pcllm_test1 / trainer_state.json
JoeyLin1's picture
fix: renormalize LFS files
3174b26
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9984,
"global_step": 468,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 2e-06,
"loss": 2.4992,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4e-06,
"loss": 2.5133,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 6e-06,
"loss": 1.9138,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 8e-06,
"loss": 1.3926,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 9.999999999999999e-06,
"loss": 1.04,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 1.2e-05,
"loss": 0.839,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 1.4e-05,
"loss": 0.7951,
"step": 7
},
{
"epoch": 0.02,
"learning_rate": 1.6e-05,
"loss": 0.776,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 1.8e-05,
"loss": 0.7137,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 1.9999999999999998e-05,
"loss": 0.6901,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 2.2e-05,
"loss": 0.6603,
"step": 11
},
{
"epoch": 0.03,
"learning_rate": 2.4e-05,
"loss": 0.6495,
"step": 12
},
{
"epoch": 0.03,
"learning_rate": 2.6000000000000002e-05,
"loss": 0.6096,
"step": 13
},
{
"epoch": 0.03,
"learning_rate": 2.8e-05,
"loss": 0.6309,
"step": 14
},
{
"epoch": 0.03,
"learning_rate": 3e-05,
"loss": 0.5983,
"step": 15
},
{
"epoch": 0.03,
"learning_rate": 2.999963928611156e-05,
"loss": 0.6011,
"step": 16
},
{
"epoch": 0.04,
"learning_rate": 2.9998557161794857e-05,
"loss": 0.5982,
"step": 17
},
{
"epoch": 0.04,
"learning_rate": 2.9996753679094844e-05,
"loss": 0.5863,
"step": 18
},
{
"epoch": 0.04,
"learning_rate": 2.9994228924750363e-05,
"loss": 0.5939,
"step": 19
},
{
"epoch": 0.04,
"learning_rate": 2.9990983020189943e-05,
"loss": 0.5674,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 2.9987016121525965e-05,
"loss": 0.574,
"step": 21
},
{
"epoch": 0.05,
"learning_rate": 2.9982328419547154e-05,
"loss": 0.5602,
"step": 22
},
{
"epoch": 0.05,
"learning_rate": 2.99769201397094e-05,
"loss": 0.5723,
"step": 23
},
{
"epoch": 0.05,
"learning_rate": 2.9970791542124925e-05,
"loss": 0.5662,
"step": 24
},
{
"epoch": 0.05,
"learning_rate": 2.9963942921549767e-05,
"loss": 0.5683,
"step": 25
},
{
"epoch": 0.06,
"learning_rate": 2.99563746073696e-05,
"loss": 0.5539,
"step": 26
},
{
"epoch": 0.06,
"learning_rate": 2.9948086963583895e-05,
"loss": 0.5468,
"step": 27
},
{
"epoch": 0.06,
"learning_rate": 2.9939080388788412e-05,
"loss": 0.555,
"step": 28
},
{
"epoch": 0.06,
"learning_rate": 2.9929355316156036e-05,
"loss": 0.5614,
"step": 29
},
{
"epoch": 0.06,
"learning_rate": 2.9918912213415933e-05,
"loss": 0.5401,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 2.9907751582831066e-05,
"loss": 0.5571,
"step": 31
},
{
"epoch": 0.07,
"learning_rate": 2.9895873961174027e-05,
"loss": 0.5562,
"step": 32
},
{
"epoch": 0.07,
"learning_rate": 2.9883279919701227e-05,
"loss": 0.5495,
"step": 33
},
{
"epoch": 0.07,
"learning_rate": 2.9869970064125424e-05,
"loss": 0.5486,
"step": 34
},
{
"epoch": 0.07,
"learning_rate": 2.9855945034586584e-05,
"loss": 0.5411,
"step": 35
},
{
"epoch": 0.08,
"learning_rate": 2.9841205505621104e-05,
"loss": 0.5376,
"step": 36
},
{
"epoch": 0.08,
"learning_rate": 2.9825752186129355e-05,
"loss": 0.5393,
"step": 37
},
{
"epoch": 0.08,
"learning_rate": 2.98095858193416e-05,
"loss": 0.532,
"step": 38
},
{
"epoch": 0.08,
"learning_rate": 2.979270718278224e-05,
"loss": 0.5242,
"step": 39
},
{
"epoch": 0.09,
"learning_rate": 2.977511708823242e-05,
"loss": 0.5452,
"step": 40
},
{
"epoch": 0.09,
"learning_rate": 2.9756816381691003e-05,
"loss": 0.5531,
"step": 41
},
{
"epoch": 0.09,
"learning_rate": 2.9737805943333857e-05,
"loss": 0.537,
"step": 42
},
{
"epoch": 0.09,
"learning_rate": 2.971808668747153e-05,
"loss": 0.5144,
"step": 43
},
{
"epoch": 0.09,
"learning_rate": 2.9697659562505286e-05,
"loss": 0.5443,
"step": 44
},
{
"epoch": 0.1,
"learning_rate": 2.9676525550881482e-05,
"loss": 0.5299,
"step": 45
},
{
"epoch": 0.1,
"learning_rate": 2.9654685669044313e-05,
"loss": 0.5374,
"step": 46
},
{
"epoch": 0.1,
"learning_rate": 2.963214096738695e-05,
"loss": 0.5354,
"step": 47
},
{
"epoch": 0.1,
"learning_rate": 2.960889253020099e-05,
"loss": 0.5374,
"step": 48
},
{
"epoch": 0.1,
"learning_rate": 2.9584941475624315e-05,
"loss": 0.5297,
"step": 49
},
{
"epoch": 0.11,
"learning_rate": 2.9560288955587334e-05,
"loss": 0.5401,
"step": 50
},
{
"epoch": 0.11,
"learning_rate": 2.9534936155757568e-05,
"loss": 0.5427,
"step": 51
},
{
"epoch": 0.11,
"learning_rate": 2.9508884295482606e-05,
"loss": 0.5443,
"step": 52
},
{
"epoch": 0.11,
"learning_rate": 2.94821346277315e-05,
"loss": 0.5414,
"step": 53
},
{
"epoch": 0.12,
"learning_rate": 2.9454688439034475e-05,
"loss": 0.5168,
"step": 54
},
{
"epoch": 0.12,
"learning_rate": 2.9426547049421047e-05,
"loss": 0.5337,
"step": 55
},
{
"epoch": 0.12,
"learning_rate": 2.9397711812356565e-05,
"loss": 0.5168,
"step": 56
},
{
"epoch": 0.12,
"learning_rate": 2.936818411467709e-05,
"loss": 0.5147,
"step": 57
},
{
"epoch": 0.12,
"learning_rate": 2.9337965376522717e-05,
"loss": 0.5343,
"step": 58
},
{
"epoch": 0.13,
"learning_rate": 2.9307057051269242e-05,
"loss": 0.5167,
"step": 59
},
{
"epoch": 0.13,
"learning_rate": 2.9275460625458298e-05,
"loss": 0.521,
"step": 60
},
{
"epoch": 0.13,
"learning_rate": 2.9243177618725826e-05,
"loss": 0.5258,
"step": 61
},
{
"epoch": 0.13,
"learning_rate": 2.921020958372902e-05,
"loss": 0.5284,
"step": 62
},
{
"epoch": 0.13,
"learning_rate": 2.9176558106071615e-05,
"loss": 0.5291,
"step": 63
},
{
"epoch": 0.14,
"learning_rate": 2.914222480422767e-05,
"loss": 0.5145,
"step": 64
},
{
"epoch": 0.14,
"learning_rate": 2.9107211329463685e-05,
"loss": 0.5349,
"step": 65
},
{
"epoch": 0.14,
"learning_rate": 2.9071519365759214e-05,
"loss": 0.5243,
"step": 66
},
{
"epoch": 0.14,
"learning_rate": 2.9035150629725858e-05,
"loss": 0.5288,
"step": 67
},
{
"epoch": 0.15,
"learning_rate": 2.8998106870524714e-05,
"loss": 0.5376,
"step": 68
},
{
"epoch": 0.15,
"learning_rate": 2.896038986978223e-05,
"loss": 0.5104,
"step": 69
},
{
"epoch": 0.15,
"learning_rate": 2.8922001441504544e-05,
"loss": 0.5206,
"step": 70
},
{
"epoch": 0.15,
"learning_rate": 2.888294343199022e-05,
"loss": 0.5061,
"step": 71
},
{
"epoch": 0.15,
"learning_rate": 2.8843217719741458e-05,
"loss": 0.5431,
"step": 72
},
{
"epoch": 0.16,
"learning_rate": 2.8802826215373742e-05,
"loss": 0.5365,
"step": 73
},
{
"epoch": 0.16,
"learning_rate": 2.876177086152395e-05,
"loss": 0.5281,
"step": 74
},
{
"epoch": 0.16,
"learning_rate": 2.8720053632756923e-05,
"loss": 0.5243,
"step": 75
},
{
"epoch": 0.16,
"learning_rate": 2.867767653547051e-05,
"loss": 0.5075,
"step": 76
},
{
"epoch": 0.16,
"learning_rate": 2.8634641607799046e-05,
"loss": 0.5178,
"step": 77
},
{
"epoch": 0.17,
"learning_rate": 2.859095091951534e-05,
"loss": 0.5498,
"step": 78
},
{
"epoch": 0.17,
"learning_rate": 2.8546606571931142e-05,
"loss": 0.5254,
"step": 79
},
{
"epoch": 0.17,
"learning_rate": 2.8501610697796044e-05,
"loss": 0.5181,
"step": 80
},
{
"epoch": 0.17,
"learning_rate": 2.8455965461194963e-05,
"loss": 0.5138,
"step": 81
},
{
"epoch": 0.17,
"learning_rate": 2.8409673057443985e-05,
"loss": 0.5192,
"step": 82
},
{
"epoch": 0.18,
"learning_rate": 2.8362735712984848e-05,
"loss": 0.5205,
"step": 83
},
{
"epoch": 0.18,
"learning_rate": 2.8315155685277816e-05,
"loss": 0.5093,
"step": 84
},
{
"epoch": 0.18,
"learning_rate": 2.8266935262693138e-05,
"loss": 0.5084,
"step": 85
},
{
"epoch": 0.18,
"learning_rate": 2.821807676440096e-05,
"loss": 0.5074,
"step": 86
},
{
"epoch": 0.19,
"learning_rate": 2.8168582540259803e-05,
"loss": 0.5122,
"step": 87
},
{
"epoch": 0.19,
"learning_rate": 2.811845497070354e-05,
"loss": 0.5265,
"step": 88
},
{
"epoch": 0.19,
"learning_rate": 2.806769646662691e-05,
"loss": 0.523,
"step": 89
},
{
"epoch": 0.19,
"learning_rate": 2.801630946926956e-05,
"loss": 0.5117,
"step": 90
},
{
"epoch": 0.19,
"learning_rate": 2.7964296450098646e-05,
"loss": 0.5034,
"step": 91
},
{
"epoch": 0.2,
"learning_rate": 2.7911659910689947e-05,
"loss": 0.5263,
"step": 92
},
{
"epoch": 0.2,
"learning_rate": 2.7858402382607577e-05,
"loss": 0.5161,
"step": 93
},
{
"epoch": 0.2,
"learning_rate": 2.78045264272822e-05,
"loss": 0.484,
"step": 94
},
{
"epoch": 0.2,
"learning_rate": 2.7750034635887874e-05,
"loss": 0.4986,
"step": 95
},
{
"epoch": 0.2,
"learning_rate": 2.769492962921738e-05,
"loss": 0.5019,
"step": 96
},
{
"epoch": 0.21,
"learning_rate": 2.7639214057556227e-05,
"loss": 0.5266,
"step": 97
},
{
"epoch": 0.21,
"learning_rate": 2.758289060055514e-05,
"loss": 0.5144,
"step": 98
},
{
"epoch": 0.21,
"learning_rate": 2.7525961967101216e-05,
"loss": 0.5084,
"step": 99
},
{
"epoch": 0.21,
"learning_rate": 2.7468430895187616e-05,
"loss": 0.5138,
"step": 100
},
{
"epoch": 0.22,
"learning_rate": 2.741030015178189e-05,
"loss": 0.5159,
"step": 101
},
{
"epoch": 0.22,
"learning_rate": 2.7351572532692916e-05,
"loss": 0.5136,
"step": 102
},
{
"epoch": 0.22,
"learning_rate": 2.7292250862436397e-05,
"loss": 0.5048,
"step": 103
},
{
"epoch": 0.22,
"learning_rate": 2.7232337994099044e-05,
"loss": 0.5153,
"step": 104
},
{
"epoch": 0.22,
"learning_rate": 2.7171836809201357e-05,
"loss": 0.5041,
"step": 105
},
{
"epoch": 0.23,
"learning_rate": 2.711075021755902e-05,
"loss": 0.5017,
"step": 106
},
{
"epoch": 0.23,
"learning_rate": 2.704908115714297e-05,
"loss": 0.5295,
"step": 107
},
{
"epoch": 0.23,
"learning_rate": 2.6986832593938088e-05,
"loss": 0.5036,
"step": 108
},
{
"epoch": 0.23,
"learning_rate": 2.6924007521800533e-05,
"loss": 0.5113,
"step": 109
},
{
"epoch": 0.23,
"learning_rate": 2.686060896231379e-05,
"loss": 0.5024,
"step": 110
},
{
"epoch": 0.24,
"learning_rate": 2.6796639964643306e-05,
"loss": 0.5115,
"step": 111
},
{
"epoch": 0.24,
"learning_rate": 2.6732103605389876e-05,
"loss": 0.5231,
"step": 112
},
{
"epoch": 0.24,
"learning_rate": 2.6667002988441638e-05,
"loss": 0.5252,
"step": 113
},
{
"epoch": 0.24,
"learning_rate": 2.660134124482482e-05,
"loss": 0.5157,
"step": 114
},
{
"epoch": 0.25,
"learning_rate": 2.6535121532553135e-05,
"loss": 0.5136,
"step": 115
},
{
"epoch": 0.25,
"learning_rate": 2.6468347036475902e-05,
"loss": 0.5108,
"step": 116
},
{
"epoch": 0.25,
"learning_rate": 2.6401020968124874e-05,
"loss": 0.5173,
"step": 117
},
{
"epoch": 0.25,
"learning_rate": 2.6333146565559775e-05,
"loss": 0.5163,
"step": 118
},
{
"epoch": 0.25,
"learning_rate": 2.6264727093212554e-05,
"loss": 0.5105,
"step": 119
},
{
"epoch": 0.26,
"learning_rate": 2.6195765841730404e-05,
"loss": 0.5158,
"step": 120
},
{
"epoch": 0.26,
"learning_rate": 2.6126266127817483e-05,
"loss": 0.5165,
"step": 121
},
{
"epoch": 0.26,
"learning_rate": 2.6056231294075393e-05,
"loss": 0.5177,
"step": 122
},
{
"epoch": 0.26,
"learning_rate": 2.5985664708842438e-05,
"loss": 0.493,
"step": 123
},
{
"epoch": 0.26,
"learning_rate": 2.5914569766031586e-05,
"loss": 0.5119,
"step": 124
},
{
"epoch": 0.27,
"learning_rate": 2.584294988496728e-05,
"loss": 0.5003,
"step": 125
},
{
"epoch": 0.27,
"learning_rate": 2.5770808510220957e-05,
"loss": 0.5101,
"step": 126
},
{
"epoch": 0.27,
"learning_rate": 2.569814911144539e-05,
"loss": 0.5188,
"step": 127
},
{
"epoch": 0.27,
"learning_rate": 2.5624975183207813e-05,
"loss": 0.5102,
"step": 128
},
{
"epoch": 0.28,
"learning_rate": 2.5551290244821856e-05,
"loss": 0.5066,
"step": 129
},
{
"epoch": 0.28,
"learning_rate": 2.547709784017826e-05,
"loss": 0.4941,
"step": 130
},
{
"epoch": 0.28,
"learning_rate": 2.5402401537574476e-05,
"loss": 0.4988,
"step": 131
},
{
"epoch": 0.28,
"learning_rate": 2.5327204929543e-05,
"loss": 0.5019,
"step": 132
},
{
"epoch": 0.28,
"learning_rate": 2.5251511632678613e-05,
"loss": 0.513,
"step": 133
},
{
"epoch": 0.29,
"learning_rate": 2.5175325287464444e-05,
"loss": 0.5045,
"step": 134
},
{
"epoch": 0.29,
"learning_rate": 2.5098649558096864e-05,
"loss": 0.5137,
"step": 135
},
{
"epoch": 0.29,
"learning_rate": 2.5021488132309282e-05,
"loss": 0.5149,
"step": 136
},
{
"epoch": 0.29,
"learning_rate": 2.4943844721194745e-05,
"loss": 0.5009,
"step": 137
},
{
"epoch": 0.29,
"learning_rate": 2.486572305902749e-05,
"loss": 0.508,
"step": 138
},
{
"epoch": 0.3,
"learning_rate": 2.4787126903083323e-05,
"loss": 0.5011,
"step": 139
},
{
"epoch": 0.3,
"learning_rate": 2.4708060033458908e-05,
"loss": 0.4966,
"step": 140
},
{
"epoch": 0.3,
"learning_rate": 2.4628526252889985e-05,
"loss": 0.498,
"step": 141
},
{
"epoch": 0.3,
"learning_rate": 2.454852938656845e-05,
"loss": 0.5132,
"step": 142
},
{
"epoch": 0.31,
"learning_rate": 2.4468073281958393e-05,
"loss": 0.4918,
"step": 143
},
{
"epoch": 0.31,
"learning_rate": 2.438716180861106e-05,
"loss": 0.5032,
"step": 144
},
{
"epoch": 0.31,
"learning_rate": 2.4305798857978756e-05,
"loss": 0.5182,
"step": 145
},
{
"epoch": 0.31,
"learning_rate": 2.4223988343227638e-05,
"loss": 0.5097,
"step": 146
},
{
"epoch": 0.31,
"learning_rate": 2.4141734199049564e-05,
"loss": 0.5133,
"step": 147
},
{
"epoch": 0.32,
"learning_rate": 2.405904038147282e-05,
"loss": 0.4896,
"step": 148
},
{
"epoch": 0.32,
"learning_rate": 2.397591086767188e-05,
"loss": 0.5178,
"step": 149
},
{
"epoch": 0.32,
"learning_rate": 2.3892349655776095e-05,
"loss": 0.495,
"step": 150
},
{
"epoch": 0.32,
"learning_rate": 2.3808360764677416e-05,
"loss": 0.4903,
"step": 151
},
{
"epoch": 0.32,
"learning_rate": 2.3723948233837116e-05,
"loss": 0.5125,
"step": 152
},
{
"epoch": 0.33,
"learning_rate": 2.363911612309149e-05,
"loss": 0.4956,
"step": 153
},
{
"epoch": 0.33,
"learning_rate": 2.3553868512456604e-05,
"loss": 0.5071,
"step": 154
},
{
"epoch": 0.33,
"learning_rate": 2.346820950193208e-05,
"loss": 0.5151,
"step": 155
},
{
"epoch": 0.33,
"learning_rate": 2.3382143211303894e-05,
"loss": 0.5136,
"step": 156
},
{
"epoch": 0.33,
"learning_rate": 2.3295673779946207e-05,
"loss": 0.5037,
"step": 157
},
{
"epoch": 0.34,
"learning_rate": 2.3208805366622342e-05,
"loss": 0.5074,
"step": 158
},
{
"epoch": 0.34,
"learning_rate": 2.3121542149284712e-05,
"loss": 0.5107,
"step": 159
},
{
"epoch": 0.34,
"learning_rate": 2.303388832487391e-05,
"loss": 0.4903,
"step": 160
},
{
"epoch": 0.34,
"learning_rate": 2.294584810911686e-05,
"loss": 0.4975,
"step": 161
},
{
"epoch": 0.35,
"learning_rate": 2.2857425736324024e-05,
"loss": 0.4947,
"step": 162
},
{
"epoch": 0.35,
"learning_rate": 2.276862545918579e-05,
"loss": 0.4999,
"step": 163
},
{
"epoch": 0.35,
"learning_rate": 2.267945154856793e-05,
"loss": 0.5,
"step": 164
},
{
"epoch": 0.35,
"learning_rate": 2.258990829330619e-05,
"loss": 0.4934,
"step": 165
},
{
"epoch": 0.35,
"learning_rate": 2.25e-05,
"loss": 0.511,
"step": 166
},
{
"epoch": 0.36,
"learning_rate": 2.2409730992805378e-05,
"loss": 0.5291,
"step": 167
},
{
"epoch": 0.36,
"learning_rate": 2.2319105613226925e-05,
"loss": 0.5124,
"step": 168
},
{
"epoch": 0.36,
"learning_rate": 2.2228128219909057e-05,
"loss": 0.5061,
"step": 169
},
{
"epoch": 0.36,
"learning_rate": 2.2136803188426344e-05,
"loss": 0.509,
"step": 170
},
{
"epoch": 0.36,
"learning_rate": 2.204513491107309e-05,
"loss": 0.4965,
"step": 171
},
{
"epoch": 0.37,
"learning_rate": 2.1953127796652057e-05,
"loss": 0.4923,
"step": 172
},
{
"epoch": 0.37,
"learning_rate": 2.1860786270262444e-05,
"loss": 0.4864,
"step": 173
},
{
"epoch": 0.37,
"learning_rate": 2.1768114773087063e-05,
"loss": 0.496,
"step": 174
},
{
"epoch": 0.37,
"learning_rate": 2.167511776217872e-05,
"loss": 0.5027,
"step": 175
},
{
"epoch": 0.38,
"learning_rate": 2.158179971024588e-05,
"loss": 0.4988,
"step": 176
},
{
"epoch": 0.38,
"learning_rate": 2.1488165105437516e-05,
"loss": 0.4905,
"step": 177
},
{
"epoch": 0.38,
"learning_rate": 2.139421845112729e-05,
"loss": 0.512,
"step": 178
},
{
"epoch": 0.38,
"learning_rate": 2.1299964265696923e-05,
"loss": 0.4978,
"step": 179
},
{
"epoch": 0.38,
"learning_rate": 2.1205407082318925e-05,
"loss": 0.4977,
"step": 180
},
{
"epoch": 0.39,
"learning_rate": 2.111055144873852e-05,
"loss": 0.4985,
"step": 181
},
{
"epoch": 0.39,
"learning_rate": 2.1015401927054977e-05,
"loss": 0.5058,
"step": 182
},
{
"epoch": 0.39,
"learning_rate": 2.0919963093502146e-05,
"loss": 0.4959,
"step": 183
},
{
"epoch": 0.39,
"learning_rate": 2.0824239538228404e-05,
"loss": 0.5018,
"step": 184
},
{
"epoch": 0.39,
"learning_rate": 2.0728235865075865e-05,
"loss": 0.4961,
"step": 185
},
{
"epoch": 0.4,
"learning_rate": 2.0631956691358952e-05,
"loss": 0.4984,
"step": 186
},
{
"epoch": 0.4,
"learning_rate": 2.053540664764235e-05,
"loss": 0.4878,
"step": 187
},
{
"epoch": 0.4,
"learning_rate": 2.0438590377518292e-05,
"loss": 0.5144,
"step": 188
},
{
"epoch": 0.4,
"learning_rate": 2.0341512537383202e-05,
"loss": 0.4977,
"step": 189
},
{
"epoch": 0.41,
"learning_rate": 2.024417779621379e-05,
"loss": 0.4948,
"step": 190
},
{
"epoch": 0.41,
"learning_rate": 2.0146590835342436e-05,
"loss": 0.4942,
"step": 191
},
{
"epoch": 0.41,
"learning_rate": 2.0048756348232097e-05,
"loss": 0.4892,
"step": 192
},
{
"epoch": 0.41,
"learning_rate": 1.9950679040250536e-05,
"loss": 0.4957,
"step": 193
},
{
"epoch": 0.41,
"learning_rate": 1.9852363628444042e-05,
"loss": 0.5053,
"step": 194
},
{
"epoch": 0.42,
"learning_rate": 1.9753814841310544e-05,
"loss": 0.4974,
"step": 195
},
{
"epoch": 0.42,
"learning_rate": 1.9655037418572202e-05,
"loss": 0.4973,
"step": 196
},
{
"epoch": 0.42,
"learning_rate": 1.955603611094745e-05,
"loss": 0.4996,
"step": 197
},
{
"epoch": 0.42,
"learning_rate": 1.9456815679922512e-05,
"loss": 0.5048,
"step": 198
},
{
"epoch": 0.42,
"learning_rate": 1.9357380897522376e-05,
"loss": 0.5041,
"step": 199
},
{
"epoch": 0.43,
"learning_rate": 1.925773654608132e-05,
"loss": 0.5117,
"step": 200
},
{
"epoch": 0.43,
"learning_rate": 1.915788741801286e-05,
"loss": 0.4922,
"step": 201
},
{
"epoch": 0.43,
"learning_rate": 1.9057838315579305e-05,
"loss": 0.4847,
"step": 202
},
{
"epoch": 0.43,
"learning_rate": 1.8957594050660752e-05,
"loss": 0.4898,
"step": 203
},
{
"epoch": 0.44,
"learning_rate": 1.8857159444523688e-05,
"loss": 0.4801,
"step": 204
},
{
"epoch": 0.44,
"learning_rate": 1.8756539327589087e-05,
"loss": 0.4944,
"step": 205
},
{
"epoch": 0.44,
"learning_rate": 1.8655738539200092e-05,
"loss": 0.4923,
"step": 206
},
{
"epoch": 0.44,
"learning_rate": 1.8554761927389283e-05,
"loss": 0.4856,
"step": 207
},
{
"epoch": 0.44,
"learning_rate": 1.8453614348645504e-05,
"loss": 0.5041,
"step": 208
},
{
"epoch": 0.45,
"learning_rate": 1.8352300667680277e-05,
"loss": 0.4845,
"step": 209
},
{
"epoch": 0.45,
"learning_rate": 1.8250825757193848e-05,
"loss": 0.4947,
"step": 210
},
{
"epoch": 0.45,
"learning_rate": 1.8149194497640817e-05,
"loss": 0.5005,
"step": 211
},
{
"epoch": 0.45,
"learning_rate": 1.8047411776995424e-05,
"loss": 0.4906,
"step": 212
},
{
"epoch": 0.45,
"learning_rate": 1.7945482490516465e-05,
"loss": 0.5018,
"step": 213
},
{
"epoch": 0.46,
"learning_rate": 1.784341154051184e-05,
"loss": 0.4992,
"step": 214
},
{
"epoch": 0.46,
"learning_rate": 1.7741203836102794e-05,
"loss": 0.4932,
"step": 215
},
{
"epoch": 0.46,
"learning_rate": 1.763886429298778e-05,
"loss": 0.4859,
"step": 216
},
{
"epoch": 0.46,
"learning_rate": 1.7536397833206082e-05,
"loss": 0.4872,
"step": 217
},
{
"epoch": 0.47,
"learning_rate": 1.743380938490104e-05,
"loss": 0.4988,
"step": 218
},
{
"epoch": 0.47,
"learning_rate": 1.7331103882083075e-05,
"loss": 0.4847,
"step": 219
},
{
"epoch": 0.47,
"learning_rate": 1.7228286264392357e-05,
"loss": 0.4849,
"step": 220
},
{
"epoch": 0.47,
"learning_rate": 1.7125361476861237e-05,
"loss": 0.4913,
"step": 221
},
{
"epoch": 0.47,
"learning_rate": 1.7022334469676433e-05,
"loss": 0.4875,
"step": 222
},
{
"epoch": 0.48,
"learning_rate": 1.691921019794093e-05,
"loss": 0.4859,
"step": 223
},
{
"epoch": 0.48,
"learning_rate": 1.6815993621435656e-05,
"loss": 0.4922,
"step": 224
},
{
"epoch": 0.48,
"learning_rate": 1.6712689704380978e-05,
"loss": 0.4849,
"step": 225
},
{
"epoch": 0.48,
"learning_rate": 1.6609303415197904e-05,
"loss": 0.4892,
"step": 226
},
{
"epoch": 0.48,
"learning_rate": 1.6505839726269153e-05,
"loss": 0.4749,
"step": 227
},
{
"epoch": 0.49,
"learning_rate": 1.64023036137e-05,
"loss": 0.4793,
"step": 228
},
{
"epoch": 0.49,
"learning_rate": 1.6298700057078945e-05,
"loss": 0.4966,
"step": 229
},
{
"epoch": 0.49,
"learning_rate": 1.619503403923823e-05,
"loss": 0.5009,
"step": 230
},
{
"epoch": 0.49,
"learning_rate": 1.6091310546014162e-05,
"loss": 0.4898,
"step": 231
},
{
"epoch": 0.49,
"learning_rate": 1.598753456600735e-05,
"loss": 0.4885,
"step": 232
},
{
"epoch": 0.5,
"learning_rate": 1.5883711090342767e-05,
"loss": 0.4881,
"step": 233
},
{
"epoch": 0.5,
"learning_rate": 1.5779845112429706e-05,
"loss": 0.5056,
"step": 234
},
{
"epoch": 0.5,
"learning_rate": 1.567594162772159e-05,
"loss": 0.4927,
"step": 235
},
{
"epoch": 0.5,
"learning_rate": 1.5572005633475758e-05,
"loss": 0.4958,
"step": 236
},
{
"epoch": 0.51,
"learning_rate": 1.5468042128513085e-05,
"loss": 0.5028,
"step": 237
},
{
"epoch": 0.51,
"learning_rate": 1.53640561129776e-05,
"loss": 0.5009,
"step": 238
},
{
"epoch": 0.51,
"learning_rate": 1.5260052588095966e-05,
"loss": 0.4787,
"step": 239
},
{
"epoch": 0.51,
"learning_rate": 1.5156036555936963e-05,
"loss": 0.4865,
"step": 240
},
{
"epoch": 0.51,
"learning_rate": 1.5052013019170917e-05,
"loss": 0.4838,
"step": 241
},
{
"epoch": 0.52,
"learning_rate": 1.4947986980829084e-05,
"loss": 0.4934,
"step": 242
},
{
"epoch": 0.52,
"learning_rate": 1.484396344406304e-05,
"loss": 0.4955,
"step": 243
},
{
"epoch": 0.52,
"learning_rate": 1.4739947411904036e-05,
"loss": 0.4953,
"step": 244
},
{
"epoch": 0.52,
"learning_rate": 1.4635943887022402e-05,
"loss": 0.4914,
"step": 245
},
{
"epoch": 0.52,
"learning_rate": 1.453195787148691e-05,
"loss": 0.4856,
"step": 246
},
{
"epoch": 0.53,
"learning_rate": 1.4427994366524248e-05,
"loss": 0.498,
"step": 247
},
{
"epoch": 0.53,
"learning_rate": 1.4324058372278415e-05,
"loss": 0.4895,
"step": 248
},
{
"epoch": 0.53,
"learning_rate": 1.4220154887570298e-05,
"loss": 0.4873,
"step": 249
},
{
"epoch": 0.53,
"learning_rate": 1.4116288909657232e-05,
"loss": 0.4877,
"step": 250
},
{
"epoch": 0.54,
"learning_rate": 1.4012465433992651e-05,
"loss": 0.4962,
"step": 251
},
{
"epoch": 0.54,
"learning_rate": 1.3908689453985844e-05,
"loss": 0.5092,
"step": 252
},
{
"epoch": 0.54,
"learning_rate": 1.3804965960761774e-05,
"loss": 0.4829,
"step": 253
},
{
"epoch": 0.54,
"learning_rate": 1.3701299942921052e-05,
"loss": 0.4868,
"step": 254
},
{
"epoch": 0.54,
"learning_rate": 1.3597696386299997e-05,
"loss": 0.4912,
"step": 255
},
{
"epoch": 0.55,
"learning_rate": 1.3494160273730844e-05,
"loss": 0.4825,
"step": 256
},
{
"epoch": 0.55,
"learning_rate": 1.33906965848021e-05,
"loss": 0.5019,
"step": 257
},
{
"epoch": 0.55,
"learning_rate": 1.3287310295619027e-05,
"loss": 0.4891,
"step": 258
},
{
"epoch": 0.55,
"learning_rate": 1.3184006378564348e-05,
"loss": 0.4858,
"step": 259
},
{
"epoch": 0.55,
"learning_rate": 1.3080789802059076e-05,
"loss": 0.4808,
"step": 260
},
{
"epoch": 0.56,
"learning_rate": 1.2977665530323568e-05,
"loss": 0.4965,
"step": 261
},
{
"epoch": 0.56,
"learning_rate": 1.2874638523138764e-05,
"loss": 0.4884,
"step": 262
},
{
"epoch": 0.56,
"learning_rate": 1.2771713735607647e-05,
"loss": 0.4935,
"step": 263
},
{
"epoch": 0.56,
"learning_rate": 1.2668896117916928e-05,
"loss": 0.4744,
"step": 264
},
{
"epoch": 0.57,
"learning_rate": 1.256619061509896e-05,
"loss": 0.5041,
"step": 265
},
{
"epoch": 0.57,
"learning_rate": 1.2463602166793919e-05,
"loss": 0.4933,
"step": 266
},
{
"epoch": 0.57,
"learning_rate": 1.2361135707012223e-05,
"loss": 0.4844,
"step": 267
},
{
"epoch": 0.57,
"learning_rate": 1.2258796163897212e-05,
"loss": 0.4839,
"step": 268
},
{
"epoch": 0.57,
"learning_rate": 1.215658845948816e-05,
"loss": 0.5059,
"step": 269
},
{
"epoch": 0.58,
"learning_rate": 1.2054517509483537e-05,
"loss": 0.4756,
"step": 270
},
{
"epoch": 0.58,
"learning_rate": 1.195258822300458e-05,
"loss": 0.4766,
"step": 271
},
{
"epoch": 0.58,
"learning_rate": 1.1850805502359189e-05,
"loss": 0.5047,
"step": 272
},
{
"epoch": 0.58,
"learning_rate": 1.1749174242806153e-05,
"loss": 0.4672,
"step": 273
},
{
"epoch": 0.58,
"learning_rate": 1.164769933231972e-05,
"loss": 0.4784,
"step": 274
},
{
"epoch": 0.59,
"learning_rate": 1.1546385651354495e-05,
"loss": 0.5188,
"step": 275
},
{
"epoch": 0.59,
"learning_rate": 1.144523807261072e-05,
"loss": 0.4915,
"step": 276
},
{
"epoch": 0.59,
"learning_rate": 1.1344261460799914e-05,
"loss": 0.4741,
"step": 277
},
{
"epoch": 0.59,
"learning_rate": 1.1243460672410919e-05,
"loss": 0.5056,
"step": 278
},
{
"epoch": 0.6,
"learning_rate": 1.1142840555476313e-05,
"loss": 0.4824,
"step": 279
},
{
"epoch": 0.6,
"learning_rate": 1.1042405949339247e-05,
"loss": 0.4941,
"step": 280
},
{
"epoch": 0.6,
"learning_rate": 1.0942161684420697e-05,
"loss": 0.4867,
"step": 281
},
{
"epoch": 0.6,
"learning_rate": 1.0842112581987143e-05,
"loss": 0.4774,
"step": 282
},
{
"epoch": 0.6,
"learning_rate": 1.0742263453918684e-05,
"loss": 0.4695,
"step": 283
},
{
"epoch": 0.61,
"learning_rate": 1.0642619102477623e-05,
"loss": 0.4899,
"step": 284
},
{
"epoch": 0.61,
"learning_rate": 1.054318432007749e-05,
"loss": 0.4964,
"step": 285
},
{
"epoch": 0.61,
"learning_rate": 1.0443963889052553e-05,
"loss": 0.4747,
"step": 286
},
{
"epoch": 0.61,
"learning_rate": 1.0344962581427802e-05,
"loss": 0.4979,
"step": 287
},
{
"epoch": 0.61,
"learning_rate": 1.024618515868946e-05,
"loss": 0.487,
"step": 288
},
{
"epoch": 0.62,
"learning_rate": 1.0147636371555964e-05,
"loss": 0.4785,
"step": 289
},
{
"epoch": 0.62,
"learning_rate": 1.0049320959749467e-05,
"loss": 0.4875,
"step": 290
},
{
"epoch": 0.62,
"learning_rate": 9.951243651767909e-06,
"loss": 0.5079,
"step": 291
},
{
"epoch": 0.62,
"learning_rate": 9.853409164657566e-06,
"loss": 0.4956,
"step": 292
},
{
"epoch": 0.63,
"learning_rate": 9.755822203786213e-06,
"loss": 0.4936,
"step": 293
},
{
"epoch": 0.63,
"learning_rate": 9.658487462616795e-06,
"loss": 0.4805,
"step": 294
},
{
"epoch": 0.63,
"learning_rate": 9.56140962248171e-06,
"loss": 0.4746,
"step": 295
},
{
"epoch": 0.63,
"learning_rate": 9.464593352357654e-06,
"loss": 0.4797,
"step": 296
},
{
"epoch": 0.63,
"learning_rate": 9.368043308641054e-06,
"loss": 0.4793,
"step": 297
},
{
"epoch": 0.64,
"learning_rate": 9.27176413492414e-06,
"loss": 0.4775,
"step": 298
},
{
"epoch": 0.64,
"learning_rate": 9.175760461771597e-06,
"loss": 0.4923,
"step": 299
},
{
"epoch": 0.64,
"learning_rate": 9.080036906497855e-06,
"loss": 0.4719,
"step": 300
},
{
"epoch": 0.64,
"learning_rate": 8.984598072945029e-06,
"loss": 0.478,
"step": 301
},
{
"epoch": 0.64,
"learning_rate": 8.889448551261481e-06,
"loss": 0.4788,
"step": 302
},
{
"epoch": 0.65,
"learning_rate": 8.794592917681079e-06,
"loss": 0.4825,
"step": 303
},
{
"epoch": 0.65,
"learning_rate": 8.700035734303074e-06,
"loss": 0.491,
"step": 304
},
{
"epoch": 0.65,
"learning_rate": 8.605781548872711e-06,
"loss": 0.5009,
"step": 305
},
{
"epoch": 0.65,
"learning_rate": 8.511834894562488e-06,
"loss": 0.4684,
"step": 306
},
{
"epoch": 0.65,
"learning_rate": 8.418200289754128e-06,
"loss": 0.4667,
"step": 307
},
{
"epoch": 0.66,
"learning_rate": 8.324882237821284e-06,
"loss": 0.4856,
"step": 308
},
{
"epoch": 0.66,
"learning_rate": 8.231885226912944e-06,
"loss": 0.4908,
"step": 309
},
{
"epoch": 0.66,
"learning_rate": 8.139213729737562e-06,
"loss": 0.4947,
"step": 310
},
{
"epoch": 0.66,
"learning_rate": 8.046872203347943e-06,
"loss": 0.4788,
"step": 311
},
{
"epoch": 0.67,
"learning_rate": 7.95486508892691e-06,
"loss": 0.4785,
"step": 312
},
{
"epoch": 0.67,
"learning_rate": 7.863196811573651e-06,
"loss": 0.4669,
"step": 313
},
{
"epoch": 0.67,
"learning_rate": 7.771871780090942e-06,
"loss": 0.4988,
"step": 314
},
{
"epoch": 0.67,
"learning_rate": 7.680894386773074e-06,
"loss": 0.4925,
"step": 315
},
{
"epoch": 0.67,
"learning_rate": 7.59026900719463e-06,
"loss": 0.4828,
"step": 316
},
{
"epoch": 0.68,
"learning_rate": 7.500000000000004e-06,
"loss": 0.4891,
"step": 317
},
{
"epoch": 0.68,
"learning_rate": 7.410091706693814e-06,
"loss": 0.4688,
"step": 318
},
{
"epoch": 0.68,
"learning_rate": 7.320548451432072e-06,
"loss": 0.4708,
"step": 319
},
{
"epoch": 0.68,
"learning_rate": 7.231374540814215e-06,
"loss": 0.4664,
"step": 320
},
{
"epoch": 0.68,
"learning_rate": 7.1425742636759835e-06,
"loss": 0.4917,
"step": 321
},
{
"epoch": 0.69,
"learning_rate": 7.054151890883147e-06,
"loss": 0.485,
"step": 322
},
{
"epoch": 0.69,
"learning_rate": 6.96611167512609e-06,
"loss": 0.4848,
"step": 323
},
{
"epoch": 0.69,
"learning_rate": 6.878457850715293e-06,
"loss": 0.4797,
"step": 324
},
{
"epoch": 0.69,
"learning_rate": 6.791194633377658e-06,
"loss": 0.4794,
"step": 325
},
{
"epoch": 0.7,
"learning_rate": 6.704326220053796e-06,
"loss": 0.4867,
"step": 326
},
{
"epoch": 0.7,
"learning_rate": 6.617856788696111e-06,
"loss": 0.5003,
"step": 327
},
{
"epoch": 0.7,
"learning_rate": 6.5317904980679176e-06,
"loss": 0.4704,
"step": 328
},
{
"epoch": 0.7,
"learning_rate": 6.446131487543397e-06,
"loss": 0.4849,
"step": 329
},
{
"epoch": 0.7,
"learning_rate": 6.360883876908513e-06,
"loss": 0.4706,
"step": 330
},
{
"epoch": 0.71,
"learning_rate": 6.276051766162887e-06,
"loss": 0.4791,
"step": 331
},
{
"epoch": 0.71,
"learning_rate": 6.191639235322586e-06,
"loss": 0.485,
"step": 332
},
{
"epoch": 0.71,
"learning_rate": 6.107650344223909e-06,
"loss": 0.5052,
"step": 333
},
{
"epoch": 0.71,
"learning_rate": 6.024089132328122e-06,
"loss": 0.4835,
"step": 334
},
{
"epoch": 0.71,
"learning_rate": 5.940959618527182e-06,
"loss": 0.4718,
"step": 335
},
{
"epoch": 0.72,
"learning_rate": 5.8582658009504395e-06,
"loss": 0.4741,
"step": 336
},
{
"epoch": 0.72,
"learning_rate": 5.776011656772364e-06,
"loss": 0.4605,
"step": 337
},
{
"epoch": 0.72,
"learning_rate": 5.694201142021245e-06,
"loss": 0.4758,
"step": 338
},
{
"epoch": 0.72,
"learning_rate": 5.612838191388936e-06,
"loss": 0.494,
"step": 339
},
{
"epoch": 0.73,
"learning_rate": 5.531926718041609e-06,
"loss": 0.4833,
"step": 340
},
{
"epoch": 0.73,
"learning_rate": 5.451470613431554e-06,
"loss": 0.4759,
"step": 341
},
{
"epoch": 0.73,
"learning_rate": 5.371473747110016e-06,
"loss": 0.4713,
"step": 342
},
{
"epoch": 0.73,
"learning_rate": 5.291939966541091e-06,
"loss": 0.4675,
"step": 343
},
{
"epoch": 0.73,
"learning_rate": 5.212873096916677e-06,
"loss": 0.4797,
"step": 344
},
{
"epoch": 0.74,
"learning_rate": 5.134276940972514e-06,
"loss": 0.4722,
"step": 345
},
{
"epoch": 0.74,
"learning_rate": 5.05615527880526e-06,
"loss": 0.4626,
"step": 346
},
{
"epoch": 0.74,
"learning_rate": 4.978511867690725e-06,
"loss": 0.5055,
"step": 347
},
{
"epoch": 0.74,
"learning_rate": 4.901350441903135e-06,
"loss": 0.4758,
"step": 348
},
{
"epoch": 0.74,
"learning_rate": 4.824674712535559e-06,
"loss": 0.4899,
"step": 349
},
{
"epoch": 0.75,
"learning_rate": 4.748488367321388e-06,
"loss": 0.4975,
"step": 350
},
{
"epoch": 0.75,
"learning_rate": 4.672795070457e-06,
"loss": 0.4901,
"step": 351
},
{
"epoch": 0.75,
"learning_rate": 4.597598462425523e-06,
"loss": 0.4809,
"step": 352
},
{
"epoch": 0.75,
"learning_rate": 4.522902159821737e-06,
"loss": 0.4712,
"step": 353
},
{
"epoch": 0.76,
"learning_rate": 4.4487097551781465e-06,
"loss": 0.4807,
"step": 354
},
{
"epoch": 0.76,
"learning_rate": 4.3750248167921925e-06,
"loss": 0.4743,
"step": 355
},
{
"epoch": 0.76,
"learning_rate": 4.301850888554617e-06,
"loss": 0.4886,
"step": 356
},
{
"epoch": 0.76,
"learning_rate": 4.229191489779048e-06,
"loss": 0.4991,
"step": 357
},
{
"epoch": 0.76,
"learning_rate": 4.157050115032724e-06,
"loss": 0.4873,
"step": 358
},
{
"epoch": 0.77,
"learning_rate": 4.085430233968418e-06,
"loss": 0.4837,
"step": 359
},
{
"epoch": 0.77,
"learning_rate": 4.014335291157566e-06,
"loss": 0.4655,
"step": 360
},
{
"epoch": 0.77,
"learning_rate": 3.9437687059246065e-06,
"loss": 0.4809,
"step": 361
},
{
"epoch": 0.77,
"learning_rate": 3.873733872182515e-06,
"loss": 0.4875,
"step": 362
},
{
"epoch": 0.77,
"learning_rate": 3.804234158269593e-06,
"loss": 0.4705,
"step": 363
},
{
"epoch": 0.78,
"learning_rate": 3.7352729067874436e-06,
"loss": 0.4761,
"step": 364
},
{
"epoch": 0.78,
"learning_rate": 3.6668534344402287e-06,
"loss": 0.495,
"step": 365
},
{
"epoch": 0.78,
"learning_rate": 3.598979031875127e-06,
"loss": 0.5018,
"step": 366
},
{
"epoch": 0.78,
"learning_rate": 3.5316529635240997e-06,
"loss": 0.4753,
"step": 367
},
{
"epoch": 0.79,
"learning_rate": 3.4648784674468687e-06,
"loss": 0.5023,
"step": 368
},
{
"epoch": 0.79,
"learning_rate": 3.398658755175183e-06,
"loss": 0.4863,
"step": 369
},
{
"epoch": 0.79,
"learning_rate": 3.3329970115583637e-06,
"loss": 0.4678,
"step": 370
},
{
"epoch": 0.79,
"learning_rate": 3.2678963946101275e-06,
"loss": 0.4765,
"step": 371
},
{
"epoch": 0.79,
"learning_rate": 3.203360035356695e-06,
"loss": 0.4708,
"step": 372
},
{
"epoch": 0.8,
"learning_rate": 3.139391037686214e-06,
"loss": 0.4732,
"step": 373
},
{
"epoch": 0.8,
"learning_rate": 3.0759924781994702e-06,
"loss": 0.4845,
"step": 374
},
{
"epoch": 0.8,
"learning_rate": 3.013167406061916e-06,
"loss": 0.4746,
"step": 375
},
{
"epoch": 0.8,
"learning_rate": 2.9509188428570287e-06,
"loss": 0.458,
"step": 376
},
{
"epoch": 0.8,
"learning_rate": 2.889249782440979e-06,
"loss": 0.4893,
"step": 377
},
{
"epoch": 0.81,
"learning_rate": 2.8281631907986445e-06,
"loss": 0.4691,
"step": 378
},
{
"epoch": 0.81,
"learning_rate": 2.767662005900958e-06,
"loss": 0.4837,
"step": 379
},
{
"epoch": 0.81,
"learning_rate": 2.7077491375636074e-06,
"loss": 0.4632,
"step": 380
},
{
"epoch": 0.81,
"learning_rate": 2.6484274673070864e-06,
"loss": 0.4821,
"step": 381
},
{
"epoch": 0.81,
"learning_rate": 2.5896998482181104e-06,
"loss": 0.4624,
"step": 382
},
{
"epoch": 0.82,
"learning_rate": 2.5315691048123886e-06,
"loss": 0.4631,
"step": 383
},
{
"epoch": 0.82,
"learning_rate": 2.474038032898786e-06,
"loss": 0.4778,
"step": 384
},
{
"epoch": 0.82,
"learning_rate": 2.4171093994448634e-06,
"loss": 0.4773,
"step": 385
},
{
"epoch": 0.82,
"learning_rate": 2.3607859424437745e-06,
"loss": 0.4731,
"step": 386
},
{
"epoch": 0.83,
"learning_rate": 2.3050703707826187e-06,
"loss": 0.494,
"step": 387
},
{
"epoch": 0.83,
"learning_rate": 2.2499653641121275e-06,
"loss": 0.4681,
"step": 388
},
{
"epoch": 0.83,
"learning_rate": 2.195473572717797e-06,
"loss": 0.4674,
"step": 389
},
{
"epoch": 0.83,
"learning_rate": 2.141597617392425e-06,
"loss": 0.4534,
"step": 390
},
{
"epoch": 0.83,
"learning_rate": 2.0883400893100535e-06,
"loss": 0.4915,
"step": 391
},
{
"epoch": 0.84,
"learning_rate": 2.0357035499013548e-06,
"loss": 0.4764,
"step": 392
},
{
"epoch": 0.84,
"learning_rate": 1.983690530730439e-06,
"loss": 0.4859,
"step": 393
},
{
"epoch": 0.84,
"learning_rate": 1.9323035333730898e-06,
"loss": 0.4738,
"step": 394
},
{
"epoch": 0.84,
"learning_rate": 1.8815450292964625e-06,
"loss": 0.4834,
"step": 395
},
{
"epoch": 0.84,
"learning_rate": 1.8314174597401995e-06,
"loss": 0.4655,
"step": 396
},
{
"epoch": 0.85,
"learning_rate": 1.7819232355990428e-06,
"loss": 0.473,
"step": 397
},
{
"epoch": 0.85,
"learning_rate": 1.7330647373068642e-06,
"loss": 0.4844,
"step": 398
},
{
"epoch": 0.85,
"learning_rate": 1.6848443147221832e-06,
"loss": 0.4715,
"step": 399
},
{
"epoch": 0.85,
"learning_rate": 1.6372642870151538e-06,
"loss": 0.4625,
"step": 400
},
{
"epoch": 0.86,
"learning_rate": 1.5903269425560146e-06,
"loss": 0.4679,
"step": 401
},
{
"epoch": 0.86,
"learning_rate": 1.5440345388050393e-06,
"loss": 0.4685,
"step": 402
},
{
"epoch": 0.86,
"learning_rate": 1.4983893022039518e-06,
"loss": 0.4678,
"step": 403
},
{
"epoch": 0.86,
"learning_rate": 1.4533934280688632e-06,
"loss": 0.4801,
"step": 404
},
{
"epoch": 0.86,
"learning_rate": 1.4090490804846612e-06,
"loss": 0.4823,
"step": 405
},
{
"epoch": 0.87,
"learning_rate": 1.3653583922009576e-06,
"loss": 0.4874,
"step": 406
},
{
"epoch": 0.87,
"learning_rate": 1.3223234645294907e-06,
"loss": 0.4695,
"step": 407
},
{
"epoch": 0.87,
"learning_rate": 1.2799463672430766e-06,
"loss": 0.4806,
"step": 408
},
{
"epoch": 0.87,
"learning_rate": 1.2382291384760536e-06,
"loss": 0.4865,
"step": 409
},
{
"epoch": 0.87,
"learning_rate": 1.197173784626261e-06,
"loss": 0.4874,
"step": 410
},
{
"epoch": 0.88,
"learning_rate": 1.1567822802585436e-06,
"loss": 0.4961,
"step": 411
},
{
"epoch": 0.88,
"learning_rate": 1.11705656800978e-06,
"loss": 0.4695,
"step": 412
},
{
"epoch": 0.88,
"learning_rate": 1.0779985584954556e-06,
"loss": 0.4672,
"step": 413
},
{
"epoch": 0.88,
"learning_rate": 1.0396101302177725e-06,
"loss": 0.4753,
"step": 414
},
{
"epoch": 0.89,
"learning_rate": 1.0018931294752897e-06,
"loss": 0.4681,
"step": 415
},
{
"epoch": 0.89,
"learning_rate": 9.648493702741429e-07,
"loss": 0.4865,
"step": 416
},
{
"epoch": 0.89,
"learning_rate": 9.284806342407881e-07,
"loss": 0.4697,
"step": 417
},
{
"epoch": 0.89,
"learning_rate": 8.927886705363181e-07,
"loss": 0.488,
"step": 418
},
{
"epoch": 0.89,
"learning_rate": 8.577751957723339e-07,
"loss": 0.4819,
"step": 419
},
{
"epoch": 0.9,
"learning_rate": 8.234418939283866e-07,
"loss": 0.4722,
"step": 420
},
{
"epoch": 0.9,
"learning_rate": 7.897904162709846e-07,
"loss": 0.4689,
"step": 421
},
{
"epoch": 0.9,
"learning_rate": 7.568223812741743e-07,
"loss": 0.4722,
"step": 422
},
{
"epoch": 0.9,
"learning_rate": 7.245393745417039e-07,
"loss": 0.4823,
"step": 423
},
{
"epoch": 0.9,
"learning_rate": 6.929429487307581e-07,
"loss": 0.4659,
"step": 424
},
{
"epoch": 0.91,
"learning_rate": 6.620346234772839e-07,
"loss": 0.4677,
"step": 425
},
{
"epoch": 0.91,
"learning_rate": 6.318158853229095e-07,
"loss": 0.4634,
"step": 426
},
{
"epoch": 0.91,
"learning_rate": 6.022881876434377e-07,
"loss": 0.4893,
"step": 427
},
{
"epoch": 0.91,
"learning_rate": 5.73452950578956e-07,
"loss": 0.4806,
"step": 428
},
{
"epoch": 0.92,
"learning_rate": 5.453115609655285e-07,
"loss": 0.4919,
"step": 429
},
{
"epoch": 0.92,
"learning_rate": 5.178653722684984e-07,
"loss": 0.4676,
"step": 430
},
{
"epoch": 0.92,
"learning_rate": 4.911157045173931e-07,
"loss": 0.4819,
"step": 431
},
{
"epoch": 0.92,
"learning_rate": 4.650638442424349e-07,
"loss": 0.479,
"step": 432
},
{
"epoch": 0.92,
"learning_rate": 4.3971104441266565e-07,
"loss": 0.4836,
"step": 433
},
{
"epoch": 0.93,
"learning_rate": 4.1505852437568823e-07,
"loss": 0.4667,
"step": 434
},
{
"epoch": 0.93,
"learning_rate": 3.911074697990136e-07,
"loss": 0.4864,
"step": 435
},
{
"epoch": 0.93,
"learning_rate": 3.678590326130493e-07,
"loss": 0.4743,
"step": 436
},
{
"epoch": 0.93,
"learning_rate": 3.4531433095568445e-07,
"loss": 0.4719,
"step": 437
},
{
"epoch": 0.93,
"learning_rate": 3.234744491185204e-07,
"loss": 0.4767,
"step": 438
},
{
"epoch": 0.94,
"learning_rate": 3.023404374947147e-07,
"loss": 0.4714,
"step": 439
},
{
"epoch": 0.94,
"learning_rate": 2.8191331252847117e-07,
"loss": 0.4781,
"step": 440
},
{
"epoch": 0.94,
"learning_rate": 2.62194056666144e-07,
"loss": 0.4685,
"step": 441
},
{
"epoch": 0.94,
"learning_rate": 2.431836183089975e-07,
"loss": 0.4696,
"step": 442
},
{
"epoch": 0.95,
"learning_rate": 2.2488291176758047e-07,
"loss": 0.4513,
"step": 443
},
{
"epoch": 0.95,
"learning_rate": 2.0729281721776682e-07,
"loss": 0.4784,
"step": 444
},
{
"epoch": 0.95,
"learning_rate": 1.904141806584042e-07,
"loss": 0.4916,
"step": 445
},
{
"epoch": 0.95,
"learning_rate": 1.7424781387064658e-07,
"loss": 0.4572,
"step": 446
},
{
"epoch": 0.95,
"learning_rate": 1.5879449437889547e-07,
"loss": 0.4676,
"step": 447
},
{
"epoch": 0.96,
"learning_rate": 1.440549654134149e-07,
"loss": 0.4801,
"step": 448
},
{
"epoch": 0.96,
"learning_rate": 1.3002993587457656e-07,
"loss": 0.4615,
"step": 449
},
{
"epoch": 0.96,
"learning_rate": 1.167200802987739e-07,
"loss": 0.4805,
"step": 450
},
{
"epoch": 0.96,
"learning_rate": 1.041260388259746e-07,
"loss": 0.4831,
"step": 451
},
{
"epoch": 0.96,
"learning_rate": 9.224841716893362e-08,
"loss": 0.4841,
"step": 452
},
{
"epoch": 0.97,
"learning_rate": 8.108778658406646e-08,
"loss": 0.468,
"step": 453
},
{
"epoch": 0.97,
"learning_rate": 7.06446838439645e-08,
"loss": 0.4702,
"step": 454
},
{
"epoch": 0.97,
"learning_rate": 6.091961121158896e-08,
"loss": 0.4761,
"step": 455
},
{
"epoch": 0.97,
"learning_rate": 5.1913036416106895e-08,
"loss": 0.4567,
"step": 456
},
{
"epoch": 0.97,
"learning_rate": 4.3625392630400884e-08,
"loss": 0.4556,
"step": 457
},
{
"epoch": 0.98,
"learning_rate": 3.605707845023232e-08,
"loss": 0.4701,
"step": 458
},
{
"epoch": 0.98,
"learning_rate": 2.920845787507509e-08,
"loss": 0.47,
"step": 459
},
{
"epoch": 0.98,
"learning_rate": 2.3079860290601272e-08,
"loss": 0.4785,
"step": 460
},
{
"epoch": 0.98,
"learning_rate": 1.7671580452848777e-08,
"loss": 0.4803,
"step": 461
},
{
"epoch": 0.99,
"learning_rate": 1.2983878474034372e-08,
"loss": 0.4768,
"step": 462
},
{
"epoch": 0.99,
"learning_rate": 9.016979810055337e-09,
"loss": 0.4613,
"step": 463
},
{
"epoch": 0.99,
"learning_rate": 5.771075249634827e-09,
"loss": 0.4798,
"step": 464
},
{
"epoch": 0.99,
"learning_rate": 3.246320905155864e-09,
"loss": 0.4797,
"step": 465
},
{
"epoch": 0.99,
"learning_rate": 1.44283820514568e-09,
"loss": 0.4639,
"step": 466
},
{
"epoch": 1.0,
"learning_rate": 3.6071388843705867e-10,
"loss": 0.4692,
"step": 467
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.4776,
"step": 468
},
{
"epoch": 1.0,
"step": 468,
"total_flos": 1.7416145444232233e+18,
"train_loss": 0.5158002531299224,
"train_runtime": 41353.3131,
"train_samples_per_second": 1.451,
"train_steps_per_second": 0.011
}
],
"max_steps": 468,
"num_train_epochs": 1,
"total_flos": 1.7416145444232233e+18,
"trial_name": null,
"trial_params": null
}