{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 705,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 9.090909090909091e-06,
      "loss": 1.574,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 1.618,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 1.5453,
      "step": 3
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 1.4047,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.545454545454546e-05,
      "loss": 1.4909,
      "step": 5
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.4545454545454546e-05,
      "loss": 1.4884,
      "step": 6
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.363636363636364e-05,
      "loss": 1.449,
      "step": 7
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.272727272727273e-05,
      "loss": 1.4063,
      "step": 8
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.181818181818183e-05,
      "loss": 1.4595,
      "step": 9
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.090909090909092e-05,
      "loss": 1.3918,
      "step": 10
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001,
      "loss": 1.3847,
      "step": 11
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00010909090909090909,
      "loss": 1.4178,
      "step": 12
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001181818181818182,
      "loss": 1.3751,
      "step": 13
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00012727272727272728,
      "loss": 1.3377,
      "step": 14
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00013636363636363637,
      "loss": 1.3761,
      "step": 15
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00014545454545454546,
      "loss": 1.3358,
      "step": 16
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00015454545454545454,
      "loss": 1.3477,
      "step": 17
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00016363636363636366,
      "loss": 1.3174,
      "step": 18
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00017272727272727275,
      "loss": 1.3381,
      "step": 19
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00018181818181818183,
      "loss": 1.3524,
      "step": 20
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019090909090909092,
      "loss": 1.3306,
      "step": 21
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0002,
      "loss": 1.3218,
      "step": 22
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019999894214150818,
      "loss": 1.3003,
      "step": 23
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019999576858841395,
      "loss": 1.3422,
      "step": 24
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019999047940786073,
      "loss": 1.3054,
      "step": 25
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019998307471175264,
      "loss": 1.3046,
      "step": 26
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019997355465675205,
      "loss": 1.3248,
      "step": 27
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019996191944427638,
      "loss": 1.3328,
      "step": 28
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019994816932049383,
      "loss": 1.3285,
      "step": 29
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001999323045763181,
      "loss": 1.3223,
      "step": 30
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019991432554740225,
      "loss": 1.3395,
      "step": 31
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001998942326141317,
      "loss": 1.3085,
      "step": 32
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.000199872026201616,
      "loss": 1.329,
      "step": 33
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019984770677968,
      "loss": 1.2921,
      "step": 34
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019982127486285384,
      "loss": 1.3047,
      "step": 35
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001997927310103621,
      "loss": 1.2923,
      "step": 36
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019976207582611189,
      "loss": 1.2481,
      "step": 37
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019972930995868014,
      "loss": 1.2574,
      "step": 38
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001996944341012999,
      "loss": 1.3083,
      "step": 39
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001996574489918456,
      "loss": 1.2613,
      "step": 40
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019961835541281746,
      "loss": 1.2648,
      "step": 41
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019957715419132498,
      "loss": 1.2976,
      "step": 42
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019953384619906945,
      "loss": 1.3284,
      "step": 43
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019948843235232535,
      "loss": 1.2759,
      "step": 44
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001994409136119212,
      "loss": 1.2619,
      "step": 45
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019939129098321904,
      "loss": 1.2225,
      "step": 46
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019933956551609322,
      "loss": 1.3009,
      "step": 47
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019928573830490826,
      "loss": 1.3523,
      "step": 48
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019922981048849564,
      "loss": 1.2688,
      "step": 49
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019917178325012963,
      "loss": 1.2939,
      "step": 50
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019911165781750237,
      "loss": 1.2932,
      "step": 51
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019904943546269785,
      "loss": 1.2886,
      "step": 52
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019898511750216505,
      "loss": 1.2656,
      "step": 53
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001989187052966899,
      "loss": 1.2864,
      "step": 54
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019885020025136677,
      "loss": 1.287,
      "step": 55
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001987796038155685,
      "loss": 1.2534,
      "step": 56
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001987069174829159,
      "loss": 1.3241,
      "step": 57
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019863214279124608,
      "loss": 1.2915,
      "step": 58
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019855528132257984,
      "loss": 1.2197,
      "step": 59
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019847633470308833,
      "loss": 1.2212,
      "step": 60
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019839530460305862,
      "loss": 1.2421,
      "step": 61
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019831219273685826,
      "loss": 1.2741,
      "step": 62
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019822700086289915,
      "loss": 1.2633,
      "step": 63
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019813973078360025,
      "loss": 1.2529,
      "step": 64
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001980503843453494,
      "loss": 1.2905,
      "step": 65
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019795896343846437,
      "loss": 1.2651,
      "step": 66
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019786546999715285,
      "loss": 1.2507,
      "step": 67
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019776990599947147,
      "loss": 1.2113,
      "step": 68
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019767227346728392,
      "loss": 1.262,
      "step": 69
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019757257446621827,
      "loss": 1.2706,
      "step": 70
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019747081110562322,
      "loss": 1.2491,
      "step": 71
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001973669855385235,
      "loss": 1.3059,
      "step": 72
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019726109996157424,
      "loss": 1.2589,
      "step": 73
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001971531566150145,
      "loss": 1.2999,
      "step": 74
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019704315778262016,
      "loss": 1.2421,
      "step": 75
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019693110579165513,
      "loss": 1.3066,
      "step": 76
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019681700301282234,
      "loss": 1.2259,
      "step": 77
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019670085186021375,
      "loss": 1.2563,
      "step": 78
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.000196582654791259,
      "loss": 1.2243,
      "step": 79
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019646241430667353,
      "loss": 1.2172,
      "step": 80
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001963401329504057,
      "loss": 1.2653,
      "step": 81
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019621581330958295,
      "loss": 1.3283,
      "step": 82
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.000196089458014457,
      "loss": 1.2118,
      "step": 83
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019596106973834835,
      "loss": 1.2633,
      "step": 84
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001958306511975895,
      "loss": 1.2357,
      "step": 85
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019569820515146768,
      "loss": 1.2624,
      "step": 86
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001955637344021664,
      "loss": 1.2221,
      "step": 87
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019542724179470616,
      "loss": 1.2304,
      "step": 88
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001952887302168842,
      "loss": 1.235,
      "step": 89
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019514820259921352,
      "loss": 1.2106,
      "step": 90
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019500566191486075,
      "loss": 1.2013,
      "step": 91
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019486111117958342,
      "loss": 1.2796,
      "step": 92
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019471455345166595,
      "loss": 1.195,
      "step": 93
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019456599183185507,
      "loss": 1.2813,
      "step": 94
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019441542946329422,
      "loss": 1.2356,
      "step": 95
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019426286953145704,
      "loss": 1.2815,
      "step": 96
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019410831526407984,
      "loss": 1.2286,
      "step": 97
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019395176993109356,
      "loss": 1.2427,
      "step": 98
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001937932368445544,
      "loss": 1.2076,
      "step": 99
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019363271935857372,
      "loss": 1.2544,
      "step": 100
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019347022086924732,
      "loss": 1.2632,
      "step": 101
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019330574481458333,
      "loss": 1.2088,
      "step": 102
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019313929467442952,
      "loss": 1.2348,
      "step": 103
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019297087397039984,
      "loss": 1.2186,
      "step": 104
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019280048626579962,
      "loss": 1.2039,
      "step": 105
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001926281351655506,
      "loss": 1.2665,
      "step": 106
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001924538243161142,
      "loss": 1.2293,
      "step": 107
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001922775574054147,
      "loss": 1.2712,
      "step": 108
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019209933816276102,
      "loss": 1.2019,
      "step": 109
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019191917035876798,
      "loss": 1.2248,
      "step": 110
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019173705780527642,
      "loss": 1.2718,
      "step": 111
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019155300435527256,
      "loss": 1.1894,
      "step": 112
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019136701390280644,
      "loss": 1.2603,
      "step": 113
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019117909038290974,
      "loss": 1.2292,
      "step": 114
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019098923777151222,
      "loss": 1.2264,
      "step": 115
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019079746008535784,
      "loss": 1.2553,
      "step": 116
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0001906037613819197,
      "loss": 1.287,
      "step": 117
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019040814575931413,
      "loss": 1.1868,
      "step": 118
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019021061735621412,
      "loss": 1.2465,
      "step": 119
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019001118035176162,
      "loss": 1.2324,
      "step": 120
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0001898098389654792,
      "loss": 1.242,
      "step": 121
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0001896065974571808,
      "loss": 1.2202,
      "step": 122
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00018940146012688146,
      "loss": 1.2269,
      "step": 123
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00018919443131470658,
      "loss": 1.2456,
      "step": 124
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00018898551540079989,
      "loss": 1.1917,
      "step": 125
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00018877471680523082,
      "loss": 1.2336,
      "step": 126
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00018856203998790112,
      "loss": 1.2126,
      "step": 127
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00018834748944845028,
      "loss": 1.2487,
      "step": 128
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00018813106972616055,
      "loss": 1.2732,
      "step": 129
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001879127853998607,
      "loss": 1.2385,
      "step": 130
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00018769264108782933,
      "loss": 1.2294,
      "step": 131
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00018747064144769703,
      "loss": 1.2875,
      "step": 132
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001872467911763479,
      "loss": 1.266,
      "step": 133
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001870210950098201,
      "loss": 1.1919,
      "step": 134
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00018679355772320585,
      "loss": 1.2646,
      "step": 135
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00018656418413055007,
      "loss": 1.2151,
      "step": 136
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0001863329790847488,
      "loss": 1.224,
      "step": 137
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001860999474774466,
      "loss": 1.2197,
      "step": 138
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00018586509423893267,
      "loss": 1.2487,
      "step": 139
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00018562842433803687,
      "loss": 1.2164,
      "step": 140
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00018538994278202448,
      "loss": 1.2395,
      "step": 141
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001851496546164903,
      "loss": 1.2454,
      "step": 142
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00018490756492525187,
      "loss": 1.2096,
      "step": 143
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00018466367883024186,
      "loss": 1.1975,
      "step": 144
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00018441800149139988,
      "loss": 1.19,
      "step": 145
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00018417053810656302,
      "loss": 1.2228,
      "step": 146
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001839212939113562,
      "loss": 1.2044,
      "step": 147
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00018367027417908117,
      "loss": 1.2381,
      "step": 148
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00018341748422060503,
      "loss": 1.225,
      "step": 149
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00018316292938424787,
      "loss": 1.2104,
      "step": 150
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00018290661505566963,
      "loss": 1.229,
      "step": 151
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00018264854665775605,
      "loss": 1.2396,
      "step": 152
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0001823887296505041,
      "loss": 1.2132,
      "step": 153
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00018212716953090624,
      "loss": 1.2255,
      "step": 154
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00018186387183283443,
      "loss": 1.2393,
      "step": 155
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00018159884212692274,
      "loss": 1.273,
      "step": 156
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00018133208602044972,
      "loss": 1.2314,
      "step": 157
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00018106360915721956,
      "loss": 1.1983,
      "step": 158
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0001807934172174429,
      "loss": 1.1759,
      "step": 159
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00018052151591761644,
      "loss": 1.2081,
      "step": 160
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0001802479110104022,
      "loss": 1.1594,
      "step": 161
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0001799726082845057,
      "loss": 1.2388,
      "step": 162
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00017969561356455336,
      "loss": 1.2052,
      "step": 163
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00017941693271096966,
      "loss": 1.2385,
      "step": 164
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00017913657161985268,
      "loss": 1.225,
      "step": 165
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0001788545362228496,
      "loss": 1.2173,
      "step": 166
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00017857083248703126,
      "loss": 1.1866,
      "step": 167
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00017828546641476578,
      "loss": 1.2196,
      "step": 168
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0001779984440435916,
      "loss": 1.2535,
      "step": 169
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00017770977144608978,
      "loss": 1.2124,
      "step": 170
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001774194547297555,
      "loss": 1.2213,
      "step": 171
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00017712750003686883,
      "loss": 1.165,
      "step": 172
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0001768339135443648,
      "loss": 1.2336,
      "step": 173
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00017653870146370267,
      "loss": 1.2574,
      "step": 174
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00017624187004073463,
      "loss": 1.201,
      "step": 175
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001759434255555734,
      "loss": 1.2127,
      "step": 176
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00017564337432245976,
      "loss": 1.1924,
      "step": 177
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00017534172268962852,
      "loss": 1.293,
      "step": 178
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00017503847703917455,
      "loss": 1.3086,
      "step": 179
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0001747336437869176,
      "loss": 1.2584,
      "step": 180
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00017442722938226647,
      "loss": 1.242,
      "step": 181
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00017411924030808284,
      "loss": 1.1669,
      "step": 182
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00017380968308054385,
      "loss": 1.2288,
      "step": 183
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0001734985642490043,
      "loss": 1.2583,
      "step": 184
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00017318589039585816,
      "loss": 1.2701,
      "step": 185
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0001728716681363993,
      "loss": 1.2294,
      "step": 186
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00017255590411868136,
      "loss": 1.1843,
      "step": 187
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00017223860502337733,
      "loss": 1.2367,
      "step": 188
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00017191977756363808,
      "loss": 1.2128,
      "step": 189
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00017159942848495025,
      "loss": 1.1754,
      "step": 190
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00017127756456499372,
      "loss": 1.1917,
      "step": 191
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0001709541926134982,
      "loss": 1.2122,
      "step": 192
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0001706293194720989,
      "loss": 1.2336,
      "step": 193
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00017030295201419206,
      "loss": 1.2285,
      "step": 194
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00016997509714478944,
      "loss": 1.2509,
      "step": 195
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00016964576180037217,
      "loss": 1.217,
      "step": 196
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00016931495294874408,
      "loss": 1.2189,
      "step": 197
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00016898267758888423,
      "loss": 1.1784,
      "step": 198
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00016864894275079882,
      "loss": 1.2197,
      "step": 199
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00016831375549537252,
      "loss": 1.2552,
      "step": 200
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00016797712291421904,
      "loss": 1.1743,
      "step": 201
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00016763905212953102,
      "loss": 1.2396,
      "step": 202
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001672995502939295,
      "loss": 1.2283,
      "step": 203
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00016695862459031248,
      "loss": 1.1892,
      "step": 204
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00016661628223170295,
      "loss": 1.1739,
      "step": 205
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00016627253046109638,
      "loss": 1.1211,
      "step": 206
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001659273765513073,
      "loss": 1.1466,
      "step": 207
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00016558082780481563,
      "loss": 1.1761,
      "step": 208
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00016523289155361204,
      "loss": 1.2146,
      "step": 209
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00016488357515904295,
      "loss": 1.2032,
      "step": 210
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0001645328860116546,
      "loss": 1.2129,
      "step": 211
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00016418083153103683,
      "loss": 1.2211,
      "step": 212
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001638274191656661,
      "loss": 1.2022,
      "step": 213
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00016347265639274778,
      "loss": 1.2057,
      "step": 214
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00016311655071805822,
      "loss": 1.229,
      "step": 215
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00016275910967578558,
      "loss": 1.1658,
      "step": 216
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00016240034082837078,
      "loss": 1.2102,
      "step": 217
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00016204025176634712,
      "loss": 1.2057,
      "step": 218
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00016167885010818017,
      "loss": 1.222,
      "step": 219
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00016131614350010614,
      "loss": 1.1824,
      "step": 220
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00016095213961597033,
      "loss": 1.2212,
      "step": 221
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00016058684615706477,
      "loss": 1.2101,
      "step": 222
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00016022027085196516,
      "loss": 1.153,
      "step": 223
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001598524214563675,
      "loss": 1.1471,
      "step": 224
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00015948330575292401,
      "loss": 1.1859,
      "step": 225
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0001591129315510782,
      "loss": 1.1612,
      "step": 226
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00015874130668690003,
      "loss": 1.1691,
      "step": 227
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00015836843902291984,
      "loss": 1.1859,
      "step": 228
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00015799433644796216,
      "loss": 1.1831,
      "step": 229
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00015761900687697865,
      "loss": 1.226,
      "step": 230
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00015724245825088086,
      "loss": 1.233,
      "step": 231
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00015686469853637192,
      "loss": 1.2146,
      "step": 232
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00015648573572577839,
      "loss": 1.1979,
      "step": 233
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001561055778368807,
      "loss": 1.1544,
      "step": 234
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00015572423291274393,
      "loss": 1.1036,
      "step": 235
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00015534170902154742,
      "loss": 0.9431,
      "step": 236
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00015495801425641407,
      "loss": 0.9798,
      "step": 237
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0001545731567352392,
      "loss": 0.9381,
      "step": 238
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00015418714460051875,
      "loss": 0.9564,
      "step": 239
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00015379998601917704,
      "loss": 0.9039,
      "step": 240
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0001534116891823939,
      "loss": 0.8984,
      "step": 241
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00015302226230543147,
      "loss": 0.952,
      "step": 242
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00015263171362746026,
      "loss": 0.94,
      "step": 243
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0001522400514113851,
      "loss": 0.9551,
      "step": 244
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00015184728394366988,
      "loss": 1.0139,
      "step": 245
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00015145341953416271,
      "loss": 0.9337,
      "step": 246
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001510584665159198,
      "loss": 0.9504,
      "step": 247
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00015066243324502918,
      "loss": 0.9361,
      "step": 248
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00015026532810043407,
      "loss": 0.9395,
      "step": 249
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00014986715948375542,
      "loss": 0.9442,
      "step": 250
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00014946793581911428,
      "loss": 0.9664,
      "step": 251
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00014906766555295358,
      "loss": 0.8896,
      "step": 252
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00014866635715385927,
      "loss": 0.9328,
      "step": 253
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0001482640191123813,
      "loss": 0.8727,
      "step": 254
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00014786065994085396,
      "loss": 0.8948,
      "step": 255
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00014745628817321578,
      "loss": 0.9564,
      "step": 256
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00014705091236482887,
      "loss": 0.9485,
      "step": 257
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00014664454109229808,
      "loss": 0.931,
      "step": 258
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00014623718295328944,
      "loss": 0.9614,
      "step": 259
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00014582884656634827,
      "loss": 0.9347,
      "step": 260
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00014541954057071692,
      "loss": 0.8861,
      "step": 261
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00014500927362615177,
      "loss": 0.9045,
      "step": 262
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00014459805441274028,
      "loss": 0.9271,
      "step": 263
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00014418589163071722,
      "loss": 0.8829,
      "step": 264
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00014377279400028053,
      "loss": 0.9252,
      "step": 265
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00014335877026140688,
      "loss": 0.9299,
      "step": 266
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.000142943829173667,
      "loss": 0.8597,
      "step": 267
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00014252797951603977,
      "loss": 0.8919,
      "step": 268
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00014211123008672712,
      "loss": 0.9356,
      "step": 269
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0001416935897029675,
      "loss": 0.8959,
      "step": 270
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0001412750672008494,
      "loss": 0.8936,
      "step": 271
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00014085567143512457,
      "loss": 0.858,
      "step": 272
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00014043541127902037,
      "loss": 0.8618,
      "step": 273
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00014001429562405225,
      "loss": 0.89,
      "step": 274
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00013959233337983582,
      "loss": 0.9272,
      "step": 275
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00013916953347389776,
      "loss": 0.8889,
      "step": 276
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0001387459048514876,
      "loss": 0.8864,
      "step": 277
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00013832145647538799,
      "loss": 0.9299,
      "step": 278
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00013789619732572538,
      "loss": 0.9379,
      "step": 279
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00013747013639977973,
      "loss": 0.8988,
      "step": 280
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001370432827117945,
      "loss": 0.8828,
      "step": 281
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001366156452927856,
      "loss": 0.8929,
      "step": 282
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00013618723319035056,
      "loss": 0.903,
      "step": 283
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.000135758055468477,
      "loss": 0.946,
      "step": 284
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00013532812120735087,
      "loss": 0.8985,
      "step": 285
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0001348974395031643,
      "loss": 0.9617,
      "step": 286
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00013446601946792334,
      "loss": 0.8937,
      "step": 287
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00013403387022925488,
      "loss": 0.9223,
      "step": 288
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00013360100093021376,
      "loss": 0.8876,
      "step": 289
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00013316742072908927,
      "loss": 0.8735,
      "step": 290
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0001327331387992114,
      "loss": 0.8766,
      "step": 291
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00013229816432875664,
      "loss": 0.8885,
      "step": 292
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00013186250652055378,
      "loss": 0.8816,
      "step": 293
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00013142617459188899,
      "loss": 0.8838,
      "step": 294
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0001309891777743111,
      "loss": 0.8981,
      "step": 295
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00013055152531343592,
      "loss": 0.8672,
      "step": 296
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00013011322646875088,
      "loss": 0.9303,
      "step": 297
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00012967429051341913,
      "loss": 0.8727,
      "step": 298
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00012923472673408322,
      "loss": 0.896,
      "step": 299
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0001287945444306686,
      "loss": 0.9047,
      "step": 300
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00012835375291618716,
      "loss": 0.8641,
      "step": 301
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00012791236151653973,
      "loss": 0.8577,
      "step": 302
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00012747037957031916,
      "loss": 0.8815,
      "step": 303
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00012702781642861253,
      "loss": 0.9196,
      "step": 304
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00012658468145480337,
      "loss": 0.8782,
      "step": 305
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00012614098402437366,
      "loss": 0.953,
      "step": 306
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00012569673352470523,
      "loss": 0.9381,
      "step": 307
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00012525193935488137,
      "loss": 0.8818,
      "step": 308
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00012480661092548786,
      "loss": 0.9057,
      "step": 309
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012436075765841396,
      "loss": 0.8975,
      "step": 310
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012391438898665287,
      "loss": 0.918,
      "step": 311
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00012346751435410248,
      "loss": 0.9211,
      "step": 312
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0001230201432153653,
      "loss": 0.9608,
      "step": 313
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00012257228503554835,
      "loss": 0.8869,
      "step": 314
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00012212394929006336,
      "loss": 0.897,
      "step": 315
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00012167514546442576,
      "loss": 0.8858,
      "step": 316
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00012122588305405434,
      "loss": 0.8305,
      "step": 317
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001207761715640702,
      "loss": 0.9205,
      "step": 318
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00012032602050909574,
      "loss": 0.922,
      "step": 319
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00011987543941305321,
      "loss": 0.9019,
      "step": 320
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00011942443780896351,
      "loss": 0.9028,
      "step": 321
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00011897302523874405,
      "loss": 0.9307,
      "step": 322
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0001185212112530073,
      "loss": 0.9475,
      "step": 323
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001180690054108585,
      "loss": 0.9168,
      "step": 324
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011761641727969343,
      "loss": 0.8799,
      "step": 325
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011716345643499608,
      "loss": 0.9363,
      "step": 326
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011671013246013596,
      "loss": 0.852,
      "step": 327
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00011625645494616535,
      "loss": 0.9343,
      "step": 328
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001158024334916165,
      "loss": 0.916,
      "step": 329
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00011534807770229845,
      "loss": 0.9824,
      "step": 330
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00011489339719109378,
      "loss": 0.9317,
      "step": 331
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00011443840157775527,
      "loss": 0.9233,
      "step": 332
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00011398310048870247,
      "loss": 0.9104,
      "step": 333
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00011352750355681772,
      "loss": 0.9006,
      "step": 334
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00011307162042124277,
      "loss": 0.926,
      "step": 335
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00011261546072717454,
      "loss": 0.9303,
      "step": 336
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00011215903412566111,
      "loss": 0.8762,
      "step": 337
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00011170235027339766,
      "loss": 0.8965,
      "step": 338
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00011124541883252198,
      "loss": 0.9352,
      "step": 339
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00011078824947041016,
      "loss": 0.9081,
      "step": 340
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00011033085185947208,
      "loss": 0.9391,
      "step": 341
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010987323567694661,
      "loss": 0.8827,
      "step": 342
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010941541060469712,
      "loss": 0.9139,
      "step": 343
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00010895738632900636,
      "loss": 0.8791,
      "step": 344
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00010849917254037174,
      "loss": 0.8964,
      "step": 345
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00010804077893330023,
      "loss": 0.924,
      "step": 346
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00010758221520610321,
      "loss": 0.8887,
      "step": 347
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00010712349106069131,
      "loss": 0.9306,
      "step": 348
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00010666461620236922,
      "loss": 0.8802,
      "step": 349
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00010620560033963025,
      "loss": 0.897,
      "step": 350
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00010574645318395095,
      "loss": 0.9017,
      "step": 351
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00010528718444958567,
      "loss": 0.8665,
      "step": 352
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00010482780385336106,
      "loss": 0.9159,
      "step": 353
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00010436832111447034,
      "loss": 0.8568,
      "step": 354
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00010390874595426794,
      "loss": 0.9258,
      "step": 355
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00010344908809606353,
      "loss": 0.8684,
      "step": 356
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00010298935726491648,
      "loss": 0.8991,
      "step": 357
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00010252956318743006,
      "loss": 0.8716,
      "step": 358
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0001020697155915457,
      "loss": 0.9192,
      "step": 359
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.000101609824206337,
      "loss": 0.8241,
      "step": 360
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00010114989876180423,
      "loss": 0.8991,
      "step": 361
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00010068994898866804,
      "loss": 0.8729,
      "step": 362
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00010022998461816389,
      "loss": 0.9218,
      "step": 363
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.977001538183616e-05,
      "loss": 0.8121,
      "step": 364
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.9310051011332e-05,
      "loss": 0.8794,
      "step": 365
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.88501012381958e-05,
      "loss": 0.9085,
      "step": 366
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.839017579366299e-05,
      "loss": 0.9281,
      "step": 367
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.793028440845434e-05,
      "loss": 0.8583,
      "step": 368
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.747043681256996e-05,
      "loss": 0.8277,
      "step": 369
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.701064273508356e-05,
      "loss": 0.8954,
      "step": 370
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.65509119039365e-05,
      "loss": 0.9306,
      "step": 371
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.609125404573211e-05,
      "loss": 0.9431,
      "step": 372
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.563167888552968e-05,
      "loss": 0.9396,
      "step": 373
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.517219614663896e-05,
      "loss": 0.8953,
      "step": 374
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.471281555041432e-05,
      "loss": 0.8922,
      "step": 375
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.425354681604907e-05,
      "loss": 0.8737,
      "step": 376
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.379439966036977e-05,
      "loss": 0.8988,
      "step": 377
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.333538379763079e-05,
      "loss": 0.8868,
      "step": 378
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.28765089393087e-05,
      "loss": 0.9303,
      "step": 379
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.241778479389683e-05,
      "loss": 0.934,
      "step": 380
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.195922106669981e-05,
      "loss": 0.8812,
      "step": 381
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.150082745962828e-05,
      "loss": 0.867,
      "step": 382
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.104261367099365e-05,
      "loss": 0.8649,
      "step": 383
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.058458939530295e-05,
      "loss": 0.9221,
      "step": 384
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.01267643230534e-05,
      "loss": 0.859,
      "step": 385
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.966914814052796e-05,
      "loss": 0.9446,
      "step": 386
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.921175052958985e-05,
      "loss": 0.885,
      "step": 387
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.875458116747806e-05,
      "loss": 0.8868,
      "step": 388
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.829764972660237e-05,
      "loss": 0.9057,
      "step": 389
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.78409658743389e-05,
      "loss": 0.8473,
      "step": 390
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.738453927282548e-05,
      "loss": 0.9346,
      "step": 391
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.692837957875725e-05,
      "loss": 0.9158,
      "step": 392
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.647249644318232e-05,
      "loss": 0.8827,
      "step": 393
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.601689951129757e-05,
      "loss": 0.8713,
      "step": 394
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.556159842224472e-05,
      "loss": 0.923,
      "step": 395
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.510660280890625e-05,
      "loss": 0.8329,
      "step": 396
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.465192229770156e-05,
      "loss": 0.8489,
      "step": 397
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.41975665083835e-05,
      "loss": 0.8915,
      "step": 398
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.374354505383467e-05,
      "loss": 0.8903,
      "step": 399
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.328986753986409e-05,
      "loss": 0.913,
      "step": 400
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.283654356500394e-05,
      "loss": 0.8553,
      "step": 401
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.238358272030658e-05,
      "loss": 0.905,
      "step": 402
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.193099458914148e-05,
      "loss": 0.8946,
      "step": 403
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.147878874699274e-05,
      "loss": 0.9124,
      "step": 404
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.102697476125597e-05,
      "loss": 0.8419,
      "step": 405
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.057556219103653e-05,
      "loss": 0.905,
      "step": 406
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.012456058694678e-05,
      "loss": 0.8736,
      "step": 407
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.967397949090431e-05,
      "loss": 0.8963,
      "step": 408
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.922382843592984e-05,
      "loss": 0.9139,
      "step": 409
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.87741169459457e-05,
      "loss": 0.8537,
      "step": 410
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.832485453557424e-05,
      "loss": 0.8808,
      "step": 411
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.787605070993668e-05,
      "loss": 0.867,
      "step": 412
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.742771496445167e-05,
      "loss": 0.8718,
      "step": 413
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.697985678463476e-05,
      "loss": 0.8623,
      "step": 414
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.653248564589751e-05,
      "loss": 0.8879,
      "step": 415
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.608561101334714e-05,
      "loss": 0.854,
      "step": 416
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.563924234158607e-05,
      "loss": 0.9452,
      "step": 417
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.519338907451215e-05,
      "loss": 0.8661,
      "step": 418
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.474806064511863e-05,
      "loss": 0.8852,
      "step": 419
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.43032664752948e-05,
      "loss": 0.8795,
      "step": 420
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.385901597562637e-05,
      "loss": 0.8973,
      "step": 421
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.341531854519664e-05,
      "loss": 0.8834,
      "step": 422
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.29721835713875e-05,
      "loss": 0.8979,
      "step": 423
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.25296204296809e-05,
      "loss": 0.8892,
      "step": 424
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.208763848346029e-05,
      "loss": 0.8541,
      "step": 425
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.164624708381285e-05,
      "loss": 0.8822,
      "step": 426
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.120545556933138e-05,
      "loss": 0.854,
      "step": 427
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.076527326591682e-05,
      "loss": 0.8699,
      "step": 428
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.03257094865809e-05,
      "loss": 0.9127,
      "step": 429
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.988677353124913e-05,
      "loss": 0.894,
      "step": 430
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.94484746865641e-05,
      "loss": 0.9312,
      "step": 431
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.901082222568895e-05,
      "loss": 0.9396,
      "step": 432
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.857382540811101e-05,
      "loss": 0.8763,
      "step": 433
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.813749347944625e-05,
      "loss": 0.8556,
      "step": 434
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.770183567124337e-05,
      "loss": 0.8683,
      "step": 435
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.726686120078862e-05,
      "loss": 0.8735,
      "step": 436
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.683257927091074e-05,
      "loss": 0.9203,
      "step": 437
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.639899906978626e-05,
      "loss": 0.9083,
      "step": 438
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.596612977074515e-05,
      "loss": 0.9017,
      "step": 439
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.553398053207671e-05,
      "loss": 0.8662,
      "step": 440
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.510256049683571e-05,
      "loss": 0.8486,
      "step": 441
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.467187879264916e-05,
      "loss": 0.8897,
      "step": 442
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.4241944531523e-05,
      "loss": 0.9247,
      "step": 443
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.381276680964947e-05,
      "loss": 0.922,
      "step": 444
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.338435470721442e-05,
      "loss": 0.897,
      "step": 445
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.295671728820553e-05,
      "loss": 0.898,
      "step": 446
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.252986360022029e-05,
      "loss": 0.8948,
      "step": 447
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.210380267427467e-05,
      "loss": 0.8923,
      "step": 448
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.167854352461202e-05,
      "loss": 0.8976,
      "step": 449
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.125409514851243e-05,
      "loss": 0.9255,
      "step": 450
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.083046652610224e-05,
      "loss": 0.8501,
      "step": 451
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.040766662016424e-05,
      "loss": 0.9446,
      "step": 452
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.998570437594775e-05,
      "loss": 0.8504,
      "step": 453
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.9564588720979655e-05,
      "loss": 0.8629,
      "step": 454
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.914432856487544e-05,
      "loss": 0.826,
      "step": 455
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.8724932799150586e-05,
      "loss": 0.883,
      "step": 456
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.830641029703254e-05,
      "loss": 0.8657,
      "step": 457
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.788876991327288e-05,
      "loss": 0.893,
      "step": 458
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.747202048396023e-05,
      "loss": 0.8951,
      "step": 459
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.705617082633306e-05,
      "loss": 0.8968,
      "step": 460
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.664122973859313e-05,
      "loss": 0.849,
      "step": 461
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.622720599971952e-05,
      "loss": 0.8614,
      "step": 462
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.5814108369282824e-05,
      "loss": 0.8236,
      "step": 463
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.540194558725973e-05,
      "loss": 0.8196,
      "step": 464
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.4990726373848243e-05,
      "loss": 0.8639,
      "step": 465
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.458045942928309e-05,
      "loss": 0.8689,
      "step": 466
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.417115343365171e-05,
      "loss": 0.8379,
      "step": 467
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.37628170467106e-05,
      "loss": 0.8694,
      "step": 468
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.3355458907701925e-05,
      "loss": 0.8661,
      "step": 469
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.2949087635171144e-05,
      "loss": 0.6885,
      "step": 470
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.254371182678424e-05,
      "loss": 0.6118,
      "step": 471
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.213934005914607e-05,
      "loss": 0.6482,
      "step": 472
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.173598088761874e-05,
      "loss": 0.5909,
      "step": 473
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.133364284614077e-05,
      "loss": 0.6283,
      "step": 474
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.093233444704641e-05,
      "loss": 0.5609,
      "step": 475
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.053206418088572e-05,
      "loss": 0.5663,
      "step": 476
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.0132840516244604e-05,
      "loss": 0.5876,
      "step": 477
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.9734671899565955e-05,
      "loss": 0.6185,
      "step": 478
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.933756675497082e-05,
      "loss": 0.6254,
      "step": 479
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.894153348408021e-05,
      "loss": 0.5577,
      "step": 480
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.8546580465837274e-05,
      "loss": 0.6386,
      "step": 481
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.815271605633012e-05,
      "loss": 0.5778,
      "step": 482
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.775994858861492e-05,
      "loss": 0.6156,
      "step": 483
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.7368286372539775e-05,
      "loss": 0.6072,
      "step": 484
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.697773769456859e-05,
      "loss": 0.6396,
      "step": 485
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.658831081760614e-05,
      "loss": 0.585,
      "step": 486
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.6200013980822954e-05,
      "loss": 0.5711,
      "step": 487
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.5812855399481256e-05,
      "loss": 0.6314,
      "step": 488
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.542684326476082e-05,
      "loss": 0.6257,
      "step": 489
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.504198574358596e-05,
      "loss": 0.6016,
      "step": 490
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.465829097845261e-05,
      "loss": 0.5966,
      "step": 491
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.427576708725609e-05,
      "loss": 0.5442,
      "step": 492
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.389442216311933e-05,
      "loss": 0.6366,
      "step": 493
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.351426427422165e-05,
      "loss": 0.5575,
      "step": 494
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.313530146362809e-05,
      "loss": 0.6181,
      "step": 495
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.275754174911921e-05,
      "loss": 0.6203,
      "step": 496
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.2380993123021385e-05,
      "loss": 0.5802,
      "step": 497
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.200566355203784e-05,
      "loss": 0.5784,
      "step": 498
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.163156097708014e-05,
      "loss": 0.6171,
      "step": 499
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.1258693313099996e-05,
      "loss": 0.5918,
      "step": 500
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.088706844892182e-05,
      "loss": 0.5546,
      "step": 501
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.051669424707602e-05,
      "loss": 0.6072,
      "step": 502
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.014757854363249e-05,
      "loss": 0.618,
      "step": 503
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.977972914803486e-05,
      "loss": 0.5796,
      "step": 504
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.9413153842935255e-05,
      "loss": 0.5466,
      "step": 505
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.9047860384029675e-05,
      "loss": 0.5826,
      "step": 506
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.868385649989388e-05,
      "loss": 0.5873,
      "step": 507
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.832114989181988e-05,
      "loss": 0.5975,
      "step": 508
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.795974823365287e-05,
      "loss": 0.6079,
      "step": 509
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.759965917162925e-05,
      "loss": 0.5623,
      "step": 510
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.724089032421441e-05,
      "loss": 0.5978,
      "step": 511
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.688344928194181e-05,
      "loss": 0.5926,
      "step": 512
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.652734360725224e-05,
      "loss": 0.5724,
      "step": 513
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.617258083433396e-05,
      "loss": 0.5538,
      "step": 514
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.581916846896318e-05,
      "loss": 0.5624,
      "step": 515
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.546711398834543e-05,
      "loss": 0.5711,
      "step": 516
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.5116424840957065e-05,
      "loss": 0.5629,
      "step": 517
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.476710844638795e-05,
      "loss": 0.5501,
      "step": 518
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.441917219518438e-05,
      "loss": 0.6033,
      "step": 519
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.407262344869272e-05,
      "loss": 0.5593,
      "step": 520
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3727469538903646e-05,
      "loss": 0.5355,
      "step": 521
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.338371776829705e-05,
      "loss": 0.5959,
      "step": 522
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.3041375409687526e-05,
      "loss": 0.5828,
      "step": 523
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.2700449706070534e-05,
      "loss": 0.5922,
      "step": 524
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.236094787046901e-05,
      "loss": 0.6114,
      "step": 525
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.202287708578097e-05,
      "loss": 0.605,
      "step": 526
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.168624450462746e-05,
      "loss": 0.6079,
      "step": 527
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.13510572492012e-05,
      "loss": 0.5634,
      "step": 528
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.10173224111158e-05,
      "loss": 0.4974,
      "step": 529
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0685047051255946e-05,
      "loss": 0.551,
      "step": 530
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.035423819962785e-05,
      "loss": 0.6034,
      "step": 531
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.002490285521059e-05,
      "loss": 0.5759,
      "step": 532
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9697047985807958e-05,
      "loss": 0.5502,
      "step": 533
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9370680527901116e-05,
      "loss": 0.5722,
      "step": 534
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.904580738650181e-05,
      "loss": 0.559,
      "step": 535
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.872243543500629e-05,
      "loss": 0.5352,
      "step": 536
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.840057151504979e-05,
      "loss": 0.5857,
      "step": 537
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.8080222436361934e-05,
      "loss": 0.5819,
      "step": 538
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.7761394976622658e-05,
      "loss": 0.5741,
      "step": 539
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7444095881318656e-05,
      "loss": 0.5694,
      "step": 540
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.712833186360072e-05,
      "loss": 0.5797,
      "step": 541
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6814109604141848e-05,
      "loss": 0.5583,
      "step": 542
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6501435750995727e-05,
      "loss": 0.5771,
      "step": 543
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.619031691945618e-05,
      "loss": 0.5822,
      "step": 544
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.588075969191718e-05,
      "loss": 0.5831,
      "step": 545
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.5572770617733544e-05,
      "loss": 0.5801,
      "step": 546
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5266356213082433e-05,
      "loss": 0.5578,
      "step": 547
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.496152296082548e-05,
      "loss": 0.6124,
      "step": 548
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.465827731037147e-05,
      "loss": 0.6306,
      "step": 549
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4356625677540233e-05,
      "loss": 0.5585,
      "step": 550
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.405657444442657e-05,
      "loss": 0.5284,
      "step": 551
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.3758129959265407e-05,
      "loss": 0.6024,
      "step": 552
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.3461298536297328e-05,
      "loss": 0.5512,
      "step": 553
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3166086455635218e-05,
      "loss": 0.5346,
      "step": 554
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.2872499963131155e-05,
      "loss": 0.5329,
      "step": 555
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.258054527024451e-05,
      "loss": 0.5496,
      "step": 556
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2290228553910242e-05,
      "loss": 0.5877,
      "step": 557
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2001555956408428e-05,
      "loss": 0.5671,
      "step": 558
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.1714533585234244e-05,
      "loss": 0.5922,
      "step": 559
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.142916751296876e-05,
      "loss": 0.6145,
      "step": 560
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.114546377715042e-05,
      "loss": 0.5264,
      "step": 561
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0863428380147344e-05,
      "loss": 0.5404,
      "step": 562
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0583067289030335e-05,
      "loss": 0.6048,
      "step": 563
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.030438643544663e-05,
      "loss": 0.5456,
      "step": 564
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0027391715494347e-05,
      "loss": 0.5801,
      "step": 565
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9752088989597795e-05,
      "loss": 0.5813,
      "step": 566
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9478484082383562e-05,
      "loss": 0.5641,
      "step": 567
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9206582782557136e-05,
      "loss": 0.5711,
      "step": 568
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.893639084278046e-05,
      "loss": 0.5707,
      "step": 569
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.86679139795503e-05,
      "loss": 0.5731,
      "step": 570
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8401157873077257e-05,
      "loss": 0.5849,
      "step": 571
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8136128167165578e-05,
      "loss": 0.581,
      "step": 572
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.787283046909376e-05,
      "loss": 0.613,
      "step": 573
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.7611270349495924e-05,
      "loss": 0.59,
      "step": 574
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.735145334224394e-05,
      "loss": 0.5961,
      "step": 575
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7093384944330393e-05,
      "loss": 0.5323,
      "step": 576
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.6837070615752115e-05,
      "loss": 0.5652,
      "step": 577
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.6582515779394968e-05,
      "loss": 0.5533,
      "step": 578
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.632972582091884e-05,
      "loss": 0.5579,
      "step": 579
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6078706088643836e-05,
      "loss": 0.5813,
      "step": 580
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.5829461893437015e-05,
      "loss": 0.5158,
      "step": 581
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.558199850860016e-05,
      "loss": 0.5997,
      "step": 582
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.533632116975814e-05,
      "loss": 0.5398,
      "step": 583
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5092435074748146e-05,
      "loss": 0.599,
      "step": 584
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.48503453835097e-05,
      "loss": 0.6448,
      "step": 585
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.4610057217975526e-05,
      "loss": 0.589,
      "step": 586
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4371575661963143e-05,
      "loss": 0.5743,
      "step": 587
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4134905761067329e-05,
      "loss": 0.5822,
      "step": 588
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3900052522553397e-05,
      "loss": 0.5397,
      "step": 589
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3667020915251173e-05,
      "loss": 0.5798,
      "step": 590
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3435815869449964e-05,
      "loss": 0.5432,
      "step": 591
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3206442276794207e-05,
      "loss": 0.6011,
      "step": 592
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.297890499017992e-05,
      "loss": 0.5629,
      "step": 593
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2753208823652141e-05,
      "loss": 0.5176,
      "step": 594
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2529358552302972e-05,
      "loss": 0.5281,
      "step": 595
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2307358912170686e-05,
      "loss": 0.5945,
      "step": 596
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2087214600139308e-05,
      "loss": 0.551,
      "step": 597
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.1868930273839473e-05,
      "loss": 0.5338,
      "step": 598
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1652510551549723e-05,
      "loss": 0.5745,
      "step": 599
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1437960012098892e-05,
      "loss": 0.5542,
      "step": 600
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1225283194769176e-05,
      "loss": 0.5808,
      "step": 601
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1014484599200125e-05,
      "loss": 0.5374,
      "step": 602
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0805568685293422e-05,
      "loss": 0.5708,
      "step": 603
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0598539873118552e-05,
      "loss": 0.5936,
      "step": 604
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0393402542819231e-05,
      "loss": 0.5621,
      "step": 605
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0190161034520795e-05,
      "loss": 0.5412,
      "step": 606
    },
    {
      "epoch": 2.58,
      "learning_rate": 9.988819648238379e-06,
      "loss": 0.5525,
      "step": 607
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.789382643785895e-06,
      "loss": 0.5622,
      "step": 608
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.591854240685882e-06,
      "loss": 0.5662,
      "step": 609
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.396238618080322e-06,
      "loss": 0.5728,
      "step": 610
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.202539914642182e-06,
      "loss": 0.5531,
      "step": 611
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.010762228487813e-06,
      "loss": 0.5631,
      "step": 612
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.820909617090289e-06,
      "loss": 0.5265,
      "step": 613
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.632986097193573e-06,
      "loss": 0.5755,
      "step": 614
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.446995644727473e-06,
      "loss": 0.5626,
      "step": 615
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.2629421947236e-06,
      "loss": 0.618,
      "step": 616
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.080829641232013e-06,
      "loss": 0.5434,
      "step": 617
    },
    {
      "epoch": 2.63,
      "learning_rate": 7.900661837238977e-06,
      "loss": 0.5201,
      "step": 618
    },
    {
      "epoch": 2.63,
      "learning_rate": 7.72244259458531e-06,
      "loss": 0.5886,
      "step": 619
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.546175683885814e-06,
      "loss": 0.6115,
      "step": 620
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.371864834449405e-06,
      "loss": 0.5531,
      "step": 621
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.199513734200369e-06,
      "loss": 0.5619,
      "step": 622
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.029126029600197e-06,
      "loss": 0.5434,
      "step": 623
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.860705325570494e-06,
      "loss": 0.5542,
      "step": 624
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.694255185416687e-06,
      "loss": 0.5529,
      "step": 625
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.529779130752678e-06,
      "loss": 0.539,
      "step": 626
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.3672806414262765e-06,
      "loss": 0.5577,
      "step": 627
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.206763155445627e-06,
      "loss": 0.5307,
      "step": 628
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.0482300689064466e-06,
      "loss": 0.5418,
      "step": 629
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.891684735920167e-06,
      "loss": 0.5613,
      "step": 630
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.737130468542972e-06,
      "loss": 0.5926,
      "step": 631
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.58457053670578e-06,
      "loss": 0.5289,
      "step": 632
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.434008168144944e-06,
      "loss": 0.5981,
      "step": 633
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.2854465483340725e-06,
      "loss": 0.5581,
      "step": 634
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.1388888204165875e-06,
      "loss": 0.5844,
      "step": 635
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.9943380851392604e-06,
      "loss": 0.5871,
      "step": 636
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.851797400786506e-06,
      "loss": 0.5835,
      "step": 637
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.7112697831158126e-06,
      "loss": 0.5522,
      "step": 638
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.572758205293848e-06,
      "loss": 0.6108,
      "step": 639
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.4362655978336e-06,
      "loss": 0.5668,
      "step": 640
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.3017948485323255e-06,
      "loss": 0.5755,
      "step": 641
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.169348802410522e-06,
      "loss": 0.5658,
      "step": 642
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.038930261651674e-06,
      "loss": 0.5555,
      "step": 643
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.910541985543014e-06,
      "loss": 0.5277,
      "step": 644
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.7841866904170798e-06,
      "loss": 0.5733,
      "step": 645
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.6598670495943123e-06,
      "loss": 0.6289,
      "step": 646
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.537585693326484e-06,
      "loss": 0.5355,
      "step": 647
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.4173452087410187e-06,
      "loss": 0.6279,
      "step": 648
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.2991481397862568e-06,
      "loss": 0.5598,
      "step": 649
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.1829969871776555e-06,
      "loss": 0.5882,
      "step": 650
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.0688942083448967e-06,
      "loss": 0.5646,
      "step": 651
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.9568422173798294e-06,
      "loss": 0.6285,
      "step": 652
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.846843384985476e-06,
      "loss": 0.5294,
      "step": 653
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.7389000384257955e-06,
      "loss": 0.5308,
      "step": 654
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.633014461476524e-06,
      "loss": 0.5364,
      "step": 655
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.5291888943767992e-06,
      "loss": 0.5462,
      "step": 656
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.427425533781746e-06,
      "loss": 0.6125,
      "step": 657
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.3277265327160904e-06,
      "loss": 0.5888,
      "step": 658
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.2300940005285374e-06,
      "loss": 0.5855,
      "step": 659
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.134530002847146e-06,
      "loss": 0.5628,
      "step": 660
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.0410365615356365e-06,
      "loss": 0.5872,
      "step": 661
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.9496156546506274e-06,
      "loss": 0.5642,
      "step": 662
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.8602692163997681e-06,
      "loss": 0.5872,
      "step": 663
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.7729991371008502e-06,
      "loss": 0.6044,
      "step": 664
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.6878072631417386e-06,
      "loss": 0.5491,
      "step": 665
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.6046953969413915e-06,
      "loss": 0.5291,
      "step": 666
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.5236652969116804e-06,
      "loss": 0.539,
      "step": 667
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.444718677420176e-06,
      "loss": 0.5518,
      "step": 668
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.367857208753931e-06,
      "loss": 0.605,
      "step": 669
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.2930825170840877e-06,
      "loss": 0.5938,
      "step": 670
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.2203961844315048e-06,
      "loss": 0.5656,
      "step": 671
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1497997486332513e-06,
      "loss": 0.5663,
      "step": 672
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0812947033101207e-06,
      "loss": 0.5731,
      "step": 673
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0148824978349792e-06,
      "loss": 0.5909,
      "step": 674
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.505645373021455e-07,
      "loss": 0.5732,
      "step": 675
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.883421824976479e-07,
      "loss": 0.5457,
      "step": 676
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.282167498703918e-07,
      "loss": 0.559,
      "step": 677
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.701895115043822e-07,
      "loss": 0.5706,
      "step": 678
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.142616950917446e-07,
      "loss": 0.5509,
      "step": 679
    },
    {
      "epoch": 2.89,
      "learning_rate": 6.604344839068021e-07,
      "loss": 0.524,
      "step": 680
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.087090167809839e-07,
      "loss": 0.5933,
      "step": 681
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.590863880788111e-07,
      "loss": 0.5565,
      "step": 682
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.115676476746489e-07,
      "loss": 0.5286,
      "step": 683
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.661538009305577e-07,
      "loss": 0.629,
      "step": 684
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.2284580867500976e-07,
      "loss": 0.5606,
      "step": 685
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.8164458718255025e-07,
      "loss": 0.5308,
      "step": 686
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.4255100815442365e-07,
      "loss": 0.5966,
      "step": 687
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.0556589870012196e-07,
      "loss": 0.5551,
      "step": 688
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.7069004131987653e-07,
      "loss": 0.5597,
      "step": 689
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.379241738881377e-07,
      "loss": 0.5856,
      "step": 690
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.0726898963793205e-07,
      "loss": 0.5698,
      "step": 691
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.78725137146174e-07,
      "loss": 0.5685,
      "step": 692
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.5229322032002115e-07,
      "loss": 0.5562,
      "step": 693
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.27973798384029e-07,
      "loss": 0.5893,
      "step": 694
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.0576738586831614e-07,
      "loss": 0.5793,
      "step": 695
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.567445259775042e-08,
      "loss": 0.5354,
      "step": 696
    },
    {
      "epoch": 2.97,
      "learning_rate": 6.769542368190162e-08,
      "loss": 0.5889,
      "step": 697
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.183067950617071e-08,
      "loss": 0.5501,
      "step": 698
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.808055572362967e-08,
      "loss": 0.5596,
      "step": 699
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.6445343247982755e-08,
      "loss": 0.5942,
      "step": 700
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.6925288247393588e-08,
      "loss": 0.5747,
      "step": 701
    },
    {
      "epoch": 2.99,
      "learning_rate": 9.52059213927825e-09,
      "loss": 0.5802,
      "step": 702
    },
    {
      "epoch": 2.99,
      "learning_rate": 4.231411586064216e-09,
      "loss": 0.5792,
      "step": 703
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.0578584918374823e-09,
      "loss": 0.5585,
      "step": 704
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 0.4152,
      "step": 705
    },
    {
      "epoch": 3.0,
      "step": 705,
      "total_flos": 59407669002240.0,
      "train_loss": 0.9080784771036594,
      "train_runtime": 4849.5227,
      "train_samples_per_second": 18.568,
      "train_steps_per_second": 0.145
    }
  ],
  "max_steps": 705,
  "num_train_epochs": 3,
  "total_flos": 59407669002240.0,
  "trial_name": null,
  "trial_params": null
}