llava_7B_full_ft / trainer_state.json
kailinjiang's picture
Add llava_7B_full_ft
5f896e1
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 6.0,
"eval_steps": 500,
"global_step": 594,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.111111111111111e-06,
"loss": 2.0656,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 2.222222222222222e-06,
"loss": 2.0598,
"step": 2
},
{
"epoch": 0.03,
"learning_rate": 3.3333333333333333e-06,
"loss": 2.0373,
"step": 3
},
{
"epoch": 0.04,
"learning_rate": 4.444444444444444e-06,
"loss": 2.011,
"step": 4
},
{
"epoch": 0.05,
"learning_rate": 5.555555555555557e-06,
"loss": 2.0331,
"step": 5
},
{
"epoch": 0.06,
"learning_rate": 6.666666666666667e-06,
"loss": 1.9345,
"step": 6
},
{
"epoch": 0.07,
"learning_rate": 7.77777777777778e-06,
"loss": 1.8867,
"step": 7
},
{
"epoch": 0.08,
"learning_rate": 8.888888888888888e-06,
"loss": 1.7814,
"step": 8
},
{
"epoch": 0.09,
"learning_rate": 1e-05,
"loss": 1.8128,
"step": 9
},
{
"epoch": 0.1,
"learning_rate": 1.1111111111111113e-05,
"loss": 1.7915,
"step": 10
},
{
"epoch": 0.11,
"learning_rate": 1.2222222222222224e-05,
"loss": 1.7218,
"step": 11
},
{
"epoch": 0.12,
"learning_rate": 1.3333333333333333e-05,
"loss": 1.7265,
"step": 12
},
{
"epoch": 0.13,
"learning_rate": 1.4444444444444446e-05,
"loss": 1.6642,
"step": 13
},
{
"epoch": 0.14,
"learning_rate": 1.555555555555556e-05,
"loss": 1.7161,
"step": 14
},
{
"epoch": 0.15,
"learning_rate": 1.6666666666666667e-05,
"loss": 1.6933,
"step": 15
},
{
"epoch": 0.16,
"learning_rate": 1.7777777777777777e-05,
"loss": 1.6913,
"step": 16
},
{
"epoch": 0.17,
"learning_rate": 1.888888888888889e-05,
"loss": 1.6784,
"step": 17
},
{
"epoch": 0.18,
"learning_rate": 2e-05,
"loss": 1.7374,
"step": 18
},
{
"epoch": 0.19,
"learning_rate": 1.999985126139422e-05,
"loss": 1.651,
"step": 19
},
{
"epoch": 0.2,
"learning_rate": 1.99994050500015e-05,
"loss": 1.611,
"step": 20
},
{
"epoch": 0.21,
"learning_rate": 1.9998661379095622e-05,
"loss": 1.6064,
"step": 21
},
{
"epoch": 0.22,
"learning_rate": 1.999762027079909e-05,
"loss": 1.6288,
"step": 22
},
{
"epoch": 0.23,
"learning_rate": 1.999628175608252e-05,
"loss": 1.6289,
"step": 23
},
{
"epoch": 0.24,
"learning_rate": 1.9994645874763657e-05,
"loss": 1.632,
"step": 24
},
{
"epoch": 0.25,
"learning_rate": 1.9992712675506253e-05,
"loss": 1.6529,
"step": 25
},
{
"epoch": 0.26,
"learning_rate": 1.999048221581858e-05,
"loss": 1.6194,
"step": 26
},
{
"epoch": 0.27,
"learning_rate": 1.9987954562051724e-05,
"loss": 1.579,
"step": 27
},
{
"epoch": 0.28,
"learning_rate": 1.9985129789397633e-05,
"loss": 1.6402,
"step": 28
},
{
"epoch": 0.29,
"learning_rate": 1.998200798188685e-05,
"loss": 1.596,
"step": 29
},
{
"epoch": 0.3,
"learning_rate": 1.9978589232386036e-05,
"loss": 1.5806,
"step": 30
},
{
"epoch": 0.31,
"learning_rate": 1.99748736425952e-05,
"loss": 1.5996,
"step": 31
},
{
"epoch": 0.32,
"learning_rate": 1.9970861323044667e-05,
"loss": 1.5413,
"step": 32
},
{
"epoch": 0.33,
"learning_rate": 1.9966552393091804e-05,
"loss": 1.5971,
"step": 33
},
{
"epoch": 0.34,
"learning_rate": 1.9961946980917457e-05,
"loss": 1.6125,
"step": 34
},
{
"epoch": 0.35,
"learning_rate": 1.995704522352214e-05,
"loss": 1.6194,
"step": 35
},
{
"epoch": 0.36,
"learning_rate": 1.995184726672197e-05,
"loss": 1.5184,
"step": 36
},
{
"epoch": 0.37,
"learning_rate": 1.9946353265144315e-05,
"loss": 1.5696,
"step": 37
},
{
"epoch": 0.38,
"learning_rate": 1.9940563382223196e-05,
"loss": 1.6241,
"step": 38
},
{
"epoch": 0.39,
"learning_rate": 1.9934477790194445e-05,
"loss": 1.6131,
"step": 39
},
{
"epoch": 0.4,
"learning_rate": 1.9928096670090552e-05,
"loss": 1.6116,
"step": 40
},
{
"epoch": 0.41,
"learning_rate": 1.99214202117353e-05,
"loss": 1.5936,
"step": 41
},
{
"epoch": 0.42,
"learning_rate": 1.9914448613738107e-05,
"loss": 1.6011,
"step": 42
},
{
"epoch": 0.43,
"learning_rate": 1.9907182083488127e-05,
"loss": 1.583,
"step": 43
},
{
"epoch": 0.44,
"learning_rate": 1.989962083714808e-05,
"loss": 1.5147,
"step": 44
},
{
"epoch": 0.45,
"learning_rate": 1.989176509964781e-05,
"loss": 1.5635,
"step": 45
},
{
"epoch": 0.46,
"learning_rate": 1.988361510467761e-05,
"loss": 1.5906,
"step": 46
},
{
"epoch": 0.47,
"learning_rate": 1.9875171094681248e-05,
"loss": 1.6465,
"step": 47
},
{
"epoch": 0.48,
"learning_rate": 1.9866433320848793e-05,
"loss": 1.5781,
"step": 48
},
{
"epoch": 0.49,
"learning_rate": 1.985740204310909e-05,
"loss": 1.5766,
"step": 49
},
{
"epoch": 0.51,
"learning_rate": 1.9848077530122083e-05,
"loss": 1.5489,
"step": 50
},
{
"epoch": 0.52,
"learning_rate": 1.9838460059270775e-05,
"loss": 1.6406,
"step": 51
},
{
"epoch": 0.53,
"learning_rate": 1.9828549916653013e-05,
"loss": 1.5429,
"step": 52
},
{
"epoch": 0.54,
"learning_rate": 1.9818347397072954e-05,
"loss": 1.5426,
"step": 53
},
{
"epoch": 0.55,
"learning_rate": 1.9807852804032306e-05,
"loss": 1.5771,
"step": 54
},
{
"epoch": 0.56,
"learning_rate": 1.9797066449721295e-05,
"loss": 1.5727,
"step": 55
},
{
"epoch": 0.57,
"learning_rate": 1.9785988655009386e-05,
"loss": 1.5436,
"step": 56
},
{
"epoch": 0.58,
"learning_rate": 1.977461974943572e-05,
"loss": 1.5526,
"step": 57
},
{
"epoch": 0.59,
"learning_rate": 1.9762960071199334e-05,
"loss": 1.4772,
"step": 58
},
{
"epoch": 0.6,
"learning_rate": 1.975100996714909e-05,
"loss": 1.5065,
"step": 59
},
{
"epoch": 0.61,
"learning_rate": 1.9738769792773338e-05,
"loss": 1.5692,
"step": 60
},
{
"epoch": 0.62,
"learning_rate": 1.9726239912189382e-05,
"loss": 1.5579,
"step": 61
},
{
"epoch": 0.63,
"learning_rate": 1.9713420698132614e-05,
"loss": 1.5486,
"step": 62
},
{
"epoch": 0.64,
"learning_rate": 1.9700312531945444e-05,
"loss": 1.5538,
"step": 63
},
{
"epoch": 0.65,
"learning_rate": 1.9686915803565934e-05,
"loss": 1.5331,
"step": 64
},
{
"epoch": 0.66,
"learning_rate": 1.967323091151623e-05,
"loss": 1.5404,
"step": 65
},
{
"epoch": 0.67,
"learning_rate": 1.9659258262890683e-05,
"loss": 1.5666,
"step": 66
},
{
"epoch": 0.68,
"learning_rate": 1.9644998273343753e-05,
"loss": 1.5203,
"step": 67
},
{
"epoch": 0.69,
"learning_rate": 1.963045136707763e-05,
"loss": 1.5571,
"step": 68
},
{
"epoch": 0.7,
"learning_rate": 1.9615617976829622e-05,
"loss": 1.5172,
"step": 69
},
{
"epoch": 0.71,
"learning_rate": 1.960049854385929e-05,
"loss": 1.5794,
"step": 70
},
{
"epoch": 0.72,
"learning_rate": 1.9585093517935308e-05,
"loss": 1.5698,
"step": 71
},
{
"epoch": 0.73,
"learning_rate": 1.956940335732209e-05,
"loss": 1.5834,
"step": 72
},
{
"epoch": 0.74,
"learning_rate": 1.9553428528766163e-05,
"loss": 1.539,
"step": 73
},
{
"epoch": 0.75,
"learning_rate": 1.953716950748227e-05,
"loss": 1.5476,
"step": 74
},
{
"epoch": 0.76,
"learning_rate": 1.9520626777139243e-05,
"loss": 1.5608,
"step": 75
},
{
"epoch": 0.77,
"learning_rate": 1.9503800829845613e-05,
"loss": 1.5345,
"step": 76
},
{
"epoch": 0.78,
"learning_rate": 1.9486692166134964e-05,
"loss": 1.5821,
"step": 77
},
{
"epoch": 0.79,
"learning_rate": 1.946930129495106e-05,
"loss": 1.5458,
"step": 78
},
{
"epoch": 0.8,
"learning_rate": 1.945162873363268e-05,
"loss": 1.5698,
"step": 79
},
{
"epoch": 0.81,
"learning_rate": 1.9433675007898255e-05,
"loss": 1.5513,
"step": 80
},
{
"epoch": 0.82,
"learning_rate": 1.941544065183021e-05,
"loss": 1.4755,
"step": 81
},
{
"epoch": 0.83,
"learning_rate": 1.9396926207859085e-05,
"loss": 1.5616,
"step": 82
},
{
"epoch": 0.84,
"learning_rate": 1.93781322267474e-05,
"loss": 1.5531,
"step": 83
},
{
"epoch": 0.85,
"learning_rate": 1.935905926757326e-05,
"loss": 1.5437,
"step": 84
},
{
"epoch": 0.86,
"learning_rate": 1.9339707897713737e-05,
"loss": 1.5212,
"step": 85
},
{
"epoch": 0.87,
"learning_rate": 1.932007869282799e-05,
"loss": 1.5164,
"step": 86
},
{
"epoch": 0.88,
"learning_rate": 1.930017223684012e-05,
"loss": 1.5727,
"step": 87
},
{
"epoch": 0.89,
"learning_rate": 1.9279989121921846e-05,
"loss": 1.5399,
"step": 88
},
{
"epoch": 0.9,
"learning_rate": 1.9259529948474833e-05,
"loss": 1.5235,
"step": 89
},
{
"epoch": 0.91,
"learning_rate": 1.9238795325112867e-05,
"loss": 1.5604,
"step": 90
},
{
"epoch": 0.92,
"learning_rate": 1.921778586864375e-05,
"loss": 1.5699,
"step": 91
},
{
"epoch": 0.93,
"learning_rate": 1.9196502204050925e-05,
"loss": 1.5291,
"step": 92
},
{
"epoch": 0.94,
"learning_rate": 1.9174944964474914e-05,
"loss": 1.507,
"step": 93
},
{
"epoch": 0.95,
"learning_rate": 1.9153114791194475e-05,
"loss": 1.5863,
"step": 94
},
{
"epoch": 0.96,
"learning_rate": 1.9131012333607507e-05,
"loss": 1.5001,
"step": 95
},
{
"epoch": 0.97,
"learning_rate": 1.910863824921176e-05,
"loss": 1.4875,
"step": 96
},
{
"epoch": 0.98,
"learning_rate": 1.9085993203585257e-05,
"loss": 1.5459,
"step": 97
},
{
"epoch": 0.99,
"learning_rate": 1.9063077870366504e-05,
"loss": 1.5156,
"step": 98
},
{
"epoch": 1.0,
"learning_rate": 1.9039892931234434e-05,
"loss": 1.3387,
"step": 99
},
{
"epoch": 1.01,
"learning_rate": 1.901643907588816e-05,
"loss": 1.1024,
"step": 100
},
{
"epoch": 1.02,
"learning_rate": 1.8992717002026433e-05,
"loss": 1.0518,
"step": 101
},
{
"epoch": 1.03,
"learning_rate": 1.8968727415326885e-05,
"loss": 1.0785,
"step": 102
},
{
"epoch": 1.04,
"learning_rate": 1.8944471029425052e-05,
"loss": 1.0259,
"step": 103
},
{
"epoch": 1.05,
"learning_rate": 1.8919948565893144e-05,
"loss": 0.9982,
"step": 104
},
{
"epoch": 1.06,
"learning_rate": 1.8895160754218562e-05,
"loss": 1.0523,
"step": 105
},
{
"epoch": 1.07,
"learning_rate": 1.887010833178222e-05,
"loss": 1.0001,
"step": 106
},
{
"epoch": 1.08,
"learning_rate": 1.8844792043836592e-05,
"loss": 1.0063,
"step": 107
},
{
"epoch": 1.09,
"learning_rate": 1.881921264348355e-05,
"loss": 1.0043,
"step": 108
},
{
"epoch": 1.1,
"learning_rate": 1.8793370891651973e-05,
"loss": 1.0166,
"step": 109
},
{
"epoch": 1.11,
"learning_rate": 1.876726755707508e-05,
"loss": 1.0636,
"step": 110
},
{
"epoch": 1.12,
"learning_rate": 1.874090341626759e-05,
"loss": 1.0159,
"step": 111
},
{
"epoch": 1.13,
"learning_rate": 1.8714279253502616e-05,
"loss": 1.0415,
"step": 112
},
{
"epoch": 1.14,
"learning_rate": 1.8687395860788325e-05,
"loss": 0.9992,
"step": 113
},
{
"epoch": 1.15,
"learning_rate": 1.866025403784439e-05,
"loss": 1.0228,
"step": 114
},
{
"epoch": 1.16,
"learning_rate": 1.8632854592078185e-05,
"loss": 0.9896,
"step": 115
},
{
"epoch": 1.17,
"learning_rate": 1.860519833856079e-05,
"loss": 1.0282,
"step": 116
},
{
"epoch": 1.18,
"learning_rate": 1.8577286100002723e-05,
"loss": 1.033,
"step": 117
},
{
"epoch": 1.19,
"learning_rate": 1.854911870672947e-05,
"loss": 1.0227,
"step": 118
},
{
"epoch": 1.2,
"learning_rate": 1.8520696996656787e-05,
"loss": 0.9903,
"step": 119
},
{
"epoch": 1.21,
"learning_rate": 1.849202181526579e-05,
"loss": 1.0171,
"step": 120
},
{
"epoch": 1.22,
"learning_rate": 1.8463094015577772e-05,
"loss": 0.9954,
"step": 121
},
{
"epoch": 1.23,
"learning_rate": 1.843391445812886e-05,
"loss": 1.0108,
"step": 122
},
{
"epoch": 1.24,
"learning_rate": 1.840448401094438e-05,
"loss": 1.0312,
"step": 123
},
{
"epoch": 1.25,
"learning_rate": 1.837480354951308e-05,
"loss": 1.0058,
"step": 124
},
{
"epoch": 1.26,
"learning_rate": 1.8344873956761045e-05,
"loss": 1.0308,
"step": 125
},
{
"epoch": 1.27,
"learning_rate": 1.8314696123025456e-05,
"loss": 1.0252,
"step": 126
},
{
"epoch": 1.28,
"learning_rate": 1.8284270946028092e-05,
"loss": 0.995,
"step": 127
},
{
"epoch": 1.29,
"learning_rate": 1.8253599330848638e-05,
"loss": 0.9775,
"step": 128
},
{
"epoch": 1.3,
"learning_rate": 1.822268218989775e-05,
"loss": 1.0477,
"step": 129
},
{
"epoch": 1.31,
"learning_rate": 1.819152044288992e-05,
"loss": 0.9951,
"step": 130
},
{
"epoch": 1.32,
"learning_rate": 1.81601150168161e-05,
"loss": 1.0224,
"step": 131
},
{
"epoch": 1.33,
"learning_rate": 1.8128466845916156e-05,
"loss": 1.0196,
"step": 132
},
{
"epoch": 1.34,
"learning_rate": 1.809657687165104e-05,
"loss": 1.0027,
"step": 133
},
{
"epoch": 1.35,
"learning_rate": 1.806444604267483e-05,
"loss": 1.0451,
"step": 134
},
{
"epoch": 1.36,
"learning_rate": 1.803207531480645e-05,
"loss": 1.0222,
"step": 135
},
{
"epoch": 1.37,
"learning_rate": 1.7999465651001297e-05,
"loss": 1.001,
"step": 136
},
{
"epoch": 1.38,
"learning_rate": 1.7966618021322558e-05,
"loss": 1.0296,
"step": 137
},
{
"epoch": 1.39,
"learning_rate": 1.7933533402912354e-05,
"loss": 0.9797,
"step": 138
},
{
"epoch": 1.4,
"learning_rate": 1.790021277996269e-05,
"loss": 1.0326,
"step": 139
},
{
"epoch": 1.41,
"learning_rate": 1.786665714368617e-05,
"loss": 1.0295,
"step": 140
},
{
"epoch": 1.42,
"learning_rate": 1.7832867492286506e-05,
"loss": 0.9919,
"step": 141
},
{
"epoch": 1.43,
"learning_rate": 1.7798844830928818e-05,
"loss": 1.0372,
"step": 142
},
{
"epoch": 1.44,
"learning_rate": 1.776459017170976e-05,
"loss": 1.0156,
"step": 143
},
{
"epoch": 1.45,
"learning_rate": 1.773010453362737e-05,
"loss": 1.0052,
"step": 144
},
{
"epoch": 1.46,
"learning_rate": 1.7695388942550807e-05,
"loss": 1.0215,
"step": 145
},
{
"epoch": 1.47,
"learning_rate": 1.766044443118978e-05,
"loss": 1.0225,
"step": 146
},
{
"epoch": 1.48,
"learning_rate": 1.7625272039063884e-05,
"loss": 1.0192,
"step": 147
},
{
"epoch": 1.49,
"learning_rate": 1.758987281247162e-05,
"loss": 1.0132,
"step": 148
},
{
"epoch": 1.51,
"learning_rate": 1.7554247804459317e-05,
"loss": 1.0388,
"step": 149
},
{
"epoch": 1.52,
"learning_rate": 1.7518398074789776e-05,
"loss": 1.0305,
"step": 150
},
{
"epoch": 1.53,
"learning_rate": 1.748232468991076e-05,
"loss": 1.0158,
"step": 151
},
{
"epoch": 1.54,
"learning_rate": 1.7446028722923266e-05,
"loss": 1.0081,
"step": 152
},
{
"epoch": 1.55,
"learning_rate": 1.7409511253549592e-05,
"loss": 1.0356,
"step": 153
},
{
"epoch": 1.56,
"learning_rate": 1.737277336810124e-05,
"loss": 0.9761,
"step": 154
},
{
"epoch": 1.57,
"learning_rate": 1.7335816159446585e-05,
"loss": 1.0206,
"step": 155
},
{
"epoch": 1.58,
"learning_rate": 1.7298640726978357e-05,
"loss": 1.0653,
"step": 156
},
{
"epoch": 1.59,
"learning_rate": 1.726124817658096e-05,
"loss": 1.0368,
"step": 157
},
{
"epoch": 1.6,
"learning_rate": 1.7223639620597556e-05,
"loss": 0.986,
"step": 158
},
{
"epoch": 1.61,
"learning_rate": 1.718581617779698e-05,
"loss": 1.0022,
"step": 159
},
{
"epoch": 1.62,
"learning_rate": 1.7147778973340466e-05,
"loss": 1.0442,
"step": 160
},
{
"epoch": 1.63,
"learning_rate": 1.7109529138748156e-05,
"loss": 1.0237,
"step": 161
},
{
"epoch": 1.64,
"learning_rate": 1.7071067811865477e-05,
"loss": 1.0118,
"step": 162
},
{
"epoch": 1.65,
"learning_rate": 1.7032396136829247e-05,
"loss": 1.0765,
"step": 163
},
{
"epoch": 1.66,
"learning_rate": 1.699351526403367e-05,
"loss": 0.9896,
"step": 164
},
{
"epoch": 1.67,
"learning_rate": 1.6954426350096118e-05,
"loss": 0.9537,
"step": 165
},
{
"epoch": 1.68,
"learning_rate": 1.6915130557822698e-05,
"loss": 1.0103,
"step": 166
},
{
"epoch": 1.69,
"learning_rate": 1.6875629056173674e-05,
"loss": 1.0174,
"step": 167
},
{
"epoch": 1.7,
"learning_rate": 1.6835923020228714e-05,
"loss": 1.0319,
"step": 168
},
{
"epoch": 1.71,
"learning_rate": 1.6796013631151898e-05,
"loss": 1.0299,
"step": 169
},
{
"epoch": 1.72,
"learning_rate": 1.6755902076156606e-05,
"loss": 0.9957,
"step": 170
},
{
"epoch": 1.73,
"learning_rate": 1.6715589548470187e-05,
"loss": 1.0193,
"step": 171
},
{
"epoch": 1.74,
"learning_rate": 1.6675077247298475e-05,
"loss": 1.0055,
"step": 172
},
{
"epoch": 1.75,
"learning_rate": 1.6634366377790113e-05,
"loss": 1.0481,
"step": 173
},
{
"epoch": 1.76,
"learning_rate": 1.659345815100069e-05,
"loss": 1.037,
"step": 174
},
{
"epoch": 1.77,
"learning_rate": 1.6552353783856733e-05,
"loss": 1.0183,
"step": 175
},
{
"epoch": 1.78,
"learning_rate": 1.6511054499119493e-05,
"loss": 1.0188,
"step": 176
},
{
"epoch": 1.79,
"learning_rate": 1.6469561525348576e-05,
"loss": 0.9937,
"step": 177
},
{
"epoch": 1.8,
"learning_rate": 1.6427876096865394e-05,
"loss": 1.0199,
"step": 178
},
{
"epoch": 1.81,
"learning_rate": 1.6385999453716453e-05,
"loss": 1.0348,
"step": 179
},
{
"epoch": 1.82,
"learning_rate": 1.6343932841636455e-05,
"loss": 1.0251,
"step": 180
},
{
"epoch": 1.83,
"learning_rate": 1.6301677512011248e-05,
"loss": 1.0115,
"step": 181
},
{
"epoch": 1.84,
"learning_rate": 1.6259234721840595e-05,
"loss": 1.0165,
"step": 182
},
{
"epoch": 1.85,
"learning_rate": 1.6216605733700776e-05,
"loss": 1.0315,
"step": 183
},
{
"epoch": 1.86,
"learning_rate": 1.6173791815707053e-05,
"loss": 1.0241,
"step": 184
},
{
"epoch": 1.87,
"learning_rate": 1.6130794241475912e-05,
"loss": 1.0063,
"step": 185
},
{
"epoch": 1.88,
"learning_rate": 1.608761429008721e-05,
"loss": 1.0121,
"step": 186
},
{
"epoch": 1.89,
"learning_rate": 1.604425324604609e-05,
"loss": 1.0086,
"step": 187
},
{
"epoch": 1.9,
"learning_rate": 1.6000712399244813e-05,
"loss": 0.9998,
"step": 188
},
{
"epoch": 1.91,
"learning_rate": 1.5956993044924334e-05,
"loss": 1.0212,
"step": 189
},
{
"epoch": 1.92,
"learning_rate": 1.5913096483635827e-05,
"loss": 1.0013,
"step": 190
},
{
"epoch": 1.93,
"learning_rate": 1.586902402120195e-05,
"loss": 1.0096,
"step": 191
},
{
"epoch": 1.94,
"learning_rate": 1.5824776968678024e-05,
"loss": 1.0396,
"step": 192
},
{
"epoch": 1.95,
"learning_rate": 1.5780356642313034e-05,
"loss": 1.0267,
"step": 193
},
{
"epoch": 1.96,
"learning_rate": 1.573576436351046e-05,
"loss": 1.0212,
"step": 194
},
{
"epoch": 1.97,
"learning_rate": 1.5691001458788984e-05,
"loss": 1.0177,
"step": 195
},
{
"epoch": 1.98,
"learning_rate": 1.5646069259743007e-05,
"loss": 1.0202,
"step": 196
},
{
"epoch": 1.99,
"learning_rate": 1.5600969103003056e-05,
"loss": 1.0059,
"step": 197
},
{
"epoch": 2.0,
"learning_rate": 1.5555702330196024e-05,
"loss": 0.7256,
"step": 198
},
{
"epoch": 2.01,
"learning_rate": 1.5510270287905243e-05,
"loss": 0.5414,
"step": 199
},
{
"epoch": 2.02,
"learning_rate": 1.5464674327630437e-05,
"loss": 0.5543,
"step": 200
},
{
"epoch": 2.03,
"learning_rate": 1.5418915805747518e-05,
"loss": 0.5303,
"step": 201
},
{
"epoch": 2.04,
"learning_rate": 1.5372996083468242e-05,
"loss": 0.4704,
"step": 202
},
{
"epoch": 2.05,
"learning_rate": 1.532691652679969e-05,
"loss": 0.4854,
"step": 203
},
{
"epoch": 2.06,
"learning_rate": 1.528067850650368e-05,
"loss": 0.4688,
"step": 204
},
{
"epoch": 2.07,
"learning_rate": 1.523428339805594e-05,
"loss": 0.4973,
"step": 205
},
{
"epoch": 2.08,
"learning_rate": 1.5187732581605217e-05,
"loss": 0.46,
"step": 206
},
{
"epoch": 2.09,
"learning_rate": 1.5141027441932217e-05,
"loss": 0.4851,
"step": 207
},
{
"epoch": 2.1,
"learning_rate": 1.509416936840842e-05,
"loss": 0.4596,
"step": 208
},
{
"epoch": 2.11,
"learning_rate": 1.5047159754954721e-05,
"loss": 0.472,
"step": 209
},
{
"epoch": 2.12,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.4691,
"step": 210
},
{
"epoch": 2.13,
"learning_rate": 1.4952691506439497e-05,
"loss": 0.4607,
"step": 211
},
{
"epoch": 2.14,
"learning_rate": 1.4905235681593079e-05,
"loss": 0.4816,
"step": 212
},
{
"epoch": 2.15,
"learning_rate": 1.4857633937163402e-05,
"loss": 0.4583,
"step": 213
},
{
"epoch": 2.16,
"learning_rate": 1.4809887689193878e-05,
"loss": 0.467,
"step": 214
},
{
"epoch": 2.17,
"learning_rate": 1.4761998358026581e-05,
"loss": 0.4515,
"step": 215
},
{
"epoch": 2.18,
"learning_rate": 1.4713967368259981e-05,
"loss": 0.4563,
"step": 216
},
{
"epoch": 2.19,
"learning_rate": 1.4665796148706561e-05,
"loss": 0.4662,
"step": 217
},
{
"epoch": 2.2,
"learning_rate": 1.4617486132350343e-05,
"loss": 0.4559,
"step": 218
},
{
"epoch": 2.21,
"learning_rate": 1.4569038756304209e-05,
"loss": 0.4685,
"step": 219
},
{
"epoch": 2.22,
"learning_rate": 1.45204554617672e-05,
"loss": 0.4384,
"step": 220
},
{
"epoch": 2.23,
"learning_rate": 1.447173769398161e-05,
"loss": 0.4842,
"step": 221
},
{
"epoch": 2.24,
"learning_rate": 1.4422886902190014e-05,
"loss": 0.4467,
"step": 222
},
{
"epoch": 2.25,
"learning_rate": 1.4373904539592145e-05,
"loss": 0.4587,
"step": 223
},
{
"epoch": 2.26,
"learning_rate": 1.4324792063301662e-05,
"loss": 0.4582,
"step": 224
},
{
"epoch": 2.27,
"learning_rate": 1.4275550934302822e-05,
"loss": 0.4456,
"step": 225
},
{
"epoch": 2.28,
"learning_rate": 1.4226182617406996e-05,
"loss": 0.4544,
"step": 226
},
{
"epoch": 2.29,
"learning_rate": 1.4176688581209109e-05,
"loss": 0.4455,
"step": 227
},
{
"epoch": 2.3,
"learning_rate": 1.4127070298043949e-05,
"loss": 0.4548,
"step": 228
},
{
"epoch": 2.31,
"learning_rate": 1.4077329243942368e-05,
"loss": 0.4477,
"step": 229
},
{
"epoch": 2.32,
"learning_rate": 1.4027466898587375e-05,
"loss": 0.4529,
"step": 230
},
{
"epoch": 2.33,
"learning_rate": 1.3977484745270112e-05,
"loss": 0.4133,
"step": 231
},
{
"epoch": 2.34,
"learning_rate": 1.3927384270845744e-05,
"loss": 0.4699,
"step": 232
},
{
"epoch": 2.35,
"learning_rate": 1.3877166965689206e-05,
"loss": 0.4647,
"step": 233
},
{
"epoch": 2.36,
"learning_rate": 1.3826834323650899e-05,
"loss": 0.4583,
"step": 234
},
{
"epoch": 2.37,
"learning_rate": 1.3776387842012217e-05,
"loss": 0.437,
"step": 235
},
{
"epoch": 2.38,
"learning_rate": 1.372582902144103e-05,
"loss": 0.466,
"step": 236
},
{
"epoch": 2.39,
"learning_rate": 1.3675159365947038e-05,
"loss": 0.4734,
"step": 237
},
{
"epoch": 2.4,
"learning_rate": 1.3624380382837017e-05,
"loss": 0.4734,
"step": 238
},
{
"epoch": 2.41,
"learning_rate": 1.3573493582670003e-05,
"loss": 0.4715,
"step": 239
},
{
"epoch": 2.42,
"learning_rate": 1.3522500479212337e-05,
"loss": 0.4509,
"step": 240
},
{
"epoch": 2.43,
"learning_rate": 1.347140258939264e-05,
"loss": 0.4268,
"step": 241
},
{
"epoch": 2.44,
"learning_rate": 1.342020143325669e-05,
"loss": 0.4791,
"step": 242
},
{
"epoch": 2.45,
"learning_rate": 1.3368898533922202e-05,
"loss": 0.4794,
"step": 243
},
{
"epoch": 2.46,
"learning_rate": 1.3317495417533523e-05,
"loss": 0.4776,
"step": 244
},
{
"epoch": 2.47,
"learning_rate": 1.3265993613216223e-05,
"loss": 0.4395,
"step": 245
},
{
"epoch": 2.48,
"learning_rate": 1.3214394653031616e-05,
"loss": 0.4477,
"step": 246
},
{
"epoch": 2.49,
"learning_rate": 1.3162700071931185e-05,
"loss": 0.4426,
"step": 247
},
{
"epoch": 2.51,
"learning_rate": 1.3110911407710909e-05,
"loss": 0.4512,
"step": 248
},
{
"epoch": 2.52,
"learning_rate": 1.3059030200965536e-05,
"loss": 0.4932,
"step": 249
},
{
"epoch": 2.53,
"learning_rate": 1.300705799504273e-05,
"loss": 0.4552,
"step": 250
},
{
"epoch": 2.54,
"learning_rate": 1.295499633599719e-05,
"loss": 0.4856,
"step": 251
},
{
"epoch": 2.55,
"learning_rate": 1.2902846772544625e-05,
"loss": 0.4616,
"step": 252
},
{
"epoch": 2.56,
"learning_rate": 1.285061085601571e-05,
"loss": 0.4826,
"step": 253
},
{
"epoch": 2.57,
"learning_rate": 1.2798290140309924e-05,
"loss": 0.4546,
"step": 254
},
{
"epoch": 2.58,
"learning_rate": 1.2745886181849325e-05,
"loss": 0.4744,
"step": 255
},
{
"epoch": 2.59,
"learning_rate": 1.2693400539532263e-05,
"loss": 0.4564,
"step": 256
},
{
"epoch": 2.6,
"learning_rate": 1.2640834774686985e-05,
"loss": 0.4737,
"step": 257
},
{
"epoch": 2.61,
"learning_rate": 1.2588190451025209e-05,
"loss": 0.4698,
"step": 258
},
{
"epoch": 2.62,
"learning_rate": 1.2535469134595598e-05,
"loss": 0.4656,
"step": 259
},
{
"epoch": 2.63,
"learning_rate": 1.2482672393737164e-05,
"loss": 0.4543,
"step": 260
},
{
"epoch": 2.64,
"learning_rate": 1.242980179903264e-05,
"loss": 0.4604,
"step": 261
},
{
"epoch": 2.65,
"learning_rate": 1.2376858923261732e-05,
"loss": 0.4853,
"step": 262
},
{
"epoch": 2.66,
"learning_rate": 1.2323845341354347e-05,
"loss": 0.4727,
"step": 263
},
{
"epoch": 2.67,
"learning_rate": 1.2270762630343734e-05,
"loss": 0.4655,
"step": 264
},
{
"epoch": 2.68,
"learning_rate": 1.221761236931958e-05,
"loss": 0.4848,
"step": 265
},
{
"epoch": 2.69,
"learning_rate": 1.2164396139381029e-05,
"loss": 0.4718,
"step": 266
},
{
"epoch": 2.7,
"learning_rate": 1.2111115523589651e-05,
"loss": 0.4646,
"step": 267
},
{
"epoch": 2.71,
"learning_rate": 1.205777210692235e-05,
"loss": 0.4864,
"step": 268
},
{
"epoch": 2.72,
"learning_rate": 1.2004367476224206e-05,
"loss": 0.4613,
"step": 269
},
{
"epoch": 2.73,
"learning_rate": 1.1950903220161286e-05,
"loss": 0.4384,
"step": 270
},
{
"epoch": 2.74,
"learning_rate": 1.1897380929173365e-05,
"loss": 0.4537,
"step": 271
},
{
"epoch": 2.75,
"learning_rate": 1.1843802195426634e-05,
"loss": 0.4551,
"step": 272
},
{
"epoch": 2.76,
"learning_rate": 1.1790168612766331e-05,
"loss": 0.4861,
"step": 273
},
{
"epoch": 2.77,
"learning_rate": 1.1736481776669307e-05,
"loss": 0.4632,
"step": 274
},
{
"epoch": 2.78,
"learning_rate": 1.1682743284196595e-05,
"loss": 0.4581,
"step": 275
},
{
"epoch": 2.79,
"learning_rate": 1.162895473394589e-05,
"loss": 0.4441,
"step": 276
},
{
"epoch": 2.8,
"learning_rate": 1.1575117726003979e-05,
"loss": 0.4707,
"step": 277
},
{
"epoch": 2.81,
"learning_rate": 1.1521233861899168e-05,
"loss": 0.4659,
"step": 278
},
{
"epoch": 2.82,
"learning_rate": 1.1467304744553618e-05,
"loss": 0.4787,
"step": 279
},
{
"epoch": 2.83,
"learning_rate": 1.1413331978235677e-05,
"loss": 0.4743,
"step": 280
},
{
"epoch": 2.84,
"learning_rate": 1.1359317168512143e-05,
"loss": 0.4753,
"step": 281
},
{
"epoch": 2.85,
"learning_rate": 1.130526192220052e-05,
"loss": 0.483,
"step": 282
},
{
"epoch": 2.86,
"learning_rate": 1.1251167847321194e-05,
"loss": 0.4675,
"step": 283
},
{
"epoch": 2.87,
"learning_rate": 1.1197036553049626e-05,
"loss": 0.4936,
"step": 284
},
{
"epoch": 2.88,
"learning_rate": 1.1142869649668467e-05,
"loss": 0.4801,
"step": 285
},
{
"epoch": 2.89,
"learning_rate": 1.1088668748519646e-05,
"loss": 0.4533,
"step": 286
},
{
"epoch": 2.9,
"learning_rate": 1.1034435461956465e-05,
"loss": 0.4461,
"step": 287
},
{
"epoch": 2.91,
"learning_rate": 1.098017140329561e-05,
"loss": 0.5007,
"step": 288
},
{
"epoch": 2.92,
"learning_rate": 1.0925878186769159e-05,
"loss": 0.4827,
"step": 289
},
{
"epoch": 2.93,
"learning_rate": 1.0871557427476585e-05,
"loss": 0.4558,
"step": 290
},
{
"epoch": 2.94,
"learning_rate": 1.0817210741336684e-05,
"loss": 0.4405,
"step": 291
},
{
"epoch": 2.95,
"learning_rate": 1.0762839745039526e-05,
"loss": 0.4622,
"step": 292
},
{
"epoch": 2.96,
"learning_rate": 1.0708446055998342e-05,
"loss": 0.4697,
"step": 293
},
{
"epoch": 2.97,
"learning_rate": 1.0654031292301432e-05,
"loss": 0.4578,
"step": 294
},
{
"epoch": 2.98,
"learning_rate": 1.0599597072664012e-05,
"loss": 0.4447,
"step": 295
},
{
"epoch": 2.99,
"learning_rate": 1.0545145016380065e-05,
"loss": 0.5112,
"step": 296
},
{
"epoch": 3.0,
"learning_rate": 1.0490676743274181e-05,
"loss": 0.2749,
"step": 297
},
{
"epoch": 3.01,
"learning_rate": 1.0436193873653362e-05,
"loss": 0.1844,
"step": 298
},
{
"epoch": 3.02,
"learning_rate": 1.0381698028258817e-05,
"loss": 0.1774,
"step": 299
},
{
"epoch": 3.03,
"learning_rate": 1.0327190828217763e-05,
"loss": 0.1709,
"step": 300
},
{
"epoch": 3.04,
"learning_rate": 1.0272673894995187e-05,
"loss": 0.1601,
"step": 301
},
{
"epoch": 3.05,
"learning_rate": 1.0218148850345613e-05,
"loss": 0.1526,
"step": 302
},
{
"epoch": 3.06,
"learning_rate": 1.0163617316264869e-05,
"loss": 0.1425,
"step": 303
},
{
"epoch": 3.07,
"learning_rate": 1.0109080914941825e-05,
"loss": 0.1597,
"step": 304
},
{
"epoch": 3.08,
"learning_rate": 1.0054541268710139e-05,
"loss": 0.1534,
"step": 305
},
{
"epoch": 3.09,
"learning_rate": 1e-05,
"loss": 0.1495,
"step": 306
},
{
"epoch": 3.1,
"learning_rate": 9.945458731289863e-06,
"loss": 0.1472,
"step": 307
},
{
"epoch": 3.11,
"learning_rate": 9.890919085058179e-06,
"loss": 0.1375,
"step": 308
},
{
"epoch": 3.12,
"learning_rate": 9.836382683735133e-06,
"loss": 0.1383,
"step": 309
},
{
"epoch": 3.13,
"learning_rate": 9.78185114965439e-06,
"loss": 0.1597,
"step": 310
},
{
"epoch": 3.14,
"learning_rate": 9.727326105004818e-06,
"loss": 0.1443,
"step": 311
},
{
"epoch": 3.15,
"learning_rate": 9.67280917178224e-06,
"loss": 0.1381,
"step": 312
},
{
"epoch": 3.16,
"learning_rate": 9.618301971741185e-06,
"loss": 0.1394,
"step": 313
},
{
"epoch": 3.17,
"learning_rate": 9.563806126346643e-06,
"loss": 0.153,
"step": 314
},
{
"epoch": 3.18,
"learning_rate": 9.50932325672582e-06,
"loss": 0.1357,
"step": 315
},
{
"epoch": 3.19,
"learning_rate": 9.454854983619936e-06,
"loss": 0.1409,
"step": 316
},
{
"epoch": 3.2,
"learning_rate": 9.400402927335992e-06,
"loss": 0.1415,
"step": 317
},
{
"epoch": 3.21,
"learning_rate": 9.34596870769857e-06,
"loss": 0.1619,
"step": 318
},
{
"epoch": 3.22,
"learning_rate": 9.29155394400166e-06,
"loss": 0.1427,
"step": 319
},
{
"epoch": 3.23,
"learning_rate": 9.237160254960477e-06,
"loss": 0.1369,
"step": 320
},
{
"epoch": 3.24,
"learning_rate": 9.182789258663321e-06,
"loss": 0.1487,
"step": 321
},
{
"epoch": 3.25,
"learning_rate": 9.128442572523418e-06,
"loss": 0.1432,
"step": 322
},
{
"epoch": 3.26,
"learning_rate": 9.074121813230846e-06,
"loss": 0.141,
"step": 323
},
{
"epoch": 3.27,
"learning_rate": 9.019828596704394e-06,
"loss": 0.1341,
"step": 324
},
{
"epoch": 3.28,
"learning_rate": 8.965564538043535e-06,
"loss": 0.1351,
"step": 325
},
{
"epoch": 3.29,
"learning_rate": 8.911331251480357e-06,
"loss": 0.1387,
"step": 326
},
{
"epoch": 3.3,
"learning_rate": 8.857130350331535e-06,
"loss": 0.1361,
"step": 327
},
{
"epoch": 3.31,
"learning_rate": 8.802963446950378e-06,
"loss": 0.1389,
"step": 328
},
{
"epoch": 3.32,
"learning_rate": 8.74883215267881e-06,
"loss": 0.1342,
"step": 329
},
{
"epoch": 3.33,
"learning_rate": 8.694738077799487e-06,
"loss": 0.1447,
"step": 330
},
{
"epoch": 3.34,
"learning_rate": 8.64068283148786e-06,
"loss": 0.1406,
"step": 331
},
{
"epoch": 3.35,
"learning_rate": 8.586668021764328e-06,
"loss": 0.1533,
"step": 332
},
{
"epoch": 3.36,
"learning_rate": 8.532695255446384e-06,
"loss": 0.1278,
"step": 333
},
{
"epoch": 3.37,
"learning_rate": 8.478766138100834e-06,
"loss": 0.15,
"step": 334
},
{
"epoch": 3.38,
"learning_rate": 8.424882273996023e-06,
"loss": 0.1381,
"step": 335
},
{
"epoch": 3.39,
"learning_rate": 8.371045266054114e-06,
"loss": 0.1488,
"step": 336
},
{
"epoch": 3.4,
"learning_rate": 8.317256715803407e-06,
"loss": 0.1428,
"step": 337
},
{
"epoch": 3.41,
"learning_rate": 8.263518223330698e-06,
"loss": 0.1461,
"step": 338
},
{
"epoch": 3.42,
"learning_rate": 8.209831387233675e-06,
"loss": 0.1474,
"step": 339
},
{
"epoch": 3.43,
"learning_rate": 8.156197804573368e-06,
"loss": 0.1438,
"step": 340
},
{
"epoch": 3.44,
"learning_rate": 8.10261907082664e-06,
"loss": 0.1377,
"step": 341
},
{
"epoch": 3.45,
"learning_rate": 8.04909677983872e-06,
"loss": 0.132,
"step": 342
},
{
"epoch": 3.46,
"learning_rate": 7.995632523775795e-06,
"loss": 0.1411,
"step": 343
},
{
"epoch": 3.47,
"learning_rate": 7.942227893077652e-06,
"loss": 0.132,
"step": 344
},
{
"epoch": 3.48,
"learning_rate": 7.888884476410348e-06,
"loss": 0.1357,
"step": 345
},
{
"epoch": 3.49,
"learning_rate": 7.835603860618973e-06,
"loss": 0.1342,
"step": 346
},
{
"epoch": 3.51,
"learning_rate": 7.782387630680422e-06,
"loss": 0.1559,
"step": 347
},
{
"epoch": 3.52,
"learning_rate": 7.72923736965627e-06,
"loss": 0.1467,
"step": 348
},
{
"epoch": 3.53,
"learning_rate": 7.676154658645656e-06,
"loss": 0.1375,
"step": 349
},
{
"epoch": 3.54,
"learning_rate": 7.623141076738271e-06,
"loss": 0.1342,
"step": 350
},
{
"epoch": 3.55,
"learning_rate": 7.570198200967363e-06,
"loss": 0.1441,
"step": 351
},
{
"epoch": 3.56,
"learning_rate": 7.5173276062628364e-06,
"loss": 0.1452,
"step": 352
},
{
"epoch": 3.57,
"learning_rate": 7.4645308654044065e-06,
"loss": 0.1423,
"step": 353
},
{
"epoch": 3.58,
"learning_rate": 7.411809548974792e-06,
"loss": 0.1344,
"step": 354
},
{
"epoch": 3.59,
"learning_rate": 7.359165225313019e-06,
"loss": 0.1401,
"step": 355
},
{
"epoch": 3.6,
"learning_rate": 7.306599460467741e-06,
"loss": 0.1322,
"step": 356
},
{
"epoch": 3.61,
"learning_rate": 7.25411381815068e-06,
"loss": 0.1501,
"step": 357
},
{
"epoch": 3.62,
"learning_rate": 7.201709859690081e-06,
"loss": 0.1389,
"step": 358
},
{
"epoch": 3.63,
"learning_rate": 7.149389143984295e-06,
"loss": 0.1301,
"step": 359
},
{
"epoch": 3.64,
"learning_rate": 7.097153227455379e-06,
"loss": 0.1395,
"step": 360
},
{
"epoch": 3.65,
"learning_rate": 7.04500366400281e-06,
"loss": 0.1469,
"step": 361
},
{
"epoch": 3.66,
"learning_rate": 6.992942004957271e-06,
"loss": 0.1416,
"step": 362
},
{
"epoch": 3.67,
"learning_rate": 6.940969799034465e-06,
"loss": 0.1482,
"step": 363
},
{
"epoch": 3.68,
"learning_rate": 6.889088592289092e-06,
"loss": 0.1374,
"step": 364
},
{
"epoch": 3.69,
"learning_rate": 6.8372999280688175e-06,
"loss": 0.1352,
"step": 365
},
{
"epoch": 3.7,
"learning_rate": 6.785605346968387e-06,
"loss": 0.1347,
"step": 366
},
{
"epoch": 3.71,
"learning_rate": 6.73400638678378e-06,
"loss": 0.1372,
"step": 367
},
{
"epoch": 3.72,
"learning_rate": 6.682504582466482e-06,
"loss": 0.1379,
"step": 368
},
{
"epoch": 3.73,
"learning_rate": 6.631101466077801e-06,
"loss": 0.138,
"step": 369
},
{
"epoch": 3.74,
"learning_rate": 6.579798566743314e-06,
"loss": 0.1427,
"step": 370
},
{
"epoch": 3.75,
"learning_rate": 6.528597410607364e-06,
"loss": 0.1483,
"step": 371
},
{
"epoch": 3.76,
"learning_rate": 6.4774995207876654e-06,
"loss": 0.127,
"step": 372
},
{
"epoch": 3.77,
"learning_rate": 6.42650641733e-06,
"loss": 0.1327,
"step": 373
},
{
"epoch": 3.78,
"learning_rate": 6.375619617162985e-06,
"loss": 0.1319,
"step": 374
},
{
"epoch": 3.79,
"learning_rate": 6.3248406340529665e-06,
"loss": 0.1455,
"step": 375
},
{
"epoch": 3.8,
"learning_rate": 6.274170978558971e-06,
"loss": 0.1484,
"step": 376
},
{
"epoch": 3.81,
"learning_rate": 6.223612157987786e-06,
"loss": 0.1447,
"step": 377
},
{
"epoch": 3.82,
"learning_rate": 6.173165676349103e-06,
"loss": 0.1379,
"step": 378
},
{
"epoch": 3.83,
"learning_rate": 6.122833034310794e-06,
"loss": 0.1507,
"step": 379
},
{
"epoch": 3.84,
"learning_rate": 6.072615729154261e-06,
"loss": 0.1506,
"step": 380
},
{
"epoch": 3.85,
"learning_rate": 6.02251525472989e-06,
"loss": 0.1437,
"step": 381
},
{
"epoch": 3.86,
"learning_rate": 5.97253310141263e-06,
"loss": 0.156,
"step": 382
},
{
"epoch": 3.87,
"learning_rate": 5.922670756057633e-06,
"loss": 0.1322,
"step": 383
},
{
"epoch": 3.88,
"learning_rate": 5.872929701956054e-06,
"loss": 0.1383,
"step": 384
},
{
"epoch": 3.89,
"learning_rate": 5.823311418790894e-06,
"loss": 0.1378,
"step": 385
},
{
"epoch": 3.9,
"learning_rate": 5.773817382593008e-06,
"loss": 0.1506,
"step": 386
},
{
"epoch": 3.91,
"learning_rate": 5.724449065697182e-06,
"loss": 0.1453,
"step": 387
},
{
"epoch": 3.92,
"learning_rate": 5.675207936698337e-06,
"loss": 0.147,
"step": 388
},
{
"epoch": 3.93,
"learning_rate": 5.6260954604078585e-06,
"loss": 0.1403,
"step": 389
},
{
"epoch": 3.94,
"learning_rate": 5.5771130978099896e-06,
"loss": 0.1287,
"step": 390
},
{
"epoch": 3.95,
"learning_rate": 5.5282623060183945e-06,
"loss": 0.1299,
"step": 391
},
{
"epoch": 3.96,
"learning_rate": 5.479544538232804e-06,
"loss": 0.133,
"step": 392
},
{
"epoch": 3.97,
"learning_rate": 5.430961243695794e-06,
"loss": 0.139,
"step": 393
},
{
"epoch": 3.98,
"learning_rate": 5.382513867649663e-06,
"loss": 0.1268,
"step": 394
},
{
"epoch": 3.99,
"learning_rate": 5.334203851293442e-06,
"loss": 0.1441,
"step": 395
},
{
"epoch": 4.0,
"learning_rate": 5.286032631740023e-06,
"loss": 0.0683,
"step": 396
},
{
"epoch": 4.01,
"learning_rate": 5.238001641973422e-06,
"loss": 0.0522,
"step": 397
},
{
"epoch": 4.02,
"learning_rate": 5.190112310806126e-06,
"loss": 0.0511,
"step": 398
},
{
"epoch": 4.03,
"learning_rate": 5.142366062836599e-06,
"loss": 0.0536,
"step": 399
},
{
"epoch": 4.04,
"learning_rate": 5.094764318406921e-06,
"loss": 0.0448,
"step": 400
},
{
"epoch": 4.05,
"learning_rate": 5.047308493560506e-06,
"loss": 0.0459,
"step": 401
},
{
"epoch": 4.06,
"learning_rate": 5.000000000000003e-06,
"loss": 0.0483,
"step": 402
},
{
"epoch": 4.07,
"learning_rate": 4.952840245045279e-06,
"loss": 0.0452,
"step": 403
},
{
"epoch": 4.08,
"learning_rate": 4.9058306315915826e-06,
"loss": 0.0437,
"step": 404
},
{
"epoch": 4.09,
"learning_rate": 4.858972558067784e-06,
"loss": 0.036,
"step": 405
},
{
"epoch": 4.1,
"learning_rate": 4.812267418394784e-06,
"loss": 0.0435,
"step": 406
},
{
"epoch": 4.11,
"learning_rate": 4.765716601944062e-06,
"loss": 0.0519,
"step": 407
},
{
"epoch": 4.12,
"learning_rate": 4.7193214934963204e-06,
"loss": 0.045,
"step": 408
},
{
"epoch": 4.13,
"learning_rate": 4.6730834732003104e-06,
"loss": 0.0414,
"step": 409
},
{
"epoch": 4.14,
"learning_rate": 4.627003916531761e-06,
"loss": 0.0402,
"step": 410
},
{
"epoch": 4.15,
"learning_rate": 4.581084194252486e-06,
"loss": 0.0436,
"step": 411
},
{
"epoch": 4.16,
"learning_rate": 4.535325672369567e-06,
"loss": 0.0382,
"step": 412
},
{
"epoch": 4.17,
"learning_rate": 4.489729712094762e-06,
"loss": 0.0443,
"step": 413
},
{
"epoch": 4.18,
"learning_rate": 4.444297669803981e-06,
"loss": 0.0419,
"step": 414
},
{
"epoch": 4.19,
"learning_rate": 4.399030896996945e-06,
"loss": 0.0403,
"step": 415
},
{
"epoch": 4.2,
"learning_rate": 4.353930740256997e-06,
"loss": 0.0422,
"step": 416
},
{
"epoch": 4.21,
"learning_rate": 4.308998541211016e-06,
"loss": 0.0449,
"step": 417
},
{
"epoch": 4.22,
"learning_rate": 4.264235636489542e-06,
"loss": 0.0419,
"step": 418
},
{
"epoch": 4.23,
"learning_rate": 4.219643357686968e-06,
"loss": 0.0406,
"step": 419
},
{
"epoch": 4.24,
"learning_rate": 4.17522303132198e-06,
"loss": 0.0371,
"step": 420
},
{
"epoch": 4.25,
"learning_rate": 4.1309759787980565e-06,
"loss": 0.0415,
"step": 421
},
{
"epoch": 4.26,
"learning_rate": 4.086903516364179e-06,
"loss": 0.0383,
"step": 422
},
{
"epoch": 4.27,
"learning_rate": 4.043006955075667e-06,
"loss": 0.0383,
"step": 423
},
{
"epoch": 4.28,
"learning_rate": 3.999287600755192e-06,
"loss": 0.0369,
"step": 424
},
{
"epoch": 4.29,
"learning_rate": 3.955746753953912e-06,
"loss": 0.0391,
"step": 425
},
{
"epoch": 4.3,
"learning_rate": 3.912385709912794e-06,
"loss": 0.042,
"step": 426
},
{
"epoch": 4.31,
"learning_rate": 3.869205758524091e-06,
"loss": 0.0409,
"step": 427
},
{
"epoch": 4.32,
"learning_rate": 3.826208184292952e-06,
"loss": 0.0381,
"step": 428
},
{
"epoch": 4.33,
"learning_rate": 3.7833942662992286e-06,
"loss": 0.0434,
"step": 429
},
{
"epoch": 4.34,
"learning_rate": 3.7407652781594094e-06,
"loss": 0.0405,
"step": 430
},
{
"epoch": 4.35,
"learning_rate": 3.698322487988755e-06,
"loss": 0.0413,
"step": 431
},
{
"epoch": 4.36,
"learning_rate": 3.6560671583635467e-06,
"loss": 0.0416,
"step": 432
},
{
"epoch": 4.37,
"learning_rate": 3.614000546283547e-06,
"loss": 0.0382,
"step": 433
},
{
"epoch": 4.38,
"learning_rate": 3.5721239031346067e-06,
"loss": 0.0461,
"step": 434
},
{
"epoch": 4.39,
"learning_rate": 3.5304384746514273e-06,
"loss": 0.0404,
"step": 435
},
{
"epoch": 4.4,
"learning_rate": 3.4889455008805107e-06,
"loss": 0.042,
"step": 436
},
{
"epoch": 4.41,
"learning_rate": 3.4476462161432678e-06,
"loss": 0.0426,
"step": 437
},
{
"epoch": 4.42,
"learning_rate": 3.4065418489993118e-06,
"loss": 0.0378,
"step": 438
},
{
"epoch": 4.43,
"learning_rate": 3.3656336222098907e-06,
"loss": 0.0411,
"step": 439
},
{
"epoch": 4.44,
"learning_rate": 3.324922752701528e-06,
"loss": 0.0404,
"step": 440
},
{
"epoch": 4.45,
"learning_rate": 3.284410451529816e-06,
"loss": 0.0423,
"step": 441
},
{
"epoch": 4.46,
"learning_rate": 3.2440979238433977e-06,
"loss": 0.0393,
"step": 442
},
{
"epoch": 4.47,
"learning_rate": 3.2039863688481055e-06,
"loss": 0.0401,
"step": 443
},
{
"epoch": 4.48,
"learning_rate": 3.1640769797712865e-06,
"loss": 0.0383,
"step": 444
},
{
"epoch": 4.49,
"learning_rate": 3.124370943826326e-06,
"loss": 0.0396,
"step": 445
},
{
"epoch": 4.51,
"learning_rate": 3.0848694421773075e-06,
"loss": 0.0391,
"step": 446
},
{
"epoch": 4.52,
"learning_rate": 3.0455736499038847e-06,
"loss": 0.0407,
"step": 447
},
{
"epoch": 4.53,
"learning_rate": 3.0064847359663284e-06,
"loss": 0.0404,
"step": 448
},
{
"epoch": 4.54,
"learning_rate": 2.967603863170759e-06,
"loss": 0.0406,
"step": 449
},
{
"epoch": 4.55,
"learning_rate": 2.9289321881345257e-06,
"loss": 0.0451,
"step": 450
},
{
"epoch": 4.56,
"learning_rate": 2.8904708612518404e-06,
"loss": 0.0368,
"step": 451
},
{
"epoch": 4.57,
"learning_rate": 2.8522210266595386e-06,
"loss": 0.0428,
"step": 452
},
{
"epoch": 4.58,
"learning_rate": 2.8141838222030195e-06,
"loss": 0.0406,
"step": 453
},
{
"epoch": 4.59,
"learning_rate": 2.776360379402445e-06,
"loss": 0.0387,
"step": 454
},
{
"epoch": 4.6,
"learning_rate": 2.7387518234190414e-06,
"loss": 0.0396,
"step": 455
},
{
"epoch": 4.61,
"learning_rate": 2.7013592730216464e-06,
"loss": 0.0394,
"step": 456
},
{
"epoch": 4.62,
"learning_rate": 2.664183840553417e-06,
"loss": 0.0423,
"step": 457
},
{
"epoch": 4.63,
"learning_rate": 2.6272266318987606e-06,
"loss": 0.0381,
"step": 458
},
{
"epoch": 4.64,
"learning_rate": 2.5904887464504115e-06,
"loss": 0.0362,
"step": 459
},
{
"epoch": 4.65,
"learning_rate": 2.5539712770767377e-06,
"loss": 0.0397,
"step": 460
},
{
"epoch": 4.66,
"learning_rate": 2.5176753100892426e-06,
"loss": 0.0382,
"step": 461
},
{
"epoch": 4.67,
"learning_rate": 2.4816019252102274e-06,
"loss": 0.039,
"step": 462
},
{
"epoch": 4.68,
"learning_rate": 2.4457521955406872e-06,
"loss": 0.0367,
"step": 463
},
{
"epoch": 4.69,
"learning_rate": 2.4101271875283818e-06,
"loss": 0.0365,
"step": 464
},
{
"epoch": 4.7,
"learning_rate": 2.3747279609361197e-06,
"loss": 0.0415,
"step": 465
},
{
"epoch": 4.71,
"learning_rate": 2.339555568810221e-06,
"loss": 0.0377,
"step": 466
},
{
"epoch": 4.72,
"learning_rate": 2.3046110574491986e-06,
"loss": 0.038,
"step": 467
},
{
"epoch": 4.73,
"learning_rate": 2.26989546637263e-06,
"loss": 0.0421,
"step": 468
},
{
"epoch": 4.74,
"learning_rate": 2.2354098282902446e-06,
"loss": 0.0411,
"step": 469
},
{
"epoch": 4.75,
"learning_rate": 2.201155169071184e-06,
"loss": 0.0386,
"step": 470
},
{
"epoch": 4.76,
"learning_rate": 2.1671325077134963e-06,
"loss": 0.0451,
"step": 471
},
{
"epoch": 4.77,
"learning_rate": 2.1333428563138304e-06,
"loss": 0.0393,
"step": 472
},
{
"epoch": 4.78,
"learning_rate": 2.0997872200373114e-06,
"loss": 0.0374,
"step": 473
},
{
"epoch": 4.79,
"learning_rate": 2.0664665970876496e-06,
"loss": 0.041,
"step": 474
},
{
"epoch": 4.8,
"learning_rate": 2.0333819786774446e-06,
"loss": 0.0384,
"step": 475
},
{
"epoch": 4.81,
"learning_rate": 2.0005343489987038e-06,
"loss": 0.0399,
"step": 476
},
{
"epoch": 4.82,
"learning_rate": 1.967924685193552e-06,
"loss": 0.0413,
"step": 477
},
{
"epoch": 4.83,
"learning_rate": 1.9355539573251737e-06,
"loss": 0.0386,
"step": 478
},
{
"epoch": 4.84,
"learning_rate": 1.903423128348959e-06,
"loss": 0.0378,
"step": 479
},
{
"epoch": 4.85,
"learning_rate": 1.8715331540838488e-06,
"loss": 0.0403,
"step": 480
},
{
"epoch": 4.86,
"learning_rate": 1.8398849831839017e-06,
"loss": 0.0394,
"step": 481
},
{
"epoch": 4.87,
"learning_rate": 1.808479557110081e-06,
"loss": 0.0446,
"step": 482
},
{
"epoch": 4.88,
"learning_rate": 1.7773178101022514e-06,
"loss": 0.043,
"step": 483
},
{
"epoch": 4.89,
"learning_rate": 1.7464006691513624e-06,
"loss": 0.0384,
"step": 484
},
{
"epoch": 4.9,
"learning_rate": 1.7157290539719108e-06,
"loss": 0.0369,
"step": 485
},
{
"epoch": 4.91,
"learning_rate": 1.6853038769745466e-06,
"loss": 0.0399,
"step": 486
},
{
"epoch": 4.92,
"learning_rate": 1.655126043238957e-06,
"loss": 0.0393,
"step": 487
},
{
"epoch": 4.93,
"learning_rate": 1.6251964504869221e-06,
"loss": 0.0421,
"step": 488
},
{
"epoch": 4.94,
"learning_rate": 1.5955159890556182e-06,
"loss": 0.0374,
"step": 489
},
{
"epoch": 4.95,
"learning_rate": 1.566085541871145e-06,
"loss": 0.0358,
"step": 490
},
{
"epoch": 4.96,
"learning_rate": 1.5369059844222279e-06,
"loss": 0.0377,
"step": 491
},
{
"epoch": 4.97,
"learning_rate": 1.5079781847342122e-06,
"loss": 0.0383,
"step": 492
},
{
"epoch": 4.98,
"learning_rate": 1.4793030033432143e-06,
"loss": 0.0421,
"step": 493
},
{
"epoch": 4.99,
"learning_rate": 1.4508812932705364e-06,
"loss": 0.0346,
"step": 494
},
{
"epoch": 5.0,
"learning_rate": 1.4227138999972801e-06,
"loss": 0.0219,
"step": 495
},
{
"epoch": 5.01,
"learning_rate": 1.3948016614392113e-06,
"loss": 0.0232,
"step": 496
},
{
"epoch": 5.02,
"learning_rate": 1.3671454079218171e-06,
"loss": 0.0237,
"step": 497
},
{
"epoch": 5.03,
"learning_rate": 1.339745962155613e-06,
"loss": 0.0228,
"step": 498
},
{
"epoch": 5.04,
"learning_rate": 1.3126041392116774e-06,
"loss": 0.0219,
"step": 499
},
{
"epoch": 5.05,
"learning_rate": 1.2857207464973876e-06,
"loss": 0.022,
"step": 500
},
{
"epoch": 5.06,
"learning_rate": 1.2590965837324132e-06,
"loss": 0.023,
"step": 501
},
{
"epoch": 5.07,
"learning_rate": 1.2327324429249232e-06,
"loss": 0.0199,
"step": 502
},
{
"epoch": 5.08,
"learning_rate": 1.2066291083480297e-06,
"loss": 0.0203,
"step": 503
},
{
"epoch": 5.09,
"learning_rate": 1.1807873565164507e-06,
"loss": 0.0189,
"step": 504
},
{
"epoch": 5.1,
"learning_rate": 1.1552079561634111e-06,
"loss": 0.0185,
"step": 505
},
{
"epoch": 5.11,
"learning_rate": 1.129891668217783e-06,
"loss": 0.0202,
"step": 506
},
{
"epoch": 5.12,
"learning_rate": 1.1048392457814406e-06,
"loss": 0.0247,
"step": 507
},
{
"epoch": 5.13,
"learning_rate": 1.0800514341068592e-06,
"loss": 0.0215,
"step": 508
},
{
"epoch": 5.14,
"learning_rate": 1.0555289705749483e-06,
"loss": 0.0195,
"step": 509
},
{
"epoch": 5.15,
"learning_rate": 1.0312725846731174e-06,
"loss": 0.0176,
"step": 510
},
{
"epoch": 5.16,
"learning_rate": 1.0072829979735698e-06,
"loss": 0.0206,
"step": 511
},
{
"epoch": 5.17,
"learning_rate": 9.835609241118404e-07,
"loss": 0.0178,
"step": 512
},
{
"epoch": 5.18,
"learning_rate": 9.601070687655667e-07,
"loss": 0.0173,
"step": 513
},
{
"epoch": 5.19,
"learning_rate": 9.369221296335007e-07,
"loss": 0.0203,
"step": 514
},
{
"epoch": 5.2,
"learning_rate": 9.140067964147447e-07,
"loss": 0.0222,
"step": 515
},
{
"epoch": 5.21,
"learning_rate": 8.91361750788241e-07,
"loss": 0.0205,
"step": 516
},
{
"epoch": 5.22,
"learning_rate": 8.689876663924957e-07,
"loss": 0.0193,
"step": 517
},
{
"epoch": 5.23,
"learning_rate": 8.468852088055291e-07,
"loss": 0.0185,
"step": 518
},
{
"epoch": 5.24,
"learning_rate": 8.250550355250875e-07,
"loss": 0.0224,
"step": 519
},
{
"epoch": 5.25,
"learning_rate": 8.034977959490775e-07,
"loss": 0.0203,
"step": 520
},
{
"epoch": 5.26,
"learning_rate": 7.822141313562548e-07,
"loss": 0.0219,
"step": 521
},
{
"epoch": 5.27,
"learning_rate": 7.612046748871327e-07,
"loss": 0.0192,
"step": 522
},
{
"epoch": 5.28,
"learning_rate": 7.404700515251672e-07,
"loss": 0.0168,
"step": 523
},
{
"epoch": 5.29,
"learning_rate": 7.200108780781556e-07,
"loss": 0.0189,
"step": 524
},
{
"epoch": 5.3,
"learning_rate": 6.998277631598793e-07,
"loss": 0.0187,
"step": 525
},
{
"epoch": 5.31,
"learning_rate": 6.799213071720156e-07,
"loss": 0.0193,
"step": 526
},
{
"epoch": 5.32,
"learning_rate": 6.602921022862663e-07,
"loss": 0.0214,
"step": 527
},
{
"epoch": 5.33,
"learning_rate": 6.409407324267448e-07,
"loss": 0.0199,
"step": 528
},
{
"epoch": 5.34,
"learning_rate": 6.218677732526035e-07,
"loss": 0.0186,
"step": 529
},
{
"epoch": 5.35,
"learning_rate": 6.030737921409169e-07,
"loss": 0.0201,
"step": 530
},
{
"epoch": 5.36,
"learning_rate": 5.845593481697931e-07,
"loss": 0.0176,
"step": 531
},
{
"epoch": 5.37,
"learning_rate": 5.663249921017477e-07,
"loss": 0.019,
"step": 532
},
{
"epoch": 5.38,
"learning_rate": 5.483712663673224e-07,
"loss": 0.0208,
"step": 533
},
{
"epoch": 5.39,
"learning_rate": 5.306987050489442e-07,
"loss": 0.0192,
"step": 534
},
{
"epoch": 5.4,
"learning_rate": 5.133078338650376e-07,
"loss": 0.0199,
"step": 535
},
{
"epoch": 5.41,
"learning_rate": 4.961991701543889e-07,
"loss": 0.0205,
"step": 536
},
{
"epoch": 5.42,
"learning_rate": 4.793732228607573e-07,
"loss": 0.0186,
"step": 537
},
{
"epoch": 5.43,
"learning_rate": 4.628304925177318e-07,
"loss": 0.0172,
"step": 538
},
{
"epoch": 5.44,
"learning_rate": 4.465714712338398e-07,
"loss": 0.0224,
"step": 539
},
{
"epoch": 5.45,
"learning_rate": 4.305966426779118e-07,
"loss": 0.0192,
"step": 540
},
{
"epoch": 5.46,
"learning_rate": 4.149064820646953e-07,
"loss": 0.0198,
"step": 541
},
{
"epoch": 5.47,
"learning_rate": 3.99501456140714e-07,
"loss": 0.0192,
"step": 542
},
{
"epoch": 5.48,
"learning_rate": 3.8438202317037987e-07,
"loss": 0.0191,
"step": 543
},
{
"epoch": 5.49,
"learning_rate": 3.6954863292237297e-07,
"loss": 0.0182,
"step": 544
},
{
"epoch": 5.51,
"learning_rate": 3.550017266562489e-07,
"loss": 0.0212,
"step": 545
},
{
"epoch": 5.52,
"learning_rate": 3.4074173710931804e-07,
"loss": 0.0217,
"step": 546
},
{
"epoch": 5.53,
"learning_rate": 3.2676908848377263e-07,
"loss": 0.0187,
"step": 547
},
{
"epoch": 5.54,
"learning_rate": 3.1308419643406915e-07,
"loss": 0.0196,
"step": 548
},
{
"epoch": 5.55,
"learning_rate": 2.996874680545603e-07,
"loss": 0.0211,
"step": 549
},
{
"epoch": 5.56,
"learning_rate": 2.865793018673857e-07,
"loss": 0.0203,
"step": 550
},
{
"epoch": 5.57,
"learning_rate": 2.7376008781061835e-07,
"loss": 0.02,
"step": 551
},
{
"epoch": 5.58,
"learning_rate": 2.612302072266637e-07,
"loss": 0.0208,
"step": 552
},
{
"epoch": 5.59,
"learning_rate": 2.489900328509154e-07,
"loss": 0.0233,
"step": 553
},
{
"epoch": 5.6,
"learning_rate": 2.370399288006664e-07,
"loss": 0.018,
"step": 554
},
{
"epoch": 5.61,
"learning_rate": 2.2538025056428216e-07,
"loss": 0.0163,
"step": 555
},
{
"epoch": 5.62,
"learning_rate": 2.140113449906167e-07,
"loss": 0.0204,
"step": 556
},
{
"epoch": 5.63,
"learning_rate": 2.0293355027870554e-07,
"loss": 0.023,
"step": 557
},
{
"epoch": 5.64,
"learning_rate": 1.921471959676957e-07,
"loss": 0.0188,
"step": 558
},
{
"epoch": 5.65,
"learning_rate": 1.8165260292704712e-07,
"loss": 0.0195,
"step": 559
},
{
"epoch": 5.66,
"learning_rate": 1.7145008334698898e-07,
"loss": 0.0202,
"step": 560
},
{
"epoch": 5.67,
"learning_rate": 1.615399407292251e-07,
"loss": 0.0196,
"step": 561
},
{
"epoch": 5.68,
"learning_rate": 1.519224698779198e-07,
"loss": 0.0191,
"step": 562
},
{
"epoch": 5.69,
"learning_rate": 1.4259795689090972e-07,
"loss": 0.0205,
"step": 563
},
{
"epoch": 5.7,
"learning_rate": 1.3356667915121025e-07,
"loss": 0.0199,
"step": 564
},
{
"epoch": 5.71,
"learning_rate": 1.2482890531875124e-07,
"loss": 0.0178,
"step": 565
},
{
"epoch": 5.72,
"learning_rate": 1.1638489532239339e-07,
"loss": 0.0222,
"step": 566
},
{
"epoch": 5.73,
"learning_rate": 1.0823490035218986e-07,
"loss": 0.0216,
"step": 567
},
{
"epoch": 5.74,
"learning_rate": 1.0037916285192129e-07,
"loss": 0.0197,
"step": 568
},
{
"epoch": 5.75,
"learning_rate": 9.281791651187366e-08,
"loss": 0.0211,
"step": 569
},
{
"epoch": 5.76,
"learning_rate": 8.555138626189619e-08,
"loss": 0.019,
"step": 570
},
{
"epoch": 5.77,
"learning_rate": 7.857978826470325e-08,
"loss": 0.0186,
"step": 571
},
{
"epoch": 5.78,
"learning_rate": 7.19033299094496e-08,
"loss": 0.0202,
"step": 572
},
{
"epoch": 5.79,
"learning_rate": 6.552220980555635e-08,
"loss": 0.0187,
"step": 573
},
{
"epoch": 5.8,
"learning_rate": 5.943661777680354e-08,
"loss": 0.0167,
"step": 574
},
{
"epoch": 5.81,
"learning_rate": 5.364673485568794e-08,
"loss": 0.0196,
"step": 575
},
{
"epoch": 5.82,
"learning_rate": 4.815273327803183e-08,
"loss": 0.0191,
"step": 576
},
{
"epoch": 5.83,
"learning_rate": 4.295477647786039e-08,
"loss": 0.0208,
"step": 577
},
{
"epoch": 5.84,
"learning_rate": 3.805301908254455e-08,
"loss": 0.0187,
"step": 578
},
{
"epoch": 5.85,
"learning_rate": 3.3447606908196815e-08,
"loss": 0.0176,
"step": 579
},
{
"epoch": 5.86,
"learning_rate": 2.9138676955333676e-08,
"loss": 0.0206,
"step": 580
},
{
"epoch": 5.87,
"learning_rate": 2.512635740480218e-08,
"loss": 0.0177,
"step": 581
},
{
"epoch": 5.88,
"learning_rate": 2.1410767613965212e-08,
"loss": 0.0181,
"step": 582
},
{
"epoch": 5.89,
"learning_rate": 1.79920181131521e-08,
"loss": 0.0199,
"step": 583
},
{
"epoch": 5.9,
"learning_rate": 1.487021060236904e-08,
"loss": 0.0202,
"step": 584
},
{
"epoch": 5.91,
"learning_rate": 1.2045437948275952e-08,
"loss": 0.0212,
"step": 585
},
{
"epoch": 5.92,
"learning_rate": 9.517784181422018e-09,
"loss": 0.0189,
"step": 586
},
{
"epoch": 5.93,
"learning_rate": 7.2873244937476935e-09,
"loss": 0.0204,
"step": 587
},
{
"epoch": 5.94,
"learning_rate": 5.354125236343155e-09,
"loss": 0.0187,
"step": 588
},
{
"epoch": 5.95,
"learning_rate": 3.7182439174832106e-09,
"loss": 0.0187,
"step": 589
},
{
"epoch": 5.96,
"learning_rate": 2.379729200908676e-09,
"loss": 0.0173,
"step": 590
},
{
"epoch": 5.97,
"learning_rate": 1.3386209043819708e-09,
"loss": 0.0218,
"step": 591
},
{
"epoch": 5.98,
"learning_rate": 5.949499985025142e-10,
"loss": 0.0191,
"step": 592
},
{
"epoch": 5.99,
"learning_rate": 1.487386057841267e-10,
"loss": 0.0206,
"step": 593
},
{
"epoch": 6.0,
"learning_rate": 0.0,
"loss": 0.0174,
"step": 594
},
{
"epoch": 6.0,
"step": 594,
"total_flos": 76624596172800.0,
"train_loss": 0.5495017380155749,
"train_runtime": 4329.0457,
"train_samples_per_second": 13.059,
"train_steps_per_second": 0.137
}
],
"logging_steps": 1.0,
"max_steps": 594,
"num_input_tokens_seen": 0,
"num_train_epochs": 6,
"save_steps": 50000,
"total_flos": 76624596172800.0,
"train_batch_size": 24,
"trial_name": null,
"trial_params": null
}