{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 710,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 2.3505215219626858e-06,
      "loss": 1.5844,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7010430439253716e-06,
      "loss": 1.1717,
      "step": 2
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.051564565888058e-06,
      "loss": 1.1708,
      "step": 3
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.402086087850743e-06,
      "loss": 1.5208,
      "step": 4
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.175260760981343e-05,
      "loss": 1.2781,
      "step": 5
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.4103129131776116e-05,
      "loss": 1.3885,
      "step": 6
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.64536506537388e-05,
      "loss": 1.4596,
      "step": 7
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.8804172175701486e-05,
      "loss": 1.2032,
      "step": 8
    },
    {
      "epoch": 0.06,
      "learning_rate": 2.1154693697664172e-05,
      "loss": 1.3789,
      "step": 9
    },
    {
      "epoch": 0.07,
      "learning_rate": 2.350521521962686e-05,
      "loss": 1.1589,
      "step": 10
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.585573674158954e-05,
      "loss": 1.3547,
      "step": 11
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.820625826355223e-05,
      "loss": 1.6605,
      "step": 12
    },
    {
      "epoch": 0.09,
      "learning_rate": 3.055677978551492e-05,
      "loss": 1.2712,
      "step": 13
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.29073013074776e-05,
      "loss": 1.2314,
      "step": 14
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.525782282944028e-05,
      "loss": 1.3365,
      "step": 15
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.760834435140297e-05,
      "loss": 1.3434,
      "step": 16
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.9958865873365655e-05,
      "loss": 1.4921,
      "step": 17
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.2309387395328345e-05,
      "loss": 1.3429,
      "step": 18
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.4659908917291035e-05,
      "loss": 1.3414,
      "step": 19
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.701043043925372e-05,
      "loss": 1.3829,
      "step": 20
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.93609519612164e-05,
      "loss": 1.0409,
      "step": 21
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.171147348317908e-05,
      "loss": 1.4251,
      "step": 22
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.406199500514177e-05,
      "loss": 1.3312,
      "step": 23
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.641251652710446e-05,
      "loss": 1.5272,
      "step": 24
    },
    {
      "epoch": 0.18,
      "learning_rate": 5.8763038049067145e-05,
      "loss": 1.3403,
      "step": 25
    },
    {
      "epoch": 0.18,
      "learning_rate": 6.111355957102983e-05,
      "loss": 1.2074,
      "step": 26
    },
    {
      "epoch": 0.19,
      "learning_rate": 6.346408109299252e-05,
      "loss": 1.2367,
      "step": 27
    },
    {
      "epoch": 0.2,
      "learning_rate": 6.58146026149552e-05,
      "loss": 1.0457,
      "step": 28
    },
    {
      "epoch": 0.2,
      "learning_rate": 6.816512413691789e-05,
      "loss": 1.2032,
      "step": 29
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.051564565888057e-05,
      "loss": 1.2936,
      "step": 30
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.286616718084326e-05,
      "loss": 1.4325,
      "step": 31
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.521668870280595e-05,
      "loss": 1.1163,
      "step": 32
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.756721022476862e-05,
      "loss": 1.2707,
      "step": 33
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.991773174673131e-05,
      "loss": 1.2504,
      "step": 34
    },
    {
      "epoch": 0.25,
      "learning_rate": 8.2268253268694e-05,
      "loss": 1.3381,
      "step": 35
    },
    {
      "epoch": 0.25,
      "learning_rate": 8.461877479065669e-05,
      "loss": 1.4586,
      "step": 36
    },
    {
      "epoch": 0.26,
      "learning_rate": 8.696929631261938e-05,
      "loss": 1.3332,
      "step": 37
    },
    {
      "epoch": 0.27,
      "learning_rate": 8.931981783458207e-05,
      "loss": 1.2071,
      "step": 38
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.167033935654475e-05,
      "loss": 1.2606,
      "step": 39
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.402086087850743e-05,
      "loss": 1.2386,
      "step": 40
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.637138240047012e-05,
      "loss": 1.163,
      "step": 41
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.87219039224328e-05,
      "loss": 1.2538,
      "step": 42
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.999988438329202e-05,
      "loss": 1.1698,
      "step": 43
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.999882216302589e-05,
      "loss": 1.3658,
      "step": 44
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.999664914761532e-05,
      "loss": 1.0131,
      "step": 45
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.999336538533694e-05,
      "loss": 1.4838,
      "step": 46
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.998897094914424e-05,
      "loss": 1.1761,
      "step": 47
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.998346593666587e-05,
      "loss": 1.4351,
      "step": 48
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.997685047020353e-05,
      "loss": 1.3387,
      "step": 49
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.996912469672925e-05,
      "loss": 1.2005,
      "step": 50
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.99602887878821e-05,
      "loss": 1.3915,
      "step": 51
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.995034293996442e-05,
      "loss": 1.2307,
      "step": 52
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.993928737393738e-05,
      "loss": 1.1234,
      "step": 53
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.992712233541616e-05,
      "loss": 1.2151,
      "step": 54
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.991384809466445e-05,
      "loss": 1.0929,
      "step": 55
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.989946494658839e-05,
      "loss": 1.4025,
      "step": 56
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.98839732107302e-05,
      "loss": 1.1941,
      "step": 57
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.986737323126084e-05,
      "loss": 1.2325,
      "step": 58
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.984966537697254e-05,
      "loss": 1.2219,
      "step": 59
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.983085004127055e-05,
      "loss": 1.256,
      "step": 60
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.981092764216437e-05,
      "loss": 1.3723,
      "step": 61
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.978989862225853e-05,
      "loss": 1.1166,
      "step": 62
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.976776344874268e-05,
      "loss": 1.2826,
      "step": 63
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.974452261338128e-05,
      "loss": 1.1347,
      "step": 64
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.972017663250264e-05,
      "loss": 1.1066,
      "step": 65
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.969472604698742e-05,
      "loss": 1.3199,
      "step": 66
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.96681714222567e-05,
      "loss": 1.3186,
      "step": 67
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.964051334825937e-05,
      "loss": 1.3577,
      "step": 68
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.961175243945893e-05,
      "loss": 1.1913,
      "step": 69
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.958188933482003e-05,
      "loss": 1.3416,
      "step": 70
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.95509246977941e-05,
      "loss": 1.4361,
      "step": 71
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.951885921630475e-05,
      "loss": 1.2034,
      "step": 72
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.948569360273236e-05,
      "loss": 1.147,
      "step": 73
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.945142859389835e-05,
      "loss": 1.5043,
      "step": 74
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.941606495104877e-05,
      "loss": 1.2877,
      "step": 75
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.937960345983736e-05,
      "loss": 1.2024,
      "step": 76
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.934204493030816e-05,
      "loss": 1.2718,
      "step": 77
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.930339019687746e-05,
      "loss": 1.1279,
      "step": 78
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.92636401183153e-05,
      "loss": 1.1138,
      "step": 79
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.922279557772639e-05,
      "loss": 1.2727,
      "step": 80
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.918085748253044e-05,
      "loss": 1.1073,
      "step": 81
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.9137826764442e-05,
      "loss": 1.1801,
      "step": 82
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.909370437944993e-05,
      "loss": 1.4271,
      "step": 83
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.904849130779588e-05,
      "loss": 1.45,
      "step": 84
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.900218855395276e-05,
      "loss": 1.3001,
      "step": 85
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.895479714660226e-05,
      "loss": 1.3452,
      "step": 86
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.890631813861207e-05,
      "loss": 1.2141,
      "step": 87
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.885675260701254e-05,
      "loss": 1.213,
      "step": 88
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.880610165297262e-05,
      "loss": 1.3351,
      "step": 89
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.87543664017755e-05,
      "loss": 1.063,
      "step": 90
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.870154800279354e-05,
      "loss": 1.0482,
      "step": 91
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.864764762946285e-05,
      "loss": 1.3549,
      "step": 92
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.859266647925709e-05,
      "loss": 1.2104,
      "step": 93
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.853660577366092e-05,
      "loss": 1.3416,
      "step": 94
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.847946675814286e-05,
      "loss": 1.0623,
      "step": 95
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.842125070212765e-05,
      "loss": 1.0585,
      "step": 96
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.8361958898968e-05,
      "loss": 1.123,
      "step": 97
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.830159266591589e-05,
      "loss": 1.2343,
      "step": 98
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.824015334409325e-05,
      "loss": 1.2386,
      "step": 99
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.817764229846228e-05,
      "loss": 0.9396,
      "step": 100
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.811406091779501e-05,
      "loss": 1.0644,
      "step": 101
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.804941061464246e-05,
      "loss": 1.2164,
      "step": 102
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.798369282530334e-05,
      "loss": 1.2597,
      "step": 103
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.79169090097921e-05,
      "loss": 1.0657,
      "step": 104
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.784906065180639e-05,
      "loss": 1.3115,
      "step": 105
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.77801492586943e-05,
      "loss": 1.1609,
      "step": 106
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.771017636142069e-05,
      "loss": 1.2352,
      "step": 107
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.763914351453328e-05,
      "loss": 1.2551,
      "step": 108
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.756705229612806e-05,
      "loss": 1.1841,
      "step": 109
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.749390430781428e-05,
      "loss": 1.4215,
      "step": 110
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.741970117467885e-05,
      "loss": 1.3392,
      "step": 111
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.734444454525015e-05,
      "loss": 1.2668,
      "step": 112
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.726813609146156e-05,
      "loss": 1.2107,
      "step": 113
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.719077750861422e-05,
      "loss": 1.0586,
      "step": 114
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.711237051533939e-05,
      "loss": 1.2463,
      "step": 115
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.703291685356022e-05,
      "loss": 1.0728,
      "step": 116
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.695241828845314e-05,
      "loss": 1.0965,
      "step": 117
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.68708766084086e-05,
      "loss": 1.4591,
      "step": 118
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.678829362499131e-05,
      "loss": 1.0267,
      "step": 119
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.670467117290003e-05,
      "loss": 1.2758,
      "step": 120
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.662001110992679e-05,
      "loss": 1.2481,
      "step": 121
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.653431531691564e-05,
      "loss": 1.1977,
      "step": 122
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.644758569772084e-05,
      "loss": 1.4683,
      "step": 123
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.635982417916459e-05,
      "loss": 1.2774,
      "step": 124
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.627103271099418e-05,
      "loss": 1.1035,
      "step": 125
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.618121326583874e-05,
      "loss": 1.1808,
      "step": 126
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.609036783916529e-05,
      "loss": 1.1988,
      "step": 127
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.59984984492346e-05,
      "loss": 1.3405,
      "step": 128
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.590560713705619e-05,
      "loss": 1.31,
      "step": 129
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.581169596634302e-05,
      "loss": 1.2651,
      "step": 130
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.571676702346571e-05,
      "loss": 1.2317,
      "step": 131
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.56208224174061e-05,
      "loss": 1.2728,
      "step": 132
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.552386427971049e-05,
      "loss": 1.2959,
      "step": 133
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.542589476444215e-05,
      "loss": 1.2706,
      "step": 134
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.53269160481336e-05,
      "loss": 1.1602,
      "step": 135
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.52269303297382e-05,
      "loss": 1.2293,
      "step": 136
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.512593983058129e-05,
      "loss": 1.2822,
      "step": 137
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.502394679431084e-05,
      "loss": 1.0499,
      "step": 138
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.49209534868476e-05,
      "loss": 1.3138,
      "step": 139
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.481696219633481e-05,
      "loss": 1.4097,
      "step": 140
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.47119752330873e-05,
      "loss": 1.1332,
      "step": 141
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.460599492954018e-05,
      "loss": 1.1116,
      "step": 142
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.449902364019708e-05,
      "loss": 1.3985,
      "step": 143
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.439106374157773e-05,
      "loss": 1.0012,
      "step": 144
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.428211763216526e-05,
      "loss": 0.9964,
      "step": 145
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.417218773235286e-05,
      "loss": 1.339,
      "step": 146
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.406127648439008e-05,
      "loss": 1.0616,
      "step": 147
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.394938635232844e-05,
      "loss": 1.2071,
      "step": 148
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.383651982196682e-05,
      "loss": 1.289,
      "step": 149
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.372267940079617e-05,
      "loss": 1.0109,
      "step": 150
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.36078676179438e-05,
      "loss": 1.1774,
      "step": 151
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.349208702411722e-05,
      "loss": 0.9881,
      "step": 152
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.337534019154745e-05,
      "loss": 1.1558,
      "step": 153
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.32576297139319e-05,
      "loss": 1.468,
      "step": 154
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.313895820637669e-05,
      "loss": 1.1095,
      "step": 155
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.301932830533862e-05,
      "loss": 1.0915,
      "step": 156
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.289874266856655e-05,
      "loss": 1.1427,
      "step": 157
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.277720397504243e-05,
      "loss": 1.2019,
      "step": 158
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.265471492492163e-05,
      "loss": 1.3101,
      "step": 159
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.253127823947311e-05,
      "loss": 1.1997,
      "step": 160
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.240689666101892e-05,
      "loss": 1.1938,
      "step": 161
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.228157295287317e-05,
      "loss": 1.2199,
      "step": 162
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.215530989928081e-05,
      "loss": 0.9386,
      "step": 163
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.202811030535565e-05,
      "loss": 1.2752,
      "step": 164
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.189997699701807e-05,
      "loss": 1.2073,
      "step": 165
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.177091282093231e-05,
      "loss": 1.3951,
      "step": 166
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.164092064444306e-05,
      "loss": 1.1749,
      "step": 167
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.151000335551194e-05,
      "loss": 1.0863,
      "step": 168
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.137816386265327e-05,
      "loss": 1.1223,
      "step": 169
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.124540509486939e-05,
      "loss": 0.948,
      "step": 170
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.111173000158566e-05,
      "loss": 1.1039,
      "step": 171
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.097714155258497e-05,
      "loss": 1.1733,
      "step": 172
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.084164273794163e-05,
      "loss": 1.3096,
      "step": 173
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.070523656795508e-05,
      "loss": 0.995,
      "step": 174
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.056792607308293e-05,
      "loss": 1.1716,
      "step": 175
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.042971430387366e-05,
      "loss": 1.1152,
      "step": 176
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.029060433089885e-05,
      "loss": 1.2403,
      "step": 177
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.015059924468494e-05,
      "loss": 1.3455,
      "step": 178
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.000970215564462e-05,
      "loss": 1.1923,
      "step": 179
    },
    {
      "epoch": 1.27,
      "learning_rate": 8.986791619400768e-05,
      "loss": 1.1189,
      "step": 180
    },
    {
      "epoch": 1.27,
      "learning_rate": 8.972524450975143e-05,
      "loss": 1.166,
      "step": 181
    },
    {
      "epoch": 1.28,
      "learning_rate": 8.958169027253087e-05,
      "loss": 1.1154,
      "step": 182
    },
    {
      "epoch": 1.29,
      "learning_rate": 8.94372566716081e-05,
      "loss": 1.0369,
      "step": 183
    },
    {
      "epoch": 1.3,
      "learning_rate": 8.929194691578158e-05,
      "loss": 1.1153,
      "step": 184
    },
    {
      "epoch": 1.3,
      "learning_rate": 8.914576423331475e-05,
      "loss": 1.0748,
      "step": 185
    },
    {
      "epoch": 1.31,
      "learning_rate": 8.89987118718644e-05,
      "loss": 1.2589,
      "step": 186
    },
    {
      "epoch": 1.32,
      "learning_rate": 8.885079309840848e-05,
      "loss": 0.9202,
      "step": 187
    },
    {
      "epoch": 1.32,
      "learning_rate": 8.870201119917354e-05,
      "loss": 1.3762,
      "step": 188
    },
    {
      "epoch": 1.33,
      "learning_rate": 8.855236947956161e-05,
      "loss": 1.0431,
      "step": 189
    },
    {
      "epoch": 1.34,
      "learning_rate": 8.840187126407695e-05,
      "loss": 1.2946,
      "step": 190
    },
    {
      "epoch": 1.35,
      "learning_rate": 8.825051989625206e-05,
      "loss": 1.1731,
      "step": 191
    },
    {
      "epoch": 1.35,
      "learning_rate": 8.809831873857344e-05,
      "loss": 1.0521,
      "step": 192
    },
    {
      "epoch": 1.36,
      "learning_rate": 8.794527117240688e-05,
      "loss": 1.2539,
      "step": 193
    },
    {
      "epoch": 1.37,
      "learning_rate": 8.779138059792232e-05,
      "loss": 1.1325,
      "step": 194
    },
    {
      "epoch": 1.37,
      "learning_rate": 8.76366504340184e-05,
      "loss": 0.9917,
      "step": 195
    },
    {
      "epoch": 1.38,
      "learning_rate": 8.748108411824636e-05,
      "loss": 1.0688,
      "step": 196
    },
    {
      "epoch": 1.39,
      "learning_rate": 8.732468510673379e-05,
      "loss": 0.882,
      "step": 197
    },
    {
      "epoch": 1.39,
      "learning_rate": 8.716745687410779e-05,
      "loss": 1.2479,
      "step": 198
    },
    {
      "epoch": 1.4,
      "learning_rate": 8.700940291341782e-05,
      "loss": 1.0746,
      "step": 199
    },
    {
      "epoch": 1.41,
      "learning_rate": 8.685052673605798e-05,
      "loss": 1.0906,
      "step": 200
    },
    {
      "epoch": 1.42,
      "learning_rate": 8.66908318716892e-05,
      "loss": 1.0399,
      "step": 201
    },
    {
      "epoch": 1.42,
      "learning_rate": 8.653032186816062e-05,
      "loss": 1.1061,
      "step": 202
    },
    {
      "epoch": 1.43,
      "learning_rate": 8.636900029143091e-05,
      "loss": 1.2413,
      "step": 203
    },
    {
      "epoch": 1.44,
      "learning_rate": 8.620687072548899e-05,
      "loss": 0.9968,
      "step": 204
    },
    {
      "epoch": 1.44,
      "learning_rate": 8.604393677227436e-05,
      "loss": 1.0685,
      "step": 205
    },
    {
      "epoch": 1.45,
      "learning_rate": 8.588020205159724e-05,
      "loss": 1.0171,
      "step": 206
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.57156702010579e-05,
      "loss": 0.9696,
      "step": 207
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.555034487596612e-05,
      "loss": 1.1729,
      "step": 208
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.538422974925975e-05,
      "loss": 1.1588,
      "step": 209
    },
    {
      "epoch": 1.48,
      "learning_rate": 8.521732851142327e-05,
      "loss": 1.2008,
      "step": 210
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.50496448704057e-05,
      "loss": 1.0552,
      "step": 211
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.488118255153831e-05,
      "loss": 1.1742,
      "step": 212
    },
    {
      "epoch": 1.5,
      "learning_rate": 8.471194529745176e-05,
      "loss": 1.2478,
      "step": 213
    },
    {
      "epoch": 1.51,
      "learning_rate": 8.454193686799306e-05,
      "loss": 1.0792,
      "step": 214
    },
    {
      "epoch": 1.51,
      "learning_rate": 8.437116104014187e-05,
      "loss": 0.9578,
      "step": 215
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.419962160792683e-05,
      "loss": 1.3535,
      "step": 216
    },
    {
      "epoch": 1.53,
      "learning_rate": 8.402732238234105e-05,
      "loss": 1.0785,
      "step": 217
    },
    {
      "epoch": 1.54,
      "learning_rate": 8.385426719125759e-05,
      "loss": 1.0145,
      "step": 218
    },
    {
      "epoch": 1.54,
      "learning_rate": 8.368045987934432e-05,
      "loss": 1.1375,
      "step": 219
    },
    {
      "epoch": 1.55,
      "learning_rate": 8.350590430797858e-05,
      "loss": 0.976,
      "step": 220
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.333060435516134e-05,
      "loss": 0.9385,
      "step": 221
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.315456391543106e-05,
      "loss": 1.1096,
      "step": 222
    },
    {
      "epoch": 1.57,
      "learning_rate": 8.297778689977719e-05,
      "loss": 0.9635,
      "step": 223
    },
    {
      "epoch": 1.58,
      "learning_rate": 8.280027723555327e-05,
      "loss": 1.0106,
      "step": 224
    },
    {
      "epoch": 1.58,
      "learning_rate": 8.262203886638964e-05,
      "loss": 1.2063,
      "step": 225
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.24430757521059e-05,
      "loss": 1.2897,
      "step": 226
    },
    {
      "epoch": 1.6,
      "learning_rate": 8.226339186862286e-05,
      "loss": 1.1298,
      "step": 227
    },
    {
      "epoch": 1.61,
      "learning_rate": 8.208299120787429e-05,
      "loss": 1.1559,
      "step": 228
    },
    {
      "epoch": 1.61,
      "learning_rate": 8.190187777771812e-05,
      "loss": 1.0587,
      "step": 229
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.172005560184755e-05,
      "loss": 1.0188,
      "step": 230
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.153752871970151e-05,
      "loss": 1.1215,
      "step": 231
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.135430118637504e-05,
      "loss": 0.9491,
      "step": 232
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.117037707252911e-05,
      "loss": 0.9292,
      "step": 233
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.098576046430023e-05,
      "loss": 1.1974,
      "step": 234
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.080045546320963e-05,
      "loss": 1.0007,
      "step": 235
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.061446618607222e-05,
      "loss": 1.1207,
      "step": 236
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.042779676490506e-05,
      "loss": 0.8835,
      "step": 237
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.024045134683559e-05,
      "loss": 0.8961,
      "step": 238
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.005243409400945e-05,
      "loss": 0.9905,
      "step": 239
    },
    {
      "epoch": 1.69,
      "learning_rate": 7.986374918349804e-05,
      "loss": 1.0613,
      "step": 240
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.967440080720581e-05,
      "loss": 1.0781,
      "step": 241
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.948439317177696e-05,
      "loss": 0.7686,
      "step": 242
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.929373049850214e-05,
      "loss": 0.941,
      "step": 243
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.910241702322452e-05,
      "loss": 1.0858,
      "step": 244
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.891045699624586e-05,
      "loss": 1.0956,
      "step": 245
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.87178546822319e-05,
      "loss": 0.9023,
      "step": 246
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.852461436011777e-05,
      "loss": 1.1251,
      "step": 247
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.833074032301279e-05,
      "loss": 1.0181,
      "step": 248
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.813623687810528e-05,
      "loss": 1.0848,
      "step": 249
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.79411083465666e-05,
      "loss": 1.0787,
      "step": 250
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.774535906345541e-05,
      "loss": 1.0356,
      "step": 251
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.754899337762122e-05,
      "loss": 1.1425,
      "step": 252
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.73520156516078e-05,
      "loss": 1.1107,
      "step": 253
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.715443026155628e-05,
      "loss": 1.1205,
      "step": 254
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.695624159710788e-05,
      "loss": 1.0453,
      "step": 255
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.675745406130647e-05,
      "loss": 0.8335,
      "step": 256
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.655807207050066e-05,
      "loss": 1.0367,
      "step": 257
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.635810005424575e-05,
      "loss": 0.9137,
      "step": 258
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.615754245520526e-05,
      "loss": 0.9815,
      "step": 259
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.595640372905235e-05,
      "loss": 1.1748,
      "step": 260
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.575468834437065e-05,
      "loss": 0.8648,
      "step": 261
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.555240078255512e-05,
      "loss": 1.0501,
      "step": 262
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.53495455377125e-05,
      "loss": 1.0069,
      "step": 263
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.514612711656132e-05,
      "loss": 1.0022,
      "step": 264
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.494215003833197e-05,
      "loss": 1.2586,
      "step": 265
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.473761883466614e-05,
      "loss": 1.1238,
      "step": 266
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.453253804951623e-05,
      "loss": 0.9192,
      "step": 267
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.432691223904439e-05,
      "loss": 1.0445,
      "step": 268
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.412074597152125e-05,
      "loss": 1.0032,
      "step": 269
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.39140438272245e-05,
      "loss": 1.147,
      "step": 270
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.370681039833706e-05,
      "loss": 1.1687,
      "step": 271
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.349905028884516e-05,
      "loss": 1.1281,
      "step": 272
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.329076811443589e-05,
      "loss": 1.0926,
      "step": 273
    },
    {
      "epoch": 1.93,
      "learning_rate": 7.308196850239487e-05,
      "loss": 1.0286,
      "step": 274
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.287265609150325e-05,
      "loss": 1.1251,
      "step": 275
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.266283553193475e-05,
      "loss": 1.0484,
      "step": 276
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.245251148515237e-05,
      "loss": 0.9548,
      "step": 277
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.224168862380482e-05,
      "loss": 1.0533,
      "step": 278
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.203037163162258e-05,
      "loss": 1.1724,
      "step": 279
    },
    {
      "epoch": 1.97,
      "learning_rate": 7.181856520331407e-05,
      "loss": 0.8868,
      "step": 280
    },
    {
      "epoch": 1.98,
      "learning_rate": 7.160627404446115e-05,
      "loss": 1.1498,
      "step": 281
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.13935028714147e-05,
      "loss": 1.2512,
      "step": 282
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.118025641118981e-05,
      "loss": 0.9648,
      "step": 283
    },
    {
      "epoch": 2.0,
      "learning_rate": 7.096653940136069e-05,
      "loss": 0.8932,
      "step": 284
    },
    {
      "epoch": 2.01,
      "learning_rate": 7.075235658995558e-05,
      "loss": 1.1335,
      "step": 285
    },
    {
      "epoch": 2.01,
      "learning_rate": 7.053771273535108e-05,
      "loss": 0.8604,
      "step": 286
    },
    {
      "epoch": 2.02,
      "learning_rate": 7.032261260616654e-05,
      "loss": 0.858,
      "step": 287
    },
    {
      "epoch": 2.03,
      "learning_rate": 7.010706098115812e-05,
      "loss": 1.152,
      "step": 288
    },
    {
      "epoch": 2.04,
      "learning_rate": 6.989106264911262e-05,
      "loss": 0.8461,
      "step": 289
    },
    {
      "epoch": 2.04,
      "learning_rate": 6.967462240874104e-05,
      "loss": 1.0286,
      "step": 290
    },
    {
      "epoch": 2.05,
      "learning_rate": 6.9457745068572e-05,
      "loss": 1.1287,
      "step": 291
    },
    {
      "epoch": 2.06,
      "learning_rate": 6.924043544684494e-05,
      "loss": 0.828,
      "step": 292
    },
    {
      "epoch": 2.06,
      "learning_rate": 6.902269837140304e-05,
      "loss": 0.9707,
      "step": 293
    },
    {
      "epoch": 2.07,
      "learning_rate": 6.880453867958599e-05,
      "loss": 0.8406,
      "step": 294
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.858596121812247e-05,
      "loss": 0.9251,
      "step": 295
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.836697084302253e-05,
      "loss": 1.1901,
      "step": 296
    },
    {
      "epoch": 2.09,
      "learning_rate": 6.814757241946969e-05,
      "loss": 0.9653,
      "step": 297
    },
    {
      "epoch": 2.1,
      "learning_rate": 6.792777082171286e-05,
      "loss": 0.9213,
      "step": 298
    },
    {
      "epoch": 2.11,
      "learning_rate": 6.770757093295798e-05,
      "loss": 0.8884,
      "step": 299
    },
    {
      "epoch": 2.11,
      "learning_rate": 6.74869776452597e-05,
      "loss": 1.0057,
      "step": 300
    },
    {
      "epoch": 2.12,
      "learning_rate": 6.726599585941247e-05,
      "loss": 1.0599,
      "step": 301
    },
    {
      "epoch": 2.13,
      "learning_rate": 6.704463048484184e-05,
      "loss": 1.0271,
      "step": 302
    },
    {
      "epoch": 2.13,
      "learning_rate": 6.682288643949534e-05,
      "loss": 0.9883,
      "step": 303
    },
    {
      "epoch": 2.14,
      "learning_rate": 6.66007686497332e-05,
      "loss": 1.0209,
      "step": 304
    },
    {
      "epoch": 2.15,
      "learning_rate": 6.637828205021891e-05,
      "loss": 0.8159,
      "step": 305
    },
    {
      "epoch": 2.15,
      "learning_rate": 6.615543158380956e-05,
      "loss": 1.0447,
      "step": 306
    },
    {
      "epoch": 2.16,
      "learning_rate": 6.593222220144617e-05,
      "loss": 1.0394,
      "step": 307
    },
    {
      "epoch": 2.17,
      "learning_rate": 6.570865886204347e-05,
      "loss": 1.146,
      "step": 308
    },
    {
      "epoch": 2.18,
      "learning_rate": 6.54847465323799e-05,
      "loss": 0.9289,
      "step": 309
    },
    {
      "epoch": 2.18,
      "learning_rate": 6.526049018698721e-05,
      "loss": 0.8888,
      "step": 310
    },
    {
      "epoch": 2.19,
      "learning_rate": 6.503589480803998e-05,
      "loss": 0.9346,
      "step": 311
    },
    {
      "epoch": 2.2,
      "learning_rate": 6.481096538524487e-05,
      "loss": 0.7988,
      "step": 312
    },
    {
      "epoch": 2.2,
      "learning_rate": 6.458570691572979e-05,
      "loss": 0.9219,
      "step": 313
    },
    {
      "epoch": 2.21,
      "learning_rate": 6.436012440393294e-05,
      "loss": 0.9391,
      "step": 314
    },
    {
      "epoch": 2.22,
      "learning_rate": 6.413422286149154e-05,
      "loss": 1.0681,
      "step": 315
    },
    {
      "epoch": 2.23,
      "learning_rate": 6.390800730713055e-05,
      "loss": 0.7624,
      "step": 316
    },
    {
      "epoch": 2.23,
      "learning_rate": 6.368148276655113e-05,
      "loss": 0.9216,
      "step": 317
    },
    {
      "epoch": 2.24,
      "learning_rate": 6.345465427231904e-05,
      "loss": 0.8041,
      "step": 318
    },
    {
      "epoch": 2.25,
      "learning_rate": 6.32275268637528e-05,
      "loss": 1.0059,
      "step": 319
    },
    {
      "epoch": 2.25,
      "learning_rate": 6.300010558681171e-05,
      "loss": 1.0607,
      "step": 320
    },
    {
      "epoch": 2.26,
      "learning_rate": 6.27723954939838e-05,
      "loss": 0.8981,
      "step": 321
    },
    {
      "epoch": 2.27,
      "learning_rate": 6.254440164417356e-05,
      "loss": 0.9132,
      "step": 322
    },
    {
      "epoch": 2.27,
      "learning_rate": 6.231612910258952e-05,
      "loss": 0.9141,
      "step": 323
    },
    {
      "epoch": 2.28,
      "learning_rate": 6.208758294063176e-05,
      "loss": 0.8164,
      "step": 324
    },
    {
      "epoch": 2.29,
      "learning_rate": 6.185876823577923e-05,
      "loss": 0.8132,
      "step": 325
    },
    {
      "epoch": 2.3,
      "learning_rate": 6.162969007147697e-05,
      "loss": 0.8809,
      "step": 326
    },
    {
      "epoch": 2.3,
      "learning_rate": 6.14003535370231e-05,
      "loss": 0.8769,
      "step": 327
    },
    {
      "epoch": 2.31,
      "learning_rate": 6.117076372745579e-05,
      "loss": 1.0372,
      "step": 328
    },
    {
      "epoch": 2.32,
      "learning_rate": 6.094092574344017e-05,
      "loss": 0.7325,
      "step": 329
    },
    {
      "epoch": 2.32,
      "learning_rate": 6.0710844691154824e-05,
      "loss": 1.1814,
      "step": 330
    },
    {
      "epoch": 2.33,
      "learning_rate": 6.0480525682178455e-05,
      "loss": 0.8201,
      "step": 331
    },
    {
      "epoch": 2.34,
      "learning_rate": 6.024997383337637e-05,
      "loss": 1.0517,
      "step": 332
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.0019194266786715e-05,
      "loss": 0.8776,
      "step": 333
    },
    {
      "epoch": 2.35,
      "learning_rate": 5.97881921095067e-05,
      "loss": 0.8195,
      "step": 334
    },
    {
      "epoch": 2.36,
      "learning_rate": 5.955697249357869e-05,
      "loss": 0.9821,
      "step": 335
    },
    {
      "epoch": 2.37,
      "learning_rate": 5.932554055587624e-05,
      "loss": 0.9625,
      "step": 336
    },
    {
      "epoch": 2.37,
      "learning_rate": 5.909390143798995e-05,
      "loss": 0.7567,
      "step": 337
    },
    {
      "epoch": 2.38,
      "learning_rate": 5.8862060286113165e-05,
      "loss": 0.8004,
      "step": 338
    },
    {
      "epoch": 2.39,
      "learning_rate": 5.8630022250927716e-05,
      "loss": 0.5626,
      "step": 339
    },
    {
      "epoch": 2.39,
      "learning_rate": 5.839779248748954e-05,
      "loss": 1.015,
      "step": 340
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.816537615511404e-05,
      "loss": 0.8443,
      "step": 341
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.79327784172615e-05,
      "loss": 0.8472,
      "step": 342
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.7700004441422395e-05,
      "loss": 0.7691,
      "step": 343
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.746705939900262e-05,
      "loss": 0.8565,
      "step": 344
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.723394846520852e-05,
      "loss": 1.0142,
      "step": 345
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.7000676818931944e-05,
      "loss": 0.7965,
      "step": 346
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.676724964263521e-05,
      "loss": 0.8013,
      "step": 347
    },
    {
      "epoch": 2.45,
      "learning_rate": 5.653367212223599e-05,
      "loss": 0.8015,
      "step": 348
    },
    {
      "epoch": 2.46,
      "learning_rate": 5.6299949446992026e-05,
      "loss": 0.742,
      "step": 349
    },
    {
      "epoch": 2.46,
      "learning_rate": 5.6066086809385873e-05,
      "loss": 0.9467,
      "step": 350
    },
    {
      "epoch": 2.47,
      "learning_rate": 5.583208940500959e-05,
      "loss": 0.8903,
      "step": 351
    },
    {
      "epoch": 2.48,
      "learning_rate": 5.559796243244926e-05,
      "loss": 0.9451,
      "step": 352
    },
    {
      "epoch": 2.49,
      "learning_rate": 5.536371109316947e-05,
      "loss": 0.8239,
      "step": 353
    },
    {
      "epoch": 2.49,
      "learning_rate": 5.5129340591397804e-05,
      "loss": 0.9108,
      "step": 354
    },
    {
      "epoch": 2.5,
      "learning_rate": 5.489485613400929e-05,
      "loss": 0.9606,
      "step": 355
    },
    {
      "epoch": 2.51,
      "learning_rate": 5.466026293041053e-05,
      "loss": 0.8744,
      "step": 356
    },
    {
      "epoch": 2.51,
      "learning_rate": 5.442556619242414e-05,
      "loss": 0.7085,
      "step": 357
    },
    {
      "epoch": 2.52,
      "learning_rate": 5.419077113417288e-05,
      "loss": 1.0677,
      "step": 358
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.3955882971963856e-05,
      "loss": 0.8227,
      "step": 359
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.3720906924172596e-05,
      "loss": 0.7129,
      "step": 360
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.348584821112711e-05,
      "loss": 0.8936,
      "step": 361
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.325071205499198e-05,
      "loss": 0.7626,
      "step": 362
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.301550367965228e-05,
      "loss": 0.7006,
      "step": 363
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.278022831059751e-05,
      "loss": 0.8868,
      "step": 364
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.254489117480556e-05,
      "loss": 0.7274,
      "step": 365
    },
    {
      "epoch": 2.58,
      "learning_rate": 5.230949750062655e-05,
      "loss": 0.7568,
      "step": 366
    },
    {
      "epoch": 2.58,
      "learning_rate": 5.2074052517666685e-05,
      "loss": 0.8734,
      "step": 367
    },
    {
      "epoch": 2.59,
      "learning_rate": 5.183856145667204e-05,
      "loss": 1.0261,
      "step": 368
    },
    {
      "epoch": 2.6,
      "learning_rate": 5.160302954941241e-05,
      "loss": 0.9876,
      "step": 369
    },
    {
      "epoch": 2.61,
      "learning_rate": 5.136746202856506e-05,
      "loss": 0.9113,
      "step": 370
    },
    {
      "epoch": 2.61,
      "learning_rate": 5.113186412759842e-05,
      "loss": 0.8247,
      "step": 371
    },
    {
      "epoch": 2.62,
      "learning_rate": 5.089624108065587e-05,
      "loss": 0.7352,
      "step": 372
    },
    {
      "epoch": 2.63,
      "learning_rate": 5.066059812243946e-05,
      "loss": 0.8319,
      "step": 373
    },
    {
      "epoch": 2.63,
      "learning_rate": 5.042494048809361e-05,
      "loss": 0.7573,
      "step": 374
    },
    {
      "epoch": 2.64,
      "learning_rate": 5.018927341308874e-05,
      "loss": 0.729,
      "step": 375
    },
    {
      "epoch": 2.65,
      "learning_rate": 4.9953602133105056e-05,
      "loss": 0.9483,
      "step": 376
    },
    {
      "epoch": 2.65,
      "learning_rate": 4.971793188391615e-05,
      "loss": 0.6948,
      "step": 377
    },
    {
      "epoch": 2.66,
      "learning_rate": 4.948226790127273e-05,
      "loss": 0.7836,
      "step": 378
    },
    {
      "epoch": 2.67,
      "learning_rate": 4.9246615420786315e-05,
      "loss": 0.647,
      "step": 379
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.901097967781281e-05,
      "loss": 0.6558,
      "step": 380
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.8775365907336345e-05,
      "loss": 0.7926,
      "step": 381
    },
    {
      "epoch": 2.69,
      "learning_rate": 4.8539779343852884e-05,
      "loss": 0.7926,
      "step": 382
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.8304225221253895e-05,
      "loss": 0.818,
      "step": 383
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.80687087727102e-05,
      "loss": 0.5522,
      "step": 384
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.783323523055559e-05,
      "loss": 0.7327,
      "step": 385
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.7597809826170646e-05,
      "loss": 0.8745,
      "step": 386
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.736243778986651e-05,
      "loss": 0.8599,
      "step": 387
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.7127124350768646e-05,
      "loss": 0.6869,
      "step": 388
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.689187473670073e-05,
      "loss": 0.8579,
      "step": 389
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.665669417406847e-05,
      "loss": 0.7942,
      "step": 390
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.642158788774345e-05,
      "loss": 0.8683,
      "step": 391
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.618656110094717e-05,
      "loss": 0.8248,
      "step": 392
    },
    {
      "epoch": 2.77,
      "learning_rate": 4.5951619035134914e-05,
      "loss": 0.8052,
      "step": 393
    },
    {
      "epoch": 2.77,
      "learning_rate": 4.571676690987972e-05,
      "loss": 0.7724,
      "step": 394
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.548200994275652e-05,
      "loss": 0.8052,
      "step": 395
    },
    {
      "epoch": 2.79,
      "learning_rate": 4.524735334922615e-05,
      "loss": 0.8825,
      "step": 396
    },
    {
      "epoch": 2.8,
      "learning_rate": 4.501280234251952e-05,
      "loss": 0.8237,
      "step": 397
    },
    {
      "epoch": 2.8,
      "learning_rate": 4.477836213352178e-05,
      "loss": 0.5574,
      "step": 398
    },
    {
      "epoch": 2.81,
      "learning_rate": 4.4544037930656526e-05,
      "loss": 0.7633,
      "step": 399
    },
    {
      "epoch": 2.82,
      "learning_rate": 4.430983493977012e-05,
      "loss": 0.6846,
      "step": 400
    },
    {
      "epoch": 2.82,
      "learning_rate": 4.407575836401605e-05,
      "loss": 0.7732,
      "step": 401
    },
    {
      "epoch": 2.83,
      "learning_rate": 4.384181340373927e-05,
      "loss": 0.8842,
      "step": 402
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.360800525636076e-05,
      "loss": 0.618,
      "step": 403
    },
    {
      "epoch": 2.85,
      "learning_rate": 4.337433911626198e-05,
      "loss": 0.7508,
      "step": 404
    },
    {
      "epoch": 2.85,
      "learning_rate": 4.314082017466947e-05,
      "loss": 0.6947,
      "step": 405
    },
    {
      "epoch": 2.86,
      "learning_rate": 4.2907453619539596e-05,
      "loss": 0.7075,
      "step": 406
    },
    {
      "epoch": 2.87,
      "learning_rate": 4.267424463544318e-05,
      "loss": 0.9874,
      "step": 407
    },
    {
      "epoch": 2.87,
      "learning_rate": 4.244119840345045e-05,
      "loss": 0.8875,
      "step": 408
    },
    {
      "epoch": 2.88,
      "learning_rate": 4.2208320101015816e-05,
      "loss": 0.6921,
      "step": 409
    },
    {
      "epoch": 2.89,
      "learning_rate": 4.197561490186291e-05,
      "loss": 0.8342,
      "step": 410
    },
    {
      "epoch": 2.89,
      "learning_rate": 4.174308797586961e-05,
      "loss": 0.7022,
      "step": 411
    },
    {
      "epoch": 2.9,
      "learning_rate": 4.151074448895324e-05,
      "loss": 0.8656,
      "step": 412
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.1278589602955706e-05,
      "loss": 0.9449,
      "step": 413
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.1046628475528936e-05,
      "loss": 0.9049,
      "step": 414
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.081486626002021e-05,
      "loss": 0.8814,
      "step": 415
    },
    {
      "epoch": 2.93,
      "learning_rate": 4.058330810535766e-05,
      "loss": 0.7045,
      "step": 416
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.035195915593598e-05,
      "loss": 0.8558,
      "step": 417
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.012082455150201e-05,
      "loss": 0.772,
      "step": 418
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.988990942704066e-05,
      "loss": 0.7281,
      "step": 419
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.9659218912660766e-05,
      "loss": 0.7919,
      "step": 420
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.9428758133481106e-05,
      "loss": 0.9592,
      "step": 421
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.919853220951658e-05,
      "loss": 0.6567,
      "step": 422
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.896854625556448e-05,
      "loss": 0.9045,
      "step": 423
    },
    {
      "epoch": 2.99,
      "learning_rate": 3.873880538109073e-05,
      "loss": 0.9526,
      "step": 424
    },
    {
      "epoch": 2.99,
      "learning_rate": 3.850931469011656e-05,
      "loss": 0.6942,
      "step": 425
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.8280079281104996e-05,
      "loss": 0.604,
      "step": 426
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.805110424684754e-05,
      "loss": 0.7876,
      "step": 427
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.782239467435119e-05,
      "loss": 0.6762,
      "step": 428
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.7593955644725263e-05,
      "loss": 0.6553,
      "step": 429
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.736579223306863e-05,
      "loss": 0.9052,
      "step": 430
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.713790950835691e-05,
      "loss": 0.6197,
      "step": 431
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.691031253332981e-05,
      "loss": 0.8027,
      "step": 432
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.6683006364378745e-05,
      "loss": 0.8929,
      "step": 433
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.64559960514345e-05,
      "loss": 0.5936,
      "step": 434
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.6229286637854865e-05,
      "loss": 0.6819,
      "step": 435
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.6002883160312865e-05,
      "loss": 0.6546,
      "step": 436
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.577679064868466e-05,
      "loss": 0.6387,
      "step": 437
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.555101412593781e-05,
      "loss": 0.8402,
      "step": 438
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.532555860801982e-05,
      "loss": 0.7592,
      "step": 439
    },
    {
      "epoch": 3.1,
      "learning_rate": 3.5100429103746547e-05,
      "loss": 0.7132,
      "step": 440
    },
    {
      "epoch": 3.11,
      "learning_rate": 3.487563061469102e-05,
      "loss": 0.6006,
      "step": 441
    },
    {
      "epoch": 3.11,
      "learning_rate": 3.465116813507232e-05,
      "loss": 0.758,
      "step": 442
    },
    {
      "epoch": 3.12,
      "learning_rate": 3.4427046651644523e-05,
      "loss": 0.7745,
      "step": 443
    },
    {
      "epoch": 3.13,
      "learning_rate": 3.420327114358606e-05,
      "loss": 0.8207,
      "step": 444
    },
    {
      "epoch": 3.13,
      "learning_rate": 3.397984658238902e-05,
      "loss": 0.7432,
      "step": 445
    },
    {
      "epoch": 3.14,
      "learning_rate": 3.375677793174862e-05,
      "loss": 0.7818,
      "step": 446
    },
    {
      "epoch": 3.15,
      "learning_rate": 3.3534070147453135e-05,
      "loss": 0.6546,
      "step": 447
    },
    {
      "epoch": 3.15,
      "learning_rate": 3.331172817727363e-05,
      "loss": 0.8025,
      "step": 448
    },
    {
      "epoch": 3.16,
      "learning_rate": 3.308975696085405e-05,
      "loss": 0.854,
      "step": 449
    },
    {
      "epoch": 3.17,
      "learning_rate": 3.2868161429601584e-05,
      "loss": 0.8708,
      "step": 450
    },
    {
      "epoch": 3.18,
      "learning_rate": 3.264694650657698e-05,
      "loss": 0.6904,
      "step": 451
    },
    {
      "epoch": 3.18,
      "learning_rate": 3.242611710638526e-05,
      "loss": 0.6629,
      "step": 452
    },
    {
      "epoch": 3.19,
      "learning_rate": 3.220567813506653e-05,
      "loss": 0.7047,
      "step": 453
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.198563448998691e-05,
      "loss": 0.6103,
      "step": 454
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.176599105972982e-05,
      "loss": 0.7088,
      "step": 455
    },
    {
      "epoch": 3.21,
      "learning_rate": 3.154675272398734e-05,
      "loss": 0.692,
      "step": 456
    },
    {
      "epoch": 3.22,
      "learning_rate": 3.132792435345173e-05,
      "loss": 0.7956,
      "step": 457
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.1109510809707395e-05,
      "loss": 0.5122,
      "step": 458
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.0891516945122735e-05,
      "loss": 0.6636,
      "step": 459
    },
    {
      "epoch": 3.24,
      "learning_rate": 3.06739476027423e-05,
      "loss": 0.5402,
      "step": 460
    },
    {
      "epoch": 3.25,
      "learning_rate": 3.0456807616179385e-05,
      "loss": 0.7909,
      "step": 461
    },
    {
      "epoch": 3.25,
      "learning_rate": 3.0240101809508447e-05,
      "loss": 0.7802,
      "step": 462
    },
    {
      "epoch": 3.26,
      "learning_rate": 3.0023834997158084e-05,
      "loss": 0.6254,
      "step": 463
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.9808011983803974e-05,
      "loss": 0.7085,
      "step": 464
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.9592637564262155e-05,
      "loss": 0.6821,
      "step": 465
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.9377716523382548e-05,
      "loss": 0.5261,
      "step": 466
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.916325363594261e-05,
      "loss": 0.6004,
      "step": 467
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.894925366654125e-05,
      "loss": 0.7009,
      "step": 468
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.8735721369492998e-05,
      "loss": 0.6891,
      "step": 469
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.8522661488722402e-05,
      "loss": 0.7982,
      "step": 470
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.8310078757658552e-05,
      "loss": 0.5411,
      "step": 471
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.809797789913004e-05,
      "loss": 0.9861,
      "step": 472
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.788636362525995e-05,
      "loss": 0.6231,
      "step": 473
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.7675240637361127e-05,
      "loss": 0.8455,
      "step": 474
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.7464613625831947e-05,
      "loss": 0.6272,
      "step": 475
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.725448727005181e-05,
      "loss": 0.6168,
      "step": 476
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.704486623827742e-05,
      "loss": 0.7557,
      "step": 477
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.683575518753899e-05,
      "loss": 0.8038,
      "step": 478
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.662715876353672e-05,
      "loss": 0.5731,
      "step": 479
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.6419081600537676e-05,
      "loss": 0.5925,
      "step": 480
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.6211528321272815e-05,
      "loss": 0.3563,
      "step": 481
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.600450353683426e-05,
      "loss": 0.8047,
      "step": 482
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.5798011846572845e-05,
      "loss": 0.6462,
      "step": 483
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5592057837996008e-05,
      "loss": 0.6358,
      "step": 484
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.5386646086665756e-05,
      "loss": 0.5355,
      "step": 485
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.518178115609714e-05,
      "loss": 0.6574,
      "step": 486
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.4977467597656733e-05,
      "loss": 0.8254,
      "step": 487
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.4773709950461665e-05,
      "loss": 0.6084,
      "step": 488
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.4570512741278672e-05,
      "loss": 0.5879,
      "step": 489
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.436788048442356e-05,
      "loss": 0.5913,
      "step": 490
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.4165817681660923e-05,
      "loss": 0.5537,
      "step": 491
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.3964328822104127e-05,
      "loss": 0.7386,
      "step": 492
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.3763418382115522e-05,
      "loss": 0.6689,
      "step": 493
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.356309082520709e-05,
      "loss": 0.72,
      "step": 494
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.3363350601941254e-05,
      "loss": 0.6096,
      "step": 495
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.3164202149831875e-05,
      "loss": 0.6886,
      "step": 496
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.2965649893245943e-05,
      "loss": 0.75,
      "step": 497
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.276769824330496e-05,
      "loss": 0.7009,
      "step": 498
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.2570351597787166e-05,
      "loss": 0.527,
      "step": 499
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.2373614341029776e-05,
      "loss": 0.8298,
      "step": 500
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.217749084383151e-05,
      "loss": 0.6151,
      "step": 501
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.1981985463355587e-05,
      "loss": 0.4804,
      "step": 502
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.178710254303288e-05,
      "loss": 0.6926,
      "step": 503
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.1592846412465408e-05,
      "loss": 0.5709,
      "step": 504
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.139922138733016e-05,
      "loss": 0.4891,
      "step": 505
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.1206231769283253e-05,
      "loss": 0.7093,
      "step": 506
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.101388184586426e-05,
      "loss": 0.5319,
      "step": 507
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.0822175890401102e-05,
      "loss": 0.5793,
      "step": 508
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.0631118161914965e-05,
      "loss": 0.6307,
      "step": 509
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.0440712905025788e-05,
      "loss": 0.7734,
      "step": 510
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.025096434985793e-05,
      "loss": 0.7496,
      "step": 511
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.006187671194616e-05,
      "loss": 0.7035,
      "step": 512
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.9873454192142038e-05,
      "loss": 0.6277,
      "step": 513
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.9685700976520605e-05,
      "loss": 0.504,
      "step": 514
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.9498621236287275e-05,
      "loss": 0.6165,
      "step": 515
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.931221912768532e-05,
      "loss": 0.5932,
      "step": 516
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.9126498791903464e-05,
      "loss": 0.5616,
      "step": 517
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.894146435498378e-05,
      "loss": 0.7417,
      "step": 518
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.8757119927730266e-05,
      "loss": 0.4589,
      "step": 519
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.8573469605617234e-05,
      "loss": 0.5317,
      "step": 520
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.839051746869854e-05,
      "loss": 0.4569,
      "step": 521
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.8208267581516853e-05,
      "loss": 0.4805,
      "step": 522
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.8026723993013317e-05,
      "loss": 0.6394,
      "step": 523
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.7845890736437686e-05,
      "loss": 0.5725,
      "step": 524
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.766577182925867e-05,
      "loss": 0.6262,
      "step": 525
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.748637127307468e-05,
      "loss": 0.4061,
      "step": 526
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.7307693053524948e-05,
      "loss": 0.5609,
      "step": 527
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.7129741140200966e-05,
      "loss": 0.6813,
      "step": 528
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.6952519486558255e-05,
      "loss": 0.6705,
      "step": 529
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.6776032029828652e-05,
      "loss": 0.5283,
      "step": 530
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.6600282690932685e-05,
      "loss": 0.647,
      "step": 531
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.642527537439259e-05,
      "loss": 0.6204,
      "step": 532
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.625101396824551e-05,
      "loss": 0.7142,
      "step": 533
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.6077502343957117e-05,
      "loss": 0.6346,
      "step": 534
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.5904744356335615e-05,
      "loss": 0.6174,
      "step": 535
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.5732743843446112e-05,
      "loss": 0.5119,
      "step": 536
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.5561504626525274e-05,
      "loss": 0.5757,
      "step": 537
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.539103050989656e-05,
      "loss": 0.6991,
      "step": 538
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.5221325280885606e-05,
      "loss": 0.6397,
      "step": 539
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.5052392709736101e-05,
      "loss": 0.3825,
      "step": 540
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.488423654952606e-05,
      "loss": 0.5771,
      "step": 541
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.4716860536084403e-05,
      "loss": 0.5234,
      "step": 542
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.4550268387907996e-05,
      "loss": 0.6009,
      "step": 543
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.4384463806079024e-05,
      "loss": 0.7015,
      "step": 544
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.4219450474182722e-05,
      "loss": 0.4498,
      "step": 545
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.4055232058225609e-05,
      "loss": 0.553,
      "step": 546
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.3891812206554023e-05,
      "loss": 0.503,
      "step": 547
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.3729194549773039e-05,
      "loss": 0.5165,
      "step": 548
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.3567382700665826e-05,
      "loss": 0.8116,
      "step": 549
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.3406380254113421e-05,
      "loss": 0.7193,
      "step": 550
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.3246190787014767e-05,
      "loss": 0.5503,
      "step": 551
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.3086817858207368e-05,
      "loss": 0.6821,
      "step": 552
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.2928265008388119e-05,
      "loss": 0.5011,
      "step": 553
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.2770535760034718e-05,
      "loss": 0.6798,
      "step": 554
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.2613633617327386e-05,
      "loss": 0.7772,
      "step": 555
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.245756206607101e-05,
      "loss": 0.7508,
      "step": 556
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.2302324573617674e-05,
      "loss": 0.7464,
      "step": 557
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.21479245887897e-05,
      "loss": 0.494,
      "step": 558
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.1994365541802938e-05,
      "loss": 0.6598,
      "step": 559
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.1841650844190639e-05,
      "loss": 0.6079,
      "step": 560
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.1689783888727612e-05,
      "loss": 0.5854,
      "step": 561
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.1538768049354886e-05,
      "loss": 0.6023,
      "step": 562
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.1388606681104701e-05,
      "loss": 0.8083,
      "step": 563
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.1239303120026045e-05,
      "loss": 0.5125,
      "step": 564
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.1090860683110432e-05,
      "loss": 0.7413,
      "step": 565
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.094328266821833e-05,
      "loss": 0.7644,
      "step": 566
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.0796572354005836e-05,
      "loss": 0.5222,
      "step": 567
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.0650732999851793e-05,
      "loss": 0.4381,
      "step": 568
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.050576784578547e-05,
      "loss": 0.5834,
      "step": 569
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.0361680112414518e-05,
      "loss": 0.5592,
      "step": 570
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.0218473000853435e-05,
      "loss": 0.54,
      "step": 571
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.007614969265247e-05,
      "loss": 0.7499,
      "step": 572
    },
    {
      "epoch": 4.04,
      "learning_rate": 9.93471334972687e-06,
      "loss": 0.5057,
      "step": 573
    },
    {
      "epoch": 4.04,
      "learning_rate": 9.794167114286728e-06,
      "loss": 0.6646,
      "step": 574
    },
    {
      "epoch": 4.05,
      "learning_rate": 9.654514108767122e-06,
      "loss": 0.7452,
      "step": 575
    },
    {
      "epoch": 4.06,
      "learning_rate": 9.515757435758743e-06,
      "loss": 0.4576,
      "step": 576
    },
    {
      "epoch": 4.06,
      "learning_rate": 9.377900177938981e-06,
      "loss": 0.5076,
      "step": 577
    },
    {
      "epoch": 4.07,
      "learning_rate": 9.240945398003448e-06,
      "loss": 0.5335,
      "step": 578
    },
    {
      "epoch": 4.08,
      "learning_rate": 9.104896138597884e-06,
      "loss": 0.4748,
      "step": 579
    },
    {
      "epoch": 4.08,
      "learning_rate": 8.969755422250642e-06,
      "loss": 0.6551,
      "step": 580
    },
    {
      "epoch": 4.09,
      "learning_rate": 8.835526251305487e-06,
      "loss": 0.6218,
      "step": 581
    },
    {
      "epoch": 4.1,
      "learning_rate": 8.702211607854871e-06,
      "loss": 0.5859,
      "step": 582
    },
    {
      "epoch": 4.11,
      "learning_rate": 8.569814453673747e-06,
      "loss": 0.4408,
      "step": 583
    },
    {
      "epoch": 4.11,
      "learning_rate": 8.438337730153728e-06,
      "loss": 0.6113,
      "step": 584
    },
    {
      "epoch": 4.12,
      "learning_rate": 8.307784358237736e-06,
      "loss": 0.6236,
      "step": 585
    },
    {
      "epoch": 4.13,
      "learning_rate": 8.178157238355149e-06,
      "loss": 0.6943,
      "step": 586
    },
    {
      "epoch": 4.13,
      "learning_rate": 8.049459250357284e-06,
      "loss": 0.6044,
      "step": 587
    },
    {
      "epoch": 4.14,
      "learning_rate": 7.92169325345351e-06,
      "loss": 0.6424,
      "step": 588
    },
    {
      "epoch": 4.15,
      "learning_rate": 7.794862086147686e-06,
      "loss": 0.5592,
      "step": 589
    },
    {
      "epoch": 4.15,
      "learning_rate": 7.668968566175067e-06,
      "loss": 0.6441,
      "step": 590
    },
    {
      "epoch": 4.16,
      "learning_rate": 7.544015490439743e-06,
      "loss": 0.7385,
      "step": 591
    },
    {
      "epoch": 4.17,
      "learning_rate": 7.420005634952548e-06,
      "loss": 0.7177,
      "step": 592
    },
    {
      "epoch": 4.18,
      "learning_rate": 7.296941754769243e-06,
      "loss": 0.5613,
      "step": 593
    },
    {
      "epoch": 4.18,
      "learning_rate": 7.174826583929478e-06,
      "loss": 0.5203,
      "step": 594
    },
    {
      "epoch": 4.19,
      "learning_rate": 7.053662835395908e-06,
      "loss": 0.5635,
      "step": 595
    },
    {
      "epoch": 4.2,
      "learning_rate": 6.933453200994033e-06,
      "loss": 0.4943,
      "step": 596
    },
    {
      "epoch": 4.2,
      "learning_rate": 6.814200351352323e-06,
      "loss": 0.5714,
      "step": 597
    },
    {
      "epoch": 4.21,
      "learning_rate": 6.695906935842916e-06,
      "loss": 0.5431,
      "step": 598
    },
    {
      "epoch": 4.22,
      "learning_rate": 6.5785755825227554e-06,
      "loss": 0.6152,
      "step": 599
    },
    {
      "epoch": 4.23,
      "learning_rate": 6.462208898075201e-06,
      "loss": 0.3967,
      "step": 600
    },
    {
      "epoch": 4.23,
      "learning_rate": 6.346809467752101e-06,
      "loss": 0.51,
      "step": 601
    },
    {
      "epoch": 4.24,
      "learning_rate": 6.232379855316395e-06,
      "loss": 0.3982,
      "step": 602
    },
    {
      "epoch": 4.25,
      "learning_rate": 6.118922602985133e-06,
      "loss": 0.6577,
      "step": 603
    },
    {
      "epoch": 4.25,
      "learning_rate": 6.006440231372973e-06,
      "loss": 0.6096,
      "step": 604
    },
    {
      "epoch": 4.26,
      "learning_rate": 5.894935239436239e-06,
      "loss": 0.4768,
      "step": 605
    },
    {
      "epoch": 4.27,
      "learning_rate": 5.78441010441736e-06,
      "loss": 0.582,
      "step": 606
    },
    {
      "epoch": 4.27,
      "learning_rate": 5.674867281789842e-06,
      "loss": 0.5339,
      "step": 607
    },
    {
      "epoch": 4.28,
      "learning_rate": 5.566309205203735e-06,
      "loss": 0.3802,
      "step": 608
    },
    {
      "epoch": 4.29,
      "learning_rate": 5.458738286431525e-06,
      "loss": 0.4993,
      "step": 609
    },
    {
      "epoch": 4.3,
      "learning_rate": 5.352156915314594e-06,
      "loss": 0.5956,
      "step": 610
    },
    {
      "epoch": 4.3,
      "learning_rate": 5.2465674597101275e-06,
      "loss": 0.5637,
      "step": 611
    },
    {
      "epoch": 4.31,
      "learning_rate": 5.141972265438455e-06,
      "loss": 0.6436,
      "step": 612
    },
    {
      "epoch": 4.32,
      "learning_rate": 5.0383736562310105e-06,
      "loss": 0.4381,
      "step": 613
    },
    {
      "epoch": 4.32,
      "learning_rate": 4.935773933678645e-06,
      "loss": 0.8673,
      "step": 614
    },
    {
      "epoch": 4.33,
      "learning_rate": 4.8341753771805275e-06,
      "loss": 0.5128,
      "step": 615
    },
    {
      "epoch": 4.34,
      "learning_rate": 4.733580243893509e-06,
      "loss": 0.7217,
      "step": 616
    },
    {
      "epoch": 4.35,
      "learning_rate": 4.633990768681928e-06,
      "loss": 0.5057,
      "step": 617
    },
    {
      "epoch": 4.35,
      "learning_rate": 4.535409164068022e-06,
      "loss": 0.4874,
      "step": 618
    },
    {
      "epoch": 4.36,
      "learning_rate": 4.437837620182745e-06,
      "loss": 0.5992,
      "step": 619
    },
    {
      "epoch": 4.37,
      "learning_rate": 4.341278304717078e-06,
      "loss": 0.7009,
      "step": 620
    },
    {
      "epoch": 4.37,
      "learning_rate": 4.2457333628739605e-06,
      "loss": 0.4728,
      "step": 621
    },
    {
      "epoch": 4.38,
      "learning_rate": 4.1512049173205445e-06,
      "loss": 0.4825,
      "step": 622
    },
    {
      "epoch": 4.39,
      "learning_rate": 4.057695068141043e-06,
      "loss": 0.264,
      "step": 623
    },
    {
      "epoch": 4.39,
      "learning_rate": 3.9652058927901415e-06,
      "loss": 0.6728,
      "step": 624
    },
    {
      "epoch": 4.4,
      "learning_rate": 3.873739446046787e-06,
      "loss": 0.5413,
      "step": 625
    },
    {
      "epoch": 4.41,
      "learning_rate": 3.7832977599685236e-06,
      "loss": 0.5163,
      "step": 626
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.693882843846419e-06,
      "loss": 0.4204,
      "step": 627
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.605496684160348e-06,
      "loss": 0.5551,
      "step": 628
    },
    {
      "epoch": 4.43,
      "learning_rate": 3.518141244534917e-06,
      "loss": 0.7123,
      "step": 629
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.4318184656958074e-06,
      "loss": 0.5059,
      "step": 630
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.3465302654266616e-06,
      "loss": 0.4831,
      "step": 631
    },
    {
      "epoch": 4.45,
      "learning_rate": 3.262278538526492e-06,
      "loss": 0.4672,
      "step": 632
    },
    {
      "epoch": 4.46,
      "learning_rate": 3.17906515676758e-06,
      "loss": 0.4423,
      "step": 633
    },
    {
      "epoch": 4.46,
      "learning_rate": 3.0968919688538656e-06,
      "loss": 0.6304,
      "step": 634
    },
    {
      "epoch": 4.47,
      "learning_rate": 3.0157608003799355e-06,
      "loss": 0.547,
      "step": 635
    },
    {
      "epoch": 4.48,
      "learning_rate": 2.9356734537903997e-06,
      "loss": 0.6087,
      "step": 636
    },
    {
      "epoch": 4.49,
      "learning_rate": 2.8566317083398907e-06,
      "loss": 0.5036,
      "step": 637
    },
    {
      "epoch": 4.49,
      "learning_rate": 2.7786373200535175e-06,
      "loss": 0.5607,
      "step": 638
    },
    {
      "epoch": 4.5,
      "learning_rate": 2.701692021687846e-06,
      "loss": 0.6446,
      "step": 639
    },
    {
      "epoch": 4.51,
      "learning_rate": 2.625797522692425e-06,
      "loss": 0.6112,
      "step": 640
    },
    {
      "epoch": 4.51,
      "learning_rate": 2.550955509171804e-06,
      "loss": 0.4442,
      "step": 641
    },
    {
      "epoch": 4.52,
      "learning_rate": 2.477167643848033e-06,
      "loss": 0.7012,
      "step": 642
    },
    {
      "epoch": 4.53,
      "learning_rate": 2.404435566023794e-06,
      "loss": 0.4942,
      "step": 643
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.3327608915459378e-06,
      "loss": 0.3784,
      "step": 644
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.262145212769562e-06,
      "loss": 0.5823,
      "step": 645
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.192590098522701e-06,
      "loss": 0.4819,
      "step": 646
    },
    {
      "epoch": 4.56,
      "learning_rate": 2.1240970940714146e-06,
      "loss": 0.3844,
      "step": 647
    },
    {
      "epoch": 4.56,
      "learning_rate": 2.056667721085481e-06,
      "loss": 0.622,
      "step": 648
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.9903034776045935e-06,
      "loss": 0.438,
      "step": 649
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.9250058380050674e-06,
      "loss": 0.5024,
      "step": 650
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.8607762529671059e-06,
      "loss": 0.5313,
      "step": 651
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.7976161494425493e-06,
      "loss": 0.6647,
      "step": 652
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.7355269306231602e-06,
      "loss": 0.6355,
      "step": 653
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.6745099759094996e-06,
      "loss": 0.6087,
      "step": 654
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.6145666408802496e-06,
      "loss": 0.5331,
      "step": 655
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.5556982572620783e-06,
      "loss": 0.419,
      "step": 656
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.49790613290009e-06,
      "loss": 0.5193,
      "step": 657
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.4411915517287543e-06,
      "loss": 0.5199,
      "step": 658
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.38555577374338e-06,
      "loss": 0.4913,
      "step": 659
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.3310000349721297e-06,
      "loss": 0.6488,
      "step": 660
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.2775255474485436e-06,
      "loss": 0.3613,
      "step": 661
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.225133499184633e-06,
      "loss": 0.4291,
      "step": 662
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.1738250541444728e-06,
      "loss": 0.3886,
      "step": 663
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.1236013522183341e-06,
      "loss": 0.4071,
      "step": 664
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.0744635091974098e-06,
      "loss": 0.5659,
      "step": 665
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.0264126167489508e-06,
      "loss": 0.4816,
      "step": 666
    },
    {
      "epoch": 4.7,
      "learning_rate": 9.794497423920635e-07,
      "loss": 0.5466,
      "step": 667
    },
    {
      "epoch": 4.7,
      "learning_rate": 9.335759294739899e-07,
      "loss": 0.3509,
      "step": 668
    },
    {
      "epoch": 4.71,
      "learning_rate": 8.887921971469037e-07,
      "loss": 0.485,
      "step": 669
    },
    {
      "epoch": 4.72,
      "learning_rate": 8.450995403452955e-07,
      "loss": 0.6001,
      "step": 670
    },
    {
      "epoch": 4.73,
      "learning_rate": 8.024989297638452e-07,
      "loss": 0.5909,
      "step": 671
    },
    {
      "epoch": 4.73,
      "learning_rate": 7.609913118358791e-07,
      "loss": 0.4665,
      "step": 672
    },
    {
      "epoch": 4.74,
      "learning_rate": 7.205776087123362e-07,
      "loss": 0.5521,
      "step": 673
    },
    {
      "epoch": 4.75,
      "learning_rate": 6.812587182412622e-07,
      "loss": 0.5523,
      "step": 674
    },
    {
      "epoch": 4.75,
      "learning_rate": 6.430355139478871e-07,
      "loss": 0.6478,
      "step": 675
    },
    {
      "epoch": 4.76,
      "learning_rate": 6.059088450152128e-07,
      "loss": 0.5499,
      "step": 676
    },
    {
      "epoch": 4.77,
      "learning_rate": 5.698795362651443e-07,
      "loss": 0.5434,
      "step": 677
    },
    {
      "epoch": 4.77,
      "learning_rate": 5.349483881401496e-07,
      "loss": 0.4236,
      "step": 678
    },
    {
      "epoch": 4.78,
      "learning_rate": 5.011161766855177e-07,
      "loss": 0.4867,
      "step": 679
    },
    {
      "epoch": 4.79,
      "learning_rate": 4.683836535320674e-07,
      "loss": 0.6196,
      "step": 680
    },
    {
      "epoch": 4.8,
      "learning_rate": 4.3675154587949907e-07,
      "loss": 0.5592,
      "step": 681
    },
    {
      "epoch": 4.8,
      "learning_rate": 4.0622055648018573e-07,
      "loss": 0.3231,
      "step": 682
    },
    {
      "epoch": 4.81,
      "learning_rate": 3.7679136362360757e-07,
      "loss": 0.5,
      "step": 683
    },
    {
      "epoch": 4.82,
      "learning_rate": 3.484646211212528e-07,
      "loss": 0.4595,
      "step": 684
    },
    {
      "epoch": 4.82,
      "learning_rate": 3.212409582921128e-07,
      "loss": 0.5326,
      "step": 685
    },
    {
      "epoch": 4.83,
      "learning_rate": 2.951209799486765e-07,
      "loss": 0.6091,
      "step": 686
    },
    {
      "epoch": 4.84,
      "learning_rate": 2.7010526638351906e-07,
      "loss": 0.3778,
      "step": 687
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.4619437335640647e-07,
      "loss": 0.4767,
      "step": 688
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.233888320819111e-07,
      "loss": 0.4401,
      "step": 689
    },
    {
      "epoch": 4.86,
      "learning_rate": 2.0168914921767644e-07,
      "loss": 0.4482,
      "step": 690
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.8109580685309324e-07,
      "loss": 0.7395,
      "step": 691
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.616092624986354e-07,
      "loss": 0.6509,
      "step": 692
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.4322994907568498e-07,
      "loss": 0.495,
      "step": 693
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.259582749068955e-07,
      "loss": 0.6137,
      "step": 694
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.0979462370714344e-07,
      "loss": 0.4285,
      "step": 695
    },
    {
      "epoch": 4.9,
      "learning_rate": 9.473935457500194e-08,
      "loss": 0.5902,
      "step": 696
    },
    {
      "epoch": 4.91,
      "learning_rate": 8.079280198473038e-08,
      "loss": 0.702,
      "step": 697
    },
    {
      "epoch": 4.92,
      "learning_rate": 6.79552757789026e-08,
      "loss": 0.6918,
      "step": 698
    },
    {
      "epoch": 4.92,
      "learning_rate": 5.622706116146237e-08,
      "loss": 0.6861,
      "step": 699
    },
    {
      "epoch": 4.93,
      "learning_rate": 4.560841869142851e-08,
      "loss": 0.4192,
      "step": 700
    },
    {
      "epoch": 4.94,
      "learning_rate": 3.609958427708282e-08,
      "loss": 0.5755,
      "step": 701
    },
    {
      "epoch": 4.94,
      "learning_rate": 2.770076917075204e-08,
      "loss": 0.5318,
      "step": 702
    },
    {
      "epoch": 4.95,
      "learning_rate": 2.0412159964083853e-08,
      "loss": 0.5324,
      "step": 703
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4233918583939077e-08,
      "loss": 0.5296,
      "step": 704
    },
    {
      "epoch": 4.96,
      "learning_rate": 9.166182288750103e-09,
      "loss": 0.7441,
      "step": 705
    },
    {
      "epoch": 4.97,
      "learning_rate": 5.209063665506664e-09,
      "loss": 0.4587,
      "step": 706
    },
    {
      "epoch": 4.98,
      "learning_rate": 2.362650627252272e-09,
      "loss": 0.671,
      "step": 707
    },
    {
      "epoch": 4.99,
      "learning_rate": 6.270064111024709e-10,
      "loss": 0.6712,
      "step": 708
    },
    {
      "epoch": 4.99,
      "learning_rate": 2.169576873711776e-12,
      "loss": 0.4572,
      "step": 709
    },
    {
      "epoch": 5.0,
      "learning_rate": 4.881540062118273e-10,
      "loss": 0.3704,
      "step": 710
    }
  ],
  "logging_steps": 1,
  "max_steps": 710,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 5.732084547059712e+16,
  "trial_name": null,
  "trial_params": null
}