范昊天
Initial commit
7ff52a6
raw
history blame contribute delete
No virus
53.2 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 438,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.4285714285714286e-06,
"loss": 0.1214,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 2.8571428571428573e-06,
"loss": 0.1163,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 4.2857142857142855e-06,
"loss": 0.1077,
"step": 3
},
{
"epoch": 0.02,
"learning_rate": 5.7142857142857145e-06,
"loss": 0.1602,
"step": 4
},
{
"epoch": 0.02,
"learning_rate": 7.1428571428571436e-06,
"loss": 0.1192,
"step": 5
},
{
"epoch": 0.03,
"learning_rate": 8.571428571428571e-06,
"loss": 0.061,
"step": 6
},
{
"epoch": 0.03,
"learning_rate": 1e-05,
"loss": 0.1013,
"step": 7
},
{
"epoch": 0.04,
"learning_rate": 1.1428571428571429e-05,
"loss": 0.1474,
"step": 8
},
{
"epoch": 0.04,
"learning_rate": 1.2857142857142859e-05,
"loss": 0.1459,
"step": 9
},
{
"epoch": 0.05,
"learning_rate": 1.4285714285714287e-05,
"loss": 0.1334,
"step": 10
},
{
"epoch": 0.05,
"learning_rate": 1.5714285714285715e-05,
"loss": 0.0921,
"step": 11
},
{
"epoch": 0.05,
"learning_rate": 1.7142857142857142e-05,
"loss": 0.0993,
"step": 12
},
{
"epoch": 0.06,
"learning_rate": 1.8571428571428575e-05,
"loss": 0.085,
"step": 13
},
{
"epoch": 0.06,
"learning_rate": 2e-05,
"loss": 0.0922,
"step": 14
},
{
"epoch": 0.07,
"learning_rate": 1.9999725503981394e-05,
"loss": 0.0963,
"step": 15
},
{
"epoch": 0.07,
"learning_rate": 1.999890203099519e-05,
"loss": 0.1059,
"step": 16
},
{
"epoch": 0.08,
"learning_rate": 1.99975296262494e-05,
"loss": 0.0809,
"step": 17
},
{
"epoch": 0.08,
"learning_rate": 1.9995608365087945e-05,
"loss": 0.0963,
"step": 18
},
{
"epoch": 0.09,
"learning_rate": 1.9993138352986538e-05,
"loss": 0.0908,
"step": 19
},
{
"epoch": 0.09,
"learning_rate": 1.999011972554688e-05,
"loss": 0.0766,
"step": 20
},
{
"epoch": 0.1,
"learning_rate": 1.9986552648489212e-05,
"loss": 0.0667,
"step": 21
},
{
"epoch": 0.1,
"learning_rate": 1.9982437317643218e-05,
"loss": 0.0804,
"step": 22
},
{
"epoch": 0.11,
"learning_rate": 1.997777395893729e-05,
"loss": 0.0719,
"step": 23
},
{
"epoch": 0.11,
"learning_rate": 1.99725628283861e-05,
"loss": 0.0781,
"step": 24
},
{
"epoch": 0.11,
"learning_rate": 1.9966804212076583e-05,
"loss": 0.0693,
"step": 25
},
{
"epoch": 0.12,
"learning_rate": 1.996049842615217e-05,
"loss": 0.0576,
"step": 26
},
{
"epoch": 0.12,
"learning_rate": 1.99536458167955e-05,
"loss": 0.0791,
"step": 27
},
{
"epoch": 0.13,
"learning_rate": 1.994624676020936e-05,
"loss": 0.0627,
"step": 28
},
{
"epoch": 0.13,
"learning_rate": 1.9938301662596075e-05,
"loss": 0.0806,
"step": 29
},
{
"epoch": 0.14,
"learning_rate": 1.992981096013517e-05,
"loss": 0.0479,
"step": 30
},
{
"epoch": 0.14,
"learning_rate": 1.9920775118959454e-05,
"loss": 0.0611,
"step": 31
},
{
"epoch": 0.15,
"learning_rate": 1.991119463512941e-05,
"loss": 0.07,
"step": 32
},
{
"epoch": 0.15,
"learning_rate": 1.9901070034605968e-05,
"loss": 0.0728,
"step": 33
},
{
"epoch": 0.16,
"learning_rate": 1.9890401873221642e-05,
"loss": 0.0687,
"step": 34
},
{
"epoch": 0.16,
"learning_rate": 1.9879190736649992e-05,
"loss": 0.0762,
"step": 35
},
{
"epoch": 0.16,
"learning_rate": 1.986743724037349e-05,
"loss": 0.0688,
"step": 36
},
{
"epoch": 0.17,
"learning_rate": 1.9855142029649725e-05,
"loss": 0.0878,
"step": 37
},
{
"epoch": 0.17,
"learning_rate": 1.984230577947597e-05,
"loss": 0.0652,
"step": 38
},
{
"epoch": 0.18,
"learning_rate": 1.9828929194552143e-05,
"loss": 0.0712,
"step": 39
},
{
"epoch": 0.18,
"learning_rate": 1.9815013009242103e-05,
"loss": 0.044,
"step": 40
},
{
"epoch": 0.19,
"learning_rate": 1.980055798753334e-05,
"loss": 0.058,
"step": 41
},
{
"epoch": 0.19,
"learning_rate": 1.9785564922995042e-05,
"loss": 0.0882,
"step": 42
},
{
"epoch": 0.2,
"learning_rate": 1.9770034638734506e-05,
"loss": 0.0663,
"step": 43
},
{
"epoch": 0.2,
"learning_rate": 1.975396798735198e-05,
"loss": 0.0649,
"step": 44
},
{
"epoch": 0.21,
"learning_rate": 1.973736585089382e-05,
"loss": 0.065,
"step": 45
},
{
"epoch": 0.21,
"learning_rate": 1.972022914080411e-05,
"loss": 0.0672,
"step": 46
},
{
"epoch": 0.21,
"learning_rate": 1.970255879787458e-05,
"loss": 0.0572,
"step": 47
},
{
"epoch": 0.22,
"learning_rate": 1.968435579219299e-05,
"loss": 0.063,
"step": 48
},
{
"epoch": 0.22,
"learning_rate": 1.966562112308985e-05,
"loss": 0.0687,
"step": 49
},
{
"epoch": 0.23,
"learning_rate": 1.964635581908359e-05,
"loss": 0.0669,
"step": 50
},
{
"epoch": 0.23,
"learning_rate": 1.9626560937824047e-05,
"loss": 0.0639,
"step": 51
},
{
"epoch": 0.24,
"learning_rate": 1.9606237566034443e-05,
"loss": 0.066,
"step": 52
},
{
"epoch": 0.24,
"learning_rate": 1.9585386819451707e-05,
"loss": 0.0606,
"step": 53
},
{
"epoch": 0.25,
"learning_rate": 1.9564009842765225e-05,
"loss": 0.0653,
"step": 54
},
{
"epoch": 0.25,
"learning_rate": 1.9542107809553995e-05,
"loss": 0.0504,
"step": 55
},
{
"epoch": 0.26,
"learning_rate": 1.9519681922222195e-05,
"loss": 0.0849,
"step": 56
},
{
"epoch": 0.26,
"learning_rate": 1.949673341193319e-05,
"loss": 0.0651,
"step": 57
},
{
"epoch": 0.26,
"learning_rate": 1.9473263538541916e-05,
"loss": 0.071,
"step": 58
},
{
"epoch": 0.27,
"learning_rate": 1.9449273590525736e-05,
"loss": 0.0608,
"step": 59
},
{
"epoch": 0.27,
"learning_rate": 1.942476488491369e-05,
"loss": 0.0665,
"step": 60
},
{
"epoch": 0.28,
"learning_rate": 1.93997387672142e-05,
"loss": 0.0623,
"step": 61
},
{
"epoch": 0.28,
"learning_rate": 1.9374196611341212e-05,
"loss": 0.0574,
"step": 62
},
{
"epoch": 0.29,
"learning_rate": 1.934813981953873e-05,
"loss": 0.0652,
"step": 63
},
{
"epoch": 0.29,
"learning_rate": 1.932156982230388e-05,
"loss": 0.0627,
"step": 64
},
{
"epoch": 0.3,
"learning_rate": 1.9294488078308355e-05,
"loss": 0.066,
"step": 65
},
{
"epoch": 0.3,
"learning_rate": 1.9266896074318335e-05,
"loss": 0.0588,
"step": 66
},
{
"epoch": 0.31,
"learning_rate": 1.9238795325112867e-05,
"loss": 0.0664,
"step": 67
},
{
"epoch": 0.31,
"learning_rate": 1.921018737340071e-05,
"loss": 0.0471,
"step": 68
},
{
"epoch": 0.32,
"learning_rate": 1.918107378973563e-05,
"loss": 0.0509,
"step": 69
},
{
"epoch": 0.32,
"learning_rate": 1.9151456172430186e-05,
"loss": 0.0732,
"step": 70
},
{
"epoch": 0.32,
"learning_rate": 1.9121336147467987e-05,
"loss": 0.0527,
"step": 71
},
{
"epoch": 0.33,
"learning_rate": 1.909071536841442e-05,
"loss": 0.0673,
"step": 72
},
{
"epoch": 0.33,
"learning_rate": 1.905959551632587e-05,
"loss": 0.0535,
"step": 73
},
{
"epoch": 0.34,
"learning_rate": 1.9027978299657436e-05,
"loss": 0.0478,
"step": 74
},
{
"epoch": 0.34,
"learning_rate": 1.8995865454169142e-05,
"loss": 0.0688,
"step": 75
},
{
"epoch": 0.35,
"learning_rate": 1.896325874283063e-05,
"loss": 0.0644,
"step": 76
},
{
"epoch": 0.35,
"learning_rate": 1.8930159955724386e-05,
"loss": 0.0671,
"step": 77
},
{
"epoch": 0.36,
"learning_rate": 1.8896570909947477e-05,
"loss": 0.0611,
"step": 78
},
{
"epoch": 0.36,
"learning_rate": 1.8862493449511757e-05,
"loss": 0.0616,
"step": 79
},
{
"epoch": 0.37,
"learning_rate": 1.8827929445242678e-05,
"loss": 0.0892,
"step": 80
},
{
"epoch": 0.37,
"learning_rate": 1.8792880794676546e-05,
"loss": 0.0615,
"step": 81
},
{
"epoch": 0.37,
"learning_rate": 1.875734942195637e-05,
"loss": 0.0632,
"step": 82
},
{
"epoch": 0.38,
"learning_rate": 1.872133727772622e-05,
"loss": 0.0588,
"step": 83
},
{
"epoch": 0.38,
"learning_rate": 1.8684846339024145e-05,
"loss": 0.0524,
"step": 84
},
{
"epoch": 0.39,
"learning_rate": 1.8647878609173612e-05,
"loss": 0.058,
"step": 85
},
{
"epoch": 0.39,
"learning_rate": 1.8610436117673557e-05,
"loss": 0.0738,
"step": 86
},
{
"epoch": 0.4,
"learning_rate": 1.857252092008695e-05,
"loss": 0.0566,
"step": 87
},
{
"epoch": 0.4,
"learning_rate": 1.853413509792795e-05,
"loss": 0.0728,
"step": 88
},
{
"epoch": 0.41,
"learning_rate": 1.849528075854762e-05,
"loss": 0.0569,
"step": 89
},
{
"epoch": 0.41,
"learning_rate": 1.845596003501826e-05,
"loss": 0.0746,
"step": 90
},
{
"epoch": 0.42,
"learning_rate": 1.8416175086016284e-05,
"loss": 0.0588,
"step": 91
},
{
"epoch": 0.42,
"learning_rate": 1.8375928095703703e-05,
"loss": 0.0632,
"step": 92
},
{
"epoch": 0.42,
"learning_rate": 1.833522127360824e-05,
"loss": 0.0509,
"step": 93
},
{
"epoch": 0.43,
"learning_rate": 1.829405685450202e-05,
"loss": 0.0506,
"step": 94
},
{
"epoch": 0.43,
"learning_rate": 1.8252437098278867e-05,
"loss": 0.0562,
"step": 95
},
{
"epoch": 0.44,
"learning_rate": 1.821036428983026e-05,
"loss": 0.056,
"step": 96
},
{
"epoch": 0.44,
"learning_rate": 1.8167840738919883e-05,
"loss": 0.0513,
"step": 97
},
{
"epoch": 0.45,
"learning_rate": 1.8124868780056814e-05,
"loss": 0.0697,
"step": 98
},
{
"epoch": 0.45,
"learning_rate": 1.8081450772367382e-05,
"loss": 0.033,
"step": 99
},
{
"epoch": 0.46,
"learning_rate": 1.8037589099465637e-05,
"loss": 0.0685,
"step": 100
},
{
"epoch": 0.46,
"learning_rate": 1.799328616932249e-05,
"loss": 0.0391,
"step": 101
},
{
"epoch": 0.47,
"learning_rate": 1.7948544414133534e-05,
"loss": 0.0387,
"step": 102
},
{
"epoch": 0.47,
"learning_rate": 1.7903366290185498e-05,
"loss": 0.0742,
"step": 103
},
{
"epoch": 0.47,
"learning_rate": 1.7857754277721417e-05,
"loss": 0.062,
"step": 104
},
{
"epoch": 0.48,
"learning_rate": 1.781171088080445e-05,
"loss": 0.0751,
"step": 105
},
{
"epoch": 0.48,
"learning_rate": 1.7765238627180424e-05,
"loss": 0.0893,
"step": 106
},
{
"epoch": 0.49,
"learning_rate": 1.7718340068139066e-05,
"loss": 0.0585,
"step": 107
},
{
"epoch": 0.49,
"learning_rate": 1.7671017778373914e-05,
"loss": 0.0788,
"step": 108
},
{
"epoch": 0.5,
"learning_rate": 1.7623274355841e-05,
"loss": 0.0539,
"step": 109
},
{
"epoch": 0.5,
"learning_rate": 1.7575112421616203e-05,
"loss": 0.0675,
"step": 110
},
{
"epoch": 0.51,
"learning_rate": 1.752653461975136e-05,
"loss": 0.0656,
"step": 111
},
{
"epoch": 0.51,
"learning_rate": 1.747754361712911e-05,
"loss": 0.0584,
"step": 112
},
{
"epoch": 0.52,
"learning_rate": 1.7428142103316493e-05,
"loss": 0.0672,
"step": 113
},
{
"epoch": 0.52,
"learning_rate": 1.7378332790417275e-05,
"loss": 0.0648,
"step": 114
},
{
"epoch": 0.53,
"learning_rate": 1.7328118412923073e-05,
"loss": 0.0565,
"step": 115
},
{
"epoch": 0.53,
"learning_rate": 1.7277501727563227e-05,
"loss": 0.0679,
"step": 116
},
{
"epoch": 0.53,
"learning_rate": 1.7226485513153458e-05,
"loss": 0.0653,
"step": 117
},
{
"epoch": 0.54,
"learning_rate": 1.717507257044331e-05,
"loss": 0.068,
"step": 118
},
{
"epoch": 0.54,
"learning_rate": 1.712326572196241e-05,
"loss": 0.0697,
"step": 119
},
{
"epoch": 0.55,
"learning_rate": 1.7071067811865477e-05,
"loss": 0.0467,
"step": 120
},
{
"epoch": 0.55,
"learning_rate": 1.701848170577621e-05,
"loss": 0.0602,
"step": 121
},
{
"epoch": 0.56,
"learning_rate": 1.6965510290629973e-05,
"loss": 0.0532,
"step": 122
},
{
"epoch": 0.56,
"learning_rate": 1.6912156474515264e-05,
"loss": 0.0567,
"step": 123
},
{
"epoch": 0.57,
"learning_rate": 1.6858423186514108e-05,
"loss": 0.0544,
"step": 124
},
{
"epoch": 0.57,
"learning_rate": 1.6804313376541226e-05,
"loss": 0.0449,
"step": 125
},
{
"epoch": 0.58,
"learning_rate": 1.6749830015182106e-05,
"loss": 0.0536,
"step": 126
},
{
"epoch": 0.58,
"learning_rate": 1.6694976093529896e-05,
"loss": 0.0799,
"step": 127
},
{
"epoch": 0.58,
"learning_rate": 1.6639754623021227e-05,
"loss": 0.0612,
"step": 128
},
{
"epoch": 0.59,
"learning_rate": 1.658416863527084e-05,
"loss": 0.0532,
"step": 129
},
{
"epoch": 0.59,
"learning_rate": 1.6528221181905217e-05,
"loss": 0.0502,
"step": 130
},
{
"epoch": 0.6,
"learning_rate": 1.647191533439499e-05,
"loss": 0.0526,
"step": 131
},
{
"epoch": 0.6,
"learning_rate": 1.6415254183886356e-05,
"loss": 0.0453,
"step": 132
},
{
"epoch": 0.61,
"learning_rate": 1.635824084103135e-05,
"loss": 0.0566,
"step": 133
},
{
"epoch": 0.61,
"learning_rate": 1.6300878435817115e-05,
"loss": 0.06,
"step": 134
},
{
"epoch": 0.62,
"learning_rate": 1.6243170117394e-05,
"loss": 0.0769,
"step": 135
},
{
"epoch": 0.62,
"learning_rate": 1.6185119053902748e-05,
"loss": 0.0412,
"step": 136
},
{
"epoch": 0.63,
"learning_rate": 1.6126728432300516e-05,
"loss": 0.0488,
"step": 137
},
{
"epoch": 0.63,
"learning_rate": 1.6068001458185934e-05,
"loss": 0.0629,
"step": 138
},
{
"epoch": 0.63,
"learning_rate": 1.600894135562312e-05,
"loss": 0.0652,
"step": 139
},
{
"epoch": 0.64,
"learning_rate": 1.5949551366964675e-05,
"loss": 0.0533,
"step": 140
},
{
"epoch": 0.64,
"learning_rate": 1.588983475267368e-05,
"loss": 0.0566,
"step": 141
},
{
"epoch": 0.65,
"learning_rate": 1.5829794791144723e-05,
"loss": 0.0787,
"step": 142
},
{
"epoch": 0.65,
"learning_rate": 1.576943477852387e-05,
"loss": 0.0614,
"step": 143
},
{
"epoch": 0.66,
"learning_rate": 1.5708758028527754e-05,
"loss": 0.0578,
"step": 144
},
{
"epoch": 0.66,
"learning_rate": 1.5647767872261633e-05,
"loss": 0.0504,
"step": 145
},
{
"epoch": 0.67,
"learning_rate": 1.5586467658036526e-05,
"loss": 0.0694,
"step": 146
},
{
"epoch": 0.67,
"learning_rate": 1.5524860751185377e-05,
"loss": 0.0673,
"step": 147
},
{
"epoch": 0.68,
"learning_rate": 1.5462950533878318e-05,
"loss": 0.0514,
"step": 148
},
{
"epoch": 0.68,
"learning_rate": 1.540074040493698e-05,
"loss": 0.0645,
"step": 149
},
{
"epoch": 0.68,
"learning_rate": 1.533823377964791e-05,
"loss": 0.0475,
"step": 150
},
{
"epoch": 0.69,
"learning_rate": 1.5275434089575055e-05,
"loss": 0.0499,
"step": 151
},
{
"epoch": 0.69,
"learning_rate": 1.52123447823714e-05,
"loss": 0.048,
"step": 152
},
{
"epoch": 0.7,
"learning_rate": 1.5148969321589672e-05,
"loss": 0.0323,
"step": 153
},
{
"epoch": 0.7,
"learning_rate": 1.5085311186492206e-05,
"loss": 0.0502,
"step": 154
},
{
"epoch": 0.71,
"learning_rate": 1.5021373871859926e-05,
"loss": 0.0705,
"step": 155
},
{
"epoch": 0.71,
"learning_rate": 1.4957160887800494e-05,
"loss": 0.0406,
"step": 156
},
{
"epoch": 0.72,
"learning_rate": 1.4892675759555607e-05,
"loss": 0.052,
"step": 157
},
{
"epoch": 0.72,
"learning_rate": 1.482792202730745e-05,
"loss": 0.0488,
"step": 158
},
{
"epoch": 0.73,
"learning_rate": 1.4762903245984369e-05,
"loss": 0.0607,
"step": 159
},
{
"epoch": 0.73,
"learning_rate": 1.469762298506568e-05,
"loss": 0.0578,
"step": 160
},
{
"epoch": 0.74,
"learning_rate": 1.4632084828385732e-05,
"loss": 0.0479,
"step": 161
},
{
"epoch": 0.74,
"learning_rate": 1.4566292373937133e-05,
"loss": 0.0426,
"step": 162
},
{
"epoch": 0.74,
"learning_rate": 1.4500249233673247e-05,
"loss": 0.0465,
"step": 163
},
{
"epoch": 0.75,
"learning_rate": 1.4433959033309888e-05,
"loss": 0.0587,
"step": 164
},
{
"epoch": 0.75,
"learning_rate": 1.4367425412126266e-05,
"loss": 0.0422,
"step": 165
},
{
"epoch": 0.76,
"learning_rate": 1.4300652022765207e-05,
"loss": 0.0657,
"step": 166
},
{
"epoch": 0.76,
"learning_rate": 1.4233642531032614e-05,
"loss": 0.0734,
"step": 167
},
{
"epoch": 0.77,
"learning_rate": 1.4166400615696232e-05,
"loss": 0.0463,
"step": 168
},
{
"epoch": 0.77,
"learning_rate": 1.4098929968283661e-05,
"loss": 0.0411,
"step": 169
},
{
"epoch": 0.78,
"learning_rate": 1.4031234292879726e-05,
"loss": 0.0439,
"step": 170
},
{
"epoch": 0.78,
"learning_rate": 1.3963317305923095e-05,
"loss": 0.0497,
"step": 171
},
{
"epoch": 0.79,
"learning_rate": 1.3895182736002276e-05,
"loss": 0.0678,
"step": 172
},
{
"epoch": 0.79,
"learning_rate": 1.3826834323650899e-05,
"loss": 0.0539,
"step": 173
},
{
"epoch": 0.79,
"learning_rate": 1.3758275821142382e-05,
"loss": 0.0533,
"step": 174
},
{
"epoch": 0.8,
"learning_rate": 1.3689510992283923e-05,
"loss": 0.0463,
"step": 175
},
{
"epoch": 0.8,
"learning_rate": 1.3620543612209861e-05,
"loss": 0.0466,
"step": 176
},
{
"epoch": 0.81,
"learning_rate": 1.3551377467174455e-05,
"loss": 0.0466,
"step": 177
},
{
"epoch": 0.81,
"learning_rate": 1.348201635434399e-05,
"loss": 0.0609,
"step": 178
},
{
"epoch": 0.82,
"learning_rate": 1.3412464081588323e-05,
"loss": 0.0426,
"step": 179
},
{
"epoch": 0.82,
"learning_rate": 1.3342724467271852e-05,
"loss": 0.0638,
"step": 180
},
{
"epoch": 0.83,
"learning_rate": 1.3272801340043867e-05,
"loss": 0.0531,
"step": 181
},
{
"epoch": 0.83,
"learning_rate": 1.3202698538628376e-05,
"loss": 0.0552,
"step": 182
},
{
"epoch": 0.84,
"learning_rate": 1.3132419911613358e-05,
"loss": 0.0803,
"step": 183
},
{
"epoch": 0.84,
"learning_rate": 1.306196931723947e-05,
"loss": 0.0496,
"step": 184
},
{
"epoch": 0.84,
"learning_rate": 1.2991350623188245e-05,
"loss": 0.0506,
"step": 185
},
{
"epoch": 0.85,
"learning_rate": 1.292056770636976e-05,
"loss": 0.0549,
"step": 186
},
{
"epoch": 0.85,
"learning_rate": 1.284962445270978e-05,
"loss": 0.0632,
"step": 187
},
{
"epoch": 0.86,
"learning_rate": 1.2778524756936438e-05,
"loss": 0.0399,
"step": 188
},
{
"epoch": 0.86,
"learning_rate": 1.2707272522366428e-05,
"loss": 0.0522,
"step": 189
},
{
"epoch": 0.87,
"learning_rate": 1.2635871660690677e-05,
"loss": 0.0569,
"step": 190
},
{
"epoch": 0.87,
"learning_rate": 1.2564326091759647e-05,
"loss": 0.0725,
"step": 191
},
{
"epoch": 0.88,
"learning_rate": 1.2492639743368096e-05,
"loss": 0.0521,
"step": 192
},
{
"epoch": 0.88,
"learning_rate": 1.242081655103947e-05,
"loss": 0.058,
"step": 193
},
{
"epoch": 0.89,
"learning_rate": 1.234886045780984e-05,
"loss": 0.0701,
"step": 194
},
{
"epoch": 0.89,
"learning_rate": 1.2276775414011421e-05,
"loss": 0.0691,
"step": 195
},
{
"epoch": 0.89,
"learning_rate": 1.220456537705572e-05,
"loss": 0.0509,
"step": 196
},
{
"epoch": 0.9,
"learning_rate": 1.2132234311216271e-05,
"loss": 0.0619,
"step": 197
},
{
"epoch": 0.9,
"learning_rate": 1.2059786187410984e-05,
"loss": 0.0537,
"step": 198
},
{
"epoch": 0.91,
"learning_rate": 1.1987224982984176e-05,
"loss": 0.0443,
"step": 199
},
{
"epoch": 0.91,
"learning_rate": 1.1914554681488188e-05,
"loss": 0.0605,
"step": 200
},
{
"epoch": 0.92,
"learning_rate": 1.1841779272464703e-05,
"loss": 0.0583,
"step": 201
},
{
"epoch": 0.92,
"learning_rate": 1.176890275122573e-05,
"loss": 0.0387,
"step": 202
},
{
"epoch": 0.93,
"learning_rate": 1.1695929118634257e-05,
"loss": 0.0745,
"step": 203
},
{
"epoch": 0.93,
"learning_rate": 1.16228623808846e-05,
"loss": 0.055,
"step": 204
},
{
"epoch": 0.94,
"learning_rate": 1.1549706549282486e-05,
"loss": 0.0484,
"step": 205
},
{
"epoch": 0.94,
"learning_rate": 1.1476465640024814e-05,
"loss": 0.0412,
"step": 206
},
{
"epoch": 0.95,
"learning_rate": 1.1403143673979182e-05,
"loss": 0.0678,
"step": 207
},
{
"epoch": 0.95,
"learning_rate": 1.1329744676463144e-05,
"loss": 0.0558,
"step": 208
},
{
"epoch": 0.95,
"learning_rate": 1.1256272677023214e-05,
"loss": 0.0401,
"step": 209
},
{
"epoch": 0.96,
"learning_rate": 1.1182731709213658e-05,
"loss": 0.0658,
"step": 210
},
{
"epoch": 0.96,
"learning_rate": 1.1109125810375054e-05,
"loss": 0.0353,
"step": 211
},
{
"epoch": 0.97,
"learning_rate": 1.1035459021412631e-05,
"loss": 0.0496,
"step": 212
},
{
"epoch": 0.97,
"learning_rate": 1.0961735386574447e-05,
"loss": 0.068,
"step": 213
},
{
"epoch": 0.98,
"learning_rate": 1.0887958953229349e-05,
"loss": 0.049,
"step": 214
},
{
"epoch": 0.98,
"learning_rate": 1.0814133771644783e-05,
"loss": 0.0496,
"step": 215
},
{
"epoch": 0.99,
"learning_rate": 1.074026389476443e-05,
"loss": 0.0527,
"step": 216
},
{
"epoch": 0.99,
"learning_rate": 1.0666353377985712e-05,
"loss": 0.0551,
"step": 217
},
{
"epoch": 1.0,
"learning_rate": 1.0592406278937143e-05,
"loss": 0.054,
"step": 218
},
{
"epoch": 1.0,
"learning_rate": 1.0518426657255585e-05,
"loss": 0.0499,
"step": 219
},
{
"epoch": 1.0,
"learning_rate": 1.0444418574363354e-05,
"loss": 0.0361,
"step": 220
},
{
"epoch": 1.01,
"learning_rate": 1.0370386093245271e-05,
"loss": 0.0428,
"step": 221
},
{
"epoch": 1.01,
"learning_rate": 1.0296333278225599e-05,
"loss": 0.0379,
"step": 222
},
{
"epoch": 1.02,
"learning_rate": 1.0222264194744918e-05,
"loss": 0.0334,
"step": 223
},
{
"epoch": 1.02,
"learning_rate": 1.0148182909136928e-05,
"loss": 0.0351,
"step": 224
},
{
"epoch": 1.03,
"learning_rate": 1.0074093488405223e-05,
"loss": 0.0353,
"step": 225
},
{
"epoch": 1.03,
"learning_rate": 1e-05,
"loss": 0.0304,
"step": 226
},
{
"epoch": 1.04,
"learning_rate": 9.92590651159478e-06,
"loss": 0.0247,
"step": 227
},
{
"epoch": 1.04,
"learning_rate": 9.851817090863073e-06,
"loss": 0.0341,
"step": 228
},
{
"epoch": 1.05,
"learning_rate": 9.777735805255087e-06,
"loss": 0.0319,
"step": 229
},
{
"epoch": 1.05,
"learning_rate": 9.703666721774403e-06,
"loss": 0.0254,
"step": 230
},
{
"epoch": 1.05,
"learning_rate": 9.629613906754732e-06,
"loss": 0.0253,
"step": 231
},
{
"epoch": 1.06,
"learning_rate": 9.555581425636648e-06,
"loss": 0.0299,
"step": 232
},
{
"epoch": 1.06,
"learning_rate": 9.481573342744418e-06,
"loss": 0.0312,
"step": 233
},
{
"epoch": 1.07,
"learning_rate": 9.407593721062858e-06,
"loss": 0.0244,
"step": 234
},
{
"epoch": 1.07,
"learning_rate": 9.33364662201429e-06,
"loss": 0.0185,
"step": 235
},
{
"epoch": 1.08,
"learning_rate": 9.259736105235572e-06,
"loss": 0.0324,
"step": 236
},
{
"epoch": 1.08,
"learning_rate": 9.18586622835522e-06,
"loss": 0.0195,
"step": 237
},
{
"epoch": 1.09,
"learning_rate": 9.112041046770653e-06,
"loss": 0.0257,
"step": 238
},
{
"epoch": 1.09,
"learning_rate": 9.038264613425556e-06,
"loss": 0.0244,
"step": 239
},
{
"epoch": 1.1,
"learning_rate": 8.96454097858737e-06,
"loss": 0.0256,
"step": 240
},
{
"epoch": 1.1,
"learning_rate": 8.890874189624951e-06,
"loss": 0.0266,
"step": 241
},
{
"epoch": 1.11,
"learning_rate": 8.817268290786343e-06,
"loss": 0.0357,
"step": 242
},
{
"epoch": 1.11,
"learning_rate": 8.743727322976788e-06,
"loss": 0.037,
"step": 243
},
{
"epoch": 1.11,
"learning_rate": 8.670255323536858e-06,
"loss": 0.0588,
"step": 244
},
{
"epoch": 1.12,
"learning_rate": 8.596856326020822e-06,
"loss": 0.0282,
"step": 245
},
{
"epoch": 1.12,
"learning_rate": 8.52353435997519e-06,
"loss": 0.0649,
"step": 246
},
{
"epoch": 1.13,
"learning_rate": 8.450293450717517e-06,
"loss": 0.0501,
"step": 247
},
{
"epoch": 1.13,
"learning_rate": 8.377137619115401e-06,
"loss": 0.0224,
"step": 248
},
{
"epoch": 1.14,
"learning_rate": 8.304070881365748e-06,
"loss": 0.0262,
"step": 249
},
{
"epoch": 1.14,
"learning_rate": 8.231097248774273e-06,
"loss": 0.0475,
"step": 250
},
{
"epoch": 1.15,
"learning_rate": 8.158220727535298e-06,
"loss": 0.0304,
"step": 251
},
{
"epoch": 1.15,
"learning_rate": 8.085445318511814e-06,
"loss": 0.0272,
"step": 252
},
{
"epoch": 1.16,
"learning_rate": 8.012775017015829e-06,
"loss": 0.0404,
"step": 253
},
{
"epoch": 1.16,
"learning_rate": 7.940213812589018e-06,
"loss": 0.015,
"step": 254
},
{
"epoch": 1.16,
"learning_rate": 7.867765688783732e-06,
"loss": 0.0322,
"step": 255
},
{
"epoch": 1.17,
"learning_rate": 7.795434622944282e-06,
"loss": 0.0326,
"step": 256
},
{
"epoch": 1.17,
"learning_rate": 7.723224585988585e-06,
"loss": 0.0276,
"step": 257
},
{
"epoch": 1.18,
"learning_rate": 7.651139542190164e-06,
"loss": 0.0267,
"step": 258
},
{
"epoch": 1.18,
"learning_rate": 7.579183448960532e-06,
"loss": 0.0179,
"step": 259
},
{
"epoch": 1.19,
"learning_rate": 7.5073602566319045e-06,
"loss": 0.0291,
"step": 260
},
{
"epoch": 1.19,
"learning_rate": 7.435673908240357e-06,
"loss": 0.0252,
"step": 261
},
{
"epoch": 1.2,
"learning_rate": 7.364128339309326e-06,
"loss": 0.0284,
"step": 262
},
{
"epoch": 1.2,
"learning_rate": 7.292727477633576e-06,
"loss": 0.0417,
"step": 263
},
{
"epoch": 1.21,
"learning_rate": 7.2214752430635625e-06,
"loss": 0.0286,
"step": 264
},
{
"epoch": 1.21,
"learning_rate": 7.1503755472902245e-06,
"loss": 0.0194,
"step": 265
},
{
"epoch": 1.21,
"learning_rate": 7.079432293630244e-06,
"loss": 0.0236,
"step": 266
},
{
"epoch": 1.22,
"learning_rate": 7.008649376811756e-06,
"loss": 0.0462,
"step": 267
},
{
"epoch": 1.22,
"learning_rate": 6.938030682760532e-06,
"loss": 0.0333,
"step": 268
},
{
"epoch": 1.23,
"learning_rate": 6.867580088386646e-06,
"loss": 0.0372,
"step": 269
},
{
"epoch": 1.23,
"learning_rate": 6.797301461371626e-06,
"loss": 0.0206,
"step": 270
},
{
"epoch": 1.24,
"learning_rate": 6.727198659956133e-06,
"loss": 0.0367,
"step": 271
},
{
"epoch": 1.24,
"learning_rate": 6.6572755327281506e-06,
"loss": 0.0441,
"step": 272
},
{
"epoch": 1.25,
"learning_rate": 6.58753591841168e-06,
"loss": 0.0324,
"step": 273
},
{
"epoch": 1.25,
"learning_rate": 6.517983645656014e-06,
"loss": 0.0313,
"step": 274
},
{
"epoch": 1.26,
"learning_rate": 6.448622532825546e-06,
"loss": 0.0333,
"step": 275
},
{
"epoch": 1.26,
"learning_rate": 6.3794563877901385e-06,
"loss": 0.0267,
"step": 276
},
{
"epoch": 1.26,
"learning_rate": 6.310489007716083e-06,
"loss": 0.0294,
"step": 277
},
{
"epoch": 1.27,
"learning_rate": 6.241724178857621e-06,
"loss": 0.027,
"step": 278
},
{
"epoch": 1.27,
"learning_rate": 6.173165676349103e-06,
"loss": 0.0246,
"step": 279
},
{
"epoch": 1.28,
"learning_rate": 6.104817263997727e-06,
"loss": 0.0328,
"step": 280
},
{
"epoch": 1.28,
"learning_rate": 6.0366826940769075e-06,
"loss": 0.0207,
"step": 281
},
{
"epoch": 1.29,
"learning_rate": 5.96876570712028e-06,
"loss": 0.0324,
"step": 282
},
{
"epoch": 1.29,
"learning_rate": 5.90107003171634e-06,
"loss": 0.0156,
"step": 283
},
{
"epoch": 1.3,
"learning_rate": 5.8335993843037695e-06,
"loss": 0.0182,
"step": 284
},
{
"epoch": 1.3,
"learning_rate": 5.766357468967385e-06,
"loss": 0.0261,
"step": 285
},
{
"epoch": 1.31,
"learning_rate": 5.699347977234799e-06,
"loss": 0.0271,
"step": 286
},
{
"epoch": 1.31,
"learning_rate": 5.632574587873736e-06,
"loss": 0.0231,
"step": 287
},
{
"epoch": 1.32,
"learning_rate": 5.566040966690115e-06,
"loss": 0.0164,
"step": 288
},
{
"epoch": 1.32,
"learning_rate": 5.4997507663267546e-06,
"loss": 0.028,
"step": 289
},
{
"epoch": 1.32,
"learning_rate": 5.43370762606287e-06,
"loss": 0.0297,
"step": 290
},
{
"epoch": 1.33,
"learning_rate": 5.367915171614273e-06,
"loss": 0.0354,
"step": 291
},
{
"epoch": 1.33,
"learning_rate": 5.302377014934322e-06,
"loss": 0.0241,
"step": 292
},
{
"epoch": 1.34,
"learning_rate": 5.237096754015635e-06,
"loss": 0.0126,
"step": 293
},
{
"epoch": 1.34,
"learning_rate": 5.172077972692553e-06,
"loss": 0.0517,
"step": 294
},
{
"epoch": 1.35,
"learning_rate": 5.107324240444398e-06,
"loss": 0.0301,
"step": 295
},
{
"epoch": 1.35,
"learning_rate": 5.042839112199509e-06,
"loss": 0.0299,
"step": 296
},
{
"epoch": 1.36,
"learning_rate": 4.978626128140074e-06,
"loss": 0.0315,
"step": 297
},
{
"epoch": 1.36,
"learning_rate": 4.914688813507798e-06,
"loss": 0.045,
"step": 298
},
{
"epoch": 1.37,
"learning_rate": 4.851030678410331e-06,
"loss": 0.0352,
"step": 299
},
{
"epoch": 1.37,
"learning_rate": 4.7876552176286005e-06,
"loss": 0.0348,
"step": 300
},
{
"epoch": 1.37,
"learning_rate": 4.724565910424946e-06,
"loss": 0.0308,
"step": 301
},
{
"epoch": 1.38,
"learning_rate": 4.661766220352098e-06,
"loss": 0.0241,
"step": 302
},
{
"epoch": 1.38,
"learning_rate": 4.59925959506302e-06,
"loss": 0.0334,
"step": 303
},
{
"epoch": 1.39,
"learning_rate": 4.5370494661216835e-06,
"loss": 0.0358,
"step": 304
},
{
"epoch": 1.39,
"learning_rate": 4.475139248814625e-06,
"loss": 0.0273,
"step": 305
},
{
"epoch": 1.4,
"learning_rate": 4.413532341963477e-06,
"loss": 0.0186,
"step": 306
},
{
"epoch": 1.4,
"learning_rate": 4.3522321277383686e-06,
"loss": 0.0225,
"step": 307
},
{
"epoch": 1.41,
"learning_rate": 4.29124197147225e-06,
"loss": 0.0265,
"step": 308
},
{
"epoch": 1.41,
"learning_rate": 4.2305652214761305e-06,
"loss": 0.0302,
"step": 309
},
{
"epoch": 1.42,
"learning_rate": 4.170205208855281e-06,
"loss": 0.0239,
"step": 310
},
{
"epoch": 1.42,
"learning_rate": 4.11016524732632e-06,
"loss": 0.0337,
"step": 311
},
{
"epoch": 1.42,
"learning_rate": 4.050448633035326e-06,
"loss": 0.0261,
"step": 312
},
{
"epoch": 1.43,
"learning_rate": 3.991058644376881e-06,
"loss": 0.0191,
"step": 313
},
{
"epoch": 1.43,
"learning_rate": 3.931998541814069e-06,
"loss": 0.0236,
"step": 314
},
{
"epoch": 1.44,
"learning_rate": 3.873271567699485e-06,
"loss": 0.0384,
"step": 315
},
{
"epoch": 1.44,
"learning_rate": 3.814880946097253e-06,
"loss": 0.0356,
"step": 316
},
{
"epoch": 1.45,
"learning_rate": 3.756829882606001e-06,
"loss": 0.0272,
"step": 317
},
{
"epoch": 1.45,
"learning_rate": 3.6991215641828903e-06,
"loss": 0.0346,
"step": 318
},
{
"epoch": 1.46,
"learning_rate": 3.64175915896865e-06,
"loss": 0.024,
"step": 319
},
{
"epoch": 1.46,
"learning_rate": 3.5847458161136485e-06,
"loss": 0.0254,
"step": 320
},
{
"epoch": 1.47,
"learning_rate": 3.5280846656050127e-06,
"loss": 0.028,
"step": 321
},
{
"epoch": 1.47,
"learning_rate": 3.4717788180947855e-06,
"loss": 0.0294,
"step": 322
},
{
"epoch": 1.47,
"learning_rate": 3.415831364729161e-06,
"loss": 0.0326,
"step": 323
},
{
"epoch": 1.48,
"learning_rate": 3.360245376978779e-06,
"loss": 0.0153,
"step": 324
},
{
"epoch": 1.48,
"learning_rate": 3.305023906470102e-06,
"loss": 0.0369,
"step": 325
},
{
"epoch": 1.49,
"learning_rate": 3.250169984817897e-06,
"loss": 0.031,
"step": 326
},
{
"epoch": 1.49,
"learning_rate": 3.195686623458777e-06,
"loss": 0.0189,
"step": 327
},
{
"epoch": 1.5,
"learning_rate": 3.1415768134858947e-06,
"loss": 0.0202,
"step": 328
},
{
"epoch": 1.5,
"learning_rate": 3.0878435254847393e-06,
"loss": 0.0189,
"step": 329
},
{
"epoch": 1.51,
"learning_rate": 3.0344897093700333e-06,
"loss": 0.027,
"step": 330
},
{
"epoch": 1.51,
"learning_rate": 2.9815182942237885e-06,
"loss": 0.0259,
"step": 331
},
{
"epoch": 1.52,
"learning_rate": 2.9289321881345257e-06,
"loss": 0.0373,
"step": 332
},
{
"epoch": 1.52,
"learning_rate": 2.876734278037593e-06,
"loss": 0.0419,
"step": 333
},
{
"epoch": 1.53,
"learning_rate": 2.8249274295566863e-06,
"loss": 0.0152,
"step": 334
},
{
"epoch": 1.53,
"learning_rate": 2.773514486846546e-06,
"loss": 0.024,
"step": 335
},
{
"epoch": 1.53,
"learning_rate": 2.7224982724367775e-06,
"loss": 0.0134,
"step": 336
},
{
"epoch": 1.54,
"learning_rate": 2.671881587076929e-06,
"loss": 0.0287,
"step": 337
},
{
"epoch": 1.54,
"learning_rate": 2.6216672095827267e-06,
"loss": 0.0249,
"step": 338
},
{
"epoch": 1.55,
"learning_rate": 2.5718578966835116e-06,
"loss": 0.0108,
"step": 339
},
{
"epoch": 1.55,
"learning_rate": 2.5224563828708904e-06,
"loss": 0.0217,
"step": 340
},
{
"epoch": 1.56,
"learning_rate": 2.473465380248643e-06,
"loss": 0.035,
"step": 341
},
{
"epoch": 1.56,
"learning_rate": 2.424887578383799e-06,
"loss": 0.0238,
"step": 342
},
{
"epoch": 1.57,
"learning_rate": 2.3767256441590015e-06,
"loss": 0.0146,
"step": 343
},
{
"epoch": 1.57,
"learning_rate": 2.328982221626087e-06,
"loss": 0.0221,
"step": 344
},
{
"epoch": 1.58,
"learning_rate": 2.2816599318609367e-06,
"loss": 0.0284,
"step": 345
},
{
"epoch": 1.58,
"learning_rate": 2.234761372819577e-06,
"loss": 0.021,
"step": 346
},
{
"epoch": 1.58,
"learning_rate": 2.1882891191955535e-06,
"loss": 0.0165,
"step": 347
},
{
"epoch": 1.59,
"learning_rate": 2.1422457222785875e-06,
"loss": 0.0224,
"step": 348
},
{
"epoch": 1.59,
"learning_rate": 2.096633709814504e-06,
"loss": 0.0135,
"step": 349
},
{
"epoch": 1.6,
"learning_rate": 2.0514555858664663e-06,
"loss": 0.0173,
"step": 350
},
{
"epoch": 1.6,
"learning_rate": 2.0067138306775126e-06,
"loss": 0.0221,
"step": 351
},
{
"epoch": 1.61,
"learning_rate": 1.962410900534367e-06,
"loss": 0.0191,
"step": 352
},
{
"epoch": 1.61,
"learning_rate": 1.918549227632619e-06,
"loss": 0.0221,
"step": 353
},
{
"epoch": 1.62,
"learning_rate": 1.875131219943187e-06,
"loss": 0.0198,
"step": 354
},
{
"epoch": 1.62,
"learning_rate": 1.832159261080122e-06,
"loss": 0.0251,
"step": 355
},
{
"epoch": 1.63,
"learning_rate": 1.7896357101697405e-06,
"loss": 0.0196,
"step": 356
},
{
"epoch": 1.63,
"learning_rate": 1.7475629017211348e-06,
"loss": 0.0138,
"step": 357
},
{
"epoch": 1.63,
"learning_rate": 1.7059431454979825e-06,
"loss": 0.0297,
"step": 358
},
{
"epoch": 1.64,
"learning_rate": 1.6647787263917615e-06,
"loss": 0.0287,
"step": 359
},
{
"epoch": 1.64,
"learning_rate": 1.6240719042963004e-06,
"loss": 0.0183,
"step": 360
},
{
"epoch": 1.65,
"learning_rate": 1.5838249139837204e-06,
"loss": 0.0191,
"step": 361
},
{
"epoch": 1.65,
"learning_rate": 1.5440399649817384e-06,
"loss": 0.0245,
"step": 362
},
{
"epoch": 1.66,
"learning_rate": 1.5047192414523814e-06,
"loss": 0.0253,
"step": 363
},
{
"epoch": 1.66,
"learning_rate": 1.4658649020720539e-06,
"loss": 0.0166,
"step": 364
},
{
"epoch": 1.67,
"learning_rate": 1.42747907991305e-06,
"loss": 0.0169,
"step": 365
},
{
"epoch": 1.67,
"learning_rate": 1.3895638823264447e-06,
"loss": 0.0289,
"step": 366
},
{
"epoch": 1.68,
"learning_rate": 1.3521213908263931e-06,
"loss": 0.0153,
"step": 367
},
{
"epoch": 1.68,
"learning_rate": 1.3151536609758586e-06,
"loss": 0.0108,
"step": 368
},
{
"epoch": 1.68,
"learning_rate": 1.2786627222737801e-06,
"loss": 0.0218,
"step": 369
},
{
"epoch": 1.69,
"learning_rate": 1.2426505780436326e-06,
"loss": 0.0296,
"step": 370
},
{
"epoch": 1.69,
"learning_rate": 1.2071192053234581e-06,
"loss": 0.0157,
"step": 371
},
{
"epoch": 1.7,
"learning_rate": 1.1720705547573263e-06,
"loss": 0.0103,
"step": 372
},
{
"epoch": 1.7,
"learning_rate": 1.1375065504882465e-06,
"loss": 0.019,
"step": 373
},
{
"epoch": 1.71,
"learning_rate": 1.1034290900525279e-06,
"loss": 0.0184,
"step": 374
},
{
"epoch": 1.71,
"learning_rate": 1.069840044275615e-06,
"loss": 0.0289,
"step": 375
},
{
"epoch": 1.72,
"learning_rate": 1.0367412571693746e-06,
"loss": 0.0117,
"step": 376
},
{
"epoch": 1.72,
"learning_rate": 1.0041345458308615e-06,
"loss": 0.0144,
"step": 377
},
{
"epoch": 1.73,
"learning_rate": 9.720217003425648e-07,
"loss": 0.0173,
"step": 378
},
{
"epoch": 1.73,
"learning_rate": 9.404044836741343e-07,
"loss": 0.0244,
"step": 379
},
{
"epoch": 1.74,
"learning_rate": 9.092846315855841e-07,
"loss": 0.0432,
"step": 380
},
{
"epoch": 1.74,
"learning_rate": 8.786638525320146e-07,
"loss": 0.0213,
"step": 381
},
{
"epoch": 1.74,
"learning_rate": 8.485438275698154e-07,
"loss": 0.0206,
"step": 382
},
{
"epoch": 1.75,
"learning_rate": 8.189262102643747e-07,
"loss": 0.0205,
"step": 383
},
{
"epoch": 1.75,
"learning_rate": 7.898126265992912e-07,
"loss": 0.0255,
"step": 384
},
{
"epoch": 1.76,
"learning_rate": 7.612046748871327e-07,
"loss": 0.0162,
"step": 385
},
{
"epoch": 1.76,
"learning_rate": 7.331039256816664e-07,
"loss": 0.0275,
"step": 386
},
{
"epoch": 1.77,
"learning_rate": 7.055119216916451e-07,
"loss": 0.0199,
"step": 387
},
{
"epoch": 1.77,
"learning_rate": 6.784301776961222e-07,
"loss": 0.0245,
"step": 388
},
{
"epoch": 1.78,
"learning_rate": 6.518601804612734e-07,
"loss": 0.0379,
"step": 389
},
{
"epoch": 1.78,
"learning_rate": 6.258033886587911e-07,
"loss": 0.0318,
"step": 390
},
{
"epoch": 1.79,
"learning_rate": 6.002612327857971e-07,
"loss": 0.0247,
"step": 391
},
{
"epoch": 1.79,
"learning_rate": 5.752351150863133e-07,
"loss": 0.0221,
"step": 392
},
{
"epoch": 1.79,
"learning_rate": 5.50726409474267e-07,
"loss": 0.0192,
"step": 393
},
{
"epoch": 1.8,
"learning_rate": 5.267364614580861e-07,
"loss": 0.02,
"step": 394
},
{
"epoch": 1.8,
"learning_rate": 5.03266588066812e-07,
"loss": 0.017,
"step": 395
},
{
"epoch": 1.81,
"learning_rate": 4.80318077777805e-07,
"loss": 0.0193,
"step": 396
},
{
"epoch": 1.81,
"learning_rate": 4.578921904460076e-07,
"loss": 0.023,
"step": 397
},
{
"epoch": 1.82,
"learning_rate": 4.359901572347758e-07,
"loss": 0.0194,
"step": 398
},
{
"epoch": 1.82,
"learning_rate": 4.1461318054829446e-07,
"loss": 0.0233,
"step": 399
},
{
"epoch": 1.83,
"learning_rate": 3.937624339655599e-07,
"loss": 0.0208,
"step": 400
},
{
"epoch": 1.83,
"learning_rate": 3.734390621759565e-07,
"loss": 0.022,
"step": 401
},
{
"epoch": 1.84,
"learning_rate": 3.5364418091641374e-07,
"loss": 0.0333,
"step": 402
},
{
"epoch": 1.84,
"learning_rate": 3.343788769101486e-07,
"loss": 0.018,
"step": 403
},
{
"epoch": 1.84,
"learning_rate": 3.1564420780701435e-07,
"loss": 0.0302,
"step": 404
},
{
"epoch": 1.85,
"learning_rate": 2.974412021254236e-07,
"loss": 0.0167,
"step": 405
},
{
"epoch": 1.85,
"learning_rate": 2.7977085919589253e-07,
"loss": 0.0159,
"step": 406
},
{
"epoch": 1.86,
"learning_rate": 2.626341491061801e-07,
"loss": 0.024,
"step": 407
},
{
"epoch": 1.86,
"learning_rate": 2.4603201264802423e-07,
"loss": 0.0185,
"step": 408
},
{
"epoch": 1.87,
"learning_rate": 2.2996536126549395e-07,
"loss": 0.0395,
"step": 409
},
{
"epoch": 1.87,
"learning_rate": 2.1443507700495968e-07,
"loss": 0.0223,
"step": 410
},
{
"epoch": 1.88,
"learning_rate": 1.9944201246666072e-07,
"loss": 0.0207,
"step": 411
},
{
"epoch": 1.88,
"learning_rate": 1.8498699075789938e-07,
"loss": 0.0183,
"step": 412
},
{
"epoch": 1.89,
"learning_rate": 1.7107080544785937e-07,
"loss": 0.0287,
"step": 413
},
{
"epoch": 1.89,
"learning_rate": 1.5769422052403172e-07,
"loss": 0.0243,
"step": 414
},
{
"epoch": 1.89,
"learning_rate": 1.4485797035027728e-07,
"loss": 0.0133,
"step": 415
},
{
"epoch": 1.9,
"learning_rate": 1.3256275962651223e-07,
"loss": 0.0473,
"step": 416
},
{
"epoch": 1.9,
"learning_rate": 1.208092633500102e-07,
"loss": 0.0303,
"step": 417
},
{
"epoch": 1.91,
"learning_rate": 1.0959812677835968e-07,
"loss": 0.018,
"step": 418
},
{
"epoch": 1.91,
"learning_rate": 9.892996539403232e-08,
"loss": 0.0131,
"step": 419
},
{
"epoch": 1.92,
"learning_rate": 8.880536487059333e-08,
"loss": 0.0191,
"step": 420
},
{
"epoch": 1.92,
"learning_rate": 7.922488104054826e-08,
"loss": 0.0216,
"step": 421
},
{
"epoch": 1.93,
"learning_rate": 7.018903986483083e-08,
"loss": 0.0149,
"step": 422
},
{
"epoch": 1.93,
"learning_rate": 6.169833740392594e-08,
"loss": 0.0386,
"step": 423
},
{
"epoch": 1.94,
"learning_rate": 5.375323979063929e-08,
"loss": 0.0128,
"step": 424
},
{
"epoch": 1.94,
"learning_rate": 4.6354183204503355e-08,
"loss": 0.0156,
"step": 425
},
{
"epoch": 1.95,
"learning_rate": 3.950157384783104e-08,
"loss": 0.0189,
"step": 426
},
{
"epoch": 1.95,
"learning_rate": 3.319578792342126e-08,
"loss": 0.0321,
"step": 427
},
{
"epoch": 1.95,
"learning_rate": 2.7437171613898806e-08,
"loss": 0.0183,
"step": 428
},
{
"epoch": 1.96,
"learning_rate": 2.2226041062715087e-08,
"loss": 0.0106,
"step": 429
},
{
"epoch": 1.96,
"learning_rate": 1.7562682356786488e-08,
"loss": 0.0264,
"step": 430
},
{
"epoch": 1.97,
"learning_rate": 1.3447351510792461e-08,
"loss": 0.0187,
"step": 431
},
{
"epoch": 1.97,
"learning_rate": 9.88027445312123e-09,
"loss": 0.0279,
"step": 432
},
{
"epoch": 1.98,
"learning_rate": 6.861647013461925e-09,
"loss": 0.0223,
"step": 433
},
{
"epoch": 1.98,
"learning_rate": 4.39163491205652e-09,
"loss": 0.0236,
"step": 434
},
{
"epoch": 1.99,
"learning_rate": 2.4703737506037715e-09,
"loss": 0.0209,
"step": 435
},
{
"epoch": 1.99,
"learning_rate": 1.0979690048107395e-09,
"loss": 0.0174,
"step": 436
},
{
"epoch": 2.0,
"learning_rate": 2.7449601860629614e-10,
"loss": 0.0156,
"step": 437
},
{
"epoch": 2.0,
"learning_rate": 0.0,
"loss": 0.0377,
"step": 438
},
{
"epoch": 2.0,
"step": 438,
"total_flos": 28999390887936.0,
"train_loss": 0.0449560657483802,
"train_runtime": 898.5819,
"train_samples_per_second": 31.16,
"train_steps_per_second": 0.487
}
],
"logging_steps": 1.0,
"max_steps": 438,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 1100,
"total_flos": 28999390887936.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}