|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 1.0, |
|
"eval_steps": 500, |
|
"global_step": 827, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.7619047619047623e-07, |
|
"loss": 0.0, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 9.523809523809525e-07, |
|
"loss": 0.0, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 1.4285714285714286e-06, |
|
"loss": 0.0, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 1.904761904761905e-06, |
|
"loss": 0.0, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 2.380952380952381e-06, |
|
"loss": 0.0, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 2.8571428571428573e-06, |
|
"loss": 0.0, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 3.3333333333333333e-06, |
|
"loss": 0.0, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 3.80952380952381e-06, |
|
"loss": 0.0, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.2857142857142855e-06, |
|
"loss": 0.0, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.761904761904762e-06, |
|
"loss": 0.0, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 5.2380952380952384e-06, |
|
"loss": 0.0, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 5.7142857142857145e-06, |
|
"loss": 0.0, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 6.1904761904761914e-06, |
|
"loss": 0.0, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 6.666666666666667e-06, |
|
"loss": 0.0, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 7.1428571428571436e-06, |
|
"loss": 0.0, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 7.61904761904762e-06, |
|
"loss": 0.0, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.095238095238097e-06, |
|
"loss": 0.0, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.571428571428571e-06, |
|
"loss": 0.0, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 9.047619047619049e-06, |
|
"loss": 0.0, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 9.523809523809525e-06, |
|
"loss": 0.0, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1e-05, |
|
"loss": 0.0, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.0476190476190477e-05, |
|
"loss": 0.0, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.0952380952380955e-05, |
|
"loss": 0.0, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.1428571428571429e-05, |
|
"loss": 0.0, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.1904761904761905e-05, |
|
"loss": 0.0, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.2380952380952383e-05, |
|
"loss": 0.0, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.2857142857142859e-05, |
|
"loss": 0.0, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.3333333333333333e-05, |
|
"loss": 0.0, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.3809523809523811e-05, |
|
"loss": 0.0, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.4285714285714287e-05, |
|
"loss": 0.0, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.4761904761904763e-05, |
|
"loss": 0.0, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.523809523809524e-05, |
|
"loss": 0.0, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.5714285714285715e-05, |
|
"loss": 0.0, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.6190476190476193e-05, |
|
"loss": 0.0, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.6666666666666667e-05, |
|
"loss": 0.0, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.7142857142857142e-05, |
|
"loss": 0.0, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.761904761904762e-05, |
|
"loss": 0.0, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.8095238095238097e-05, |
|
"loss": 0.0, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.8571428571428575e-05, |
|
"loss": 0.0, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.904761904761905e-05, |
|
"loss": 0.0, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9523809523809524e-05, |
|
"loss": 0.0, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 2e-05, |
|
"loss": 0.0, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.999991991893198e-05, |
|
"loss": 0.0, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9999679677010514e-05, |
|
"loss": 0.0, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9999279278083362e-05, |
|
"loss": 0.0, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9998718728563407e-05, |
|
"loss": 0.0, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9997998037428528e-05, |
|
"loss": 0.0, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9997117216221465e-05, |
|
"loss": 0.0, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.999607627904964e-05, |
|
"loss": 0.0, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9994875242584926e-05, |
|
"loss": 0.0, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.999351412606338e-05, |
|
"loss": 0.0, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.999199295128493e-05, |
|
"loss": 0.0, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9990311742613047e-05, |
|
"loss": 0.0, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.9988470526974316e-05, |
|
"loss": 0.0, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.9986469333858047e-05, |
|
"loss": 0.0, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.9984308195315775e-05, |
|
"loss": 0.0, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.998198714596076e-05, |
|
"loss": 0.0, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.9979506222967414e-05, |
|
"loss": 0.0, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.9976865466070737e-05, |
|
"loss": 0.0, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.9974064917565656e-05, |
|
"loss": 0.0, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.9971104622306355e-05, |
|
"loss": 0.0, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.996798462770555e-05, |
|
"loss": 0.0, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.9964704983733748e-05, |
|
"loss": 0.0, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.9961265742918418e-05, |
|
"loss": 0.0, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.9957666960343185e-05, |
|
"loss": 0.0, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.9953908693646913e-05, |
|
"loss": 0.0, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.994999100302281e-05, |
|
"loss": 0.0, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.9945913951217438e-05, |
|
"loss": 0.0, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.994167760352973e-05, |
|
"loss": 0.0, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.993728202780995e-05, |
|
"loss": 0.0, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.993272729445856e-05, |
|
"loss": 0.0, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.992801347642515e-05, |
|
"loss": 0.0, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.992314064920724e-05, |
|
"loss": 0.0, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.9918108890849064e-05, |
|
"loss": 0.0, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.991291828194034e-05, |
|
"loss": 0.0, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.9907568905614974e-05, |
|
"loss": 0.0, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.9902060847549716e-05, |
|
"loss": 0.0, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.9896394195962802e-05, |
|
"loss": 0.0, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9890569041612534e-05, |
|
"loss": 0.0, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9884585477795827e-05, |
|
"loss": 0.0, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.987844360034672e-05, |
|
"loss": 0.0, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.987214350763483e-05, |
|
"loss": 0.0, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9865685300563794e-05, |
|
"loss": 0.0, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.985906908256963e-05, |
|
"loss": 0.0, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.98522949596191e-05, |
|
"loss": 0.0, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9845363040208004e-05, |
|
"loss": 0.0, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9838273435359447e-05, |
|
"loss": 0.0, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.983102625862205e-05, |
|
"loss": 0.0, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9823621626068145e-05, |
|
"loss": 0.0, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.981605965629191e-05, |
|
"loss": 0.0, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9808340470407473e-05, |
|
"loss": 0.0, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9800464192046956e-05, |
|
"loss": 0.0, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9792430947358517e-05, |
|
"loss": 0.0, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9784240865004322e-05, |
|
"loss": 0.0, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9775894076158476e-05, |
|
"loss": 0.0, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9767390714504935e-05, |
|
"loss": 0.0, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9758730916235356e-05, |
|
"loss": 0.0, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9749914820046915e-05, |
|
"loss": 0.0, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9740942567140093e-05, |
|
"loss": 0.0, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9731814301216407e-05, |
|
"loss": 0.0, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.972253016847612e-05, |
|
"loss": 0.0, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9713090317615877e-05, |
|
"loss": 0.0, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.970349489982635e-05, |
|
"loss": 0.0, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.9693744068789803e-05, |
|
"loss": 0.0, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.9683837980677623e-05, |
|
"loss": 0.0, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.9673776794147838e-05, |
|
"loss": 0.0, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.966356067034256e-05, |
|
"loss": 0.0, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.96531897728854e-05, |
|
"loss": 0.0, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.964266426787888e-05, |
|
"loss": 0.0, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.9631984323901734e-05, |
|
"loss": 0.0, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.962115011200622e-05, |
|
"loss": 0.0, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.96101618057154e-05, |
|
"loss": 0.0, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.959901958102032e-05, |
|
"loss": 0.0, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.9587723616377245e-05, |
|
"loss": 0.0, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.9576274092704747e-05, |
|
"loss": 0.0, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.956467119338085e-05, |
|
"loss": 0.0, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.9552915104240067e-05, |
|
"loss": 0.0, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.9541006013570428e-05, |
|
"loss": 0.0, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.952894411211048e-05, |
|
"loss": 0.0, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.951672959304621e-05, |
|
"loss": 0.0, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.950436265200796e-05, |
|
"loss": 0.0, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9491843487067305e-05, |
|
"loss": 0.0, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9479172298733865e-05, |
|
"loss": 0.0, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9466349289952095e-05, |
|
"loss": 0.0, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9453374666098048e-05, |
|
"loss": 0.0, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9440248634976062e-05, |
|
"loss": 0.0, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9426971406815464e-05, |
|
"loss": 0.0, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9413543194267173e-05, |
|
"loss": 0.0, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.939996421240031e-05, |
|
"loss": 0.0, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.9386234678698753e-05, |
|
"loss": 0.0, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.9372354813057638e-05, |
|
"loss": 0.0, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.9358324837779864e-05, |
|
"loss": 0.0, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.934414497757251e-05, |
|
"loss": 0.0, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.9329815459543245e-05, |
|
"loss": 0.0, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.9315336513196693e-05, |
|
"loss": 0.0, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.930070837043075e-05, |
|
"loss": 0.0, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9285931265532875e-05, |
|
"loss": 0.0, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9271005435176335e-05, |
|
"loss": 0.0, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9255931118416424e-05, |
|
"loss": 0.0, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9240708556686613e-05, |
|
"loss": 0.0, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9225337993794702e-05, |
|
"loss": 0.0, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.920981967591891e-05, |
|
"loss": 0.0, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9194153851603935e-05, |
|
"loss": 0.0, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.917834077175696e-05, |
|
"loss": 0.0, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.9162380689643655e-05, |
|
"loss": 0.0, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.91462738608841e-05, |
|
"loss": 0.0, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.9130020543448705e-05, |
|
"loss": 0.0, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.9113620997654078e-05, |
|
"loss": 0.0, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.909707548615884e-05, |
|
"loss": 0.0, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.9080384273959445e-05, |
|
"loss": 0.0, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.9063547628385907e-05, |
|
"loss": 0.0, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.9046565819097546e-05, |
|
"loss": 0.0, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.9029439118078637e-05, |
|
"loss": 0.0, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.9012167799634085e-05, |
|
"loss": 0.0, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.899475214038502e-05, |
|
"loss": 0.0, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.8977192419264358e-05, |
|
"loss": 0.0, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.895948891751234e-05, |
|
"loss": 0.0, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.8941641918672037e-05, |
|
"loss": 0.0, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.8923651708584788e-05, |
|
"loss": 0.0, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.890551857538565e-05, |
|
"loss": 0.0, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.8887242809498748e-05, |
|
"loss": 0.0, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.8868824703632658e-05, |
|
"loss": 0.0, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.8850264552775693e-05, |
|
"loss": 0.0, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.8831562654191203e-05, |
|
"loss": 0.0, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.881271930741278e-05, |
|
"loss": 0.0, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.8793734814239496e-05, |
|
"loss": 0.0, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.8774609478731048e-05, |
|
"loss": 0.0, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.8755343607202894e-05, |
|
"loss": 0.0, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.873593750822135e-05, |
|
"loss": 0.0, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.871639149259864e-05, |
|
"loss": 0.0, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8696705873387927e-05, |
|
"loss": 0.0, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8676880965878293e-05, |
|
"loss": 0.0, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.865691708758969e-05, |
|
"loss": 0.0, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.863681455826786e-05, |
|
"loss": 0.0, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.86165736998792e-05, |
|
"loss": 0.0, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8596194836605633e-05, |
|
"loss": 0.0, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.857567829483937e-05, |
|
"loss": 0.0, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8555024403177742e-05, |
|
"loss": 0.0, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8534233492417887e-05, |
|
"loss": 0.0, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.851330589555146e-05, |
|
"loss": 0.0, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.849224194775934e-05, |
|
"loss": 0.0, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.84710419864062e-05, |
|
"loss": 0.0, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8449706351035157e-05, |
|
"loss": 0.0, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8428235383362304e-05, |
|
"loss": 0.0, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8406629427271242e-05, |
|
"loss": 0.0, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8384888828807585e-05, |
|
"loss": 0.0, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8363013936173393e-05, |
|
"loss": 0.0, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8341005099721626e-05, |
|
"loss": 0.0, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.831886267195051e-05, |
|
"loss": 0.0, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8296587007497893e-05, |
|
"loss": 0.0, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8274178463135578e-05, |
|
"loss": 0.0, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8251637397763597e-05, |
|
"loss": 0.0, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8228964172404472e-05, |
|
"loss": 0.0, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8206159150197417e-05, |
|
"loss": 0.0, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.8183222696392545e-05, |
|
"loss": 0.0, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.8160155178344992e-05, |
|
"loss": 0.0, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.8136956965509064e-05, |
|
"loss": 0.0, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.8113628429432286e-05, |
|
"loss": 0.0, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.8090169943749477e-05, |
|
"loss": 0.0, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.8066581884176754e-05, |
|
"loss": 0.0, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.8042864628505516e-05, |
|
"loss": 0.0, |
|
"step": 201 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.8019018556596402e-05, |
|
"loss": 0.0, |
|
"step": 202 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.7995044050373185e-05, |
|
"loss": 0.0, |
|
"step": 203 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.7970941493816684e-05, |
|
"loss": 0.0, |
|
"step": 204 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.794671127295859e-05, |
|
"loss": 0.0, |
|
"step": 205 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.79223537758753e-05, |
|
"loss": 0.0, |
|
"step": 206 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.7897869392681685e-05, |
|
"loss": 0.0, |
|
"step": 207 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.7873258515524857e-05, |
|
"loss": 0.0, |
|
"step": 208 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.7848521538577888e-05, |
|
"loss": 0.0, |
|
"step": 209 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.7823658858033474e-05, |
|
"loss": 0.0, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7798670872097625e-05, |
|
"loss": 0.0, |
|
"step": 211 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7773557980983264e-05, |
|
"loss": 0.0, |
|
"step": 212 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7748320586903805e-05, |
|
"loss": 0.0, |
|
"step": 213 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.772295909406676e-05, |
|
"loss": 0.0, |
|
"step": 214 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7697473908667205e-05, |
|
"loss": 0.0, |
|
"step": 215 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7671865438881316e-05, |
|
"loss": 0.0, |
|
"step": 216 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7646134094859816e-05, |
|
"loss": 0.0, |
|
"step": 217 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7620280288721402e-05, |
|
"loss": 0.0, |
|
"step": 218 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7594304434546163e-05, |
|
"loss": 0.0, |
|
"step": 219 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.7568206948368925e-05, |
|
"loss": 0.0, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.75419882481726e-05, |
|
"loss": 0.0, |
|
"step": 221 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.7515648753881495e-05, |
|
"loss": 0.0, |
|
"step": 222 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.7489188887354567e-05, |
|
"loss": 0.0, |
|
"step": 223 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.74626090723787e-05, |
|
"loss": 0.0, |
|
"step": 224 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.7435909734661885e-05, |
|
"loss": 0.0, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.740909130182642e-05, |
|
"loss": 0.0, |
|
"step": 226 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.738215420340205e-05, |
|
"loss": 0.0, |
|
"step": 227 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.7355098870819097e-05, |
|
"loss": 0.0, |
|
"step": 228 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.7327925737401546e-05, |
|
"loss": 0.0, |
|
"step": 229 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.7300635238360114e-05, |
|
"loss": 0.0, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.7273227810785256e-05, |
|
"loss": 0.0, |
|
"step": 231 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.7245703893640188e-05, |
|
"loss": 0.0, |
|
"step": 232 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.7218063927753848e-05, |
|
"loss": 0.0, |
|
"step": 233 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.719030835581383e-05, |
|
"loss": 0.0, |
|
"step": 234 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.716243762235931e-05, |
|
"loss": 0.0, |
|
"step": 235 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.7134452173773897e-05, |
|
"loss": 0.0, |
|
"step": 236 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.7106352458278524e-05, |
|
"loss": 0.0, |
|
"step": 237 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.707813892592423e-05, |
|
"loss": 0.0, |
|
"step": 238 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.7049812028584977e-05, |
|
"loss": 0.0, |
|
"step": 239 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.7021372219950403e-05, |
|
"loss": 0.0, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.6992819955518562e-05, |
|
"loss": 0.0, |
|
"step": 241 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.696415569258862e-05, |
|
"loss": 0.0, |
|
"step": 242 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.6935379890253528e-05, |
|
"loss": 0.0, |
|
"step": 243 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.690649300939269e-05, |
|
"loss": 0.0, |
|
"step": 244 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.6877495512664555e-05, |
|
"loss": 0.0, |
|
"step": 245 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.6848387864499228e-05, |
|
"loss": 0.0, |
|
"step": 246 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.6819170531091018e-05, |
|
"loss": 0.0, |
|
"step": 247 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.678984398039098e-05, |
|
"loss": 0.0, |
|
"step": 248 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.676040868209941e-05, |
|
"loss": 0.0, |
|
"step": 249 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.673086510765834e-05, |
|
"loss": 0.0, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.670121373024396e-05, |
|
"loss": 0.0, |
|
"step": 251 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.667145502475907e-05, |
|
"loss": 0.0, |
|
"step": 252 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.6641589467825455e-05, |
|
"loss": 0.0, |
|
"step": 253 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.6611617537776246e-05, |
|
"loss": 0.0, |
|
"step": 254 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.6581539714648288e-05, |
|
"loss": 0.0, |
|
"step": 255 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.655135648017441e-05, |
|
"loss": 0.0, |
|
"step": 256 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.6521068317775756e-05, |
|
"loss": 0.0, |
|
"step": 257 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.649067571255399e-05, |
|
"loss": 0.0, |
|
"step": 258 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.646017915128358e-05, |
|
"loss": 0.0, |
|
"step": 259 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.6429579122403957e-05, |
|
"loss": 0.0, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6398876116011727e-05, |
|
"loss": 0.0, |
|
"step": 261 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6368070623852794e-05, |
|
"loss": 0.0, |
|
"step": 262 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6337163139314506e-05, |
|
"loss": 0.0, |
|
"step": 263 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6306154157417733e-05, |
|
"loss": 0.0, |
|
"step": 264 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6275044174808948e-05, |
|
"loss": 0.0, |
|
"step": 265 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.624383368975229e-05, |
|
"loss": 0.0, |
|
"step": 266 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6212523202121547e-05, |
|
"loss": 0.0, |
|
"step": 267 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6181113213392176e-05, |
|
"loss": 0.0, |
|
"step": 268 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.6149604226633267e-05, |
|
"loss": 0.0, |
|
"step": 269 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.611799674649948e-05, |
|
"loss": 0.0, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.6086291279222975e-05, |
|
"loss": 0.0, |
|
"step": 271 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.6054488332605282e-05, |
|
"loss": 0.0, |
|
"step": 272 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.6022588416009192e-05, |
|
"loss": 0.0, |
|
"step": 273 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.599059204035058e-05, |
|
"loss": 0.0, |
|
"step": 274 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.5958499718090233e-05, |
|
"loss": 0.0, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.592631196322564e-05, |
|
"loss": 0.0, |
|
"step": 276 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.589402929128276e-05, |
|
"loss": 0.0, |
|
"step": 277 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.5861652219307763e-05, |
|
"loss": 0.0, |
|
"step": 278 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.5829181265858745e-05, |
|
"loss": 0.0, |
|
"step": 279 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.5796616950997433e-05, |
|
"loss": 0.0, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.5763959796280855e-05, |
|
"loss": 0.0, |
|
"step": 281 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.573121032475297e-05, |
|
"loss": 0.0, |
|
"step": 282 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.569836906093632e-05, |
|
"loss": 0.0, |
|
"step": 283 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.566543653082359e-05, |
|
"loss": 0.0, |
|
"step": 284 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.5632413261869223e-05, |
|
"loss": 0.0, |
|
"step": 285 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5599299782980945e-05, |
|
"loss": 0.0, |
|
"step": 286 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5566096624511306e-05, |
|
"loss": 0.0, |
|
"step": 287 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.553280431824919e-05, |
|
"loss": 0.0, |
|
"step": 288 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5499423397411283e-05, |
|
"loss": 0.0, |
|
"step": 289 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5465954396633543e-05, |
|
"loss": 0.0, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.543239785196263e-05, |
|
"loss": 0.0, |
|
"step": 291 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5398754300847346e-05, |
|
"loss": 0.0, |
|
"step": 292 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5365024282129976e-05, |
|
"loss": 0.0, |
|
"step": 293 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.533120833603771e-05, |
|
"loss": 0.0, |
|
"step": 294 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.5297307004173974e-05, |
|
"loss": 0.0, |
|
"step": 295 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.5263320829509724e-05, |
|
"loss": 0.0, |
|
"step": 296 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.5229250356374804e-05, |
|
"loss": 0.0, |
|
"step": 297 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.5195096130449188e-05, |
|
"loss": 0.0, |
|
"step": 298 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.5160858698754245e-05, |
|
"loss": 0.0, |
|
"step": 299 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.512653860964401e-05, |
|
"loss": 0.0, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.509213641279635e-05, |
|
"loss": 0.0, |
|
"step": 301 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.5057652659204198e-05, |
|
"loss": 0.0, |
|
"step": 302 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.5023087901166724e-05, |
|
"loss": 0.0, |
|
"step": 303 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4988442692280468e-05, |
|
"loss": 0.0, |
|
"step": 304 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4953717587430504e-05, |
|
"loss": 0.0, |
|
"step": 305 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4918913142781524e-05, |
|
"loss": 0.0, |
|
"step": 306 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4884029915768945e-05, |
|
"loss": 0.0, |
|
"step": 307 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4849068465089986e-05, |
|
"loss": 0.0, |
|
"step": 308 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4814029350694712e-05, |
|
"loss": 0.0, |
|
"step": 309 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4778913133777057e-05, |
|
"loss": 0.0, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.474372037676586e-05, |
|
"loss": 0.0, |
|
"step": 311 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4708451643315827e-05, |
|
"loss": 0.0, |
|
"step": 312 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4673107498298527e-05, |
|
"loss": 0.0, |
|
"step": 313 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4637688507793342e-05, |
|
"loss": 0.0, |
|
"step": 314 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4602195239078385e-05, |
|
"loss": 0.0, |
|
"step": 315 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4566628260621432e-05, |
|
"loss": 0.0, |
|
"step": 316 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4530988142070802e-05, |
|
"loss": 0.0, |
|
"step": 317 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4495275454246255e-05, |
|
"loss": 0.0, |
|
"step": 318 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.4459490769129818e-05, |
|
"loss": 0.0, |
|
"step": 319 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.442363465985666e-05, |
|
"loss": 0.0, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.4387707700705882e-05, |
|
"loss": 0.0, |
|
"step": 321 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.4351710467091337e-05, |
|
"loss": 0.0, |
|
"step": 322 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.4315643535552407e-05, |
|
"loss": 0.0, |
|
"step": 323 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.427950748374477e-05, |
|
"loss": 0.0, |
|
"step": 324 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.4243302890431155e-05, |
|
"loss": 0.0, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.420703033547206e-05, |
|
"loss": 0.0, |
|
"step": 326 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.4170690399816469e-05, |
|
"loss": 0.0, |
|
"step": 327 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.4134283665492558e-05, |
|
"loss": 0.0, |
|
"step": 328 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.4097810715598362e-05, |
|
"loss": 0.0, |
|
"step": 329 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.4061272134292434e-05, |
|
"loss": 0.0, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.40246685067845e-05, |
|
"loss": 0.0, |
|
"step": 331 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.3988000419326073e-05, |
|
"loss": 0.0, |
|
"step": 332 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.3951268459201078e-05, |
|
"loss": 0.0, |
|
"step": 333 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.3914473214716431e-05, |
|
"loss": 0.0, |
|
"step": 334 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.387761527519263e-05, |
|
"loss": 0.0, |
|
"step": 335 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3840695230954306e-05, |
|
"loss": 0.0, |
|
"step": 336 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3803713673320773e-05, |
|
"loss": 0.0, |
|
"step": 337 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3766671194596559e-05, |
|
"loss": 0.0, |
|
"step": 338 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3729568388061916e-05, |
|
"loss": 0.0, |
|
"step": 339 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3692405847963321e-05, |
|
"loss": 0.0, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.365518416950395e-05, |
|
"loss": 0.0, |
|
"step": 341 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3617903948834155e-05, |
|
"loss": 0.0, |
|
"step": 342 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3580565783041916e-05, |
|
"loss": 0.0, |
|
"step": 343 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3543170270143275e-05, |
|
"loss": 0.0, |
|
"step": 344 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3505718009072752e-05, |
|
"loss": 0.0, |
|
"step": 345 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3468209599673755e-05, |
|
"loss": 0.0, |
|
"step": 346 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.343064564268899e-05, |
|
"loss": 0.0, |
|
"step": 347 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3393026739750807e-05, |
|
"loss": 0.0, |
|
"step": 348 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3355353493371595e-05, |
|
"loss": 0.0, |
|
"step": 349 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.331762650693412e-05, |
|
"loss": 0.0, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3279846384681846e-05, |
|
"loss": 0.0, |
|
"step": 351 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.324201373170929e-05, |
|
"loss": 0.0, |
|
"step": 352 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.3204129153952296e-05, |
|
"loss": 0.0, |
|
"step": 353 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.3166193258178356e-05, |
|
"loss": 0.0, |
|
"step": 354 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.3128206651976884e-05, |
|
"loss": 0.0, |
|
"step": 355 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.3090169943749475e-05, |
|
"loss": 0.0, |
|
"step": 356 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.3052083742700172e-05, |
|
"loss": 0.0, |
|
"step": 357 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.301394865882571e-05, |
|
"loss": 0.0, |
|
"step": 358 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.2975765302905734e-05, |
|
"loss": 0.0, |
|
"step": 359 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.293753428649303e-05, |
|
"loss": 0.0, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2899256221903728e-05, |
|
"loss": 0.0, |
|
"step": 361 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.286093172220748e-05, |
|
"loss": 0.0, |
|
"step": 362 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2822561401217662e-05, |
|
"loss": 0.0, |
|
"step": 363 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2784145873481529e-05, |
|
"loss": 0.0, |
|
"step": 364 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2745685754270383e-05, |
|
"loss": 0.0, |
|
"step": 365 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2707181659569706e-05, |
|
"loss": 0.0, |
|
"step": 366 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2668634206069305e-05, |
|
"loss": 0.0, |
|
"step": 367 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2630044011153426e-05, |
|
"loss": 0.0, |
|
"step": 368 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2591411692890879e-05, |
|
"loss": 0.0, |
|
"step": 369 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2552737870025118e-05, |
|
"loss": 0.0, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2514023161964358e-05, |
|
"loss": 0.0, |
|
"step": 371 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2475268188771628e-05, |
|
"loss": 0.0, |
|
"step": 372 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2436473571154862e-05, |
|
"loss": 0.0, |
|
"step": 373 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2397639930456936e-05, |
|
"loss": 0.0, |
|
"step": 374 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2358767888645737e-05, |
|
"loss": 0.0, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2319858068304195e-05, |
|
"loss": 0.0, |
|
"step": 376 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2280911092620298e-05, |
|
"loss": 0.0, |
|
"step": 377 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2241927585377127e-05, |
|
"loss": 0.0, |
|
"step": 378 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2202908170942865e-05, |
|
"loss": 0.0, |
|
"step": 379 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2163853474260785e-05, |
|
"loss": 0.0, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2124764120839256e-05, |
|
"loss": 0.0, |
|
"step": 381 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2085640736741708e-05, |
|
"loss": 0.0, |
|
"step": 382 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2046483948576616e-05, |
|
"loss": 0.0, |
|
"step": 383 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2007294383487468e-05, |
|
"loss": 0.0, |
|
"step": 384 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1968072669142708e-05, |
|
"loss": 0.0, |
|
"step": 385 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1928819433725691e-05, |
|
"loss": 0.0, |
|
"step": 386 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1889535305924619e-05, |
|
"loss": 0.0, |
|
"step": 387 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1850220914922472e-05, |
|
"loss": 0.0, |
|
"step": 388 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.181087689038694e-05, |
|
"loss": 0.0, |
|
"step": 389 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1771503862460318e-05, |
|
"loss": 0.0, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1732102461749438e-05, |
|
"loss": 0.0, |
|
"step": 391 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1692673319315541e-05, |
|
"loss": 0.0, |
|
"step": 392 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1653217066664202e-05, |
|
"loss": 0.0, |
|
"step": 393 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.161373433573519e-05, |
|
"loss": 0.0, |
|
"step": 394 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1574225758892354e-05, |
|
"loss": 0.0, |
|
"step": 395 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1534691968913503e-05, |
|
"loss": 0.0, |
|
"step": 396 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1495133598980263e-05, |
|
"loss": 0.0, |
|
"step": 397 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1455551282667936e-05, |
|
"loss": 0.0, |
|
"step": 398 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.141594565393535e-05, |
|
"loss": 0.0, |
|
"step": 399 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.137631734711472e-05, |
|
"loss": 0.0, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1336666996901471e-05, |
|
"loss": 0.0, |
|
"step": 401 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1296995238344084e-05, |
|
"loss": 0.0, |
|
"step": 402 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1257302706833914e-05, |
|
"loss": 0.0, |
|
"step": 403 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1217590038095027e-05, |
|
"loss": 0.0, |
|
"step": 404 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1177857868174006e-05, |
|
"loss": 0.0, |
|
"step": 405 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1138106833429773e-05, |
|
"loss": 0.0, |
|
"step": 406 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1098337570523397e-05, |
|
"loss": 0.0, |
|
"step": 407 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1058550716407876e-05, |
|
"loss": 0.0, |
|
"step": 408 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1018746908317972e-05, |
|
"loss": 0.0, |
|
"step": 409 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.0978926783759973e-05, |
|
"loss": 0.0, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.0939090980501505e-05, |
|
"loss": 0.0, |
|
"step": 411 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.08992401365613e-05, |
|
"loss": 0.0, |
|
"step": 412 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.0859374890198987e-05, |
|
"loss": 0.0, |
|
"step": 413 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.0819495879904864e-05, |
|
"loss": 0.0, |
|
"step": 414 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.077960374438968e-05, |
|
"loss": 0.0, |
|
"step": 415 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.07396991225744e-05, |
|
"loss": 0.0, |
|
"step": 416 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.0699782653579973e-05, |
|
"loss": 0.0, |
|
"step": 417 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0659854976717088e-05, |
|
"loss": 0.0, |
|
"step": 418 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0619916731475946e-05, |
|
"loss": 0.0, |
|
"step": 419 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0579968557516018e-05, |
|
"loss": 0.0, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0540011094655787e-05, |
|
"loss": 0.0, |
|
"step": 421 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0500044982862519e-05, |
|
"loss": 0.0, |
|
"step": 422 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0460070862241991e-05, |
|
"loss": 0.0, |
|
"step": 423 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0420089373028259e-05, |
|
"loss": 0.0, |
|
"step": 424 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0380101155573398e-05, |
|
"loss": 0.0, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0340106850337237e-05, |
|
"loss": 0.0, |
|
"step": 426 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0300107097877114e-05, |
|
"loss": 0.0, |
|
"step": 427 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0260102538837608e-05, |
|
"loss": 0.0, |
|
"step": 428 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.022009381394028e-05, |
|
"loss": 0.0, |
|
"step": 429 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0180081563973412e-05, |
|
"loss": 0.0, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0140066429781753e-05, |
|
"loss": 0.0, |
|
"step": 431 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0100049052256236e-05, |
|
"loss": 0.0, |
|
"step": 432 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0060030072323726e-05, |
|
"loss": 0.0, |
|
"step": 433 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0020010130936753e-05, |
|
"loss": 0.0, |
|
"step": 434 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.97998986906325e-06, |
|
"loss": 0.0, |
|
"step": 435 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.939969927676279e-06, |
|
"loss": 0.0, |
|
"step": 436 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.899950947743767e-06, |
|
"loss": 0.0, |
|
"step": 437 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.85993357021825e-06, |
|
"loss": 0.0, |
|
"step": 438 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.81991843602659e-06, |
|
"loss": 0.0, |
|
"step": 439 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.779906186059724e-06, |
|
"loss": 0.0, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.739897461162395e-06, |
|
"loss": 0.0, |
|
"step": 441 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.699892902122887e-06, |
|
"loss": 0.0, |
|
"step": 442 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.659893149662766e-06, |
|
"loss": 0.0, |
|
"step": 443 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.619898844426605e-06, |
|
"loss": 0.0, |
|
"step": 444 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.579910626971743e-06, |
|
"loss": 0.0, |
|
"step": 445 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.539929137758012e-06, |
|
"loss": 0.0, |
|
"step": 446 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.499955017137485e-06, |
|
"loss": 0.0, |
|
"step": 447 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.459988905344216e-06, |
|
"loss": 0.0, |
|
"step": 448 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.420031442483985e-06, |
|
"loss": 0.0, |
|
"step": 449 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.380083268524056e-06, |
|
"loss": 0.0, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.340145023282917e-06, |
|
"loss": 0.0, |
|
"step": 451 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.300217346420029e-06, |
|
"loss": 0.0, |
|
"step": 452 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.260300877425602e-06, |
|
"loss": 0.0, |
|
"step": 453 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.220396255610321e-06, |
|
"loss": 0.0, |
|
"step": 454 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.180504120095139e-06, |
|
"loss": 0.0, |
|
"step": 455 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.140625109801017e-06, |
|
"loss": 0.0, |
|
"step": 456 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.100759863438702e-06, |
|
"loss": 0.0, |
|
"step": 457 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.060909019498498e-06, |
|
"loss": 0.0, |
|
"step": 458 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 9.021073216240029e-06, |
|
"loss": 0.0, |
|
"step": 459 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.981253091682031e-06, |
|
"loss": 0.0, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.941449283592127e-06, |
|
"loss": 0.0, |
|
"step": 461 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.901662429476607e-06, |
|
"loss": 0.0, |
|
"step": 462 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.861893166570228e-06, |
|
"loss": 0.0, |
|
"step": 463 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.822142131825995e-06, |
|
"loss": 0.0, |
|
"step": 464 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.782409961904976e-06, |
|
"loss": 0.0, |
|
"step": 465 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.742697293166087e-06, |
|
"loss": 0.0, |
|
"step": 466 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.703004761655918e-06, |
|
"loss": 0.0, |
|
"step": 467 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.663333003098532e-06, |
|
"loss": 0.0, |
|
"step": 468 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.623682652885284e-06, |
|
"loss": 0.0, |
|
"step": 469 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.584054346064652e-06, |
|
"loss": 0.0, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.544448717332068e-06, |
|
"loss": 0.0, |
|
"step": 471 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.504866401019738e-06, |
|
"loss": 0.0, |
|
"step": 472 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.465308031086499e-06, |
|
"loss": 0.0, |
|
"step": 473 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.425774241107648e-06, |
|
"loss": 0.0, |
|
"step": 474 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.386265664264815e-06, |
|
"loss": 0.0, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.346782933335801e-06, |
|
"loss": 0.0, |
|
"step": 476 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.30732668068446e-06, |
|
"loss": 0.0, |
|
"step": 477 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.267897538250567e-06, |
|
"loss": 0.0, |
|
"step": 478 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.228496137539685e-06, |
|
"loss": 0.0, |
|
"step": 479 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.189123109613064e-06, |
|
"loss": 0.0, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.14977908507753e-06, |
|
"loss": 0.0, |
|
"step": 481 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.110464694075383e-06, |
|
"loss": 0.0, |
|
"step": 482 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.07118056627431e-06, |
|
"loss": 0.0, |
|
"step": 483 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.031927330857294e-06, |
|
"loss": 0.0, |
|
"step": 484 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 7.992705616512533e-06, |
|
"loss": 0.0, |
|
"step": 485 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 7.953516051423385e-06, |
|
"loss": 0.0, |
|
"step": 486 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 7.914359263258295e-06, |
|
"loss": 0.0, |
|
"step": 487 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 7.875235879160747e-06, |
|
"loss": 0.0, |
|
"step": 488 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 7.836146525739217e-06, |
|
"loss": 0.0, |
|
"step": 489 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 7.797091829057138e-06, |
|
"loss": 0.0, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 7.758072414622876e-06, |
|
"loss": 0.0, |
|
"step": 491 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 7.719088907379705e-06, |
|
"loss": 0.0, |
|
"step": 492 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.680141931695808e-06, |
|
"loss": 0.0, |
|
"step": 493 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.641232111354265e-06, |
|
"loss": 0.0, |
|
"step": 494 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.602360069543068e-06, |
|
"loss": 0.0, |
|
"step": 495 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.563526428845143e-06, |
|
"loss": 0.0, |
|
"step": 496 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.524731811228374e-06, |
|
"loss": 0.0, |
|
"step": 497 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.485976838035645e-06, |
|
"loss": 0.0, |
|
"step": 498 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.447262129974886e-06, |
|
"loss": 0.0, |
|
"step": 499 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.4085883071091255e-06, |
|
"loss": 0.0, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.369955988846577e-06, |
|
"loss": 0.0, |
|
"step": 501 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.331365793930698e-06, |
|
"loss": 0.0, |
|
"step": 502 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.292818340430297e-06, |
|
"loss": 0.0, |
|
"step": 503 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.2543142457296215e-06, |
|
"loss": 0.0, |
|
"step": 504 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.215854126518474e-06, |
|
"loss": 0.0, |
|
"step": 505 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.177438598782342e-06, |
|
"loss": 0.0, |
|
"step": 506 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.139068277792524e-06, |
|
"loss": 0.0, |
|
"step": 507 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.100743778096275e-06, |
|
"loss": 0.0, |
|
"step": 508 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 7.062465713506972e-06, |
|
"loss": 0.0, |
|
"step": 509 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 7.024234697094268e-06, |
|
"loss": 0.0, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 6.9860513411742934e-06, |
|
"loss": 0.0, |
|
"step": 511 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 6.947916257299829e-06, |
|
"loss": 0.0, |
|
"step": 512 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 6.909830056250527e-06, |
|
"loss": 0.0, |
|
"step": 513 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 6.871793348023119e-06, |
|
"loss": 0.0, |
|
"step": 514 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 6.833806741821644e-06, |
|
"loss": 0.0, |
|
"step": 515 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 6.795870846047707e-06, |
|
"loss": 0.0, |
|
"step": 516 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.757986268290713e-06, |
|
"loss": 0.0, |
|
"step": 517 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.7201536153181545e-06, |
|
"loss": 0.0, |
|
"step": 518 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.682373493065883e-06, |
|
"loss": 0.0, |
|
"step": 519 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.644646506628407e-06, |
|
"loss": 0.0, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.606973260249197e-06, |
|
"loss": 0.0, |
|
"step": 521 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.569354357311015e-06, |
|
"loss": 0.0, |
|
"step": 522 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.531790400326246e-06, |
|
"loss": 0.0, |
|
"step": 523 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.494281990927251e-06, |
|
"loss": 0.0, |
|
"step": 524 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.456829729856728e-06, |
|
"loss": 0.0, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.419434216958084e-06, |
|
"loss": 0.0, |
|
"step": 526 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.382096051165847e-06, |
|
"loss": 0.0, |
|
"step": 527 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.344815830496054e-06, |
|
"loss": 0.0, |
|
"step": 528 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.307594152036681e-06, |
|
"loss": 0.0, |
|
"step": 529 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.270431611938086e-06, |
|
"loss": 0.0, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.233328805403442e-06, |
|
"loss": 0.0, |
|
"step": 531 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.196286326679231e-06, |
|
"loss": 0.0, |
|
"step": 532 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.159304769045698e-06, |
|
"loss": 0.0, |
|
"step": 533 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.122384724807374e-06, |
|
"loss": 0.0, |
|
"step": 534 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.085526785283575e-06, |
|
"loss": 0.0, |
|
"step": 535 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.048731540798928e-06, |
|
"loss": 0.0, |
|
"step": 536 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.0119995806739316e-06, |
|
"loss": 0.0, |
|
"step": 537 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 5.975331493215505e-06, |
|
"loss": 0.0, |
|
"step": 538 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 5.93872786570757e-06, |
|
"loss": 0.0, |
|
"step": 539 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 5.902189284401642e-06, |
|
"loss": 0.0, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 5.865716334507445e-06, |
|
"loss": 0.0, |
|
"step": 541 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.829309600183536e-06, |
|
"loss": 0.0, |
|
"step": 542 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.792969664527946e-06, |
|
"loss": 0.0, |
|
"step": 543 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.756697109568849e-06, |
|
"loss": 0.0, |
|
"step": 544 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.720492516255233e-06, |
|
"loss": 0.0, |
|
"step": 545 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.6843564644475966e-06, |
|
"loss": 0.0, |
|
"step": 546 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.648289532908666e-06, |
|
"loss": 0.0, |
|
"step": 547 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.612292299294124e-06, |
|
"loss": 0.0, |
|
"step": 548 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.576365340143345e-06, |
|
"loss": 0.0, |
|
"step": 549 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.540509230870186e-06, |
|
"loss": 0.0, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.504724545753751e-06, |
|
"loss": 0.0, |
|
"step": 551 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.469011857929202e-06, |
|
"loss": 0.0, |
|
"step": 552 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.433371739378573e-06, |
|
"loss": 0.0, |
|
"step": 553 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.397804760921619e-06, |
|
"loss": 0.0, |
|
"step": 554 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.3623114922066626e-06, |
|
"loss": 0.0, |
|
"step": 555 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.326892501701475e-06, |
|
"loss": 0.0, |
|
"step": 556 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.291548356684177e-06, |
|
"loss": 0.0, |
|
"step": 557 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.256279623234147e-06, |
|
"loss": 0.0, |
|
"step": 558 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.221086866222946e-06, |
|
"loss": 0.0, |
|
"step": 559 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.185970649305293e-06, |
|
"loss": 0.0, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.150931534910018e-06, |
|
"loss": 0.0, |
|
"step": 561 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.115970084231059e-06, |
|
"loss": 0.0, |
|
"step": 562 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.081086857218482e-06, |
|
"loss": 0.0, |
|
"step": 563 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.0462824125695005e-06, |
|
"loss": 0.0, |
|
"step": 564 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.011557307719535e-06, |
|
"loss": 0.0, |
|
"step": 565 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 4.97691209883328e-06, |
|
"loss": 0.0, |
|
"step": 566 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 4.942347340795803e-06, |
|
"loss": 0.0, |
|
"step": 567 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 4.907863587203657e-06, |
|
"loss": 0.0, |
|
"step": 568 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 4.873461390355996e-06, |
|
"loss": 0.0, |
|
"step": 569 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 4.839141301245757e-06, |
|
"loss": 0.0, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 4.804903869550819e-06, |
|
"loss": 0.0, |
|
"step": 571 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 4.7707496436252e-06, |
|
"loss": 0.0, |
|
"step": 572 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 4.736679170490279e-06, |
|
"loss": 0.0, |
|
"step": 573 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 4.702692995826032e-06, |
|
"loss": 0.0, |
|
"step": 574 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.66879166396229e-06, |
|
"loss": 0.0, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.634975717870027e-06, |
|
"loss": 0.0, |
|
"step": 576 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.601245699152659e-06, |
|
"loss": 0.0, |
|
"step": 577 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.56760214803737e-06, |
|
"loss": 0.0, |
|
"step": 578 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.534045603366465e-06, |
|
"loss": 0.0, |
|
"step": 579 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.500576602588723e-06, |
|
"loss": 0.0, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.467195681750814e-06, |
|
"loss": 0.0, |
|
"step": 581 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.4339033754886974e-06, |
|
"loss": 0.0, |
|
"step": 582 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.40070021701906e-06, |
|
"loss": 0.0, |
|
"step": 583 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.367586738130782e-06, |
|
"loss": 0.0, |
|
"step": 584 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.334563469176414e-06, |
|
"loss": 0.0, |
|
"step": 585 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.3016309390636835e-06, |
|
"loss": 0.0, |
|
"step": 586 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.268789675247029e-06, |
|
"loss": 0.0, |
|
"step": 587 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.236040203719146e-06, |
|
"loss": 0.0, |
|
"step": 588 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.203383049002572e-06, |
|
"loss": 0.0, |
|
"step": 589 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.170818734141262e-06, |
|
"loss": 0.0, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.1383477806922435e-06, |
|
"loss": 0.0, |
|
"step": 591 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.105970708717244e-06, |
|
"loss": 0.0, |
|
"step": 592 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.0736880367743625e-06, |
|
"loss": 0.0, |
|
"step": 593 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.041500281909771e-06, |
|
"loss": 0.0, |
|
"step": 594 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.009407959649424e-06, |
|
"loss": 0.0, |
|
"step": 595 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 3.977411583990811e-06, |
|
"loss": 0.0, |
|
"step": 596 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 3.945511667394719e-06, |
|
"loss": 0.0, |
|
"step": 597 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 3.913708720777027e-06, |
|
"loss": 0.0, |
|
"step": 598 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 3.882003253500524e-06, |
|
"loss": 0.0, |
|
"step": 599 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.850395773366739e-06, |
|
"loss": 0.0, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.81888678660783e-06, |
|
"loss": 0.0, |
|
"step": 601 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.787476797878459e-06, |
|
"loss": 0.0, |
|
"step": 602 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.756166310247713e-06, |
|
"loss": 0.0, |
|
"step": 603 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.724955825191052e-06, |
|
"loss": 0.0, |
|
"step": 604 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.6938458425822722e-06, |
|
"loss": 0.0, |
|
"step": 605 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.6628368606854968e-06, |
|
"loss": 0.0, |
|
"step": 606 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.6319293761472073e-06, |
|
"loss": 0.0, |
|
"step": 607 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.6011238839882745e-06, |
|
"loss": 0.0, |
|
"step": 608 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.570420877596048e-06, |
|
"loss": 0.0, |
|
"step": 609 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.5398208487164266e-06, |
|
"loss": 0.0, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.5093242874460142e-06, |
|
"loss": 0.0, |
|
"step": 611 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.47893168222425e-06, |
|
"loss": 0.0, |
|
"step": 612 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.4486435198255906e-06, |
|
"loss": 0.0, |
|
"step": 613 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.418460285351716e-06, |
|
"loss": 0.0, |
|
"step": 614 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.388382462223756e-06, |
|
"loss": 0.0, |
|
"step": 615 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.3584105321745498e-06, |
|
"loss": 0.0, |
|
"step": 616 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.3285449752409315e-06, |
|
"loss": 0.0, |
|
"step": 617 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.2987862697560423e-06, |
|
"loss": 0.0, |
|
"step": 618 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.269134892341663e-06, |
|
"loss": 0.0, |
|
"step": 619 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.2395913179005934e-06, |
|
"loss": 0.0, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.210156019609025e-06, |
|
"loss": 0.0, |
|
"step": 621 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.1808294689089856e-06, |
|
"loss": 0.0, |
|
"step": 622 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.1516121355007757e-06, |
|
"loss": 0.0, |
|
"step": 623 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.1225044873354492e-06, |
|
"loss": 0.0, |
|
"step": 624 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.0935069906073144e-06, |
|
"loss": 0.0, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.0646201097464756e-06, |
|
"loss": 0.0, |
|
"step": 626 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.035844307411384e-06, |
|
"loss": 0.0, |
|
"step": 627 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.0071800444814394e-06, |
|
"loss": 0.0, |
|
"step": 628 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 2.9786277800495977e-06, |
|
"loss": 0.0, |
|
"step": 629 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 2.9501879714150294e-06, |
|
"loss": 0.0, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 2.9218610740757734e-06, |
|
"loss": 0.0, |
|
"step": 631 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 2.8936475417214795e-06, |
|
"loss": 0.0, |
|
"step": 632 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.865547826226104e-06, |
|
"loss": 0.0, |
|
"step": 633 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.8375623776406935e-06, |
|
"loss": 0.0, |
|
"step": 634 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.8096916441861687e-06, |
|
"loss": 0.0, |
|
"step": 635 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.781936072246152e-06, |
|
"loss": 0.0, |
|
"step": 636 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.754296106359811e-06, |
|
"loss": 0.0, |
|
"step": 637 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.726772189214746e-06, |
|
"loss": 0.0, |
|
"step": 638 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.6993647616398876e-06, |
|
"loss": 0.0, |
|
"step": 639 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.672074262598454e-06, |
|
"loss": 0.0, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.644901129180907e-06, |
|
"loss": 0.0, |
|
"step": 641 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.6178457965979543e-06, |
|
"loss": 0.0, |
|
"step": 642 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.5909086981735834e-06, |
|
"loss": 0.0, |
|
"step": 643 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.564090265338115e-06, |
|
"loss": 0.0, |
|
"step": 644 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.5373909276212994e-06, |
|
"loss": 0.0, |
|
"step": 645 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.5108111126454326e-06, |
|
"loss": 0.0, |
|
"step": 646 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.484351246118507e-06, |
|
"loss": 0.0, |
|
"step": 647 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.4580117518274017e-06, |
|
"loss": 0.0, |
|
"step": 648 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.431793051631076e-06, |
|
"loss": 0.0, |
|
"step": 649 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.4056955654538373e-06, |
|
"loss": 0.0, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.3797197112785975e-06, |
|
"loss": 0.0, |
|
"step": 651 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.353865905140187e-06, |
|
"loss": 0.0, |
|
"step": 652 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.3281345611186847e-06, |
|
"loss": 0.0, |
|
"step": 653 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.3025260913327953e-06, |
|
"loss": 0.0, |
|
"step": 654 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.277040905933241e-06, |
|
"loss": 0.0, |
|
"step": 655 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.2516794130961926e-06, |
|
"loss": 0.0, |
|
"step": 656 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.226442019016739e-06, |
|
"loss": 0.0, |
|
"step": 657 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.2013291279023763e-06, |
|
"loss": 0.0, |
|
"step": 658 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.176341141966528e-06, |
|
"loss": 0.0, |
|
"step": 659 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.151478461422116e-06, |
|
"loss": 0.0, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.126741484475143e-06, |
|
"loss": 0.0, |
|
"step": 661 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.1021306073183166e-06, |
|
"loss": 0.0, |
|
"step": 662 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.077646224124702e-06, |
|
"loss": 0.0, |
|
"step": 663 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.0532887270414104e-06, |
|
"loss": 0.0, |
|
"step": 664 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.029058506183317e-06, |
|
"loss": 0.0, |
|
"step": 665 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.0049559496268154e-06, |
|
"loss": 0.0, |
|
"step": 666 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.9809814434036e-06, |
|
"loss": 0.0, |
|
"step": 667 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.9571353714944842e-06, |
|
"loss": 0.0, |
|
"step": 668 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.93341811582325e-06, |
|
"loss": 0.0, |
|
"step": 669 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.9098300562505266e-06, |
|
"loss": 0.0, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.8863715705677167e-06, |
|
"loss": 0.0, |
|
"step": 671 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.8630430344909378e-06, |
|
"loss": 0.0, |
|
"step": 672 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.839844821655008e-06, |
|
"loss": 0.0, |
|
"step": 673 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.8167773036074576e-06, |
|
"loss": 0.0, |
|
"step": 674 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.793840849802584e-06, |
|
"loss": 0.0, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.7710358275955297e-06, |
|
"loss": 0.0, |
|
"step": 676 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.748362602236403e-06, |
|
"loss": 0.0, |
|
"step": 677 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.7258215368644226e-06, |
|
"loss": 0.0, |
|
"step": 678 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.7034129925021103e-06, |
|
"loss": 0.0, |
|
"step": 679 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.6811373280494946e-06, |
|
"loss": 0.0, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.6589949002783768e-06, |
|
"loss": 0.0, |
|
"step": 681 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.63698606382661e-06, |
|
"loss": 0.0, |
|
"step": 682 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.6151111711924194e-06, |
|
"loss": 0.0, |
|
"step": 683 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.5933705727287597e-06, |
|
"loss": 0.0, |
|
"step": 684 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.5717646166376988e-06, |
|
"loss": 0.0, |
|
"step": 685 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.5502936489648434e-06, |
|
"loss": 0.0, |
|
"step": 686 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.528958013593801e-06, |
|
"loss": 0.0, |
|
"step": 687 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.507758052240662e-06, |
|
"loss": 0.0, |
|
"step": 688 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.4866941044485405e-06, |
|
"loss": 0.0, |
|
"step": 689 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.465766507582117e-06, |
|
"loss": 0.0, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.4449755968222578e-06, |
|
"loss": 0.0, |
|
"step": 691 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.4243217051606285e-06, |
|
"loss": 0.0, |
|
"step": 692 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.4038051633943705e-06, |
|
"loss": 0.0, |
|
"step": 693 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.3834263001207993e-06, |
|
"loss": 0.0, |
|
"step": 694 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.3631854417321422e-06, |
|
"loss": 0.0, |
|
"step": 695 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.3430829124103106e-06, |
|
"loss": 0.0, |
|
"step": 696 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.3231190341217081e-06, |
|
"loss": 0.0, |
|
"step": 697 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.3032941266120736e-06, |
|
"loss": 0.0, |
|
"step": 698 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.2836085074013627e-06, |
|
"loss": 0.0, |
|
"step": 699 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.2640624917786525e-06, |
|
"loss": 0.0, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.2446563927971079e-06, |
|
"loss": 0.0, |
|
"step": 701 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.2253905212689554e-06, |
|
"loss": 0.0, |
|
"step": 702 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.2062651857605069e-06, |
|
"loss": 0.0, |
|
"step": 703 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.1872806925872226e-06, |
|
"loss": 0.0, |
|
"step": 704 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.1684373458088006e-06, |
|
"loss": 0.0, |
|
"step": 705 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.149735447224306e-06, |
|
"loss": 0.0, |
|
"step": 706 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.1311752963673438e-06, |
|
"loss": 0.0, |
|
"step": 707 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.1127571905012534e-06, |
|
"loss": 0.0, |
|
"step": 708 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.094481424614352e-06, |
|
"loss": 0.0, |
|
"step": 709 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.0763482914152136e-06, |
|
"loss": 0.0, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.0583580813279671e-06, |
|
"loss": 0.0, |
|
"step": 711 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.0405110824876619e-06, |
|
"loss": 0.0, |
|
"step": 712 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.0228075807356453e-06, |
|
"loss": 0.0, |
|
"step": 713 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.0052478596149817e-06, |
|
"loss": 0.0, |
|
"step": 714 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 9.878322003659157e-07, |
|
"loss": 0.0, |
|
"step": 715 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 9.705608819213664e-07, |
|
"loss": 0.0, |
|
"step": 716 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 9.534341809024583e-07, |
|
"loss": 0.0, |
|
"step": 717 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 9.364523716140939e-07, |
|
"loss": 0.0, |
|
"step": 718 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 9.196157260405569e-07, |
|
"loss": 0.0, |
|
"step": 719 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 9.029245138411635e-07, |
|
"loss": 0.0, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 8.863790023459274e-07, |
|
"loss": 0.0, |
|
"step": 721 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 8.699794565512976e-07, |
|
"loss": 0.0, |
|
"step": 722 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 8.537261391159035e-07, |
|
"loss": 0.0, |
|
"step": 723 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 8.376193103563479e-07, |
|
"loss": 0.0, |
|
"step": 724 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 8.216592282430413e-07, |
|
"loss": 0.0, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 8.058461483960667e-07, |
|
"loss": 0.0, |
|
"step": 726 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 7.901803240810901e-07, |
|
"loss": 0.0, |
|
"step": 727 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 7.746620062053e-07, |
|
"loss": 0.0, |
|
"step": 728 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 7.592914433133891e-07, |
|
"loss": 0.0, |
|
"step": 729 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 7.440688815835795e-07, |
|
"loss": 0.0, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 7.289945648236651e-07, |
|
"loss": 0.0, |
|
"step": 731 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 7.140687344671282e-07, |
|
"loss": 0.0, |
|
"step": 732 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 6.992916295692531e-07, |
|
"loss": 0.0, |
|
"step": 733 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 6.846634868033097e-07, |
|
"loss": 0.0, |
|
"step": 734 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 6.701845404567575e-07, |
|
"loss": 0.0, |
|
"step": 735 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 6.55855022427494e-07, |
|
"loss": 0.0, |
|
"step": 736 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 6.416751622201389e-07, |
|
"loss": 0.0, |
|
"step": 737 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 6.276451869423639e-07, |
|
"loss": 0.0, |
|
"step": 738 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 6.137653213012507e-07, |
|
"loss": 0.0, |
|
"step": 739 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 6.000357875996909e-07, |
|
"loss": 0.0, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 5.86456805732829e-07, |
|
"loss": 0.0, |
|
"step": 741 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 5.730285931845381e-07, |
|
"loss": 0.0, |
|
"step": 742 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 5.597513650239405e-07, |
|
"loss": 0.0, |
|
"step": 743 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 5.466253339019578e-07, |
|
"loss": 0.0, |
|
"step": 744 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 5.336507100479071e-07, |
|
"loss": 0.0, |
|
"step": 745 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 5.208277012661378e-07, |
|
"loss": 0.0, |
|
"step": 746 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 5.08156512932696e-07, |
|
"loss": 0.0, |
|
"step": 747 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 4.956373479920407e-07, |
|
"loss": 0.0, |
|
"step": 748 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 4.83270406953793e-07, |
|
"loss": 0.0, |
|
"step": 749 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 4.7105588788952107e-07, |
|
"loss": 0.0, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 4.5899398642957225e-07, |
|
"loss": 0.0, |
|
"step": 751 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 4.4708489575993496e-07, |
|
"loss": 0.0, |
|
"step": 752 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 4.353288066191508e-07, |
|
"loss": 0.0, |
|
"step": 753 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 4.2372590729525374e-07, |
|
"loss": 0.0, |
|
"step": 754 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 4.1227638362275836e-07, |
|
"loss": 0.0, |
|
"step": 755 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 4.0098041897968065e-07, |
|
"loss": 0.0, |
|
"step": 756 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.8983819428460414e-07, |
|
"loss": 0.0, |
|
"step": 757 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.7884988799377965e-07, |
|
"loss": 0.0, |
|
"step": 758 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.680156760982667e-07, |
|
"loss": 0.0, |
|
"step": 759 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.5733573212111907e-07, |
|
"loss": 0.0, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.468102271146001e-07, |
|
"loss": 0.0, |
|
"step": 761 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.364393296574453e-07, |
|
"loss": 0.0, |
|
"step": 762 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.2622320585216417e-07, |
|
"loss": 0.0, |
|
"step": 763 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.161620193223769e-07, |
|
"loss": 0.0, |
|
"step": 764 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 3.062559312101987e-07, |
|
"loss": 0.0, |
|
"step": 765 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 2.9650510017364965e-07, |
|
"loss": 0.0, |
|
"step": 766 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 2.8690968238412444e-07, |
|
"loss": 0.0, |
|
"step": 767 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 2.774698315238833e-07, |
|
"loss": 0.0, |
|
"step": 768 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 2.68185698783594e-07, |
|
"loss": 0.0, |
|
"step": 769 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 2.590574328599094e-07, |
|
"loss": 0.0, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 2.500851799530879e-07, |
|
"loss": 0.0, |
|
"step": 771 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 2.41269083764647e-07, |
|
"loss": 0.0, |
|
"step": 772 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 2.3260928549506568e-07, |
|
"loss": 0.0, |
|
"step": 773 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 2.241059238415255e-07, |
|
"loss": 0.0, |
|
"step": 774 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 2.1575913499568003e-07, |
|
"loss": 0.0, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 2.075690526414842e-07, |
|
"loss": 0.0, |
|
"step": 776 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 1.9953580795304628e-07, |
|
"loss": 0.0, |
|
"step": 777 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 1.9165952959252943e-07, |
|
"loss": 0.0, |
|
"step": 778 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 1.8394034370808888e-07, |
|
"loss": 0.0, |
|
"step": 779 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 1.7637837393185474e-07, |
|
"loss": 0.0, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 1.6897374137795351e-07, |
|
"loss": 0.0, |
|
"step": 781 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 1.6172656464055748e-07, |
|
"loss": 0.0, |
|
"step": 782 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 1.5463695979199833e-07, |
|
"loss": 0.0, |
|
"step": 783 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 1.4770504038090328e-07, |
|
"loss": 0.0, |
|
"step": 784 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 1.4093091743037302e-07, |
|
"loss": 0.0, |
|
"step": 785 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 1.3431469943620877e-07, |
|
"loss": 0.0, |
|
"step": 786 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 1.278564923651704e-07, |
|
"loss": 0.0, |
|
"step": 787 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 1.2155639965328202e-07, |
|
"loss": 0.0, |
|
"step": 788 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 1.1541452220417359e-07, |
|
"loss": 0.0, |
|
"step": 789 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 1.0943095838746753e-07, |
|
"loss": 0.0, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 1.03605804037199e-07, |
|
"loss": 0.0, |
|
"step": 791 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 9.793915245028595e-08, |
|
"loss": 0.0, |
|
"step": 792 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 9.243109438502817e-08, |
|
"loss": 0.0, |
|
"step": 793 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 8.70817180596617e-08, |
|
"loss": 0.0, |
|
"step": 794 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 8.189110915094e-08, |
|
"loss": 0.0, |
|
"step": 795 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 7.685935079276397e-08, |
|
"loss": 0.0, |
|
"step": 796 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 7.19865235748507e-08, |
|
"loss": 0.0, |
|
"step": 797 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 6.727270554144127e-08, |
|
"loss": 0.0, |
|
"step": 798 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 6.271797219005393e-08, |
|
"loss": 0.0, |
|
"step": 799 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 5.832239647026838e-08, |
|
"loss": 0.0, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 5.408604878256563e-08, |
|
"loss": 0.0, |
|
"step": 801 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 5.000899697719552e-08, |
|
"loss": 0.0, |
|
"step": 802 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 4.6091306353088785e-08, |
|
"loss": 0.0, |
|
"step": 803 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 4.2333039656817784e-08, |
|
"loss": 0.0, |
|
"step": 804 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 3.873425708158296e-08, |
|
"loss": 0.0, |
|
"step": 805 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 3.529501626625465e-08, |
|
"loss": 0.0, |
|
"step": 806 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 3.201537229445051e-08, |
|
"loss": 0.0, |
|
"step": 807 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 2.8895377693646255e-08, |
|
"loss": 0.0, |
|
"step": 808 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 2.5935082434344063e-08, |
|
"loss": 0.0, |
|
"step": 809 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 2.3134533929263235e-08, |
|
"loss": 0.0, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 2.049377703258748e-08, |
|
"loss": 0.0, |
|
"step": 811 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.8012854039244354e-08, |
|
"loss": 0.0, |
|
"step": 812 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.569180468422582e-08, |
|
"loss": 0.0, |
|
"step": 813 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.3530666141953198e-08, |
|
"loss": 0.0, |
|
"step": 814 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 1.1529473025684301e-08, |
|
"loss": 0.0, |
|
"step": 815 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 9.688257386955003e-09, |
|
"loss": 0.0, |
|
"step": 816 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 8.007048715068522e-09, |
|
"loss": 0.0, |
|
"step": 817 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 6.4858739366224775e-09, |
|
"loss": 0.0, |
|
"step": 818 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 5.1247574150747885e-09, |
|
"loss": 0.0, |
|
"step": 819 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 3.923720950361754e-09, |
|
"loss": 0.0, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 2.8827837785372346e-09, |
|
"loss": 0.0, |
|
"step": 821 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 2.0019625714740032e-09, |
|
"loss": 0.0, |
|
"step": 822 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 1.281271436593956e-09, |
|
"loss": 0.0, |
|
"step": 823 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 7.207219166383006e-10, |
|
"loss": 0.0, |
|
"step": 824 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 3.2032298948880824e-10, |
|
"loss": 0.0, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.008106802126492e-11, |
|
"loss": 0.0, |
|
"step": 826 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 0.0, |
|
"loss": 0.0, |
|
"step": 827 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"step": 827, |
|
"total_flos": 234215613644800.0, |
|
"train_loss": 0.0, |
|
"train_runtime": 2509.8536, |
|
"train_samples_per_second": 5.268, |
|
"train_steps_per_second": 0.33 |
|
} |
|
], |
|
"logging_steps": 1.0, |
|
"max_steps": 827, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 250, |
|
"total_flos": 234215613644800.0, |
|
"train_batch_size": 4, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|