vila1.5-3b_cropped-city / trainer_state.json
tollea1234's picture
Upload folder using huggingface_hub
22a5f39 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9991941982272361,
"eval_steps": 500,
"global_step": 620,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5.263157894736842e-06,
"loss": 3.5906,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 1.0526315789473684e-05,
"loss": 3.5784,
"step": 2
},
{
"epoch": 0.0,
"learning_rate": 1.5789473684210526e-05,
"loss": 3.1344,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 2.105263157894737e-05,
"loss": 2.2119,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 2.6315789473684212e-05,
"loss": 1.6495,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 3.157894736842105e-05,
"loss": 0.666,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 3.6842105263157895e-05,
"loss": 0.2982,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 4.210526315789474e-05,
"loss": 0.2059,
"step": 8
},
{
"epoch": 0.01,
"learning_rate": 4.736842105263158e-05,
"loss": 0.1887,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 5.2631578947368424e-05,
"loss": 0.1531,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 5.789473684210527e-05,
"loss": 0.183,
"step": 11
},
{
"epoch": 0.02,
"learning_rate": 6.31578947368421e-05,
"loss": 0.1053,
"step": 12
},
{
"epoch": 0.02,
"learning_rate": 6.842105263157895e-05,
"loss": 0.1282,
"step": 13
},
{
"epoch": 0.02,
"learning_rate": 7.368421052631579e-05,
"loss": 0.1039,
"step": 14
},
{
"epoch": 0.02,
"learning_rate": 7.894736842105263e-05,
"loss": 0.106,
"step": 15
},
{
"epoch": 0.03,
"learning_rate": 8.421052631578948e-05,
"loss": 0.119,
"step": 16
},
{
"epoch": 0.03,
"learning_rate": 8.947368421052632e-05,
"loss": 0.0991,
"step": 17
},
{
"epoch": 0.03,
"learning_rate": 9.473684210526316e-05,
"loss": 0.0935,
"step": 18
},
{
"epoch": 0.03,
"learning_rate": 0.0001,
"loss": 0.0925,
"step": 19
},
{
"epoch": 0.03,
"learning_rate": 9.99993168912927e-05,
"loss": 0.0825,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 9.999726758383623e-05,
"loss": 0.0999,
"step": 21
},
{
"epoch": 0.04,
"learning_rate": 9.999385213362663e-05,
"loss": 0.0725,
"step": 22
},
{
"epoch": 0.04,
"learning_rate": 9.998907063398883e-05,
"loss": 0.0795,
"step": 23
},
{
"epoch": 0.04,
"learning_rate": 9.998292321557421e-05,
"loss": 0.0766,
"step": 24
},
{
"epoch": 0.04,
"learning_rate": 9.997541004635695e-05,
"loss": 0.083,
"step": 25
},
{
"epoch": 0.04,
"learning_rate": 9.99665313316295e-05,
"loss": 0.073,
"step": 26
},
{
"epoch": 0.04,
"learning_rate": 9.995628731399699e-05,
"loss": 0.0987,
"step": 27
},
{
"epoch": 0.05,
"learning_rate": 9.994467827337048e-05,
"loss": 0.0668,
"step": 28
},
{
"epoch": 0.05,
"learning_rate": 9.993170452695948e-05,
"loss": 0.0797,
"step": 29
},
{
"epoch": 0.05,
"learning_rate": 9.991736642926311e-05,
"loss": 0.0713,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 9.990166437206058e-05,
"loss": 0.0659,
"step": 31
},
{
"epoch": 0.05,
"learning_rate": 9.988459878440038e-05,
"loss": 0.0609,
"step": 32
},
{
"epoch": 0.05,
"learning_rate": 9.986617013258853e-05,
"loss": 0.078,
"step": 33
},
{
"epoch": 0.05,
"learning_rate": 9.984637892017596e-05,
"loss": 0.0569,
"step": 34
},
{
"epoch": 0.06,
"learning_rate": 9.982522568794465e-05,
"loss": 0.0569,
"step": 35
},
{
"epoch": 0.06,
"learning_rate": 9.980271101389287e-05,
"loss": 0.0481,
"step": 36
},
{
"epoch": 0.06,
"learning_rate": 9.977883551321942e-05,
"loss": 0.0719,
"step": 37
},
{
"epoch": 0.06,
"learning_rate": 9.975359983830681e-05,
"loss": 0.0593,
"step": 38
},
{
"epoch": 0.06,
"learning_rate": 9.972700467870339e-05,
"loss": 0.0605,
"step": 39
},
{
"epoch": 0.06,
"learning_rate": 9.969905076110458e-05,
"loss": 0.0595,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 9.966973884933298e-05,
"loss": 0.0588,
"step": 41
},
{
"epoch": 0.07,
"learning_rate": 9.963906974431742e-05,
"loss": 0.0598,
"step": 42
},
{
"epoch": 0.07,
"learning_rate": 9.960704428407124e-05,
"loss": 0.059,
"step": 43
},
{
"epoch": 0.07,
"learning_rate": 9.957366334366928e-05,
"loss": 0.0653,
"step": 44
},
{
"epoch": 0.07,
"learning_rate": 9.953892783522396e-05,
"loss": 0.0459,
"step": 45
},
{
"epoch": 0.07,
"learning_rate": 9.950283870786043e-05,
"loss": 0.0518,
"step": 46
},
{
"epoch": 0.08,
"learning_rate": 9.946539694769058e-05,
"loss": 0.0491,
"step": 47
},
{
"epoch": 0.08,
"learning_rate": 9.942660357778607e-05,
"loss": 0.0419,
"step": 48
},
{
"epoch": 0.08,
"learning_rate": 9.938645965815049e-05,
"loss": 0.0497,
"step": 49
},
{
"epoch": 0.08,
"learning_rate": 9.934496628569026e-05,
"loss": 0.0511,
"step": 50
},
{
"epoch": 0.08,
"learning_rate": 9.930212459418475e-05,
"loss": 0.0574,
"step": 51
},
{
"epoch": 0.08,
"learning_rate": 9.925793575425525e-05,
"loss": 0.0703,
"step": 52
},
{
"epoch": 0.09,
"learning_rate": 9.921240097333303e-05,
"loss": 0.046,
"step": 53
},
{
"epoch": 0.09,
"learning_rate": 9.916552149562628e-05,
"loss": 0.0599,
"step": 54
},
{
"epoch": 0.09,
"learning_rate": 9.911729860208619e-05,
"loss": 0.0501,
"step": 55
},
{
"epoch": 0.09,
"learning_rate": 9.90677336103719e-05,
"loss": 0.0636,
"step": 56
},
{
"epoch": 0.09,
"learning_rate": 9.901682787481452e-05,
"loss": 0.0454,
"step": 57
},
{
"epoch": 0.09,
"learning_rate": 9.896458278638008e-05,
"loss": 0.054,
"step": 58
},
{
"epoch": 0.1,
"learning_rate": 9.891099977263156e-05,
"loss": 0.0563,
"step": 59
},
{
"epoch": 0.1,
"learning_rate": 9.885608029768991e-05,
"loss": 0.0391,
"step": 60
},
{
"epoch": 0.1,
"learning_rate": 9.879982586219401e-05,
"loss": 0.0505,
"step": 61
},
{
"epoch": 0.1,
"learning_rate": 9.874223800325961e-05,
"loss": 0.0478,
"step": 62
},
{
"epoch": 0.1,
"learning_rate": 9.868331829443744e-05,
"loss": 0.0474,
"step": 63
},
{
"epoch": 0.1,
"learning_rate": 9.862306834567014e-05,
"loss": 0.0411,
"step": 64
},
{
"epoch": 0.1,
"learning_rate": 9.856148980324831e-05,
"loss": 0.0413,
"step": 65
},
{
"epoch": 0.11,
"learning_rate": 9.849858434976549e-05,
"loss": 0.04,
"step": 66
},
{
"epoch": 0.11,
"learning_rate": 9.843435370407219e-05,
"loss": 0.0437,
"step": 67
},
{
"epoch": 0.11,
"learning_rate": 9.836879962122895e-05,
"loss": 0.0417,
"step": 68
},
{
"epoch": 0.11,
"learning_rate": 9.830192389245835e-05,
"loss": 0.0427,
"step": 69
},
{
"epoch": 0.11,
"learning_rate": 9.82337283450961e-05,
"loss": 0.0506,
"step": 70
},
{
"epoch": 0.11,
"learning_rate": 9.81642148425411e-05,
"loss": 0.0414,
"step": 71
},
{
"epoch": 0.12,
"learning_rate": 9.809338528420448e-05,
"loss": 0.0498,
"step": 72
},
{
"epoch": 0.12,
"learning_rate": 9.80212416054578e-05,
"loss": 0.0373,
"step": 73
},
{
"epoch": 0.12,
"learning_rate": 9.794778577758004e-05,
"loss": 0.0414,
"step": 74
},
{
"epoch": 0.12,
"learning_rate": 9.787301980770382e-05,
"loss": 0.0492,
"step": 75
},
{
"epoch": 0.12,
"learning_rate": 9.779694573876055e-05,
"loss": 0.0438,
"step": 76
},
{
"epoch": 0.12,
"learning_rate": 9.771956564942459e-05,
"loss": 0.0459,
"step": 77
},
{
"epoch": 0.13,
"learning_rate": 9.764088165405645e-05,
"loss": 0.0396,
"step": 78
},
{
"epoch": 0.13,
"learning_rate": 9.7560895902645e-05,
"loss": 0.036,
"step": 79
},
{
"epoch": 0.13,
"learning_rate": 9.747961058074881e-05,
"loss": 0.0502,
"step": 80
},
{
"epoch": 0.13,
"learning_rate": 9.739702790943631e-05,
"loss": 0.0323,
"step": 81
},
{
"epoch": 0.13,
"learning_rate": 9.731315014522517e-05,
"loss": 0.0437,
"step": 82
},
{
"epoch": 0.13,
"learning_rate": 9.722797958002064e-05,
"loss": 0.0413,
"step": 83
},
{
"epoch": 0.14,
"learning_rate": 9.71415185410529e-05,
"loss": 0.038,
"step": 84
},
{
"epoch": 0.14,
"learning_rate": 9.705376939081348e-05,
"loss": 0.0447,
"step": 85
},
{
"epoch": 0.14,
"learning_rate": 9.696473452699075e-05,
"loss": 0.0427,
"step": 86
},
{
"epoch": 0.14,
"learning_rate": 9.687441638240433e-05,
"loss": 0.0381,
"step": 87
},
{
"epoch": 0.14,
"learning_rate": 9.678281742493866e-05,
"loss": 0.0546,
"step": 88
},
{
"epoch": 0.14,
"learning_rate": 9.668994015747556e-05,
"loss": 0.034,
"step": 89
},
{
"epoch": 0.15,
"learning_rate": 9.659578711782583e-05,
"loss": 0.0398,
"step": 90
},
{
"epoch": 0.15,
"learning_rate": 9.650036087865992e-05,
"loss": 0.0473,
"step": 91
},
{
"epoch": 0.15,
"learning_rate": 9.640366404743761e-05,
"loss": 0.0344,
"step": 92
},
{
"epoch": 0.15,
"learning_rate": 9.630569926633682e-05,
"loss": 0.0393,
"step": 93
},
{
"epoch": 0.15,
"learning_rate": 9.620646921218134e-05,
"loss": 0.0401,
"step": 94
},
{
"epoch": 0.15,
"learning_rate": 9.610597659636773e-05,
"loss": 0.0378,
"step": 95
},
{
"epoch": 0.15,
"learning_rate": 9.60042241647912e-05,
"loss": 0.0464,
"step": 96
},
{
"epoch": 0.16,
"learning_rate": 9.590121469777069e-05,
"loss": 0.0486,
"step": 97
},
{
"epoch": 0.16,
"learning_rate": 9.57969510099727e-05,
"loss": 0.0307,
"step": 98
},
{
"epoch": 0.16,
"learning_rate": 9.569143595033456e-05,
"loss": 0.0356,
"step": 99
},
{
"epoch": 0.16,
"learning_rate": 9.558467240198655e-05,
"loss": 0.037,
"step": 100
},
{
"epoch": 0.16,
"learning_rate": 9.547666328217298e-05,
"loss": 0.0335,
"step": 101
},
{
"epoch": 0.16,
"learning_rate": 9.536741154217272e-05,
"loss": 0.0344,
"step": 102
},
{
"epoch": 0.17,
"learning_rate": 9.525692016721834e-05,
"loss": 0.0334,
"step": 103
},
{
"epoch": 0.17,
"learning_rate": 9.514519217641465e-05,
"loss": 0.0359,
"step": 104
},
{
"epoch": 0.17,
"learning_rate": 9.503223062265618e-05,
"loss": 0.0311,
"step": 105
},
{
"epoch": 0.17,
"learning_rate": 9.491803859254378e-05,
"loss": 0.0332,
"step": 106
},
{
"epoch": 0.17,
"learning_rate": 9.480261920630027e-05,
"loss": 0.0351,
"step": 107
},
{
"epoch": 0.17,
"learning_rate": 9.468597561768513e-05,
"loss": 0.0415,
"step": 108
},
{
"epoch": 0.18,
"learning_rate": 9.45681110139084e-05,
"loss": 0.03,
"step": 109
},
{
"epoch": 0.18,
"learning_rate": 9.444902861554359e-05,
"loss": 0.0368,
"step": 110
},
{
"epoch": 0.18,
"learning_rate": 9.43287316764396e-05,
"loss": 0.029,
"step": 111
},
{
"epoch": 0.18,
"learning_rate": 9.420722348363194e-05,
"loss": 0.0292,
"step": 112
},
{
"epoch": 0.18,
"learning_rate": 9.408450735725273e-05,
"loss": 0.0443,
"step": 113
},
{
"epoch": 0.18,
"learning_rate": 9.396058665044018e-05,
"loss": 0.0306,
"step": 114
},
{
"epoch": 0.19,
"learning_rate": 9.383546474924684e-05,
"loss": 0.0379,
"step": 115
},
{
"epoch": 0.19,
"learning_rate": 9.370914507254714e-05,
"loss": 0.0322,
"step": 116
},
{
"epoch": 0.19,
"learning_rate": 9.358163107194389e-05,
"loss": 0.0353,
"step": 117
},
{
"epoch": 0.19,
"learning_rate": 9.345292623167405e-05,
"loss": 0.0372,
"step": 118
},
{
"epoch": 0.19,
"learning_rate": 9.332303406851353e-05,
"loss": 0.0305,
"step": 119
},
{
"epoch": 0.19,
"learning_rate": 9.319195813168102e-05,
"loss": 0.0308,
"step": 120
},
{
"epoch": 0.2,
"learning_rate": 9.305970200274108e-05,
"loss": 0.0347,
"step": 121
},
{
"epoch": 0.2,
"learning_rate": 9.292626929550623e-05,
"loss": 0.0349,
"step": 122
},
{
"epoch": 0.2,
"learning_rate": 9.279166365593823e-05,
"loss": 0.0398,
"step": 123
},
{
"epoch": 0.2,
"learning_rate": 9.26558887620485e-05,
"loss": 0.0347,
"step": 124
},
{
"epoch": 0.2,
"learning_rate": 9.251894832379747e-05,
"loss": 0.0337,
"step": 125
},
{
"epoch": 0.2,
"learning_rate": 9.238084608299343e-05,
"loss": 0.032,
"step": 126
},
{
"epoch": 0.2,
"learning_rate": 9.224158581319004e-05,
"loss": 0.0336,
"step": 127
},
{
"epoch": 0.21,
"learning_rate": 9.210117131958348e-05,
"loss": 0.0385,
"step": 128
},
{
"epoch": 0.21,
"learning_rate": 9.195960643890824e-05,
"loss": 0.0389,
"step": 129
},
{
"epoch": 0.21,
"learning_rate": 9.181689503933243e-05,
"loss": 0.0356,
"step": 130
},
{
"epoch": 0.21,
"learning_rate": 9.167304102035207e-05,
"loss": 0.0317,
"step": 131
},
{
"epoch": 0.21,
"learning_rate": 9.152804831268444e-05,
"loss": 0.0362,
"step": 132
},
{
"epoch": 0.21,
"learning_rate": 9.138192087816079e-05,
"loss": 0.0257,
"step": 133
},
{
"epoch": 0.22,
"learning_rate": 9.123466270961806e-05,
"loss": 0.0408,
"step": 134
},
{
"epoch": 0.22,
"learning_rate": 9.108627783078971e-05,
"loss": 0.0314,
"step": 135
},
{
"epoch": 0.22,
"learning_rate": 9.093677029619585e-05,
"loss": 0.0305,
"step": 136
},
{
"epoch": 0.22,
"learning_rate": 9.078614419103245e-05,
"loss": 0.0335,
"step": 137
},
{
"epoch": 0.22,
"learning_rate": 9.063440363105965e-05,
"loss": 0.0301,
"step": 138
},
{
"epoch": 0.22,
"learning_rate": 9.048155276248938e-05,
"loss": 0.0295,
"step": 139
},
{
"epoch": 0.23,
"learning_rate": 9.032759576187197e-05,
"loss": 0.0293,
"step": 140
},
{
"epoch": 0.23,
"learning_rate": 9.017253683598218e-05,
"loss": 0.0333,
"step": 141
},
{
"epoch": 0.23,
"learning_rate": 9.001638022170407e-05,
"loss": 0.0277,
"step": 142
},
{
"epoch": 0.23,
"learning_rate": 8.985913018591536e-05,
"loss": 0.0302,
"step": 143
},
{
"epoch": 0.23,
"learning_rate": 8.970079102537083e-05,
"loss": 0.0354,
"step": 144
},
{
"epoch": 0.23,
"learning_rate": 8.95413670665848e-05,
"loss": 0.0297,
"step": 145
},
{
"epoch": 0.24,
"learning_rate": 8.938086266571309e-05,
"loss": 0.0275,
"step": 146
},
{
"epoch": 0.24,
"learning_rate": 8.921928220843381e-05,
"loss": 0.0241,
"step": 147
},
{
"epoch": 0.24,
"learning_rate": 8.90566301098277e-05,
"loss": 0.0319,
"step": 148
},
{
"epoch": 0.24,
"learning_rate": 8.889291081425732e-05,
"loss": 0.0277,
"step": 149
},
{
"epoch": 0.24,
"learning_rate": 8.872812879524573e-05,
"loss": 0.041,
"step": 150
},
{
"epoch": 0.24,
"learning_rate": 8.856228855535422e-05,
"loss": 0.0257,
"step": 151
},
{
"epoch": 0.24,
"learning_rate": 8.839539462605926e-05,
"loss": 0.0317,
"step": 152
},
{
"epoch": 0.25,
"learning_rate": 8.822745156762869e-05,
"loss": 0.0353,
"step": 153
},
{
"epoch": 0.25,
"learning_rate": 8.805846396899715e-05,
"loss": 0.0276,
"step": 154
},
{
"epoch": 0.25,
"learning_rate": 8.788843644764064e-05,
"loss": 0.0334,
"step": 155
},
{
"epoch": 0.25,
"learning_rate": 8.771737364945036e-05,
"loss": 0.0306,
"step": 156
},
{
"epoch": 0.25,
"learning_rate": 8.75452802486058e-05,
"loss": 0.0242,
"step": 157
},
{
"epoch": 0.25,
"learning_rate": 8.737216094744697e-05,
"loss": 0.0365,
"step": 158
},
{
"epoch": 0.26,
"learning_rate": 8.719802047634596e-05,
"loss": 0.0322,
"step": 159
},
{
"epoch": 0.26,
"learning_rate": 8.702286359357765e-05,
"loss": 0.0293,
"step": 160
},
{
"epoch": 0.26,
"learning_rate": 8.684669508518974e-05,
"loss": 0.0333,
"step": 161
},
{
"epoch": 0.26,
"learning_rate": 8.666951976487187e-05,
"loss": 0.032,
"step": 162
},
{
"epoch": 0.26,
"learning_rate": 8.649134247382421e-05,
"loss": 0.0252,
"step": 163
},
{
"epoch": 0.26,
"learning_rate": 8.631216808062514e-05,
"loss": 0.038,
"step": 164
},
{
"epoch": 0.27,
"learning_rate": 8.613200148109816e-05,
"loss": 0.0343,
"step": 165
},
{
"epoch": 0.27,
"learning_rate": 8.59508475981782e-05,
"loss": 0.0297,
"step": 166
},
{
"epoch": 0.27,
"learning_rate": 8.576871138177705e-05,
"loss": 0.0281,
"step": 167
},
{
"epoch": 0.27,
"learning_rate": 8.558559780864813e-05,
"loss": 0.0281,
"step": 168
},
{
"epoch": 0.27,
"learning_rate": 8.540151188225046e-05,
"loss": 0.0326,
"step": 169
},
{
"epoch": 0.27,
"learning_rate": 8.521645863261205e-05,
"loss": 0.028,
"step": 170
},
{
"epoch": 0.28,
"learning_rate": 8.503044311619232e-05,
"loss": 0.0367,
"step": 171
},
{
"epoch": 0.28,
"learning_rate": 8.484347041574402e-05,
"loss": 0.0209,
"step": 172
},
{
"epoch": 0.28,
"learning_rate": 8.465554564017437e-05,
"loss": 0.0352,
"step": 173
},
{
"epoch": 0.28,
"learning_rate": 8.446667392440538e-05,
"loss": 0.0301,
"step": 174
},
{
"epoch": 0.28,
"learning_rate": 8.427686042923358e-05,
"loss": 0.0312,
"step": 175
},
{
"epoch": 0.28,
"learning_rate": 8.408611034118902e-05,
"loss": 0.0353,
"step": 176
},
{
"epoch": 0.29,
"learning_rate": 8.389442887239357e-05,
"loss": 0.0241,
"step": 177
},
{
"epoch": 0.29,
"learning_rate": 8.370182126041841e-05,
"loss": 0.0379,
"step": 178
},
{
"epoch": 0.29,
"learning_rate": 8.350829276814103e-05,
"loss": 0.0272,
"step": 179
},
{
"epoch": 0.29,
"learning_rate": 8.331384868360138e-05,
"loss": 0.0282,
"step": 180
},
{
"epoch": 0.29,
"learning_rate": 8.31184943198573e-05,
"loss": 0.0251,
"step": 181
},
{
"epoch": 0.29,
"learning_rate": 8.292223501483952e-05,
"loss": 0.031,
"step": 182
},
{
"epoch": 0.29,
"learning_rate": 8.27250761312056e-05,
"loss": 0.0327,
"step": 183
},
{
"epoch": 0.3,
"learning_rate": 8.252702305619357e-05,
"loss": 0.0238,
"step": 184
},
{
"epoch": 0.3,
"learning_rate": 8.232808120147464e-05,
"loss": 0.0345,
"step": 185
},
{
"epoch": 0.3,
"learning_rate": 8.21282560030053e-05,
"loss": 0.0268,
"step": 186
},
{
"epoch": 0.3,
"learning_rate": 8.192755292087893e-05,
"loss": 0.0342,
"step": 187
},
{
"epoch": 0.3,
"learning_rate": 8.172597743917639e-05,
"loss": 0.0229,
"step": 188
},
{
"epoch": 0.3,
"learning_rate": 8.152353506581637e-05,
"loss": 0.0321,
"step": 189
},
{
"epoch": 0.31,
"learning_rate": 8.132023133240481e-05,
"loss": 0.0267,
"step": 190
},
{
"epoch": 0.31,
"learning_rate": 8.111607179408369e-05,
"loss": 0.0264,
"step": 191
},
{
"epoch": 0.31,
"learning_rate": 8.09110620293794e-05,
"loss": 0.0276,
"step": 192
},
{
"epoch": 0.31,
"learning_rate": 8.07052076400501e-05,
"loss": 0.0282,
"step": 193
},
{
"epoch": 0.31,
"learning_rate": 8.049851425093284e-05,
"loss": 0.0296,
"step": 194
},
{
"epoch": 0.31,
"learning_rate": 8.02909875097898e-05,
"loss": 0.0274,
"step": 195
},
{
"epoch": 0.32,
"learning_rate": 8.008263308715389e-05,
"loss": 0.0282,
"step": 196
},
{
"epoch": 0.32,
"learning_rate": 7.987345667617395e-05,
"loss": 0.031,
"step": 197
},
{
"epoch": 0.32,
"learning_rate": 7.966346399245909e-05,
"loss": 0.0248,
"step": 198
},
{
"epoch": 0.32,
"learning_rate": 7.945266077392253e-05,
"loss": 0.0229,
"step": 199
},
{
"epoch": 0.32,
"learning_rate": 7.924105278062481e-05,
"loss": 0.0385,
"step": 200
},
{
"epoch": 0.32,
"learning_rate": 7.90286457946165e-05,
"loss": 0.0261,
"step": 201
},
{
"epoch": 0.33,
"learning_rate": 7.881544561978002e-05,
"loss": 0.0252,
"step": 202
},
{
"epoch": 0.33,
"learning_rate": 7.86014580816712e-05,
"loss": 0.0293,
"step": 203
},
{
"epoch": 0.33,
"learning_rate": 7.83866890273601e-05,
"loss": 0.0228,
"step": 204
},
{
"epoch": 0.33,
"learning_rate": 7.817114432527113e-05,
"loss": 0.0224,
"step": 205
},
{
"epoch": 0.33,
"learning_rate": 7.79548298650228e-05,
"loss": 0.0274,
"step": 206
},
{
"epoch": 0.33,
"learning_rate": 7.77377515572668e-05,
"loss": 0.0249,
"step": 207
},
{
"epoch": 0.34,
"learning_rate": 7.751991533352636e-05,
"loss": 0.0354,
"step": 208
},
{
"epoch": 0.34,
"learning_rate": 7.730132714603438e-05,
"loss": 0.0256,
"step": 209
},
{
"epoch": 0.34,
"learning_rate": 7.708199296757061e-05,
"loss": 0.0277,
"step": 210
},
{
"epoch": 0.34,
"learning_rate": 7.686191879129853e-05,
"loss": 0.0273,
"step": 211
},
{
"epoch": 0.34,
"learning_rate": 7.664111063060158e-05,
"loss": 0.0233,
"step": 212
},
{
"epoch": 0.34,
"learning_rate": 7.641957451891886e-05,
"loss": 0.0255,
"step": 213
},
{
"epoch": 0.34,
"learning_rate": 7.619731650958023e-05,
"loss": 0.0295,
"step": 214
},
{
"epoch": 0.35,
"learning_rate": 7.597434267564097e-05,
"loss": 0.0219,
"step": 215
},
{
"epoch": 0.35,
"learning_rate": 7.575065910971575e-05,
"loss": 0.0297,
"step": 216
},
{
"epoch": 0.35,
"learning_rate": 7.552627192381227e-05,
"loss": 0.0203,
"step": 217
},
{
"epoch": 0.35,
"learning_rate": 7.530118724916411e-05,
"loss": 0.0254,
"step": 218
},
{
"epoch": 0.35,
"learning_rate": 7.507541123606333e-05,
"loss": 0.0302,
"step": 219
},
{
"epoch": 0.35,
"learning_rate": 7.484895005369237e-05,
"loss": 0.0165,
"step": 220
},
{
"epoch": 0.36,
"learning_rate": 7.462180988995542e-05,
"loss": 0.0303,
"step": 221
},
{
"epoch": 0.36,
"learning_rate": 7.439399695130945e-05,
"loss": 0.0234,
"step": 222
},
{
"epoch": 0.36,
"learning_rate": 7.416551746259452e-05,
"loss": 0.0303,
"step": 223
},
{
"epoch": 0.36,
"learning_rate": 7.393637766686378e-05,
"loss": 0.0273,
"step": 224
},
{
"epoch": 0.36,
"learning_rate": 7.37065838252128e-05,
"loss": 0.0294,
"step": 225
},
{
"epoch": 0.36,
"learning_rate": 7.347614221660855e-05,
"loss": 0.0264,
"step": 226
},
{
"epoch": 0.37,
"learning_rate": 7.32450591377178e-05,
"loss": 0.0227,
"step": 227
},
{
"epoch": 0.37,
"learning_rate": 7.301334090273509e-05,
"loss": 0.0267,
"step": 228
},
{
"epoch": 0.37,
"learning_rate": 7.278099384321017e-05,
"loss": 0.0334,
"step": 229
},
{
"epoch": 0.37,
"learning_rate": 7.254802430787502e-05,
"loss": 0.0299,
"step": 230
},
{
"epoch": 0.37,
"learning_rate": 7.231443866247039e-05,
"loss": 0.027,
"step": 231
},
{
"epoch": 0.37,
"learning_rate": 7.208024328957176e-05,
"loss": 0.023,
"step": 232
},
{
"epoch": 0.38,
"learning_rate": 7.184544458841513e-05,
"loss": 0.0193,
"step": 233
},
{
"epoch": 0.38,
"learning_rate": 7.161004897472192e-05,
"loss": 0.0223,
"step": 234
},
{
"epoch": 0.38,
"learning_rate": 7.137406288052393e-05,
"loss": 0.0197,
"step": 235
},
{
"epoch": 0.38,
"learning_rate": 7.113749275398735e-05,
"loss": 0.0251,
"step": 236
},
{
"epoch": 0.38,
"learning_rate": 7.090034505923673e-05,
"loss": 0.024,
"step": 237
},
{
"epoch": 0.38,
"learning_rate": 7.066262627617826e-05,
"loss": 0.0219,
"step": 238
},
{
"epoch": 0.39,
"learning_rate": 7.042434290032279e-05,
"loss": 0.0224,
"step": 239
},
{
"epoch": 0.39,
"learning_rate": 7.018550144260827e-05,
"loss": 0.0296,
"step": 240
},
{
"epoch": 0.39,
"learning_rate": 6.994610842922184e-05,
"loss": 0.0258,
"step": 241
},
{
"epoch": 0.39,
"learning_rate": 6.970617040142164e-05,
"loss": 0.0265,
"step": 242
},
{
"epoch": 0.39,
"learning_rate": 6.946569391535785e-05,
"loss": 0.0228,
"step": 243
},
{
"epoch": 0.39,
"learning_rate": 6.922468554189377e-05,
"loss": 0.0272,
"step": 244
},
{
"epoch": 0.39,
"learning_rate": 6.898315186642612e-05,
"loss": 0.0222,
"step": 245
},
{
"epoch": 0.4,
"learning_rate": 6.874109948870517e-05,
"loss": 0.0245,
"step": 246
},
{
"epoch": 0.4,
"learning_rate": 6.849853502265442e-05,
"loss": 0.0253,
"step": 247
},
{
"epoch": 0.4,
"learning_rate": 6.825546509618978e-05,
"loss": 0.0273,
"step": 248
},
{
"epoch": 0.4,
"learning_rate": 6.801189635103864e-05,
"loss": 0.0277,
"step": 249
},
{
"epoch": 0.4,
"learning_rate": 6.776783544255817e-05,
"loss": 0.0183,
"step": 250
},
{
"epoch": 0.4,
"learning_rate": 6.752328903955367e-05,
"loss": 0.0205,
"step": 251
},
{
"epoch": 0.41,
"learning_rate": 6.72782638240962e-05,
"loss": 0.0277,
"step": 252
},
{
"epoch": 0.41,
"learning_rate": 6.703276649134013e-05,
"loss": 0.0221,
"step": 253
},
{
"epoch": 0.41,
"learning_rate": 6.678680374934004e-05,
"loss": 0.0241,
"step": 254
},
{
"epoch": 0.41,
"learning_rate": 6.654038231886759e-05,
"loss": 0.0219,
"step": 255
},
{
"epoch": 0.41,
"learning_rate": 6.629350893322778e-05,
"loss": 0.0175,
"step": 256
},
{
"epoch": 0.41,
"learning_rate": 6.604619033807494e-05,
"loss": 0.0183,
"step": 257
},
{
"epoch": 0.42,
"learning_rate": 6.579843329122856e-05,
"loss": 0.0232,
"step": 258
},
{
"epoch": 0.42,
"learning_rate": 6.555024456248843e-05,
"loss": 0.0211,
"step": 259
},
{
"epoch": 0.42,
"learning_rate": 6.530163093344986e-05,
"loss": 0.0269,
"step": 260
},
{
"epoch": 0.42,
"learning_rate": 6.505259919731819e-05,
"loss": 0.0233,
"step": 261
},
{
"epoch": 0.42,
"learning_rate": 6.480315615872335e-05,
"loss": 0.0237,
"step": 262
},
{
"epoch": 0.42,
"learning_rate": 6.45533086335338e-05,
"loss": 0.0239,
"step": 263
},
{
"epoch": 0.43,
"learning_rate": 6.430306344867033e-05,
"loss": 0.0196,
"step": 264
},
{
"epoch": 0.43,
"learning_rate": 6.405242744191954e-05,
"loss": 0.0207,
"step": 265
},
{
"epoch": 0.43,
"learning_rate": 6.380140746174697e-05,
"loss": 0.018,
"step": 266
},
{
"epoch": 0.43,
"learning_rate": 6.355001036710998e-05,
"loss": 0.0219,
"step": 267
},
{
"epoch": 0.43,
"learning_rate": 6.329824302727035e-05,
"loss": 0.0152,
"step": 268
},
{
"epoch": 0.43,
"learning_rate": 6.304611232160657e-05,
"loss": 0.0255,
"step": 269
},
{
"epoch": 0.44,
"learning_rate": 6.279362513942584e-05,
"loss": 0.016,
"step": 270
},
{
"epoch": 0.44,
"learning_rate": 6.254078837977589e-05,
"loss": 0.0236,
"step": 271
},
{
"epoch": 0.44,
"learning_rate": 6.228760895125636e-05,
"loss": 0.0259,
"step": 272
},
{
"epoch": 0.44,
"learning_rate": 6.203409377183017e-05,
"loss": 0.0254,
"step": 273
},
{
"epoch": 0.44,
"learning_rate": 6.178024976863439e-05,
"loss": 0.0193,
"step": 274
},
{
"epoch": 0.44,
"learning_rate": 6.152608387779092e-05,
"loss": 0.0254,
"step": 275
},
{
"epoch": 0.44,
"learning_rate": 6.127160304421716e-05,
"loss": 0.0271,
"step": 276
},
{
"epoch": 0.45,
"learning_rate": 6.101681422143598e-05,
"loss": 0.0248,
"step": 277
},
{
"epoch": 0.45,
"learning_rate": 6.076172437138595e-05,
"loss": 0.0265,
"step": 278
},
{
"epoch": 0.45,
"learning_rate": 6.050634046423096e-05,
"loss": 0.024,
"step": 279
},
{
"epoch": 0.45,
"learning_rate": 6.025066947816985e-05,
"loss": 0.02,
"step": 280
},
{
"epoch": 0.45,
"learning_rate": 5.999471839924567e-05,
"loss": 0.0199,
"step": 281
},
{
"epoch": 0.45,
"learning_rate": 5.973849422115486e-05,
"loss": 0.0251,
"step": 282
},
{
"epoch": 0.46,
"learning_rate": 5.948200394505612e-05,
"loss": 0.022,
"step": 283
},
{
"epoch": 0.46,
"learning_rate": 5.9225254579379065e-05,
"loss": 0.0262,
"step": 284
},
{
"epoch": 0.46,
"learning_rate": 5.89682531396328e-05,
"loss": 0.0224,
"step": 285
},
{
"epoch": 0.46,
"learning_rate": 5.871100664821416e-05,
"loss": 0.0261,
"step": 286
},
{
"epoch": 0.46,
"learning_rate": 5.84535221342159e-05,
"loss": 0.0187,
"step": 287
},
{
"epoch": 0.46,
"learning_rate": 5.819580663323452e-05,
"loss": 0.0184,
"step": 288
},
{
"epoch": 0.47,
"learning_rate": 5.7937867187178163e-05,
"loss": 0.0289,
"step": 289
},
{
"epoch": 0.47,
"learning_rate": 5.7679710844074095e-05,
"loss": 0.0247,
"step": 290
},
{
"epoch": 0.47,
"learning_rate": 5.742134465787612e-05,
"loss": 0.0141,
"step": 291
},
{
"epoch": 0.47,
"learning_rate": 5.716277568827193e-05,
"loss": 0.016,
"step": 292
},
{
"epoch": 0.47,
"learning_rate": 5.690401100049008e-05,
"loss": 0.0288,
"step": 293
},
{
"epoch": 0.47,
"learning_rate": 5.6645057665107035e-05,
"loss": 0.0265,
"step": 294
},
{
"epoch": 0.48,
"learning_rate": 5.6385922757853915e-05,
"loss": 0.025,
"step": 295
},
{
"epoch": 0.48,
"learning_rate": 5.6126613359423206e-05,
"loss": 0.0163,
"step": 296
},
{
"epoch": 0.48,
"learning_rate": 5.5867136555275204e-05,
"loss": 0.0184,
"step": 297
},
{
"epoch": 0.48,
"learning_rate": 5.5607499435444476e-05,
"loss": 0.0248,
"step": 298
},
{
"epoch": 0.48,
"learning_rate": 5.534770909434613e-05,
"loss": 0.0279,
"step": 299
},
{
"epoch": 0.48,
"learning_rate": 5.508777263058191e-05,
"loss": 0.0207,
"step": 300
},
{
"epoch": 0.49,
"learning_rate": 5.482769714674631e-05,
"loss": 0.0206,
"step": 301
},
{
"epoch": 0.49,
"learning_rate": 5.4567489749232415e-05,
"loss": 0.0209,
"step": 302
},
{
"epoch": 0.49,
"learning_rate": 5.43071575480378e-05,
"loss": 0.0277,
"step": 303
},
{
"epoch": 0.49,
"learning_rate": 5.4046707656570174e-05,
"loss": 0.0249,
"step": 304
},
{
"epoch": 0.49,
"learning_rate": 5.378614719145312e-05,
"loss": 0.0197,
"step": 305
},
{
"epoch": 0.49,
"learning_rate": 5.35254832723315e-05,
"loss": 0.0177,
"step": 306
},
{
"epoch": 0.49,
"learning_rate": 5.3264723021677065e-05,
"loss": 0.0283,
"step": 307
},
{
"epoch": 0.5,
"learning_rate": 5.3003873564593696e-05,
"loss": 0.0205,
"step": 308
},
{
"epoch": 0.5,
"learning_rate": 5.274294202862281e-05,
"loss": 0.0131,
"step": 309
},
{
"epoch": 0.5,
"learning_rate": 5.248193554354861e-05,
"loss": 0.0157,
"step": 310
},
{
"epoch": 0.5,
"learning_rate": 5.222086124120317e-05,
"loss": 0.0258,
"step": 311
},
{
"epoch": 0.5,
"learning_rate": 5.1959726255271655e-05,
"loss": 0.0241,
"step": 312
},
{
"epoch": 0.5,
"learning_rate": 5.16985377210974e-05,
"loss": 0.018,
"step": 313
},
{
"epoch": 0.51,
"learning_rate": 5.143730277548685e-05,
"loss": 0.012,
"step": 314
},
{
"epoch": 0.51,
"learning_rate": 5.117602855651467e-05,
"loss": 0.0307,
"step": 315
},
{
"epoch": 0.51,
"learning_rate": 5.091472220332859e-05,
"loss": 0.0176,
"step": 316
},
{
"epoch": 0.51,
"learning_rate": 5.0653390855954465e-05,
"loss": 0.0212,
"step": 317
},
{
"epoch": 0.51,
"learning_rate": 5.0392041655100994e-05,
"loss": 0.0202,
"step": 318
},
{
"epoch": 0.51,
"learning_rate": 5.013068174196481e-05,
"loss": 0.023,
"step": 319
},
{
"epoch": 0.52,
"learning_rate": 4.9869318258035194e-05,
"loss": 0.0214,
"step": 320
},
{
"epoch": 0.52,
"learning_rate": 4.960795834489901e-05,
"loss": 0.0168,
"step": 321
},
{
"epoch": 0.52,
"learning_rate": 4.9346609144045546e-05,
"loss": 0.0176,
"step": 322
},
{
"epoch": 0.52,
"learning_rate": 4.908527779667142e-05,
"loss": 0.0191,
"step": 323
},
{
"epoch": 0.52,
"learning_rate": 4.882397144348534e-05,
"loss": 0.0211,
"step": 324
},
{
"epoch": 0.52,
"learning_rate": 4.856269722451316e-05,
"loss": 0.02,
"step": 325
},
{
"epoch": 0.53,
"learning_rate": 4.830146227890262e-05,
"loss": 0.022,
"step": 326
},
{
"epoch": 0.53,
"learning_rate": 4.804027374472835e-05,
"loss": 0.0205,
"step": 327
},
{
"epoch": 0.53,
"learning_rate": 4.777913875879685e-05,
"loss": 0.021,
"step": 328
},
{
"epoch": 0.53,
"learning_rate": 4.75180644564514e-05,
"loss": 0.0184,
"step": 329
},
{
"epoch": 0.53,
"learning_rate": 4.7257057971377195e-05,
"loss": 0.0214,
"step": 330
},
{
"epoch": 0.53,
"learning_rate": 4.6996126435406316e-05,
"loss": 0.0254,
"step": 331
},
{
"epoch": 0.54,
"learning_rate": 4.6735276978322946e-05,
"loss": 0.0187,
"step": 332
},
{
"epoch": 0.54,
"learning_rate": 4.64745167276685e-05,
"loss": 0.0133,
"step": 333
},
{
"epoch": 0.54,
"learning_rate": 4.621385280854689e-05,
"loss": 0.0192,
"step": 334
},
{
"epoch": 0.54,
"learning_rate": 4.595329234342983e-05,
"loss": 0.0272,
"step": 335
},
{
"epoch": 0.54,
"learning_rate": 4.569284245196222e-05,
"loss": 0.0204,
"step": 336
},
{
"epoch": 0.54,
"learning_rate": 4.54325102507676e-05,
"loss": 0.0188,
"step": 337
},
{
"epoch": 0.54,
"learning_rate": 4.51723028532537e-05,
"loss": 0.0194,
"step": 338
},
{
"epoch": 0.55,
"learning_rate": 4.4912227369418106e-05,
"loss": 0.0243,
"step": 339
},
{
"epoch": 0.55,
"learning_rate": 4.4652290905653894e-05,
"loss": 0.0164,
"step": 340
},
{
"epoch": 0.55,
"learning_rate": 4.439250056455554e-05,
"loss": 0.0153,
"step": 341
},
{
"epoch": 0.55,
"learning_rate": 4.413286344472483e-05,
"loss": 0.0227,
"step": 342
},
{
"epoch": 0.55,
"learning_rate": 4.387338664057681e-05,
"loss": 0.0217,
"step": 343
},
{
"epoch": 0.55,
"learning_rate": 4.36140772421461e-05,
"loss": 0.0298,
"step": 344
},
{
"epoch": 0.56,
"learning_rate": 4.335494233489299e-05,
"loss": 0.0179,
"step": 345
},
{
"epoch": 0.56,
"learning_rate": 4.309598899950995e-05,
"loss": 0.019,
"step": 346
},
{
"epoch": 0.56,
"learning_rate": 4.28372243117281e-05,
"loss": 0.0213,
"step": 347
},
{
"epoch": 0.56,
"learning_rate": 4.257865534212389e-05,
"loss": 0.0149,
"step": 348
},
{
"epoch": 0.56,
"learning_rate": 4.232028915592593e-05,
"loss": 0.0198,
"step": 349
},
{
"epoch": 0.56,
"learning_rate": 4.206213281282185e-05,
"loss": 0.0118,
"step": 350
},
{
"epoch": 0.57,
"learning_rate": 4.18041933667655e-05,
"loss": 0.0174,
"step": 351
},
{
"epoch": 0.57,
"learning_rate": 4.1546477865784125e-05,
"loss": 0.0184,
"step": 352
},
{
"epoch": 0.57,
"learning_rate": 4.1288993351785866e-05,
"loss": 0.0153,
"step": 353
},
{
"epoch": 0.57,
"learning_rate": 4.103174686036722e-05,
"loss": 0.0156,
"step": 354
},
{
"epoch": 0.57,
"learning_rate": 4.077474542062096e-05,
"loss": 0.0162,
"step": 355
},
{
"epoch": 0.57,
"learning_rate": 4.05179960549439e-05,
"loss": 0.0203,
"step": 356
},
{
"epoch": 0.58,
"learning_rate": 4.026150577884514e-05,
"loss": 0.0172,
"step": 357
},
{
"epoch": 0.58,
"learning_rate": 4.000528160075433e-05,
"loss": 0.0138,
"step": 358
},
{
"epoch": 0.58,
"learning_rate": 3.974933052183016e-05,
"loss": 0.0164,
"step": 359
},
{
"epoch": 0.58,
"learning_rate": 3.949365953576904e-05,
"loss": 0.0147,
"step": 360
},
{
"epoch": 0.58,
"learning_rate": 3.9238275628614055e-05,
"loss": 0.0262,
"step": 361
},
{
"epoch": 0.58,
"learning_rate": 3.898318577856401e-05,
"loss": 0.0158,
"step": 362
},
{
"epoch": 0.59,
"learning_rate": 3.872839695578284e-05,
"loss": 0.0136,
"step": 363
},
{
"epoch": 0.59,
"learning_rate": 3.847391612220906e-05,
"loss": 0.0217,
"step": 364
},
{
"epoch": 0.59,
"learning_rate": 3.821975023136561e-05,
"loss": 0.0205,
"step": 365
},
{
"epoch": 0.59,
"learning_rate": 3.7965906228169824e-05,
"loss": 0.0206,
"step": 366
},
{
"epoch": 0.59,
"learning_rate": 3.771239104874363e-05,
"loss": 0.0197,
"step": 367
},
{
"epoch": 0.59,
"learning_rate": 3.745921162022412e-05,
"loss": 0.02,
"step": 368
},
{
"epoch": 0.59,
"learning_rate": 3.720637486057416e-05,
"loss": 0.0179,
"step": 369
},
{
"epoch": 0.6,
"learning_rate": 3.695388767839343e-05,
"loss": 0.021,
"step": 370
},
{
"epoch": 0.6,
"learning_rate": 3.670175697272965e-05,
"loss": 0.0145,
"step": 371
},
{
"epoch": 0.6,
"learning_rate": 3.6449989632890016e-05,
"loss": 0.0186,
"step": 372
},
{
"epoch": 0.6,
"learning_rate": 3.619859253825304e-05,
"loss": 0.0111,
"step": 373
},
{
"epoch": 0.6,
"learning_rate": 3.594757255808047e-05,
"loss": 0.0146,
"step": 374
},
{
"epoch": 0.6,
"learning_rate": 3.5696936551329685e-05,
"loss": 0.0197,
"step": 375
},
{
"epoch": 0.61,
"learning_rate": 3.544669136646621e-05,
"loss": 0.0123,
"step": 376
},
{
"epoch": 0.61,
"learning_rate": 3.519684384127667e-05,
"loss": 0.0215,
"step": 377
},
{
"epoch": 0.61,
"learning_rate": 3.494740080268183e-05,
"loss": 0.0173,
"step": 378
},
{
"epoch": 0.61,
"learning_rate": 3.469836906655016e-05,
"loss": 0.0183,
"step": 379
},
{
"epoch": 0.61,
"learning_rate": 3.4449755437511586e-05,
"loss": 0.0175,
"step": 380
},
{
"epoch": 0.61,
"learning_rate": 3.4201566708771456e-05,
"loss": 0.0142,
"step": 381
},
{
"epoch": 0.62,
"learning_rate": 3.395380966192507e-05,
"loss": 0.0186,
"step": 382
},
{
"epoch": 0.62,
"learning_rate": 3.370649106677224e-05,
"loss": 0.0186,
"step": 383
},
{
"epoch": 0.62,
"learning_rate": 3.345961768113242e-05,
"loss": 0.0186,
"step": 384
},
{
"epoch": 0.62,
"learning_rate": 3.3213196250659965e-05,
"loss": 0.0111,
"step": 385
},
{
"epoch": 0.62,
"learning_rate": 3.296723350865988e-05,
"loss": 0.0162,
"step": 386
},
{
"epoch": 0.62,
"learning_rate": 3.272173617590381e-05,
"loss": 0.0185,
"step": 387
},
{
"epoch": 0.63,
"learning_rate": 3.247671096044634e-05,
"loss": 0.017,
"step": 388
},
{
"epoch": 0.63,
"learning_rate": 3.223216455744184e-05,
"loss": 0.009,
"step": 389
},
{
"epoch": 0.63,
"learning_rate": 3.1988103648961375e-05,
"loss": 0.0144,
"step": 390
},
{
"epoch": 0.63,
"learning_rate": 3.174453490381022e-05,
"loss": 0.0168,
"step": 391
},
{
"epoch": 0.63,
"learning_rate": 3.1501464977345596e-05,
"loss": 0.02,
"step": 392
},
{
"epoch": 0.63,
"learning_rate": 3.125890051129484e-05,
"loss": 0.0176,
"step": 393
},
{
"epoch": 0.63,
"learning_rate": 3.101684813357389e-05,
"loss": 0.0199,
"step": 394
},
{
"epoch": 0.64,
"learning_rate": 3.0775314458106245e-05,
"loss": 0.0123,
"step": 395
},
{
"epoch": 0.64,
"learning_rate": 3.053430608464216e-05,
"loss": 0.0149,
"step": 396
},
{
"epoch": 0.64,
"learning_rate": 3.029382959857837e-05,
"loss": 0.0172,
"step": 397
},
{
"epoch": 0.64,
"learning_rate": 3.0053891570778165e-05,
"loss": 0.0163,
"step": 398
},
{
"epoch": 0.64,
"learning_rate": 2.9814498557391747e-05,
"loss": 0.014,
"step": 399
},
{
"epoch": 0.64,
"learning_rate": 2.9575657099677222e-05,
"loss": 0.0171,
"step": 400
},
{
"epoch": 0.65,
"learning_rate": 2.9337373723821747e-05,
"loss": 0.0134,
"step": 401
},
{
"epoch": 0.65,
"learning_rate": 2.909965494076329e-05,
"loss": 0.0114,
"step": 402
},
{
"epoch": 0.65,
"learning_rate": 2.886250724601266e-05,
"loss": 0.0156,
"step": 403
},
{
"epoch": 0.65,
"learning_rate": 2.8625937119476082e-05,
"loss": 0.0203,
"step": 404
},
{
"epoch": 0.65,
"learning_rate": 2.838995102527808e-05,
"loss": 0.0153,
"step": 405
},
{
"epoch": 0.65,
"learning_rate": 2.8154555411584903e-05,
"loss": 0.0156,
"step": 406
},
{
"epoch": 0.66,
"learning_rate": 2.7919756710428247e-05,
"loss": 0.013,
"step": 407
},
{
"epoch": 0.66,
"learning_rate": 2.768556133752963e-05,
"loss": 0.0195,
"step": 408
},
{
"epoch": 0.66,
"learning_rate": 2.7451975692124976e-05,
"loss": 0.0166,
"step": 409
},
{
"epoch": 0.66,
"learning_rate": 2.7219006156789828e-05,
"loss": 0.0186,
"step": 410
},
{
"epoch": 0.66,
"learning_rate": 2.6986659097264934e-05,
"loss": 0.0158,
"step": 411
},
{
"epoch": 0.66,
"learning_rate": 2.6754940862282213e-05,
"loss": 0.0156,
"step": 412
},
{
"epoch": 0.67,
"learning_rate": 2.652385778339146e-05,
"loss": 0.0143,
"step": 413
},
{
"epoch": 0.67,
"learning_rate": 2.6293416174787205e-05,
"loss": 0.0194,
"step": 414
},
{
"epoch": 0.67,
"learning_rate": 2.6063622333136238e-05,
"loss": 0.0138,
"step": 415
},
{
"epoch": 0.67,
"learning_rate": 2.583448253740549e-05,
"loss": 0.014,
"step": 416
},
{
"epoch": 0.67,
"learning_rate": 2.5606003048690564e-05,
"loss": 0.0109,
"step": 417
},
{
"epoch": 0.67,
"learning_rate": 2.5378190110044586e-05,
"loss": 0.0175,
"step": 418
},
{
"epoch": 0.68,
"learning_rate": 2.5151049946307636e-05,
"loss": 0.0121,
"step": 419
},
{
"epoch": 0.68,
"learning_rate": 2.492458876393669e-05,
"loss": 0.0097,
"step": 420
},
{
"epoch": 0.68,
"learning_rate": 2.4698812750835914e-05,
"loss": 0.0182,
"step": 421
},
{
"epoch": 0.68,
"learning_rate": 2.4473728076187756e-05,
"loss": 0.0129,
"step": 422
},
{
"epoch": 0.68,
"learning_rate": 2.424934089028426e-05,
"loss": 0.0165,
"step": 423
},
{
"epoch": 0.68,
"learning_rate": 2.4025657324359065e-05,
"loss": 0.0112,
"step": 424
},
{
"epoch": 0.68,
"learning_rate": 2.3802683490419793e-05,
"loss": 0.0102,
"step": 425
},
{
"epoch": 0.69,
"learning_rate": 2.3580425481081165e-05,
"loss": 0.0153,
"step": 426
},
{
"epoch": 0.69,
"learning_rate": 2.3358889369398442e-05,
"loss": 0.0107,
"step": 427
},
{
"epoch": 0.69,
"learning_rate": 2.3138081208701484e-05,
"loss": 0.0131,
"step": 428
},
{
"epoch": 0.69,
"learning_rate": 2.2918007032429418e-05,
"loss": 0.0152,
"step": 429
},
{
"epoch": 0.69,
"learning_rate": 2.2698672853965634e-05,
"loss": 0.0197,
"step": 430
},
{
"epoch": 0.69,
"learning_rate": 2.248008466647365e-05,
"loss": 0.0103,
"step": 431
},
{
"epoch": 0.7,
"learning_rate": 2.2262248442733218e-05,
"loss": 0.016,
"step": 432
},
{
"epoch": 0.7,
"learning_rate": 2.20451701349772e-05,
"loss": 0.0127,
"step": 433
},
{
"epoch": 0.7,
"learning_rate": 2.1828855674728876e-05,
"loss": 0.012,
"step": 434
},
{
"epoch": 0.7,
"learning_rate": 2.16133109726399e-05,
"loss": 0.0183,
"step": 435
},
{
"epoch": 0.7,
"learning_rate": 2.1398541918328785e-05,
"loss": 0.0179,
"step": 436
},
{
"epoch": 0.7,
"learning_rate": 2.1184554380219996e-05,
"loss": 0.013,
"step": 437
},
{
"epoch": 0.71,
"learning_rate": 2.097135420538351e-05,
"loss": 0.0155,
"step": 438
},
{
"epoch": 0.71,
"learning_rate": 2.075894721937518e-05,
"loss": 0.0156,
"step": 439
},
{
"epoch": 0.71,
"learning_rate": 2.0547339226077477e-05,
"loss": 0.0137,
"step": 440
},
{
"epoch": 0.71,
"learning_rate": 2.0336536007540907e-05,
"loss": 0.0122,
"step": 441
},
{
"epoch": 0.71,
"learning_rate": 2.012654332382605e-05,
"loss": 0.0131,
"step": 442
},
{
"epoch": 0.71,
"learning_rate": 1.9917366912846113e-05,
"loss": 0.011,
"step": 443
},
{
"epoch": 0.72,
"learning_rate": 1.9709012490210205e-05,
"loss": 0.0179,
"step": 444
},
{
"epoch": 0.72,
"learning_rate": 1.9501485749067145e-05,
"loss": 0.0098,
"step": 445
},
{
"epoch": 0.72,
"learning_rate": 1.929479235994991e-05,
"loss": 0.0079,
"step": 446
},
{
"epoch": 0.72,
"learning_rate": 1.908893797062061e-05,
"loss": 0.0142,
"step": 447
},
{
"epoch": 0.72,
"learning_rate": 1.8883928205916302e-05,
"loss": 0.0078,
"step": 448
},
{
"epoch": 0.72,
"learning_rate": 1.86797686675952e-05,
"loss": 0.0105,
"step": 449
},
{
"epoch": 0.73,
"learning_rate": 1.847646493418363e-05,
"loss": 0.0102,
"step": 450
},
{
"epoch": 0.73,
"learning_rate": 1.827402256082363e-05,
"loss": 0.0128,
"step": 451
},
{
"epoch": 0.73,
"learning_rate": 1.8072447079121092e-05,
"loss": 0.0057,
"step": 452
},
{
"epoch": 0.73,
"learning_rate": 1.78717439969947e-05,
"loss": 0.008,
"step": 453
},
{
"epoch": 0.73,
"learning_rate": 1.767191879852537e-05,
"loss": 0.0101,
"step": 454
},
{
"epoch": 0.73,
"learning_rate": 1.747297694380644e-05,
"loss": 0.0157,
"step": 455
},
{
"epoch": 0.73,
"learning_rate": 1.7274923868794412e-05,
"loss": 0.0133,
"step": 456
},
{
"epoch": 0.74,
"learning_rate": 1.7077764985160493e-05,
"loss": 0.012,
"step": 457
},
{
"epoch": 0.74,
"learning_rate": 1.68815056801427e-05,
"loss": 0.0125,
"step": 458
},
{
"epoch": 0.74,
"learning_rate": 1.668615131639863e-05,
"loss": 0.0116,
"step": 459
},
{
"epoch": 0.74,
"learning_rate": 1.6491707231858982e-05,
"loss": 0.0108,
"step": 460
},
{
"epoch": 0.74,
"learning_rate": 1.6298178739581606e-05,
"loss": 0.0149,
"step": 461
},
{
"epoch": 0.74,
"learning_rate": 1.610557112760645e-05,
"loss": 0.0133,
"step": 462
},
{
"epoch": 0.75,
"learning_rate": 1.5913889658810988e-05,
"loss": 0.0113,
"step": 463
},
{
"epoch": 0.75,
"learning_rate": 1.572313957076643e-05,
"loss": 0.0163,
"step": 464
},
{
"epoch": 0.75,
"learning_rate": 1.5533326075594638e-05,
"loss": 0.0132,
"step": 465
},
{
"epoch": 0.75,
"learning_rate": 1.534445435982564e-05,
"loss": 0.0117,
"step": 466
},
{
"epoch": 0.75,
"learning_rate": 1.515652958425598e-05,
"loss": 0.0187,
"step": 467
},
{
"epoch": 0.75,
"learning_rate": 1.496955688380769e-05,
"loss": 0.0149,
"step": 468
},
{
"epoch": 0.76,
"learning_rate": 1.4783541367387971e-05,
"loss": 0.0112,
"step": 469
},
{
"epoch": 0.76,
"learning_rate": 1.4598488117749548e-05,
"loss": 0.0118,
"step": 470
},
{
"epoch": 0.76,
"learning_rate": 1.441440219135189e-05,
"loss": 0.0125,
"step": 471
},
{
"epoch": 0.76,
"learning_rate": 1.423128861822296e-05,
"loss": 0.006,
"step": 472
},
{
"epoch": 0.76,
"learning_rate": 1.4049152401821803e-05,
"loss": 0.0088,
"step": 473
},
{
"epoch": 0.76,
"learning_rate": 1.386799851890186e-05,
"loss": 0.0092,
"step": 474
},
{
"epoch": 0.77,
"learning_rate": 1.3687831919374876e-05,
"loss": 0.0111,
"step": 475
},
{
"epoch": 0.77,
"learning_rate": 1.35086575261758e-05,
"loss": 0.0072,
"step": 476
},
{
"epoch": 0.77,
"learning_rate": 1.333048023512814e-05,
"loss": 0.0102,
"step": 477
},
{
"epoch": 0.77,
"learning_rate": 1.3153304914810283e-05,
"loss": 0.0072,
"step": 478
},
{
"epoch": 0.77,
"learning_rate": 1.2977136406422353e-05,
"loss": 0.0105,
"step": 479
},
{
"epoch": 0.77,
"learning_rate": 1.280197952365405e-05,
"loss": 0.0162,
"step": 480
},
{
"epoch": 0.78,
"learning_rate": 1.2627839052553042e-05,
"loss": 0.0121,
"step": 481
},
{
"epoch": 0.78,
"learning_rate": 1.2454719751394211e-05,
"loss": 0.0138,
"step": 482
},
{
"epoch": 0.78,
"learning_rate": 1.2282626350549658e-05,
"loss": 0.0093,
"step": 483
},
{
"epoch": 0.78,
"learning_rate": 1.2111563552359378e-05,
"loss": 0.0104,
"step": 484
},
{
"epoch": 0.78,
"learning_rate": 1.1941536031002859e-05,
"loss": 0.0101,
"step": 485
},
{
"epoch": 0.78,
"learning_rate": 1.1772548432371311e-05,
"loss": 0.0115,
"step": 486
},
{
"epoch": 0.78,
"learning_rate": 1.1604605373940758e-05,
"loss": 0.0083,
"step": 487
},
{
"epoch": 0.79,
"learning_rate": 1.1437711444645793e-05,
"loss": 0.0111,
"step": 488
},
{
"epoch": 0.79,
"learning_rate": 1.1271871204754276e-05,
"loss": 0.011,
"step": 489
},
{
"epoch": 0.79,
"learning_rate": 1.1107089185742686e-05,
"loss": 0.0133,
"step": 490
},
{
"epoch": 0.79,
"learning_rate": 1.0943369890172301e-05,
"loss": 0.008,
"step": 491
},
{
"epoch": 0.79,
"learning_rate": 1.0780717791566197e-05,
"loss": 0.0163,
"step": 492
},
{
"epoch": 0.79,
"learning_rate": 1.0619137334286927e-05,
"loss": 0.0095,
"step": 493
},
{
"epoch": 0.8,
"learning_rate": 1.0458632933415208e-05,
"loss": 0.0124,
"step": 494
},
{
"epoch": 0.8,
"learning_rate": 1.0299208974629182e-05,
"loss": 0.0112,
"step": 495
},
{
"epoch": 0.8,
"learning_rate": 1.0140869814084636e-05,
"loss": 0.0095,
"step": 496
},
{
"epoch": 0.8,
"learning_rate": 9.983619778295945e-06,
"loss": 0.0108,
"step": 497
},
{
"epoch": 0.8,
"learning_rate": 9.827463164017836e-06,
"loss": 0.0152,
"step": 498
},
{
"epoch": 0.8,
"learning_rate": 9.67240423812803e-06,
"loss": 0.0103,
"step": 499
},
{
"epoch": 0.81,
"learning_rate": 9.518447237510636e-06,
"loss": 0.0076,
"step": 500
},
{
"epoch": 0.81,
"learning_rate": 9.365596368940365e-06,
"loss": 0.0099,
"step": 501
},
{
"epoch": 0.81,
"learning_rate": 9.213855808967565e-06,
"loss": 0.0091,
"step": 502
},
{
"epoch": 0.81,
"learning_rate": 9.063229703804154e-06,
"loss": 0.0168,
"step": 503
},
{
"epoch": 0.81,
"learning_rate": 8.913722169210304e-06,
"loss": 0.0139,
"step": 504
},
{
"epoch": 0.81,
"learning_rate": 8.765337290381948e-06,
"loss": 0.0084,
"step": 505
},
{
"epoch": 0.82,
"learning_rate": 8.618079121839217e-06,
"loss": 0.0056,
"step": 506
},
{
"epoch": 0.82,
"learning_rate": 8.471951687315572e-06,
"loss": 0.0137,
"step": 507
},
{
"epoch": 0.82,
"learning_rate": 8.326958979647926e-06,
"loss": 0.0076,
"step": 508
},
{
"epoch": 0.82,
"learning_rate": 8.183104960667564e-06,
"loss": 0.0095,
"step": 509
},
{
"epoch": 0.82,
"learning_rate": 8.040393561091764e-06,
"loss": 0.0141,
"step": 510
},
{
"epoch": 0.82,
"learning_rate": 7.898828680416526e-06,
"loss": 0.0108,
"step": 511
},
{
"epoch": 0.83,
"learning_rate": 7.758414186809959e-06,
"loss": 0.0115,
"step": 512
},
{
"epoch": 0.83,
"learning_rate": 7.6191539170065826e-06,
"loss": 0.0087,
"step": 513
},
{
"epoch": 0.83,
"learning_rate": 7.481051676202538e-06,
"loss": 0.0063,
"step": 514
},
{
"epoch": 0.83,
"learning_rate": 7.3441112379515175e-06,
"loss": 0.0122,
"step": 515
},
{
"epoch": 0.83,
"learning_rate": 7.208336344061767e-06,
"loss": 0.0078,
"step": 516
},
{
"epoch": 0.83,
"learning_rate": 7.0737307044937805e-06,
"loss": 0.0084,
"step": 517
},
{
"epoch": 0.83,
"learning_rate": 6.940297997258938e-06,
"loss": 0.0095,
"step": 518
},
{
"epoch": 0.84,
"learning_rate": 6.808041868318993e-06,
"loss": 0.012,
"step": 519
},
{
"epoch": 0.84,
"learning_rate": 6.676965931486479e-06,
"loss": 0.0079,
"step": 520
},
{
"epoch": 0.84,
"learning_rate": 6.547073768325956e-06,
"loss": 0.0065,
"step": 521
},
{
"epoch": 0.84,
"learning_rate": 6.418368928056123e-06,
"loss": 0.0122,
"step": 522
},
{
"epoch": 0.84,
"learning_rate": 6.290854927452872e-06,
"loss": 0.0097,
"step": 523
},
{
"epoch": 0.84,
"learning_rate": 6.164535250753156e-06,
"loss": 0.0077,
"step": 524
},
{
"epoch": 0.85,
"learning_rate": 6.0394133495598195e-06,
"loss": 0.0069,
"step": 525
},
{
"epoch": 0.85,
"learning_rate": 5.9154926427472725e-06,
"loss": 0.0085,
"step": 526
},
{
"epoch": 0.85,
"learning_rate": 5.79277651636807e-06,
"loss": 0.0118,
"step": 527
},
{
"epoch": 0.85,
"learning_rate": 5.671268323560397e-06,
"loss": 0.0098,
"step": 528
},
{
"epoch": 0.85,
"learning_rate": 5.550971384456416e-06,
"loss": 0.0077,
"step": 529
},
{
"epoch": 0.85,
"learning_rate": 5.431888986091604e-06,
"loss": 0.007,
"step": 530
},
{
"epoch": 0.86,
"learning_rate": 5.31402438231488e-06,
"loss": 0.0127,
"step": 531
},
{
"epoch": 0.86,
"learning_rate": 5.197380793699746e-06,
"loss": 0.0076,
"step": 532
},
{
"epoch": 0.86,
"learning_rate": 5.0819614074562205e-06,
"loss": 0.0069,
"step": 533
},
{
"epoch": 0.86,
"learning_rate": 4.967769377343828e-06,
"loss": 0.0089,
"step": 534
},
{
"epoch": 0.86,
"learning_rate": 4.8548078235853665e-06,
"loss": 0.0077,
"step": 535
},
{
"epoch": 0.86,
"learning_rate": 4.743079832781672e-06,
"loss": 0.0077,
"step": 536
},
{
"epoch": 0.87,
"learning_rate": 4.632588457827297e-06,
"loss": 0.0116,
"step": 537
},
{
"epoch": 0.87,
"learning_rate": 4.523336717827026e-06,
"loss": 0.0182,
"step": 538
},
{
"epoch": 0.87,
"learning_rate": 4.415327598013469e-06,
"loss": 0.0088,
"step": 539
},
{
"epoch": 0.87,
"learning_rate": 4.308564049665437e-06,
"loss": 0.0054,
"step": 540
},
{
"epoch": 0.87,
"learning_rate": 4.203048990027314e-06,
"loss": 0.0073,
"step": 541
},
{
"epoch": 0.87,
"learning_rate": 4.098785302229325e-06,
"loss": 0.0063,
"step": 542
},
{
"epoch": 0.88,
"learning_rate": 3.995775835208798e-06,
"loss": 0.0059,
"step": 543
},
{
"epoch": 0.88,
"learning_rate": 3.894023403632285e-06,
"loss": 0.0134,
"step": 544
},
{
"epoch": 0.88,
"learning_rate": 3.7935307878186664e-06,
"loss": 0.0096,
"step": 545
},
{
"epoch": 0.88,
"learning_rate": 3.694300733663186e-06,
"loss": 0.0134,
"step": 546
},
{
"epoch": 0.88,
"learning_rate": 3.596335952562396e-06,
"loss": 0.0062,
"step": 547
},
{
"epoch": 0.88,
"learning_rate": 3.4996391213400925e-06,
"loss": 0.0087,
"step": 548
},
{
"epoch": 0.88,
"learning_rate": 3.4042128821741736e-06,
"loss": 0.007,
"step": 549
},
{
"epoch": 0.89,
"learning_rate": 3.3100598425244376e-06,
"loss": 0.0068,
"step": 550
},
{
"epoch": 0.89,
"learning_rate": 3.2171825750613395e-06,
"loss": 0.0041,
"step": 551
},
{
"epoch": 0.89,
"learning_rate": 3.125583617595668e-06,
"loss": 0.0059,
"step": 552
},
{
"epoch": 0.89,
"learning_rate": 3.035265473009252e-06,
"loss": 0.009,
"step": 553
},
{
"epoch": 0.89,
"learning_rate": 2.9462306091865232e-06,
"loss": 0.0085,
"step": 554
},
{
"epoch": 0.89,
"learning_rate": 2.858481458947121e-06,
"loss": 0.0103,
"step": 555
},
{
"epoch": 0.9,
"learning_rate": 2.7720204199793763e-06,
"loss": 0.0102,
"step": 556
},
{
"epoch": 0.9,
"learning_rate": 2.6868498547748366e-06,
"loss": 0.0091,
"step": 557
},
{
"epoch": 0.9,
"learning_rate": 2.6029720905636933e-06,
"loss": 0.0071,
"step": 558
},
{
"epoch": 0.9,
"learning_rate": 2.52038941925119e-06,
"loss": 0.0089,
"step": 559
},
{
"epoch": 0.9,
"learning_rate": 2.439104097355005e-06,
"loss": 0.0072,
"step": 560
},
{
"epoch": 0.9,
"learning_rate": 2.359118345943567e-06,
"loss": 0.0072,
"step": 561
},
{
"epoch": 0.91,
"learning_rate": 2.280434350575422e-06,
"loss": 0.0066,
"step": 562
},
{
"epoch": 0.91,
"learning_rate": 2.2030542612394568e-06,
"loss": 0.0052,
"step": 563
},
{
"epoch": 0.91,
"learning_rate": 2.1269801922961908e-06,
"loss": 0.008,
"step": 564
},
{
"epoch": 0.91,
"learning_rate": 2.0522142224199735e-06,
"loss": 0.0086,
"step": 565
},
{
"epoch": 0.91,
"learning_rate": 1.978758394542207e-06,
"loss": 0.0075,
"step": 566
},
{
"epoch": 0.91,
"learning_rate": 1.9066147157955128e-06,
"loss": 0.0162,
"step": 567
},
{
"epoch": 0.92,
"learning_rate": 1.835785157458908e-06,
"loss": 0.009,
"step": 568
},
{
"epoch": 0.92,
"learning_rate": 1.766271654903906e-06,
"loss": 0.0104,
"step": 569
},
{
"epoch": 0.92,
"learning_rate": 1.6980761075416629e-06,
"loss": 0.0053,
"step": 570
},
{
"epoch": 0.92,
"learning_rate": 1.6312003787710639e-06,
"loss": 0.0074,
"step": 571
},
{
"epoch": 0.92,
"learning_rate": 1.565646295927814e-06,
"loss": 0.0081,
"step": 572
},
{
"epoch": 0.92,
"learning_rate": 1.5014156502345167e-06,
"loss": 0.0057,
"step": 573
},
{
"epoch": 0.93,
"learning_rate": 1.4385101967516911e-06,
"loss": 0.0079,
"step": 574
},
{
"epoch": 0.93,
"learning_rate": 1.3769316543298604e-06,
"loss": 0.0057,
"step": 575
},
{
"epoch": 0.93,
"learning_rate": 1.316681705562567e-06,
"loss": 0.0096,
"step": 576
},
{
"epoch": 0.93,
"learning_rate": 1.257761996740403e-06,
"loss": 0.005,
"step": 577
},
{
"epoch": 0.93,
"learning_rate": 1.2001741378060095e-06,
"loss": 0.0081,
"step": 578
},
{
"epoch": 0.93,
"learning_rate": 1.1439197023100923e-06,
"loss": 0.0129,
"step": 579
},
{
"epoch": 0.93,
"learning_rate": 1.0890002273684474e-06,
"loss": 0.0107,
"step": 580
},
{
"epoch": 0.94,
"learning_rate": 1.035417213619938e-06,
"loss": 0.0104,
"step": 581
},
{
"epoch": 0.94,
"learning_rate": 9.831721251854885e-07,
"loss": 0.0083,
"step": 582
},
{
"epoch": 0.94,
"learning_rate": 9.322663896281003e-07,
"loss": 0.0084,
"step": 583
},
{
"epoch": 0.94,
"learning_rate": 8.827013979138099e-07,
"loss": 0.0099,
"step": 584
},
{
"epoch": 0.94,
"learning_rate": 8.344785043737257e-07,
"loss": 0.007,
"step": 585
},
{
"epoch": 0.94,
"learning_rate": 7.875990266669852e-07,
"loss": 0.007,
"step": 586
},
{
"epoch": 0.95,
"learning_rate": 7.420642457447558e-07,
"loss": 0.0106,
"step": 587
},
{
"epoch": 0.95,
"learning_rate": 6.978754058152626e-07,
"loss": 0.0091,
"step": 588
},
{
"epoch": 0.95,
"learning_rate": 6.55033714309744e-07,
"loss": 0.0103,
"step": 589
},
{
"epoch": 0.95,
"learning_rate": 6.135403418495167e-07,
"loss": 0.0085,
"step": 590
},
{
"epoch": 0.95,
"learning_rate": 5.733964222139343e-07,
"loss": 0.0104,
"step": 591
},
{
"epoch": 0.95,
"learning_rate": 5.346030523094348e-07,
"loss": 0.0065,
"step": 592
},
{
"epoch": 0.96,
"learning_rate": 4.971612921395696e-07,
"loss": 0.0073,
"step": 593
},
{
"epoch": 0.96,
"learning_rate": 4.6107216477603856e-07,
"loss": 0.0063,
"step": 594
},
{
"epoch": 0.96,
"learning_rate": 4.263366563307281e-07,
"loss": 0.0101,
"step": 595
},
{
"epoch": 0.96,
"learning_rate": 3.9295571592876666e-07,
"loss": 0.007,
"step": 596
},
{
"epoch": 0.96,
"learning_rate": 3.6093025568259e-07,
"loss": 0.0068,
"step": 597
},
{
"epoch": 0.96,
"learning_rate": 3.302611506670383e-07,
"loss": 0.0096,
"step": 598
},
{
"epoch": 0.97,
"learning_rate": 3.0094923889541493e-07,
"loss": 0.0067,
"step": 599
},
{
"epoch": 0.97,
"learning_rate": 2.7299532129660965e-07,
"loss": 0.0068,
"step": 600
},
{
"epoch": 0.97,
"learning_rate": 2.4640016169319433e-07,
"loss": 0.0055,
"step": 601
},
{
"epoch": 0.97,
"learning_rate": 2.2116448678058398e-07,
"loss": 0.0071,
"step": 602
},
{
"epoch": 0.97,
"learning_rate": 1.9728898610713587e-07,
"loss": 0.0116,
"step": 603
},
{
"epoch": 0.97,
"learning_rate": 1.747743120553591e-07,
"loss": 0.0059,
"step": 604
},
{
"epoch": 0.98,
"learning_rate": 1.536210798240456e-07,
"loss": 0.0118,
"step": 605
},
{
"epoch": 0.98,
"learning_rate": 1.338298674114724e-07,
"loss": 0.0076,
"step": 606
},
{
"epoch": 0.98,
"learning_rate": 1.154012155996309e-07,
"loss": 0.0088,
"step": 607
},
{
"epoch": 0.98,
"learning_rate": 9.833562793941098e-08,
"loss": 0.0057,
"step": 608
},
{
"epoch": 0.98,
"learning_rate": 8.26335707368897e-08,
"loss": 0.0079,
"step": 609
},
{
"epoch": 0.98,
"learning_rate": 6.829547304053053e-08,
"loss": 0.0055,
"step": 610
},
{
"epoch": 0.98,
"learning_rate": 5.532172662952029e-08,
"loss": 0.0102,
"step": 611
},
{
"epoch": 0.99,
"learning_rate": 4.371268600301681e-08,
"loss": 0.0063,
"step": 612
},
{
"epoch": 0.99,
"learning_rate": 3.3468668370489895e-08,
"loss": 0.0052,
"step": 613
},
{
"epoch": 0.99,
"learning_rate": 2.458995364305605e-08,
"loss": 0.0091,
"step": 614
},
{
"epoch": 0.99,
"learning_rate": 1.7076784425795745e-08,
"loss": 0.0032,
"step": 615
},
{
"epoch": 0.99,
"learning_rate": 1.0929366011169784e-08,
"loss": 0.0086,
"step": 616
},
{
"epoch": 0.99,
"learning_rate": 6.147866373373834e-09,
"loss": 0.007,
"step": 617
},
{
"epoch": 1.0,
"learning_rate": 2.7324161637753977e-09,
"loss": 0.0092,
"step": 618
},
{
"epoch": 1.0,
"learning_rate": 6.831087073166931e-10,
"loss": 0.0044,
"step": 619
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.0031,
"step": 620
},
{
"epoch": 1.0,
"step": 620,
"total_flos": 0.0,
"train_loss": 0.0495266400213774,
"train_runtime": 18187.982,
"train_samples_per_second": 17.476,
"train_steps_per_second": 0.034
}
],
"logging_steps": 1.0,
"max_steps": 620,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 100,
"total_flos": 0.0,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}