vicuna-7b-lora-fin-1epoch / trainer_state.json
wza's picture
upload
bb47802
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"global_step": 791,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 0,
"loss": 1.4791,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 0,
"loss": 1.5012,
"step": 2
},
{
"epoch": 0.0,
"learning_rate": 0,
"loss": 1.5116,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 0,
"loss": 1.4817,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 0,
"loss": 1.4658,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 0,
"loss": 1.4871,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 0,
"loss": 1.4805,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 0,
"loss": 1.538,
"step": 8
},
{
"epoch": 0.01,
"learning_rate": 0,
"loss": 1.4946,
"step": 9
},
{
"epoch": 0.01,
"learning_rate": 0,
"loss": 1.4747,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 0,
"loss": 1.4675,
"step": 11
},
{
"epoch": 0.02,
"learning_rate": 0,
"loss": 1.5055,
"step": 12
},
{
"epoch": 0.02,
"learning_rate": 0,
"loss": 1.4786,
"step": 13
},
{
"epoch": 0.02,
"learning_rate": 0,
"loss": 1.5139,
"step": 14
},
{
"epoch": 0.02,
"learning_rate": 0,
"loss": 1.473,
"step": 15
},
{
"epoch": 0.02,
"learning_rate": 0,
"loss": 1.4784,
"step": 16
},
{
"epoch": 0.02,
"learning_rate": 0,
"loss": 1.4888,
"step": 17
},
{
"epoch": 0.02,
"learning_rate": 0.0,
"loss": 1.4879,
"step": 18
},
{
"epoch": 0.02,
"learning_rate": 3.581044635020827e-06,
"loss": 1.5187,
"step": 19
},
{
"epoch": 0.03,
"learning_rate": 5.675821459916692e-06,
"loss": 1.4978,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 7.162089270041654e-06,
"loss": 1.5043,
"step": 21
},
{
"epoch": 0.03,
"learning_rate": 8.31492814710052e-06,
"loss": 1.5099,
"step": 22
},
{
"epoch": 0.03,
"learning_rate": 9.256866094937518e-06,
"loss": 1.5065,
"step": 23
},
{
"epoch": 0.03,
"learning_rate": 1.0053263282233697e-05,
"loss": 1.4681,
"step": 24
},
{
"epoch": 0.03,
"learning_rate": 1.0743133905062482e-05,
"loss": 1.4753,
"step": 25
},
{
"epoch": 0.03,
"learning_rate": 1.1351642919833384e-05,
"loss": 1.5138,
"step": 26
},
{
"epoch": 0.03,
"learning_rate": 1.1895972782121347e-05,
"loss": 1.4987,
"step": 27
},
{
"epoch": 0.04,
"learning_rate": 1.2388379038142511e-05,
"loss": 1.4604,
"step": 28
},
{
"epoch": 0.04,
"learning_rate": 1.2837910729958346e-05,
"loss": 1.5164,
"step": 29
},
{
"epoch": 0.04,
"learning_rate": 1.325143979986714e-05,
"loss": 1.5106,
"step": 30
},
{
"epoch": 0.04,
"learning_rate": 1.3634307917254523e-05,
"loss": 1.4675,
"step": 31
},
{
"epoch": 0.04,
"learning_rate": 1.399074960701721e-05,
"loss": 1.4595,
"step": 32
},
{
"epoch": 0.04,
"learning_rate": 1.4324178540083309e-05,
"loss": 1.4644,
"step": 33
},
{
"epoch": 0.04,
"learning_rate": 1.4637386878506516e-05,
"loss": 1.5077,
"step": 34
},
{
"epoch": 0.04,
"learning_rate": 1.4932687554854209e-05,
"loss": 1.5314,
"step": 35
},
{
"epoch": 0.05,
"learning_rate": 1.5212018031974513e-05,
"loss": 1.4735,
"step": 36
},
{
"epoch": 0.05,
"learning_rate": 1.5477017417142176e-05,
"loss": 1.4847,
"step": 37
},
{
"epoch": 0.05,
"learning_rate": 1.5729084742150388e-05,
"loss": 1.4679,
"step": 38
},
{
"epoch": 0.05,
"learning_rate": 1.596942367316334e-05,
"loss": 1.5153,
"step": 39
},
{
"epoch": 0.05,
"learning_rate": 1.6199077273922286e-05,
"loss": 1.4827,
"step": 40
},
{
"epoch": 0.05,
"learning_rate": 1.6418955364979175e-05,
"loss": 1.4994,
"step": 41
},
{
"epoch": 0.05,
"learning_rate": 1.662985629420104e-05,
"loss": 1.5087,
"step": 42
},
{
"epoch": 0.05,
"learning_rate": 1.6832484434887967e-05,
"loss": 1.4974,
"step": 43
},
{
"epoch": 0.06,
"learning_rate": 1.7027464379750072e-05,
"loss": 1.4484,
"step": 44
},
{
"epoch": 0.06,
"learning_rate": 1.721535255227535e-05,
"loss": 1.4335,
"step": 45
},
{
"epoch": 0.06,
"learning_rate": 1.7396646779634733e-05,
"loss": 1.4854,
"step": 46
},
{
"epoch": 0.06,
"learning_rate": 1.757179424203804e-05,
"loss": 1.4985,
"step": 47
},
{
"epoch": 0.06,
"learning_rate": 1.7741198118150898e-05,
"loss": 1.4907,
"step": 48
},
{
"epoch": 0.06,
"learning_rate": 1.7905223175104135e-05,
"loss": 1.4369,
"step": 49
},
{
"epoch": 0.06,
"learning_rate": 1.80642004980592e-05,
"loss": 1.5002,
"step": 50
},
{
"epoch": 0.06,
"learning_rate": 1.8218431513527344e-05,
"loss": 1.4749,
"step": 51
},
{
"epoch": 0.07,
"learning_rate": 1.8368191429334215e-05,
"loss": 1.5021,
"step": 52
},
{
"epoch": 0.07,
"learning_rate": 1.8513732189875035e-05,
"loss": 1.4545,
"step": 53
},
{
"epoch": 0.07,
"learning_rate": 1.8655285026376745e-05,
"loss": 1.449,
"step": 54
},
{
"epoch": 0.07,
"learning_rate": 1.8793062666995343e-05,
"loss": 1.4276,
"step": 55
},
{
"epoch": 0.07,
"learning_rate": 1.892726125978383e-05,
"loss": 1.4625,
"step": 56
},
{
"epoch": 0.07,
"learning_rate": 1.9058062052163002e-05,
"loss": 1.4697,
"step": 57
},
{
"epoch": 0.07,
"learning_rate": 1.9185632862982668e-05,
"loss": 1.4584,
"step": 58
},
{
"epoch": 0.07,
"learning_rate": 1.9310129377171218e-05,
"loss": 1.4801,
"step": 59
},
{
"epoch": 0.08,
"learning_rate": 1.9431696288028554e-05,
"loss": 1.4368,
"step": 60
},
{
"epoch": 0.08,
"learning_rate": 1.9550468308184164e-05,
"loss": 1.4424,
"step": 61
},
{
"epoch": 0.08,
"learning_rate": 1.96665710669339e-05,
"loss": 1.418,
"step": 62
},
{
"epoch": 0.08,
"learning_rate": 1.9780121908943112e-05,
"loss": 1.4525,
"step": 63
},
{
"epoch": 0.08,
"learning_rate": 1.98912306070467e-05,
"loss": 1.4115,
"step": 64
},
{
"epoch": 0.08,
"learning_rate": 2e-05,
"loss": 1.4584,
"step": 65
},
{
"epoch": 0.08,
"learning_rate": 2e-05,
"loss": 1.3983,
"step": 66
},
{
"epoch": 0.08,
"learning_rate": 1.9986962190352023e-05,
"loss": 1.4412,
"step": 67
},
{
"epoch": 0.09,
"learning_rate": 1.9973924380704045e-05,
"loss": 1.4321,
"step": 68
},
{
"epoch": 0.09,
"learning_rate": 1.9960886571056063e-05,
"loss": 1.429,
"step": 69
},
{
"epoch": 0.09,
"learning_rate": 1.9947848761408084e-05,
"loss": 1.4509,
"step": 70
},
{
"epoch": 0.09,
"learning_rate": 1.9934810951760106e-05,
"loss": 1.4283,
"step": 71
},
{
"epoch": 0.09,
"learning_rate": 1.9921773142112127e-05,
"loss": 1.4553,
"step": 72
},
{
"epoch": 0.09,
"learning_rate": 1.990873533246415e-05,
"loss": 1.3925,
"step": 73
},
{
"epoch": 0.09,
"learning_rate": 1.9895697522816167e-05,
"loss": 1.4257,
"step": 74
},
{
"epoch": 0.09,
"learning_rate": 1.988265971316819e-05,
"loss": 1.4058,
"step": 75
},
{
"epoch": 0.1,
"learning_rate": 1.986962190352021e-05,
"loss": 1.3916,
"step": 76
},
{
"epoch": 0.1,
"learning_rate": 1.9856584093872232e-05,
"loss": 1.4246,
"step": 77
},
{
"epoch": 0.1,
"learning_rate": 1.984354628422425e-05,
"loss": 1.434,
"step": 78
},
{
"epoch": 0.1,
"learning_rate": 1.9830508474576275e-05,
"loss": 1.3815,
"step": 79
},
{
"epoch": 0.1,
"learning_rate": 1.9817470664928293e-05,
"loss": 1.386,
"step": 80
},
{
"epoch": 0.1,
"learning_rate": 1.9804432855280314e-05,
"loss": 1.3976,
"step": 81
},
{
"epoch": 0.1,
"learning_rate": 1.9791395045632336e-05,
"loss": 1.4288,
"step": 82
},
{
"epoch": 0.1,
"learning_rate": 1.9778357235984354e-05,
"loss": 1.4148,
"step": 83
},
{
"epoch": 0.11,
"learning_rate": 1.976531942633638e-05,
"loss": 1.3914,
"step": 84
},
{
"epoch": 0.11,
"learning_rate": 1.9752281616688397e-05,
"loss": 1.4066,
"step": 85
},
{
"epoch": 0.11,
"learning_rate": 1.973924380704042e-05,
"loss": 1.408,
"step": 86
},
{
"epoch": 0.11,
"learning_rate": 1.972620599739244e-05,
"loss": 1.395,
"step": 87
},
{
"epoch": 0.11,
"learning_rate": 1.9713168187744462e-05,
"loss": 1.421,
"step": 88
},
{
"epoch": 0.11,
"learning_rate": 1.970013037809648e-05,
"loss": 1.3871,
"step": 89
},
{
"epoch": 0.11,
"learning_rate": 1.96870925684485e-05,
"loss": 1.4153,
"step": 90
},
{
"epoch": 0.12,
"learning_rate": 1.9674054758800523e-05,
"loss": 1.3899,
"step": 91
},
{
"epoch": 0.12,
"learning_rate": 1.9661016949152545e-05,
"loss": 1.4215,
"step": 92
},
{
"epoch": 0.12,
"learning_rate": 1.9647979139504566e-05,
"loss": 1.3961,
"step": 93
},
{
"epoch": 0.12,
"learning_rate": 1.9634941329856584e-05,
"loss": 1.4343,
"step": 94
},
{
"epoch": 0.12,
"learning_rate": 1.9621903520208606e-05,
"loss": 1.4023,
"step": 95
},
{
"epoch": 0.12,
"learning_rate": 1.9608865710560627e-05,
"loss": 1.4044,
"step": 96
},
{
"epoch": 0.12,
"learning_rate": 1.959582790091265e-05,
"loss": 1.4089,
"step": 97
},
{
"epoch": 0.12,
"learning_rate": 1.958279009126467e-05,
"loss": 1.4286,
"step": 98
},
{
"epoch": 0.13,
"learning_rate": 1.956975228161669e-05,
"loss": 1.3584,
"step": 99
},
{
"epoch": 0.13,
"learning_rate": 1.9556714471968713e-05,
"loss": 1.4116,
"step": 100
},
{
"epoch": 0.13,
"learning_rate": 1.954367666232073e-05,
"loss": 1.3701,
"step": 101
},
{
"epoch": 0.13,
"learning_rate": 1.9530638852672753e-05,
"loss": 1.4187,
"step": 102
},
{
"epoch": 0.13,
"learning_rate": 1.9517601043024775e-05,
"loss": 1.3879,
"step": 103
},
{
"epoch": 0.13,
"learning_rate": 1.9504563233376793e-05,
"loss": 1.3561,
"step": 104
},
{
"epoch": 0.13,
"learning_rate": 1.9491525423728814e-05,
"loss": 1.3715,
"step": 105
},
{
"epoch": 0.13,
"learning_rate": 1.9478487614080836e-05,
"loss": 1.3772,
"step": 106
},
{
"epoch": 0.14,
"learning_rate": 1.9465449804432857e-05,
"loss": 1.3914,
"step": 107
},
{
"epoch": 0.14,
"learning_rate": 1.9452411994784876e-05,
"loss": 1.4004,
"step": 108
},
{
"epoch": 0.14,
"learning_rate": 1.94393741851369e-05,
"loss": 1.3425,
"step": 109
},
{
"epoch": 0.14,
"learning_rate": 1.942633637548892e-05,
"loss": 1.3588,
"step": 110
},
{
"epoch": 0.14,
"learning_rate": 1.941329856584094e-05,
"loss": 1.4044,
"step": 111
},
{
"epoch": 0.14,
"learning_rate": 1.940026075619296e-05,
"loss": 1.3721,
"step": 112
},
{
"epoch": 0.14,
"learning_rate": 1.938722294654498e-05,
"loss": 1.3787,
"step": 113
},
{
"epoch": 0.14,
"learning_rate": 1.9374185136897005e-05,
"loss": 1.3646,
"step": 114
},
{
"epoch": 0.15,
"learning_rate": 1.9361147327249023e-05,
"loss": 1.3621,
"step": 115
},
{
"epoch": 0.15,
"learning_rate": 1.9348109517601044e-05,
"loss": 1.3546,
"step": 116
},
{
"epoch": 0.15,
"learning_rate": 1.9335071707953066e-05,
"loss": 1.385,
"step": 117
},
{
"epoch": 0.15,
"learning_rate": 1.9322033898305087e-05,
"loss": 1.3973,
"step": 118
},
{
"epoch": 0.15,
"learning_rate": 1.930899608865711e-05,
"loss": 1.376,
"step": 119
},
{
"epoch": 0.15,
"learning_rate": 1.9295958279009127e-05,
"loss": 1.3938,
"step": 120
},
{
"epoch": 0.15,
"learning_rate": 1.928292046936115e-05,
"loss": 1.3755,
"step": 121
},
{
"epoch": 0.15,
"learning_rate": 1.926988265971317e-05,
"loss": 1.3964,
"step": 122
},
{
"epoch": 0.16,
"learning_rate": 1.9256844850065192e-05,
"loss": 1.3369,
"step": 123
},
{
"epoch": 0.16,
"learning_rate": 1.924380704041721e-05,
"loss": 1.3865,
"step": 124
},
{
"epoch": 0.16,
"learning_rate": 1.923076923076923e-05,
"loss": 1.3689,
"step": 125
},
{
"epoch": 0.16,
"learning_rate": 1.9217731421121253e-05,
"loss": 1.3519,
"step": 126
},
{
"epoch": 0.16,
"learning_rate": 1.9204693611473274e-05,
"loss": 1.3663,
"step": 127
},
{
"epoch": 0.16,
"learning_rate": 1.9191655801825296e-05,
"loss": 1.3595,
"step": 128
},
{
"epoch": 0.16,
"learning_rate": 1.9178617992177314e-05,
"loss": 1.3594,
"step": 129
},
{
"epoch": 0.16,
"learning_rate": 1.916558018252934e-05,
"loss": 1.3555,
"step": 130
},
{
"epoch": 0.17,
"learning_rate": 1.9152542372881357e-05,
"loss": 1.3864,
"step": 131
},
{
"epoch": 0.17,
"learning_rate": 1.913950456323338e-05,
"loss": 1.4041,
"step": 132
},
{
"epoch": 0.17,
"learning_rate": 1.91264667535854e-05,
"loss": 1.3717,
"step": 133
},
{
"epoch": 0.17,
"learning_rate": 1.911342894393742e-05,
"loss": 1.3451,
"step": 134
},
{
"epoch": 0.17,
"learning_rate": 1.910039113428944e-05,
"loss": 1.3484,
"step": 135
},
{
"epoch": 0.17,
"learning_rate": 1.908735332464146e-05,
"loss": 1.4012,
"step": 136
},
{
"epoch": 0.17,
"learning_rate": 1.9074315514993483e-05,
"loss": 1.3636,
"step": 137
},
{
"epoch": 0.17,
"learning_rate": 1.9061277705345505e-05,
"loss": 1.3663,
"step": 138
},
{
"epoch": 0.18,
"learning_rate": 1.9048239895697526e-05,
"loss": 1.3757,
"step": 139
},
{
"epoch": 0.18,
"learning_rate": 1.9035202086049544e-05,
"loss": 1.3837,
"step": 140
},
{
"epoch": 0.18,
"learning_rate": 1.9022164276401566e-05,
"loss": 1.3621,
"step": 141
},
{
"epoch": 0.18,
"learning_rate": 1.9009126466753587e-05,
"loss": 1.3636,
"step": 142
},
{
"epoch": 0.18,
"learning_rate": 1.8996088657105605e-05,
"loss": 1.3512,
"step": 143
},
{
"epoch": 0.18,
"learning_rate": 1.898305084745763e-05,
"loss": 1.3347,
"step": 144
},
{
"epoch": 0.18,
"learning_rate": 1.897001303780965e-05,
"loss": 1.3534,
"step": 145
},
{
"epoch": 0.18,
"learning_rate": 1.895697522816167e-05,
"loss": 1.3199,
"step": 146
},
{
"epoch": 0.19,
"learning_rate": 1.894393741851369e-05,
"loss": 1.3707,
"step": 147
},
{
"epoch": 0.19,
"learning_rate": 1.8930899608865713e-05,
"loss": 1.3481,
"step": 148
},
{
"epoch": 0.19,
"learning_rate": 1.8917861799217735e-05,
"loss": 1.3829,
"step": 149
},
{
"epoch": 0.19,
"learning_rate": 1.8904823989569753e-05,
"loss": 1.3748,
"step": 150
},
{
"epoch": 0.19,
"learning_rate": 1.8891786179921774e-05,
"loss": 1.3545,
"step": 151
},
{
"epoch": 0.19,
"learning_rate": 1.8878748370273796e-05,
"loss": 1.4016,
"step": 152
},
{
"epoch": 0.19,
"learning_rate": 1.8865710560625817e-05,
"loss": 1.3523,
"step": 153
},
{
"epoch": 0.19,
"learning_rate": 1.8852672750977836e-05,
"loss": 1.3497,
"step": 154
},
{
"epoch": 0.2,
"learning_rate": 1.8839634941329857e-05,
"loss": 1.344,
"step": 155
},
{
"epoch": 0.2,
"learning_rate": 1.882659713168188e-05,
"loss": 1.347,
"step": 156
},
{
"epoch": 0.2,
"learning_rate": 1.88135593220339e-05,
"loss": 1.306,
"step": 157
},
{
"epoch": 0.2,
"learning_rate": 1.880052151238592e-05,
"loss": 1.3613,
"step": 158
},
{
"epoch": 0.2,
"learning_rate": 1.878748370273794e-05,
"loss": 1.335,
"step": 159
},
{
"epoch": 0.2,
"learning_rate": 1.8774445893089965e-05,
"loss": 1.3098,
"step": 160
},
{
"epoch": 0.2,
"learning_rate": 1.8761408083441983e-05,
"loss": 1.3601,
"step": 161
},
{
"epoch": 0.2,
"learning_rate": 1.8748370273794004e-05,
"loss": 1.3298,
"step": 162
},
{
"epoch": 0.21,
"learning_rate": 1.8735332464146026e-05,
"loss": 1.382,
"step": 163
},
{
"epoch": 0.21,
"learning_rate": 1.8722294654498044e-05,
"loss": 1.3274,
"step": 164
},
{
"epoch": 0.21,
"learning_rate": 1.870925684485007e-05,
"loss": 1.3522,
"step": 165
},
{
"epoch": 0.21,
"learning_rate": 1.8696219035202087e-05,
"loss": 1.3281,
"step": 166
},
{
"epoch": 0.21,
"learning_rate": 1.868318122555411e-05,
"loss": 1.3514,
"step": 167
},
{
"epoch": 0.21,
"learning_rate": 1.867014341590613e-05,
"loss": 1.3655,
"step": 168
},
{
"epoch": 0.21,
"learning_rate": 1.8657105606258152e-05,
"loss": 1.3663,
"step": 169
},
{
"epoch": 0.21,
"learning_rate": 1.864406779661017e-05,
"loss": 1.3439,
"step": 170
},
{
"epoch": 0.22,
"learning_rate": 1.863102998696219e-05,
"loss": 1.3763,
"step": 171
},
{
"epoch": 0.22,
"learning_rate": 1.8617992177314213e-05,
"loss": 1.3954,
"step": 172
},
{
"epoch": 0.22,
"learning_rate": 1.860495436766623e-05,
"loss": 1.3251,
"step": 173
},
{
"epoch": 0.22,
"learning_rate": 1.8591916558018256e-05,
"loss": 1.3619,
"step": 174
},
{
"epoch": 0.22,
"learning_rate": 1.8578878748370274e-05,
"loss": 1.3516,
"step": 175
},
{
"epoch": 0.22,
"learning_rate": 1.8565840938722296e-05,
"loss": 1.3173,
"step": 176
},
{
"epoch": 0.22,
"learning_rate": 1.8552803129074317e-05,
"loss": 1.3649,
"step": 177
},
{
"epoch": 0.23,
"learning_rate": 1.853976531942634e-05,
"loss": 1.3279,
"step": 178
},
{
"epoch": 0.23,
"learning_rate": 1.852672750977836e-05,
"loss": 1.3372,
"step": 179
},
{
"epoch": 0.23,
"learning_rate": 1.851368970013038e-05,
"loss": 1.3315,
"step": 180
},
{
"epoch": 0.23,
"learning_rate": 1.85006518904824e-05,
"loss": 1.3341,
"step": 181
},
{
"epoch": 0.23,
"learning_rate": 1.848761408083442e-05,
"loss": 1.332,
"step": 182
},
{
"epoch": 0.23,
"learning_rate": 1.8474576271186443e-05,
"loss": 1.3092,
"step": 183
},
{
"epoch": 0.23,
"learning_rate": 1.8461538461538465e-05,
"loss": 1.3214,
"step": 184
},
{
"epoch": 0.23,
"learning_rate": 1.8448500651890483e-05,
"loss": 1.3785,
"step": 185
},
{
"epoch": 0.24,
"learning_rate": 1.8435462842242504e-05,
"loss": 1.3363,
"step": 186
},
{
"epoch": 0.24,
"learning_rate": 1.8422425032594526e-05,
"loss": 1.3572,
"step": 187
},
{
"epoch": 0.24,
"learning_rate": 1.8409387222946547e-05,
"loss": 1.3531,
"step": 188
},
{
"epoch": 0.24,
"learning_rate": 1.8396349413298566e-05,
"loss": 1.3387,
"step": 189
},
{
"epoch": 0.24,
"learning_rate": 1.838331160365059e-05,
"loss": 1.3566,
"step": 190
},
{
"epoch": 0.24,
"learning_rate": 1.837027379400261e-05,
"loss": 1.3018,
"step": 191
},
{
"epoch": 0.24,
"learning_rate": 1.835723598435463e-05,
"loss": 1.276,
"step": 192
},
{
"epoch": 0.24,
"learning_rate": 1.834419817470665e-05,
"loss": 1.3368,
"step": 193
},
{
"epoch": 0.25,
"learning_rate": 1.833116036505867e-05,
"loss": 1.3339,
"step": 194
},
{
"epoch": 0.25,
"learning_rate": 1.8318122555410695e-05,
"loss": 1.2972,
"step": 195
},
{
"epoch": 0.25,
"learning_rate": 1.8305084745762713e-05,
"loss": 1.3182,
"step": 196
},
{
"epoch": 0.25,
"learning_rate": 1.8292046936114734e-05,
"loss": 1.3167,
"step": 197
},
{
"epoch": 0.25,
"learning_rate": 1.8279009126466756e-05,
"loss": 1.3792,
"step": 198
},
{
"epoch": 0.25,
"learning_rate": 1.8265971316818777e-05,
"loss": 1.3479,
"step": 199
},
{
"epoch": 0.25,
"learning_rate": 1.8252933507170796e-05,
"loss": 1.3412,
"step": 200
},
{
"epoch": 0.25,
"learning_rate": 1.8239895697522817e-05,
"loss": 1.3534,
"step": 201
},
{
"epoch": 0.26,
"learning_rate": 1.822685788787484e-05,
"loss": 1.3856,
"step": 202
},
{
"epoch": 0.26,
"learning_rate": 1.821382007822686e-05,
"loss": 1.3234,
"step": 203
},
{
"epoch": 0.26,
"learning_rate": 1.8200782268578882e-05,
"loss": 1.3444,
"step": 204
},
{
"epoch": 0.26,
"learning_rate": 1.81877444589309e-05,
"loss": 1.3439,
"step": 205
},
{
"epoch": 0.26,
"learning_rate": 1.817470664928292e-05,
"loss": 1.3241,
"step": 206
},
{
"epoch": 0.26,
"learning_rate": 1.8161668839634943e-05,
"loss": 1.366,
"step": 207
},
{
"epoch": 0.26,
"learning_rate": 1.8148631029986964e-05,
"loss": 1.2926,
"step": 208
},
{
"epoch": 0.26,
"learning_rate": 1.8135593220338986e-05,
"loss": 1.3304,
"step": 209
},
{
"epoch": 0.27,
"learning_rate": 1.8122555410691004e-05,
"loss": 1.3566,
"step": 210
},
{
"epoch": 0.27,
"learning_rate": 1.8109517601043026e-05,
"loss": 1.2843,
"step": 211
},
{
"epoch": 0.27,
"learning_rate": 1.8096479791395047e-05,
"loss": 1.3363,
"step": 212
},
{
"epoch": 0.27,
"learning_rate": 1.808344198174707e-05,
"loss": 1.3334,
"step": 213
},
{
"epoch": 0.27,
"learning_rate": 1.807040417209909e-05,
"loss": 1.3555,
"step": 214
},
{
"epoch": 0.27,
"learning_rate": 1.805736636245111e-05,
"loss": 1.2939,
"step": 215
},
{
"epoch": 0.27,
"learning_rate": 1.804432855280313e-05,
"loss": 1.3147,
"step": 216
},
{
"epoch": 0.27,
"learning_rate": 1.803129074315515e-05,
"loss": 1.2948,
"step": 217
},
{
"epoch": 0.28,
"learning_rate": 1.8018252933507173e-05,
"loss": 1.3479,
"step": 218
},
{
"epoch": 0.28,
"learning_rate": 1.800521512385919e-05,
"loss": 1.3534,
"step": 219
},
{
"epoch": 0.28,
"learning_rate": 1.7992177314211213e-05,
"loss": 1.3074,
"step": 220
},
{
"epoch": 0.28,
"learning_rate": 1.7979139504563234e-05,
"loss": 1.3268,
"step": 221
},
{
"epoch": 0.28,
"learning_rate": 1.7966101694915256e-05,
"loss": 1.3584,
"step": 222
},
{
"epoch": 0.28,
"learning_rate": 1.7953063885267277e-05,
"loss": 1.33,
"step": 223
},
{
"epoch": 0.28,
"learning_rate": 1.7940026075619295e-05,
"loss": 1.3715,
"step": 224
},
{
"epoch": 0.28,
"learning_rate": 1.792698826597132e-05,
"loss": 1.3215,
"step": 225
},
{
"epoch": 0.29,
"learning_rate": 1.791395045632334e-05,
"loss": 1.3071,
"step": 226
},
{
"epoch": 0.29,
"learning_rate": 1.790091264667536e-05,
"loss": 1.3423,
"step": 227
},
{
"epoch": 0.29,
"learning_rate": 1.788787483702738e-05,
"loss": 1.3315,
"step": 228
},
{
"epoch": 0.29,
"learning_rate": 1.78748370273794e-05,
"loss": 1.3218,
"step": 229
},
{
"epoch": 0.29,
"learning_rate": 1.7861799217731425e-05,
"loss": 1.3346,
"step": 230
},
{
"epoch": 0.29,
"learning_rate": 1.7848761408083443e-05,
"loss": 1.2703,
"step": 231
},
{
"epoch": 0.29,
"learning_rate": 1.7835723598435464e-05,
"loss": 1.3326,
"step": 232
},
{
"epoch": 0.29,
"learning_rate": 1.7822685788787486e-05,
"loss": 1.3458,
"step": 233
},
{
"epoch": 0.3,
"learning_rate": 1.7809647979139507e-05,
"loss": 1.3073,
"step": 234
},
{
"epoch": 0.3,
"learning_rate": 1.7796610169491526e-05,
"loss": 1.3069,
"step": 235
},
{
"epoch": 0.3,
"learning_rate": 1.7783572359843547e-05,
"loss": 1.3293,
"step": 236
},
{
"epoch": 0.3,
"learning_rate": 1.777053455019557e-05,
"loss": 1.3202,
"step": 237
},
{
"epoch": 0.3,
"learning_rate": 1.7757496740547587e-05,
"loss": 1.3181,
"step": 238
},
{
"epoch": 0.3,
"learning_rate": 1.774445893089961e-05,
"loss": 1.3342,
"step": 239
},
{
"epoch": 0.3,
"learning_rate": 1.773142112125163e-05,
"loss": 1.2949,
"step": 240
},
{
"epoch": 0.3,
"learning_rate": 1.771838331160365e-05,
"loss": 1.3351,
"step": 241
},
{
"epoch": 0.31,
"learning_rate": 1.7705345501955673e-05,
"loss": 1.3513,
"step": 242
},
{
"epoch": 0.31,
"learning_rate": 1.7692307692307694e-05,
"loss": 1.2994,
"step": 243
},
{
"epoch": 0.31,
"learning_rate": 1.7679269882659716e-05,
"loss": 1.3175,
"step": 244
},
{
"epoch": 0.31,
"learning_rate": 1.7666232073011734e-05,
"loss": 1.2964,
"step": 245
},
{
"epoch": 0.31,
"learning_rate": 1.7653194263363756e-05,
"loss": 1.3344,
"step": 246
},
{
"epoch": 0.31,
"learning_rate": 1.7640156453715777e-05,
"loss": 1.318,
"step": 247
},
{
"epoch": 0.31,
"learning_rate": 1.76271186440678e-05,
"loss": 1.3169,
"step": 248
},
{
"epoch": 0.31,
"learning_rate": 1.761408083441982e-05,
"loss": 1.2988,
"step": 249
},
{
"epoch": 0.32,
"learning_rate": 1.760104302477184e-05,
"loss": 1.363,
"step": 250
},
{
"epoch": 0.32,
"learning_rate": 1.758800521512386e-05,
"loss": 1.3074,
"step": 251
},
{
"epoch": 0.32,
"learning_rate": 1.757496740547588e-05,
"loss": 1.3013,
"step": 252
},
{
"epoch": 0.32,
"learning_rate": 1.7561929595827903e-05,
"loss": 1.2819,
"step": 253
},
{
"epoch": 0.32,
"learning_rate": 1.754889178617992e-05,
"loss": 1.3164,
"step": 254
},
{
"epoch": 0.32,
"learning_rate": 1.7535853976531946e-05,
"loss": 1.3016,
"step": 255
},
{
"epoch": 0.32,
"learning_rate": 1.7522816166883964e-05,
"loss": 1.3185,
"step": 256
},
{
"epoch": 0.32,
"learning_rate": 1.7509778357235986e-05,
"loss": 1.3228,
"step": 257
},
{
"epoch": 0.33,
"learning_rate": 1.7496740547588007e-05,
"loss": 1.3274,
"step": 258
},
{
"epoch": 0.33,
"learning_rate": 1.7483702737940025e-05,
"loss": 1.2992,
"step": 259
},
{
"epoch": 0.33,
"learning_rate": 1.747066492829205e-05,
"loss": 1.3154,
"step": 260
},
{
"epoch": 0.33,
"learning_rate": 1.745762711864407e-05,
"loss": 1.2939,
"step": 261
},
{
"epoch": 0.33,
"learning_rate": 1.744458930899609e-05,
"loss": 1.3522,
"step": 262
},
{
"epoch": 0.33,
"learning_rate": 1.743155149934811e-05,
"loss": 1.3123,
"step": 263
},
{
"epoch": 0.33,
"learning_rate": 1.7418513689700133e-05,
"loss": 1.349,
"step": 264
},
{
"epoch": 0.34,
"learning_rate": 1.740547588005215e-05,
"loss": 1.3293,
"step": 265
},
{
"epoch": 0.34,
"learning_rate": 1.7392438070404173e-05,
"loss": 1.2897,
"step": 266
},
{
"epoch": 0.34,
"learning_rate": 1.7379400260756194e-05,
"loss": 1.3026,
"step": 267
},
{
"epoch": 0.34,
"learning_rate": 1.7366362451108216e-05,
"loss": 1.3231,
"step": 268
},
{
"epoch": 0.34,
"learning_rate": 1.7353324641460237e-05,
"loss": 1.3545,
"step": 269
},
{
"epoch": 0.34,
"learning_rate": 1.7340286831812255e-05,
"loss": 1.3335,
"step": 270
},
{
"epoch": 0.34,
"learning_rate": 1.7327249022164277e-05,
"loss": 1.2864,
"step": 271
},
{
"epoch": 0.34,
"learning_rate": 1.73142112125163e-05,
"loss": 1.3365,
"step": 272
},
{
"epoch": 0.35,
"learning_rate": 1.730117340286832e-05,
"loss": 1.3276,
"step": 273
},
{
"epoch": 0.35,
"learning_rate": 1.728813559322034e-05,
"loss": 1.2736,
"step": 274
},
{
"epoch": 0.35,
"learning_rate": 1.727509778357236e-05,
"loss": 1.3193,
"step": 275
},
{
"epoch": 0.35,
"learning_rate": 1.726205997392438e-05,
"loss": 1.3533,
"step": 276
},
{
"epoch": 0.35,
"learning_rate": 1.7249022164276403e-05,
"loss": 1.319,
"step": 277
},
{
"epoch": 0.35,
"learning_rate": 1.7235984354628424e-05,
"loss": 1.3008,
"step": 278
},
{
"epoch": 0.35,
"learning_rate": 1.7222946544980446e-05,
"loss": 1.2969,
"step": 279
},
{
"epoch": 0.35,
"learning_rate": 1.7209908735332464e-05,
"loss": 1.3213,
"step": 280
},
{
"epoch": 0.36,
"learning_rate": 1.7196870925684486e-05,
"loss": 1.2974,
"step": 281
},
{
"epoch": 0.36,
"learning_rate": 1.7183833116036507e-05,
"loss": 1.325,
"step": 282
},
{
"epoch": 0.36,
"learning_rate": 1.717079530638853e-05,
"loss": 1.3225,
"step": 283
},
{
"epoch": 0.36,
"learning_rate": 1.7157757496740547e-05,
"loss": 1.3142,
"step": 284
},
{
"epoch": 0.36,
"learning_rate": 1.714471968709257e-05,
"loss": 1.3547,
"step": 285
},
{
"epoch": 0.36,
"learning_rate": 1.713168187744459e-05,
"loss": 1.2861,
"step": 286
},
{
"epoch": 0.36,
"learning_rate": 1.711864406779661e-05,
"loss": 1.3414,
"step": 287
},
{
"epoch": 0.36,
"learning_rate": 1.7105606258148633e-05,
"loss": 1.2943,
"step": 288
},
{
"epoch": 0.37,
"learning_rate": 1.709256844850065e-05,
"loss": 1.3063,
"step": 289
},
{
"epoch": 0.37,
"learning_rate": 1.7079530638852676e-05,
"loss": 1.3047,
"step": 290
},
{
"epoch": 0.37,
"learning_rate": 1.7066492829204694e-05,
"loss": 1.2827,
"step": 291
},
{
"epoch": 0.37,
"learning_rate": 1.7053455019556716e-05,
"loss": 1.3425,
"step": 292
},
{
"epoch": 0.37,
"learning_rate": 1.7040417209908737e-05,
"loss": 1.3091,
"step": 293
},
{
"epoch": 0.37,
"learning_rate": 1.702737940026076e-05,
"loss": 1.3136,
"step": 294
},
{
"epoch": 0.37,
"learning_rate": 1.7014341590612777e-05,
"loss": 1.3516,
"step": 295
},
{
"epoch": 0.37,
"learning_rate": 1.70013037809648e-05,
"loss": 1.3174,
"step": 296
},
{
"epoch": 0.38,
"learning_rate": 1.698826597131682e-05,
"loss": 1.2878,
"step": 297
},
{
"epoch": 0.38,
"learning_rate": 1.697522816166884e-05,
"loss": 1.2913,
"step": 298
},
{
"epoch": 0.38,
"learning_rate": 1.6962190352020863e-05,
"loss": 1.3191,
"step": 299
},
{
"epoch": 0.38,
"learning_rate": 1.694915254237288e-05,
"loss": 1.3154,
"step": 300
},
{
"epoch": 0.38,
"learning_rate": 1.6936114732724903e-05,
"loss": 1.3379,
"step": 301
},
{
"epoch": 0.38,
"learning_rate": 1.6923076923076924e-05,
"loss": 1.3069,
"step": 302
},
{
"epoch": 0.38,
"learning_rate": 1.6910039113428946e-05,
"loss": 1.2919,
"step": 303
},
{
"epoch": 0.38,
"learning_rate": 1.6897001303780967e-05,
"loss": 1.3109,
"step": 304
},
{
"epoch": 0.39,
"learning_rate": 1.6883963494132985e-05,
"loss": 1.3032,
"step": 305
},
{
"epoch": 0.39,
"learning_rate": 1.687092568448501e-05,
"loss": 1.3502,
"step": 306
},
{
"epoch": 0.39,
"learning_rate": 1.685788787483703e-05,
"loss": 1.3052,
"step": 307
},
{
"epoch": 0.39,
"learning_rate": 1.684485006518905e-05,
"loss": 1.2963,
"step": 308
},
{
"epoch": 0.39,
"learning_rate": 1.683181225554107e-05,
"loss": 1.3191,
"step": 309
},
{
"epoch": 0.39,
"learning_rate": 1.681877444589309e-05,
"loss": 1.3306,
"step": 310
},
{
"epoch": 0.39,
"learning_rate": 1.680573663624511e-05,
"loss": 1.2985,
"step": 311
},
{
"epoch": 0.39,
"learning_rate": 1.6792698826597133e-05,
"loss": 1.2842,
"step": 312
},
{
"epoch": 0.4,
"learning_rate": 1.6779661016949154e-05,
"loss": 1.2462,
"step": 313
},
{
"epoch": 0.4,
"learning_rate": 1.6766623207301176e-05,
"loss": 1.3252,
"step": 314
},
{
"epoch": 0.4,
"learning_rate": 1.6753585397653197e-05,
"loss": 1.3273,
"step": 315
},
{
"epoch": 0.4,
"learning_rate": 1.6740547588005215e-05,
"loss": 1.3298,
"step": 316
},
{
"epoch": 0.4,
"learning_rate": 1.6727509778357237e-05,
"loss": 1.3062,
"step": 317
},
{
"epoch": 0.4,
"learning_rate": 1.671447196870926e-05,
"loss": 1.297,
"step": 318
},
{
"epoch": 0.4,
"learning_rate": 1.6701434159061277e-05,
"loss": 1.2899,
"step": 319
},
{
"epoch": 0.4,
"learning_rate": 1.66883963494133e-05,
"loss": 1.3318,
"step": 320
},
{
"epoch": 0.41,
"learning_rate": 1.667535853976532e-05,
"loss": 1.2843,
"step": 321
},
{
"epoch": 0.41,
"learning_rate": 1.666232073011734e-05,
"loss": 1.3131,
"step": 322
},
{
"epoch": 0.41,
"learning_rate": 1.6649282920469363e-05,
"loss": 1.3368,
"step": 323
},
{
"epoch": 0.41,
"learning_rate": 1.6636245110821384e-05,
"loss": 1.278,
"step": 324
},
{
"epoch": 0.41,
"learning_rate": 1.6623207301173406e-05,
"loss": 1.3033,
"step": 325
},
{
"epoch": 0.41,
"learning_rate": 1.6610169491525424e-05,
"loss": 1.3167,
"step": 326
},
{
"epoch": 0.41,
"learning_rate": 1.6597131681877446e-05,
"loss": 1.308,
"step": 327
},
{
"epoch": 0.41,
"learning_rate": 1.6584093872229467e-05,
"loss": 1.2753,
"step": 328
},
{
"epoch": 0.42,
"learning_rate": 1.657105606258149e-05,
"loss": 1.2986,
"step": 329
},
{
"epoch": 0.42,
"learning_rate": 1.6558018252933507e-05,
"loss": 1.3022,
"step": 330
},
{
"epoch": 0.42,
"learning_rate": 1.654498044328553e-05,
"loss": 1.2979,
"step": 331
},
{
"epoch": 0.42,
"learning_rate": 1.653194263363755e-05,
"loss": 1.3029,
"step": 332
},
{
"epoch": 0.42,
"learning_rate": 1.651890482398957e-05,
"loss": 1.2787,
"step": 333
},
{
"epoch": 0.42,
"learning_rate": 1.6505867014341593e-05,
"loss": 1.2913,
"step": 334
},
{
"epoch": 0.42,
"learning_rate": 1.649282920469361e-05,
"loss": 1.2938,
"step": 335
},
{
"epoch": 0.42,
"learning_rate": 1.6479791395045636e-05,
"loss": 1.2875,
"step": 336
},
{
"epoch": 0.43,
"learning_rate": 1.6466753585397654e-05,
"loss": 1.2859,
"step": 337
},
{
"epoch": 0.43,
"learning_rate": 1.6453715775749676e-05,
"loss": 1.3157,
"step": 338
},
{
"epoch": 0.43,
"learning_rate": 1.6440677966101697e-05,
"loss": 1.3082,
"step": 339
},
{
"epoch": 0.43,
"learning_rate": 1.6427640156453715e-05,
"loss": 1.2784,
"step": 340
},
{
"epoch": 0.43,
"learning_rate": 1.6414602346805737e-05,
"loss": 1.3204,
"step": 341
},
{
"epoch": 0.43,
"learning_rate": 1.640156453715776e-05,
"loss": 1.3344,
"step": 342
},
{
"epoch": 0.43,
"learning_rate": 1.638852672750978e-05,
"loss": 1.2981,
"step": 343
},
{
"epoch": 0.43,
"learning_rate": 1.63754889178618e-05,
"loss": 1.2943,
"step": 344
},
{
"epoch": 0.44,
"learning_rate": 1.6362451108213823e-05,
"loss": 1.2911,
"step": 345
},
{
"epoch": 0.44,
"learning_rate": 1.634941329856584e-05,
"loss": 1.3058,
"step": 346
},
{
"epoch": 0.44,
"learning_rate": 1.6336375488917863e-05,
"loss": 1.2999,
"step": 347
},
{
"epoch": 0.44,
"learning_rate": 1.6323337679269884e-05,
"loss": 1.3075,
"step": 348
},
{
"epoch": 0.44,
"learning_rate": 1.6310299869621902e-05,
"loss": 1.307,
"step": 349
},
{
"epoch": 0.44,
"learning_rate": 1.6297262059973927e-05,
"loss": 1.2651,
"step": 350
},
{
"epoch": 0.44,
"learning_rate": 1.6284224250325945e-05,
"loss": 1.307,
"step": 351
},
{
"epoch": 0.45,
"learning_rate": 1.6271186440677967e-05,
"loss": 1.2931,
"step": 352
},
{
"epoch": 0.45,
"learning_rate": 1.625814863102999e-05,
"loss": 1.3058,
"step": 353
},
{
"epoch": 0.45,
"learning_rate": 1.624511082138201e-05,
"loss": 1.3273,
"step": 354
},
{
"epoch": 0.45,
"learning_rate": 1.623207301173403e-05,
"loss": 1.2512,
"step": 355
},
{
"epoch": 0.45,
"learning_rate": 1.621903520208605e-05,
"loss": 1.2958,
"step": 356
},
{
"epoch": 0.45,
"learning_rate": 1.620599739243807e-05,
"loss": 1.291,
"step": 357
},
{
"epoch": 0.45,
"learning_rate": 1.6192959582790093e-05,
"loss": 1.3268,
"step": 358
},
{
"epoch": 0.45,
"learning_rate": 1.6179921773142114e-05,
"loss": 1.334,
"step": 359
},
{
"epoch": 0.46,
"learning_rate": 1.6166883963494132e-05,
"loss": 1.2966,
"step": 360
},
{
"epoch": 0.46,
"learning_rate": 1.6153846153846154e-05,
"loss": 1.3442,
"step": 361
},
{
"epoch": 0.46,
"learning_rate": 1.6140808344198176e-05,
"loss": 1.2639,
"step": 362
},
{
"epoch": 0.46,
"learning_rate": 1.6127770534550197e-05,
"loss": 1.2777,
"step": 363
},
{
"epoch": 0.46,
"learning_rate": 1.611473272490222e-05,
"loss": 1.288,
"step": 364
},
{
"epoch": 0.46,
"learning_rate": 1.6101694915254237e-05,
"loss": 1.3018,
"step": 365
},
{
"epoch": 0.46,
"learning_rate": 1.608865710560626e-05,
"loss": 1.2993,
"step": 366
},
{
"epoch": 0.46,
"learning_rate": 1.607561929595828e-05,
"loss": 1.2826,
"step": 367
},
{
"epoch": 0.47,
"learning_rate": 1.60625814863103e-05,
"loss": 1.3124,
"step": 368
},
{
"epoch": 0.47,
"learning_rate": 1.6049543676662323e-05,
"loss": 1.3529,
"step": 369
},
{
"epoch": 0.47,
"learning_rate": 1.603650586701434e-05,
"loss": 1.2919,
"step": 370
},
{
"epoch": 0.47,
"learning_rate": 1.6023468057366366e-05,
"loss": 1.3044,
"step": 371
},
{
"epoch": 0.47,
"learning_rate": 1.6010430247718384e-05,
"loss": 1.2424,
"step": 372
},
{
"epoch": 0.47,
"learning_rate": 1.5997392438070406e-05,
"loss": 1.2787,
"step": 373
},
{
"epoch": 0.47,
"learning_rate": 1.5984354628422427e-05,
"loss": 1.2701,
"step": 374
},
{
"epoch": 0.47,
"learning_rate": 1.597131681877445e-05,
"loss": 1.2924,
"step": 375
},
{
"epoch": 0.48,
"learning_rate": 1.5958279009126467e-05,
"loss": 1.2633,
"step": 376
},
{
"epoch": 0.48,
"learning_rate": 1.594524119947849e-05,
"loss": 1.2288,
"step": 377
},
{
"epoch": 0.48,
"learning_rate": 1.593220338983051e-05,
"loss": 1.2877,
"step": 378
},
{
"epoch": 0.48,
"learning_rate": 1.5919165580182528e-05,
"loss": 1.2767,
"step": 379
},
{
"epoch": 0.48,
"learning_rate": 1.5906127770534553e-05,
"loss": 1.2788,
"step": 380
},
{
"epoch": 0.48,
"learning_rate": 1.589308996088657e-05,
"loss": 1.2822,
"step": 381
},
{
"epoch": 0.48,
"learning_rate": 1.5880052151238593e-05,
"loss": 1.3312,
"step": 382
},
{
"epoch": 0.48,
"learning_rate": 1.5867014341590614e-05,
"loss": 1.2997,
"step": 383
},
{
"epoch": 0.49,
"learning_rate": 1.5853976531942636e-05,
"loss": 1.2441,
"step": 384
},
{
"epoch": 0.49,
"learning_rate": 1.5840938722294657e-05,
"loss": 1.3047,
"step": 385
},
{
"epoch": 0.49,
"learning_rate": 1.5827900912646675e-05,
"loss": 1.3101,
"step": 386
},
{
"epoch": 0.49,
"learning_rate": 1.5814863102998697e-05,
"loss": 1.2845,
"step": 387
},
{
"epoch": 0.49,
"learning_rate": 1.580182529335072e-05,
"loss": 1.2832,
"step": 388
},
{
"epoch": 0.49,
"learning_rate": 1.578878748370274e-05,
"loss": 1.3157,
"step": 389
},
{
"epoch": 0.49,
"learning_rate": 1.577574967405476e-05,
"loss": 1.248,
"step": 390
},
{
"epoch": 0.49,
"learning_rate": 1.576271186440678e-05,
"loss": 1.2858,
"step": 391
},
{
"epoch": 0.5,
"learning_rate": 1.57496740547588e-05,
"loss": 1.283,
"step": 392
},
{
"epoch": 0.5,
"learning_rate": 1.5736636245110823e-05,
"loss": 1.3123,
"step": 393
},
{
"epoch": 0.5,
"learning_rate": 1.5723598435462844e-05,
"loss": 1.2941,
"step": 394
},
{
"epoch": 0.5,
"learning_rate": 1.5710560625814862e-05,
"loss": 1.3051,
"step": 395
},
{
"epoch": 0.5,
"learning_rate": 1.5697522816166887e-05,
"loss": 1.3005,
"step": 396
},
{
"epoch": 0.5,
"learning_rate": 1.5684485006518905e-05,
"loss": 1.2645,
"step": 397
},
{
"epoch": 0.5,
"learning_rate": 1.5671447196870927e-05,
"loss": 1.2991,
"step": 398
},
{
"epoch": 0.5,
"learning_rate": 1.565840938722295e-05,
"loss": 1.2974,
"step": 399
},
{
"epoch": 0.51,
"learning_rate": 1.5645371577574967e-05,
"loss": 1.2877,
"step": 400
},
{
"epoch": 0.51,
"learning_rate": 1.563233376792699e-05,
"loss": 1.2832,
"step": 401
},
{
"epoch": 0.51,
"learning_rate": 1.561929595827901e-05,
"loss": 1.2827,
"step": 402
},
{
"epoch": 0.51,
"learning_rate": 1.560625814863103e-05,
"loss": 1.3064,
"step": 403
},
{
"epoch": 0.51,
"learning_rate": 1.5593220338983053e-05,
"loss": 1.3051,
"step": 404
},
{
"epoch": 0.51,
"learning_rate": 1.5580182529335074e-05,
"loss": 1.2562,
"step": 405
},
{
"epoch": 0.51,
"learning_rate": 1.5567144719687092e-05,
"loss": 1.3186,
"step": 406
},
{
"epoch": 0.51,
"learning_rate": 1.5554106910039114e-05,
"loss": 1.2731,
"step": 407
},
{
"epoch": 0.52,
"learning_rate": 1.5541069100391136e-05,
"loss": 1.2955,
"step": 408
},
{
"epoch": 0.52,
"learning_rate": 1.5528031290743157e-05,
"loss": 1.2938,
"step": 409
},
{
"epoch": 0.52,
"learning_rate": 1.551499348109518e-05,
"loss": 1.2505,
"step": 410
},
{
"epoch": 0.52,
"learning_rate": 1.5501955671447197e-05,
"loss": 1.3368,
"step": 411
},
{
"epoch": 0.52,
"learning_rate": 1.5488917861799218e-05,
"loss": 1.2964,
"step": 412
},
{
"epoch": 0.52,
"learning_rate": 1.547588005215124e-05,
"loss": 1.2986,
"step": 413
},
{
"epoch": 0.52,
"learning_rate": 1.546284224250326e-05,
"loss": 1.2694,
"step": 414
},
{
"epoch": 0.52,
"learning_rate": 1.5449804432855283e-05,
"loss": 1.3032,
"step": 415
},
{
"epoch": 0.53,
"learning_rate": 1.54367666232073e-05,
"loss": 1.3099,
"step": 416
},
{
"epoch": 0.53,
"learning_rate": 1.5423728813559326e-05,
"loss": 1.2749,
"step": 417
},
{
"epoch": 0.53,
"learning_rate": 1.5410691003911344e-05,
"loss": 1.3003,
"step": 418
},
{
"epoch": 0.53,
"learning_rate": 1.5397653194263366e-05,
"loss": 1.3204,
"step": 419
},
{
"epoch": 0.53,
"learning_rate": 1.5384615384615387e-05,
"loss": 1.3378,
"step": 420
},
{
"epoch": 0.53,
"learning_rate": 1.5371577574967405e-05,
"loss": 1.2841,
"step": 421
},
{
"epoch": 0.53,
"learning_rate": 1.5358539765319427e-05,
"loss": 1.2957,
"step": 422
},
{
"epoch": 0.53,
"learning_rate": 1.534550195567145e-05,
"loss": 1.2842,
"step": 423
},
{
"epoch": 0.54,
"learning_rate": 1.533246414602347e-05,
"loss": 1.2773,
"step": 424
},
{
"epoch": 0.54,
"learning_rate": 1.5319426336375488e-05,
"loss": 1.3038,
"step": 425
},
{
"epoch": 0.54,
"learning_rate": 1.5306388526727513e-05,
"loss": 1.2687,
"step": 426
},
{
"epoch": 0.54,
"learning_rate": 1.529335071707953e-05,
"loss": 1.3098,
"step": 427
},
{
"epoch": 0.54,
"learning_rate": 1.5280312907431553e-05,
"loss": 1.2704,
"step": 428
},
{
"epoch": 0.54,
"learning_rate": 1.5267275097783574e-05,
"loss": 1.2802,
"step": 429
},
{
"epoch": 0.54,
"learning_rate": 1.5254237288135594e-05,
"loss": 1.2667,
"step": 430
},
{
"epoch": 0.54,
"learning_rate": 1.5241199478487616e-05,
"loss": 1.27,
"step": 431
},
{
"epoch": 0.55,
"learning_rate": 1.5228161668839635e-05,
"loss": 1.2521,
"step": 432
},
{
"epoch": 0.55,
"learning_rate": 1.5215123859191657e-05,
"loss": 1.2719,
"step": 433
},
{
"epoch": 0.55,
"learning_rate": 1.5202086049543678e-05,
"loss": 1.3188,
"step": 434
},
{
"epoch": 0.55,
"learning_rate": 1.5189048239895698e-05,
"loss": 1.2958,
"step": 435
},
{
"epoch": 0.55,
"learning_rate": 1.517601043024772e-05,
"loss": 1.271,
"step": 436
},
{
"epoch": 0.55,
"learning_rate": 1.5162972620599741e-05,
"loss": 1.2307,
"step": 437
},
{
"epoch": 0.55,
"learning_rate": 1.5149934810951761e-05,
"loss": 1.2925,
"step": 438
},
{
"epoch": 0.55,
"learning_rate": 1.5136897001303783e-05,
"loss": 1.2579,
"step": 439
},
{
"epoch": 0.56,
"learning_rate": 1.5123859191655803e-05,
"loss": 1.3206,
"step": 440
},
{
"epoch": 0.56,
"learning_rate": 1.5110821382007822e-05,
"loss": 1.3112,
"step": 441
},
{
"epoch": 0.56,
"learning_rate": 1.5097783572359846e-05,
"loss": 1.2642,
"step": 442
},
{
"epoch": 0.56,
"learning_rate": 1.5084745762711865e-05,
"loss": 1.3029,
"step": 443
},
{
"epoch": 0.56,
"learning_rate": 1.5071707953063885e-05,
"loss": 1.2833,
"step": 444
},
{
"epoch": 0.56,
"learning_rate": 1.5058670143415907e-05,
"loss": 1.2645,
"step": 445
},
{
"epoch": 0.56,
"learning_rate": 1.5045632333767928e-05,
"loss": 1.293,
"step": 446
},
{
"epoch": 0.57,
"learning_rate": 1.503259452411995e-05,
"loss": 1.3011,
"step": 447
},
{
"epoch": 0.57,
"learning_rate": 1.501955671447197e-05,
"loss": 1.2661,
"step": 448
},
{
"epoch": 0.57,
"learning_rate": 1.500651890482399e-05,
"loss": 1.2931,
"step": 449
},
{
"epoch": 0.57,
"learning_rate": 1.4993481095176013e-05,
"loss": 1.2683,
"step": 450
},
{
"epoch": 0.57,
"learning_rate": 1.4980443285528033e-05,
"loss": 1.2996,
"step": 451
},
{
"epoch": 0.57,
"learning_rate": 1.4967405475880053e-05,
"loss": 1.3036,
"step": 452
},
{
"epoch": 0.57,
"learning_rate": 1.4954367666232074e-05,
"loss": 1.2673,
"step": 453
},
{
"epoch": 0.57,
"learning_rate": 1.4941329856584096e-05,
"loss": 1.2502,
"step": 454
},
{
"epoch": 0.58,
"learning_rate": 1.4928292046936117e-05,
"loss": 1.2681,
"step": 455
},
{
"epoch": 0.58,
"learning_rate": 1.4915254237288137e-05,
"loss": 1.3107,
"step": 456
},
{
"epoch": 0.58,
"learning_rate": 1.4902216427640157e-05,
"loss": 1.2971,
"step": 457
},
{
"epoch": 0.58,
"learning_rate": 1.488917861799218e-05,
"loss": 1.3087,
"step": 458
},
{
"epoch": 0.58,
"learning_rate": 1.48761408083442e-05,
"loss": 1.281,
"step": 459
},
{
"epoch": 0.58,
"learning_rate": 1.486310299869622e-05,
"loss": 1.2891,
"step": 460
},
{
"epoch": 0.58,
"learning_rate": 1.4850065189048241e-05,
"loss": 1.2732,
"step": 461
},
{
"epoch": 0.58,
"learning_rate": 1.4837027379400261e-05,
"loss": 1.3044,
"step": 462
},
{
"epoch": 0.59,
"learning_rate": 1.4823989569752283e-05,
"loss": 1.2866,
"step": 463
},
{
"epoch": 0.59,
"learning_rate": 1.4810951760104304e-05,
"loss": 1.3029,
"step": 464
},
{
"epoch": 0.59,
"learning_rate": 1.4797913950456324e-05,
"loss": 1.2736,
"step": 465
},
{
"epoch": 0.59,
"learning_rate": 1.4784876140808346e-05,
"loss": 1.2585,
"step": 466
},
{
"epoch": 0.59,
"learning_rate": 1.4771838331160367e-05,
"loss": 1.2804,
"step": 467
},
{
"epoch": 0.59,
"learning_rate": 1.4758800521512387e-05,
"loss": 1.2877,
"step": 468
},
{
"epoch": 0.59,
"learning_rate": 1.4745762711864408e-05,
"loss": 1.2478,
"step": 469
},
{
"epoch": 0.59,
"learning_rate": 1.4732724902216428e-05,
"loss": 1.2736,
"step": 470
},
{
"epoch": 0.6,
"learning_rate": 1.4719687092568448e-05,
"loss": 1.2902,
"step": 471
},
{
"epoch": 0.6,
"learning_rate": 1.4706649282920471e-05,
"loss": 1.2788,
"step": 472
},
{
"epoch": 0.6,
"learning_rate": 1.4693611473272491e-05,
"loss": 1.2657,
"step": 473
},
{
"epoch": 0.6,
"learning_rate": 1.4680573663624513e-05,
"loss": 1.2828,
"step": 474
},
{
"epoch": 0.6,
"learning_rate": 1.4667535853976533e-05,
"loss": 1.2885,
"step": 475
},
{
"epoch": 0.6,
"learning_rate": 1.4654498044328554e-05,
"loss": 1.2744,
"step": 476
},
{
"epoch": 0.6,
"learning_rate": 1.4641460234680576e-05,
"loss": 1.2737,
"step": 477
},
{
"epoch": 0.6,
"learning_rate": 1.4628422425032595e-05,
"loss": 1.2311,
"step": 478
},
{
"epoch": 0.61,
"learning_rate": 1.4615384615384615e-05,
"loss": 1.2834,
"step": 479
},
{
"epoch": 0.61,
"learning_rate": 1.4602346805736639e-05,
"loss": 1.2615,
"step": 480
},
{
"epoch": 0.61,
"learning_rate": 1.4589308996088658e-05,
"loss": 1.2748,
"step": 481
},
{
"epoch": 0.61,
"learning_rate": 1.4576271186440678e-05,
"loss": 1.3031,
"step": 482
},
{
"epoch": 0.61,
"learning_rate": 1.45632333767927e-05,
"loss": 1.3126,
"step": 483
},
{
"epoch": 0.61,
"learning_rate": 1.455019556714472e-05,
"loss": 1.249,
"step": 484
},
{
"epoch": 0.61,
"learning_rate": 1.4537157757496743e-05,
"loss": 1.2816,
"step": 485
},
{
"epoch": 0.61,
"learning_rate": 1.4524119947848763e-05,
"loss": 1.3064,
"step": 486
},
{
"epoch": 0.62,
"learning_rate": 1.4511082138200782e-05,
"loss": 1.299,
"step": 487
},
{
"epoch": 0.62,
"learning_rate": 1.4498044328552806e-05,
"loss": 1.2998,
"step": 488
},
{
"epoch": 0.62,
"learning_rate": 1.4485006518904826e-05,
"loss": 1.2906,
"step": 489
},
{
"epoch": 0.62,
"learning_rate": 1.4471968709256845e-05,
"loss": 1.2656,
"step": 490
},
{
"epoch": 0.62,
"learning_rate": 1.4458930899608867e-05,
"loss": 1.2729,
"step": 491
},
{
"epoch": 0.62,
"learning_rate": 1.4445893089960887e-05,
"loss": 1.2892,
"step": 492
},
{
"epoch": 0.62,
"learning_rate": 1.443285528031291e-05,
"loss": 1.2692,
"step": 493
},
{
"epoch": 0.62,
"learning_rate": 1.441981747066493e-05,
"loss": 1.288,
"step": 494
},
{
"epoch": 0.63,
"learning_rate": 1.440677966101695e-05,
"loss": 1.2915,
"step": 495
},
{
"epoch": 0.63,
"learning_rate": 1.4393741851368971e-05,
"loss": 1.2751,
"step": 496
},
{
"epoch": 0.63,
"learning_rate": 1.4380704041720993e-05,
"loss": 1.285,
"step": 497
},
{
"epoch": 0.63,
"learning_rate": 1.4367666232073013e-05,
"loss": 1.2832,
"step": 498
},
{
"epoch": 0.63,
"learning_rate": 1.4354628422425034e-05,
"loss": 1.2645,
"step": 499
},
{
"epoch": 0.63,
"learning_rate": 1.4341590612777054e-05,
"loss": 1.2577,
"step": 500
},
{
"epoch": 0.63,
"learning_rate": 1.4328552803129077e-05,
"loss": 1.252,
"step": 501
},
{
"epoch": 0.63,
"learning_rate": 1.4315514993481097e-05,
"loss": 1.236,
"step": 502
},
{
"epoch": 0.64,
"learning_rate": 1.4302477183833117e-05,
"loss": 1.2863,
"step": 503
},
{
"epoch": 0.64,
"learning_rate": 1.4289439374185138e-05,
"loss": 1.2549,
"step": 504
},
{
"epoch": 0.64,
"learning_rate": 1.4276401564537158e-05,
"loss": 1.3345,
"step": 505
},
{
"epoch": 0.64,
"learning_rate": 1.426336375488918e-05,
"loss": 1.269,
"step": 506
},
{
"epoch": 0.64,
"learning_rate": 1.4250325945241201e-05,
"loss": 1.2834,
"step": 507
},
{
"epoch": 0.64,
"learning_rate": 1.4237288135593221e-05,
"loss": 1.2853,
"step": 508
},
{
"epoch": 0.64,
"learning_rate": 1.4224250325945241e-05,
"loss": 1.2363,
"step": 509
},
{
"epoch": 0.64,
"learning_rate": 1.4211212516297264e-05,
"loss": 1.2842,
"step": 510
},
{
"epoch": 0.65,
"learning_rate": 1.4198174706649284e-05,
"loss": 1.2705,
"step": 511
},
{
"epoch": 0.65,
"learning_rate": 1.4185136897001306e-05,
"loss": 1.2667,
"step": 512
},
{
"epoch": 0.65,
"learning_rate": 1.4172099087353325e-05,
"loss": 1.2555,
"step": 513
},
{
"epoch": 0.65,
"learning_rate": 1.4159061277705345e-05,
"loss": 1.2753,
"step": 514
},
{
"epoch": 0.65,
"learning_rate": 1.4146023468057368e-05,
"loss": 1.3025,
"step": 515
},
{
"epoch": 0.65,
"learning_rate": 1.4132985658409388e-05,
"loss": 1.2666,
"step": 516
},
{
"epoch": 0.65,
"learning_rate": 1.4119947848761408e-05,
"loss": 1.2577,
"step": 517
},
{
"epoch": 0.65,
"learning_rate": 1.4106910039113431e-05,
"loss": 1.2736,
"step": 518
},
{
"epoch": 0.66,
"learning_rate": 1.4093872229465451e-05,
"loss": 1.2861,
"step": 519
},
{
"epoch": 0.66,
"learning_rate": 1.4080834419817473e-05,
"loss": 1.2975,
"step": 520
},
{
"epoch": 0.66,
"learning_rate": 1.4067796610169493e-05,
"loss": 1.2819,
"step": 521
},
{
"epoch": 0.66,
"learning_rate": 1.4054758800521512e-05,
"loss": 1.282,
"step": 522
},
{
"epoch": 0.66,
"learning_rate": 1.4041720990873536e-05,
"loss": 1.2714,
"step": 523
},
{
"epoch": 0.66,
"learning_rate": 1.4028683181225555e-05,
"loss": 1.2793,
"step": 524
},
{
"epoch": 0.66,
"learning_rate": 1.4015645371577575e-05,
"loss": 1.2814,
"step": 525
},
{
"epoch": 0.66,
"learning_rate": 1.4002607561929597e-05,
"loss": 1.3046,
"step": 526
},
{
"epoch": 0.67,
"learning_rate": 1.3989569752281618e-05,
"loss": 1.244,
"step": 527
},
{
"epoch": 0.67,
"learning_rate": 1.3976531942633638e-05,
"loss": 1.2451,
"step": 528
},
{
"epoch": 0.67,
"learning_rate": 1.396349413298566e-05,
"loss": 1.2856,
"step": 529
},
{
"epoch": 0.67,
"learning_rate": 1.395045632333768e-05,
"loss": 1.2764,
"step": 530
},
{
"epoch": 0.67,
"learning_rate": 1.3937418513689703e-05,
"loss": 1.2517,
"step": 531
},
{
"epoch": 0.67,
"learning_rate": 1.3924380704041723e-05,
"loss": 1.2867,
"step": 532
},
{
"epoch": 0.67,
"learning_rate": 1.3911342894393742e-05,
"loss": 1.245,
"step": 533
},
{
"epoch": 0.68,
"learning_rate": 1.3898305084745764e-05,
"loss": 1.2917,
"step": 534
},
{
"epoch": 0.68,
"learning_rate": 1.3885267275097784e-05,
"loss": 1.2524,
"step": 535
},
{
"epoch": 0.68,
"learning_rate": 1.3872229465449805e-05,
"loss": 1.2679,
"step": 536
},
{
"epoch": 0.68,
"learning_rate": 1.3859191655801827e-05,
"loss": 1.3221,
"step": 537
},
{
"epoch": 0.68,
"learning_rate": 1.3846153846153847e-05,
"loss": 1.2793,
"step": 538
},
{
"epoch": 0.68,
"learning_rate": 1.383311603650587e-05,
"loss": 1.2444,
"step": 539
},
{
"epoch": 0.68,
"learning_rate": 1.382007822685789e-05,
"loss": 1.2708,
"step": 540
},
{
"epoch": 0.68,
"learning_rate": 1.380704041720991e-05,
"loss": 1.2523,
"step": 541
},
{
"epoch": 0.69,
"learning_rate": 1.3794002607561931e-05,
"loss": 1.2609,
"step": 542
},
{
"epoch": 0.69,
"learning_rate": 1.3780964797913951e-05,
"loss": 1.2598,
"step": 543
},
{
"epoch": 0.69,
"learning_rate": 1.3767926988265971e-05,
"loss": 1.2975,
"step": 544
},
{
"epoch": 0.69,
"learning_rate": 1.3754889178617994e-05,
"loss": 1.2717,
"step": 545
},
{
"epoch": 0.69,
"learning_rate": 1.3741851368970014e-05,
"loss": 1.2833,
"step": 546
},
{
"epoch": 0.69,
"learning_rate": 1.3728813559322034e-05,
"loss": 1.2526,
"step": 547
},
{
"epoch": 0.69,
"learning_rate": 1.3715775749674057e-05,
"loss": 1.2871,
"step": 548
},
{
"epoch": 0.69,
"learning_rate": 1.3702737940026077e-05,
"loss": 1.2983,
"step": 549
},
{
"epoch": 0.7,
"learning_rate": 1.3689700130378098e-05,
"loss": 1.2928,
"step": 550
},
{
"epoch": 0.7,
"learning_rate": 1.3676662320730118e-05,
"loss": 1.2762,
"step": 551
},
{
"epoch": 0.7,
"learning_rate": 1.3663624511082138e-05,
"loss": 1.2646,
"step": 552
},
{
"epoch": 0.7,
"learning_rate": 1.3650586701434161e-05,
"loss": 1.2729,
"step": 553
},
{
"epoch": 0.7,
"learning_rate": 1.3637548891786181e-05,
"loss": 1.3102,
"step": 554
},
{
"epoch": 0.7,
"learning_rate": 1.3624511082138201e-05,
"loss": 1.3051,
"step": 555
},
{
"epoch": 0.7,
"learning_rate": 1.3611473272490223e-05,
"loss": 1.2805,
"step": 556
},
{
"epoch": 0.7,
"learning_rate": 1.3598435462842244e-05,
"loss": 1.2861,
"step": 557
},
{
"epoch": 0.71,
"learning_rate": 1.3585397653194266e-05,
"loss": 1.2333,
"step": 558
},
{
"epoch": 0.71,
"learning_rate": 1.3572359843546285e-05,
"loss": 1.2615,
"step": 559
},
{
"epoch": 0.71,
"learning_rate": 1.3559322033898305e-05,
"loss": 1.2748,
"step": 560
},
{
"epoch": 0.71,
"learning_rate": 1.3546284224250328e-05,
"loss": 1.2869,
"step": 561
},
{
"epoch": 0.71,
"learning_rate": 1.3533246414602348e-05,
"loss": 1.2874,
"step": 562
},
{
"epoch": 0.71,
"learning_rate": 1.3520208604954368e-05,
"loss": 1.2952,
"step": 563
},
{
"epoch": 0.71,
"learning_rate": 1.350717079530639e-05,
"loss": 1.2576,
"step": 564
},
{
"epoch": 0.71,
"learning_rate": 1.349413298565841e-05,
"loss": 1.295,
"step": 565
},
{
"epoch": 0.72,
"learning_rate": 1.3481095176010431e-05,
"loss": 1.2789,
"step": 566
},
{
"epoch": 0.72,
"learning_rate": 1.3468057366362453e-05,
"loss": 1.2804,
"step": 567
},
{
"epoch": 0.72,
"learning_rate": 1.3455019556714472e-05,
"loss": 1.2206,
"step": 568
},
{
"epoch": 0.72,
"learning_rate": 1.3441981747066496e-05,
"loss": 1.2588,
"step": 569
},
{
"epoch": 0.72,
"learning_rate": 1.3428943937418515e-05,
"loss": 1.2982,
"step": 570
},
{
"epoch": 0.72,
"learning_rate": 1.3415906127770535e-05,
"loss": 1.3011,
"step": 571
},
{
"epoch": 0.72,
"learning_rate": 1.3402868318122557e-05,
"loss": 1.2406,
"step": 572
},
{
"epoch": 0.72,
"learning_rate": 1.3389830508474577e-05,
"loss": 1.2391,
"step": 573
},
{
"epoch": 0.73,
"learning_rate": 1.3376792698826597e-05,
"loss": 1.2511,
"step": 574
},
{
"epoch": 0.73,
"learning_rate": 1.336375488917862e-05,
"loss": 1.2675,
"step": 575
},
{
"epoch": 0.73,
"learning_rate": 1.335071707953064e-05,
"loss": 1.2623,
"step": 576
},
{
"epoch": 0.73,
"learning_rate": 1.3337679269882661e-05,
"loss": 1.2775,
"step": 577
},
{
"epoch": 0.73,
"learning_rate": 1.3324641460234683e-05,
"loss": 1.2499,
"step": 578
},
{
"epoch": 0.73,
"learning_rate": 1.3311603650586703e-05,
"loss": 1.2494,
"step": 579
},
{
"epoch": 0.73,
"learning_rate": 1.3298565840938724e-05,
"loss": 1.2524,
"step": 580
},
{
"epoch": 0.73,
"learning_rate": 1.3285528031290744e-05,
"loss": 1.2797,
"step": 581
},
{
"epoch": 0.74,
"learning_rate": 1.3272490221642764e-05,
"loss": 1.282,
"step": 582
},
{
"epoch": 0.74,
"learning_rate": 1.3259452411994787e-05,
"loss": 1.2408,
"step": 583
},
{
"epoch": 0.74,
"learning_rate": 1.3246414602346807e-05,
"loss": 1.2726,
"step": 584
},
{
"epoch": 0.74,
"learning_rate": 1.3233376792698827e-05,
"loss": 1.2561,
"step": 585
},
{
"epoch": 0.74,
"learning_rate": 1.3220338983050848e-05,
"loss": 1.2661,
"step": 586
},
{
"epoch": 0.74,
"learning_rate": 1.320730117340287e-05,
"loss": 1.2561,
"step": 587
},
{
"epoch": 0.74,
"learning_rate": 1.3194263363754891e-05,
"loss": 1.243,
"step": 588
},
{
"epoch": 0.74,
"learning_rate": 1.3181225554106911e-05,
"loss": 1.2936,
"step": 589
},
{
"epoch": 0.75,
"learning_rate": 1.3168187744458931e-05,
"loss": 1.2764,
"step": 590
},
{
"epoch": 0.75,
"learning_rate": 1.3155149934810954e-05,
"loss": 1.2767,
"step": 591
},
{
"epoch": 0.75,
"learning_rate": 1.3142112125162974e-05,
"loss": 1.2394,
"step": 592
},
{
"epoch": 0.75,
"learning_rate": 1.3129074315514994e-05,
"loss": 1.2588,
"step": 593
},
{
"epoch": 0.75,
"learning_rate": 1.3116036505867015e-05,
"loss": 1.2523,
"step": 594
},
{
"epoch": 0.75,
"learning_rate": 1.3102998696219035e-05,
"loss": 1.2491,
"step": 595
},
{
"epoch": 0.75,
"learning_rate": 1.3089960886571058e-05,
"loss": 1.2787,
"step": 596
},
{
"epoch": 0.75,
"learning_rate": 1.3076923076923078e-05,
"loss": 1.2155,
"step": 597
},
{
"epoch": 0.76,
"learning_rate": 1.3063885267275098e-05,
"loss": 1.2714,
"step": 598
},
{
"epoch": 0.76,
"learning_rate": 1.305084745762712e-05,
"loss": 1.2273,
"step": 599
},
{
"epoch": 0.76,
"learning_rate": 1.3037809647979141e-05,
"loss": 1.2542,
"step": 600
},
{
"epoch": 0.76,
"learning_rate": 1.3024771838331161e-05,
"loss": 1.2642,
"step": 601
},
{
"epoch": 0.76,
"learning_rate": 1.3011734028683183e-05,
"loss": 1.2722,
"step": 602
},
{
"epoch": 0.76,
"learning_rate": 1.2998696219035202e-05,
"loss": 1.2264,
"step": 603
},
{
"epoch": 0.76,
"learning_rate": 1.2985658409387226e-05,
"loss": 1.2885,
"step": 604
},
{
"epoch": 0.76,
"learning_rate": 1.2972620599739245e-05,
"loss": 1.2732,
"step": 605
},
{
"epoch": 0.77,
"learning_rate": 1.2959582790091265e-05,
"loss": 1.2704,
"step": 606
},
{
"epoch": 0.77,
"learning_rate": 1.2946544980443287e-05,
"loss": 1.2737,
"step": 607
},
{
"epoch": 0.77,
"learning_rate": 1.2933507170795307e-05,
"loss": 1.2621,
"step": 608
},
{
"epoch": 0.77,
"learning_rate": 1.2920469361147328e-05,
"loss": 1.2648,
"step": 609
},
{
"epoch": 0.77,
"learning_rate": 1.290743155149935e-05,
"loss": 1.2455,
"step": 610
},
{
"epoch": 0.77,
"learning_rate": 1.289439374185137e-05,
"loss": 1.2411,
"step": 611
},
{
"epoch": 0.77,
"learning_rate": 1.288135593220339e-05,
"loss": 1.2219,
"step": 612
},
{
"epoch": 0.77,
"learning_rate": 1.2868318122555413e-05,
"loss": 1.266,
"step": 613
},
{
"epoch": 0.78,
"learning_rate": 1.2855280312907432e-05,
"loss": 1.2748,
"step": 614
},
{
"epoch": 0.78,
"learning_rate": 1.2842242503259454e-05,
"loss": 1.257,
"step": 615
},
{
"epoch": 0.78,
"learning_rate": 1.2829204693611474e-05,
"loss": 1.2866,
"step": 616
},
{
"epoch": 0.78,
"learning_rate": 1.2816166883963494e-05,
"loss": 1.2799,
"step": 617
},
{
"epoch": 0.78,
"learning_rate": 1.2803129074315517e-05,
"loss": 1.2476,
"step": 618
},
{
"epoch": 0.78,
"learning_rate": 1.2790091264667537e-05,
"loss": 1.2384,
"step": 619
},
{
"epoch": 0.78,
"learning_rate": 1.2777053455019557e-05,
"loss": 1.2716,
"step": 620
},
{
"epoch": 0.79,
"learning_rate": 1.276401564537158e-05,
"loss": 1.2717,
"step": 621
},
{
"epoch": 0.79,
"learning_rate": 1.27509778357236e-05,
"loss": 1.235,
"step": 622
},
{
"epoch": 0.79,
"learning_rate": 1.2737940026075621e-05,
"loss": 1.2544,
"step": 623
},
{
"epoch": 0.79,
"learning_rate": 1.2724902216427641e-05,
"loss": 1.218,
"step": 624
},
{
"epoch": 0.79,
"learning_rate": 1.2711864406779661e-05,
"loss": 1.276,
"step": 625
},
{
"epoch": 0.79,
"learning_rate": 1.2698826597131684e-05,
"loss": 1.2299,
"step": 626
},
{
"epoch": 0.79,
"learning_rate": 1.2685788787483704e-05,
"loss": 1.2955,
"step": 627
},
{
"epoch": 0.79,
"learning_rate": 1.2672750977835724e-05,
"loss": 1.2668,
"step": 628
},
{
"epoch": 0.8,
"learning_rate": 1.2659713168187745e-05,
"loss": 1.2819,
"step": 629
},
{
"epoch": 0.8,
"learning_rate": 1.2646675358539767e-05,
"loss": 1.2571,
"step": 630
},
{
"epoch": 0.8,
"learning_rate": 1.2633637548891787e-05,
"loss": 1.272,
"step": 631
},
{
"epoch": 0.8,
"learning_rate": 1.2620599739243808e-05,
"loss": 1.3043,
"step": 632
},
{
"epoch": 0.8,
"learning_rate": 1.2607561929595828e-05,
"loss": 1.2443,
"step": 633
},
{
"epoch": 0.8,
"learning_rate": 1.2594524119947851e-05,
"loss": 1.2615,
"step": 634
},
{
"epoch": 0.8,
"learning_rate": 1.2581486310299871e-05,
"loss": 1.2781,
"step": 635
},
{
"epoch": 0.8,
"learning_rate": 1.2568448500651891e-05,
"loss": 1.2577,
"step": 636
},
{
"epoch": 0.81,
"learning_rate": 1.2555410691003912e-05,
"loss": 1.2922,
"step": 637
},
{
"epoch": 0.81,
"learning_rate": 1.2542372881355932e-05,
"loss": 1.2979,
"step": 638
},
{
"epoch": 0.81,
"learning_rate": 1.2529335071707954e-05,
"loss": 1.2463,
"step": 639
},
{
"epoch": 0.81,
"learning_rate": 1.2516297262059975e-05,
"loss": 1.2788,
"step": 640
},
{
"epoch": 0.81,
"learning_rate": 1.2503259452411995e-05,
"loss": 1.2393,
"step": 641
},
{
"epoch": 0.81,
"learning_rate": 1.2490221642764018e-05,
"loss": 1.2751,
"step": 642
},
{
"epoch": 0.81,
"learning_rate": 1.2477183833116038e-05,
"loss": 1.2578,
"step": 643
},
{
"epoch": 0.81,
"learning_rate": 1.2464146023468058e-05,
"loss": 1.2604,
"step": 644
},
{
"epoch": 0.82,
"learning_rate": 1.245110821382008e-05,
"loss": 1.2474,
"step": 645
},
{
"epoch": 0.82,
"learning_rate": 1.24380704041721e-05,
"loss": 1.2448,
"step": 646
},
{
"epoch": 0.82,
"learning_rate": 1.242503259452412e-05,
"loss": 1.282,
"step": 647
},
{
"epoch": 0.82,
"learning_rate": 1.2411994784876143e-05,
"loss": 1.2609,
"step": 648
},
{
"epoch": 0.82,
"learning_rate": 1.2398956975228162e-05,
"loss": 1.2749,
"step": 649
},
{
"epoch": 0.82,
"learning_rate": 1.2385919165580182e-05,
"loss": 1.3005,
"step": 650
},
{
"epoch": 0.82,
"learning_rate": 1.2372881355932205e-05,
"loss": 1.2745,
"step": 651
},
{
"epoch": 0.82,
"learning_rate": 1.2359843546284225e-05,
"loss": 1.2719,
"step": 652
},
{
"epoch": 0.83,
"learning_rate": 1.2346805736636247e-05,
"loss": 1.2699,
"step": 653
},
{
"epoch": 0.83,
"learning_rate": 1.2333767926988267e-05,
"loss": 1.2916,
"step": 654
},
{
"epoch": 0.83,
"learning_rate": 1.2320730117340287e-05,
"loss": 1.2686,
"step": 655
},
{
"epoch": 0.83,
"learning_rate": 1.230769230769231e-05,
"loss": 1.26,
"step": 656
},
{
"epoch": 0.83,
"learning_rate": 1.229465449804433e-05,
"loss": 1.2863,
"step": 657
},
{
"epoch": 0.83,
"learning_rate": 1.228161668839635e-05,
"loss": 1.2948,
"step": 658
},
{
"epoch": 0.83,
"learning_rate": 1.2268578878748371e-05,
"loss": 1.2524,
"step": 659
},
{
"epoch": 0.83,
"learning_rate": 1.2255541069100392e-05,
"loss": 1.287,
"step": 660
},
{
"epoch": 0.84,
"learning_rate": 1.2242503259452414e-05,
"loss": 1.2449,
"step": 661
},
{
"epoch": 0.84,
"learning_rate": 1.2229465449804434e-05,
"loss": 1.2594,
"step": 662
},
{
"epoch": 0.84,
"learning_rate": 1.2216427640156454e-05,
"loss": 1.1996,
"step": 663
},
{
"epoch": 0.84,
"learning_rate": 1.2203389830508477e-05,
"loss": 1.2283,
"step": 664
},
{
"epoch": 0.84,
"learning_rate": 1.2190352020860497e-05,
"loss": 1.2709,
"step": 665
},
{
"epoch": 0.84,
"learning_rate": 1.2177314211212517e-05,
"loss": 1.265,
"step": 666
},
{
"epoch": 0.84,
"learning_rate": 1.2164276401564538e-05,
"loss": 1.2476,
"step": 667
},
{
"epoch": 0.84,
"learning_rate": 1.2151238591916558e-05,
"loss": 1.2646,
"step": 668
},
{
"epoch": 0.85,
"learning_rate": 1.213820078226858e-05,
"loss": 1.2458,
"step": 669
},
{
"epoch": 0.85,
"learning_rate": 1.2125162972620601e-05,
"loss": 1.281,
"step": 670
},
{
"epoch": 0.85,
"learning_rate": 1.2112125162972621e-05,
"loss": 1.2588,
"step": 671
},
{
"epoch": 0.85,
"learning_rate": 1.2099087353324644e-05,
"loss": 1.2638,
"step": 672
},
{
"epoch": 0.85,
"learning_rate": 1.2086049543676664e-05,
"loss": 1.2916,
"step": 673
},
{
"epoch": 0.85,
"learning_rate": 1.2073011734028684e-05,
"loss": 1.2437,
"step": 674
},
{
"epoch": 0.85,
"learning_rate": 1.2059973924380705e-05,
"loss": 1.2164,
"step": 675
},
{
"epoch": 0.85,
"learning_rate": 1.2046936114732725e-05,
"loss": 1.2417,
"step": 676
},
{
"epoch": 0.86,
"learning_rate": 1.2033898305084745e-05,
"loss": 1.2457,
"step": 677
},
{
"epoch": 0.86,
"learning_rate": 1.2020860495436768e-05,
"loss": 1.2517,
"step": 678
},
{
"epoch": 0.86,
"learning_rate": 1.2007822685788788e-05,
"loss": 1.2539,
"step": 679
},
{
"epoch": 0.86,
"learning_rate": 1.199478487614081e-05,
"loss": 1.2645,
"step": 680
},
{
"epoch": 0.86,
"learning_rate": 1.1981747066492831e-05,
"loss": 1.2408,
"step": 681
},
{
"epoch": 0.86,
"learning_rate": 1.1968709256844851e-05,
"loss": 1.2574,
"step": 682
},
{
"epoch": 0.86,
"learning_rate": 1.1955671447196873e-05,
"loss": 1.2322,
"step": 683
},
{
"epoch": 0.86,
"learning_rate": 1.1942633637548892e-05,
"loss": 1.2244,
"step": 684
},
{
"epoch": 0.87,
"learning_rate": 1.1929595827900912e-05,
"loss": 1.272,
"step": 685
},
{
"epoch": 0.87,
"learning_rate": 1.1916558018252935e-05,
"loss": 1.2791,
"step": 686
},
{
"epoch": 0.87,
"learning_rate": 1.1903520208604955e-05,
"loss": 1.2562,
"step": 687
},
{
"epoch": 0.87,
"learning_rate": 1.1890482398956975e-05,
"loss": 1.2441,
"step": 688
},
{
"epoch": 0.87,
"learning_rate": 1.1877444589308997e-05,
"loss": 1.2437,
"step": 689
},
{
"epoch": 0.87,
"learning_rate": 1.1864406779661018e-05,
"loss": 1.2618,
"step": 690
},
{
"epoch": 0.87,
"learning_rate": 1.185136897001304e-05,
"loss": 1.2552,
"step": 691
},
{
"epoch": 0.87,
"learning_rate": 1.183833116036506e-05,
"loss": 1.2729,
"step": 692
},
{
"epoch": 0.88,
"learning_rate": 1.182529335071708e-05,
"loss": 1.2729,
"step": 693
},
{
"epoch": 0.88,
"learning_rate": 1.1812255541069103e-05,
"loss": 1.2659,
"step": 694
},
{
"epoch": 0.88,
"learning_rate": 1.1799217731421122e-05,
"loss": 1.2323,
"step": 695
},
{
"epoch": 0.88,
"learning_rate": 1.1786179921773142e-05,
"loss": 1.2218,
"step": 696
},
{
"epoch": 0.88,
"learning_rate": 1.1773142112125164e-05,
"loss": 1.2419,
"step": 697
},
{
"epoch": 0.88,
"learning_rate": 1.1760104302477184e-05,
"loss": 1.2745,
"step": 698
},
{
"epoch": 0.88,
"learning_rate": 1.1747066492829207e-05,
"loss": 1.2887,
"step": 699
},
{
"epoch": 0.88,
"learning_rate": 1.1734028683181227e-05,
"loss": 1.2111,
"step": 700
},
{
"epoch": 0.89,
"learning_rate": 1.1720990873533247e-05,
"loss": 1.2789,
"step": 701
},
{
"epoch": 0.89,
"learning_rate": 1.170795306388527e-05,
"loss": 1.2781,
"step": 702
},
{
"epoch": 0.89,
"learning_rate": 1.169491525423729e-05,
"loss": 1.2439,
"step": 703
},
{
"epoch": 0.89,
"learning_rate": 1.168187744458931e-05,
"loss": 1.2841,
"step": 704
},
{
"epoch": 0.89,
"learning_rate": 1.1668839634941331e-05,
"loss": 1.2676,
"step": 705
},
{
"epoch": 0.89,
"learning_rate": 1.165580182529335e-05,
"loss": 1.2831,
"step": 706
},
{
"epoch": 0.89,
"learning_rate": 1.1642764015645374e-05,
"loss": 1.2432,
"step": 707
},
{
"epoch": 0.9,
"learning_rate": 1.1629726205997394e-05,
"loss": 1.265,
"step": 708
},
{
"epoch": 0.9,
"learning_rate": 1.1616688396349414e-05,
"loss": 1.2151,
"step": 709
},
{
"epoch": 0.9,
"learning_rate": 1.1603650586701435e-05,
"loss": 1.2303,
"step": 710
},
{
"epoch": 0.9,
"learning_rate": 1.1590612777053457e-05,
"loss": 1.2487,
"step": 711
},
{
"epoch": 0.9,
"learning_rate": 1.1577574967405477e-05,
"loss": 1.2618,
"step": 712
},
{
"epoch": 0.9,
"learning_rate": 1.1564537157757498e-05,
"loss": 1.2507,
"step": 713
},
{
"epoch": 0.9,
"learning_rate": 1.1551499348109518e-05,
"loss": 1.2502,
"step": 714
},
{
"epoch": 0.9,
"learning_rate": 1.1538461538461538e-05,
"loss": 1.2424,
"step": 715
},
{
"epoch": 0.91,
"learning_rate": 1.1525423728813561e-05,
"loss": 1.2504,
"step": 716
},
{
"epoch": 0.91,
"learning_rate": 1.1512385919165581e-05,
"loss": 1.2261,
"step": 717
},
{
"epoch": 0.91,
"learning_rate": 1.1499348109517602e-05,
"loss": 1.2887,
"step": 718
},
{
"epoch": 0.91,
"learning_rate": 1.1486310299869622e-05,
"loss": 1.2443,
"step": 719
},
{
"epoch": 0.91,
"learning_rate": 1.1473272490221644e-05,
"loss": 1.2317,
"step": 720
},
{
"epoch": 0.91,
"learning_rate": 1.1460234680573665e-05,
"loss": 1.2329,
"step": 721
},
{
"epoch": 0.91,
"learning_rate": 1.1447196870925685e-05,
"loss": 1.2781,
"step": 722
},
{
"epoch": 0.91,
"learning_rate": 1.1434159061277705e-05,
"loss": 1.2329,
"step": 723
},
{
"epoch": 0.92,
"learning_rate": 1.1421121251629728e-05,
"loss": 1.2723,
"step": 724
},
{
"epoch": 0.92,
"learning_rate": 1.1408083441981748e-05,
"loss": 1.236,
"step": 725
},
{
"epoch": 0.92,
"learning_rate": 1.139504563233377e-05,
"loss": 1.2114,
"step": 726
},
{
"epoch": 0.92,
"learning_rate": 1.138200782268579e-05,
"loss": 1.2483,
"step": 727
},
{
"epoch": 0.92,
"learning_rate": 1.136897001303781e-05,
"loss": 1.2572,
"step": 728
},
{
"epoch": 0.92,
"learning_rate": 1.1355932203389833e-05,
"loss": 1.2286,
"step": 729
},
{
"epoch": 0.92,
"learning_rate": 1.1342894393741852e-05,
"loss": 1.2546,
"step": 730
},
{
"epoch": 0.92,
"learning_rate": 1.1329856584093872e-05,
"loss": 1.238,
"step": 731
},
{
"epoch": 0.93,
"learning_rate": 1.1316818774445895e-05,
"loss": 1.246,
"step": 732
},
{
"epoch": 0.93,
"learning_rate": 1.1303780964797915e-05,
"loss": 1.2441,
"step": 733
},
{
"epoch": 0.93,
"learning_rate": 1.1290743155149935e-05,
"loss": 1.2487,
"step": 734
},
{
"epoch": 0.93,
"learning_rate": 1.1277705345501957e-05,
"loss": 1.2324,
"step": 735
},
{
"epoch": 0.93,
"learning_rate": 1.1264667535853976e-05,
"loss": 1.2577,
"step": 736
},
{
"epoch": 0.93,
"learning_rate": 1.1251629726206e-05,
"loss": 1.2184,
"step": 737
},
{
"epoch": 0.93,
"learning_rate": 1.123859191655802e-05,
"loss": 1.2179,
"step": 738
},
{
"epoch": 0.93,
"learning_rate": 1.122555410691004e-05,
"loss": 1.2626,
"step": 739
},
{
"epoch": 0.94,
"learning_rate": 1.1212516297262061e-05,
"loss": 1.2292,
"step": 740
},
{
"epoch": 0.94,
"learning_rate": 1.1199478487614082e-05,
"loss": 1.274,
"step": 741
},
{
"epoch": 0.94,
"learning_rate": 1.1186440677966102e-05,
"loss": 1.2303,
"step": 742
},
{
"epoch": 0.94,
"learning_rate": 1.1173402868318124e-05,
"loss": 1.2963,
"step": 743
},
{
"epoch": 0.94,
"learning_rate": 1.1160365058670144e-05,
"loss": 1.231,
"step": 744
},
{
"epoch": 0.94,
"learning_rate": 1.1147327249022167e-05,
"loss": 1.2573,
"step": 745
},
{
"epoch": 0.94,
"learning_rate": 1.1134289439374187e-05,
"loss": 1.2445,
"step": 746
},
{
"epoch": 0.94,
"learning_rate": 1.1121251629726207e-05,
"loss": 1.2458,
"step": 747
},
{
"epoch": 0.95,
"learning_rate": 1.1108213820078228e-05,
"loss": 1.2385,
"step": 748
},
{
"epoch": 0.95,
"learning_rate": 1.1095176010430248e-05,
"loss": 1.2599,
"step": 749
},
{
"epoch": 0.95,
"learning_rate": 1.108213820078227e-05,
"loss": 1.2517,
"step": 750
},
{
"epoch": 0.95,
"learning_rate": 1.1069100391134291e-05,
"loss": 1.255,
"step": 751
},
{
"epoch": 0.95,
"learning_rate": 1.1056062581486311e-05,
"loss": 1.2482,
"step": 752
},
{
"epoch": 0.95,
"learning_rate": 1.104302477183833e-05,
"loss": 1.2484,
"step": 753
},
{
"epoch": 0.95,
"learning_rate": 1.1029986962190354e-05,
"loss": 1.2538,
"step": 754
},
{
"epoch": 0.95,
"learning_rate": 1.1016949152542374e-05,
"loss": 1.244,
"step": 755
},
{
"epoch": 0.96,
"learning_rate": 1.1003911342894395e-05,
"loss": 1.2635,
"step": 756
},
{
"epoch": 0.96,
"learning_rate": 1.0990873533246415e-05,
"loss": 1.2307,
"step": 757
},
{
"epoch": 0.96,
"learning_rate": 1.0977835723598435e-05,
"loss": 1.2157,
"step": 758
},
{
"epoch": 0.96,
"learning_rate": 1.0964797913950458e-05,
"loss": 1.2393,
"step": 759
},
{
"epoch": 0.96,
"learning_rate": 1.0951760104302478e-05,
"loss": 1.2542,
"step": 760
},
{
"epoch": 0.96,
"learning_rate": 1.0938722294654498e-05,
"loss": 1.2732,
"step": 761
},
{
"epoch": 0.96,
"learning_rate": 1.092568448500652e-05,
"loss": 1.2625,
"step": 762
},
{
"epoch": 0.96,
"learning_rate": 1.0912646675358541e-05,
"loss": 1.2423,
"step": 763
},
{
"epoch": 0.97,
"learning_rate": 1.0899608865710562e-05,
"loss": 1.2318,
"step": 764
},
{
"epoch": 0.97,
"learning_rate": 1.0886571056062582e-05,
"loss": 1.2404,
"step": 765
},
{
"epoch": 0.97,
"learning_rate": 1.0873533246414602e-05,
"loss": 1.2513,
"step": 766
},
{
"epoch": 0.97,
"learning_rate": 1.0860495436766625e-05,
"loss": 1.2864,
"step": 767
},
{
"epoch": 0.97,
"learning_rate": 1.0847457627118645e-05,
"loss": 1.2753,
"step": 768
},
{
"epoch": 0.97,
"learning_rate": 1.0834419817470665e-05,
"loss": 1.2426,
"step": 769
},
{
"epoch": 0.97,
"learning_rate": 1.0821382007822687e-05,
"loss": 1.2604,
"step": 770
},
{
"epoch": 0.97,
"learning_rate": 1.0808344198174706e-05,
"loss": 1.2538,
"step": 771
},
{
"epoch": 0.98,
"learning_rate": 1.0795306388526728e-05,
"loss": 1.2528,
"step": 772
},
{
"epoch": 0.98,
"learning_rate": 1.078226857887875e-05,
"loss": 1.2666,
"step": 773
},
{
"epoch": 0.98,
"learning_rate": 1.076923076923077e-05,
"loss": 1.2455,
"step": 774
},
{
"epoch": 0.98,
"learning_rate": 1.0756192959582793e-05,
"loss": 1.2563,
"step": 775
},
{
"epoch": 0.98,
"learning_rate": 1.0743155149934812e-05,
"loss": 1.2577,
"step": 776
},
{
"epoch": 0.98,
"learning_rate": 1.0730117340286832e-05,
"loss": 1.2611,
"step": 777
},
{
"epoch": 0.98,
"learning_rate": 1.0717079530638854e-05,
"loss": 1.2731,
"step": 778
},
{
"epoch": 0.98,
"learning_rate": 1.0704041720990874e-05,
"loss": 1.2245,
"step": 779
},
{
"epoch": 0.99,
"learning_rate": 1.0691003911342893e-05,
"loss": 1.2378,
"step": 780
},
{
"epoch": 0.99,
"learning_rate": 1.0677966101694917e-05,
"loss": 1.2627,
"step": 781
},
{
"epoch": 0.99,
"learning_rate": 1.0664928292046937e-05,
"loss": 1.2847,
"step": 782
},
{
"epoch": 0.99,
"learning_rate": 1.0651890482398958e-05,
"loss": 1.2548,
"step": 783
},
{
"epoch": 0.99,
"learning_rate": 1.063885267275098e-05,
"loss": 1.2675,
"step": 784
},
{
"epoch": 0.99,
"learning_rate": 1.0625814863103e-05,
"loss": 1.2573,
"step": 785
},
{
"epoch": 0.99,
"learning_rate": 1.0612777053455021e-05,
"loss": 1.2499,
"step": 786
},
{
"epoch": 0.99,
"learning_rate": 1.059973924380704e-05,
"loss": 1.2209,
"step": 787
},
{
"epoch": 1.0,
"learning_rate": 1.058670143415906e-05,
"loss": 1.2124,
"step": 788
},
{
"epoch": 1.0,
"learning_rate": 1.0573663624511084e-05,
"loss": 1.2803,
"step": 789
},
{
"epoch": 1.0,
"learning_rate": 1.0560625814863104e-05,
"loss": 1.2343,
"step": 790
},
{
"epoch": 1.0,
"learning_rate": 1.0547588005215125e-05,
"loss": 1.3397,
"step": 791
}
],
"max_steps": 1582,
"num_train_epochs": 2,
"total_flos": 5423674366623744.0,
"trial_name": null,
"trial_params": null
}