_rate": 5.32246020263636e-07,
      "loss": 0.1731,
      "step": 10703
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4284620545681295,
      "learning_rate": 5.316645388711033e-07,
      "loss": 0.1595,
      "step": 10704
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4222133609898235,
      "learning_rate": 5.31083357451469e-07,
      "loss": 0.1845,
      "step": 10705
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2589204197288464,
      "learning_rate": 5.305024760437489e-07,
      "loss": 0.1529,
      "step": 10706
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5202601878350592,
      "learning_rate": 5.299218946869394e-07,
      "loss": 0.1956,
      "step": 10707
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.539906503052729,
      "learning_rate": 5.293416134200164e-07,
      "loss": 0.1833,
      "step": 10708
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4676716970709158,
      "learning_rate": 5.287616322819372e-07,
      "loss": 0.2112,
      "step": 10709
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5142322596846975,
      "learning_rate": 5.281819513116371e-07,
      "loss": 0.1984,
      "step": 10710
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2372880966812074,
      "learning_rate": 5.27602570548032e-07,
      "loss": 0.1265,
      "step": 10711
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4559158772102778,
      "learning_rate": 5.270234900300164e-07,
      "loss": 0.2008,
      "step": 10712
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2967967907460891,
      "learning_rate": 5.26444709796467e-07,
      "loss": 0.1492,
      "step": 10713
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.276132377045814,
      "learning_rate": 5.258662298862383e-07,
      "loss": 0.1441,
      "step": 10714
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.188062824533826,
      "learning_rate": 5.252880503381658e-07,
      "loss": 0.168,
      "step": 10715
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4688078496630812,
      "learning_rate": 5.247101711910646e-07,
      "loss": 0.1897,
      "step": 10716
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.519678326359524,
      "learning_rate": 5.241325924837277e-07,
      "loss": 0.1606,
      "step": 10717
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5530452508088857,
      "learning_rate": 5.235553142549338e-07,
      "loss": 0.2072,
      "step": 10718
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3116130786746796,
      "learning_rate": 5.229783365434321e-07,
      "loss": 0.1436,
      "step": 10719
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4182911164374756,
      "learning_rate": 5.22401659387961e-07,
      "loss": 0.1673,
      "step": 10720
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3902968225946,
      "learning_rate": 5.21825282827233e-07,
      "loss": 0.1654,
      "step": 10721
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3444171053538785,
      "learning_rate": 5.212492068999414e-07,
      "loss": 0.1618,
      "step": 10722
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3255262001266808,
      "learning_rate": 5.206734316447615e-07,
      "loss": 0.1678,
      "step": 10723
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3777758351988492,
      "learning_rate": 5.200979571003456e-07,
      "loss": 0.2073,
      "step": 10724
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.323300414159089,
      "learning_rate": 5.195227833053273e-07,
      "loss": 0.1373,
      "step": 10725
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.58545925403752,
      "learning_rate": 5.189479102983203e-07,
      "loss": 0.1887,
      "step": 10726
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2895156186280978,
      "learning_rate": 5.183733381179168e-07,
      "loss": 0.1829,
      "step": 10727
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4082991209466291,
      "learning_rate": 5.177990668026889e-07,
      "loss": 0.2108,
      "step": 10728
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3676593754975992,
      "learning_rate": 5.172250963911923e-07,
      "loss": 0.1914,
      "step": 10729
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3257451261519173,
      "learning_rate": 5.166514269219546e-07,
      "loss": 0.1732,
      "step": 10730
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.266770315363688,
      "learning_rate": 5.160780584334923e-07,
      "loss": 0.1641,
      "step": 10731
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.615752223659782,
      "learning_rate": 5.155049909642967e-07,
      "loss": 0.1984,
      "step": 10732
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4719864971448822,
      "learning_rate": 5.149322245528365e-07,
      "loss": 0.146,
      "step": 10733
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.396251805725455,
      "learning_rate": 5.14359759237566e-07,
      "loss": 0.4866,
      "step": 10734
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.323175011068097,
      "learning_rate": 5.13787595056916e-07,
      "loss": 0.1254,
      "step": 10735
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5753289130588009,
      "learning_rate": 5.132157320492975e-07,
      "loss": 0.1731,
      "step": 10736
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4173567935064983,
      "learning_rate": 5.12644170253101e-07,
      "loss": 0.183,
      "step": 10737
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4245674365390482,
      "learning_rate": 5.120729097066978e-07,
      "loss": 0.148,
      "step": 10738
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3092688966148205,
      "learning_rate": 5.11501950448437e-07,
      "loss": 0.1524,
      "step": 10739
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3343046111134684,
      "learning_rate": 5.10931292516652e-07,
      "loss": 0.1505,
      "step": 10740
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4272431085704738,
      "learning_rate": 5.103609359496492e-07,
      "loss": 0.1678,
      "step": 10741
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.38841904232489,
      "learning_rate": 5.097908807857193e-07,
      "loss": 0.1772,
      "step": 10742
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3657522748499518,
      "learning_rate": 5.092211270631343e-07,
      "loss": 0.21,
      "step": 10743
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.596545760535399,
      "learning_rate": 5.086516748201393e-07,
      "loss": 0.4536,
      "step": 10744
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.320915279745972,
      "learning_rate": 5.080825240949672e-07,
      "loss": 0.1655,
      "step": 10745
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4090026460298837,
      "learning_rate": 5.075136749258248e-07,
      "loss": 0.1999,
      "step": 10746
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.404031127141463,
      "learning_rate": 5.069451273509007e-07,
      "loss": 0.5957,
      "step": 10747
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4299705284045514,
      "learning_rate": 5.063768814083641e-07,
      "loss": 0.1968,
      "step": 10748
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6480755634554847,
      "learning_rate": 5.058089371363628e-07,
      "loss": 0.2095,
      "step": 10749
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2079254220657192,
      "learning_rate": 5.05241294573024e-07,
      "loss": 0.1613,
      "step": 10750
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2062723463067426,
      "learning_rate": 5.046739537564549e-07,
      "loss": 0.1426,
      "step": 10751
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3430996595014446,
      "learning_rate": 5.041069147247452e-07,
      "loss": 0.1653,
      "step": 10752
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4539651543284005,
      "learning_rate": 5.035401775159588e-07,
      "loss": 0.1716,
      "step": 10753
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2267534771353652,
      "learning_rate": 5.029737421681446e-07,
      "loss": 0.1503,
      "step": 10754
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.31927656464137,
      "learning_rate": 5.024076087193292e-07,
      "loss": 0.1661,
      "step": 10755
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3023593501530626,
      "learning_rate": 5.018417772075174e-07,
      "loss": 0.1449,
      "step": 10756
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3119511005733342,
      "learning_rate": 5.012762476706961e-07,
      "loss": 0.1639,
      "step": 10757
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.397542959684994,
      "learning_rate": 5.007110201468312e-07,
      "loss": 0.4112,
      "step": 10758
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.532501705355364,
      "learning_rate": 5.001460946738679e-07,
      "loss": 0.1848,
      "step": 10759
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.450439442903786,
      "learning_rate": 4.995814712897312e-07,
      "loss": 0.1909,
      "step": 10760
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2524635849564814,
      "learning_rate": 4.99017150032326e-07,
      "loss": 0.1929,
      "step": 10761
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.21217028037563,
      "learning_rate": 4.984531309395357e-07,
      "loss": 0.1267,
      "step": 10762
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4286413043166686,
      "learning_rate": 4.978894140492279e-07,
      "loss": 0.1704,
      "step": 10763
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2487583103839746,
      "learning_rate": 4.97325999399243e-07,
      "loss": 0.1367,
      "step": 10764
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.302547225535744,
      "learning_rate": 4.967628870274071e-07,
      "loss": 0.148,
      "step": 10765
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3612114656572132,
      "learning_rate": 4.962000769715236e-07,
      "loss": 0.1576,
      "step": 10766
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.416346413575944,
      "learning_rate": 4.956375692693732e-07,
      "loss": 0.1676,
      "step": 10767
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.424282498461266,
      "learning_rate": 4.950753639587214e-07,
      "loss": 0.1827,
      "step": 10768
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3497385518522562,
      "learning_rate": 4.9451346107731e-07,
      "loss": 0.1687,
      "step": 10769
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.228783537633953,
      "learning_rate": 4.939518606628607e-07,
      "loss": 0.5702,
      "step": 10770
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3061232290681442,
      "learning_rate": 4.933905627530761e-07,
      "loss": 0.1269,
      "step": 10771
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1823576704238843,
      "learning_rate": 4.92829567385637e-07,
      "loss": 0.1197,
      "step": 10772
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3525085853428738,
      "learning_rate": 4.922688745982046e-07,
      "loss": 0.1476,
      "step": 10773
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.473038333807972,
      "learning_rate": 4.917084844284225e-07,
      "loss": 0.1819,
      "step": 10774
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2857561415888838,
      "learning_rate": 4.911483969139086e-07,
      "loss": 0.1668,
      "step": 10775
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4338573692401042,
      "learning_rate": 4.905886120922626e-07,
      "loss": 0.1987,
      "step": 10776
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.439769701814589,
      "learning_rate": 4.900291300010684e-07,
      "loss": 0.1822,
      "step": 10777
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.639885417213558,
      "learning_rate": 4.894699506778816e-07,
      "loss": 0.1923,
      "step": 10778
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2570515971613156,
      "learning_rate": 4.889110741602437e-07,
      "loss": 0.1821,
      "step": 10779
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3190793801263405,
      "learning_rate": 4.883525004856738e-07,
      "loss": 0.1588,
      "step": 10780
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2685335737114225,
      "learning_rate": 4.877942296916704e-07,
      "loss": 0.1568,
      "step": 10781
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3179934732928813,
      "learning_rate": 4.872362618157117e-07,
      "loss": 0.1878,
      "step": 10782
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2987883971390568,
      "learning_rate": 4.866785968952559e-07,
      "loss": 0.1501,
      "step": 10783
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5489665768860346,
      "learning_rate": 4.861212349677408e-07,
      "loss": 0.1947,
      "step": 10784
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4832335057118036,
      "learning_rate": 4.855641760705837e-07,
      "loss": 0.1692,
      "step": 10785
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4413926151169387,
      "learning_rate": 4.850074202411814e-07,
      "loss": 0.2114,
      "step": 10786
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.373933697097091,
      "learning_rate": 4.844509675169101e-07,
      "loss": 0.1594,
      "step": 10787
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.525182960616275,
      "learning_rate": 4.838948179351288e-07,
      "loss": 0.1774,
      "step": 10788
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2897896412497687,
      "learning_rate": 4.833389715331699e-07,
      "loss": 0.1445,
      "step": 10789
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.37918614791825,
      "learning_rate": 4.827834283483513e-07,
      "loss": 0.1998,
      "step": 10790
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.313756174988163,
      "learning_rate": 4.822281884179681e-07,
      "loss": 0.4548,
      "step": 10791
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1670251074380418,
      "learning_rate": 4.816732517792949e-07,
      "loss": 0.149,
      "step": 10792
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2564124482055126,
      "learning_rate": 4.81118618469586e-07,
      "loss": 0.1368,
      "step": 10793
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2466733723568582,
      "learning_rate": 4.805642885260764e-07,
      "loss": 0.1549,
      "step": 10794
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4223359751903752,
      "learning_rate": 4.800102619859792e-07,
      "loss": 0.1387,
      "step": 10795
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3861661837478807,
      "learning_rate": 4.794565388864869e-07,
      "loss": 0.1563,
      "step": 10796
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.467508216815393,
      "learning_rate": 4.789031192647758e-07,
      "loss": 0.1932,
      "step": 10797
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4063786463582348,
      "learning_rate": 4.783500031579952e-07,
      "loss": 0.1854,
      "step": 10798
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3966460858230294,
      "learning_rate": 4.777971906032797e-07,
      "loss": 0.1785,
      "step": 10799
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5076328423209329,
      "learning_rate": 4.772446816377408e-07,
      "loss": 0.2023,
      "step": 10800
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.975669748014996,
      "learning_rate": 4.7669247629846926e-07,
      "loss": 0.4406,
      "step": 10801
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.270288084306112,
      "learning_rate": 4.761405746225373e-07,
      "loss": 0.1589,
      "step": 10802
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5511397758411156,
      "learning_rate": 4.755889766469951e-07,
      "loss": 0.2187,
      "step": 10803
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1754300006033265,
      "learning_rate": 4.7503768240887313e-07,
      "loss": 0.1288,
      "step": 10804
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3361724472095928,
      "learning_rate": 4.744866919451824e-07,
      "loss": 0.1438,
      "step": 10805
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.382646187136116,
      "learning_rate": 4.7393600529291096e-07,
      "loss": 0.5683,
      "step": 10806
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.525595422318339,
      "learning_rate": 4.733856224890282e-07,
      "loss": 0.1768,
      "step": 10807
    },
    {
      "epoch": 0.86,
      "grad_norm": 11.818702062891521,
      "learning_rate": 4.7283554357048566e-07,
      "loss": 0.629,
      "step": 10808
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.557497888484463,
      "learning_rate": 4.7228576857420883e-07,
      "loss": 0.6591,
      "step": 10809
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3753639150594845,
      "learning_rate": 4.7173629753710595e-07,
      "loss": 0.1187,
      "step": 10810
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.49871510291026,
      "learning_rate": 4.711871304960675e-07,
      "loss": 0.1672,
      "step": 10811
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.297628754275986,
      "learning_rate": 4.706382674879567e-07,
      "loss": 0.1963,
      "step": 10812
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4704846987786728,
      "learning_rate": 4.7008970854962363e-07,
      "loss": 0.2076,
      "step": 10813
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.184003159461943,
      "learning_rate": 4.695414537178933e-07,
      "loss": 0.1754,
      "step": 10814
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2868423575961738,
      "learning_rate": 4.689935030295717e-07,
      "loss": 0.1578,
      "step": 10815
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3664872436691053,
      "learning_rate": 4.68445856521445e-07,
      "loss": 0.1766,
      "step": 10816
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.415791641927633,
      "learning_rate": 4.678985142302778e-07,
      "loss": 0.1794,
      "step": 10817
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4222051500911845,
      "learning_rate": 4.6735147619281495e-07,
      "loss": 0.1641,
      "step": 10818
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2977619644460066,
      "learning_rate": 4.668047424457811e-07,
      "loss": 0.1562,
      "step": 10819
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3747404818840225,
      "learning_rate": 4.662583130258796e-07,
      "loss": 0.1599,
      "step": 10820
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.49567819586526,
      "learning_rate": 4.65712187969794e-07,
      "loss": 0.1815,
      "step": 10821
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.259701534891147,
      "learning_rate": 4.6516636731418875e-07,
      "loss": 0.1601,
      "step": 10822
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6270672832231368,
      "learning_rate": 4.64620851095704e-07,
      "loss": 0.1758,
      "step": 10823
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2823394066127363,
      "learning_rate": 4.640756393509638e-07,
      "loss": 0.1665,
      "step": 10824
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.293474583669807,
      "learning_rate": 4.6353073211656886e-07,
      "loss": 0.1854,
      "step": 10825
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5014117271963903,
      "learning_rate": 4.6298612942910105e-07,
      "loss": 0.2134,
      "step": 10826
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4169576569551785,
      "learning_rate": 4.624418313251211e-07,
      "loss": 0.1654,
      "step": 10827
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4653698053638866,
      "learning_rate": 4.6189783784116927e-07,
      "loss": 0.1809,
      "step": 10828
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2649154153665856,
      "learning_rate": 4.613541490137657e-07,
      "loss": 0.162,
      "step": 10829
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3698287618284197,
      "learning_rate": 4.608107648794091e-07,
      "loss": 0.183,
      "step": 10830
    },
    {
      "epoch": 0.87,
      "grad_norm": 14.837131370819016,
      "learning_rate": 4.602676854745797e-07,
      "loss": 0.5171,
      "step": 10831
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3555949438615669,
      "learning_rate": 4.597249108357338e-07,
      "loss": 0.172,
      "step": 10832
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3235932939104549,
      "learning_rate": 4.591824409993134e-07,
      "loss": 0.1585,
      "step": 10833
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2457131580146856,
      "learning_rate": 4.586402760017322e-07,
      "loss": 0.162,
      "step": 10834
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.591793963448041,
      "learning_rate": 4.580984158793894e-07,
      "loss": 0.4505,
      "step": 10835
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.229849529708539,
      "learning_rate": 4.5755686066866313e-07,
      "loss": 0.1367,
      "step": 10836
    },
    {
      "epoch": 0.87,
      "grad_norm": 15.262851854442323,
      "learning_rate": 4.5701561040590534e-07,
      "loss": 0.408,
      "step": 10837
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.371548020655865,
      "learning_rate": 4.564746651274554e-07,
      "loss": 0.6073,
      "step": 10838
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3818505976355937,
      "learning_rate": 4.5593402486962757e-07,
      "loss": 0.1495,
      "step": 10839
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.319194960967024,
      "learning_rate": 4.553936896687161e-07,
      "loss": 0.1602,
      "step": 10840
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5173042284823384,
      "learning_rate": 4.548536595609959e-07,
      "loss": 0.1856,
      "step": 10841
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2479264111781918,
      "learning_rate": 4.543139345827219e-07,
      "loss": 0.1477,
      "step": 10842
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4781360673786108,
      "learning_rate": 4.5377451477012446e-07,
      "loss": 0.1487,
      "step": 10843
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.714765408001884,
      "learning_rate": 4.532354001594197e-07,
      "loss": 0.7003,
      "step": 10844
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5462426691874387,
      "learning_rate": 4.5269659078679973e-07,
      "loss": 0.216,
      "step": 10845
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3214280166045052,
      "learning_rate": 4.5215808668843284e-07,
      "loss": 0.1818,
      "step": 10846
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5339132592964004,
      "learning_rate": 4.516198879004746e-07,
      "loss": 0.1742,
      "step": 10847
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3870471557520077,
      "learning_rate": 4.5108199445905387e-07,
      "loss": 0.1821,
      "step": 10848
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3420032327808804,
      "learning_rate": 4.5054440640028173e-07,
      "loss": 0.1571,
      "step": 10849
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4189815382376711,
      "learning_rate": 4.5000712376024826e-07,
      "loss": 0.1713,
      "step": 10850
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5888482292040853,
      "learning_rate": 4.494701465750217e-07,
      "loss": 0.2116,
      "step": 10851
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3505876452660421,
      "learning_rate": 4.489334748806518e-07,
      "loss": 0.1467,
      "step": 10852
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.350048776561533,
      "learning_rate": 4.4839710871316776e-07,
      "loss": 0.6005,
      "step": 10853
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.127126537109826,
      "learning_rate": 4.478610481085766e-07,
      "loss": 0.1335,
      "step": 10854
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.490094509043298,
      "learning_rate": 4.473252931028643e-07,
      "loss": 0.1862,
      "step": 10855
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4997920218893166,
      "learning_rate": 4.4678984373200117e-07,
      "loss": 0.1784,
      "step": 10856
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.557147374164674,
      "learning_rate": 4.462547000319295e-07,
      "loss": 0.2157,
      "step": 10857
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3226962037582275,
      "learning_rate": 4.4571986203857777e-07,
      "loss": 0.1489,
      "step": 10858
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3237092307333724,
      "learning_rate": 4.451853297878511e-07,
      "loss": 0.1453,
      "step": 10859
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2593725537732356,
      "learning_rate": 4.446511033156337e-07,
      "loss": 0.1249,
      "step": 10860
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2740839148635874,
      "learning_rate": 4.4411718265778946e-07,
      "loss": 0.1708,
      "step": 10861
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.560343060647224,
      "learning_rate": 4.435835678501632e-07,
      "loss": 0.1947,
      "step": 10862
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.414233442714087,
      "learning_rate": 4.430502589285768e-07,
      "loss": 0.1787,
      "step": 10863
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.240280994101839,
      "learning_rate": 4.4251725592883374e-07,
      "loss": 0.1377,
      "step": 10864
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4033701362237134,
      "learning_rate": 4.419845588867161e-07,
      "loss": 0.172,
      "step": 10865
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.216382774076751,
      "learning_rate": 4.4145216783798404e-07,
      "loss": 0.1315,
      "step": 10866
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3779752815586535,
      "learning_rate": 4.4092008281838183e-07,
      "loss": 0.1866,
      "step": 10867
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4186357924391155,
      "learning_rate": 4.403883038636264e-07,
      "loss": 0.2068,
      "step": 10868
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4300458050000626,
      "learning_rate": 4.3985683100941976e-07,
      "loss": 0.189,
      "step": 10869
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4693112917705078,
      "learning_rate": 4.3932566429144175e-07,
      "loss": 0.2197,
      "step": 10870
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2472156478581446,
      "learning_rate": 4.387948037453482e-07,
      "loss": 0.1389,
      "step": 10871
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3319265849427564,
      "learning_rate": 4.382642494067801e-07,
      "loss": 0.1287,
      "step": 10872
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3022825219310923,
      "learning_rate": 4.377340013113551e-07,
      "loss": 0.1584,
      "step": 10873
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3414097197806183,
      "learning_rate": 4.372040594946692e-07,
      "loss": 0.1943,
      "step": 10874
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5054932744572576,
      "learning_rate": 4.3667442399229985e-07,
      "loss": 0.2043,
      "step": 10875
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3250410655203841,
      "learning_rate": 4.361450948398027e-07,
      "loss": 0.1562,
      "step": 10876
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2595182993460818,
      "learning_rate": 4.356160720727121e-07,
      "loss": 0.146,
      "step": 10877
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2216303329216809,
      "learning_rate": 4.3508735572654505e-07,
      "loss": 0.1702,
      "step": 10878
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.302197874130371,
      "learning_rate": 4.3455894583679606e-07,
      "loss": 0.1535,
      "step": 10879
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4700381280069024,
      "learning_rate": 4.3403084243893556e-07,
      "loss": 0.1799,
      "step": 10880
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8225266585250548,
      "learning_rate": 4.3350304556842024e-07,
      "loss": 0.1535,
      "step": 10881
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3879002615969367,
      "learning_rate": 4.3297555526068124e-07,
      "loss": 0.1784,
      "step": 10882
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3632825827868034,
      "learning_rate": 4.3244837155113073e-07,
      "loss": 0.1903,
      "step": 10883
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3470137072999415,
      "learning_rate": 4.3192149447516043e-07,
      "loss": 0.1692,
      "step": 10884
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.981848215763743,
      "learning_rate": 4.3139492406814086e-07,
      "loss": 0.2152,
      "step": 10885
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3341174056670826,
      "learning_rate": 4.308686603654211e-07,
      "loss": 0.1705,
      "step": 10886
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.6414093777774434,
      "learning_rate": 4.3034270340233444e-07,
      "loss": 0.5674,
      "step": 10887
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8388214873263853,
      "learning_rate": 4.2981705321418596e-07,
      "loss": 0.1773,
      "step": 10888
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3487742879912015,
      "learning_rate": 4.292917098362653e-07,
      "loss": 0.1644,
      "step": 10889
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4780184186332737,
      "learning_rate": 4.2876667330384315e-07,
      "loss": 0.1746,
      "step": 10890
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.473981077003753,
      "learning_rate": 4.282419436521618e-07,
      "loss": 0.4217,
      "step": 10891
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.423379379268459,
      "learning_rate": 4.2771752091645246e-07,
      "loss": 0.2166,
      "step": 10892
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4533622077121109,
      "learning_rate": 4.271934051319188e-07,
      "loss": 0.1994,
      "step": 10893
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.517224369234004,
      "learning_rate": 4.2666959633374693e-07,
      "loss": 0.509,
      "step": 10894
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.727207902450283,
      "learning_rate": 4.261460945571017e-07,
      "loss": 0.4944,
      "step": 10895
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3708848107942913,
      "learning_rate": 4.256228998371276e-07,
      "loss": 0.1584,
      "step": 10896
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5022227487567108,
      "learning_rate": 4.251000122089477e-07,
      "loss": 0.1804,
      "step": 10897
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3010065062471772,
      "learning_rate": 4.2457743170766565e-07,
      "loss": 0.1498,
      "step": 10898
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2065311568646544,
      "learning_rate": 4.2405515836836384e-07,
      "loss": 0.1544,
      "step": 10899
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.345209053360188,
      "learning_rate": 4.2353319222610265e-07,
      "loss": 0.5977,
      "step": 10900
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.489633700460956,
      "learning_rate": 4.2301153331592627e-07,
      "loss": 0.1559,
      "step": 10901
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2685997682648158,
      "learning_rate": 4.2249018167285104e-07,
      "loss": 0.1727,
      "step": 10902
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3209471690839587,
      "learning_rate": 4.219691373318807e-07,
      "loss": 0.1632,
      "step": 10903
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4165597418682605,
      "learning_rate": 4.2144840032799283e-07,
      "loss": 0.1975,
      "step": 10904
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5326772159473,
      "learning_rate": 4.2092797069614667e-07,
      "loss": 0.1728,
      "step": 10905
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3957537367840924,
      "learning_rate": 4.2040784847127925e-07,
      "loss": 0.1991,
      "step": 10906
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3673420129629266,
      "learning_rate": 4.1988803368830934e-07,
      "loss": 0.1513,
      "step": 10907
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2924918648047796,
      "learning_rate": 4.193685263821323e-07,
      "loss": 0.1297,
      "step": 10908
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5949097165949266,
      "learning_rate": 4.1884932658762536e-07,
      "loss": 0.1897,
      "step": 10909
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.67150128914668,
      "learning_rate": 4.183304343396427e-07,
      "loss": 0.2001,
      "step": 10910
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7539800429875156,
      "learning_rate": 4.1781184967301934e-07,
      "loss": 0.2051,
      "step": 10911
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4904017167281327,
      "learning_rate": 4.1729357262257185e-07,
      "loss": 0.1808,
      "step": 10912
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2448141904210848,
      "learning_rate": 4.167756032230902e-07,
      "loss": 0.1479,
      "step": 10913
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.411363680582119,
      "learning_rate": 4.1625794150934986e-07,
      "loss": 0.1636,
      "step": 10914
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.438723899241405,
      "learning_rate": 4.15740587516103e-07,
      "loss": 0.1894,
      "step": 10915
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3504406299230292,
      "learning_rate": 4.1522354127807807e-07,
      "loss": 0.1583,
      "step": 10916
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4667402732974089,
      "learning_rate": 4.147068028299894e-07,
      "loss": 0.2029,
      "step": 10917
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5541346607212239,
      "learning_rate": 4.14190372206526e-07,
      "loss": 0.2054,
      "step": 10918
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.0853948355321,
      "learning_rate": 4.136742494423573e-07,
      "loss": 0.5319,
      "step": 10919
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5027997779344031,
      "learning_rate": 4.131584345721312e-07,
      "loss": 0.2093,
      "step": 10920
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3504262149292312,
      "learning_rate": 4.1264292763047886e-07,
      "loss": 0.169,
      "step": 10921
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5029131287446185,
      "learning_rate": 4.1212772865200365e-07,
      "loss": 0.168,
      "step": 10922
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.379462950587084,
      "learning_rate": 4.1161283767129636e-07,
      "loss": 0.161,
      "step": 10923
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2674442395480743,
      "learning_rate": 4.1109825472292196e-07,
      "loss": 0.1572,
      "step": 10924
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4053414716144796,
      "learning_rate": 4.1058397984142405e-07,
      "loss": 0.185,
      "step": 10925
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.362549981955955,
      "learning_rate": 4.1007001306132943e-07,
      "loss": 0.1858,
      "step": 10926
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.437342282821647,
      "learning_rate": 4.0955635441714216e-07,
      "loss": 0.1721,
      "step": 10927
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3327197547260632,
      "learning_rate": 4.0904300394334515e-07,
      "loss": 0.1596,
      "step": 10928
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.399636838110795,
      "learning_rate": 4.0852996167440153e-07,
      "loss": 0.1865,
      "step": 10929
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3132224334418692,
      "learning_rate": 4.0801722764475303e-07,
      "loss": 0.1499,
      "step": 10930
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3943536674854309,
      "learning_rate": 4.075048018888211e-07,
      "loss": 0.1754,
      "step": 10931
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3014529381268565,
      "learning_rate": 4.069926844410066e-07,
      "loss": 0.1609,
      "step": 10932
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.430002171585103,
      "learning_rate": 4.0648087533568913e-07,
      "loss": 0.1678,
      "step": 10933
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.85925357905726,
      "learning_rate": 4.0596937460722795e-07,
      "loss": 0.5752,
      "step": 10934
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5819819719618007,
      "learning_rate": 4.0545818228996336e-07,
      "loss": 0.225,
      "step": 10935
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5142029765709426,
      "learning_rate": 4.049472984182107e-07,
      "loss": 0.1638,
      "step": 10936
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3899194760551206,
      "learning_rate": 4.0443672302626857e-07,
      "loss": 0.1684,
      "step": 10937
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3746808389307428,
      "learning_rate": 4.039264561484135e-07,
      "loss": 0.1997,
      "step": 10938
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3135159870906599,
      "learning_rate": 4.0341649781890144e-07,
      "loss": 0.1478,
      "step": 10939
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3534165482166853,
      "learning_rate": 4.0290684807196667e-07,
      "loss": 0.1611,
      "step": 10940
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3825277015934818,
      "learning_rate": 4.0239750694182346e-07,
      "loss": 0.1515,
      "step": 10941
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4366822210585461,
      "learning_rate": 4.018884744626661e-07,
      "loss": 0.1666,
      "step": 10942
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2599282634047564,
      "learning_rate": 4.013797506686673e-07,
      "loss": 0.1571,
      "step": 10943
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3949266783620244,
      "learning_rate": 4.008713355939786e-07,
      "loss": 0.1468,
      "step": 10944
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3526273193698493,
      "learning_rate": 4.003632292727316e-07,
      "loss": 0.1822,
      "step": 10945
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.417770019203867,
      "learning_rate": 3.998554317390385e-07,
      "loss": 0.6612,
      "step": 10946
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4436500361770703,
      "learning_rate": 3.99347943026987e-07,
      "loss": 0.1624,
      "step": 10947
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5919952013677723,
      "learning_rate": 3.9884076317064813e-07,
      "loss": 0.1585,
      "step": 10948
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3413592588917502,
      "learning_rate": 3.983338922040708e-07,
      "loss": 0.1771,
      "step": 10949
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3399280680857606,
      "learning_rate": 3.9782733016128006e-07,
      "loss": 0.1822,
      "step": 10950
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4227833007043245,
      "learning_rate": 3.973210770762853e-07,
      "loss": 0.1545,
      "step": 10951
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2689849886012383,
      "learning_rate": 3.9681513298307216e-07,
      "loss": 0.1341,
      "step": 10952
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.353718525355096,
      "learning_rate": 3.963094979156068e-07,
      "loss": 0.1858,
      "step": 10953
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.207401303646797,
      "learning_rate": 3.9580417190783316e-07,
      "loss": 0.162,
      "step": 10954
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2977639500196645,
      "learning_rate": 3.952991549936752e-07,
      "loss": 0.1709,
      "step": 10955
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.622246039667715,
      "learning_rate": 3.947944472070364e-07,
      "loss": 0.5607,
      "step": 10956
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3724633708381222,
      "learning_rate": 3.9429004858180076e-07,
      "loss": 0.1379,
      "step": 10957
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.342505536913543,
      "learning_rate": 3.9378595915182837e-07,
      "loss": 0.1543,
      "step": 10958
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2877394075074835,
      "learning_rate": 3.9328217895095997e-07,
      "loss": 0.1452,
      "step": 10959
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2962800859117072,
      "learning_rate": 3.9277870801301855e-07,
      "loss": 0.1679,
      "step": 10960
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4926300779522514,
      "learning_rate": 3.9227554637179977e-07,
      "loss": 0.1926,
      "step": 10961
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.457895022612262,
      "learning_rate": 3.9177269406108553e-07,
      "loss": 0.19,
      "step": 10962
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2102066230924624,
      "learning_rate": 3.912701511146322e-07,
      "loss": 0.1324,
      "step": 10963
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.381145396280551,
      "learning_rate": 3.907679175661783e-07,
      "loss": 0.149,
      "step": 10964
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4051858911562365,
      "learning_rate": 3.9026599344943794e-07,
      "loss": 0.1831,
      "step": 10965
    },
    {
      "epoch": 0.88,
      "grad_norm": 10.409093828637229,
      "learning_rate": 3.8976437879811035e-07,
      "loss": 0.5226,
      "step": 10966
    },
    {
      "epoch": 0.88,
      "grad_norm": 10.291491207008816,
      "learning_rate": 3.8926307364586745e-07,
      "loss": 0.5507,
      "step": 10967
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4195971334040238,
      "learning_rate": 3.88762078026364e-07,
      "loss": 0.1792,
      "step": 10968
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5042481847713542,
      "learning_rate": 3.882613919732353e-07,
      "loss": 0.2042,
      "step": 10969
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2757365404604653,
      "learning_rate": 3.877610155200906e-07,
      "loss": 0.1368,
      "step": 10970
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3993723628481347,
      "learning_rate": 3.8726094870052413e-07,
      "loss": 0.1566,
      "step": 10971
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2711664213883518,
      "learning_rate": 3.867611915481062e-07,
      "loss": 0.1828,
      "step": 10972
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3470967548281911,
      "learning_rate": 3.8626174409638726e-07,
      "loss": 0.1787,
      "step": 10973
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3822747247011236,
      "learning_rate": 3.8576260637889664e-07,
      "loss": 0.1679,
      "step": 10974
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5288873216807632,
      "learning_rate": 3.852637784291424e-07,
      "loss": 0.1669,
      "step": 10975
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.469797136464258,
      "learning_rate": 3.847652602806129e-07,
      "loss": 0.2143,
      "step": 10976
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.374641653582868,
      "learning_rate": 3.8426705196677517e-07,
      "loss": 0.1727,
      "step": 10977
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4824597654974832,
      "learning_rate": 3.837691535210758e-07,
      "loss": 0.188,
      "step": 10978
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.301449845871071,
      "learning_rate": 3.832715649769381e-07,
      "loss": 0.1682,
      "step": 10979
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2392884380458786,
      "learning_rate": 3.827742863677708e-07,
      "loss": 0.1705,
      "step": 10980
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.395701858577252,
      "learning_rate": 3.8227731772695344e-07,
      "loss": 0.16,
      "step": 10981
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3329215481184697,
      "learning_rate": 3.8178065908785143e-07,
      "loss": 0.1438,
      "step": 10982
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.789037380138889,
      "learning_rate": 3.8128431048380653e-07,
      "loss": 0.5914,
      "step": 10983
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4309208445004962,
      "learning_rate": 3.807882719481398e-07,
      "loss": 0.1536,
      "step": 10984
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3788356746513588,
      "learning_rate": 3.802925435141525e-07,
      "loss": 0.1615,
      "step": 10985
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4389441480773226,
      "learning_rate": 3.797971252151239e-07,
      "loss": 0.1565,
      "step": 10986
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3997453481955369,
      "learning_rate": 3.793020170843126e-07,
      "loss": 0.2511,
      "step": 10987
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3211223075899494,
      "learning_rate": 3.7880721915495744e-07,
      "loss": 0.1345,
      "step": 10988
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.8411297044490285,
      "learning_rate": 3.783127314602747e-07,
      "loss": 0.5914,
      "step": 10989
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3990802706992915,
      "learning_rate": 3.778185540334611e-07,
      "loss": 0.1615,
      "step": 10990
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5766375201593792,
      "learning_rate": 3.7732468690769406e-07,
      "loss": 0.2197,
      "step": 10991
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3719197011153383,
      "learning_rate": 3.7683113011612593e-07,
      "loss": 0.1848,
      "step": 10992
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7242142893124688,
      "learning_rate": 3.763378836918918e-07,
      "loss": 0.1884,
      "step": 10993
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3563799719243221,
      "learning_rate": 3.758449476681059e-07,
      "loss": 0.171,
      "step": 10994
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5387660926425042,
      "learning_rate": 3.753523220778571e-07,
      "loss": 0.1997,
      "step": 10995
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2170233167367548,
      "learning_rate": 3.748600069542202e-07,
      "loss": 0.1321,
      "step": 10996
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5861987192498954,
      "learning_rate": 3.7436800233024485e-07,
      "loss": 0.199,
      "step": 10997
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4837344927632734,
      "learning_rate": 3.7387630823896013e-07,
      "loss": 0.1934,
      "step": 10998
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.220479886428112,
      "learning_rate": 3.7338492471337585e-07,
      "loss": 0.1544,
      "step": 10999
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3230158311643934,
      "learning_rate": 3.728938517864794e-07,
      "loss": 0.1988,
      "step": 11000
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.317205359221635,
      "learning_rate": 3.7240308949123835e-07,
      "loss": 0.1909,
      "step": 11001
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.36334413043963,
      "learning_rate": 3.7191263786059917e-07,
      "loss": 0.1745,
      "step": 11002
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.271028503771703,
      "learning_rate": 3.7142249692748713e-07,
      "loss": 0.1783,
      "step": 11003
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4745772432840127,
      "learning_rate": 3.7093266672480586e-07,
      "loss": 0.1851,
      "step": 11004
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3262805059430038,
      "learning_rate": 3.7044314728544196e-07,
      "loss": 0.1559,
      "step": 11005
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5003210951088684,
      "learning_rate": 3.6995393864225457e-07,
      "loss": 0.1955,
      "step": 11006
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4049914377776864,
      "learning_rate": 3.694650408280892e-07,
      "loss": 0.181,
      "step": 11007
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3088945212744225,
      "learning_rate": 3.689764538757651e-07,
      "loss": 0.1968,
      "step": 11008
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5727815688238391,
      "learning_rate": 3.684881778180832e-07,
      "loss": 0.176,
      "step": 11009
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.464961185025448,
      "learning_rate": 3.6800021268782293e-07,
      "loss": 0.7439,
      "step": 11010
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3276180684554897,
      "learning_rate": 3.67512558517743e-07,
      "loss": 0.1547,
      "step": 11011
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4771211394590342,
      "learning_rate": 3.6702521534058065e-07,
      "loss": 0.1764,
      "step": 11012
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1966089048599553,
      "learning_rate": 3.665381831890519e-07,
      "loss": 0.1559,
      "step": 11013
    },
    {
      "epoch": 0.88,
      "grad_norm": 44.03278836023848,
      "learning_rate": 3.6605146209585605e-07,
      "loss": 0.5717,
      "step": 11014
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2491527112479215,
      "learning_rate": 3.655650520936638e-07,
      "loss": 0.159,
      "step": 11015
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4174246236840933,
      "learning_rate": 3.650789532151322e-07,
      "loss": 0.1686,
      "step": 11016
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2616323945758376,
      "learning_rate": 3.645931654928936e-07,
      "loss": 0.1766,
      "step": 11017
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.385534418225101,
      "learning_rate": 3.641076889595607e-07,
      "loss": 0.1856,
      "step": 11018
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.217597760212756,
      "learning_rate": 3.636225236477248e-07,
      "loss": 0.5081,
      "step": 11019
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.666725604506887,
      "learning_rate": 3.6313766958995635e-07,
      "loss": 0.5583,
      "step": 11020
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2896188538520112,
      "learning_rate": 3.626531268188055e-07,
      "loss": 0.15,
      "step": 11021
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3671465091312707,
      "learning_rate": 3.6216889536680125e-07,
      "loss": 0.1287,
      "step": 11022
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3987110119034194,
      "learning_rate": 3.616849752664503e-07,
      "loss": 0.1885,
      "step": 11023
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.416102596938885,
      "learning_rate": 3.612013665502401e-07,
      "loss": 0.2125,
      "step": 11024
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3298266166149644,
      "learning_rate": 3.60718069250639e-07,
      "loss": 0.1799,
      "step": 11025
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4291490805647944,
      "learning_rate": 3.602350834000884e-07,
      "loss": 0.1688,
      "step": 11026
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2591676795184357,
      "learning_rate": 3.597524090310156e-07,
      "loss": 0.1568,
      "step": 11027
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.215410419605292,
      "learning_rate": 3.5927004617582363e-07,
      "loss": 0.1536,
      "step": 11028
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3920706912398622,
      "learning_rate": 3.587879948668926e-07,
      "loss": 0.2054,
      "step": 11029
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.430658811985136,
      "learning_rate": 3.5830625513658677e-07,
      "loss": 0.1725,
      "step": 11030
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0397914583044003,
      "learning_rate": 3.5782482701724575e-07,
      "loss": 0.2074,
      "step": 11031
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.443383237556742,
      "learning_rate": 3.573437105411892e-07,
      "loss": 0.1318,
      "step": 11032
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.222184831502122,
      "learning_rate": 3.5686290574071637e-07,
      "loss": 0.1332,
      "step": 11033
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.465555494750924,
      "learning_rate": 3.563824126481047e-07,
      "loss": 0.1733,
      "step": 11034
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3676026107522443,
      "learning_rate": 3.5590223129561063e-07,
      "loss": 0.1733,
      "step": 11035
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4452617528897223,
      "learning_rate": 3.554223617154723e-07,
      "loss": 0.1753,
      "step": 11036
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.544648516317134,
      "learning_rate": 3.549428039399022e-07,
      "loss": 0.4559,
      "step": 11037
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2562725218811925,
      "learning_rate": 3.544635580010952e-07,
      "loss": 0.1619,
      "step": 11038
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4287710654552075,
      "learning_rate": 3.539846239312267e-07,
      "loss": 0.1852,
      "step": 11039
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.402688958356042,
      "learning_rate": 3.535060017624453e-07,
      "loss": 0.1801,
      "step": 11040
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.205857472891897,
      "learning_rate": 3.5302769152688543e-07,
      "loss": 0.1681,
      "step": 11041
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2762026035082723,
      "learning_rate": 3.5254969325665576e-07,
      "loss": 0.1533,
      "step": 11042
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2679095645245506,
      "learning_rate": 3.520720069838474e-07,
      "loss": 0.1804,
      "step": 11043
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4587740545207275,
      "learning_rate": 3.515946327405273e-07,
      "loss": 0.144,
      "step": 11044
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2461141742161097,
      "learning_rate": 3.511175705587433e-07,
      "loss": 0.1489,
      "step": 11045
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4816179584005076,
      "learning_rate": 3.506408204705225e-07,
      "loss": 0.2099,
      "step": 11046
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.21467246003251,
      "learning_rate": 3.5016438250787097e-07,
      "loss": 0.1103,
      "step": 11047
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.039574422964098,
      "learning_rate": 3.496882567027726e-07,
      "loss": 0.6468,
      "step": 11048
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4056976353816089,
      "learning_rate": 3.4921244308719014e-07,
      "loss": 0.1383,
      "step": 11049
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2845727549568058,
      "learning_rate": 3.4873694169306915e-07,
      "loss": 0.1647,
      "step": 11050
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4997488741741116,
      "learning_rate": 3.482617525523296e-07,
      "loss": 0.2078,
      "step": 11051
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.386253062336796,
      "learning_rate": 3.477868756968733e-07,
      "loss": 0.2082,
      "step": 11052
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3453813991414216,
      "learning_rate": 3.473123111585791e-07,
      "loss": 0.1863,
      "step": 11053
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5657840770247213,
      "learning_rate": 3.4683805896930713e-07,
      "loss": 0.1707,
      "step": 11054
    },
    {
      "epoch": 0.88,
      "grad_norm": 10.470699246511801,
      "learning_rate": 3.4636411916089465e-07,
      "loss": 0.5715,
      "step": 11055
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2985639075382216,
      "learning_rate": 3.4589049176515844e-07,
      "loss": 0.139,
      "step": 11056
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.426930214315953,
      "learning_rate": 3.4541717681389475e-07,
      "loss": 0.1523,
      "step": 11057
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5630127519827672,
      "learning_rate": 3.4494417433887874e-07,
      "loss": 0.198,
      "step": 11058
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4551814617907926,
      "learning_rate": 3.444714843718655e-07,
      "loss": 0.1526,
      "step": 11059
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4627558181325298,
      "learning_rate": 3.4399910694458583e-07,
      "loss": 0.18,
      "step": 11060
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.558874424708463,
      "learning_rate": 3.435270420887543e-07,
      "loss": 0.5888,
      "step": 11061
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3633184816651975,
      "learning_rate": 3.430552898360617e-07,
      "loss": 0.1679,
      "step": 11062
    },
    {
      "epoch": 0.89,
      "grad_norm": 10.272409880032354,
      "learning_rate": 3.4258385021817555e-07,
      "loss": 0.6198,
      "step": 11063
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2833901003105128,
      "learning_rate": 3.4211272326674816e-07,
      "loss": 0.1655,
      "step": 11064
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3870952801446848,
      "learning_rate": 3.41641909013406e-07,
      "loss": 0.1563,
      "step": 11065
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3942311187350314,
      "learning_rate": 3.411714074897571e-07,
      "loss": 0.1731,
      "step": 11066
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3201721977710539,
      "learning_rate": 3.4070121872738727e-07,
      "loss": 0.1457,
      "step": 11067
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4562341523126676,
      "learning_rate": 3.4023134275786184e-07,
      "loss": 0.2079,
      "step": 11068
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2836211048998314,
      "learning_rate": 3.3976177961272395e-07,
      "loss": 0.1639,
      "step": 11069
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4275956787815223,
      "learning_rate": 3.3929252932349944e-07,
      "loss": 0.1686,
      "step": 11070
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3213251128933132,
      "learning_rate": 3.388235919216876e-07,
      "loss": 0.1838,
      "step": 11071
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4243463840839043,
      "learning_rate": 3.383549674387704e-07,
      "loss": 0.1882,
      "step": 11072
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2774908023058051,
      "learning_rate": 3.378866559062105e-07,
      "loss": 0.1544,
      "step": 11073
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3253873546663835,
      "learning_rate": 3.374186573554428e-07,
      "loss": 0.1609,
      "step": 11074
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4527610039297811,
      "learning_rate": 3.369509718178887e-07,
      "loss": 0.2088,
      "step": 11075
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1784328852258756,
      "learning_rate": 3.364835993249443e-07,
      "loss": 0.1316,
      "step": 11076
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3477742521436449,
      "learning_rate": 3.3601653990798566e-07,
      "loss": 0.1638,
      "step": 11077
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3399956660126535,
      "learning_rate": 3.3554979359836816e-07,
      "loss": 0.1538,
      "step": 11078
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2741122550977966,
      "learning_rate": 3.350833604274256e-07,
      "loss": 0.1583,
      "step": 11079
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3418466177857096,
      "learning_rate": 3.3461724042647136e-07,
      "loss": 0.1518,
      "step": 11080
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4152361416780388,
      "learning_rate": 3.3415143362679704e-07,
      "loss": 0.1291,
      "step": 11081
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2158615842321112,
      "learning_rate": 3.336859400596743e-07,
      "loss": 0.1446,
      "step": 11082
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.340204107858864,
      "learning_rate": 3.3322075975635194e-07,
      "loss": 0.1926,
      "step": 11083
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.432024641806054,
      "learning_rate": 3.3275589274806117e-07,
      "loss": 0.1982,
      "step": 11084
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3892252764458355,
      "learning_rate": 3.3229133906600706e-07,
      "loss": 0.1572,
      "step": 11085
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2932078274644476,
      "learning_rate": 3.3182709874137854e-07,
      "loss": 0.1575,
      "step": 11086
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5323624056715146,
      "learning_rate": 3.313631718053412e-07,
      "loss": 0.186,
      "step": 11087
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3295573604417814,
      "learning_rate": 3.3089955828903895e-07,
      "loss": 0.1748,
      "step": 11088
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3811967715112745,
      "learning_rate": 3.30436258223597e-07,
      "loss": 0.1992,
      "step": 11089
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3233296382122004,
      "learning_rate": 3.299732716401166e-07,
      "loss": 0.1651,
      "step": 11090
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3481720028068371,
      "learning_rate": 3.295105985696806e-07,
      "loss": 0.1834,
      "step": 11091
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1764927131863134,
      "learning_rate": 3.2904823904334804e-07,
      "loss": 0.1417,
      "step": 11092
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2141261313428235,
      "learning_rate": 3.285861930921608e-07,
      "loss": 0.1517,
      "step": 11093
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.25421314163318,
      "learning_rate": 3.2812446074713513e-07,
      "loss": 0.6064,
      "step": 11094
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4175774048839866,
      "learning_rate": 3.276630420392707e-07,
      "loss": 0.2164,
      "step": 11095
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3645605837282677,
      "learning_rate": 3.272019369995422e-07,
      "loss": 0.1494,
      "step": 11096
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5182162987960348,
      "learning_rate": 3.267411456589059e-07,
      "loss": 0.179,
      "step": 11097
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5123252274058185,
      "learning_rate": 3.2628066804829604e-07,
      "loss": 0.1896,
      "step": 11098
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3349536137495188,
      "learning_rate": 3.258205041986262e-07,
      "loss": 0.145,
      "step": 11099
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.537807137846857,
      "learning_rate": 3.2536065414078724e-07,
      "loss": 0.6531,
      "step": 11100
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2637114981561115,
      "learning_rate": 3.249011179056516e-07,
      "loss": 0.1337,
      "step": 11101
    },
    {
      "epoch": 0.89,
      "grad_norm": 13.717426333235315,
      "learning_rate": 3.2444189552406914e-07,
      "loss": 0.6826,
      "step": 11102
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3208547401926929,
      "learning_rate": 3.2398298702686736e-07,
      "loss": 0.1561,
      "step": 11103
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4993039738045884,
      "learning_rate": 3.2352439244485723e-07,
      "loss": 0.1856,
      "step": 11104
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2196306290947323,
      "learning_rate": 3.230661118088219e-07,
      "loss": 0.1897,
      "step": 11105
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.21585471262804,
      "learning_rate": 3.2260814514953053e-07,
      "loss": 0.6915,
      "step": 11106
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4078083556275283,
      "learning_rate": 3.2215049249772645e-07,
      "loss": 0.1501,
      "step": 11107
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4117956060357606,
      "learning_rate": 3.216931538841317e-07,
      "loss": 0.1462,
      "step": 11108
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3046957063643916,
      "learning_rate": 3.212361293394511e-07,
      "loss": 0.1645,
      "step": 11109
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2266206781493447,
      "learning_rate": 3.2077941889436525e-07,
      "loss": 0.156,
      "step": 11110
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2246394431719136,
      "learning_rate": 3.203230225795345e-07,
      "loss": 0.1405,
      "step": 11111
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4544179726239115,
      "learning_rate": 3.198669404255977e-07,
      "loss": 0.1564,
      "step": 11112
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.318958098398374,
      "learning_rate": 3.1941117246317433e-07,
      "loss": 0.649,
      "step": 11113
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.533541404980899,
      "learning_rate": 3.1895571872285926e-07,
      "loss": 0.1803,
      "step": 11114
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.251346296631035,
      "learning_rate": 3.185005792352308e-07,
      "loss": 0.1422,
      "step": 11115
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5429566960988912,
      "learning_rate": 3.1804575403084293e-07,
      "loss": 0.1752,
      "step": 11116
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3996174578535134,
      "learning_rate": 3.1759124314022835e-07,
      "loss": 0.1632,
      "step": 11117
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.107689015086432,
      "learning_rate": 3.171370465939022e-07,
      "loss": 0.13,
      "step": 11118
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.252095183458183,
      "learning_rate": 3.166831644223528e-07,
      "loss": 0.1337,
      "step": 11119
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4629233393971817,
      "learning_rate": 3.162295966560536e-07,
      "loss": 0.1736,
      "step": 11120
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4730333068431403,
      "learning_rate": 3.1577634332545294e-07,
      "loss": 0.1974,
      "step": 11121
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2654861225452234,
      "learning_rate": 3.1532340446097874e-07,
      "loss": 0.1514,
      "step": 11122
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8531602252401644,
      "learning_rate": 3.148707800930384e-07,
      "loss": 0.18,
      "step": 11123
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3135201466264634,
      "learning_rate": 3.1441847025201754e-07,
      "loss": 0.1482,
      "step": 11124
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1459905357761362,
      "learning_rate": 3.1396647496828245e-07,
      "loss": 0.1283,
      "step": 11125
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4546993747524704,
      "learning_rate": 3.135147942721756e-07,
      "loss": 0.1962,
      "step": 11126
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2675305180466545,
      "learning_rate": 3.130634281940198e-07,
      "loss": 0.1776,
      "step": 11127
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3387171026259423,
      "learning_rate": 3.126123767641165e-07,
      "loss": 0.1613,
      "step": 11128
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5483004018780033,
      "learning_rate": 3.121616400127481e-07,
      "loss": 0.1796,
      "step": 11129
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.331678741762243,
      "learning_rate": 3.1171121797017036e-07,
      "loss": 0.1943,
      "step": 11130
    },
    {
      "epoch": 0.89,
      "grad_norm": 14.226704254575495,
      "learning_rate": 3.1126111066662467e-07,
      "loss": 0.5818,
      "step": 11131
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3554157878099566,
      "learning_rate": 3.10811318132328e-07,
      "loss": 0.1667,
      "step": 11132
    },
    {
      "epoch": 0.89,
      "grad_norm": 10.789609420766846,
      "learning_rate": 3.103618403974734e-07,
      "loss": 0.7467,
      "step": 11133
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.218761279217407,
      "learning_rate": 3.0991267749223787e-07,
      "loss": 0.152,
      "step": 11134
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4001744893932715,
      "learning_rate": 3.09463829446775e-07,
      "loss": 0.2006,
      "step": 11135
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4691793704122154,
      "learning_rate": 3.090152962912174e-07,
      "loss": 0.1917,
      "step": 11136
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3000315431550218,
      "learning_rate": 3.085670780556749e-07,
      "loss": 0.1803,
      "step": 11137
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2598946649620286,
      "learning_rate": 3.0811917477024057e-07,
      "loss": 0.1389,
      "step": 11138
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.94081004043715,
      "learning_rate": 3.076715864649804e-07,
      "loss": 0.501,
      "step": 11139
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.460900402587475,
      "learning_rate": 3.072243131699443e-07,
      "loss": 0.59,
      "step": 11140
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4067151448244295,
      "learning_rate": 3.067773549151598e-07,
      "loss": 0.1744,
      "step": 11141
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2396228480282168,
      "learning_rate": 3.0633071173062966e-07,
      "loss": 0.1495,
      "step": 11142
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.89290509373479,
      "learning_rate": 3.05884383646341e-07,
      "loss": 0.5554,
      "step": 11143
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.254318194169198,
      "learning_rate": 3.054383706922559e-07,
      "loss": 0.1508,
      "step": 11144
    },
    {
      "epoch": 0.89,
      "grad_norm": 11.242087523337672,
      "learning_rate": 3.049926728983171e-07,
      "loss": 0.4199,
      "step": 11145
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4413295071220684,
      "learning_rate": 3.045472902944452e-07,
      "loss": 0.1848,
      "step": 11146
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6188726010420798,
      "learning_rate": 3.041022229105406e-07,
      "loss": 0.2121,
      "step": 11147
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4643359019414128,
      "learning_rate": 3.036574707764811e-07,
      "loss": 0.2102,
      "step": 11148
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3398425913420022,
      "learning_rate": 3.032130339221268e-07,
      "loss": 0.1545,
      "step": 11149
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.259954941119145,
      "learning_rate": 3.0276891237731085e-07,
      "loss": 0.1633,
      "step": 11150
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.797083454564077,
      "learning_rate": 3.023251061718496e-07,
      "loss": 0.5474,
      "step": 11151
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4232150218390427,
      "learning_rate": 3.0188161533553864e-07,
      "loss": 0.1568,
      "step": 11152
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.410387009940278,
      "learning_rate": 3.014384398981479e-07,
      "loss": 0.1911,
      "step": 11153
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.309069093701047,
      "learning_rate": 3.009955798894315e-07,
      "loss": 0.1773,
      "step": 11154
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4405097712256496,
      "learning_rate": 3.005530353391195e-07,
      "loss": 0.1991,
      "step": 11155
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.079036034296953,
      "learning_rate": 3.001108062769209e-07,
      "loss": 0.5024,
      "step": 11156
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1491660571543736,
      "learning_rate": 2.996688927325242e-07,
      "loss": 0.1409,
      "step": 11157
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4011063665403745,
      "learning_rate": 2.9922729473559564e-07,
      "loss": 0.165,
      "step": 11158
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.515201636351137,
      "learning_rate": 2.9878601231578207e-07,
      "loss": 0.2126,
      "step": 11159
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.41828655010213,
      "learning_rate": 2.9834504550270706e-07,
      "loss": 0.1487,
      "step": 11160
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5116500148704706,
      "learning_rate": 2.9790439432597517e-07,
      "loss": 0.1635,
      "step": 11161
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2852916893514048,
      "learning_rate": 2.9746405881516727e-07,
      "loss": 0.1609,
      "step": 11162
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2531764113769908,
      "learning_rate": 2.9702403899984635e-07,
      "loss": 0.1759,
      "step": 11163
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.446296957653437,
      "learning_rate": 2.965843349095499e-07,
      "loss": 0.1819,
      "step": 11164
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.423817906458392,
      "learning_rate": 2.9614494657379865e-07,
      "loss": 0.1825,
      "step": 11165
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4649724112846072,
      "learning_rate": 2.957058740220886e-07,
      "loss": 0.1837,
      "step": 11166
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4910796802988728,
      "learning_rate": 2.952671172838972e-07,
      "loss": 0.1804,
      "step": 11167
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4760678774805092,
      "learning_rate": 2.948286763886793e-07,
      "loss": 0.1746,
      "step": 11168
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.418580867201188,
      "learning_rate": 2.943905513658679e-07,
      "loss": 0.1882,
      "step": 11169
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.287768415634642,
      "learning_rate": 2.939527422448768e-07,
      "loss": 0.1696,
      "step": 11170
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4511559305315846,
      "learning_rate": 2.9351524905509634e-07,
      "loss": 0.1773,
      "step": 11171
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.243612767275687,
      "learning_rate": 2.9307807182589743e-07,
      "loss": 0.5028,
      "step": 11172
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.148265095834209,
      "learning_rate": 2.926412105866283e-07,
      "loss": 0.1495,
      "step": 11173
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.296998698100775,
      "learning_rate": 2.9220466536661885e-07,
      "loss": 0.1506,
      "step": 11174
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3369539130876749,
      "learning_rate": 2.917684361951728e-07,
      "loss": 0.1791,
      "step": 11175
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4148968677225944,
      "learning_rate": 2.913325231015779e-07,
      "loss": 0.1748,
      "step": 11176
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.328571471896504,
      "learning_rate": 2.908969261150979e-07,
      "loss": 0.1806,
      "step": 11177
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.559644767050513,
      "learning_rate": 2.904616452649739e-07,
      "loss": 0.2056,
      "step": 11178
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3342893807781893,
      "learning_rate": 2.9002668058042973e-07,
      "loss": 0.1829,
      "step": 11179
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5132020511015358,
      "learning_rate": 2.8959203209066477e-07,
      "loss": 0.1813,
      "step": 11180
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.315377947579541,
      "learning_rate": 2.891576998248591e-07,
      "loss": 0.1698,
      "step": 11181
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3975496276756696,
      "learning_rate": 2.8872368381216984e-07,
      "loss": 0.1804,
      "step": 11182
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5575378891100844,
      "learning_rate": 2.882899840817349e-07,
      "loss": 0.1828,
      "step": 11183
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.215020347320121,
      "learning_rate": 2.878566006626693e-07,
      "loss": 0.4571,
      "step": 11184
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.354633052374859,
      "learning_rate": 2.874235335840664e-07,
      "loss": 0.1732,
      "step": 11185
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.385517589919848,
      "learning_rate": 2.869907828750018e-07,
      "loss": 0.1549,
      "step": 11186
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6360534661396309,
      "learning_rate": 2.8655834856452404e-07,
      "loss": 0.1946,
      "step": 11187
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3263504680774265,
      "learning_rate": 2.8612623068166646e-07,
      "loss": 0.1641,
      "step": 11188
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5332972515360006,
      "learning_rate": 2.85694429255437e-07,
      "loss": 0.2095,
      "step": 11189
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.306270618784906,
      "learning_rate": 2.852629443148247e-07,
      "loss": 0.1625,
      "step": 11190
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.323675804102063,
      "learning_rate": 2.848317758887958e-07,
      "loss": 0.165,
      "step": 11191
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.391800786274174,
      "learning_rate": 2.8440092400629673e-07,
      "loss": 0.1997,
      "step": 11192
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3876283897055879,
      "learning_rate": 2.839703886962508e-07,
      "loss": 0.1549,
      "step": 11193
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0982690536142847,
      "learning_rate": 2.8354016998756174e-07,
      "loss": 0.1604,
      "step": 11194
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4975301551982976,
      "learning_rate": 2.831102679091113e-07,
      "loss": 0.1653,
      "step": 11195
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.087906569363012,
      "learning_rate": 2.826806824897599e-07,
      "loss": 0.5338,
      "step": 11196
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4753092377997294,
      "learning_rate": 2.822514137583482e-07,
      "loss": 0.1852,
      "step": 11197
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3112978564947928,
      "learning_rate": 2.818224617436921e-07,
      "loss": 0.1412,
      "step": 11198
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3520446186875188,
      "learning_rate": 2.8139382647459025e-07,
      "loss": 0.1926,
      "step": 11199
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3077274941590027,
      "learning_rate": 2.809655079798179e-07,
      "loss": 0.1452,
      "step": 11200
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5967930585525536,
      "learning_rate": 2.805375062881288e-07,
      "loss": 0.1759,
      "step": 11201
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3451385067030859,
      "learning_rate": 2.801098214282566e-07,
      "loss": 0.1635,
      "step": 11202
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2862528581895054,
      "learning_rate": 2.796824534289122e-07,
      "loss": 0.1757,
      "step": 11203
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3925230983974384,
      "learning_rate": 2.792554023187871e-07,
      "loss": 0.1683,
      "step": 11204
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.733450523254797,
      "learning_rate": 2.7882866812655006e-07,
      "loss": 0.6001,
      "step": 11205
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.694024746934279,
      "learning_rate": 2.7840225088084927e-07,
      "loss": 0.5553,
      "step": 11206
    },
    {
      "epoch": 0.9,
      "grad_norm": 14.651615728885222,
      "learning_rate": 2.779761506103107e-07,
      "loss": 0.522,
      "step": 11207
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2901443145978497,
      "learning_rate": 2.775503673435415e-07,
      "loss": 0.1701,
      "step": 11208
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3730175919097642,
      "learning_rate": 2.771249011091226e-07,
      "loss": 0.1758,
      "step": 11209
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8400465603811433,
      "learning_rate": 2.7669975193562013e-07,
      "loss": 0.255,
      "step": 11210
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4243811786710665,
      "learning_rate": 2.762749198515752e-07,
      "loss": 0.1577,
      "step": 11211
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5835270772356915,
      "learning_rate": 2.758504048855054e-07,
      "loss": 0.1827,
      "step": 11212
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5407018723513857,
      "learning_rate": 2.754262070659125e-07,
      "loss": 0.1648,
      "step": 11213
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2361681416791075,
      "learning_rate": 2.750023264212731e-07,
      "loss": 0.1874,
      "step": 11214
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3295495383107927,
      "learning_rate": 2.7457876298004393e-07,
      "loss": 0.1657,
      "step": 11215
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4016942045911114,
      "learning_rate": 2.7415551677065943e-07,
      "loss": 0.155,
      "step": 11216
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2793645361885995,
      "learning_rate": 2.737325878215341e-07,
      "loss": 0.1602,
      "step": 11217
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5005097158778191,
      "learning_rate": 2.733099761610597e-07,
      "loss": 0.188,
      "step": 11218
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2643817258749224,
      "learning_rate": 2.7288768181760904e-07,
      "loss": 0.1613,
      "step": 11219
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4258418527834953,
      "learning_rate": 2.7246570481953004e-07,
      "loss": 0.1726,
      "step": 11220
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2752776027513757,
      "learning_rate": 2.720440451951517e-07,
      "loss": 0.1323,
      "step": 11221
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4529624644026562,
      "learning_rate": 2.716227029727819e-07,
      "loss": 0.1793,
      "step": 11222
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4750083057856656,
      "learning_rate": 2.7120167818070695e-07,
      "loss": 0.1733,
      "step": 11223
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2516795667807024,
      "learning_rate": 2.707809708471909e-07,
      "loss": 0.1593,
      "step": 11224
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5290373768374597,
      "learning_rate": 2.7036058100047723e-07,
      "loss": 0.151,
      "step": 11225
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4296213240426994,
      "learning_rate": 2.699405086687884e-07,
      "loss": 0.1781,
      "step": 11226
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4517576008756699,
      "learning_rate": 2.695207538803235e-07,
      "loss": 0.1977,
      "step": 11227
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4830281465735342,
      "learning_rate": 2.69101316663265e-07,
      "loss": 0.1933,
      "step": 11228
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.242334605681596,
      "learning_rate": 2.686821970457681e-07,
      "loss": 0.1587,
      "step": 11229
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2721309371086087,
      "learning_rate": 2.682633950559699e-07,
      "loss": 0.1637,
      "step": 11230
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4717923791003538,
      "learning_rate": 2.6784491072198824e-07,
      "loss": 0.1401,
      "step": 11231
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.291579186756562,
      "learning_rate": 2.6742674407191416e-07,
      "loss": 0.17,
      "step": 11232
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.233129772261389,
      "learning_rate": 2.6700889513382233e-07,
      "loss": 0.5044,
      "step": 11233
    },
    {
      "epoch": 0.9,
      "grad_norm": 12.155976609685988,
      "learning_rate": 2.6659136393576423e-07,
      "loss": 0.5389,
      "step": 11234
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.403403485571084,
      "learning_rate": 2.661741505057691e-07,
      "loss": 0.1962,
      "step": 11235
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3619770184361806,
      "learning_rate": 2.657572548718462e-07,
      "loss": 0.1917,
      "step": 11236
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.255701772370654,
      "learning_rate": 2.653406770619832e-07,
      "loss": 0.129,
      "step": 11237
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.308736207511838,
      "learning_rate": 2.64924417104146e-07,
      "loss": 0.1675,
      "step": 11238
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3199373540189938,
      "learning_rate": 2.6450847502627883e-07,
      "loss": 0.1402,
      "step": 11239
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.259438699294948,
      "learning_rate": 2.640928508563062e-07,
      "loss": 0.6889,
      "step": 11240
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4930885147163255,
      "learning_rate": 2.63677544622129e-07,
      "loss": 0.1832,
      "step": 11241
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.283363145290632,
      "learning_rate": 2.6326255635163e-07,
      "loss": 0.1841,
      "step": 11242
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0185935890238516,
      "learning_rate": 2.6284788607266575e-07,
      "loss": 0.1894,
      "step": 11243
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.262108635097663,
      "learning_rate": 2.6243353381307736e-07,
      "loss": 0.4882,
      "step": 11244
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3874852569905642,
      "learning_rate": 2.620194996006803e-07,
      "loss": 0.2046,
      "step": 11245
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4284294551158938,
      "learning_rate": 2.616057834632679e-07,
      "loss": 0.2049,
      "step": 11246
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4427296521748576,
      "learning_rate": 2.6119238542861746e-07,
      "loss": 0.1551,
      "step": 11247
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2150311113623231,
      "learning_rate": 2.607793055244795e-07,
      "loss": 0.1572,
      "step": 11248
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3241562533462572,
      "learning_rate": 2.603665437785868e-07,
      "loss": 0.161,
      "step": 11249
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3563488547256068,
      "learning_rate": 2.599541002186479e-07,
      "loss": 0.2204,
      "step": 11250
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.501295912379093,
      "learning_rate": 2.595419748723521e-07,
      "loss": 0.1834,
      "step": 11251
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5079538826457175,
      "learning_rate": 2.591301677673663e-07,
      "loss": 0.1804,
      "step": 11252
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2117070089435333,
      "learning_rate": 2.587186789313378e-07,
      "loss": 0.1619,
      "step": 11253
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2387238116416834,
      "learning_rate": 2.58307508391889e-07,
      "loss": 0.1357,
      "step": 11254
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.137241958271838,
      "learning_rate": 2.578966561766233e-07,
      "loss": 0.6801,
      "step": 11255
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2455089060626816,
      "learning_rate": 2.5748612231312476e-07,
      "loss": 0.1523,
      "step": 11256
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3726570472408386,
      "learning_rate": 2.570759068289502e-07,
      "loss": 0.1467,
      "step": 11257
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.304875651541526,
      "learning_rate": 2.566660097516416e-07,
      "loss": 0.1764,
      "step": 11258
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4302420917468852,
      "learning_rate": 2.562564311087151e-07,
      "loss": 0.1691,
      "step": 11259
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.156362734284608,
      "learning_rate": 2.5584717092766774e-07,
      "loss": 0.5535,
      "step": 11260
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.625379630242133,
      "learning_rate": 2.5543822923597305e-07,
      "loss": 0.5441,
      "step": 11261
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.425631249694794,
      "learning_rate": 2.550296060610868e-07,
      "loss": 0.6162,
      "step": 11262
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.373690359634014,
      "learning_rate": 2.546213014304394e-07,
      "loss": 0.1595,
      "step": 11263
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.608065901132052,
      "learning_rate": 2.54213315371441e-07,
      "loss": 0.5138,
      "step": 11264
    },
    {
      "epoch": 0.9,
      "grad_norm": 12.538600530275065,
      "learning_rate": 2.5380564791148364e-07,
      "loss": 0.5417,
      "step": 11265
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3468750836968264,
      "learning_rate": 2.533982990779316e-07,
      "loss": 0.1771,
      "step": 11266
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.360446351749534,
      "learning_rate": 2.5299126889813454e-07,
      "loss": 0.1904,
      "step": 11267
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5155586404153307,
      "learning_rate": 2.5258455739941623e-07,
      "loss": 0.1834,
      "step": 11268
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4738491296319747,
      "learning_rate": 2.5217816460908094e-07,
      "loss": 0.1733,
      "step": 11269
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.260438200463216,
      "learning_rate": 2.517720905544102e-07,
      "loss": 0.426,
      "step": 11270
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3242362176274711,
      "learning_rate": 2.513663352626661e-07,
      "loss": 0.18,
      "step": 11271
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3125636685125515,
      "learning_rate": 2.5096089876108784e-07,
      "loss": 0.2008,
      "step": 11272
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4872015756791364,
      "learning_rate": 2.5055578107689326e-07,
      "loss": 0.1958,
      "step": 11273
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5479871507838738,
      "learning_rate": 2.501509822372788e-07,
      "loss": 0.1794,
      "step": 11274
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3542164714482559,
      "learning_rate": 2.497465022694207e-07,
      "loss": 0.164,
      "step": 11275
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.575557847446362,
      "learning_rate": 2.4934234120047306e-07,
      "loss": 0.4692,
      "step": 11276
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.366116668322423,
      "learning_rate": 2.489384990575672e-07,
      "loss": 0.1926,
      "step": 11277
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5124471526161982,
      "learning_rate": 2.485349758678152e-07,
      "loss": 0.1864,
      "step": 11278
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.343282577393237,
      "learning_rate": 2.48131771658307e-07,
      "loss": 0.5763,
      "step": 11279
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2765130945823198,
      "learning_rate": 2.477288864561106e-07,
      "loss": 0.1541,
      "step": 11280
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3667983399347927,
      "learning_rate": 2.4732632028827255e-07,
      "loss": 0.1746,
      "step": 11281
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3074991472304878,
      "learning_rate": 2.46924073181819e-07,
      "loss": 0.1751,
      "step": 11282
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6123227552756643,
      "learning_rate": 2.465221451637534e-07,
      "loss": 0.1489,
      "step": 11283
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4770941867483163,
      "learning_rate": 2.461205362610591e-07,
      "loss": 0.1703,
      "step": 11284
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.543111835060115,
      "learning_rate": 2.4571924650069634e-07,
      "loss": 0.4469,
      "step": 11285
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3415549581439412,
      "learning_rate": 2.453182759096051e-07,
      "loss": 0.1782,
      "step": 11286
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.429112839847905,
      "learning_rate": 2.4491762451470556e-07,
      "loss": 0.6321,
      "step": 11287
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2451587676931866,
      "learning_rate": 2.445172923428912e-07,
      "loss": 0.1513,
      "step": 11288
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3143353382703578,
      "learning_rate": 2.4411727942104115e-07,
      "loss": 0.1789,
      "step": 11289
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.412337338748157,
      "learning_rate": 2.437175857760077e-07,
      "loss": 0.1729,
      "step": 11290
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4056652859777632,
      "learning_rate": 2.433182114346228e-07,
      "loss": 0.1476,
      "step": 11291
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3626645754418352,
      "learning_rate": 2.4291915642369934e-07,
      "loss": 0.1857,
      "step": 11292
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3428383347917274,
      "learning_rate": 2.42520420770026e-07,
      "loss": 0.1678,
      "step": 11293
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4118610901162105,
      "learning_rate": 2.421220045003714e-07,
      "loss": 0.1727,
      "step": 11294
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.278524901478949,
      "learning_rate": 2.417239076414829e-07,
      "loss": 0.2124,
      "step": 11295
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3389404788255934,
      "learning_rate": 2.413261302200848e-07,
      "loss": 0.1838,
      "step": 11296
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3553019883794628,
      "learning_rate": 2.4092867226288173e-07,
      "loss": 0.1792,
      "step": 11297
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3334678646912719,
      "learning_rate": 2.4053153379655693e-07,
      "loss": 0.1495,
      "step": 11298
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3868336794763116,
      "learning_rate": 2.4013471484777063e-07,
      "loss": 0.1765,
      "step": 11299
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2548148723380272,
      "learning_rate": 2.397382154431621e-07,
      "loss": 0.141,
      "step": 11300
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3415066768795039,
      "learning_rate": 2.3934203560935174e-07,
      "loss": 0.187,
      "step": 11301
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.378302616017471,
      "learning_rate": 2.389461753729333e-07,
      "loss": 0.1716,
      "step": 11302
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.320515498694177,
      "learning_rate": 2.385506347604838e-07,
      "loss": 0.1596,
      "step": 11303
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3852956920895878,
      "learning_rate": 2.38155413798557e-07,
      "loss": 0.172,
      "step": 11304
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3757911350499652,
      "learning_rate": 2.3776051251368505e-07,
      "loss": 0.2087,
      "step": 11305
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.509730190985472,
      "learning_rate": 2.3736593093237948e-07,
      "loss": 0.2175,
      "step": 11306
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3880319628435038,
      "learning_rate": 2.3697166908112855e-07,
      "loss": 0.174,
      "step": 11307
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3601592483856686,
      "learning_rate": 2.3657772698640113e-07,
      "loss": 0.1746,
      "step": 11308
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.887079865979303,
      "learning_rate": 2.361841046746427e-07,
      "loss": 0.4426,
      "step": 11309
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2900978896642583,
      "learning_rate": 2.3579080217228046e-07,
      "loss": 0.1559,
      "step": 11310
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3303123338746785,
      "learning_rate": 2.3539781950571494e-07,
      "loss": 0.1905,
      "step": 11311
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4537985118973686,
      "learning_rate": 2.3500515670133117e-07,
      "loss": 0.1813,
      "step": 11312
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2868874499746372,
      "learning_rate": 2.346128137854886e-07,
      "loss": 0.1766,
      "step": 11313
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2836052704329015,
      "learning_rate": 2.342207907845262e-07,
      "loss": 0.1629,
      "step": 11314
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1469195728899055,
      "learning_rate": 2.3382908772476175e-07,
      "loss": 0.1327,
      "step": 11315
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3133155600217268,
      "learning_rate": 2.3343770463249204e-07,
      "loss": 0.1714,
      "step": 11316
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2828180306492585,
      "learning_rate": 2.3304664153399048e-07,
      "loss": 0.1751,
      "step": 11317
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4480467815344436,
      "learning_rate": 2.3265589845551218e-07,
      "loss": 0.1842,
      "step": 11318
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5155134125418652,
      "learning_rate": 2.3226547542328724e-07,
      "loss": 0.1856,
      "step": 11319
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.390828331936354,
      "learning_rate": 2.3187537246352587e-07,
      "loss": 0.1605,
      "step": 11320
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.205682173666545,
      "learning_rate": 2.3148558960241929e-07,
      "loss": 0.1549,
      "step": 11321
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.480276546162343,
      "learning_rate": 2.3109612686613158e-07,
      "loss": 0.1839,
      "step": 11322
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4709713576882006,
      "learning_rate": 2.3070698428081074e-07,
      "loss": 0.1926,
      "step": 11323
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1651707156886693,
      "learning_rate": 2.3031816187258138e-07,
      "loss": 0.1337,
      "step": 11324
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.714345901868557,
      "learning_rate": 2.2992965966754378e-07,
      "loss": 0.6461,
      "step": 11325
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.044866187826981,
      "learning_rate": 2.2954147769178148e-07,
      "loss": 0.5367,
      "step": 11326
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2210531438022616,
      "learning_rate": 2.2915361597135366e-07,
      "loss": 0.1684,
      "step": 11327
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3653871403414761,
      "learning_rate": 2.2876607453229893e-07,
      "loss": 0.1696,
      "step": 11328
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3860088379045408,
      "learning_rate": 2.283788534006337e-07,
      "loss": 0.1581,
      "step": 11329
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2186318068934878,
      "learning_rate": 2.279919526023533e-07,
      "loss": 0.1596,
      "step": 11330
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3176731839071112,
      "learning_rate": 2.2760537216343136e-07,
      "loss": 0.1824,
      "step": 11331
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3375797865317043,
      "learning_rate": 2.272191121098216e-07,
      "loss": 0.1836,
      "step": 11332
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.278116095053253,
      "learning_rate": 2.2683317246745328e-07,
      "loss": 0.1424,
      "step": 11333
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.014855159222185,
      "learning_rate": 2.2644755326223567e-07,
      "loss": 0.4228,
      "step": 11334
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3722299052231866,
      "learning_rate": 2.260622545200586e-07,
      "loss": 0.1475,
      "step": 11335
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.228068908603046,
      "learning_rate": 2.2567727626678527e-07,
      "loss": 0.1833,
      "step": 11336
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2446707372354253,
      "learning_rate": 2.2529261852826279e-07,
      "loss": 0.1691,
      "step": 11337
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2662142530203966,
      "learning_rate": 2.249082813303133e-07,
      "loss": 0.1755,
      "step": 11338
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3954235117999418,
      "learning_rate": 2.245242646987389e-07,
      "loss": 0.1643,
      "step": 11339
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2980226473618142,
      "learning_rate": 2.241405686593201e-07,
      "loss": 0.1593,
      "step": 11340
    },
    {
      "epoch": 0.91,
      "grad_norm": 35.42730954413929,
      "learning_rate": 2.237571932378152e-07,
      "loss": 0.6564,
      "step": 11341
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2531754395004402,
      "learning_rate": 2.233741384599608e-07,
      "loss": 0.139,
      "step": 11342
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3081334325517342,
      "learning_rate": 2.229914043514736e-07,
      "loss": 0.1649,
      "step": 11343
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3914059702502306,
      "learning_rate": 2.226089909380469e-07,
      "loss": 0.1522,
      "step": 11344
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.469126633590488,
      "learning_rate": 2.2222689824535294e-07,
      "loss": 0.1648,
      "step": 11345
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2423789168728636,
      "learning_rate": 2.2184512629904508e-07,
      "loss": 0.1501,
      "step": 11346
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2404927337037424,
      "learning_rate": 2.2146367512474955e-07,
      "loss": 0.1391,
      "step": 11347
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5025268019206122,
      "learning_rate": 2.2108254474807632e-07,
      "loss": 0.1675,
      "step": 11348
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.366908438639587,
      "learning_rate": 2.207017351946117e-07,
      "loss": 0.1535,
      "step": 11349
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.325470516823539,
      "learning_rate": 2.2032124648992015e-07,
      "loss": 0.1472,
      "step": 11350
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.349874325502185,
      "learning_rate": 2.1994107865954516e-07,
      "loss": 0.1872,
      "step": 11351
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.505345700678516,
      "learning_rate": 2.1956123172900855e-07,
      "loss": 0.1914,
      "step": 11352
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.282501386624574,
      "learning_rate": 2.1918170572381048e-07,
      "loss": 0.2122,
      "step": 11353
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2710376643072956,
      "learning_rate": 2.1880250066942942e-07,
      "loss": 0.1468,
      "step": 11354
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4804138448394293,
      "learning_rate": 2.1842361659132395e-07,
      "loss": 0.2016,
      "step": 11355
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5057787595821464,
      "learning_rate": 2.1804505351492755e-07,
      "loss": 0.1788,
      "step": 11356
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2201136509617896,
      "learning_rate": 2.176668114656566e-07,
      "loss": 0.1524,
      "step": 11357
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5881335013817457,
      "learning_rate": 2.1728889046890188e-07,
      "loss": 0.1976,
      "step": 11358
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.338732888422352,
      "learning_rate": 2.1691129055003535e-07,
      "loss": 0.1991,
      "step": 11359
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5490961424216514,
      "learning_rate": 2.1653401173440558e-07,
      "loss": 0.1662,
      "step": 11360
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.364717059256129,
      "learning_rate": 2.1615705404734123e-07,
      "loss": 0.1979,
      "step": 11361
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.351110415298109,
      "learning_rate": 2.1578041751414812e-07,
      "loss": 0.1949,
      "step": 11362
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2972547798748257,
      "learning_rate": 2.1540410216011166e-07,
      "loss": 0.177,
      "step": 11363
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3050948452069984,
      "learning_rate": 2.1502810801049435e-07,
      "loss": 0.1775,
      "step": 11364
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.249732013254502,
      "learning_rate": 2.1465243509053713e-07,
      "loss": 0.5292,
      "step": 11365
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1293522377783152,
      "learning_rate": 2.142770834254626e-07,
      "loss": 0.139,
      "step": 11366
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5264711838401406,
      "learning_rate": 2.1390205304046675e-07,
      "loss": 0.1855,
      "step": 11367
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5470745674084725,
      "learning_rate": 2.135273439607266e-07,
      "loss": 0.1854,
      "step": 11368
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2711768183162393,
      "learning_rate": 2.1315295621139986e-07,
      "loss": 0.17,
      "step": 11369
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.539818299926648,
      "learning_rate": 2.1277888981761753e-07,
      "loss": 0.1556,
      "step": 11370
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4243469221767162,
      "learning_rate": 2.1240514480449336e-07,
      "loss": 0.1561,
      "step": 11371
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2306508212648097,
      "learning_rate": 2.120317211971179e-07,
      "loss": 0.1412,
      "step": 11372
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3523333040255328,
      "learning_rate": 2.1165861902055995e-07,
      "loss": 0.1721,
      "step": 11373
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.296729216152383,
      "learning_rate": 2.112858382998667e-07,
      "loss": 0.1683,
      "step": 11374
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4331784653211281,
      "learning_rate": 2.109133790600648e-07,
      "loss": 0.1479,
      "step": 11375
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3148471759233555,
      "learning_rate": 2.105412413261576e-07,
      "loss": 0.1466,
      "step": 11376
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5281822435819672,
      "learning_rate": 2.1016942512312898e-07,
      "loss": 0.2459,
      "step": 11377
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5249807713625627,
      "learning_rate": 2.09797930475939e-07,
      "loss": 0.1808,
      "step": 11378
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1298143522777115,
      "learning_rate": 2.0942675740952712e-07,
      "loss": 0.1288,
      "step": 11379
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.468200947875259,
      "learning_rate": 2.0905590594881342e-07,
      "loss": 0.6689,
      "step": 11380
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2000965178762486,
      "learning_rate": 2.0868537611869132e-07,
      "loss": 0.1343,
      "step": 11381
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1935475531589255,
      "learning_rate": 2.0831516794403762e-07,
      "loss": 0.1813,
      "step": 11382
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.514907941992528,
      "learning_rate": 2.079452814497046e-07,
      "loss": 0.5754,
      "step": 11383
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.52306167020663,
      "learning_rate": 2.0757571666052466e-07,
      "loss": 0.5163,
      "step": 11384
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5487308059315799,
      "learning_rate": 2.0720647360130687e-07,
      "loss": 0.2222,
      "step": 11385
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3409813187302209,
      "learning_rate": 2.0683755229684077e-07,
      "loss": 0.1645,
      "step": 11386
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4443439127504245,
      "learning_rate": 2.0646895277189216e-07,
      "loss": 0.2067,
      "step": 11387
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.290636357865715,
      "learning_rate": 2.0610067505120622e-07,
      "loss": 0.1513,
      "step": 11388
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1834382543129705,
      "learning_rate": 2.057327191595071e-07,
      "loss": 0.1398,
      "step": 11389
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3163940697301169,
      "learning_rate": 2.053650851214961e-07,
      "loss": 0.1576,
      "step": 11390
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2865088405975913,
      "learning_rate": 2.0499777296185463e-07,
      "loss": 0.1766,
      "step": 11391
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2781280120027936,
      "learning_rate": 2.0463078270524128e-07,
      "loss": 0.1456,
      "step": 11392
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.444722876936616,
      "learning_rate": 2.0426411437629245e-07,
      "loss": 0.544,
      "step": 11393
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.393964975379059,
      "learning_rate": 2.0389776799962402e-07,
      "loss": 0.1831,
      "step": 11394
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3570879425880782,
      "learning_rate": 2.0353174359983074e-07,
      "loss": 0.1717,
      "step": 11395
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.406559724360969,
      "learning_rate": 2.0316604120148354e-07,
      "loss": 0.6578,
      "step": 11396
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4447430794356344,
      "learning_rate": 2.0280066082913385e-07,
      "loss": 0.1735,
      "step": 11397
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3122376447155761,
      "learning_rate": 2.0243560250731097e-07,
      "loss": 0.1891,
      "step": 11398
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2623000871543326,
      "learning_rate": 2.0207086626052142e-07,
      "loss": 0.167,
      "step": 11399
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2745929701369876,
      "learning_rate": 2.0170645211325335e-07,
      "loss": 0.1761,
      "step": 11400
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4543249833706569,
      "learning_rate": 2.0134236008996778e-07,
      "loss": 0.187,
      "step": 11401
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5456968944808578,
      "learning_rate": 2.0097859021510956e-07,
      "loss": 0.1965,
      "step": 11402
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2685604452349555,
      "learning_rate": 2.0061514251309978e-07,
      "loss": 0.1521,
      "step": 11403
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6966512967484293,
      "learning_rate": 2.0025201700833607e-07,
      "loss": 0.1601,
      "step": 11404
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4083702487034786,
      "learning_rate": 1.9988921372519732e-07,
      "loss": 0.1769,
      "step": 11405
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.320405015220427,
      "learning_rate": 1.9952673268803956e-07,
      "loss": 0.1456,
      "step": 11406
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3570214671225111,
      "learning_rate": 1.991645739211967e-07,
      "loss": 0.1534,
      "step": 11407
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3936849766499813,
      "learning_rate": 1.988027374489826e-07,
      "loss": 0.1677,
      "step": 11408
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2940103954318118,
      "learning_rate": 1.9844122329568728e-07,
      "loss": 0.1466,
      "step": 11409
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4664966455165562,
      "learning_rate": 1.9808003148558074e-07,
      "loss": 0.1908,
      "step": 11410
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.42781695690178,
      "learning_rate": 1.977191620429114e-07,
      "loss": 0.1783,
      "step": 11411
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3852473108245562,
      "learning_rate": 1.973586149919049e-07,
      "loss": 0.1743,
      "step": 11412
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.389177646950667,
      "learning_rate": 1.9699839035676516e-07,
      "loss": 0.5123,
      "step": 11413
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.278126115528506,
      "learning_rate": 1.9663848816167785e-07,
      "loss": 0.5899,
      "step": 11414
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3184167644848865,
      "learning_rate": 1.9627890843080034e-07,
      "loss": 0.1636,
      "step": 11415
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3006664715132283,
      "learning_rate": 1.959196511882755e-07,
      "loss": 0.1507,
      "step": 11416
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2776166239735658,
      "learning_rate": 1.9556071645822017e-07,
      "loss": 0.1734,
      "step": 11417
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2895393602465155,
      "learning_rate": 1.9520210426473063e-07,
      "loss": 0.1447,
      "step": 11418
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.07309447915683,
      "learning_rate": 1.9484381463188206e-07,
      "loss": 0.6733,
      "step": 11419
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3122534945256383,
      "learning_rate": 1.9448584758372745e-07,
      "loss": 0.1487,
      "step": 11420
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.301036612388024,
      "learning_rate": 1.941282031442976e-07,
      "loss": 0.1802,
      "step": 11421
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.15144116980991,
      "learning_rate": 1.937708813376027e-07,
      "loss": 0.6986,
      "step": 11422
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2499385185905192,
      "learning_rate": 1.934138821876308e-07,
      "loss": 0.1314,
      "step": 11423
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3804036601337681,
      "learning_rate": 1.930572057183483e-07,
      "loss": 0.2122,
      "step": 11424
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.534112555745727,
      "learning_rate": 1.9270085195370048e-07,
      "loss": 0.185,
      "step": 11425
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3593264776915632,
      "learning_rate": 1.9234482091760932e-07,
      "loss": 0.1553,
      "step": 11426
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3539667179312658,
      "learning_rate": 1.9198911263397734e-07,
      "loss": 0.147,
      "step": 11427
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.934540251060143,
      "learning_rate": 1.916337271266844e-07,
      "loss": 0.5276,
      "step": 11428
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2632505728716807,
      "learning_rate": 1.9127866441958743e-07,
      "loss": 0.1844,
      "step": 11429
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3002017700371515,
      "learning_rate": 1.9092392453652352e-07,
      "loss": 0.1891,
      "step": 11430
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2839726055450675,
      "learning_rate": 1.905695075013081e-07,
      "loss": 0.157,
      "step": 11431
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3324023563283238,
      "learning_rate": 1.902154133377332e-07,
      "loss": 0.1567,
      "step": 11432
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3098399473628979,
      "learning_rate": 1.8986164206957037e-07,
      "loss": 0.1608,
      "step": 11433
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.574573910652031,
      "learning_rate": 1.895081937205706e-07,
      "loss": 0.1883,
      "step": 11434
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4133053070200015,
      "learning_rate": 1.8915506831445996e-07,
      "loss": 0.1983,
      "step": 11435
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3655258634119631,
      "learning_rate": 1.888022658749461e-07,
      "loss": 0.1652,
      "step": 11436
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.099794613797391,
      "learning_rate": 1.8844978642571454e-07,
      "loss": 0.124,
      "step": 11437
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.242352889620964,
      "learning_rate": 1.880976299904258e-07,
      "loss": 0.1287,
      "step": 11438
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4569215421711281,
      "learning_rate": 1.8774579659272318e-07,
      "loss": 0.164,
      "step": 11439
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4125310961049473,
      "learning_rate": 1.8739428625622614e-07,
      "loss": 0.1685,
      "step": 11440
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1699636229474681,
      "learning_rate": 1.8704309900453187e-07,
      "loss": 0.1397,
      "step": 11441
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3321259395147382,
      "learning_rate": 1.866922348612171e-07,
      "loss": 0.1587,
      "step": 11442
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5587867890700113,
      "learning_rate": 1.863416938498369e-07,
      "loss": 0.1942,
      "step": 11443
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3528618112109925,
      "learning_rate": 1.8599147599392243e-07,
      "loss": 0.1509,
      "step": 11444
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5097232051060785,
      "learning_rate": 1.856415813169876e-07,
      "loss": 0.191,
      "step": 11445
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3753701599321564,
      "learning_rate": 1.8529200984251926e-07,
      "loss": 0.2013,
      "step": 11446
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4202737410647657,
      "learning_rate": 1.8494276159398582e-07,
      "loss": 0.2389,
      "step": 11447
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3293926683595405,
      "learning_rate": 1.8459383659483522e-07,
      "loss": 0.1915,
      "step": 11448
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4753297727027381,
      "learning_rate": 1.842452348684892e-07,
      "loss": 0.2012,
      "step": 11449
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.370411036750608,
      "learning_rate": 1.838969564383525e-07,
      "loss": 0.1407,
      "step": 11450
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2886527495552438,
      "learning_rate": 1.8354900132780517e-07,
      "loss": 0.1518,
      "step": 11451
    },
    {
      "epoch": 0.92,
      "grad_norm": 17.18297221343557,
      "learning_rate": 1.832013695602064e-07,
      "loss": 0.4941,
      "step": 11452
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3034576986651727,
      "learning_rate": 1.8285406115889416e-07,
      "loss": 0.1938,
      "step": 11453
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4732539637540782,
      "learning_rate": 1.8250707614718422e-07,
      "loss": 0.2083,
      "step": 11454
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5443487948506132,
      "learning_rate": 1.8216041454837075e-07,
      "loss": 0.2012,
      "step": 11455
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2788028405246241,
      "learning_rate": 1.8181407638572623e-07,
      "loss": 0.1619,
      "step": 11456
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3971423940841963,
      "learning_rate": 1.8146806168250098e-07,
      "loss": 0.1568,
      "step": 11457
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4053743817982913,
      "learning_rate": 1.8112237046192417e-07,
      "loss": 0.1887,
      "step": 11458
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5101223467114036,
      "learning_rate": 1.8077700274720388e-07,
      "loss": 0.134,
      "step": 11459
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4153147719482428,
      "learning_rate": 1.804319585615244e-07,
      "loss": 0.1577,
      "step": 11460
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.560946142223074,
      "learning_rate": 1.8008723792805049e-07,
      "loss": 0.1596,
      "step": 11461
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7437177999010351,
      "learning_rate": 1.7974284086992422e-07,
      "loss": 0.1878,
      "step": 11462
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5061473489434563,
      "learning_rate": 1.7939876741026597e-07,
      "loss": 0.1769,
      "step": 11463
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1740682537061646,
      "learning_rate": 1.7905501757217393e-07,
      "loss": 0.1226,
      "step": 11464
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2490806595606019,
      "learning_rate": 1.7871159137872573e-07,
      "loss": 0.1331,
      "step": 11465
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.545424997827313,
      "learning_rate": 1.783684888529763e-07,
      "loss": 0.2009,
      "step": 11466
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3347479475507522,
      "learning_rate": 1.780257100179589e-07,
      "loss": 0.1548,
      "step": 11467
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3075272856577411,
      "learning_rate": 1.7768325489668558e-07,
      "loss": 0.1799,
      "step": 11468
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5451895703395253,
      "learning_rate": 1.773411235121464e-07,
      "loss": 0.1643,
      "step": 11469
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2483975751150365,
      "learning_rate": 1.7699931588731012e-07,
      "loss": 0.1602,
      "step": 11470
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5080102490450584,
      "learning_rate": 1.766578320451223e-07,
      "loss": 0.1695,
      "step": 11471
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3792794733884053,
      "learning_rate": 1.7631667200850856e-07,
      "loss": 0.133,
      "step": 11472
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3084587690882636,
      "learning_rate": 1.759758358003727e-07,
      "loss": 0.1838,
      "step": 11473
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.322550186207593,
      "learning_rate": 1.7563532344359367e-07,
      "loss": 0.4574,
      "step": 11474
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3568265035408105,
      "learning_rate": 1.7529513496103322e-07,
      "loss": 0.1603,
      "step": 11475
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.350049010280239,
      "learning_rate": 1.7495527037552917e-07,
      "loss": 0.1535,
      "step": 11476
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.322810727965199,
      "learning_rate": 1.746157297098966e-07,
      "loss": 0.5451,
      "step": 11477
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5306958179578278,
      "learning_rate": 1.7427651298693004e-07,
      "loss": 0.1789,
      "step": 11478
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4535288205852983,
      "learning_rate": 1.7393762022940352e-07,
      "loss": 0.1455,
      "step": 11479
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1398275904744233,
      "learning_rate": 1.7359905146006607e-07,
      "loss": 0.1405,
      "step": 11480
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5760153073825938,
      "learning_rate": 1.732608067016478e-07,
      "loss": 0.2058,
      "step": 11481
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5230044043923348,
      "learning_rate": 1.7292288597685724e-07,
      "loss": 0.1765,
      "step": 11482
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.578727590705592,
      "learning_rate": 1.7258528930837682e-07,
      "loss": 0.5327,
      "step": 11483
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3884724461109659,
      "learning_rate": 1.7224801671887392e-07,
      "loss": 0.1789,
      "step": 11484
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.399420253818466,
      "learning_rate": 1.719110682309888e-07,
      "loss": 0.1761,
      "step": 11485
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2574402819903847,
      "learning_rate": 1.715744438673417e-07,
      "loss": 0.1559,
      "step": 11486
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.366907934093081,
      "learning_rate": 1.7123814365053227e-07,
      "loss": 0.1676,
      "step": 11487
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3948790246563503,
      "learning_rate": 1.7090216760313693e-07,
      "loss": 0.1608,
      "step": 11488
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.414944117984996,
      "learning_rate": 1.7056651574771043e-07,
      "loss": 0.2018,
      "step": 11489
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4226239270356955,
      "learning_rate": 1.702311881067864e-07,
      "loss": 0.182,
      "step": 11490
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4503277530578726,
      "learning_rate": 1.6989618470287628e-07,
      "loss": 0.1688,
      "step": 11491
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4162574904915288,
      "learning_rate": 1.6956150555846928e-07,
      "loss": 0.1519,
      "step": 11492
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3541594483226762,
      "learning_rate": 1.6922715069603524e-07,
      "loss": 0.1989,
      "step": 11493
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.374428141386888,
      "learning_rate": 1.6889312013801784e-07,
      "loss": 0.3775,
      "step": 11494
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4951721292761446,
      "learning_rate": 1.6855941390684415e-07,
      "loss": 0.1856,
      "step": 11495
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.373724817312194,
      "learning_rate": 1.6822603202491515e-07,
      "loss": 0.1341,
      "step": 11496
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2684039902744457,
      "learning_rate": 1.678929745146124e-07,
      "loss": 0.1576,
      "step": 11497
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2399231376470368,
      "learning_rate": 1.675602413982952e-07,
      "loss": 0.1379,
      "step": 11498
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.423296366320413,
      "learning_rate": 1.6722783269830068e-07,
      "loss": 0.1577,
      "step": 11499
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.165002051646205,
      "learning_rate": 1.6689574843694433e-07,
      "loss": 0.1442,
      "step": 11500
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2566012843541576,
      "learning_rate": 1.6656398863652003e-07,
      "loss": 0.1562,
      "step": 11501
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.101281550273509,
      "learning_rate": 1.662325533192999e-07,
      "loss": 0.1505,
      "step": 11502
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.431737475269822,
      "learning_rate": 1.6590144250753392e-07,
      "loss": 0.1769,
      "step": 11503
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.6575369847766,
      "learning_rate": 1.655706562234516e-07,
      "loss": 0.5614,
      "step": 11504
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2570528718895349,
      "learning_rate": 1.6524019448925788e-07,
      "loss": 0.179,
      "step": 11505
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2674761158916967,
      "learning_rate": 1.6491005732713948e-07,
      "loss": 0.1712,
      "step": 11506
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4350581209408035,
      "learning_rate": 1.645802447592587e-07,
      "loss": 0.1674,
      "step": 11507
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3593463710779512,
      "learning_rate": 1.6425075680775616e-07,
      "loss": 0.1699,
      "step": 11508
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.897923619181858,
      "learning_rate": 1.6392159349475246e-07,
      "loss": 0.4527,
      "step": 11509
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3614106361429505,
      "learning_rate": 1.6359275484234495e-07,
      "loss": 0.1399,
      "step": 11510
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3523951929676439,
      "learning_rate": 1.6326424087260984e-07,
      "loss": 0.1699,
      "step": 11511
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.351064679727098,
      "learning_rate": 1.6293605160760063e-07,
      "loss": 0.1844,
      "step": 11512
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2823307393445196,
      "learning_rate": 1.6260818706934967e-07,
      "loss": 0.1967,
      "step": 11513
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.432463421913745,
      "learning_rate": 1.622806472798677e-07,
      "loss": 0.6423,
      "step": 11514
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.385859036779389,
      "learning_rate": 1.6195343226114492e-07,
      "loss": 0.1951,
      "step": 11515
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4221183041507257,
      "learning_rate": 1.6162654203514594e-07,
      "loss": 0.1922,
      "step": 11516
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3300417037604144,
      "learning_rate": 1.6129997662381657e-07,
      "loss": 0.1599,
      "step": 11517
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1844371598770815,
      "learning_rate": 1.609737360490815e-07,
      "loss": 0.1368,
      "step": 11518
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4048293891083952,
      "learning_rate": 1.6064782033283988e-07,
      "loss": 0.1651,
      "step": 11519
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.416730853310614,
      "learning_rate": 1.6032222949697361e-07,
      "loss": 0.1787,
      "step": 11520
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3133560801310733,
      "learning_rate": 1.5999696356333916e-07,
      "loss": 0.1848,
      "step": 11521
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.283556131758709,
      "learning_rate": 1.5967202255377344e-07,
      "loss": 0.1385,
      "step": 11522
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2208994065994303,
      "learning_rate": 1.5934740649009016e-07,
      "loss": 0.1802,
      "step": 11523
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2544995440690612,
      "learning_rate": 1.59023115394083e-07,
      "loss": 0.1285,
      "step": 11524
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.594654569703901,
      "learning_rate": 1.5869914928752117e-07,
      "loss": 0.5808,
      "step": 11525
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3295716500152246,
      "learning_rate": 1.5837550819215342e-07,
      "loss": 0.173,
      "step": 11526
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4497989036883736,
      "learning_rate": 1.5805219212970846e-07,
      "loss": 0.1669,
      "step": 11527
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5607999730474693,
      "learning_rate": 1.577292011218895e-07,
      "loss": 0.1921,
      "step": 11528
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2797238916967677,
      "learning_rate": 1.5740653519038085e-07,
      "loss": 0.1825,
      "step": 11529
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3908101154257415,
      "learning_rate": 1.5708419435684463e-07,
      "loss": 0.1767,
      "step": 11530
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5597408224887392,
      "learning_rate": 1.567621786429191e-07,
      "loss": 0.1747,
      "step": 11531
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.018537429593646,
      "learning_rate": 1.564404880702236e-07,
      "loss": 0.7068,
      "step": 11532
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2621791382933296,
      "learning_rate": 1.5611912266035312e-07,
      "loss": 0.1557,
      "step": 11533
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.537506580991218,
      "learning_rate": 1.5579808243488258e-07,
      "loss": 0.1453,
      "step": 11534
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2651419077587005,
      "learning_rate": 1.5547736741536367e-07,
      "loss": 0.1483,
      "step": 11535
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.264317035471171,
      "learning_rate": 1.5515697762332804e-07,
      "loss": 0.6699,
      "step": 11536
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3128643687052466,
      "learning_rate": 1.5483691308028292e-07,
      "loss": 0.18,
      "step": 11537
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.552323275639246,
      "learning_rate": 1.5451717380771725e-07,
      "loss": 0.1566,
      "step": 11538
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2224464990258117,
      "learning_rate": 1.5419775982709383e-07,
      "loss": 0.1396,
      "step": 11539
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3181872238068628,
      "learning_rate": 1.5387867115985721e-07,
      "loss": 0.1732,
      "step": 11540
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3633695410635465,
      "learning_rate": 1.5355990782742912e-07,
      "loss": 0.1615,
      "step": 11541
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5594357890009398,
      "learning_rate": 1.5324146985120802e-07,
      "loss": 0.142,
      "step": 11542
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2094004395938132,
      "learning_rate": 1.5292335725257236e-07,
      "loss": 0.1743,
      "step": 11543
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.432113804501824,
      "learning_rate": 1.526055700528778e-07,
      "loss": 0.2102,
      "step": 11544
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4326210845457028,
      "learning_rate": 1.522881082734584e-07,
      "loss": 0.146,
      "step": 11545
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2343444107002965,
      "learning_rate": 1.5197097193562603e-07,
      "loss": 0.1147,
      "step": 11546
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5500966415520663,
      "learning_rate": 1.5165416106067143e-07,
      "loss": 0.177,
      "step": 11547
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2435392311250306,
      "learning_rate": 1.5133767566986257e-07,
      "loss": 0.1571,
      "step": 11548
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4331472407456767,
      "learning_rate": 1.5102151578444746e-07,
      "loss": 0.1916,
      "step": 11549
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.111728804777668,
      "learning_rate": 1.5070568142564912e-07,
      "loss": 0.1405,
      "step": 11550
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3167782800287304,
      "learning_rate": 1.5039017261467058e-07,
      "loss": 0.1814,
      "step": 11551
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2812087500582154,
      "learning_rate": 1.5007498937269438e-07,
      "loss": 0.1808,
      "step": 11552
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.376230391418388,
      "learning_rate": 1.49760131720878e-07,
      "loss": 0.1635,
      "step": 11553
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.461463040745427,
      "learning_rate": 1.494455996803601e-07,
      "loss": 0.188,
      "step": 11554
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.096575580315122,
      "learning_rate": 1.4913139327225546e-07,
      "loss": 0.5001,
      "step": 11555
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.28238425058102,
      "learning_rate": 1.4881751251765832e-07,
      "loss": 0.171,
      "step": 11556
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4135550828292813,
      "learning_rate": 1.4850395743764013e-07,
      "loss": 0.1713,
      "step": 11557
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2748570891991327,
      "learning_rate": 1.4819072805325074e-07,
      "loss": 0.1188,
      "step": 11558
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3664247423529206,
      "learning_rate": 1.478778243855178e-07,
      "loss": 0.161,
      "step": 11559
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4798664950350844,
      "learning_rate": 1.475652464554478e-07,
      "loss": 0.1782,
      "step": 11560
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4464052747718317,
      "learning_rate": 1.472529942840256e-07,
      "loss": 0.1801,
      "step": 11561
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2592005379057025,
      "learning_rate": 1.469410678922123e-07,
      "loss": 0.1548,
      "step": 11562
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2995115674848683,
      "learning_rate": 1.4662946730094997e-07,
      "loss": 0.1493,
      "step": 11563
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1236899007906922,
      "learning_rate": 1.4631819253115632e-07,
      "loss": 0.1062,
      "step": 11564
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3206675186667056,
      "learning_rate": 1.4600724360372853e-07,
      "loss": 0.1702,
      "step": 11565
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5256679316636497,
      "learning_rate": 1.456966205395416e-07,
      "loss": 0.1623,
      "step": 11566
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3169614037898738,
      "learning_rate": 1.453863233594488e-07,
      "loss": 0.1872,
      "step": 11567
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5806519354418795,
      "learning_rate": 1.4507635208428073e-07,
      "loss": 0.2091,
      "step": 11568
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.404001972917452,
      "learning_rate": 1.4476670673484683e-07,
      "loss": 0.1575,
      "step": 11569
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4690596319965643,
      "learning_rate": 1.4445738733193494e-07,
      "loss": 0.1739,
      "step": 11570
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3435462904088367,
      "learning_rate": 1.4414839389630954e-07,
      "loss": 0.1638,
      "step": 11571
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3598554665722264,
      "learning_rate": 1.4383972644871624e-07,
      "loss": 0.1493,
      "step": 11572
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.584575882644112,
      "learning_rate": 1.4353138500987461e-07,
      "loss": 0.1652,
      "step": 11573
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3508869907894938,
      "learning_rate": 1.4322336960048643e-07,
      "loss": 0.1576,
      "step": 11574
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.373996453732175,
      "learning_rate": 1.4291568024122848e-07,
      "loss": 0.7397,
      "step": 11575
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4594073972742634,
      "learning_rate": 1.4260831695275757e-07,
      "loss": 0.1575,
      "step": 11576
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3152737649499724,
      "learning_rate": 1.4230127975570774e-07,
      "loss": 0.1879,
      "step": 11577
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4668314630234571,
      "learning_rate": 1.4199456867069084e-07,
      "loss": 0.1735,
      "step": 11578
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5572226036038777,
      "learning_rate": 1.4168818371829817e-07,
      "loss": 0.1497,
      "step": 11579
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4322428142764396,
      "learning_rate": 1.4138212491909776e-07,
      "loss": 0.1739,
      "step": 11580
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.28411973563813,
      "learning_rate": 1.410763922936359e-07,
      "loss": 0.1444,
      "step": 11581
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7479492048656395,
      "learning_rate": 1.4077098586243786e-07,
      "loss": 0.1441,
      "step": 11582
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4321866864417385,
      "learning_rate": 1.404659056460067e-07,
      "loss": 0.1872,
      "step": 11583
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2374099928641051,
      "learning_rate": 1.4016115166482269e-07,
      "loss": 0.1576,
      "step": 11584
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3933546318095675,
      "learning_rate": 1.3985672393934557e-07,
      "loss": 0.201,
      "step": 11585
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4939498893191685,
      "learning_rate": 1.3955262249001289e-07,
      "loss": 0.1777,
      "step": 11586
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.435452472503611,
      "learning_rate": 1.392488473372383e-07,
      "loss": 0.2081,
      "step": 11587
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4705166808432428,
      "learning_rate": 1.3894539850141607e-07,
      "loss": 0.1706,
      "step": 11588
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2518529620532008,
      "learning_rate": 1.3864227600291824e-07,
      "loss": 0.1467,
      "step": 11589
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3189854991732242,
      "learning_rate": 1.383394798620935e-07,
      "loss": 0.1803,
      "step": 11590
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.335237866046854,
      "learning_rate": 1.380370100992695e-07,
      "loss": 0.1859,
      "step": 11591
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2714991006837475,
      "learning_rate": 1.3773486673475277e-07,
      "loss": 0.1474,
      "step": 11592
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.479207314942831,
      "learning_rate": 1.3743304978882543e-07,
      "loss": 0.1363,
      "step": 11593
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5434757045523793,
      "learning_rate": 1.3713155928175183e-07,
      "loss": 0.1942,
      "step": 11594
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.36410381815116,
      "learning_rate": 1.3683039523376962e-07,
      "loss": 0.1723,
      "step": 11595
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1967038207265357,
      "learning_rate": 1.3652955766509768e-07,
      "loss": 0.1015,
      "step": 11596
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5589894916195683,
      "learning_rate": 1.3622904659593316e-07,
      "loss": 0.2017,
      "step": 11597
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.100457027645962,
      "learning_rate": 1.359288620464483e-07,
      "loss": 0.4993,
      "step": 11598
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.594524097136254,
      "learning_rate": 1.356290040367969e-07,
      "loss": 0.6542,
      "step": 11599
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5272223316091265,
      "learning_rate": 1.3532947258710905e-07,
      "loss": 0.2064,
      "step": 11600
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2770427248546683,
      "learning_rate": 1.3503026771749305e-07,
      "loss": 0.195,
      "step": 11601
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3349588407368802,
      "learning_rate": 1.347313894480351e-07,
      "loss": 0.138,
      "step": 11602
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.478920998702164,
      "learning_rate": 1.3443283779880134e-07,
      "loss": 0.1978,
      "step": 11603
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.413244687397578,
      "learning_rate": 1.3413461278983241e-07,
      "loss": 0.1481,
      "step": 11604
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5458598533427121,
      "learning_rate": 1.3383671444114953e-07,
      "loss": 0.1971,
      "step": 11605
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2996549683554335,
      "learning_rate": 1.3353914277275282e-07,
      "loss": 0.1507,
      "step": 11606
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3516846603688444,
      "learning_rate": 1.3324189780461684e-07,
      "loss": 0.1942,
      "step": 11607
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3780526072681978,
      "learning_rate": 1.3294497955669894e-07,
      "loss": 0.1446,
      "step": 11608
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2647168685219725,
      "learning_rate": 1.32648388048931e-07,
      "loss": 0.1692,
      "step": 11609
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3195374089670084,
      "learning_rate": 1.3235212330122425e-07,
      "loss": 0.1661,
      "step": 11610
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5100444120355025,
      "learning_rate": 1.320561853334684e-07,
      "loss": 0.1762,
      "step": 11611
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3948257857317143,
      "learning_rate": 1.3176057416552924e-07,
      "loss": 0.1482,
      "step": 11612
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5381756774677289,
      "learning_rate": 1.314652898172536e-07,
      "loss": 0.1767,
      "step": 11613
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.182517359519336,
      "learning_rate": 1.3117033230846398e-07,
      "loss": 0.5426,
      "step": 11614
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5295353551751703,
      "learning_rate": 1.308757016589618e-07,
      "loss": 0.1819,
      "step": 11615
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2889947658943508,
      "learning_rate": 1.3058139788852563e-07,
      "loss": 0.1384,
      "step": 11616
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.281284381654974,
      "learning_rate": 1.3028742101691583e-07,
      "loss": 0.1463,
      "step": 11617
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4299306856534115,
      "learning_rate": 1.2999377106386492e-07,
      "loss": 0.1594,
      "step": 11618
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5363362481854856,
      "learning_rate": 1.2970044804908766e-07,
      "loss": 0.1932,
      "step": 11619
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.375480237019025,
      "learning_rate": 1.2940745199227666e-07,
      "loss": 0.1743,
      "step": 11620
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2567849833924285,
      "learning_rate": 1.2911478291309954e-07,
      "loss": 0.1589,
      "step": 11621
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3560608629829285,
      "learning_rate": 1.288224408312061e-07,
      "loss": 0.1962,
      "step": 11622
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4764825798809136,
      "learning_rate": 1.2853042576622066e-07,
      "loss": 0.2005,
      "step": 11623
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5354825149451514,
      "learning_rate": 1.2823873773774808e-07,
      "loss": 0.1706,
      "step": 11624
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.295288288789178,
      "learning_rate": 1.2794737676536993e-07,
      "loss": 0.1621,
      "step": 11625
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4545686386522088,
      "learning_rate": 1.2765634286864614e-07,
      "loss": 0.2197,
      "step": 11626
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4600296111379967,
      "learning_rate": 1.2736563606711384e-07,
      "loss": 0.1664,
      "step": 11627
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.214750667177578,
      "learning_rate": 1.270752563802913e-07,
      "loss": 0.1521,
      "step": 11628
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.318386643133485,
      "learning_rate": 1.2678520382767068e-07,
      "loss": 0.1536,
      "step": 11629
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3030253432667387,
      "learning_rate": 1.2649547842872367e-07,
      "loss": 0.1772,
      "step": 11630
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2744554417542768,
      "learning_rate": 1.262060802029025e-07,
      "loss": 0.1832,
      "step": 11631
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.293691205806851,
      "learning_rate": 1.2591700916963323e-07,
      "loss": 0.1906,
      "step": 11632
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.130174241335614,
      "learning_rate": 1.2562826534832372e-07,
      "loss": 0.1289,
      "step": 11633
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.276443191441171,
      "learning_rate": 1.2533984875835737e-07,
      "loss": 0.1933,
      "step": 11634
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3091566107207626,
      "learning_rate": 1.2505175941909642e-07,
      "loss": 0.1717,
      "step": 11635
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3599025772734412,
      "learning_rate": 1.2476399734988155e-07,
      "loss": 0.1699,
      "step": 11636
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3039832224328505,
      "learning_rate": 1.244765625700306e-07,
      "loss": 0.1861,
      "step": 11637
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.452215308739537,
      "learning_rate": 1.2418945509884095e-07,
      "loss": 0.1869,
      "step": 11638
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3969264812006772,
      "learning_rate": 1.239026749555855e-07,
      "loss": 0.1504,
      "step": 11639
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.031209296605902,
      "learning_rate": 1.2361622215951774e-07,
      "loss": 0.4558,
      "step": 11640
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.773914536683679,
      "learning_rate": 1.2333009672986784e-07,
      "loss": 0.6375,
      "step": 11641
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1204355130934291,
      "learning_rate": 1.2304429868584488e-07,
      "loss": 0.144,
      "step": 11642
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4784018417360747,
      "learning_rate": 1.2275882804663354e-07,
      "loss": 0.2096,
      "step": 11643
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.247350831220143,
      "learning_rate": 1.224736848314001e-07,
      "loss": 0.1651,
      "step": 11644
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3814533502398265,
      "learning_rate": 1.2218886905928652e-07,
      "loss": 0.1747,
      "step": 11645
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4813408760692128,
      "learning_rate": 1.2190438074941357e-07,
      "loss": 0.1965,
      "step": 11646
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5130654882098742,
      "learning_rate": 1.2162021992087934e-07,
      "loss": 0.1648,
      "step": 11647
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3224570741428938,
      "learning_rate": 1.2133638659276082e-07,
      "loss": 0.1456,
      "step": 11648
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3349571583502873,
      "learning_rate": 1.2105288078411214e-07,
      "loss": 0.1558,
      "step": 11649
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2338567122692776,
      "learning_rate": 1.2076970251396593e-07,
      "loss": 0.144,
      "step": 11650
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3958895075070241,
      "learning_rate": 1.2048685180133412e-07,
      "loss": 0.1649,
      "step": 11651
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4010496539570263,
      "learning_rate": 1.2020432866520325e-07,
      "loss": 0.1849,
      "step": 11652
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.503249879452813,
      "learning_rate": 1.199221331245415e-07,
      "loss": 0.1734,
      "step": 11653
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3575245845949964,
      "learning_rate": 1.1964026519829307e-07,
      "loss": 0.1626,
      "step": 11654
    },
    {
      "epoch": 0.93,
      "grad_norm": 11.979808973539093,
      "learning_rate": 1.193587249053807e-07,
      "loss": 0.5021,
      "step": 11655
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3907039782940525,
      "learning_rate": 1.1907751226470421e-07,
      "loss": 0.1872,
      "step": 11656
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4467321057726041,
      "learning_rate": 1.1879662729514352e-07,
      "loss": 0.1727,
      "step": 11657
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.143763110836109,
      "learning_rate": 1.1851607001555465e-07,
      "loss": 0.1349,
      "step": 11658
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4173322095061018,
      "learning_rate": 1.1823584044477199e-07,
      "loss": 0.202,
      "step": 11659
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3626871066482655,
      "learning_rate": 1.179559386016088e-07,
      "loss": 0.1586,
      "step": 11660
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3310918363445101,
      "learning_rate": 1.1767636450485453e-07,
      "loss": 0.1809,
      "step": 11661
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6161479866087205,
      "learning_rate": 1.1739711817328025e-07,
      "loss": 0.1901,
      "step": 11662
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2339997782035526,
      "learning_rate": 1.1711819962562987e-07,
      "loss": 0.1466,
      "step": 11663
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.327590247981248,
      "learning_rate": 1.1683960888063006e-07,
      "loss": 0.1471,
      "step": 11664
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2278666142565842,
      "learning_rate": 1.1656134595698309e-07,
      "loss": 0.1454,
      "step": 11665
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2766610652719923,
      "learning_rate": 1.1628341087336792e-07,
      "loss": 0.1124,
      "step": 11666
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2098282132851113,
      "learning_rate": 1.1600580364844516e-07,
      "loss": 0.124,
      "step": 11667
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2536296093800676,
      "learning_rate": 1.1572852430085103e-07,
      "loss": 0.1624,
      "step": 11668
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4224677729078656,
      "learning_rate": 1.154515728491995e-07,
      "loss": 0.1932,
      "step": 11669
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4758594928094455,
      "learning_rate": 1.151749493120835e-07,
      "loss": 0.1956,
      "step": 11670
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3267670904776225,
      "learning_rate": 1.1489865370807318e-07,
      "loss": 0.168,
      "step": 11671
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4064444632889166,
      "learning_rate": 1.1462268605571814e-07,
      "loss": 0.1502,
      "step": 11672
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2044490560652903,
      "learning_rate": 1.143470463735441e-07,
      "loss": 0.1525,
      "step": 11673
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3548379019308125,
      "learning_rate": 1.1407173468005572e-07,
      "loss": 0.1931,
      "step": 11674
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3123905965422187,
      "learning_rate": 1.1379675099373489e-07,
      "loss": 0.167,
      "step": 11675
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.364710543413833,
      "learning_rate": 1.1352209533304404e-07,
      "loss": 0.1612,
      "step": 11676
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2119668242147925,
      "learning_rate": 1.1324776771641898e-07,
      "loss": 0.1218,
      "step": 11677
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4953614485925484,
      "learning_rate": 1.1297376816227834e-07,
      "loss": 0.185,
      "step": 11678
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3725143923522227,
      "learning_rate": 1.1270009668901571e-07,
      "loss": 0.1674,
      "step": 11679
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.866969719005281,
      "learning_rate": 1.1242675331500363e-07,
      "loss": 0.4868,
      "step": 11680
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1670670071655291,
      "learning_rate": 1.1215373805859242e-07,
      "loss": 0.1109,
      "step": 11681
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3660391574464552,
      "learning_rate": 1.118810509381102e-07,
      "loss": 0.1512,
      "step": 11682
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.219007255532347,
      "learning_rate": 1.1160869197186342e-07,
      "loss": 0.1616,
      "step": 11683
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4423812245767982,
      "learning_rate": 1.1133666117813635e-07,
      "loss": 0.1545,
      "step": 11684
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.7599134179340705,
      "learning_rate": 1.1106495857519162e-07,
      "loss": 0.5044,
      "step": 11685
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.27294324674106,
      "learning_rate": 1.1079358418126795e-07,
      "loss": 0.1245,
      "step": 11686
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.400516955577557,
      "learning_rate": 1.1052253801458634e-07,
      "loss": 0.1793,
      "step": 11687
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.709964790224891,
      "learning_rate": 1.1025182009334001e-07,
      "loss": 0.1771,
      "step": 11688
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.476409105865249,
      "learning_rate": 1.0998143043570441e-07,
      "loss": 0.1754,
      "step": 11689
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2376053220258127,
      "learning_rate": 1.0971136905983282e-07,
      "loss": 0.1481,
      "step": 11690
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3058320902057976,
      "learning_rate": 1.0944163598385237e-07,
      "loss": 0.145,
      "step": 11691
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4384127158071447,
      "learning_rate": 1.0917223122587361e-07,
      "loss": 0.1745,
      "step": 11692
    },
    {
      "epoch": 0.94,
      "grad_norm": 14.913152952662157,
      "learning_rate": 1.089031548039815e-07,
      "loss": 0.6187,
      "step": 11693
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3470912575561913,
      "learning_rate": 1.0863440673623992e-07,
      "loss": 0.1646,
      "step": 11694
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.441832855824926,
      "learning_rate": 1.0836598704069057e-07,
      "loss": 0.1645,
      "step": 11695
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5576253411749779,
      "learning_rate": 1.0809789573535456e-07,
      "loss": 0.1778,
      "step": 11696
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4154872282832505,
      "learning_rate": 1.0783013283822752e-07,
      "loss": 0.1779,
      "step": 11697
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.382538972798503,
      "learning_rate": 1.0756269836728672e-07,
      "loss": 0.1805,
      "step": 11698
    },
    {
      "epoch": 0.94,
      "grad_norm": 56.269954542126946,
      "learning_rate": 1.0729559234048615e-07,
      "loss": 0.698,
      "step": 11699
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.255364273374407,
      "learning_rate": 1.0702881477575589e-07,
      "loss": 0.1691,
      "step": 11700
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1659005639972897,
      "learning_rate": 1.0676236569100718e-07,
      "loss": 0.1701,
      "step": 11701
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3542944143413516,
      "learning_rate": 1.0649624510412625e-07,
      "loss": 0.1702,
      "step": 11702
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.310998641833066,
      "learning_rate": 1.0623045303297941e-07,
      "loss": 0.1519,
      "step": 11703
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2372867607674312,
      "learning_rate": 1.0596498949540957e-07,
      "loss": 0.1632,
      "step": 11704
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3983616959310992,
      "learning_rate": 1.0569985450923803e-07,
      "loss": 0.1875,
      "step": 11705
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5547153997130903,
      "learning_rate": 1.0543504809226446e-07,
      "loss": 0.1753,
      "step": 11706
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.6434501216870006,
      "learning_rate": 1.0517057026226685e-07,
      "loss": 0.5303,
      "step": 11707
    },
    {
      "epoch": 0.94,
      "grad_norm": 74.11884199535422,
      "learning_rate": 1.0490642103699933e-07,
      "loss": 0.5841,
      "step": 11708
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2098872102968024,
      "learning_rate": 1.0464260043419438e-07,
      "loss": 0.1374,
      "step": 11709
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3595075573672604,
      "learning_rate": 1.0437910847156507e-07,
      "loss": 0.1941,
      "step": 11710
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2432710661925528,
      "learning_rate": 1.0411594516679835e-07,
      "loss": 0.1342,
      "step": 11711
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4001168094389072,
      "learning_rate": 1.0385311053756287e-07,
      "loss": 0.2087,
      "step": 11712
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.143803182788164,
      "learning_rate": 1.0359060460150283e-07,
      "loss": 0.1195,
      "step": 11713
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6966454642514566,
      "learning_rate": 1.0332842737624082e-07,
      "loss": 0.2324,
      "step": 11714
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4257000685944459,
      "learning_rate": 1.0306657887937833e-07,
      "loss": 0.1423,
      "step": 11715
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.33784751745277,
      "learning_rate": 1.0280505912849293e-07,
      "loss": 0.1877,
      "step": 11716
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6299263861555227,
      "learning_rate": 1.0254386814114226e-07,
      "loss": 0.2113,
      "step": 11717
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2666545341317765,
      "learning_rate": 1.0228300593486007e-07,
      "loss": 0.1752,
      "step": 11718
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2743314990649175,
      "learning_rate": 1.0202247252715902e-07,
      "loss": 0.178,
      "step": 11719
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8066721037296976,
      "learning_rate": 1.0176226793552957e-07,
      "loss": 0.1657,
      "step": 11720
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2448332861839368,
      "learning_rate": 1.0150239217744106e-07,
      "loss": 0.152,
      "step": 11721
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4096673461971014,
      "learning_rate": 1.0124284527033789e-07,
      "loss": 0.1482,
      "step": 11722
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3754035706029113,
      "learning_rate": 1.0098362723164556e-07,
      "loss": 0.1816,
      "step": 11723
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2840843206080021,
      "learning_rate": 1.007247380787657e-07,
      "loss": 0.1273,
      "step": 11724
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2431449712110232,
      "learning_rate": 1.004661778290783e-07,
      "loss": 0.1545,
      "step": 11725
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3662559613394658,
      "learning_rate": 1.002079464999417e-07,
      "loss": 0.1664,
      "step": 11726
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2559547986400617,
      "learning_rate": 9.995004410869147e-08,
      "loss": 0.1456,
      "step": 11727
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.763105304067645,
      "learning_rate": 9.969247067264099e-08,
      "loss": 0.5072,
      "step": 11728
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4274905041538866,
      "learning_rate": 9.943522620908253e-08,
      "loss": 0.1472,
      "step": 11729
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3271281774266643,
      "learning_rate": 9.917831073528506e-08,
      "loss": 0.1463,
      "step": 11730
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4211394312039303,
      "learning_rate": 9.892172426849644e-08,
      "loss": 0.203,
      "step": 11731
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.455423329644393,
      "learning_rate": 9.866546682594236e-08,
      "loss": 0.5779,
      "step": 11732
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2713255801559635,
      "learning_rate": 9.840953842482682e-08,
      "loss": 0.1225,
      "step": 11733
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7715292316507498,
      "learning_rate": 9.815393908232885e-08,
      "loss": 0.1797,
      "step": 11734
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3863715798558283,
      "learning_rate": 9.789866881560971e-08,
      "loss": 0.164,
      "step": 11735
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4807437787339865,
      "learning_rate": 9.764372764180519e-08,
      "loss": 0.1618,
      "step": 11736
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4400704952228964,
      "learning_rate": 9.738911557803154e-08,
      "loss": 0.1599,
      "step": 11737
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4236330699601027,
      "learning_rate": 9.713483264138013e-08,
      "loss": 0.1871,
      "step": 11738
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3618291288221354,
      "learning_rate": 9.688087884892284e-08,
      "loss": 0.1721,
      "step": 11739
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4380555809988866,
      "learning_rate": 9.66272542177077e-08,
      "loss": 0.185,
      "step": 11740
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.33692767645177,
      "learning_rate": 9.637395876476219e-08,
      "loss": 0.1662,
      "step": 11741
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.414480408177421,
      "learning_rate": 9.612099250708995e-08,
      "loss": 0.5949,
      "step": 11742
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.291475200489871,
      "learning_rate": 9.586835546167351e-08,
      "loss": 0.164,
      "step": 11743
    },
    {
      "epoch": 0.94,
      "grad_norm": 14.460021615551394,
      "learning_rate": 9.561604764547371e-08,
      "loss": 0.6736,
      "step": 11744
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6068235874191255,
      "learning_rate": 9.536406907542761e-08,
      "loss": 0.1803,
      "step": 11745
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4638836463804885,
      "learning_rate": 9.511241976845276e-08,
      "loss": 0.1744,
      "step": 11746
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6782197776043437,
      "learning_rate": 9.486109974144175e-08,
      "loss": 0.1939,
      "step": 11747
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2861538522534006,
      "learning_rate": 9.461010901126777e-08,
      "loss": 0.1539,
      "step": 11748
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3761552758320768,
      "learning_rate": 9.435944759477955e-08,
      "loss": 0.1536,
      "step": 11749
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2343077190634686,
      "learning_rate": 9.410911550880474e-08,
      "loss": 0.155,
      "step": 11750
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.698658656326484,
      "learning_rate": 9.385911277014991e-08,
      "loss": 0.4707,
      "step": 11751
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5147066198744654,
      "learning_rate": 9.360943939559774e-08,
      "loss": 0.1757,
      "step": 11752
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4972340936540334,
      "learning_rate": 9.336009540190927e-08,
      "loss": 0.1639,
      "step": 11753
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5525489449835235,
      "learning_rate": 9.311108080582387e-08,
      "loss": 0.14,
      "step": 11754
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.309520514414241,
      "learning_rate": 9.286239562405985e-08,
      "loss": 0.153,
      "step": 11755
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5327706412103699,
      "learning_rate": 9.261403987331052e-08,
      "loss": 0.193,
      "step": 11756
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.37571221931503,
      "learning_rate": 9.236601357024977e-08,
      "loss": 0.1653,
      "step": 11757
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4567987096461017,
      "learning_rate": 9.211831673152816e-08,
      "loss": 0.1538,
      "step": 11758
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6904854909974252,
      "learning_rate": 9.187094937377406e-08,
      "loss": 0.2043,
      "step": 11759
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2490973624499306,
      "learning_rate": 9.162391151359417e-08,
      "loss": 0.1534,
      "step": 11760
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5479756034338594,
      "learning_rate": 9.137720316757303e-08,
      "loss": 0.1856,
      "step": 11761
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2745122005590874,
      "learning_rate": 9.113082435227294e-08,
      "loss": 0.194,
      "step": 11762
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.421100319367278,
      "learning_rate": 9.088477508423343e-08,
      "loss": 0.2067,
      "step": 11763
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.266449547376562,
      "learning_rate": 9.063905537997352e-08,
      "loss": 0.6652,
      "step": 11764
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3450258706195086,
      "learning_rate": 9.039366525598781e-08,
      "loss": 0.1688,
      "step": 11765
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5287419179764004,
      "learning_rate": 9.014860472875197e-08,
      "loss": 0.1925,
      "step": 11766
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4424855994443555,
      "learning_rate": 8.99038738147151e-08,
      "loss": 0.183,
      "step": 11767
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.38429330711202,
      "learning_rate": 8.965947253030904e-08,
      "loss": 0.2049,
      "step": 11768
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4556569867386469,
      "learning_rate": 8.941540089194067e-08,
      "loss": 0.1747,
      "step": 11769
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.860342207838784,
      "learning_rate": 8.917165891599467e-08,
      "loss": 0.5741,
      "step": 11770
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.839762515804097,
      "learning_rate": 8.892824661883403e-08,
      "loss": 0.3994,
      "step": 11771
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.359713427083873,
      "learning_rate": 8.868516401680072e-08,
      "loss": 0.1655,
      "step": 11772
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4144130968296706,
      "learning_rate": 8.844241112621277e-08,
      "loss": 0.1389,
      "step": 11773
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4500565876798737,
      "learning_rate": 8.81999879633666e-08,
      "loss": 0.2155,
      "step": 11774
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4172840698664086,
      "learning_rate": 8.795789454453862e-08,
      "loss": 0.2117,
      "step": 11775
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5215896066552475,
      "learning_rate": 8.771613088597863e-08,
      "loss": 0.1756,
      "step": 11776
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3608572549513422,
      "learning_rate": 8.74746970039192e-08,
      "loss": 0.1778,
      "step": 11777
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.491995202689277,
      "learning_rate": 8.723359291456846e-08,
      "loss": 0.1617,
      "step": 11778
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3943551231388651,
      "learning_rate": 8.69928186341107e-08,
      "loss": 0.1574,
      "step": 11779
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.278201244907948,
      "learning_rate": 8.675237417871075e-08,
      "loss": 0.1174,
      "step": 11780
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.268493452103561,
      "learning_rate": 8.65122595645107e-08,
      "loss": 0.1539,
      "step": 11781
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4604603257923368,
      "learning_rate": 8.627247480763046e-08,
      "loss": 0.1778,
      "step": 11782
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3686321563197972,
      "learning_rate": 8.603301992416658e-08,
      "loss": 0.1758,
      "step": 11783
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1786353006275547,
      "learning_rate": 8.579389493019507e-08,
      "loss": 0.1223,
      "step": 11784
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4074305568440126,
      "learning_rate": 8.555509984176812e-08,
      "loss": 0.1761,
      "step": 11785
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.188634840818344,
      "learning_rate": 8.531663467491846e-08,
      "loss": 0.1489,
      "step": 11786
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2663873230029539,
      "learning_rate": 8.507849944565327e-08,
      "loss": 0.1325,
      "step": 11787
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2603704199895884,
      "learning_rate": 8.484069416995977e-08,
      "loss": 0.1229,
      "step": 11788
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3655044245529577,
      "learning_rate": 8.460321886380407e-08,
      "loss": 0.1986,
      "step": 11789
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.443027256509294,
      "learning_rate": 8.43660735431262e-08,
      "loss": 0.1683,
      "step": 11790
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4153268601765026,
      "learning_rate": 8.412925822384788e-08,
      "loss": 0.1626,
      "step": 11791
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6689826618293173,
      "learning_rate": 8.389277292186692e-08,
      "loss": 0.1604,
      "step": 11792
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3813344447179734,
      "learning_rate": 8.365661765305955e-08,
      "loss": 0.1955,
      "step": 11793
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.317158557669469,
      "learning_rate": 8.342079243327972e-08,
      "loss": 0.1912,
      "step": 11794
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.348288666630903,
      "learning_rate": 8.318529727835811e-08,
      "loss": 0.1914,
      "step": 11795
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3701256826982715,
      "learning_rate": 8.295013220410486e-08,
      "loss": 0.1792,
      "step": 11796
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7469704833515158,
      "learning_rate": 8.271529722630788e-08,
      "loss": 0.2075,
      "step": 11797
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3642710142341172,
      "learning_rate": 8.248079236073125e-08,
      "loss": 0.1616,
      "step": 11798
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2114330681994019,
      "learning_rate": 8.224661762311847e-08,
      "loss": 0.1528,
      "step": 11799
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7523648079478362,
      "learning_rate": 8.201277302919086e-08,
      "loss": 0.1645,
      "step": 11800
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2202036558095306,
      "learning_rate": 8.177925859464587e-08,
      "loss": 0.1582,
      "step": 11801
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4181966397972763,
      "learning_rate": 8.154607433516149e-08,
      "loss": 0.145,
      "step": 11802
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.383059796788842,
      "learning_rate": 8.131322026639187e-08,
      "loss": 0.4503,
      "step": 11803
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.25872582386875,
      "learning_rate": 8.108069640396843e-08,
      "loss": 0.1344,
      "step": 11804
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.213188701177112,
      "learning_rate": 8.084850276350142e-08,
      "loss": 0.5332,
      "step": 11805
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3184609466322557,
      "learning_rate": 8.061663936057895e-08,
      "loss": 0.15,
      "step": 11806
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4008237022365766,
      "learning_rate": 8.038510621076689e-08,
      "loss": 0.192,
      "step": 11807
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.302024960927299,
      "learning_rate": 8.015390332960782e-08,
      "loss": 0.158,
      "step": 11808
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.412139565159276,
      "learning_rate": 7.992303073262431e-08,
      "loss": 0.1361,
      "step": 11809
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.476782233975,
      "learning_rate": 7.969248843531452e-08,
      "loss": 0.1818,
      "step": 11810
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.098117317268838,
      "learning_rate": 7.946227645315663e-08,
      "loss": 0.619,
      "step": 11811
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5588694507718681,
      "learning_rate": 7.923239480160439e-08,
      "loss": 0.1535,
      "step": 11812
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3787995921824425,
      "learning_rate": 7.900284349609044e-08,
      "loss": 0.1798,
      "step": 11813
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4868015954810596,
      "learning_rate": 7.877362255202636e-08,
      "loss": 0.1964,
      "step": 11814
    },
    {
      "epoch": 0.95,
      "grad_norm": 31.725333076294138,
      "learning_rate": 7.854473198479928e-08,
      "loss": 0.5089,
      "step": 11815
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.594085304985128,
      "learning_rate": 7.83161718097758e-08,
      "loss": 0.5377,
      "step": 11816
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5393466634921353,
      "learning_rate": 7.808794204229975e-08,
      "loss": 0.1999,
      "step": 11817
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2690071632713675,
      "learning_rate": 7.786004269769331e-08,
      "loss": 0.1464,
      "step": 11818
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5224477129625908,
      "learning_rate": 7.763247379125482e-08,
      "loss": 0.1748,
      "step": 11819
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3937655334728871,
      "learning_rate": 7.740523533826372e-08,
      "loss": 0.1618,
      "step": 11820
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3150660386909798,
      "learning_rate": 7.717832735397335e-08,
      "loss": 0.1594,
      "step": 11821
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.386233364795769,
      "learning_rate": 7.695174985361708e-08,
      "loss": 0.181,
      "step": 11822
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3181758694792953,
      "learning_rate": 7.672550285240721e-08,
      "loss": 0.1215,
      "step": 11823
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5427027537448201,
      "learning_rate": 7.649958636552989e-08,
      "loss": 0.1674,
      "step": 11824
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2628763141232395,
      "learning_rate": 7.627400040815414e-08,
      "loss": 0.1572,
      "step": 11825
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3494593233701575,
      "learning_rate": 7.604874499542225e-08,
      "loss": 0.1713,
      "step": 11826
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.320751195883479,
      "learning_rate": 7.582382014245771e-08,
      "loss": 0.1328,
      "step": 11827
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.372544392211871,
      "learning_rate": 7.559922586435953e-08,
      "loss": 0.1969,
      "step": 11828
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4500060332112423,
      "learning_rate": 7.537496217620566e-08,
      "loss": 0.1852,
      "step": 11829
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4611062718142405,
      "learning_rate": 7.515102909305128e-08,
      "loss": 0.1972,
      "step": 11830
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2803739504199134,
      "learning_rate": 7.492742662993047e-08,
      "loss": 0.1655,
      "step": 11831
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4299403887454627,
      "learning_rate": 7.470415480185345e-08,
      "loss": 0.1638,
      "step": 11832
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.465899490176388,
      "learning_rate": 7.448121362380989e-08,
      "loss": 0.5181,
      "step": 11833
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2583204024744978,
      "learning_rate": 7.425860311076616e-08,
      "loss": 0.1286,
      "step": 11834
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4264964072415753,
      "learning_rate": 7.403632327766641e-08,
      "loss": 0.1761,
      "step": 11835
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.467687939329435,
      "learning_rate": 7.381437413943372e-08,
      "loss": 0.5684,
      "step": 11836
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.639114553828653,
      "learning_rate": 7.35927557109678e-08,
      "loss": 0.5775,
      "step": 11837
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3258592943825107,
      "learning_rate": 7.337146800714678e-08,
      "loss": 0.1398,
      "step": 11838
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2688494590841974,
      "learning_rate": 7.3150511042826e-08,
      "loss": 0.1622,
      "step": 11839
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4422449900031205,
      "learning_rate": 7.292988483283913e-08,
      "loss": 0.1753,
      "step": 11840
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3807472280583288,
      "learning_rate": 7.270958939199767e-08,
      "loss": 0.1508,
      "step": 11841
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.683136374884187,
      "learning_rate": 7.248962473509091e-08,
      "loss": 0.4974,
      "step": 11842
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.38674509247126,
      "learning_rate": 7.22699908768848e-08,
      "loss": 0.1527,
      "step": 11843
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4087857088589786,
      "learning_rate": 7.205068783212476e-08,
      "loss": 0.255,
      "step": 11844
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2762590965064187,
      "learning_rate": 7.183171561553349e-08,
      "loss": 0.1533,
      "step": 11845
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3487275156130714,
      "learning_rate": 7.16130742418103e-08,
      "loss": 0.1818,
      "step": 11846
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4676877901726253,
      "learning_rate": 7.139476372563403e-08,
      "loss": 0.1668,
      "step": 11847
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2601961299799058,
      "learning_rate": 7.117678408166128e-08,
      "loss": 0.1663,
      "step": 11848
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6415343775013025,
      "learning_rate": 7.09591353245237e-08,
      "loss": 0.1893,
      "step": 11849
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2609758007419987,
      "learning_rate": 7.074181746883402e-08,
      "loss": 0.1663,
      "step": 11850
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4582308333366587,
      "learning_rate": 7.052483052918113e-08,
      "loss": 0.1651,
      "step": 11851
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.668328731260004,
      "learning_rate": 7.030817452013227e-08,
      "loss": 0.1898,
      "step": 11852
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2341059693512457,
      "learning_rate": 7.009184945623193e-08,
      "loss": 0.1391,
      "step": 11853
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2778406269058784,
      "learning_rate": 6.987585535200292e-08,
      "loss": 0.1675,
      "step": 11854
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.834718743366357,
      "learning_rate": 6.966019222194531e-08,
      "loss": 0.5617,
      "step": 11855
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.462109830434754,
      "learning_rate": 6.944486008053697e-08,
      "loss": 0.1685,
      "step": 11856
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4835814478188611,
      "learning_rate": 6.922985894223467e-08,
      "loss": 0.2032,
      "step": 11857
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3265635060824332,
      "learning_rate": 6.901518882147129e-08,
      "loss": 0.1888,
      "step": 11858
    },
    {
      "epoch": 0.95,
      "grad_norm": 11.46863593758475,
      "learning_rate": 6.880084973265866e-08,
      "loss": 0.6125,
      "step": 11859
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.534773571685166,
      "learning_rate": 6.858684169018581e-08,
      "loss": 0.5993,
      "step": 11860
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3396505072731482,
      "learning_rate": 6.837316470842015e-08,
      "loss": 0.166,
      "step": 11861
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3750438743016877,
      "learning_rate": 6.815981880170575e-08,
      "loss": 0.1808,
      "step": 11862
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4581663242174605,
      "learning_rate": 6.794680398436615e-08,
      "loss": 0.2204,
      "step": 11863
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4268435317041939,
      "learning_rate": 6.773412027070104e-08,
      "loss": 0.2123,
      "step": 11864
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2488165644565898,
      "learning_rate": 6.752176767498841e-08,
      "loss": 0.1628,
      "step": 11865
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.475039661974663,
      "learning_rate": 6.730974621148412e-08,
      "loss": 0.1433,
      "step": 11866
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4387197920466739,
      "learning_rate": 6.709805589442175e-08,
      "loss": 0.1684,
      "step": 11867
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.5563961475422365,
      "learning_rate": 6.688669673801384e-08,
      "loss": 0.509,
      "step": 11868
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4809665338888722,
      "learning_rate": 6.667566875644849e-08,
      "loss": 0.1954,
      "step": 11869
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3069074443785949,
      "learning_rate": 6.646497196389268e-08,
      "loss": 0.1613,
      "step": 11870
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.407974081476115,
      "learning_rate": 6.625460637449122e-08,
      "loss": 0.1916,
      "step": 11871
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2927187629684935,
      "learning_rate": 6.604457200236669e-08,
      "loss": 0.15,
      "step": 11872
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3222100886207386,
      "learning_rate": 6.58348688616195e-08,
      "loss": 0.1577,
      "step": 11873
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4637226745812413,
      "learning_rate": 6.56254969663278e-08,
      "loss": 0.1477,
      "step": 11874
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5566512797458902,
      "learning_rate": 6.54164563305465e-08,
      "loss": 0.1891,
      "step": 11875
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5186953975690822,
      "learning_rate": 6.52077469683099e-08,
      "loss": 0.1634,
      "step": 11876
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4582468980484613,
      "learning_rate": 6.499936889362956e-08,
      "loss": 0.1905,
      "step": 11877
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2695336439914797,
      "learning_rate": 6.479132212049322e-08,
      "loss": 0.1661,
      "step": 11878
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.311169568961537,
      "learning_rate": 6.458360666286912e-08,
      "loss": 0.1775,
      "step": 11879
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1074794040218978,
      "learning_rate": 6.437622253470055e-08,
      "loss": 0.1329,
      "step": 11880
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2269036820881958,
      "learning_rate": 6.416916974991083e-08,
      "loss": 0.1665,
      "step": 11881
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2298001006015544,
      "learning_rate": 6.396244832240051e-08,
      "loss": 0.1564,
      "step": 11882
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4068667335196685,
      "learning_rate": 6.375605826604514e-08,
      "loss": 0.1706,
      "step": 11883
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.585625138109625,
      "learning_rate": 6.354999959470254e-08,
      "loss": 0.5553,
      "step": 11884
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.564265137659678,
      "learning_rate": 6.334427232220552e-08,
      "loss": 0.1959,
      "step": 11885
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4280273058211257,
      "learning_rate": 6.313887646236472e-08,
      "loss": 0.1612,
      "step": 11886
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6124953048659518,
      "learning_rate": 6.293381202896964e-08,
      "loss": 0.1739,
      "step": 11887
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3286016519205572,
      "learning_rate": 6.272907903578595e-08,
      "loss": 0.1898,
      "step": 11888
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3640079049012501,
      "learning_rate": 6.252467749655878e-08,
      "loss": 0.1786,
      "step": 11889
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.39774889598795,
      "learning_rate": 6.23206074250099e-08,
      "loss": 0.1613,
      "step": 11890
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3539897184371086,
      "learning_rate": 6.211686883483947e-08,
      "loss": 0.1862,
      "step": 11891
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.274897312091227,
      "learning_rate": 6.191346173972434e-08,
      "loss": 0.147,
      "step": 11892
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.352203457863679,
      "learning_rate": 6.171038615332081e-08,
      "loss": 0.1443,
      "step": 11893
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4077795221133333,
      "learning_rate": 6.150764208926074e-08,
      "loss": 0.1855,
      "step": 11894
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3117932688014544,
      "learning_rate": 6.130522956115659e-08,
      "loss": 0.1545,
      "step": 11895
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.412363925100236,
      "learning_rate": 6.110314858259581e-08,
      "loss": 0.1709,
      "step": 11896
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.713954540133612,
      "learning_rate": 6.090139916714478e-08,
      "loss": 0.5904,
      "step": 11897
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.356462327846721,
      "learning_rate": 6.069998132834764e-08,
      "loss": 0.1846,
      "step": 11898
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5278864187242538,
      "learning_rate": 6.049889507972585e-08,
      "loss": 0.2077,
      "step": 11899
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5705808268346064,
      "learning_rate": 6.029814043478022e-08,
      "loss": 0.2153,
      "step": 11900
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3169297207329678,
      "learning_rate": 6.009771740698667e-08,
      "loss": 0.1655,
      "step": 11901
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3927743587432218,
      "learning_rate": 5.989762600980053e-08,
      "loss": 0.1671,
      "step": 11902
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2645494613998252,
      "learning_rate": 5.969786625665441e-08,
      "loss": 0.1747,
      "step": 11903
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.317143416391292,
      "learning_rate": 5.949843816095924e-08,
      "loss": 0.129,
      "step": 11904
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4721098787005424,
      "learning_rate": 5.9299341736103746e-08,
      "loss": 0.1591,
      "step": 11905
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4451669317684392,
      "learning_rate": 5.910057699545224e-08,
      "loss": 0.1729,
      "step": 11906
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.324290096574141,
      "learning_rate": 5.890214395235017e-08,
      "loss": 0.1881,
      "step": 11907
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.246805943837617,
      "learning_rate": 5.870404262011742e-08,
      "loss": 0.147,
      "step": 11908
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4247928029368102,
      "learning_rate": 5.850627301205391e-08,
      "loss": 0.1654,
      "step": 11909
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3842602430613966,
      "learning_rate": 5.8308835141436814e-08,
      "loss": 0.1807,
      "step": 11910
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4656827671721269,
      "learning_rate": 5.811172902151996e-08,
      "loss": 0.1831,
      "step": 11911
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5149523393238529,
      "learning_rate": 5.791495466553609e-08,
      "loss": 0.1899,
      "step": 11912
    },
    {
      "epoch": 0.95,
      "grad_norm": 11.8023149756254,
      "learning_rate": 5.7718512086695767e-08,
      "loss": 0.5914,
      "step": 11913
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.430295278233407,
      "learning_rate": 5.752240129818565e-08,
      "loss": 0.1888,
      "step": 11914
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2904451642914454,
      "learning_rate": 5.7326622313171877e-08,
      "loss": 0.1821,
      "step": 11915
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2898162873948038,
      "learning_rate": 5.713117514479838e-08,
      "loss": 0.1549,
      "step": 11916
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3696932179407268,
      "learning_rate": 5.6936059806184105e-08,
      "loss": 0.1694,
      "step": 11917
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2702318614737542,
      "learning_rate": 5.674127631043025e-08,
      "loss": 0.1674,
      "step": 11918
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3067137746286897,
      "learning_rate": 5.6546824670611344e-08,
      "loss": 0.1505,
      "step": 11919
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.238389726238456,
      "learning_rate": 5.6352704899782506e-08,
      "loss": 0.6668,
      "step": 11920
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3543000249610455,
      "learning_rate": 5.6158917010974976e-08,
      "loss": 0.1525,
      "step": 11921
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4162895508431155,
      "learning_rate": 5.596546101719835e-08,
      "loss": 0.1469,
      "step": 11922
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1779053378458708,
      "learning_rate": 5.5772336931440574e-08,
      "loss": 0.1284,
      "step": 11923
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3600622654393317,
      "learning_rate": 5.557954476666627e-08,
      "loss": 0.1888,
      "step": 11924
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.42920131246672,
      "learning_rate": 5.538708453581787e-08,
      "loss": 0.1851,
      "step": 11925
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3929477475156726,
      "learning_rate": 5.5194956251816144e-08,
      "loss": 0.1701,
      "step": 11926
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.608129123392848,
      "learning_rate": 5.500315992755911e-08,
      "loss": 0.7568,
      "step": 11927
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4447726585552805,
      "learning_rate": 5.481169557592258e-08,
      "loss": 0.2044,
      "step": 11928
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.127699067645563,
      "learning_rate": 5.462056320976072e-08,
      "loss": 0.4706,
      "step": 11929
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4406931273244223,
      "learning_rate": 5.442976284190382e-08,
      "loss": 0.1695,
      "step": 11930
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3103668551675502,
      "learning_rate": 5.423929448516218e-08,
      "loss": 0.1159,
      "step": 11931
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.1211141393635735,
      "learning_rate": 5.404915815232115e-08,
      "loss": 0.5285,
      "step": 11932
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4288727479488013,
      "learning_rate": 5.3859353856146045e-08,
      "loss": 0.1791,
      "step": 11933
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6192437324781481,
      "learning_rate": 5.366988160937836e-08,
      "loss": 0.1886,
      "step": 11934
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3116857609879293,
      "learning_rate": 5.348074142473847e-08,
      "loss": 0.174,
      "step": 11935
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2317871056191212,
      "learning_rate": 5.329193331492399e-08,
      "loss": 0.1352,
      "step": 11936
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3377019156556629,
      "learning_rate": 5.310345729260924e-08,
      "loss": 0.1618,
      "step": 11937
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4222184493637573,
      "learning_rate": 5.2915313370449083e-08,
      "loss": 0.19,
      "step": 11938
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.945399768370425,
      "learning_rate": 5.2727501561071756e-08,
      "loss": 0.6056,
      "step": 11939
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5389058375136386,
      "learning_rate": 5.254002187708773e-08,
      "loss": 0.1759,
      "step": 11940
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5082862149830094,
      "learning_rate": 5.235287433108194e-08,
      "loss": 0.1739,
      "step": 11941
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5273411009924767,
      "learning_rate": 5.216605893561877e-08,
      "loss": 0.1672,
      "step": 11942
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3497536718479062,
      "learning_rate": 5.1979575703239304e-08,
      "loss": 0.1888,
      "step": 11943
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.467161216185006,
      "learning_rate": 5.179342464646242e-08,
      "loss": 0.199,
      "step": 11944
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.984529937390512,
      "learning_rate": 5.16076057777859e-08,
      "loss": 0.6603,
      "step": 11945
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.388086282541771,
      "learning_rate": 5.142211910968309e-08,
      "loss": 0.4956,
      "step": 11946
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3552435997528884,
      "learning_rate": 5.1236964654607926e-08,
      "loss": 0.1267,
      "step": 11947
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2360965020268753,
      "learning_rate": 5.1052142424988794e-08,
      "loss": 0.1692,
      "step": 11948
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4755246890534195,
      "learning_rate": 5.086765243323466e-08,
      "loss": 0.1734,
      "step": 11949
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.64536004619589,
      "learning_rate": 5.068349469173006e-08,
      "loss": 0.5911,
      "step": 11950
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2712322513977827,
      "learning_rate": 5.0499669212837885e-08,
      "loss": 0.1573,
      "step": 11951
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4606907262031674,
      "learning_rate": 5.031617600889993e-08,
      "loss": 0.1559,
      "step": 11952
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.273862432883727,
      "learning_rate": 5.013301509223356e-08,
      "loss": 0.5911,
      "step": 11953
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2352277630418924,
      "learning_rate": 4.995018647513561e-08,
      "loss": 0.1494,
      "step": 11954
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3406570443905086,
      "learning_rate": 4.976769016987959e-08,
      "loss": 0.14,
      "step": 11955
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2840225368469602,
      "learning_rate": 4.958552618871737e-08,
      "loss": 0.1693,
      "step": 11956
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5847924636894588,
      "learning_rate": 4.94036945438775e-08,
      "loss": 0.195,
      "step": 11957
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.450570288825283,
      "learning_rate": 4.922219524756799e-08,
      "loss": 0.1965,
      "step": 11958
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5363409185710943,
      "learning_rate": 4.904102831197188e-08,
      "loss": 0.1951,
      "step": 11959
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2329498282943006,
      "learning_rate": 4.8860193749253324e-08,
      "loss": 0.1456,
      "step": 11960
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.438794220940196,
      "learning_rate": 4.867969157155095e-08,
      "loss": 0.1674,
      "step": 11961
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3859866160844427,
      "learning_rate": 4.8499521790982276e-08,
      "loss": 0.1747,
      "step": 11962
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3750328183286025,
      "learning_rate": 4.8319684419643743e-08,
      "loss": 0.1647,
      "step": 11963
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4195078026030556,
      "learning_rate": 4.814017946960792e-08,
      "loss": 0.1419,
      "step": 11964
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.315211731286358,
      "learning_rate": 4.79610069529246e-08,
      "loss": 0.1652,
      "step": 11965
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3367000069064292,
      "learning_rate": 4.7782166881623626e-08,
      "loss": 0.1553,
      "step": 11966
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3159252485269777,
      "learning_rate": 4.7603659267710376e-08,
      "loss": 0.157,
      "step": 11967
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.789139154827396,
      "learning_rate": 4.742548412316805e-08,
      "loss": 0.5509,
      "step": 11968
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2437922974328492,
      "learning_rate": 4.72476414599593e-08,
      "loss": 0.1419,
      "step": 11969
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6004520603837116,
      "learning_rate": 4.707013129002291e-08,
      "loss": 0.1565,
      "step": 11970
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.134726028223145,
      "learning_rate": 4.689295362527435e-08,
      "loss": 0.1429,
      "step": 11971
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3213673349821737,
      "learning_rate": 4.671610847761021e-08,
      "loss": 0.159,
      "step": 11972
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3062637064638105,
      "learning_rate": 4.6539595858900446e-08,
      "loss": 0.1371,
      "step": 11973
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3642930246811105,
      "learning_rate": 4.636341578099668e-08,
      "loss": 0.2032,
      "step": 11974
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4505218549250118,
      "learning_rate": 4.618756825572612e-08,
      "loss": 0.2039,
      "step": 11975
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3423592004613445,
      "learning_rate": 4.601205329489267e-08,
      "loss": 0.1705,
      "step": 11976
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5010634829434208,
      "learning_rate": 4.583687091028077e-08,
      "loss": 0.1986,
      "step": 11977
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2958664340074666,
      "learning_rate": 4.5662021113649923e-08,
      "loss": 0.1823,
      "step": 11978
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.043083292926528,
      "learning_rate": 4.5487503916738505e-08,
      "loss": 0.1252,
      "step": 11979
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.625813694686941,
      "learning_rate": 4.5313319331262703e-08,
      "loss": 0.2111,
      "step": 11980
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3601050371379158,
      "learning_rate": 4.513946736891595e-08,
      "loss": 0.1305,
      "step": 11981
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3581968669282554,
      "learning_rate": 4.4965948041369444e-08,
      "loss": 0.1687,
      "step": 11982
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2303874547425697,
      "learning_rate": 4.479276136027222e-08,
      "loss": 0.1481,
      "step": 11983
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3018819918256086,
      "learning_rate": 4.4619907337249965e-08,
      "loss": 0.1666,
      "step": 11984
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3012908546933881,
      "learning_rate": 4.4447385983907855e-08,
      "loss": 0.1699,
      "step": 11985
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3858201731481983,
      "learning_rate": 4.427519731182772e-08,
      "loss": 0.1389,
      "step": 11986
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1107173646406219,
      "learning_rate": 4.4103341332568105e-08,
      "loss": 0.1305,
      "step": 11987
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.37185997790486,
      "learning_rate": 4.393181805766755e-08,
      "loss": 0.5384,
      "step": 11988
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.904637254738123,
      "learning_rate": 4.3760627498640184e-08,
      "loss": 0.6962,
      "step": 11989
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2649148302185147,
      "learning_rate": 4.3589769666978476e-08,
      "loss": 0.163,
      "step": 11990
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4412159237437532,
      "learning_rate": 4.3419244574153255e-08,
      "loss": 0.1953,
      "step": 11991
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1417599908904636,
      "learning_rate": 4.324905223161202e-08,
      "loss": 0.1458,
      "step": 11992
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2401118467312509,
      "learning_rate": 4.3079192650779536e-08,
      "loss": 0.1541,
      "step": 11993
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5466035441811181,
      "learning_rate": 4.2909665843060555e-08,
      "loss": 0.2007,
      "step": 11994
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2994211061589582,
      "learning_rate": 4.274047181983487e-08,
      "loss": 0.158,
      "step": 11995
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6688625719378687,
      "learning_rate": 4.257161059246118e-08,
      "loss": 0.193,
      "step": 11996
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2440328455094596,
      "learning_rate": 4.24030821722754e-08,
      "loss": 0.125,
      "step": 11997
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3409149787992523,
      "learning_rate": 4.2234886570591824e-08,
      "loss": 0.1744,
      "step": 11998
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3023544062343446,
      "learning_rate": 4.206702379870198e-08,
      "loss": 0.1592,
      "step": 11999
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.349200978983085,
      "learning_rate": 4.189949386787462e-08,
      "loss": 0.1539,
      "step": 12000
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.966447204672004,
      "learning_rate": 4.173229678935631e-08,
      "loss": 0.4401,
      "step": 12001
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3377347632907357,
      "learning_rate": 4.156543257437196e-08,
      "loss": 0.1816,
      "step": 12002
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3720423091730143,
      "learning_rate": 4.1398901234124265e-08,
      "loss": 0.1776,
      "step": 12003
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.437915317907944,
      "learning_rate": 4.123270277979208e-08,
      "loss": 0.5207,
      "step": 12004
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.481365682065454,
      "learning_rate": 4.106683722253257e-08,
      "loss": 0.1809,
      "step": 12005
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3562969908639704,
      "learning_rate": 4.090130457348185e-08,
      "loss": 0.1841,
      "step": 12006
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.879694441676661,
      "learning_rate": 4.0736104843751014e-08,
      "loss": 0.4557,
      "step": 12007
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.354566403703353,
      "learning_rate": 4.057123804443231e-08,
      "loss": 0.1804,
      "step": 12008
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1399744381798957,
      "learning_rate": 4.040670418659243e-08,
      "loss": 0.1276,
      "step": 12009
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3858615599850823,
      "learning_rate": 4.024250328127755e-08,
      "loss": 0.2334,
      "step": 12010
    },
    {
      "epoch": 0.96,
      "grad_norm": 24.288813121532584,
      "learning_rate": 4.0078635339511065e-08,
      "loss": 0.5138,
      "step": 12011
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3163628676677386,
      "learning_rate": 3.991510037229363e-08,
      "loss": 0.1449,
      "step": 12012
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3322016445372824,
      "learning_rate": 3.9751898390604226e-08,
      "loss": 0.1725,
      "step": 12013
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4141994433896983,
      "learning_rate": 3.958902940539855e-08,
      "loss": 0.1782,
      "step": 12014
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4621149743937452,
      "learning_rate": 3.9426493427611177e-08,
      "loss": 0.1683,
      "step": 12015
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4422256309450874,
      "learning_rate": 3.926429046815228e-08,
      "loss": 0.1863,
      "step": 12016
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.329864987008123,
      "learning_rate": 3.9102420537913134e-08,
      "loss": 0.7039,
      "step": 12017
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.207100151076323,
      "learning_rate": 3.894088364775894e-08,
      "loss": 0.1506,
      "step": 12018
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4950051982668435,
      "learning_rate": 3.877967980853437e-08,
      "loss": 0.1861,
      "step": 12019
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.238555975896437,
      "learning_rate": 3.8618809031061855e-08,
      "loss": 0.5917,
      "step": 12020
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2985831890474746,
      "learning_rate": 3.845827132614166e-08,
      "loss": 0.151,
      "step": 12021
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2807882841052944,
      "learning_rate": 3.8298066704550165e-08,
      "loss": 0.1598,
      "step": 12022
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.508349957569687,
      "learning_rate": 3.8138195177042644e-08,
      "loss": 0.1848,
      "step": 12023
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.370408601266554,
      "learning_rate": 3.797865675435219e-08,
      "loss": 0.17,
      "step": 12024
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1601162191682797,
      "learning_rate": 3.781945144718912e-08,
      "loss": 0.1503,
      "step": 12025
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3392262274521287,
      "learning_rate": 3.766057926624045e-08,
      "loss": 0.1871,
      "step": 12026
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4056785724764818,
      "learning_rate": 3.750204022217263e-08,
      "loss": 0.1737,
      "step": 12027
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1851661030201108,
      "learning_rate": 3.7343834325628827e-08,
      "loss": 0.12,
      "step": 12028
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2453294102838228,
      "learning_rate": 3.7185961587229424e-08,
      "loss": 0.1512,
      "step": 12029
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3343456041599573,
      "learning_rate": 3.7028422017573175e-08,
      "loss": 0.1589,
      "step": 12030
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2237028374117966,
      "learning_rate": 3.687121562723661e-08,
      "loss": 0.1098,
      "step": 12031
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3551363731750161,
      "learning_rate": 3.6714342426771856e-08,
      "loss": 0.1693,
      "step": 12032
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.333018165195099,
      "learning_rate": 3.65578024267127e-08,
      "loss": 0.1629,
      "step": 12033
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3904831247379053,
      "learning_rate": 3.640159563756629e-08,
      "loss": 0.1666,
      "step": 12034
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4007722579059874,
      "learning_rate": 3.62457220698198e-08,
      "loss": 0.1784,
      "step": 12035
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.251428677603877,
      "learning_rate": 3.609018173393763e-08,
      "loss": 0.1436,
      "step": 12036
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4125431535547044,
      "learning_rate": 3.5934974640362e-08,
      "loss": 0.2091,
      "step": 12037
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.976626649929335,
      "learning_rate": 3.578010079951178e-08,
      "loss": 0.5252,
      "step": 12038
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2178594571534171,
      "learning_rate": 3.5625560221784205e-08,
      "loss": 0.1561,
      "step": 12039
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.322820552515946,
      "learning_rate": 3.547135291755488e-08,
      "loss": 0.1639,
      "step": 12040
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4706527220410934,
      "learning_rate": 3.531747889717496e-08,
      "loss": 0.1998,
      "step": 12041
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.396940556118862,
      "learning_rate": 3.516393817097508e-08,
      "loss": 0.1989,
      "step": 12042
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3234492517612735,
      "learning_rate": 3.5010730749263086e-08,
      "loss": 0.1279,
      "step": 12043
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3079839344993724,
      "learning_rate": 3.4857856642323554e-08,
      "loss": 0.1758,
      "step": 12044
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2881367357471751,
      "learning_rate": 3.470531586042047e-08,
      "loss": 0.167,
      "step": 12045
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.155113237486573,
      "learning_rate": 3.455310841379345e-08,
      "loss": 0.5205,
      "step": 12046
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2546464502133254,
      "learning_rate": 3.440123431266151e-08,
      "loss": 0.1473,
      "step": 12047
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.350399063246769,
      "learning_rate": 3.42496935672193e-08,
      "loss": 0.1657,
      "step": 12048
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2606406846876248,
      "learning_rate": 3.409848618764089e-08,
      "loss": 0.2075,
      "step": 12049
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1886334889528178,
      "learning_rate": 3.394761218407705e-08,
      "loss": 0.135,
      "step": 12050
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5817456614390903,
      "learning_rate": 3.379707156665746e-08,
      "loss": 0.232,
      "step": 12051
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4404575633818477,
      "learning_rate": 3.364686434548625e-08,
      "loss": 0.1554,
      "step": 12052
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3618906418744414,
      "learning_rate": 3.3496990530649256e-08,
      "loss": 0.1834,
      "step": 12053
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.194238485918517,
      "learning_rate": 3.334745013220675e-08,
      "loss": 0.1276,
      "step": 12054
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1872329717793155,
      "learning_rate": 3.3198243160198486e-08,
      "loss": 0.1554,
      "step": 12055
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3792562535418866,
      "learning_rate": 3.3049369624640896e-08,
      "loss": 0.1916,
      "step": 12056
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.251789081141179,
      "learning_rate": 3.290082953552876e-08,
      "loss": 0.1438,
      "step": 12057
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3313032966704588,
      "learning_rate": 3.2752622902832984e-08,
      "loss": 0.1535,
      "step": 12058
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.349054406544141,
      "learning_rate": 3.2604749736504514e-08,
      "loss": 0.2289,
      "step": 12059
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3629849066459452,
      "learning_rate": 3.245721004646929e-08,
      "loss": 0.1766,
      "step": 12060
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.444491168269748,
      "learning_rate": 3.231000384263272e-08,
      "loss": 0.1783,
      "step": 12061
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.361216997056909,
      "learning_rate": 3.21631311348769e-08,
      "loss": 0.1794,
      "step": 12062
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5301662417876827,
      "learning_rate": 3.201659193306228e-08,
      "loss": 0.2261,
      "step": 12063
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5299945825349242,
      "learning_rate": 3.1870386247025986e-08,
      "loss": 0.172,
      "step": 12064
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4737109521963345,
      "learning_rate": 3.172451408658406e-08,
      "loss": 0.1795,
      "step": 12065
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.195948475801584,
      "learning_rate": 3.1578975461528106e-08,
      "loss": 0.1508,
      "step": 12066
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6151968153174672,
      "learning_rate": 3.1433770381629756e-08,
      "loss": 0.1991,
      "step": 12067
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3002722602315484,
      "learning_rate": 3.128889885663622e-08,
      "loss": 0.1788,
      "step": 12068
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5033514642767625,
      "learning_rate": 3.114436089627415e-08,
      "loss": 0.1792,
      "step": 12069
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2977307387035604,
      "learning_rate": 3.100015651024524e-08,
      "loss": 0.1443,
      "step": 12070
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.201616295875197,
      "learning_rate": 3.085628570823174e-08,
      "loss": 0.3964,
      "step": 12071
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3622989725437762,
      "learning_rate": 3.071274849989147e-08,
      "loss": 0.2083,
      "step": 12072
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2723276387382116,
      "learning_rate": 3.0569544894861194e-08,
      "loss": 0.1731,
      "step": 12073
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.562506940744663,
      "learning_rate": 3.04266749027543e-08,
      "loss": 0.1778,
      "step": 12074
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3578661234455802,
      "learning_rate": 3.0284138533160924e-08,
      "loss": 0.1821,
      "step": 12075
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4358501300507704,
      "learning_rate": 3.0141935795651725e-08,
      "loss": 0.1604,
      "step": 12076
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.9823365425292865,
      "learning_rate": 3.00000666997724e-08,
      "loss": 0.568,
      "step": 12077
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4925221508350677,
      "learning_rate": 2.985853125504701e-08,
      "loss": 0.1543,
      "step": 12078
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2836759499464627,
      "learning_rate": 2.9717329470977384e-08,
      "loss": 0.1332,
      "step": 12079
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2673374611178578,
      "learning_rate": 2.957646135704262e-08,
      "loss": 0.1588,
      "step": 12080
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3686982005707646,
      "learning_rate": 2.9435926922699588e-08,
      "loss": 0.1921,
      "step": 12081
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.379521332314932,
      "learning_rate": 2.929572617738352e-08,
      "loss": 0.6916,
      "step": 12082
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3617758046017843,
      "learning_rate": 2.9155859130505782e-08,
      "loss": 0.1722,
      "step": 12083
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2711545968125395,
      "learning_rate": 2.901632579145608e-08,
      "loss": 0.1709,
      "step": 12084
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5732754287839423,
      "learning_rate": 2.8877126169602477e-08,
      "loss": 0.1702,
      "step": 12085
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.409748390686803,
      "learning_rate": 2.873826027428861e-08,
      "loss": 0.157,
      "step": 12086
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.464016065618886,
      "learning_rate": 2.8599728114838134e-08,
      "loss": 0.1565,
      "step": 12087
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4756818170544415,
      "learning_rate": 2.846152970055027e-08,
      "loss": 0.1858,
      "step": 12088
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5188673246293447,
      "learning_rate": 2.8323665040703717e-08,
      "loss": 0.1746,
      "step": 12089
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.361258431014825,
      "learning_rate": 2.818613414455218e-08,
      "loss": 0.1703,
      "step": 12090
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4520797451366454,
      "learning_rate": 2.804893702133049e-08,
      "loss": 0.1699,
      "step": 12091
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3891759310034593,
      "learning_rate": 2.7912073680247398e-08,
      "loss": 0.1726,
      "step": 12092
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.321580333725119,
      "learning_rate": 2.777554413049166e-08,
      "loss": 0.1567,
      "step": 12093
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1762527955721975,
      "learning_rate": 2.7639348381229282e-08,
      "loss": 0.1426,
      "step": 12094
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.364128231709815,
      "learning_rate": 2.7503486441602388e-08,
      "loss": 0.1836,
      "step": 12095
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4035172312033404,
      "learning_rate": 2.7367958320733135e-08,
      "loss": 0.1853,
      "step": 12096
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4078442344238051,
      "learning_rate": 2.7232764027718684e-08,
      "loss": 0.1976,
      "step": 12097
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2707720743362028,
      "learning_rate": 2.709790357163622e-08,
      "loss": 0.1304,
      "step": 12098
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4421887111905762,
      "learning_rate": 2.69633769615385e-08,
      "loss": 0.1684,
      "step": 12099
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.505569400520923,
      "learning_rate": 2.6829184206457194e-08,
      "loss": 0.2069,
      "step": 12100
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2181578783457396,
      "learning_rate": 2.6695325315401198e-08,
      "loss": 0.1575,
      "step": 12101
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2886042226557122,
      "learning_rate": 2.656180029735611e-08,
      "loss": 0.1699,
      "step": 12102
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3962941558613893,
      "learning_rate": 2.642860916128642e-08,
      "loss": 0.1814,
      "step": 12103
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.913118761696666,
      "learning_rate": 2.6295751916133315e-08,
      "loss": 0.6563,
      "step": 12104
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3406454155451237,
      "learning_rate": 2.6163228570816324e-08,
      "loss": 0.1729,
      "step": 12105
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4182273240135925,
      "learning_rate": 2.6031039134231663e-08,
      "loss": 0.182,
      "step": 12106
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4333338765856458,
      "learning_rate": 2.5899183615253897e-08,
      "loss": 0.1762,
      "step": 12107
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4923643455037134,
      "learning_rate": 2.5767662022735394e-08,
      "loss": 0.1924,
      "step": 12108
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3441307105179607,
      "learning_rate": 2.5636474365504095e-08,
      "loss": 0.1884,
      "step": 12109
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2881119185703722,
      "learning_rate": 2.5505620652369058e-08,
      "loss": 0.1564,
      "step": 12110
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3008360361380846,
      "learning_rate": 2.5375100892113258e-08,
      "loss": 0.1546,
      "step": 12111
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.532645808645719,
      "learning_rate": 2.5244915093499134e-08,
      "loss": 0.2131,
      "step": 12112
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3872226656204951,
      "learning_rate": 2.511506326526747e-08,
      "loss": 0.1474,
      "step": 12113
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2666883093797752,
      "learning_rate": 2.4985545416134628e-08,
      "loss": 0.1765,
      "step": 12114
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1640033121557665,
      "learning_rate": 2.4856361554795318e-08,
      "loss": 0.1244,
      "step": 12115
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3539787683073303,
      "learning_rate": 2.4727511689923156e-08,
      "loss": 0.1647,
      "step": 12116
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.259777909976236,
      "learning_rate": 2.459899583016734e-08,
      "loss": 0.1523,
      "step": 12117
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.526114999174613,
      "learning_rate": 2.4470813984155962e-08,
      "loss": 0.1798,
      "step": 12118
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5669402329431252,
      "learning_rate": 2.4342966160494364e-08,
      "loss": 0.1919,
      "step": 12119
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3069705197312311,
      "learning_rate": 2.421545236776457e-08,
      "loss": 0.1979,
      "step": 12120
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2768006460369934,
      "learning_rate": 2.408827261452751e-08,
      "loss": 0.1609,
      "step": 12121
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.374996924499818,
      "learning_rate": 2.396142690932135e-08,
      "loss": 0.177,
      "step": 12122
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3627517008156633,
      "learning_rate": 2.3834915260661506e-08,
      "loss": 0.1713,
      "step": 12123
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.967998693571285,
      "learning_rate": 2.3708737677040628e-08,
      "loss": 0.5587,
      "step": 12124
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2928547266145998,
      "learning_rate": 2.358289416693027e-08,
      "loss": 0.1567,
      "step": 12125
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3860491595305744,
      "learning_rate": 2.3457384738777567e-08,
      "loss": 0.1698,
      "step": 12126
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3301884738101102,
      "learning_rate": 2.3332209401009664e-08,
      "loss": 0.1462,
      "step": 12127
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3183929770279312,
      "learning_rate": 2.3207368162028733e-08,
      "loss": 0.1705,
      "step": 12128
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0769143902764093,
      "learning_rate": 2.3082861030215843e-08,
      "loss": 0.1296,
      "step": 12129
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2567878788932627,
      "learning_rate": 2.2958688013930973e-08,
      "loss": 0.1765,
      "step": 12130
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3403712707072726,
      "learning_rate": 2.2834849121508574e-08,
      "loss": 0.1806,
      "step": 12131
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3600016957594245,
      "learning_rate": 2.27113443612631e-08,
      "loss": 0.1315,
      "step": 12132
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4083011512643924,
      "learning_rate": 2.2588173741485144e-08,
      "loss": 0.1875,
      "step": 12133
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2210303716329927,
      "learning_rate": 2.2465337270444754e-08,
      "loss": 0.1711,
      "step": 12134
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4334874075537845,
      "learning_rate": 2.234283495638756e-08,
      "loss": 0.1854,
      "step": 12135
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.495359517216171,
      "learning_rate": 2.2220666807537538e-08,
      "loss": 0.2015,
      "step": 12136
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2088796558583776,
      "learning_rate": 2.2098832832095906e-08,
      "loss": 0.1612,
      "step": 12137
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.115386132069599,
      "learning_rate": 2.1977333038242233e-08,
      "loss": 0.1262,
      "step": 12138
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4154215861804995,
      "learning_rate": 2.185616743413277e-08,
      "loss": 0.1524,
      "step": 12139
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.392322841923491,
      "learning_rate": 2.173533602790212e-08,
      "loss": 0.1628,
      "step": 12140
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3750747118610367,
      "learning_rate": 2.161483882766213e-08,
      "loss": 0.2158,
      "step": 12141
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5932392113946434,
      "learning_rate": 2.149467584150189e-08,
      "loss": 0.19,
      "step": 12142
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2262813156384613,
      "learning_rate": 2.137484707748827e-08,
      "loss": 0.1377,
      "step": 12143
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5471823532578715,
      "learning_rate": 2.1255352543665954e-08,
      "loss": 0.2242,
      "step": 12144
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8684449764849067,
      "learning_rate": 2.1136192248056298e-08,
      "loss": 0.2044,
      "step": 12145
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3234845516522307,
      "learning_rate": 2.1017366198659573e-08,
      "loss": 0.1678,
      "step": 12146
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.373305621190505,
      "learning_rate": 2.0898874403453284e-08,
      "loss": 0.1716,
      "step": 12147
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.573837829976918,
      "learning_rate": 2.078071687039107e-08,
      "loss": 0.1762,
      "step": 12148
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6337536955590073,
      "learning_rate": 2.0662893607406033e-08,
      "loss": 0.199,
      "step": 12149
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5128868896716055,
      "learning_rate": 2.0545404622407396e-08,
      "loss": 0.2215,
      "step": 12150
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.553462953056607,
      "learning_rate": 2.0428249923283296e-08,
      "loss": 0.1718,
      "step": 12151
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.939391718539825,
      "learning_rate": 2.0311429517897997e-08,
      "loss": 0.4344,
      "step": 12152
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5829037585418475,
      "learning_rate": 2.019494341409467e-08,
      "loss": 0.1815,
      "step": 12153
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.848698248388663,
      "learning_rate": 2.0078791619692616e-08,
      "loss": 0.3825,
      "step": 12154
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.91867027264245,
      "learning_rate": 1.9962974142490043e-08,
      "loss": 0.4767,
      "step": 12155
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.461525713350055,
      "learning_rate": 1.984749099026184e-08,
      "loss": 0.1633,
      "step": 12156
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4647316525356164,
      "learning_rate": 1.9732342170760698e-08,
      "loss": 0.1687,
      "step": 12157
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3837270126836565,
      "learning_rate": 1.9617527691717653e-08,
      "loss": 0.1553,
      "step": 12158
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.270716247017522,
      "learning_rate": 1.9503047560839316e-08,
      "loss": 0.134,
      "step": 12159
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4217707303042713,
      "learning_rate": 1.9388901785811766e-08,
      "loss": 0.2105,
      "step": 12160
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3559998798507675,
      "learning_rate": 1.9275090374298312e-08,
      "loss": 0.1594,
      "step": 12161
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2515511588089945,
      "learning_rate": 1.91616133339384e-08,
      "loss": 0.1349,
      "step": 12162
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3755356718032212,
      "learning_rate": 1.9048470672350938e-08,
      "loss": 0.1773,
      "step": 12163
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3256021393822173,
      "learning_rate": 1.8935662397131516e-08,
      "loss": 0.1424,
      "step": 12164
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2528755099029445,
      "learning_rate": 1.8823188515852964e-08,
      "loss": 0.1486,
      "step": 12165
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5578253581860955,
      "learning_rate": 1.8711049036066465e-08,
      "loss": 0.2048,
      "step": 12166
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4950695414615793,
      "learning_rate": 1.8599243965299328e-08,
      "loss": 0.1894,
      "step": 12167
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2226207761711811,
      "learning_rate": 1.848777331105833e-08,
      "loss": 0.1266,
      "step": 12168
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.341830133092988,
      "learning_rate": 1.837663708082693e-08,
      "loss": 0.1707,
      "step": 12169
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2053602885680637,
      "learning_rate": 1.826583528206527e-08,
      "loss": 0.1355,
      "step": 12170
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2104290718913862,
      "learning_rate": 1.815536792221184e-08,
      "loss": 0.1289,
      "step": 12171
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.27395936487803,
      "learning_rate": 1.8045235008682937e-08,
      "loss": 0.6108,
      "step": 12172
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.299393859648016,
      "learning_rate": 1.7935436548872643e-08,
      "loss": 0.1776,
      "step": 12173
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.103057425911352,
      "learning_rate": 1.7825972550151173e-08,
      "loss": 0.6399,
      "step": 12174
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4472560694868717,
      "learning_rate": 1.7716843019867646e-08,
      "loss": 0.1523,
      "step": 12175
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2498497336712182,
      "learning_rate": 1.7608047965347876e-08,
      "loss": 0.1727,
      "step": 12176
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4063233194479239,
      "learning_rate": 1.749958739389601e-08,
      "loss": 0.1552,
      "step": 12177
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3362612585591054,
      "learning_rate": 1.7391461312794012e-08,
      "loss": 0.1534,
      "step": 12178
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.286995834326484,
      "learning_rate": 1.7283669729298847e-08,
      "loss": 0.1637,
      "step": 12179
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3039593072071103,
      "learning_rate": 1.7176212650648616e-08,
      "loss": 0.1426,
      "step": 12180
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3119264852066315,
      "learning_rate": 1.7069090084056437e-08,
      "loss": 0.1706,
      "step": 12181
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3018367517572076,
      "learning_rate": 1.6962302036713783e-08,
      "loss": 0.1521,
      "step": 12182
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3977147282360436,
      "learning_rate": 1.685584851578992e-08,
      "loss": 0.1855,
      "step": 12183
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.461932026973442,
      "learning_rate": 1.6749729528431345e-08,
      "loss": 0.1452,
      "step": 12184
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4717675995866952,
      "learning_rate": 1.6643945081761813e-08,
      "loss": 0.1601,
      "step": 12185
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5521248266603107,
      "learning_rate": 1.6538495182883418e-08,
      "loss": 0.193,
      "step": 12186
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.25264307954479,
      "learning_rate": 1.6433379838874942e-08,
      "loss": 0.1916,
      "step": 12187
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2547476828899171,
      "learning_rate": 1.6328599056792958e-08,
      "loss": 0.1354,
      "step": 12188
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4894967671136325,
      "learning_rate": 1.622415284367296e-08,
      "loss": 0.1788,
      "step": 12189
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.341557243562213,
      "learning_rate": 1.6120041206524885e-08,
      "loss": 0.1619,
      "step": 12190
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2447323989861585,
      "learning_rate": 1.6016264152339813e-08,
      "loss": 0.1386,
      "step": 12191
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3762815605662289,
      "learning_rate": 1.5912821688083278e-08,
      "loss": 0.1193,
      "step": 12192
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5732982880052122,
      "learning_rate": 1.580971382070029e-08,
      "loss": 0.2075,
      "step": 12193
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5262706883971546,
      "learning_rate": 1.570694055711308e-08,
      "loss": 0.2018,
      "step": 12194
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3828249323637616,
      "learning_rate": 1.560450190422058e-08,
      "loss": 0.1328,
      "step": 12195
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2874134876775765,
      "learning_rate": 1.550239786889951e-08,
      "loss": 0.1758,
      "step": 12196
    },
    {
      "epoch": 0.98,
      "grad_norm": 24.55686252076405,
      "learning_rate": 1.5400628458005495e-08,
      "loss": 0.4715,
      "step": 12197
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4217719934969204,
      "learning_rate": 1.529919367836974e-08,
      "loss": 0.178,
      "step": 12198
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2759049850375173,
      "learning_rate": 1.519809353680235e-08,
      "loss": 0.1199,
      "step": 12199
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4830562542830823,
      "learning_rate": 1.509732804009012e-08,
      "loss": 0.1692,
      "step": 12200
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2995127586027977,
      "learning_rate": 1.499689719499764e-08,
      "loss": 0.1531,
      "step": 12201
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5185508529231253,
      "learning_rate": 1.4896801008267848e-08,
      "loss": 0.1656,
      "step": 12202
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2641075517430078,
      "learning_rate": 1.4797039486619814e-08,
      "loss": 0.1849,
      "step": 12203
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7144137044918875,
      "learning_rate": 1.4697612636751513e-08,
      "loss": 0.1885,
      "step": 12204
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3426685854295706,
      "learning_rate": 1.4598520465337051e-08,
      "loss": 0.1967,
      "step": 12205
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4591341307756867,
      "learning_rate": 1.4499762979028887e-08,
      "loss": 0.1809,
      "step": 12206
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3242376509794311,
      "learning_rate": 1.4401340184457268e-08,
      "loss": 0.152,
      "step": 12207
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3009264294852863,
      "learning_rate": 1.4303252088229136e-08,
      "loss": 0.1386,
      "step": 12208
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2450471758564503,
      "learning_rate": 1.4205498696930332e-08,
      "loss": 0.2055,
      "step": 12209
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4451286899028526,
      "learning_rate": 1.4108080017122272e-08,
      "loss": 0.146,
      "step": 12210
    },
    {
      "epoch": 0.98,
      "grad_norm": 13.892704659211985,
      "learning_rate": 1.4010996055345838e-08,
      "loss": 0.5407,
      "step": 12211
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.352273748660829,
      "learning_rate": 1.3914246818118039e-08,
      "loss": 0.1449,
      "step": 12212
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.2156737052469255,
      "learning_rate": 1.3817832311934232e-08,
      "loss": 0.6321,
      "step": 12213
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3658827378126372,
      "learning_rate": 1.3721752543266464e-08,
      "loss": 0.1639,
      "step": 12214
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1383944717417929,
      "learning_rate": 1.3626007518565686e-08,
      "loss": 0.13,
      "step": 12215
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4276798186314943,
      "learning_rate": 1.353059724425898e-08,
      "loss": 0.1775,
      "step": 12216
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2914492159618929,
      "learning_rate": 1.3435521726751777e-08,
      "loss": 0.1388,
      "step": 12217
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.441087741275333,
      "learning_rate": 1.3340780972426192e-08,
      "loss": 0.2028,
      "step": 12218
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4539996835838316,
      "learning_rate": 1.324637498764325e-08,
      "loss": 0.1597,
      "step": 12219
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.129070178719021,
      "learning_rate": 1.3152303778740661e-08,
      "loss": 0.1221,
      "step": 12220
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4354945699889285,
      "learning_rate": 1.3058567352033369e-08,
      "loss": 0.1842,
      "step": 12221
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2541763511749886,
      "learning_rate": 1.296516571381412e-08,
      "loss": 0.1579,
      "step": 12222
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4075754669774874,
      "learning_rate": 1.287209887035401e-08,
      "loss": 0.1585,
      "step": 12223
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4577081755652435,
      "learning_rate": 1.2779366827899708e-08,
      "loss": 0.1905,
      "step": 12224
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2628100249784184,
      "learning_rate": 1.268696959267679e-08,
      "loss": 0.1453,
      "step": 12225
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.331820262349359,
      "learning_rate": 1.2594907170889181e-08,
      "loss": 0.1525,
      "step": 12226
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3350976274236408,
      "learning_rate": 1.2503179568716938e-08,
      "loss": 0.15,
      "step": 12227
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3573646794474215,
      "learning_rate": 1.241178679231736e-08,
      "loss": 0.1817,
      "step": 12228
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.274578300393972,
      "learning_rate": 1.232072884782609e-08,
      "loss": 0.1283,
      "step": 12229
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.316245290009325,
      "learning_rate": 1.2230005741356577e-08,
      "loss": 0.443,
      "step": 12230
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2483599025012198,
      "learning_rate": 1.21396174789995e-08,
      "loss": 0.1662,
      "step": 12231
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4160588071124862,
      "learning_rate": 1.2049564066822228e-08,
      "loss": 0.1945,
      "step": 12232
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5538771749297238,
      "learning_rate": 1.1959845510870483e-08,
      "loss": 0.185,
      "step": 12233
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5526792575936301,
      "learning_rate": 1.1870461817167222e-08,
      "loss": 0.1711,
      "step": 12234
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4968994634360262,
      "learning_rate": 1.1781412991713759e-08,
      "loss": 0.1873,
      "step": 12235
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.92480709905564,
      "learning_rate": 1.1692699040487532e-08,
      "loss": 0.5893,
      "step": 12236
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4863686775880538,
      "learning_rate": 1.1604319969444334e-08,
      "loss": 0.2052,
      "step": 12237
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.601728728951114,
      "learning_rate": 1.151627578451775e-08,
      "loss": 0.5932,
      "step": 12238
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3597971860915512,
      "learning_rate": 1.1428566491618053e-08,
      "loss": 0.2024,
      "step": 12239
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.301203309283687,
      "learning_rate": 1.1341192096633313e-08,
      "loss": 0.1897,
      "step": 12240
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.415794142201371,
      "learning_rate": 1.1254152605428836e-08,
      "loss": 0.1415,
      "step": 12241
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.4361603980693,
      "learning_rate": 1.1167448023848837e-08,
      "loss": 0.4815,
      "step": 12242
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.577964480547651,
      "learning_rate": 1.1081078357713659e-08,
      "loss": 0.1894,
      "step": 12243
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.372290978759702,
      "learning_rate": 1.0995043612821443e-08,
      "loss": 0.5653,
      "step": 12244
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.548587649850906,
      "learning_rate": 1.0909343794948124e-08,
      "loss": 0.1703,
      "step": 12245
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4707505482090233,
      "learning_rate": 1.0823978909846877e-08,
      "loss": 0.1678,
      "step": 12246
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4679607928467995,
      "learning_rate": 1.0738948963248674e-08,
      "loss": 0.2064,
      "step": 12247
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3945284839322962,
      "learning_rate": 1.0654253960861727e-08,
      "loss": 0.1581,
      "step": 12248
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3792259788850962,
      "learning_rate": 1.0569893908371487e-08,
      "loss": 0.1448,
      "step": 12249
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.995862125893316,
      "learning_rate": 1.0485868811441757e-08,
      "loss": 0.6902,
      "step": 12250
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3767203136396222,
      "learning_rate": 1.0402178675713026e-08,
      "loss": 0.2001,
      "step": 12251
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2402256006405865,
      "learning_rate": 1.0318823506804687e-08,
      "loss": 0.1483,
      "step": 12252
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2703544834799045,
      "learning_rate": 1.0235803310311154e-08,
      "loss": 0.1698,
      "step": 12253
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4279018193731392,
      "learning_rate": 1.0153118091806857e-08,
      "loss": 0.1754,
      "step": 12254
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1595784870017645,
      "learning_rate": 1.00707678568418e-08,
      "loss": 0.1433,
      "step": 12255
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2705992922837306,
      "learning_rate": 9.98875261094545e-09,
      "loss": 0.1492,
      "step": 12256
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4978084942537777,
      "learning_rate": 9.907072359623404e-09,
      "loss": 0.1583,
      "step": 12257
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.404079408824503,
      "learning_rate": 9.825727108359051e-09,
      "loss": 0.2114,
      "step": 12258
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3029560689472313,
      "learning_rate": 9.744716862613024e-09,
      "loss": 0.171,
      "step": 12259
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3693483338590915,
      "learning_rate": 9.66404162782375e-09,
      "loss": 0.1587,
      "step": 12260
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4032130563104992,
      "learning_rate": 9.583701409408009e-09,
      "loss": 0.1694,
      "step": 12261
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2531816839980177,
      "learning_rate": 9.503696212758706e-09,
      "loss": 0.1708,
      "step": 12262
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4984307093106302,
      "learning_rate": 9.424026043246548e-09,
      "loss": 0.1646,
      "step": 12263
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3965366120538856,
      "learning_rate": 9.344690906220588e-09,
      "loss": 0.1884,
      "step": 12264
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5029559073182108,
      "learning_rate": 9.265690807006566e-09,
      "loss": 0.1996,
      "step": 12265
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4743878636787504,
      "learning_rate": 9.187025750908018e-09,
      "loss": 0.188,
      "step": 12266
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.29589460222466,
      "learning_rate": 9.10869574320572e-09,
      "loss": 0.1434,
      "step": 12267
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5041501804371316,
      "learning_rate": 9.030700789158798e-09,
      "loss": 0.1839,
      "step": 12268
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3793725258579925,
      "learning_rate": 8.953040894003062e-09,
      "loss": 0.1754,
      "step": 12269
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2571181271275564,
      "learning_rate": 8.875716062951566e-09,
      "loss": 0.1268,
      "step": 12270
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.816754062772981,
      "learning_rate": 8.79872630119627e-09,
      "loss": 0.5238,
      "step": 12271
    },
    {
      "epoch": 0.98,
      "grad_norm": 17.58607576034883,
      "learning_rate": 8.722071613904704e-09,
      "loss": 0.6891,
      "step": 12272
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2452210552376264,
      "learning_rate": 8.645752006223861e-09,
      "loss": 0.1432,
      "step": 12273
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.345283979362492,
      "learning_rate": 8.56976748327687e-09,
      "loss": 0.182,
      "step": 12274
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4479199907936136,
      "learning_rate": 8.494118050164646e-09,
      "loss": 0.1912,
      "step": 12275
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2887791332407421,
      "learning_rate": 8.418803711965907e-09,
      "loss": 0.1418,
      "step": 12276
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2822601742930948,
      "learning_rate": 8.343824473736606e-09,
      "loss": 0.1732,
      "step": 12277
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2564170035108555,
      "learning_rate": 8.269180340510496e-09,
      "loss": 0.1869,
      "step": 12278
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.9862912713262535,
      "learning_rate": 8.194871317299125e-09,
      "loss": 0.8268,
      "step": 12279
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2768001077003917,
      "learning_rate": 8.120897409090166e-09,
      "loss": 0.1743,
      "step": 12280
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.810197117912155,
      "learning_rate": 8.047258620850206e-09,
      "loss": 0.5019,
      "step": 12281
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.990512480880503,
      "learning_rate": 7.973954957523068e-09,
      "loss": 0.5697,
      "step": 12282
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4695344571953228,
      "learning_rate": 7.900986424029256e-09,
      "loss": 0.1927,
      "step": 12283
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2664276407804347,
      "learning_rate": 7.828353025268188e-09,
      "loss": 0.1424,
      "step": 12284
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.331401252710619,
      "learning_rate": 7.756054766114852e-09,
      "loss": 0.5963,
      "step": 12285
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4112852802567288,
      "learning_rate": 7.684091651423697e-09,
      "loss": 0.165,
      "step": 12286
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3731390589457115,
      "learning_rate": 7.612463686025306e-09,
      "loss": 0.1452,
      "step": 12287
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1929103023388103,
      "learning_rate": 7.541170874729165e-09,
      "loss": 0.1343,
      "step": 12288
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4233021652881748,
      "learning_rate": 7.470213222320332e-09,
      "loss": 0.1595,
      "step": 12289
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.225446325597461,
      "learning_rate": 7.399590733562778e-09,
      "loss": 0.1362,
      "step": 12290
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2730192885612466,
      "learning_rate": 7.329303413198263e-09,
      "loss": 0.1465,
      "step": 12291
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.017879972028917,
      "learning_rate": 7.259351265944126e-09,
      "loss": 0.7116,
      "step": 12292
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2975253074124933,
      "learning_rate": 7.189734296497719e-09,
      "loss": 0.117,
      "step": 12293
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2638526734027022,
      "learning_rate": 7.120452509531417e-09,
      "loss": 0.1541,
      "step": 12294
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4948836129436398,
      "learning_rate": 7.051505909697609e-09,
      "loss": 0.212,
      "step": 12295
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1712536099176125,
      "learning_rate": 6.982894501623705e-09,
      "loss": 0.1612,
      "step": 12296
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3713071960391352,
      "learning_rate": 6.914618289916575e-09,
      "loss": 0.1712,
      "step": 12297
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.879105410249399,
      "learning_rate": 6.846677279159775e-09,
      "loss": 0.709,
      "step": 12298
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.429659359760385,
      "learning_rate": 6.7790714739141e-09,
      "loss": 0.1861,
      "step": 12299
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3749823825771499,
      "learning_rate": 6.711800878718144e-09,
      "loss": 0.196,
      "step": 12300
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.314832207426685,
      "learning_rate": 6.644865498087738e-09,
      "loss": 0.1619,
      "step": 12301
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5540704623641748,
      "learning_rate": 6.578265336517065e-09,
      "loss": 0.1755,
      "step": 12302
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5245910919725998,
      "learning_rate": 6.512000398476992e-09,
      "loss": 0.1824,
      "step": 12303
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.257500997300339,
      "learning_rate": 6.446070688415629e-09,
      "loss": 0.1668,
      "step": 12304
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3401438360576166,
      "learning_rate": 6.38047621075999e-09,
      "loss": 0.1645,
      "step": 12305
    },
    {
      "epoch": 0.98,
      "grad_norm": 21.119012663067124,
      "learning_rate": 6.315216969912663e-09,
      "loss": 0.5201,
      "step": 12306
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.448409313037296,
      "learning_rate": 6.250292970255145e-09,
      "loss": 0.1748,
      "step": 12307
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3553161486965755,
      "learning_rate": 6.185704216145616e-09,
      "loss": 0.1431,
      "step": 12308
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.678348894200219,
      "learning_rate": 6.121450711921162e-09,
      "loss": 0.4574,
      "step": 12309
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4669329433026022,
      "learning_rate": 6.057532461893889e-09,
      "loss": 0.191,
      "step": 12310
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.393941975253871,
      "learning_rate": 5.99394947035592e-09,
      "loss": 0.188,
      "step": 12311
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5439729910392685,
      "learning_rate": 5.9307017415749516e-09,
      "loss": 0.1863,
      "step": 12312
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.50180898577787,
      "learning_rate": 5.8677892797975866e-09,
      "loss": 0.1875,
      "step": 12313
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5607833765045886,
      "learning_rate": 5.805212089247114e-09,
      "loss": 0.1987,
      "step": 12314
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.352077229514529,
      "learning_rate": 5.742970174124618e-09,
      "loss": 0.1663,
      "step": 12315
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.199383473404824,
      "learning_rate": 5.681063538608422e-09,
      "loss": 0.1519,
      "step": 12316
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2819460329864152,
      "learning_rate": 5.619492186855202e-09,
      "loss": 0.161,
      "step": 12317
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.264466058052309,
      "learning_rate": 5.558256122997207e-09,
      "loss": 0.1576,
      "step": 12318
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4920802902798858,
      "learning_rate": 5.49735535114615e-09,
      "loss": 0.2044,
      "step": 12319
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3614897920733158,
      "learning_rate": 5.436789875390425e-09,
      "loss": 0.1687,
      "step": 12320
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.296904684012285,
      "learning_rate": 5.376559699796224e-09,
      "loss": 0.1722,
      "step": 12321
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3481339893970439,
      "learning_rate": 5.316664828406426e-09,
      "loss": 0.1473,
      "step": 12322
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.576515547067488,
      "learning_rate": 5.257105265241702e-09,
      "loss": 0.1966,
      "step": 12323
    },
    {
      "epoch": 0.99,
      "grad_norm": 19.66069683916925,
      "learning_rate": 5.197881014301631e-09,
      "loss": 0.4991,
      "step": 12324
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4723995186236816,
      "learning_rate": 5.138992079561367e-09,
      "loss": 0.2246,
      "step": 12325
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5279680531601956,
      "learning_rate": 5.080438464974413e-09,
      "loss": 0.2005,
      "step": 12326
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3006601194454075,
      "learning_rate": 5.022220174471515e-09,
      "loss": 0.1537,
      "step": 12327
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.618676123596008,
      "learning_rate": 4.964337211960657e-09,
      "loss": 0.1419,
      "step": 12328
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.427657177786239,
      "learning_rate": 4.90678958132873e-09,
      "loss": 0.1752,
      "step": 12329
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.391970293093847,
      "learning_rate": 4.849577286438756e-09,
      "loss": 0.1562,
      "step": 12330
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.726878325539933,
      "learning_rate": 4.792700331130995e-09,
      "loss": 0.1889,
      "step": 12331
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3897014568421922,
      "learning_rate": 4.73615871922406e-09,
      "loss": 0.1501,
      "step": 12332
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4757281658750148,
      "learning_rate": 4.679952454514358e-09,
      "loss": 0.187,
      "step": 12333
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5806145387314678,
      "learning_rate": 4.6240815407744274e-09,
      "loss": 0.185,
      "step": 12334
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3732692878523967,
      "learning_rate": 4.568545981755157e-09,
      "loss": 0.1801,
      "step": 12335
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4647804333632826,
      "learning_rate": 4.513345781184675e-09,
      "loss": 0.1715,
      "step": 12336
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3065198763656092,
      "learning_rate": 4.458480942769461e-09,
      "loss": 0.1267,
      "step": 12337
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.498248120338687,
      "learning_rate": 4.403951470192125e-09,
      "loss": 0.1445,
      "step": 12338
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2871319985551446,
      "learning_rate": 4.349757367113627e-09,
      "loss": 0.1441,
      "step": 12339
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4926997197484004,
      "learning_rate": 4.295898637172169e-09,
      "loss": 0.2172,
      "step": 12340
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.54686938084507,
      "learning_rate": 4.242375283983191e-09,
      "loss": 0.5971,
      "step": 12341
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.560666477147416,
      "learning_rate": 4.1891873111404855e-09,
      "loss": 0.2095,
      "step": 12342
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.467135754940844,
      "learning_rate": 4.136334722213975e-09,
      "loss": 0.1857,
      "step": 12343
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4942887830872245,
      "learning_rate": 4.083817520752487e-09,
      "loss": 0.1969,
      "step": 12344
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2894693195372344,
      "learning_rate": 4.031635710281534e-09,
      "loss": 0.1976,
      "step": 12345
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4526431439360832,
      "learning_rate": 3.979789294303315e-09,
      "loss": 0.1514,
      "step": 12346
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3223293873940583,
      "learning_rate": 3.928278276300046e-09,
      "loss": 0.1666,
      "step": 12347
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2099540498933572,
      "learning_rate": 3.877102659728404e-09,
      "loss": 0.1364,
      "step": 12348
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3267124091172107,
      "learning_rate": 3.826262448024531e-09,
      "loss": 0.13,
      "step": 12349
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4496510481714342,
      "learning_rate": 3.775757644601808e-09,
      "loss": 0.173,
      "step": 12350
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4600101639645677,
      "learning_rate": 3.725588252849743e-09,
      "loss": 0.1782,
      "step": 12351
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5136698008893839,
      "learning_rate": 3.6757542761378663e-09,
      "loss": 0.2242,
      "step": 12352
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.346901645513817,
      "learning_rate": 3.6262557178101678e-09,
      "loss": 0.1741,
      "step": 12353
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2687115238432485,
      "learning_rate": 3.5770925811906554e-09,
      "loss": 0.1489,
      "step": 12354
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4988878498216378,
      "learning_rate": 3.5282648695794675e-09,
      "loss": 0.1726,
      "step": 12355
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4155570823141062,
      "learning_rate": 3.4797725862545374e-09,
      "loss": 0.1684,
      "step": 12356
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3345094877212889,
      "learning_rate": 3.431615734471594e-09,
      "loss": 0.1596,
      "step": 12357
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5817264727196219,
      "learning_rate": 3.3837943174630517e-09,
      "loss": 0.1787,
      "step": 12358
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4127084729906447,
      "learning_rate": 3.3363083384396756e-09,
      "loss": 0.1847,
      "step": 12359
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1734884922316862,
      "learning_rate": 3.2891578005889158e-09,
      "loss": 0.15,
      "step": 12360
    },
    {
      "epoch": 0.99,
      "grad_norm": 11.560635991850749,
      "learning_rate": 3.2423427070765734e-09,
      "loss": 0.553,
      "step": 12361
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3816335815303793,
      "learning_rate": 3.1958630610451346e-09,
      "loss": 0.1853,
      "step": 12362
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6024541699705677,
      "learning_rate": 3.1497188656154364e-09,
      "loss": 0.2055,
      "step": 12363
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.502855844873582,
      "learning_rate": 3.103910123885001e-09,
      "loss": 0.4776,
      "step": 12364
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.140915778075582,
      "learning_rate": 3.0584368389291465e-09,
      "loss": 0.1278,
      "step": 12365
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3163209347215794,
      "learning_rate": 3.0132990138004303e-09,
      "loss": 0.1317,
      "step": 12366
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.153940413798579,
      "learning_rate": 2.9684966515286517e-09,
      "loss": 0.5226,
      "step": 12367
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3417864872004317,
      "learning_rate": 2.924029755122515e-09,
      "loss": 0.1588,
      "step": 12368
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3600769353574524,
      "learning_rate": 2.8798983275674096e-09,
      "loss": 0.1533,
      "step": 12369
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3940917293233759,
      "learning_rate": 2.8361023718248557e-09,
      "loss": 0.1603,
      "step": 12370
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4202603775114153,
      "learning_rate": 2.7926418908352795e-09,
      "loss": 0.196,
      "step": 12371
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.462933731978273,
      "learning_rate": 2.749516887516901e-09,
      "loss": 0.587,
      "step": 12372
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2236738063619397,
      "learning_rate": 2.7067273647646274e-09,
      "loss": 0.1773,
      "step": 12373
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4349459706290002,
      "learning_rate": 2.664273325450606e-09,
      "loss": 0.1823,
      "step": 12374
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3487796835289365,
      "learning_rate": 2.6221547724253337e-09,
      "loss": 0.2131,
      "step": 12375
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1979761253788213,
      "learning_rate": 2.580371708516549e-09,
      "loss": 0.1128,
      "step": 12376
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5371049958341534,
      "learning_rate": 2.5389241365286753e-09,
      "loss": 0.1851,
      "step": 12377
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4489277692090052,
      "learning_rate": 2.4978120592450415e-09,
      "loss": 0.1966,
      "step": 12378
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.68366510113351,
      "learning_rate": 2.457035479424552e-09,
      "loss": 0.6126,
      "step": 12379
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3466698401367339,
      "learning_rate": 2.4165943998050166e-09,
      "loss": 0.1715,
      "step": 12380
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.610756926846695,
      "learning_rate": 2.376488823102041e-09,
      "loss": 0.2028,
      "step": 12381
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4663812471311686,
      "learning_rate": 2.3367187520079158e-09,
      "loss": 0.1654,
      "step": 12382
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6868530767921344,
      "learning_rate": 2.2972841891916175e-09,
      "loss": 0.193,
      "step": 12383
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4160130622822396,
      "learning_rate": 2.258185137301028e-09,
      "loss": 0.1856,
      "step": 12384
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2848517940312105,
      "learning_rate": 2.219421598961269e-09,
      "loss": 0.1735,
      "step": 12385
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5012106866844217,
      "learning_rate": 2.1809935767741484e-09,
      "loss": 0.2074,
      "step": 12386
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.589497274494362,
      "learning_rate": 2.1429010733203804e-09,
      "loss": 0.1916,
      "step": 12387
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1476062738983304,
      "learning_rate": 2.1051440911556973e-09,
      "loss": 0.1436,
      "step": 12388
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3746764929364983,
      "learning_rate": 2.0677226328164047e-09,
      "loss": 0.1958,
      "step": 12389
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2920370035860693,
      "learning_rate": 2.0306367008138263e-09,
      "loss": 0.1615,
      "step": 12390
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.757896195405598,
      "learning_rate": 1.993886297637637e-09,
      "loss": 0.5433,
      "step": 12391
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4190559361498127,
      "learning_rate": 1.9574714257553084e-09,
      "loss": 0.185,
      "step": 12392
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3618281376289567,
      "learning_rate": 1.921392087611551e-09,
      "loss": 0.1626,
      "step": 12393
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.459251675745753,
      "learning_rate": 1.8856482856277615e-09,
      "loss": 0.164,
      "step": 12394
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2817898197403625,
      "learning_rate": 1.8502400222047967e-09,
      "loss": 0.1691,
      "step": 12395
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8164649396142765,
      "learning_rate": 1.8151672997185342e-09,
      "loss": 0.1875,
      "step": 12396
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6218945840895487,
      "learning_rate": 1.7804301205243125e-09,
      "loss": 0.2111,
      "step": 12397
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3480529726296482,
      "learning_rate": 1.7460284869535992e-09,
      "loss": 0.1438,
      "step": 12398
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.368658752033586,
      "learning_rate": 1.711962401316214e-09,
      "loss": 0.2167,
      "step": 12399
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.348689541208689,
      "learning_rate": 1.6782318658992159e-09,
      "loss": 0.1454,
      "step": 12400
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3600608221739365,
      "learning_rate": 1.6448368829663497e-09,
      "loss": 0.1789,
      "step": 12401
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3864266789405137,
      "learning_rate": 1.611777454760266e-09,
      "loss": 0.1777,
      "step": 12402
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4377286808726921,
      "learning_rate": 1.5790535835003006e-09,
      "loss": 0.1721,
      "step": 12403
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4011933271877726,
      "learning_rate": 1.5466652713824748e-09,
      "loss": 0.1601,
      "step": 12404
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4537048304126532,
      "learning_rate": 1.5146125205822703e-09,
      "loss": 0.1769,
      "step": 12405
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2751143686417106,
      "learning_rate": 1.4828953332507445e-09,
      "loss": 0.118,
      "step": 12406
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.360199711863573,
      "learning_rate": 1.4515137115173051e-09,
      "loss": 0.5382,
      "step": 12407
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4751042143119324,
      "learning_rate": 1.4204676574886001e-09,
      "loss": 0.231,
      "step": 12408
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4660952568821204,
      "learning_rate": 1.3897571732496284e-09,
      "loss": 0.165,
      "step": 12409
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5423802821682449,
      "learning_rate": 1.3593822608609642e-09,
      "loss": 0.2109,
      "step": 12410
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3128584140898922,
      "learning_rate": 1.3293429223620869e-09,
      "loss": 0.1765,
      "step": 12411
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.11882325642994,
      "learning_rate": 1.2996391597697167e-09,
      "loss": 0.5815,
      "step": 12412
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3228958382020615,
      "learning_rate": 1.2702709750783693e-09,
      "loss": 0.1612,
      "step": 12413
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3176762011998675,
      "learning_rate": 1.2412383702592456e-09,
      "loss": 0.1538,
      "step": 12414
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.473617421344286,
      "learning_rate": 1.2125413472613424e-09,
      "loss": 0.1866,
      "step": 12415
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2728618120494672,
      "learning_rate": 1.1841799080114513e-09,
      "loss": 0.2002,
      "step": 12416
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5430915116808106,
      "learning_rate": 1.1561540544130502e-09,
      "loss": 0.1467,
      "step": 12417
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.450949560863337,
      "learning_rate": 1.1284637883485217e-09,
      "loss": 0.1547,
      "step": 12418
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5430801091108468,
      "learning_rate": 1.1011091116758244e-09,
      "loss": 0.1603,
      "step": 12419
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6751472628821946,
      "learning_rate": 1.074090026231267e-09,
      "loss": 0.1708,
      "step": 12420
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.303974764425101,
      "learning_rate": 1.0474065338300643e-09,
      "loss": 0.1727,
      "step": 12421
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.278845192915408,
      "learning_rate": 1.0210586362618957e-09,
      "loss": 0.1472,
      "step": 12422
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3259345134072038,
      "learning_rate": 9.95046335297012e-10,
      "loss": 0.1558,
      "step": 12423
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3629431828669294,
      "learning_rate": 9.693696326806834e-10,
      "loss": 0.1625,
      "step": 12424
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.448764721383848,
      "learning_rate": 9.440285301370865e-10,
      "loss": 0.1952,
      "step": 12425
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2563578990533695,
      "learning_rate": 9.190230293670832e-10,
      "loss": 0.1517,
      "step": 12426
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3348301019367887,
      "learning_rate": 8.943531320504406e-10,
      "loss": 0.1427,
      "step": 12427
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.221874588182704,
      "learning_rate": 8.700188398419462e-10,
      "loss": 0.1711,
      "step": 12428
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2782950854156685,
      "learning_rate": 8.460201543758484e-10,
      "loss": 0.1559,
      "step": 12429
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4578008185900555,
      "learning_rate": 8.223570772636358e-10,
      "loss": 0.1712,
      "step": 12430
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.54978810808045,
      "learning_rate": 7.990296100929273e-10,
      "loss": 0.2191,
      "step": 12431
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4407032747913722,
      "learning_rate": 7.760377544308028e-10,
      "loss": 0.1854,
      "step": 12432
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.39391860974014,
      "learning_rate": 7.533815118204724e-10,
      "loss": 0.1841,
      "step": 12433
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4293782580132575,
      "learning_rate": 7.310608837823863e-10,
      "loss": 0.164,
      "step": 12434
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4080986413261523,
      "learning_rate": 7.090758718153457e-10,
      "loss": 0.2149,
      "step": 12435
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4417525270832874,
      "learning_rate": 6.87426477395392e-10,
      "loss": 0.186,
      "step": 12436
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3546528250782155,
      "learning_rate": 6.661127019758074e-10,
      "loss": 0.1715,
      "step": 12437
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3576999772131655,
      "learning_rate": 6.451345469876691e-10,
      "loss": 0.1437,
      "step": 12438
    },
    {
      "epoch": 1.0,
      "grad_norm": 10.346874579005064,
      "learning_rate": 6.2449201383874e-10,
      "loss": 0.5354,
      "step": 12439
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.844415334165052,
      "learning_rate": 6.041851039151337e-10,
      "loss": 0.5341,
      "step": 12440
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.340737348657837,
      "learning_rate": 5.842138185807589e-10,
      "loss": 0.185,
      "step": 12441
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5325089484802543,
      "learning_rate": 5.645781591750998e-10,
      "loss": 0.1831,
      "step": 12442
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4041455854715992,
      "learning_rate": 5.452781270171015e-10,
      "loss": 0.1762,
      "step": 12443
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.278495733628024,
      "learning_rate": 5.263137234023941e-10,
      "loss": 0.1619,
      "step": 12444
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5047500870787491,
      "learning_rate": 5.076849496044034e-10,
      "loss": 0.175,
      "step": 12445
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.757268320782321,
      "learning_rate": 4.893918068726855e-10,
      "loss": 0.5709,
      "step": 12446
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2482668570882471,
      "learning_rate": 4.714342964368124e-10,
      "loss": 0.1614,
      "step": 12447
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2113480552390334,
      "learning_rate": 4.538124195008209e-10,
      "loss": 0.1384,
      "step": 12448
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3194012389859553,
      "learning_rate": 4.3652617724876387e-10,
      "loss": 0.1433,
      "step": 12449
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2482354084598875,
      "learning_rate": 4.1957557084082447e-10,
      "loss": 0.153,
      "step": 12450
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.500151594637948,
      "learning_rate": 4.0296060141498116e-10,
      "loss": 0.208,
      "step": 12451
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.904553600045467,
      "learning_rate": 3.86681270085898e-10,
      "loss": 0.5328,
      "step": 12452
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4399258407950049,
      "learning_rate": 3.707375779476996e-10,
      "loss": 0.1371,
      "step": 12453
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2919240224978263,
      "learning_rate": 3.5512952607008597e-10,
      "loss": 0.1548,
      "step": 12454
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3644810223640418,
      "learning_rate": 3.398571155011077e-10,
      "loss": 0.1568,
      "step": 12455
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5294373605825142,
      "learning_rate": 3.249203472655005e-10,
      "loss": 0.1793,
      "step": 12456
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4704870900241869,
      "learning_rate": 3.1031922236690604e-10,
      "loss": 0.1279,
      "step": 12457
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2845473309482875,
      "learning_rate": 2.9605374178454104e-10,
      "loss": 0.1521,
      "step": 12458
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3298594368596983,
      "learning_rate": 2.821239064770831e-10,
      "loss": 0.2038,
      "step": 12459
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.502680772724333,
      "learning_rate": 2.6852971737878483e-10,
      "loss": 0.1841,
      "step": 12460
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3207531587764756,
      "learning_rate": 2.5527117540280475e-10,
      "loss": 0.1786,
      "step": 12461
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3703484223685662,
      "learning_rate": 2.423482814389866e-10,
      "loss": 0.1898,
      "step": 12462
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4995343979937201,
      "learning_rate": 2.297610363549696e-10,
      "loss": 0.1911,
      "step": 12463
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.5257805567300124,
      "learning_rate": 2.1750944099618865e-10,
      "loss": 0.5361,
      "step": 12464
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4570660963391613,
      "learning_rate": 2.0559349618420877e-10,
      "loss": 0.1944,
      "step": 12465
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.038698655981214,
      "learning_rate": 1.9401320272005585e-10,
      "loss": 0.5719,
      "step": 12466
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3652835620030952,
      "learning_rate": 1.8276856138033095e-10,
      "loss": 0.2062,
      "step": 12467
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.298275905013123,
      "learning_rate": 1.7185957291998568e-10,
      "loss": 0.1709,
      "step": 12468
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.676511801464607,
      "learning_rate": 1.6128623807176726e-10,
      "loss": 0.5762,
      "step": 12469
    },
    {
      "epoch": 1.0,
      "grad_norm": 12.137816656550177,
      "learning_rate": 1.5104855754566329e-10,
      "loss": 0.5806,
      "step": 12470
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4239731564966485,
      "learning_rate": 1.4114653202834672e-10,
      "loss": 0.1686,
      "step": 12471
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3775099348967232,
      "learning_rate": 1.3158016218428605e-10,
      "loss": 0.176,
      "step": 12472
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.388389116201817,
      "learning_rate": 1.223494486568555e-10,
      "loss": 0.186,
      "step": 12473
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.277211543108001,
      "learning_rate": 1.1345439206500441e-10,
      "loss": 0.1286,
      "step": 12474
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2543232351115798,
      "learning_rate": 1.0489499300603279e-10,
      "loss": 0.1465,
      "step": 12475
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6695659682758446,
      "learning_rate": 9.667125205448102e-11,
      "loss": 0.1687,
      "step": 12476
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4460695622520532,
      "learning_rate": 8.878316976268508e-11,
      "loss": 0.1777,
      "step": 12477
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4673696087842043,
      "learning_rate": 8.123074665966624e-11,
      "loss": 0.1799,
      "step": 12478
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3636871146700074,
      "learning_rate": 7.401398325335152e-11,
      "loss": 0.1799,
      "step": 12479
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.297010273055021,
      "learning_rate": 6.713288002724305e-11,
      "loss": 0.1489,
      "step": 12480
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3352641986509013,
      "learning_rate": 6.058743744374874e-11,
      "loss": 0.1751,
      "step": 12481
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.034262881864984,
      "learning_rate": 5.437765594196176e-11,
      "loss": 0.3937,
      "step": 12482
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4119534379433494,
      "learning_rate": 4.8503535939326e-11,
      "loss": 0.1569,
      "step": 12483
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6655655272604932,
      "learning_rate": 4.296507782997061e-11,
      "loss": 0.1851,
      "step": 12484
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4135804673982388,
      "learning_rate": 3.776228198526521e-11,
      "loss": 0.1764,
      "step": 12485
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.326214058789395,
      "learning_rate": 3.289514875548516e-11,
      "loss": 0.5402,
      "step": 12486
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3324158441519371,
      "learning_rate": 2.8363678465925804e-11,
      "loss": 0.1696,
      "step": 12487
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3566381054246832,
      "learning_rate": 2.416787142245358e-11,
      "loss": 0.182,
      "step": 12488
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4884231772623906,
      "learning_rate": 2.030772790539981e-11,
      "loss": 0.1793,
      "step": 12489
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.277548275918522,
      "learning_rate": 1.6783248174556677e-11,
      "loss": 0.1348,
      "step": 12490
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.310317184661321,
      "learning_rate": 1.3594432466401686e-11,
      "loss": 0.1589,
      "step": 12491
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4874888057435114,
      "learning_rate": 1.074128099520788e-11,
      "loss": 0.1793,
      "step": 12492
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3331354664989092,
      "learning_rate": 8.223793951933623e-12,
      "loss": 0.1697,
      "step": 12493
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1352728993213603,
      "learning_rate": 6.0419715058879225e-12,
      "loss": 0.1307,
      "step": 12494
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3584496311398142,
      "learning_rate": 4.19581380417533e-12,
      "loss": 0.1526,
      "step": 12495
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5578867726168715,
      "learning_rate": 2.6853209700306025e-12,
      "loss": 0.1933,
      "step": 12496
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2706422910499398,
      "learning_rate": 1.5104931044840344e-12,
      "loss": 0.1463,
      "step": 12497
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1398165796291209,
      "learning_rate": 6.713302874716832e-13,
      "loss": 0.1258,
      "step": 12498
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3057438696186767,
      "learning_rate": 1.6783257450470047e-13,
      "loss": 0.163,
      "step": 12499
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.928702036545608,
      "learning_rate": 0.0,
      "loss": 0.36,
      "step": 12500
    },
    {
      "epoch": 1.0,
      "step": 12500,
      "total_flos": 5919025980409856.0,
      "train_loss": 0.246136455129385,
      "train_runtime": 40622.9472,
      "train_samples_per_second": 4.923,
      "train_steps_per_second": 0.308
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 12500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 5919025980409856.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}