scarlett-33b / trainer_state.json
ajibawa-2023's picture
Upload folder using huggingface_hub
330c4dd
raw
history blame
17.4 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 141,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 3.3333333333333333e-06,
"loss": 1.5693,
"step": 1
},
{
"epoch": 0.04,
"learning_rate": 6.666666666666667e-06,
"loss": 1.583,
"step": 2
},
{
"epoch": 0.06,
"learning_rate": 1e-05,
"loss": 1.5527,
"step": 3
},
{
"epoch": 0.09,
"learning_rate": 1.3333333333333333e-05,
"loss": 1.3408,
"step": 4
},
{
"epoch": 0.11,
"learning_rate": 1.6666666666666667e-05,
"loss": 1.2607,
"step": 5
},
{
"epoch": 0.13,
"learning_rate": 2e-05,
"loss": 1.21,
"step": 6
},
{
"epoch": 0.15,
"learning_rate": 1.999729241179462e-05,
"loss": 1.1992,
"step": 7
},
{
"epoch": 0.17,
"learning_rate": 1.998917111338525e-05,
"loss": 1.1406,
"step": 8
},
{
"epoch": 0.19,
"learning_rate": 1.9975640502598243e-05,
"loss": 1.1113,
"step": 9
},
{
"epoch": 0.21,
"learning_rate": 1.9956707906498046e-05,
"loss": 1.1289,
"step": 10
},
{
"epoch": 0.23,
"learning_rate": 1.9932383577419432e-05,
"loss": 1.1074,
"step": 11
},
{
"epoch": 0.26,
"learning_rate": 1.9902680687415704e-05,
"loss": 1.0898,
"step": 12
},
{
"epoch": 0.28,
"learning_rate": 1.9867615321125796e-05,
"loss": 1.0869,
"step": 13
},
{
"epoch": 0.3,
"learning_rate": 1.9827206467064133e-05,
"loss": 1.0664,
"step": 14
},
{
"epoch": 0.32,
"learning_rate": 1.9781476007338058e-05,
"loss": 1.0791,
"step": 15
},
{
"epoch": 0.34,
"learning_rate": 1.973044870579824e-05,
"loss": 1.0615,
"step": 16
},
{
"epoch": 0.36,
"learning_rate": 1.967415219462864e-05,
"loss": 1.0527,
"step": 17
},
{
"epoch": 0.38,
"learning_rate": 1.961261695938319e-05,
"loss": 1.042,
"step": 18
},
{
"epoch": 0.4,
"learning_rate": 1.954587632247732e-05,
"loss": 1.0581,
"step": 19
},
{
"epoch": 0.43,
"learning_rate": 1.9473966425143292e-05,
"loss": 1.0415,
"step": 20
},
{
"epoch": 0.45,
"learning_rate": 1.9396926207859085e-05,
"loss": 1.0454,
"step": 21
},
{
"epoch": 0.47,
"learning_rate": 1.9314797389261426e-05,
"loss": 1.0576,
"step": 22
},
{
"epoch": 0.49,
"learning_rate": 1.9227624443554425e-05,
"loss": 1.0098,
"step": 23
},
{
"epoch": 0.51,
"learning_rate": 1.913545457642601e-05,
"loss": 1.0063,
"step": 24
},
{
"epoch": 0.53,
"learning_rate": 1.9038337699485207e-05,
"loss": 1.0146,
"step": 25
},
{
"epoch": 0.55,
"learning_rate": 1.8936326403234125e-05,
"loss": 1.0127,
"step": 26
},
{
"epoch": 0.57,
"learning_rate": 1.8829475928589272e-05,
"loss": 1.0117,
"step": 27
},
{
"epoch": 0.6,
"learning_rate": 1.8717844136967626e-05,
"loss": 1.0054,
"step": 28
},
{
"epoch": 0.62,
"learning_rate": 1.860149147895366e-05,
"loss": 1.0068,
"step": 29
},
{
"epoch": 0.64,
"learning_rate": 1.848048096156426e-05,
"loss": 1.0024,
"step": 30
},
{
"epoch": 0.66,
"learning_rate": 1.8354878114129368e-05,
"loss": 1.0073,
"step": 31
},
{
"epoch": 0.68,
"learning_rate": 1.8224750952806626e-05,
"loss": 1.0137,
"step": 32
},
{
"epoch": 0.7,
"learning_rate": 1.8090169943749477e-05,
"loss": 1.0312,
"step": 33
},
{
"epoch": 0.72,
"learning_rate": 1.795120796494848e-05,
"loss": 0.9912,
"step": 34
},
{
"epoch": 0.74,
"learning_rate": 1.7807940266766595e-05,
"loss": 0.9966,
"step": 35
},
{
"epoch": 0.77,
"learning_rate": 1.766044443118978e-05,
"loss": 0.979,
"step": 36
},
{
"epoch": 0.79,
"learning_rate": 1.7508800329814993e-05,
"loss": 0.9883,
"step": 37
},
{
"epoch": 0.81,
"learning_rate": 1.735309008059829e-05,
"loss": 1.0137,
"step": 38
},
{
"epoch": 0.83,
"learning_rate": 1.7193398003386514e-05,
"loss": 0.9927,
"step": 39
},
{
"epoch": 0.85,
"learning_rate": 1.702981057425662e-05,
"loss": 0.9868,
"step": 40
},
{
"epoch": 0.87,
"learning_rate": 1.686241637868734e-05,
"loss": 0.9932,
"step": 41
},
{
"epoch": 0.89,
"learning_rate": 1.6691306063588583e-05,
"loss": 0.9741,
"step": 42
},
{
"epoch": 0.91,
"learning_rate": 1.6516572288214555e-05,
"loss": 0.981,
"step": 43
},
{
"epoch": 0.94,
"learning_rate": 1.63383096739871e-05,
"loss": 0.9893,
"step": 44
},
{
"epoch": 0.96,
"learning_rate": 1.6156614753256583e-05,
"loss": 0.9971,
"step": 45
},
{
"epoch": 0.98,
"learning_rate": 1.5971585917027864e-05,
"loss": 0.9731,
"step": 46
},
{
"epoch": 1.0,
"learning_rate": 1.5783323361679865e-05,
"loss": 0.9541,
"step": 47
},
{
"epoch": 1.02,
"learning_rate": 1.5591929034707468e-05,
"loss": 0.8223,
"step": 48
},
{
"epoch": 1.04,
"learning_rate": 1.539750657951513e-05,
"loss": 0.8345,
"step": 49
},
{
"epoch": 1.06,
"learning_rate": 1.5200161279292154e-05,
"loss": 0.8193,
"step": 50
},
{
"epoch": 1.09,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.7939,
"step": 51
},
{
"epoch": 1.11,
"learning_rate": 1.4797131132502464e-05,
"loss": 0.8115,
"step": 52
},
{
"epoch": 1.13,
"learning_rate": 1.4591664533870118e-05,
"loss": 0.7896,
"step": 53
},
{
"epoch": 1.15,
"learning_rate": 1.4383711467890776e-05,
"loss": 0.772,
"step": 54
},
{
"epoch": 1.17,
"learning_rate": 1.417338454481818e-05,
"loss": 0.8052,
"step": 55
},
{
"epoch": 1.19,
"learning_rate": 1.396079766039157e-05,
"loss": 0.7891,
"step": 56
},
{
"epoch": 1.21,
"learning_rate": 1.3746065934159123e-05,
"loss": 0.7788,
"step": 57
},
{
"epoch": 1.23,
"learning_rate": 1.3529305647138689e-05,
"loss": 0.7856,
"step": 58
},
{
"epoch": 1.26,
"learning_rate": 1.3310634178849583e-05,
"loss": 0.7515,
"step": 59
},
{
"epoch": 1.28,
"learning_rate": 1.3090169943749475e-05,
"loss": 0.751,
"step": 60
},
{
"epoch": 1.3,
"learning_rate": 1.2868032327110904e-05,
"loss": 0.7681,
"step": 61
},
{
"epoch": 1.32,
"learning_rate": 1.2644341620372025e-05,
"loss": 0.7866,
"step": 62
},
{
"epoch": 1.34,
"learning_rate": 1.2419218955996677e-05,
"loss": 0.793,
"step": 63
},
{
"epoch": 1.36,
"learning_rate": 1.2192786241879033e-05,
"loss": 0.7754,
"step": 64
},
{
"epoch": 1.38,
"learning_rate": 1.1965166095328302e-05,
"loss": 0.7715,
"step": 65
},
{
"epoch": 1.4,
"learning_rate": 1.1736481776669307e-05,
"loss": 0.7773,
"step": 66
},
{
"epoch": 1.43,
"learning_rate": 1.1506857122494832e-05,
"loss": 0.7695,
"step": 67
},
{
"epoch": 1.45,
"learning_rate": 1.127641647860595e-05,
"loss": 0.7871,
"step": 68
},
{
"epoch": 1.47,
"learning_rate": 1.1045284632676535e-05,
"loss": 0.7744,
"step": 69
},
{
"epoch": 1.49,
"learning_rate": 1.0813586746678584e-05,
"loss": 0.7705,
"step": 70
},
{
"epoch": 1.51,
"learning_rate": 1.0581448289104759e-05,
"loss": 0.7681,
"step": 71
},
{
"epoch": 1.53,
"learning_rate": 1.0348994967025012e-05,
"loss": 0.7632,
"step": 72
},
{
"epoch": 1.55,
"learning_rate": 1.0116352658013973e-05,
"loss": 0.7856,
"step": 73
},
{
"epoch": 1.57,
"learning_rate": 9.883647341986032e-06,
"loss": 0.772,
"step": 74
},
{
"epoch": 1.6,
"learning_rate": 9.651005032974994e-06,
"loss": 0.7671,
"step": 75
},
{
"epoch": 1.62,
"learning_rate": 9.418551710895243e-06,
"loss": 0.7837,
"step": 76
},
{
"epoch": 1.64,
"learning_rate": 9.18641325332142e-06,
"loss": 0.7534,
"step": 77
},
{
"epoch": 1.66,
"learning_rate": 8.954715367323468e-06,
"loss": 0.7725,
"step": 78
},
{
"epoch": 1.68,
"learning_rate": 8.723583521394054e-06,
"loss": 0.7949,
"step": 79
},
{
"epoch": 1.7,
"learning_rate": 8.49314287750517e-06,
"loss": 0.8062,
"step": 80
},
{
"epoch": 1.72,
"learning_rate": 8.263518223330698e-06,
"loss": 0.7783,
"step": 81
},
{
"epoch": 1.74,
"learning_rate": 8.034833904671698e-06,
"loss": 0.8071,
"step": 82
},
{
"epoch": 1.77,
"learning_rate": 7.807213758120965e-06,
"loss": 0.7817,
"step": 83
},
{
"epoch": 1.79,
"learning_rate": 7.580781044003324e-06,
"loss": 0.7847,
"step": 84
},
{
"epoch": 1.81,
"learning_rate": 7.355658379627981e-06,
"loss": 0.7837,
"step": 85
},
{
"epoch": 1.83,
"learning_rate": 7.131967672889101e-06,
"loss": 0.749,
"step": 86
},
{
"epoch": 1.85,
"learning_rate": 6.909830056250527e-06,
"loss": 0.7754,
"step": 87
},
{
"epoch": 1.87,
"learning_rate": 6.689365821150421e-06,
"loss": 0.771,
"step": 88
},
{
"epoch": 1.89,
"learning_rate": 6.4706943528613135e-06,
"loss": 0.7705,
"step": 89
},
{
"epoch": 1.91,
"learning_rate": 6.25393406584088e-06,
"loss": 0.7881,
"step": 90
},
{
"epoch": 1.94,
"learning_rate": 6.039202339608432e-06,
"loss": 0.7603,
"step": 91
},
{
"epoch": 1.96,
"learning_rate": 5.8266154551818225e-06,
"loss": 0.7773,
"step": 92
},
{
"epoch": 1.98,
"learning_rate": 5.616288532109225e-06,
"loss": 0.7349,
"step": 93
},
{
"epoch": 2.0,
"learning_rate": 5.4083354661298816e-06,
"loss": 0.77,
"step": 94
},
{
"epoch": 2.02,
"learning_rate": 5.202868867497542e-06,
"loss": 0.584,
"step": 95
},
{
"epoch": 2.04,
"learning_rate": 5.000000000000003e-06,
"loss": 0.5554,
"step": 96
},
{
"epoch": 2.06,
"learning_rate": 4.799838720707847e-06,
"loss": 0.6069,
"step": 97
},
{
"epoch": 2.09,
"learning_rate": 4.6024934204848745e-06,
"loss": 0.6045,
"step": 98
},
{
"epoch": 2.11,
"learning_rate": 4.408070965292534e-06,
"loss": 0.6035,
"step": 99
},
{
"epoch": 2.13,
"learning_rate": 4.216676638320135e-06,
"loss": 0.5938,
"step": 100
},
{
"epoch": 2.15,
"learning_rate": 4.028414082972141e-06,
"loss": 0.5562,
"step": 101
},
{
"epoch": 2.17,
"learning_rate": 3.8433852467434175e-06,
"loss": 0.5879,
"step": 102
},
{
"epoch": 2.19,
"learning_rate": 3.661690326012897e-06,
"loss": 0.5493,
"step": 103
},
{
"epoch": 2.21,
"learning_rate": 3.483427711785449e-06,
"loss": 0.5342,
"step": 104
},
{
"epoch": 2.23,
"learning_rate": 3.308693936411421e-06,
"loss": 0.5178,
"step": 105
},
{
"epoch": 2.26,
"learning_rate": 3.1375836213126653e-06,
"loss": 0.5571,
"step": 106
},
{
"epoch": 2.28,
"learning_rate": 2.970189425743383e-06,
"loss": 0.5796,
"step": 107
},
{
"epoch": 2.3,
"learning_rate": 2.8066019966134907e-06,
"loss": 0.5642,
"step": 108
},
{
"epoch": 2.32,
"learning_rate": 2.6469099194017144e-06,
"loss": 0.5537,
"step": 109
},
{
"epoch": 2.34,
"learning_rate": 2.4911996701850083e-06,
"loss": 0.5374,
"step": 110
},
{
"epoch": 2.36,
"learning_rate": 2.339555568810221e-06,
"loss": 0.5552,
"step": 111
},
{
"epoch": 2.38,
"learning_rate": 2.192059733233408e-06,
"loss": 0.5544,
"step": 112
},
{
"epoch": 2.4,
"learning_rate": 2.048792035051521e-06,
"loss": 0.5889,
"step": 113
},
{
"epoch": 2.43,
"learning_rate": 1.9098300562505266e-06,
"loss": 0.5674,
"step": 114
},
{
"epoch": 2.45,
"learning_rate": 1.7752490471933769e-06,
"loss": 0.5664,
"step": 115
},
{
"epoch": 2.47,
"learning_rate": 1.6451218858706374e-06,
"loss": 0.5508,
"step": 116
},
{
"epoch": 2.49,
"learning_rate": 1.5195190384357405e-06,
"loss": 0.532,
"step": 117
},
{
"epoch": 2.51,
"learning_rate": 1.3985085210463479e-06,
"loss": 0.5378,
"step": 118
},
{
"epoch": 2.53,
"learning_rate": 1.282155863032377e-06,
"loss": 0.5535,
"step": 119
},
{
"epoch": 2.55,
"learning_rate": 1.1705240714107301e-06,
"loss": 0.5469,
"step": 120
},
{
"epoch": 2.57,
"learning_rate": 1.0636735967658785e-06,
"loss": 0.5659,
"step": 121
},
{
"epoch": 2.6,
"learning_rate": 9.616623005147952e-07,
"loss": 0.5435,
"step": 122
},
{
"epoch": 2.62,
"learning_rate": 8.645454235739903e-07,
"loss": 0.5513,
"step": 123
},
{
"epoch": 2.64,
"learning_rate": 7.723755564455771e-07,
"loss": 0.5437,
"step": 124
},
{
"epoch": 2.66,
"learning_rate": 6.852026107385756e-07,
"loss": 0.5422,
"step": 125
},
{
"epoch": 2.68,
"learning_rate": 6.030737921409169e-07,
"loss": 0.5461,
"step": 126
},
{
"epoch": 2.7,
"learning_rate": 5.26033574856708e-07,
"loss": 0.5347,
"step": 127
},
{
"epoch": 2.72,
"learning_rate": 4.5412367752268094e-07,
"loss": 0.5576,
"step": 128
},
{
"epoch": 2.74,
"learning_rate": 3.8738304061681107e-07,
"loss": 0.5261,
"step": 129
},
{
"epoch": 2.77,
"learning_rate": 3.2584780537136206e-07,
"loss": 0.5452,
"step": 130
},
{
"epoch": 2.79,
"learning_rate": 2.6955129420176193e-07,
"loss": 0.5811,
"step": 131
},
{
"epoch": 2.81,
"learning_rate": 2.1852399266194312e-07,
"loss": 0.5334,
"step": 132
},
{
"epoch": 2.83,
"learning_rate": 1.7279353293586765e-07,
"loss": 0.5542,
"step": 133
},
{
"epoch": 2.85,
"learning_rate": 1.323846788742078e-07,
"loss": 0.5732,
"step": 134
},
{
"epoch": 2.87,
"learning_rate": 9.731931258429638e-08,
"loss": 0.5583,
"step": 135
},
{
"epoch": 2.89,
"learning_rate": 6.761642258056977e-08,
"loss": 0.5696,
"step": 136
},
{
"epoch": 2.91,
"learning_rate": 4.329209350195651e-08,
"loss": 0.5571,
"step": 137
},
{
"epoch": 2.94,
"learning_rate": 2.4359497401758026e-08,
"loss": 0.5334,
"step": 138
},
{
"epoch": 2.96,
"learning_rate": 1.0828886614754342e-08,
"loss": 0.5129,
"step": 139
},
{
"epoch": 2.98,
"learning_rate": 2.7075882053828605e-09,
"loss": 0.5488,
"step": 140
},
{
"epoch": 3.0,
"learning_rate": 0.0,
"loss": 0.5657,
"step": 141
},
{
"epoch": 3.0,
"step": 141,
"total_flos": 176790366584832.0,
"train_loss": 0.80621640902039,
"train_runtime": 20653.5724,
"train_samples_per_second": 0.868,
"train_steps_per_second": 0.007
}
],
"max_steps": 141,
"num_train_epochs": 3,
"total_flos": 176790366584832.0,
"trial_name": null,
"trial_params": null
}