{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9999474154703686, "eval_steps": 500, "global_step": 4754, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.3986013986013987e-07, "loss": 3.8525, "step": 1 }, { "epoch": 0.0, "learning_rate": 2.7972027972027973e-07, "loss": 3.7035, "step": 2 }, { "epoch": 0.0, "learning_rate": 4.195804195804196e-07, "loss": 3.7521, "step": 3 }, { "epoch": 0.0, "learning_rate": 5.594405594405595e-07, "loss": 3.7164, "step": 4 }, { "epoch": 0.0, "learning_rate": 6.993006993006994e-07, "loss": 3.8053, "step": 5 }, { "epoch": 0.0, "learning_rate": 8.391608391608393e-07, "loss": 3.5792, "step": 6 }, { "epoch": 0.0, "learning_rate": 9.790209790209791e-07, "loss": 3.6344, "step": 7 }, { "epoch": 0.0, "learning_rate": 1.118881118881119e-06, "loss": 3.5863, "step": 8 }, { "epoch": 0.0, "learning_rate": 1.258741258741259e-06, "loss": 3.0069, "step": 9 }, { "epoch": 0.0, "learning_rate": 1.3986013986013987e-06, "loss": 3.1007, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.5384615384615387e-06, "loss": 2.8284, "step": 11 }, { "epoch": 0.0, "learning_rate": 1.6783216783216785e-06, "loss": 2.2989, "step": 12 }, { "epoch": 0.0, "learning_rate": 1.8181818181818183e-06, "loss": 2.0953, "step": 13 }, { "epoch": 0.0, "learning_rate": 1.9580419580419583e-06, "loss": 1.9323, "step": 14 }, { "epoch": 0.0, "learning_rate": 2.0979020979020983e-06, "loss": 1.8144, "step": 15 }, { "epoch": 0.0, "learning_rate": 2.237762237762238e-06, "loss": 1.7808, "step": 16 }, { "epoch": 0.0, "learning_rate": 2.377622377622378e-06, "loss": 1.5106, "step": 17 }, { "epoch": 0.0, "learning_rate": 2.517482517482518e-06, "loss": 1.4651, "step": 18 }, { "epoch": 0.0, "learning_rate": 2.6573426573426574e-06, "loss": 1.4566, "step": 19 }, { "epoch": 0.0, "learning_rate": 2.7972027972027974e-06, "loss": 1.4512, "step": 20 }, { "epoch": 0.0, "learning_rate": 2.937062937062937e-06, "loss": 1.4492, "step": 21 }, { "epoch": 0.0, "learning_rate": 3.0769230769230774e-06, "loss": 1.4418, "step": 22 }, { "epoch": 0.0, "learning_rate": 3.216783216783217e-06, "loss": 1.4453, "step": 23 }, { "epoch": 0.01, "learning_rate": 3.356643356643357e-06, "loss": 1.3653, "step": 24 }, { "epoch": 0.01, "learning_rate": 3.4965034965034966e-06, "loss": 1.3794, "step": 25 }, { "epoch": 0.01, "learning_rate": 3.6363636363636366e-06, "loss": 1.4183, "step": 26 }, { "epoch": 0.01, "learning_rate": 3.776223776223776e-06, "loss": 1.3817, "step": 27 }, { "epoch": 0.01, "learning_rate": 3.916083916083917e-06, "loss": 1.3966, "step": 28 }, { "epoch": 0.01, "learning_rate": 4.055944055944056e-06, "loss": 1.4609, "step": 29 }, { "epoch": 0.01, "learning_rate": 4.195804195804197e-06, "loss": 1.4318, "step": 30 }, { "epoch": 0.01, "learning_rate": 4.335664335664336e-06, "loss": 1.4155, "step": 31 }, { "epoch": 0.01, "learning_rate": 4.475524475524476e-06, "loss": 1.3867, "step": 32 }, { "epoch": 0.01, "learning_rate": 4.615384615384616e-06, "loss": 1.3732, "step": 33 }, { "epoch": 0.01, "learning_rate": 4.755244755244756e-06, "loss": 1.435, "step": 34 }, { "epoch": 0.01, "learning_rate": 4.895104895104895e-06, "loss": 1.4521, "step": 35 }, { "epoch": 0.01, "learning_rate": 5.034965034965036e-06, "loss": 1.3773, "step": 36 }, { "epoch": 0.01, "learning_rate": 5.174825174825175e-06, "loss": 1.4319, "step": 37 }, { "epoch": 0.01, "learning_rate": 5.314685314685315e-06, "loss": 1.3781, "step": 38 }, { "epoch": 0.01, "learning_rate": 5.4545454545454545e-06, "loss": 1.351, "step": 39 }, { "epoch": 0.01, "learning_rate": 5.594405594405595e-06, "loss": 1.358, "step": 40 }, { "epoch": 0.01, "learning_rate": 5.7342657342657345e-06, "loss": 1.3821, "step": 41 }, { "epoch": 0.01, "learning_rate": 5.874125874125874e-06, "loss": 1.3329, "step": 42 }, { "epoch": 0.01, "learning_rate": 6.013986013986014e-06, "loss": 1.3145, "step": 43 }, { "epoch": 0.01, "learning_rate": 6.153846153846155e-06, "loss": 1.3183, "step": 44 }, { "epoch": 0.01, "learning_rate": 6.2937062937062944e-06, "loss": 1.3965, "step": 45 }, { "epoch": 0.01, "learning_rate": 6.433566433566434e-06, "loss": 1.372, "step": 46 }, { "epoch": 0.01, "learning_rate": 6.573426573426574e-06, "loss": 1.3856, "step": 47 }, { "epoch": 0.01, "learning_rate": 6.713286713286714e-06, "loss": 1.359, "step": 48 }, { "epoch": 0.01, "learning_rate": 6.853146853146854e-06, "loss": 1.3734, "step": 49 }, { "epoch": 0.01, "learning_rate": 6.993006993006993e-06, "loss": 1.3323, "step": 50 }, { "epoch": 0.01, "learning_rate": 7.132867132867134e-06, "loss": 1.4248, "step": 51 }, { "epoch": 0.01, "learning_rate": 7.272727272727273e-06, "loss": 1.3429, "step": 52 }, { "epoch": 0.01, "learning_rate": 7.412587412587413e-06, "loss": 1.3851, "step": 53 }, { "epoch": 0.01, "learning_rate": 7.552447552447552e-06, "loss": 1.3935, "step": 54 }, { "epoch": 0.01, "learning_rate": 7.692307692307694e-06, "loss": 1.396, "step": 55 }, { "epoch": 0.01, "learning_rate": 7.832167832167833e-06, "loss": 1.4116, "step": 56 }, { "epoch": 0.01, "learning_rate": 7.972027972027973e-06, "loss": 1.4268, "step": 57 }, { "epoch": 0.01, "learning_rate": 8.111888111888112e-06, "loss": 1.3697, "step": 58 }, { "epoch": 0.01, "learning_rate": 8.251748251748254e-06, "loss": 1.3953, "step": 59 }, { "epoch": 0.01, "learning_rate": 8.391608391608393e-06, "loss": 1.3102, "step": 60 }, { "epoch": 0.01, "learning_rate": 8.531468531468533e-06, "loss": 1.3684, "step": 61 }, { "epoch": 0.01, "learning_rate": 8.671328671328672e-06, "loss": 1.4022, "step": 62 }, { "epoch": 0.01, "learning_rate": 8.811188811188812e-06, "loss": 1.3667, "step": 63 }, { "epoch": 0.01, "learning_rate": 8.951048951048951e-06, "loss": 1.3922, "step": 64 }, { "epoch": 0.01, "learning_rate": 9.090909090909091e-06, "loss": 1.3328, "step": 65 }, { "epoch": 0.01, "learning_rate": 9.230769230769232e-06, "loss": 1.341, "step": 66 }, { "epoch": 0.01, "learning_rate": 9.370629370629372e-06, "loss": 1.3963, "step": 67 }, { "epoch": 0.01, "learning_rate": 9.510489510489511e-06, "loss": 1.3379, "step": 68 }, { "epoch": 0.01, "learning_rate": 9.650349650349651e-06, "loss": 1.3986, "step": 69 }, { "epoch": 0.01, "learning_rate": 9.79020979020979e-06, "loss": 1.313, "step": 70 }, { "epoch": 0.01, "learning_rate": 9.93006993006993e-06, "loss": 1.3771, "step": 71 }, { "epoch": 0.02, "learning_rate": 1.0069930069930071e-05, "loss": 1.3712, "step": 72 }, { "epoch": 0.02, "learning_rate": 1.0209790209790211e-05, "loss": 1.3607, "step": 73 }, { "epoch": 0.02, "learning_rate": 1.034965034965035e-05, "loss": 1.3464, "step": 74 }, { "epoch": 0.02, "learning_rate": 1.048951048951049e-05, "loss": 1.3674, "step": 75 }, { "epoch": 0.02, "learning_rate": 1.062937062937063e-05, "loss": 1.3891, "step": 76 }, { "epoch": 0.02, "learning_rate": 1.076923076923077e-05, "loss": 1.3269, "step": 77 }, { "epoch": 0.02, "learning_rate": 1.0909090909090909e-05, "loss": 1.3792, "step": 78 }, { "epoch": 0.02, "learning_rate": 1.1048951048951048e-05, "loss": 1.3708, "step": 79 }, { "epoch": 0.02, "learning_rate": 1.118881118881119e-05, "loss": 1.3629, "step": 80 }, { "epoch": 0.02, "learning_rate": 1.132867132867133e-05, "loss": 1.3122, "step": 81 }, { "epoch": 0.02, "learning_rate": 1.1468531468531469e-05, "loss": 1.3584, "step": 82 }, { "epoch": 0.02, "learning_rate": 1.1608391608391608e-05, "loss": 1.3342, "step": 83 }, { "epoch": 0.02, "learning_rate": 1.1748251748251748e-05, "loss": 1.4108, "step": 84 }, { "epoch": 0.02, "learning_rate": 1.1888111888111888e-05, "loss": 1.3698, "step": 85 }, { "epoch": 0.02, "learning_rate": 1.2027972027972027e-05, "loss": 1.4113, "step": 86 }, { "epoch": 0.02, "learning_rate": 1.216783216783217e-05, "loss": 1.3494, "step": 87 }, { "epoch": 0.02, "learning_rate": 1.230769230769231e-05, "loss": 1.2987, "step": 88 }, { "epoch": 0.02, "learning_rate": 1.244755244755245e-05, "loss": 1.3608, "step": 89 }, { "epoch": 0.02, "learning_rate": 1.2587412587412589e-05, "loss": 1.3636, "step": 90 }, { "epoch": 0.02, "learning_rate": 1.2727272727272728e-05, "loss": 1.3423, "step": 91 }, { "epoch": 0.02, "learning_rate": 1.2867132867132868e-05, "loss": 1.4264, "step": 92 }, { "epoch": 0.02, "learning_rate": 1.3006993006993008e-05, "loss": 1.4093, "step": 93 }, { "epoch": 0.02, "learning_rate": 1.3146853146853147e-05, "loss": 1.4413, "step": 94 }, { "epoch": 0.02, "learning_rate": 1.3286713286713288e-05, "loss": 1.2839, "step": 95 }, { "epoch": 0.02, "learning_rate": 1.3426573426573428e-05, "loss": 1.3507, "step": 96 }, { "epoch": 0.02, "learning_rate": 1.3566433566433568e-05, "loss": 1.3502, "step": 97 }, { "epoch": 0.02, "learning_rate": 1.3706293706293707e-05, "loss": 1.3816, "step": 98 }, { "epoch": 0.02, "learning_rate": 1.3846153846153847e-05, "loss": 1.35, "step": 99 }, { "epoch": 0.02, "learning_rate": 1.3986013986013986e-05, "loss": 1.3335, "step": 100 }, { "epoch": 0.02, "learning_rate": 1.4125874125874126e-05, "loss": 1.2728, "step": 101 }, { "epoch": 0.02, "learning_rate": 1.4265734265734267e-05, "loss": 1.3518, "step": 102 }, { "epoch": 0.02, "learning_rate": 1.4405594405594407e-05, "loss": 1.345, "step": 103 }, { "epoch": 0.02, "learning_rate": 1.4545454545454546e-05, "loss": 1.3138, "step": 104 }, { "epoch": 0.02, "learning_rate": 1.4685314685314686e-05, "loss": 1.3511, "step": 105 }, { "epoch": 0.02, "learning_rate": 1.4825174825174825e-05, "loss": 1.3698, "step": 106 }, { "epoch": 0.02, "learning_rate": 1.4965034965034965e-05, "loss": 1.3671, "step": 107 }, { "epoch": 0.02, "learning_rate": 1.5104895104895105e-05, "loss": 1.3366, "step": 108 }, { "epoch": 0.02, "learning_rate": 1.5244755244755244e-05, "loss": 1.3672, "step": 109 }, { "epoch": 0.02, "learning_rate": 1.5384615384615387e-05, "loss": 1.3607, "step": 110 }, { "epoch": 0.02, "learning_rate": 1.5524475524475525e-05, "loss": 1.3592, "step": 111 }, { "epoch": 0.02, "learning_rate": 1.5664335664335666e-05, "loss": 1.3234, "step": 112 }, { "epoch": 0.02, "learning_rate": 1.5804195804195804e-05, "loss": 1.3095, "step": 113 }, { "epoch": 0.02, "learning_rate": 1.5944055944055945e-05, "loss": 1.2844, "step": 114 }, { "epoch": 0.02, "learning_rate": 1.6083916083916083e-05, "loss": 1.4102, "step": 115 }, { "epoch": 0.02, "learning_rate": 1.6223776223776225e-05, "loss": 1.3583, "step": 116 }, { "epoch": 0.02, "learning_rate": 1.6363636363636366e-05, "loss": 1.3179, "step": 117 }, { "epoch": 0.02, "learning_rate": 1.6503496503496507e-05, "loss": 1.3393, "step": 118 }, { "epoch": 0.03, "learning_rate": 1.6643356643356645e-05, "loss": 1.3167, "step": 119 }, { "epoch": 0.03, "learning_rate": 1.6783216783216786e-05, "loss": 1.3557, "step": 120 }, { "epoch": 0.03, "learning_rate": 1.6923076923076924e-05, "loss": 1.3226, "step": 121 }, { "epoch": 0.03, "learning_rate": 1.7062937062937065e-05, "loss": 1.3573, "step": 122 }, { "epoch": 0.03, "learning_rate": 1.7202797202797203e-05, "loss": 1.3605, "step": 123 }, { "epoch": 0.03, "learning_rate": 1.7342657342657345e-05, "loss": 1.3461, "step": 124 }, { "epoch": 0.03, "learning_rate": 1.7482517482517486e-05, "loss": 1.3552, "step": 125 }, { "epoch": 0.03, "learning_rate": 1.7622377622377624e-05, "loss": 1.3978, "step": 126 }, { "epoch": 0.03, "learning_rate": 1.7762237762237765e-05, "loss": 1.3653, "step": 127 }, { "epoch": 0.03, "learning_rate": 1.7902097902097903e-05, "loss": 1.3946, "step": 128 }, { "epoch": 0.03, "learning_rate": 1.8041958041958044e-05, "loss": 1.4004, "step": 129 }, { "epoch": 0.03, "learning_rate": 1.8181818181818182e-05, "loss": 1.3499, "step": 130 }, { "epoch": 0.03, "learning_rate": 1.8321678321678323e-05, "loss": 1.3896, "step": 131 }, { "epoch": 0.03, "learning_rate": 1.8461538461538465e-05, "loss": 1.3121, "step": 132 }, { "epoch": 0.03, "learning_rate": 1.8601398601398602e-05, "loss": 1.308, "step": 133 }, { "epoch": 0.03, "learning_rate": 1.8741258741258744e-05, "loss": 1.3528, "step": 134 }, { "epoch": 0.03, "learning_rate": 1.888111888111888e-05, "loss": 1.2631, "step": 135 }, { "epoch": 0.03, "learning_rate": 1.9020979020979023e-05, "loss": 1.3493, "step": 136 }, { "epoch": 0.03, "learning_rate": 1.916083916083916e-05, "loss": 1.3801, "step": 137 }, { "epoch": 0.03, "learning_rate": 1.9300699300699302e-05, "loss": 1.3408, "step": 138 }, { "epoch": 0.03, "learning_rate": 1.944055944055944e-05, "loss": 1.3936, "step": 139 }, { "epoch": 0.03, "learning_rate": 1.958041958041958e-05, "loss": 1.3448, "step": 140 }, { "epoch": 0.03, "learning_rate": 1.9720279720279722e-05, "loss": 1.3019, "step": 141 }, { "epoch": 0.03, "learning_rate": 1.986013986013986e-05, "loss": 1.3405, "step": 142 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 1.3478, "step": 143 }, { "epoch": 0.03, "learning_rate": 1.9999997678976766e-05, "loss": 1.2915, "step": 144 }, { "epoch": 0.03, "learning_rate": 1.9999990715908128e-05, "loss": 1.3496, "step": 145 }, { "epoch": 0.03, "learning_rate": 1.999997911079733e-05, "loss": 1.2838, "step": 146 }, { "epoch": 0.03, "learning_rate": 1.9999962863649753e-05, "loss": 1.284, "step": 147 }, { "epoch": 0.03, "learning_rate": 1.999994197447294e-05, "loss": 1.3394, "step": 148 }, { "epoch": 0.03, "learning_rate": 1.999991644327659e-05, "loss": 1.305, "step": 149 }, { "epoch": 0.03, "learning_rate": 1.9999886270072552e-05, "loss": 1.3646, "step": 150 }, { "epoch": 0.03, "learning_rate": 1.9999851454874833e-05, "loss": 1.3571, "step": 151 }, { "epoch": 0.03, "learning_rate": 1.9999811997699595e-05, "loss": 1.3163, "step": 152 }, { "epoch": 0.03, "learning_rate": 1.9999767898565152e-05, "loss": 1.359, "step": 153 }, { "epoch": 0.03, "learning_rate": 1.9999719157491976e-05, "loss": 1.384, "step": 154 }, { "epoch": 0.03, "learning_rate": 1.9999665774502697e-05, "loss": 1.3121, "step": 155 }, { "epoch": 0.03, "learning_rate": 1.999960774962209e-05, "loss": 1.3159, "step": 156 }, { "epoch": 0.03, "learning_rate": 1.9999545082877095e-05, "loss": 1.4312, "step": 157 }, { "epoch": 0.03, "learning_rate": 1.99994777742968e-05, "loss": 1.3397, "step": 158 }, { "epoch": 0.03, "learning_rate": 1.999940582391245e-05, "loss": 1.3809, "step": 159 }, { "epoch": 0.03, "learning_rate": 1.9999329231757446e-05, "loss": 1.3526, "step": 160 }, { "epoch": 0.03, "learning_rate": 1.9999247997867338e-05, "loss": 1.3493, "step": 161 }, { "epoch": 0.03, "learning_rate": 1.999916212227984e-05, "loss": 1.3042, "step": 162 }, { "epoch": 0.03, "learning_rate": 1.9999071605034815e-05, "loss": 1.3473, "step": 163 }, { "epoch": 0.03, "learning_rate": 1.999897644617428e-05, "loss": 1.3374, "step": 164 }, { "epoch": 0.03, "learning_rate": 1.9998876645742408e-05, "loss": 1.346, "step": 165 }, { "epoch": 0.03, "learning_rate": 1.9998772203785525e-05, "loss": 1.2989, "step": 166 }, { "epoch": 0.04, "learning_rate": 1.999866312035212e-05, "loss": 1.3735, "step": 167 }, { "epoch": 0.04, "learning_rate": 1.9998549395492825e-05, "loss": 1.3525, "step": 168 }, { "epoch": 0.04, "learning_rate": 1.9998431029260434e-05, "loss": 1.2808, "step": 169 }, { "epoch": 0.04, "learning_rate": 1.999830802170989e-05, "loss": 1.3568, "step": 170 }, { "epoch": 0.04, "learning_rate": 1.99981803728983e-05, "loss": 1.3004, "step": 171 }, { "epoch": 0.04, "learning_rate": 1.999804808288491e-05, "loss": 1.3454, "step": 172 }, { "epoch": 0.04, "learning_rate": 1.9997911151731137e-05, "loss": 1.3602, "step": 173 }, { "epoch": 0.04, "learning_rate": 1.9997769579500537e-05, "loss": 1.3316, "step": 174 }, { "epoch": 0.04, "learning_rate": 1.999762336625884e-05, "loss": 1.4087, "step": 175 }, { "epoch": 0.04, "learning_rate": 1.9997472512073914e-05, "loss": 1.3493, "step": 176 }, { "epoch": 0.04, "learning_rate": 1.999731701701578e-05, "loss": 1.3445, "step": 177 }, { "epoch": 0.04, "learning_rate": 1.999715688115663e-05, "loss": 1.3481, "step": 178 }, { "epoch": 0.04, "learning_rate": 1.9996992104570788e-05, "loss": 1.4139, "step": 179 }, { "epoch": 0.04, "learning_rate": 1.9996822687334756e-05, "loss": 1.3435, "step": 180 }, { "epoch": 0.04, "learning_rate": 1.999664862952717e-05, "loss": 1.3267, "step": 181 }, { "epoch": 0.04, "learning_rate": 1.999646993122883e-05, "loss": 1.355, "step": 182 }, { "epoch": 0.04, "learning_rate": 1.9996286592522693e-05, "loss": 1.3183, "step": 183 }, { "epoch": 0.04, "learning_rate": 1.9996098613493862e-05, "loss": 1.388, "step": 184 }, { "epoch": 0.04, "learning_rate": 1.9995905994229594e-05, "loss": 1.3626, "step": 185 }, { "epoch": 0.04, "learning_rate": 1.999570873481931e-05, "loss": 1.3194, "step": 186 }, { "epoch": 0.04, "learning_rate": 1.999550683535458e-05, "loss": 1.3339, "step": 187 }, { "epoch": 0.04, "learning_rate": 1.9995300295929122e-05, "loss": 1.3063, "step": 188 }, { "epoch": 0.04, "learning_rate": 1.999508911663881e-05, "loss": 1.3289, "step": 189 }, { "epoch": 0.04, "learning_rate": 1.9994873297581682e-05, "loss": 1.4022, "step": 190 }, { "epoch": 0.04, "learning_rate": 1.9994652838857916e-05, "loss": 1.3245, "step": 191 }, { "epoch": 0.04, "learning_rate": 1.9994427740569854e-05, "loss": 1.3836, "step": 192 }, { "epoch": 0.04, "learning_rate": 1.999419800282199e-05, "loss": 1.2859, "step": 193 }, { "epoch": 0.04, "learning_rate": 1.999396362572096e-05, "loss": 1.3297, "step": 194 }, { "epoch": 0.04, "learning_rate": 1.9993724609375567e-05, "loss": 1.2884, "step": 195 }, { "epoch": 0.04, "learning_rate": 1.999348095389677e-05, "loss": 1.327, "step": 196 }, { "epoch": 0.04, "learning_rate": 1.9993232659397668e-05, "loss": 1.3711, "step": 197 }, { "epoch": 0.04, "learning_rate": 1.9992979725993524e-05, "loss": 1.3377, "step": 198 }, { "epoch": 0.04, "learning_rate": 1.9992722153801748e-05, "loss": 1.3643, "step": 199 }, { "epoch": 0.04, "learning_rate": 1.9992459942941905e-05, "loss": 1.3585, "step": 200 }, { "epoch": 0.04, "learning_rate": 1.999219309353572e-05, "loss": 1.3406, "step": 201 }, { "epoch": 0.04, "learning_rate": 1.9991921605707062e-05, "loss": 1.3742, "step": 202 }, { "epoch": 0.04, "learning_rate": 1.9991645479581957e-05, "loss": 1.3387, "step": 203 }, { "epoch": 0.04, "learning_rate": 1.9991364715288585e-05, "loss": 1.2922, "step": 204 }, { "epoch": 0.04, "learning_rate": 1.9991079312957276e-05, "loss": 1.3244, "step": 205 }, { "epoch": 0.04, "learning_rate": 1.999078927272052e-05, "loss": 1.3224, "step": 206 }, { "epoch": 0.04, "learning_rate": 1.999049459471295e-05, "loss": 1.3496, "step": 207 }, { "epoch": 0.04, "learning_rate": 1.999019527907136e-05, "loss": 1.3685, "step": 208 }, { "epoch": 0.04, "learning_rate": 1.9989891325934692e-05, "loss": 1.2866, "step": 209 }, { "epoch": 0.04, "learning_rate": 1.998958273544404e-05, "loss": 1.3414, "step": 210 }, { "epoch": 0.04, "learning_rate": 1.998926950774266e-05, "loss": 1.2783, "step": 211 }, { "epoch": 0.04, "learning_rate": 1.9988951642975948e-05, "loss": 1.3789, "step": 212 }, { "epoch": 0.04, "learning_rate": 1.9988629141291458e-05, "loss": 1.3242, "step": 213 }, { "epoch": 0.05, "learning_rate": 1.99883020028389e-05, "loss": 1.3194, "step": 214 }, { "epoch": 0.05, "learning_rate": 1.9987970227770135e-05, "loss": 1.3217, "step": 215 }, { "epoch": 0.05, "learning_rate": 1.9987633816239168e-05, "loss": 1.3302, "step": 216 }, { "epoch": 0.05, "learning_rate": 1.998729276840217e-05, "loss": 1.3165, "step": 217 }, { "epoch": 0.05, "learning_rate": 1.998694708441745e-05, "loss": 1.2886, "step": 218 }, { "epoch": 0.05, "learning_rate": 1.998659676444548e-05, "loss": 1.3289, "step": 219 }, { "epoch": 0.05, "learning_rate": 1.9986241808648884e-05, "loss": 1.2753, "step": 220 }, { "epoch": 0.05, "learning_rate": 1.9985882217192426e-05, "loss": 1.326, "step": 221 }, { "epoch": 0.05, "learning_rate": 1.9985517990243034e-05, "loss": 1.3225, "step": 222 }, { "epoch": 0.05, "learning_rate": 1.9985149127969782e-05, "loss": 1.3237, "step": 223 }, { "epoch": 0.05, "learning_rate": 1.9984775630543903e-05, "loss": 1.3681, "step": 224 }, { "epoch": 0.05, "learning_rate": 1.998439749813877e-05, "loss": 1.2982, "step": 225 }, { "epoch": 0.05, "learning_rate": 1.998401473092992e-05, "loss": 1.2878, "step": 226 }, { "epoch": 0.05, "learning_rate": 1.9983627329095028e-05, "loss": 1.3632, "step": 227 }, { "epoch": 0.05, "learning_rate": 1.998323529281393e-05, "loss": 1.3479, "step": 228 }, { "epoch": 0.05, "learning_rate": 1.9982838622268613e-05, "loss": 1.261, "step": 229 }, { "epoch": 0.05, "learning_rate": 1.9982437317643218e-05, "loss": 1.3203, "step": 230 }, { "epoch": 0.05, "learning_rate": 1.998203137912402e-05, "loss": 1.2993, "step": 231 }, { "epoch": 0.05, "learning_rate": 1.998162080689947e-05, "loss": 1.2846, "step": 232 }, { "epoch": 0.05, "learning_rate": 1.9981205601160148e-05, "loss": 1.3254, "step": 233 }, { "epoch": 0.05, "learning_rate": 1.99807857620988e-05, "loss": 1.3122, "step": 234 }, { "epoch": 0.05, "learning_rate": 1.998036128991032e-05, "loss": 1.3427, "step": 235 }, { "epoch": 0.05, "learning_rate": 1.9979932184791743e-05, "loss": 1.3794, "step": 236 }, { "epoch": 0.05, "learning_rate": 1.9979498446942263e-05, "loss": 1.3265, "step": 237 }, { "epoch": 0.05, "learning_rate": 1.9979060076563227e-05, "loss": 1.2561, "step": 238 }, { "epoch": 0.05, "learning_rate": 1.9978617073858124e-05, "loss": 1.2815, "step": 239 }, { "epoch": 0.05, "learning_rate": 1.9978169439032603e-05, "loss": 1.3074, "step": 240 }, { "epoch": 0.05, "learning_rate": 1.997771717229445e-05, "loss": 1.3638, "step": 241 }, { "epoch": 0.05, "learning_rate": 1.997726027385362e-05, "loss": 1.3235, "step": 242 }, { "epoch": 0.05, "learning_rate": 1.99767987439222e-05, "loss": 1.2649, "step": 243 }, { "epoch": 0.05, "learning_rate": 1.9976332582714434e-05, "loss": 1.2708, "step": 244 }, { "epoch": 0.05, "learning_rate": 1.997586179044672e-05, "loss": 1.3234, "step": 245 }, { "epoch": 0.05, "learning_rate": 1.9975386367337607e-05, "loss": 1.3391, "step": 246 }, { "epoch": 0.05, "learning_rate": 1.9974906313607776e-05, "loss": 1.2833, "step": 247 }, { "epoch": 0.05, "learning_rate": 1.9974421629480077e-05, "loss": 1.3444, "step": 248 }, { "epoch": 0.05, "learning_rate": 1.9973932315179502e-05, "loss": 1.3217, "step": 249 }, { "epoch": 0.05, "learning_rate": 1.997343837093319e-05, "loss": 1.3193, "step": 250 }, { "epoch": 0.05, "learning_rate": 1.9972939796970437e-05, "loss": 1.3209, "step": 251 }, { "epoch": 0.05, "learning_rate": 1.997243659352268e-05, "loss": 1.3199, "step": 252 }, { "epoch": 0.05, "learning_rate": 1.9971928760823514e-05, "loss": 1.3347, "step": 253 }, { "epoch": 0.05, "learning_rate": 1.9971416299108672e-05, "loss": 1.3526, "step": 254 }, { "epoch": 0.05, "learning_rate": 1.9970899208616042e-05, "loss": 1.2891, "step": 255 }, { "epoch": 0.05, "learning_rate": 1.9970377489585658e-05, "loss": 1.3257, "step": 256 }, { "epoch": 0.05, "learning_rate": 1.9969851142259708e-05, "loss": 1.3344, "step": 257 }, { "epoch": 0.05, "learning_rate": 1.9969320166882523e-05, "loss": 1.3588, "step": 258 }, { "epoch": 0.05, "learning_rate": 1.9968784563700586e-05, "loss": 1.3215, "step": 259 }, { "epoch": 0.05, "learning_rate": 1.9968244332962522e-05, "loss": 1.3242, "step": 260 }, { "epoch": 0.05, "learning_rate": 1.9967699474919114e-05, "loss": 1.3407, "step": 261 }, { "epoch": 0.06, "learning_rate": 1.9967149989823286e-05, "loss": 1.3236, "step": 262 }, { "epoch": 0.06, "learning_rate": 1.9966595877930106e-05, "loss": 1.3016, "step": 263 }, { "epoch": 0.06, "learning_rate": 1.9966037139496802e-05, "loss": 1.3702, "step": 264 }, { "epoch": 0.06, "learning_rate": 1.9965473774782744e-05, "loss": 1.3658, "step": 265 }, { "epoch": 0.06, "learning_rate": 1.9964905784049444e-05, "loss": 1.3318, "step": 266 }, { "epoch": 0.06, "learning_rate": 1.9964333167560567e-05, "loss": 1.3768, "step": 267 }, { "epoch": 0.06, "learning_rate": 1.9963755925581925e-05, "loss": 1.3423, "step": 268 }, { "epoch": 0.06, "learning_rate": 1.996317405838148e-05, "loss": 1.342, "step": 269 }, { "epoch": 0.06, "learning_rate": 1.9962587566229327e-05, "loss": 1.3406, "step": 270 }, { "epoch": 0.06, "learning_rate": 1.996199644939773e-05, "loss": 1.3573, "step": 271 }, { "epoch": 0.06, "learning_rate": 1.996140070816108e-05, "loss": 1.2904, "step": 272 }, { "epoch": 0.06, "learning_rate": 1.9960800342795928e-05, "loss": 1.33, "step": 273 }, { "epoch": 0.06, "learning_rate": 1.9960195353580963e-05, "loss": 1.367, "step": 274 }, { "epoch": 0.06, "learning_rate": 1.9959585740797028e-05, "loss": 1.3068, "step": 275 }, { "epoch": 0.06, "learning_rate": 1.9958971504727103e-05, "loss": 1.267, "step": 276 }, { "epoch": 0.06, "learning_rate": 1.9958352645656323e-05, "loss": 1.3478, "step": 277 }, { "epoch": 0.06, "learning_rate": 1.9957729163871964e-05, "loss": 1.3462, "step": 278 }, { "epoch": 0.06, "learning_rate": 1.9957101059663447e-05, "loss": 1.2869, "step": 279 }, { "epoch": 0.06, "learning_rate": 1.9956468333322346e-05, "loss": 1.2115, "step": 280 }, { "epoch": 0.06, "learning_rate": 1.995583098514237e-05, "loss": 1.3228, "step": 281 }, { "epoch": 0.06, "learning_rate": 1.9955189015419383e-05, "loss": 1.3299, "step": 282 }, { "epoch": 0.06, "learning_rate": 1.9954542424451386e-05, "loss": 1.2739, "step": 283 }, { "epoch": 0.06, "learning_rate": 1.9953891212538535e-05, "loss": 1.2893, "step": 284 }, { "epoch": 0.06, "learning_rate": 1.995323537998312e-05, "loss": 1.3063, "step": 285 }, { "epoch": 0.06, "learning_rate": 1.9952574927089587e-05, "loss": 1.2869, "step": 286 }, { "epoch": 0.06, "learning_rate": 1.9951909854164517e-05, "loss": 1.3313, "step": 287 }, { "epoch": 0.06, "learning_rate": 1.9951240161516643e-05, "loss": 1.274, "step": 288 }, { "epoch": 0.06, "learning_rate": 1.9950565849456836e-05, "loss": 1.3331, "step": 289 }, { "epoch": 0.06, "learning_rate": 1.994988691829812e-05, "loss": 1.3377, "step": 290 }, { "epoch": 0.06, "learning_rate": 1.9949203368355655e-05, "loss": 1.307, "step": 291 }, { "epoch": 0.06, "learning_rate": 1.9948515199946743e-05, "loss": 1.3097, "step": 292 }, { "epoch": 0.06, "learning_rate": 1.994782241339084e-05, "loss": 1.3398, "step": 293 }, { "epoch": 0.06, "learning_rate": 1.9947125009009546e-05, "loss": 1.3357, "step": 294 }, { "epoch": 0.06, "learning_rate": 1.9946422987126592e-05, "loss": 1.3061, "step": 295 }, { "epoch": 0.06, "learning_rate": 1.994571634806786e-05, "loss": 1.3184, "step": 296 }, { "epoch": 0.06, "learning_rate": 1.994500509216138e-05, "loss": 1.2707, "step": 297 }, { "epoch": 0.06, "learning_rate": 1.9944289219737316e-05, "loss": 1.3126, "step": 298 }, { "epoch": 0.06, "learning_rate": 1.994356873112798e-05, "loss": 1.339, "step": 299 }, { "epoch": 0.06, "learning_rate": 1.9942843626667825e-05, "loss": 1.3375, "step": 300 }, { "epoch": 0.06, "learning_rate": 1.994211390669345e-05, "loss": 1.3201, "step": 301 }, { "epoch": 0.06, "learning_rate": 1.9941379571543597e-05, "loss": 1.34, "step": 302 }, { "epoch": 0.06, "learning_rate": 1.9940640621559143e-05, "loss": 1.3298, "step": 303 }, { "epoch": 0.06, "learning_rate": 1.9939897057083116e-05, "loss": 1.32, "step": 304 }, { "epoch": 0.06, "learning_rate": 1.9939148878460677e-05, "loss": 1.3367, "step": 305 }, { "epoch": 0.06, "learning_rate": 1.9938396086039135e-05, "loss": 1.2482, "step": 306 }, { "epoch": 0.06, "learning_rate": 1.9937638680167945e-05, "loss": 1.2876, "step": 307 }, { "epoch": 0.06, "learning_rate": 1.9936876661198696e-05, "loss": 1.3151, "step": 308 }, { "epoch": 0.06, "learning_rate": 1.9936110029485113e-05, "loss": 1.3125, "step": 309 }, { "epoch": 0.07, "learning_rate": 1.9935338785383083e-05, "loss": 1.2912, "step": 310 }, { "epoch": 0.07, "learning_rate": 1.9934562929250614e-05, "loss": 1.3056, "step": 311 }, { "epoch": 0.07, "learning_rate": 1.9933782461447862e-05, "loss": 1.2883, "step": 312 }, { "epoch": 0.07, "learning_rate": 1.9932997382337125e-05, "loss": 1.2932, "step": 313 }, { "epoch": 0.07, "learning_rate": 1.9932207692282838e-05, "loss": 1.3265, "step": 314 }, { "epoch": 0.07, "learning_rate": 1.993141339165158e-05, "loss": 1.3159, "step": 315 }, { "epoch": 0.07, "learning_rate": 1.9930614480812073e-05, "loss": 1.2045, "step": 316 }, { "epoch": 0.07, "learning_rate": 1.992981096013517e-05, "loss": 1.3502, "step": 317 }, { "epoch": 0.07, "learning_rate": 1.9929002829993874e-05, "loss": 1.3013, "step": 318 }, { "epoch": 0.07, "learning_rate": 1.9928190090763317e-05, "loss": 1.3461, "step": 319 }, { "epoch": 0.07, "learning_rate": 1.9927372742820778e-05, "loss": 1.3649, "step": 320 }, { "epoch": 0.07, "learning_rate": 1.9926550786545674e-05, "loss": 1.3249, "step": 321 }, { "epoch": 0.07, "learning_rate": 1.9925724222319566e-05, "loss": 1.3058, "step": 322 }, { "epoch": 0.07, "learning_rate": 1.9924893050526143e-05, "loss": 1.3034, "step": 323 }, { "epoch": 0.07, "learning_rate": 1.992405727155124e-05, "loss": 1.3906, "step": 324 }, { "epoch": 0.07, "learning_rate": 1.9923216885782827e-05, "loss": 1.3401, "step": 325 }, { "epoch": 0.07, "learning_rate": 1.9922371893611023e-05, "loss": 1.2659, "step": 326 }, { "epoch": 0.07, "learning_rate": 1.992152229542807e-05, "loss": 1.3423, "step": 327 }, { "epoch": 0.07, "learning_rate": 1.9920668091628355e-05, "loss": 1.3295, "step": 328 }, { "epoch": 0.07, "learning_rate": 1.991980928260841e-05, "loss": 1.3849, "step": 329 }, { "epoch": 0.07, "learning_rate": 1.991894586876689e-05, "loss": 1.3088, "step": 330 }, { "epoch": 0.07, "learning_rate": 1.9918077850504605e-05, "loss": 1.3391, "step": 331 }, { "epoch": 0.07, "learning_rate": 1.9917205228224483e-05, "loss": 1.2776, "step": 332 }, { "epoch": 0.07, "learning_rate": 1.9916328002331605e-05, "loss": 1.342, "step": 333 }, { "epoch": 0.07, "learning_rate": 1.9915446173233184e-05, "loss": 1.3426, "step": 334 }, { "epoch": 0.07, "learning_rate": 1.9914559741338572e-05, "loss": 1.3447, "step": 335 }, { "epoch": 0.07, "learning_rate": 1.9913668707059243e-05, "loss": 1.2998, "step": 336 }, { "epoch": 0.07, "learning_rate": 1.991277307080883e-05, "loss": 1.3111, "step": 337 }, { "epoch": 0.07, "learning_rate": 1.9911872833003088e-05, "loss": 1.3328, "step": 338 }, { "epoch": 0.07, "learning_rate": 1.9910967994059916e-05, "loss": 1.309, "step": 339 }, { "epoch": 0.07, "learning_rate": 1.9910058554399333e-05, "loss": 1.2972, "step": 340 }, { "epoch": 0.07, "learning_rate": 1.990914451444352e-05, "loss": 1.3247, "step": 341 }, { "epoch": 0.07, "learning_rate": 1.9908225874616764e-05, "loss": 1.3312, "step": 342 }, { "epoch": 0.07, "learning_rate": 1.9907302635345513e-05, "loss": 1.3211, "step": 343 }, { "epoch": 0.07, "learning_rate": 1.9906374797058334e-05, "loss": 1.2922, "step": 344 }, { "epoch": 0.07, "learning_rate": 1.9905442360185937e-05, "loss": 1.3093, "step": 345 }, { "epoch": 0.07, "learning_rate": 1.990450532516116e-05, "loss": 1.3219, "step": 346 }, { "epoch": 0.07, "learning_rate": 1.9903563692418978e-05, "loss": 1.3053, "step": 347 }, { "epoch": 0.07, "learning_rate": 1.9902617462396504e-05, "loss": 1.3466, "step": 348 }, { "epoch": 0.07, "learning_rate": 1.990166663553298e-05, "loss": 1.2792, "step": 349 }, { "epoch": 0.07, "learning_rate": 1.990071121226979e-05, "loss": 1.2812, "step": 350 }, { "epoch": 0.07, "learning_rate": 1.989975119305044e-05, "loss": 1.2456, "step": 351 }, { "epoch": 0.07, "learning_rate": 1.9898786578320583e-05, "loss": 1.3227, "step": 352 }, { "epoch": 0.07, "learning_rate": 1.9897817368527987e-05, "loss": 1.3363, "step": 353 }, { "epoch": 0.07, "learning_rate": 1.989684356412257e-05, "loss": 1.3211, "step": 354 }, { "epoch": 0.07, "learning_rate": 1.9895865165556375e-05, "loss": 1.308, "step": 355 }, { "epoch": 0.07, "learning_rate": 1.9894882173283577e-05, "loss": 1.3196, "step": 356 }, { "epoch": 0.08, "learning_rate": 1.9893894587760495e-05, "loss": 1.3547, "step": 357 }, { "epoch": 0.08, "learning_rate": 1.9892902409445558e-05, "loss": 1.3052, "step": 358 }, { "epoch": 0.08, "learning_rate": 1.9891905638799345e-05, "loss": 1.3232, "step": 359 }, { "epoch": 0.08, "learning_rate": 1.9890904276284567e-05, "loss": 1.2851, "step": 360 }, { "epoch": 0.08, "learning_rate": 1.988989832236605e-05, "loss": 1.3677, "step": 361 }, { "epoch": 0.08, "learning_rate": 1.9888887777510776e-05, "loss": 1.2866, "step": 362 }, { "epoch": 0.08, "learning_rate": 1.9887872642187834e-05, "loss": 1.333, "step": 363 }, { "epoch": 0.08, "learning_rate": 1.988685291686846e-05, "loss": 1.32, "step": 364 }, { "epoch": 0.08, "learning_rate": 1.9885828602026013e-05, "loss": 1.3102, "step": 365 }, { "epoch": 0.08, "learning_rate": 1.988479969813598e-05, "loss": 1.3422, "step": 366 }, { "epoch": 0.08, "learning_rate": 1.9883766205675993e-05, "loss": 1.2943, "step": 367 }, { "epoch": 0.08, "learning_rate": 1.98827281251258e-05, "loss": 1.3602, "step": 368 }, { "epoch": 0.08, "learning_rate": 1.9881685456967277e-05, "loss": 1.2804, "step": 369 }, { "epoch": 0.08, "learning_rate": 1.9880638201684445e-05, "loss": 1.3013, "step": 370 }, { "epoch": 0.08, "learning_rate": 1.9879586359763438e-05, "loss": 1.2849, "step": 371 }, { "epoch": 0.08, "learning_rate": 1.9878529931692524e-05, "loss": 1.3706, "step": 372 }, { "epoch": 0.08, "learning_rate": 1.9877468917962107e-05, "loss": 1.3145, "step": 373 }, { "epoch": 0.08, "learning_rate": 1.9876403319064717e-05, "loss": 1.283, "step": 374 }, { "epoch": 0.08, "learning_rate": 1.9875333135495e-05, "loss": 1.3086, "step": 375 }, { "epoch": 0.08, "learning_rate": 1.9874258367749754e-05, "loss": 1.2711, "step": 376 }, { "epoch": 0.08, "learning_rate": 1.9873179016327882e-05, "loss": 1.3052, "step": 377 }, { "epoch": 0.08, "learning_rate": 1.987209508173042e-05, "loss": 1.3175, "step": 378 }, { "epoch": 0.08, "learning_rate": 1.9871006564460543e-05, "loss": 1.3261, "step": 379 }, { "epoch": 0.08, "learning_rate": 1.9869913465023548e-05, "loss": 1.3154, "step": 380 }, { "epoch": 0.08, "learning_rate": 1.9868815783926848e-05, "loss": 1.3206, "step": 381 }, { "epoch": 0.08, "learning_rate": 1.9867713521679998e-05, "loss": 1.3295, "step": 382 }, { "epoch": 0.08, "learning_rate": 1.986660667879467e-05, "loss": 1.2834, "step": 383 }, { "epoch": 0.08, "learning_rate": 1.986549525578467e-05, "loss": 1.3527, "step": 384 }, { "epoch": 0.08, "learning_rate": 1.986437925316592e-05, "loss": 1.3236, "step": 385 }, { "epoch": 0.08, "learning_rate": 1.9863258671456477e-05, "loss": 1.3187, "step": 386 }, { "epoch": 0.08, "learning_rate": 1.986213351117652e-05, "loss": 1.3208, "step": 387 }, { "epoch": 0.08, "learning_rate": 1.9861003772848355e-05, "loss": 1.3081, "step": 388 }, { "epoch": 0.08, "learning_rate": 1.9859869456996408e-05, "loss": 1.3025, "step": 389 }, { "epoch": 0.08, "learning_rate": 1.9858730564147235e-05, "loss": 1.285, "step": 390 }, { "epoch": 0.08, "learning_rate": 1.9857587094829518e-05, "loss": 1.2828, "step": 391 }, { "epoch": 0.08, "learning_rate": 1.9856439049574055e-05, "loss": 1.3134, "step": 392 }, { "epoch": 0.08, "learning_rate": 1.9855286428913786e-05, "loss": 1.3255, "step": 393 }, { "epoch": 0.08, "learning_rate": 1.9854129233383746e-05, "loss": 1.3362, "step": 394 }, { "epoch": 0.08, "learning_rate": 1.9852967463521126e-05, "loss": 1.2389, "step": 395 }, { "epoch": 0.08, "learning_rate": 1.9851801119865213e-05, "loss": 1.312, "step": 396 }, { "epoch": 0.08, "learning_rate": 1.9850630202957437e-05, "loss": 1.3307, "step": 397 }, { "epoch": 0.08, "learning_rate": 1.9849454713341337e-05, "loss": 1.2981, "step": 398 }, { "epoch": 0.08, "learning_rate": 1.9848274651562588e-05, "loss": 1.3015, "step": 399 }, { "epoch": 0.08, "learning_rate": 1.9847090018168976e-05, "loss": 1.2513, "step": 400 }, { "epoch": 0.08, "learning_rate": 1.984590081371041e-05, "loss": 1.3381, "step": 401 }, { "epoch": 0.08, "learning_rate": 1.9844707038738932e-05, "loss": 1.3364, "step": 402 }, { "epoch": 0.08, "learning_rate": 1.9843508693808694e-05, "loss": 1.2444, "step": 403 }, { "epoch": 0.08, "learning_rate": 1.984230577947597e-05, "loss": 1.321, "step": 404 }, { "epoch": 0.09, "learning_rate": 1.9841098296299162e-05, "loss": 1.2737, "step": 405 }, { "epoch": 0.09, "learning_rate": 1.983988624483879e-05, "loss": 1.3197, "step": 406 }, { "epoch": 0.09, "learning_rate": 1.983866962565749e-05, "loss": 1.3085, "step": 407 }, { "epoch": 0.09, "learning_rate": 1.9837448439320027e-05, "loss": 1.3423, "step": 408 }, { "epoch": 0.09, "learning_rate": 1.9836222686393283e-05, "loss": 1.2371, "step": 409 }, { "epoch": 0.09, "learning_rate": 1.983499236744625e-05, "loss": 1.3478, "step": 410 }, { "epoch": 0.09, "learning_rate": 1.9833757483050054e-05, "loss": 1.3268, "step": 411 }, { "epoch": 0.09, "learning_rate": 1.983251803377793e-05, "loss": 1.2929, "step": 412 }, { "epoch": 0.09, "learning_rate": 1.9831274020205242e-05, "loss": 1.2685, "step": 413 }, { "epoch": 0.09, "learning_rate": 1.9830025442909462e-05, "loss": 1.2955, "step": 414 }, { "epoch": 0.09, "learning_rate": 1.9828772302470186e-05, "loss": 1.2071, "step": 415 }, { "epoch": 0.09, "learning_rate": 1.9827514599469128e-05, "loss": 1.3418, "step": 416 }, { "epoch": 0.09, "learning_rate": 1.982625233449012e-05, "loss": 1.297, "step": 417 }, { "epoch": 0.09, "learning_rate": 1.9824985508119113e-05, "loss": 1.2968, "step": 418 }, { "epoch": 0.09, "learning_rate": 1.982371412094417e-05, "loss": 1.2161, "step": 419 }, { "epoch": 0.09, "learning_rate": 1.9822438173555478e-05, "loss": 1.3061, "step": 420 }, { "epoch": 0.09, "learning_rate": 1.9821157666545336e-05, "loss": 1.3138, "step": 421 }, { "epoch": 0.09, "learning_rate": 1.981987260050816e-05, "loss": 1.2783, "step": 422 }, { "epoch": 0.09, "learning_rate": 1.981858297604049e-05, "loss": 1.2794, "step": 423 }, { "epoch": 0.09, "learning_rate": 1.9817288793740966e-05, "loss": 1.337, "step": 424 }, { "epoch": 0.09, "learning_rate": 1.9815990054210363e-05, "loss": 1.2994, "step": 425 }, { "epoch": 0.09, "learning_rate": 1.9814686758051557e-05, "loss": 1.2692, "step": 426 }, { "epoch": 0.09, "learning_rate": 1.9813378905869542e-05, "loss": 1.2931, "step": 427 }, { "epoch": 0.09, "learning_rate": 1.981206649827143e-05, "loss": 1.3197, "step": 428 }, { "epoch": 0.09, "learning_rate": 1.9810749535866453e-05, "loss": 1.2885, "step": 429 }, { "epoch": 0.09, "learning_rate": 1.9809428019265944e-05, "loss": 1.2745, "step": 430 }, { "epoch": 0.09, "learning_rate": 1.9808101949083362e-05, "loss": 1.2845, "step": 431 }, { "epoch": 0.09, "learning_rate": 1.980677132593427e-05, "loss": 1.2979, "step": 432 }, { "epoch": 0.09, "learning_rate": 1.9805436150436352e-05, "loss": 1.3273, "step": 433 }, { "epoch": 0.09, "learning_rate": 1.98040964232094e-05, "loss": 1.3509, "step": 434 }, { "epoch": 0.09, "learning_rate": 1.980275214487533e-05, "loss": 1.2849, "step": 435 }, { "epoch": 0.09, "learning_rate": 1.9801403316058152e-05, "loss": 1.3064, "step": 436 }, { "epoch": 0.09, "learning_rate": 1.9800049937384004e-05, "loss": 1.3297, "step": 437 }, { "epoch": 0.09, "learning_rate": 1.9798692009481133e-05, "loss": 1.2855, "step": 438 }, { "epoch": 0.09, "learning_rate": 1.9797329532979887e-05, "loss": 1.284, "step": 439 }, { "epoch": 0.09, "learning_rate": 1.979596250851274e-05, "loss": 1.3387, "step": 440 }, { "epoch": 0.09, "learning_rate": 1.9794590936714276e-05, "loss": 1.2886, "step": 441 }, { "epoch": 0.09, "learning_rate": 1.9793214818221173e-05, "loss": 1.266, "step": 442 }, { "epoch": 0.09, "learning_rate": 1.979183415367224e-05, "loss": 1.3179, "step": 443 }, { "epoch": 0.09, "learning_rate": 1.9790448943708386e-05, "loss": 1.3515, "step": 444 }, { "epoch": 0.09, "learning_rate": 1.978905918897263e-05, "loss": 1.3105, "step": 445 }, { "epoch": 0.09, "learning_rate": 1.9787664890110107e-05, "loss": 1.2619, "step": 446 }, { "epoch": 0.09, "learning_rate": 1.9786266047768055e-05, "loss": 1.2984, "step": 447 }, { "epoch": 0.09, "learning_rate": 1.978486266259582e-05, "loss": 1.2376, "step": 448 }, { "epoch": 0.09, "learning_rate": 1.978345473524486e-05, "loss": 1.35, "step": 449 }, { "epoch": 0.09, "learning_rate": 1.9782042266368748e-05, "loss": 1.3298, "step": 450 }, { "epoch": 0.09, "learning_rate": 1.9780625256623153e-05, "loss": 1.2869, "step": 451 }, { "epoch": 0.1, "learning_rate": 1.9779203706665856e-05, "loss": 1.3279, "step": 452 }, { "epoch": 0.1, "learning_rate": 1.9777777617156753e-05, "loss": 1.3167, "step": 453 }, { "epoch": 0.1, "learning_rate": 1.9776346988757837e-05, "loss": 1.3233, "step": 454 }, { "epoch": 0.1, "learning_rate": 1.9774911822133217e-05, "loss": 1.3526, "step": 455 }, { "epoch": 0.1, "learning_rate": 1.9773472117949097e-05, "loss": 1.2661, "step": 456 }, { "epoch": 0.1, "learning_rate": 1.9772027876873796e-05, "loss": 1.2986, "step": 457 }, { "epoch": 0.1, "learning_rate": 1.9770579099577745e-05, "loss": 1.3201, "step": 458 }, { "epoch": 0.1, "learning_rate": 1.9769125786733463e-05, "loss": 1.2798, "step": 459 }, { "epoch": 0.1, "learning_rate": 1.9767667939015593e-05, "loss": 1.2539, "step": 460 }, { "epoch": 0.1, "learning_rate": 1.976620555710087e-05, "loss": 1.308, "step": 461 }, { "epoch": 0.1, "learning_rate": 1.9764738641668137e-05, "loss": 1.309, "step": 462 }, { "epoch": 0.1, "learning_rate": 1.9763267193398348e-05, "loss": 1.3192, "step": 463 }, { "epoch": 0.1, "learning_rate": 1.976179121297455e-05, "loss": 1.3207, "step": 464 }, { "epoch": 0.1, "learning_rate": 1.9760310701081905e-05, "loss": 1.3079, "step": 465 }, { "epoch": 0.1, "learning_rate": 1.9758825658407675e-05, "loss": 1.2781, "step": 466 }, { "epoch": 0.1, "learning_rate": 1.9757336085641218e-05, "loss": 1.3299, "step": 467 }, { "epoch": 0.1, "learning_rate": 1.9755841983474004e-05, "loss": 1.2614, "step": 468 }, { "epoch": 0.1, "learning_rate": 1.9754343352599598e-05, "loss": 1.34, "step": 469 }, { "epoch": 0.1, "learning_rate": 1.975284019371368e-05, "loss": 1.3174, "step": 470 }, { "epoch": 0.1, "learning_rate": 1.9751332507514016e-05, "loss": 1.3196, "step": 471 }, { "epoch": 0.1, "learning_rate": 1.9749820294700482e-05, "loss": 1.3337, "step": 472 }, { "epoch": 0.1, "learning_rate": 1.9748303555975056e-05, "loss": 1.2804, "step": 473 }, { "epoch": 0.1, "learning_rate": 1.9746782292041815e-05, "loss": 1.2702, "step": 474 }, { "epoch": 0.1, "learning_rate": 1.9745256503606937e-05, "loss": 1.3474, "step": 475 }, { "epoch": 0.1, "learning_rate": 1.97437261913787e-05, "loss": 1.2665, "step": 476 }, { "epoch": 0.1, "learning_rate": 1.9742191356067478e-05, "loss": 1.305, "step": 477 }, { "epoch": 0.1, "learning_rate": 1.9740651998385753e-05, "loss": 1.3163, "step": 478 }, { "epoch": 0.1, "learning_rate": 1.9739108119048103e-05, "loss": 1.2825, "step": 479 }, { "epoch": 0.1, "learning_rate": 1.97375597187712e-05, "loss": 1.237, "step": 480 }, { "epoch": 0.1, "learning_rate": 1.9736006798273823e-05, "loss": 1.3776, "step": 481 }, { "epoch": 0.1, "learning_rate": 1.973444935827684e-05, "loss": 1.3068, "step": 482 }, { "epoch": 0.1, "learning_rate": 1.9732887399503225e-05, "loss": 1.3523, "step": 483 }, { "epoch": 0.1, "learning_rate": 1.9731320922678048e-05, "loss": 1.295, "step": 484 }, { "epoch": 0.1, "learning_rate": 1.972974992852847e-05, "loss": 1.2571, "step": 485 }, { "epoch": 0.1, "learning_rate": 1.972817441778376e-05, "loss": 1.3029, "step": 486 }, { "epoch": 0.1, "learning_rate": 1.972659439117527e-05, "loss": 1.234, "step": 487 }, { "epoch": 0.1, "learning_rate": 1.9725009849436464e-05, "loss": 1.3067, "step": 488 }, { "epoch": 0.1, "learning_rate": 1.9723420793302886e-05, "loss": 1.3091, "step": 489 }, { "epoch": 0.1, "learning_rate": 1.9721827223512186e-05, "loss": 1.294, "step": 490 }, { "epoch": 0.1, "learning_rate": 1.972022914080411e-05, "loss": 1.2734, "step": 491 }, { "epoch": 0.1, "learning_rate": 1.971862654592049e-05, "loss": 1.3115, "step": 492 }, { "epoch": 0.1, "learning_rate": 1.971701943960526e-05, "loss": 1.3097, "step": 493 }, { "epoch": 0.1, "learning_rate": 1.971540782260445e-05, "loss": 1.3009, "step": 494 }, { "epoch": 0.1, "learning_rate": 1.9713791695666177e-05, "loss": 1.3039, "step": 495 }, { "epoch": 0.1, "learning_rate": 1.971217105954065e-05, "loss": 1.2825, "step": 496 }, { "epoch": 0.1, "learning_rate": 1.9710545914980182e-05, "loss": 1.302, "step": 497 }, { "epoch": 0.1, "learning_rate": 1.9708916262739172e-05, "loss": 1.3138, "step": 498 }, { "epoch": 0.1, "learning_rate": 1.970728210357411e-05, "loss": 1.301, "step": 499 }, { "epoch": 0.11, "learning_rate": 1.9705643438243586e-05, "loss": 1.2974, "step": 500 }, { "epoch": 0.11, "learning_rate": 1.9704000267508265e-05, "loss": 1.2781, "step": 501 }, { "epoch": 0.11, "learning_rate": 1.9702352592130923e-05, "loss": 1.311, "step": 502 }, { "epoch": 0.11, "learning_rate": 1.9700700412876418e-05, "loss": 1.2774, "step": 503 }, { "epoch": 0.11, "learning_rate": 1.9699043730511697e-05, "loss": 1.3088, "step": 504 }, { "epoch": 0.11, "learning_rate": 1.9697382545805798e-05, "loss": 1.2834, "step": 505 }, { "epoch": 0.11, "learning_rate": 1.9695716859529853e-05, "loss": 1.2788, "step": 506 }, { "epoch": 0.11, "learning_rate": 1.9694046672457086e-05, "loss": 1.3138, "step": 507 }, { "epoch": 0.11, "learning_rate": 1.9692371985362798e-05, "loss": 1.334, "step": 508 }, { "epoch": 0.11, "learning_rate": 1.969069279902439e-05, "loss": 1.2995, "step": 509 }, { "epoch": 0.11, "learning_rate": 1.9689009114221348e-05, "loss": 1.2996, "step": 510 }, { "epoch": 0.11, "learning_rate": 1.9687320931735244e-05, "loss": 1.2419, "step": 511 }, { "epoch": 0.11, "learning_rate": 1.9685628252349743e-05, "loss": 1.2951, "step": 512 }, { "epoch": 0.11, "learning_rate": 1.9683931076850596e-05, "loss": 1.3245, "step": 513 }, { "epoch": 0.11, "learning_rate": 1.9682229406025635e-05, "loss": 1.2737, "step": 514 }, { "epoch": 0.11, "learning_rate": 1.9680523240664786e-05, "loss": 1.2618, "step": 515 }, { "epoch": 0.11, "learning_rate": 1.967881258156006e-05, "loss": 1.3108, "step": 516 }, { "epoch": 0.11, "learning_rate": 1.9677097429505552e-05, "loss": 1.3077, "step": 517 }, { "epoch": 0.11, "learning_rate": 1.967537778529744e-05, "loss": 1.2561, "step": 518 }, { "epoch": 0.11, "learning_rate": 1.9673653649733996e-05, "loss": 1.3132, "step": 519 }, { "epoch": 0.11, "learning_rate": 1.9671925023615572e-05, "loss": 1.2549, "step": 520 }, { "epoch": 0.11, "learning_rate": 1.9670191907744602e-05, "loss": 1.2605, "step": 521 }, { "epoch": 0.11, "learning_rate": 1.9668454302925607e-05, "loss": 1.3585, "step": 522 }, { "epoch": 0.11, "learning_rate": 1.9666712209965185e-05, "loss": 1.2986, "step": 523 }, { "epoch": 0.11, "learning_rate": 1.9664965629672035e-05, "loss": 1.2551, "step": 524 }, { "epoch": 0.11, "learning_rate": 1.9663214562856924e-05, "loss": 1.2982, "step": 525 }, { "epoch": 0.11, "learning_rate": 1.9661459010332697e-05, "loss": 1.2668, "step": 526 }, { "epoch": 0.11, "learning_rate": 1.96596989729143e-05, "loss": 1.3058, "step": 527 }, { "epoch": 0.11, "learning_rate": 1.965793445141875e-05, "loss": 1.2901, "step": 528 }, { "epoch": 0.11, "learning_rate": 1.9656165446665136e-05, "loss": 1.2553, "step": 529 }, { "epoch": 0.11, "learning_rate": 1.9654391959474648e-05, "loss": 1.3094, "step": 530 }, { "epoch": 0.11, "learning_rate": 1.9652613990670542e-05, "loss": 1.2947, "step": 531 }, { "epoch": 0.11, "learning_rate": 1.9650831541078165e-05, "loss": 1.274, "step": 532 }, { "epoch": 0.11, "learning_rate": 1.9649044611524935e-05, "loss": 1.2787, "step": 533 }, { "epoch": 0.11, "learning_rate": 1.964725320284035e-05, "loss": 1.2869, "step": 534 }, { "epoch": 0.11, "learning_rate": 1.9645457315855993e-05, "loss": 1.2733, "step": 535 }, { "epoch": 0.11, "learning_rate": 1.9643656951405525e-05, "loss": 1.2771, "step": 536 }, { "epoch": 0.11, "learning_rate": 1.964185211032468e-05, "loss": 1.3024, "step": 537 }, { "epoch": 0.11, "learning_rate": 1.9640042793451276e-05, "loss": 1.2906, "step": 538 }, { "epoch": 0.11, "learning_rate": 1.9638229001625206e-05, "loss": 1.2775, "step": 539 }, { "epoch": 0.11, "learning_rate": 1.963641073568844e-05, "loss": 1.3075, "step": 540 }, { "epoch": 0.11, "learning_rate": 1.9634587996485027e-05, "loss": 1.2856, "step": 541 }, { "epoch": 0.11, "learning_rate": 1.9632760784861088e-05, "loss": 1.3041, "step": 542 }, { "epoch": 0.11, "learning_rate": 1.9630929101664826e-05, "loss": 1.3299, "step": 543 }, { "epoch": 0.11, "learning_rate": 1.9629092947746513e-05, "loss": 1.3534, "step": 544 }, { "epoch": 0.11, "learning_rate": 1.9627252323958504e-05, "loss": 1.3178, "step": 545 }, { "epoch": 0.11, "learning_rate": 1.9625407231155226e-05, "loss": 1.2976, "step": 546 }, { "epoch": 0.12, "learning_rate": 1.9623557670193175e-05, "loss": 1.3166, "step": 547 }, { "epoch": 0.12, "learning_rate": 1.9621703641930933e-05, "loss": 1.2534, "step": 548 }, { "epoch": 0.12, "learning_rate": 1.961984514722914e-05, "loss": 1.3031, "step": 549 }, { "epoch": 0.12, "learning_rate": 1.9617982186950524e-05, "loss": 1.3145, "step": 550 }, { "epoch": 0.12, "learning_rate": 1.9616114761959874e-05, "loss": 1.2246, "step": 551 }, { "epoch": 0.12, "learning_rate": 1.9614242873124067e-05, "loss": 1.3072, "step": 552 }, { "epoch": 0.12, "learning_rate": 1.961236652131203e-05, "loss": 1.2773, "step": 553 }, { "epoch": 0.12, "learning_rate": 1.9610485707394786e-05, "loss": 1.2831, "step": 554 }, { "epoch": 0.12, "learning_rate": 1.960860043224541e-05, "loss": 1.314, "step": 555 }, { "epoch": 0.12, "learning_rate": 1.9606710696739054e-05, "loss": 1.306, "step": 556 }, { "epoch": 0.12, "learning_rate": 1.960481650175295e-05, "loss": 1.2692, "step": 557 }, { "epoch": 0.12, "learning_rate": 1.9602917848166387e-05, "loss": 1.2983, "step": 558 }, { "epoch": 0.12, "learning_rate": 1.960101473686073e-05, "loss": 1.2895, "step": 559 }, { "epoch": 0.12, "learning_rate": 1.959910716871941e-05, "loss": 1.2794, "step": 560 }, { "epoch": 0.12, "learning_rate": 1.9597195144627934e-05, "loss": 1.2838, "step": 561 }, { "epoch": 0.12, "learning_rate": 1.959527866547387e-05, "loss": 1.3024, "step": 562 }, { "epoch": 0.12, "learning_rate": 1.959335773214685e-05, "loss": 1.2778, "step": 563 }, { "epoch": 0.12, "learning_rate": 1.959143234553859e-05, "loss": 1.3096, "step": 564 }, { "epoch": 0.12, "learning_rate": 1.958950250654286e-05, "loss": 1.2558, "step": 565 }, { "epoch": 0.12, "learning_rate": 1.95875682160555e-05, "loss": 1.2829, "step": 566 }, { "epoch": 0.12, "learning_rate": 1.9585629474974413e-05, "loss": 1.264, "step": 567 }, { "epoch": 0.12, "learning_rate": 1.958368628419958e-05, "loss": 1.2592, "step": 568 }, { "epoch": 0.12, "learning_rate": 1.958173864463303e-05, "loss": 1.2928, "step": 569 }, { "epoch": 0.12, "learning_rate": 1.957978655717887e-05, "loss": 1.2376, "step": 570 }, { "epoch": 0.12, "learning_rate": 1.957783002274327e-05, "loss": 1.3366, "step": 571 }, { "epoch": 0.12, "learning_rate": 1.957586904223446e-05, "loss": 1.1983, "step": 572 }, { "epoch": 0.12, "learning_rate": 1.957390361656274e-05, "loss": 1.3088, "step": 573 }, { "epoch": 0.12, "learning_rate": 1.9571933746640466e-05, "loss": 1.312, "step": 574 }, { "epoch": 0.12, "learning_rate": 1.956995943338206e-05, "loss": 1.2762, "step": 575 }, { "epoch": 0.12, "learning_rate": 1.9567980677704012e-05, "loss": 1.2317, "step": 576 }, { "epoch": 0.12, "learning_rate": 1.956599748052486e-05, "loss": 1.298, "step": 577 }, { "epoch": 0.12, "learning_rate": 1.9564009842765225e-05, "loss": 1.2511, "step": 578 }, { "epoch": 0.12, "learning_rate": 1.9562017765347773e-05, "loss": 1.3263, "step": 579 }, { "epoch": 0.12, "learning_rate": 1.9560021249197235e-05, "loss": 1.2362, "step": 580 }, { "epoch": 0.12, "learning_rate": 1.95580202952404e-05, "loss": 1.3078, "step": 581 }, { "epoch": 0.12, "learning_rate": 1.955601490440613e-05, "loss": 1.3096, "step": 582 }, { "epoch": 0.12, "learning_rate": 1.955400507762532e-05, "loss": 1.2901, "step": 583 }, { "epoch": 0.12, "learning_rate": 1.955199081583096e-05, "loss": 1.2871, "step": 584 }, { "epoch": 0.12, "learning_rate": 1.954997211995807e-05, "loss": 1.3276, "step": 585 }, { "epoch": 0.12, "learning_rate": 1.9547948990943734e-05, "loss": 1.2617, "step": 586 }, { "epoch": 0.12, "learning_rate": 1.9545921429727105e-05, "loss": 1.3083, "step": 587 }, { "epoch": 0.12, "learning_rate": 1.9543889437249386e-05, "loss": 1.3143, "step": 588 }, { "epoch": 0.12, "learning_rate": 1.9541853014453833e-05, "loss": 1.2455, "step": 589 }, { "epoch": 0.12, "learning_rate": 1.9539812162285768e-05, "loss": 1.3116, "step": 590 }, { "epoch": 0.12, "learning_rate": 1.9537766881692558e-05, "loss": 1.3624, "step": 591 }, { "epoch": 0.12, "learning_rate": 1.9535717173623638e-05, "loss": 1.3078, "step": 592 }, { "epoch": 0.12, "learning_rate": 1.953366303903049e-05, "loss": 1.2753, "step": 593 }, { "epoch": 0.12, "learning_rate": 1.9531604478866647e-05, "loss": 1.2896, "step": 594 }, { "epoch": 0.13, "learning_rate": 1.9529541494087714e-05, "loss": 1.314, "step": 595 }, { "epoch": 0.13, "learning_rate": 1.9527474085651326e-05, "loss": 1.2796, "step": 596 }, { "epoch": 0.13, "learning_rate": 1.9525402254517192e-05, "loss": 1.2588, "step": 597 }, { "epoch": 0.13, "learning_rate": 1.9523326001647057e-05, "loss": 1.2923, "step": 598 }, { "epoch": 0.13, "learning_rate": 1.952124532800474e-05, "loss": 1.3179, "step": 599 }, { "epoch": 0.13, "learning_rate": 1.951916023455609e-05, "loss": 1.2909, "step": 600 }, { "epoch": 0.13, "learning_rate": 1.9517070722269017e-05, "loss": 1.303, "step": 601 }, { "epoch": 0.13, "learning_rate": 1.9514976792113485e-05, "loss": 1.2957, "step": 602 }, { "epoch": 0.13, "learning_rate": 1.9512878445061506e-05, "loss": 1.2979, "step": 603 }, { "epoch": 0.13, "learning_rate": 1.9510775682087143e-05, "loss": 1.2548, "step": 604 }, { "epoch": 0.13, "learning_rate": 1.9508668504166507e-05, "loss": 1.2447, "step": 605 }, { "epoch": 0.13, "learning_rate": 1.9506556912277757e-05, "loss": 1.2376, "step": 606 }, { "epoch": 0.13, "learning_rate": 1.9504440907401113e-05, "loss": 1.2693, "step": 607 }, { "epoch": 0.13, "learning_rate": 1.9502320490518825e-05, "loss": 1.3026, "step": 608 }, { "epoch": 0.13, "learning_rate": 1.9500195662615202e-05, "loss": 1.3426, "step": 609 }, { "epoch": 0.13, "learning_rate": 1.9498066424676606e-05, "loss": 1.2839, "step": 610 }, { "epoch": 0.13, "learning_rate": 1.949593277769143e-05, "loss": 1.308, "step": 611 }, { "epoch": 0.13, "learning_rate": 1.9493794722650126e-05, "loss": 1.2811, "step": 612 }, { "epoch": 0.13, "learning_rate": 1.949165226054519e-05, "loss": 1.254, "step": 613 }, { "epoch": 0.13, "learning_rate": 1.9489505392371164e-05, "loss": 1.2636, "step": 614 }, { "epoch": 0.13, "learning_rate": 1.9487354119124633e-05, "loss": 1.2875, "step": 615 }, { "epoch": 0.13, "learning_rate": 1.9485198441804226e-05, "loss": 1.2923, "step": 616 }, { "epoch": 0.13, "learning_rate": 1.9483038361410622e-05, "loss": 1.3487, "step": 617 }, { "epoch": 0.13, "learning_rate": 1.9480873878946538e-05, "loss": 1.3364, "step": 618 }, { "epoch": 0.13, "learning_rate": 1.9478704995416736e-05, "loss": 1.3026, "step": 619 }, { "epoch": 0.13, "learning_rate": 1.9476531711828027e-05, "loss": 1.257, "step": 620 }, { "epoch": 0.13, "learning_rate": 1.9474354029189252e-05, "loss": 1.3002, "step": 621 }, { "epoch": 0.13, "learning_rate": 1.9472171948511306e-05, "loss": 1.2597, "step": 622 }, { "epoch": 0.13, "learning_rate": 1.9469985470807123e-05, "loss": 1.2853, "step": 623 }, { "epoch": 0.13, "learning_rate": 1.946779459709167e-05, "loss": 1.2837, "step": 624 }, { "epoch": 0.13, "learning_rate": 1.9465599328381964e-05, "loss": 1.2949, "step": 625 }, { "epoch": 0.13, "learning_rate": 1.946339966569706e-05, "loss": 1.2833, "step": 626 }, { "epoch": 0.13, "learning_rate": 1.9461195610058047e-05, "loss": 1.2948, "step": 627 }, { "epoch": 0.13, "learning_rate": 1.9458987162488064e-05, "loss": 1.3069, "step": 628 }, { "epoch": 0.13, "learning_rate": 1.9456774324012276e-05, "loss": 1.2966, "step": 629 }, { "epoch": 0.13, "learning_rate": 1.94545570956579e-05, "loss": 1.3204, "step": 630 }, { "epoch": 0.13, "learning_rate": 1.9452335478454186e-05, "loss": 1.2931, "step": 631 }, { "epoch": 0.13, "learning_rate": 1.9450109473432406e-05, "loss": 1.2797, "step": 632 }, { "epoch": 0.13, "learning_rate": 1.9447879081625892e-05, "loss": 1.2857, "step": 633 }, { "epoch": 0.13, "learning_rate": 1.944564430407e-05, "loss": 1.3086, "step": 634 }, { "epoch": 0.13, "learning_rate": 1.9443405141802122e-05, "loss": 1.2983, "step": 635 }, { "epoch": 0.13, "learning_rate": 1.944116159586169e-05, "loss": 1.3067, "step": 636 }, { "epoch": 0.13, "learning_rate": 1.943891366729017e-05, "loss": 1.2656, "step": 637 }, { "epoch": 0.13, "learning_rate": 1.9436661357131056e-05, "loss": 1.3231, "step": 638 }, { "epoch": 0.13, "learning_rate": 1.9434404666429886e-05, "loss": 1.3053, "step": 639 }, { "epoch": 0.13, "learning_rate": 1.9432143596234224e-05, "loss": 1.2248, "step": 640 }, { "epoch": 0.13, "learning_rate": 1.942987814759367e-05, "loss": 1.2919, "step": 641 }, { "epoch": 0.14, "learning_rate": 1.9427608321559848e-05, "loss": 1.3308, "step": 642 }, { "epoch": 0.14, "learning_rate": 1.9425334119186434e-05, "loss": 1.2912, "step": 643 }, { "epoch": 0.14, "learning_rate": 1.942305554152912e-05, "loss": 1.2493, "step": 644 }, { "epoch": 0.14, "learning_rate": 1.9420772589645627e-05, "loss": 1.2806, "step": 645 }, { "epoch": 0.14, "learning_rate": 1.9418485264595714e-05, "loss": 1.2946, "step": 646 }, { "epoch": 0.14, "learning_rate": 1.9416193567441174e-05, "loss": 1.2538, "step": 647 }, { "epoch": 0.14, "learning_rate": 1.9413897499245816e-05, "loss": 1.2767, "step": 648 }, { "epoch": 0.14, "learning_rate": 1.9411597061075488e-05, "loss": 1.2931, "step": 649 }, { "epoch": 0.14, "learning_rate": 1.940929225399806e-05, "loss": 1.2861, "step": 650 }, { "epoch": 0.14, "learning_rate": 1.9406983079083447e-05, "loss": 1.2902, "step": 651 }, { "epoch": 0.14, "learning_rate": 1.9404669537403563e-05, "loss": 1.2664, "step": 652 }, { "epoch": 0.14, "learning_rate": 1.9402351630032378e-05, "loss": 1.2932, "step": 653 }, { "epoch": 0.14, "learning_rate": 1.9400029358045865e-05, "loss": 1.2985, "step": 654 }, { "epoch": 0.14, "learning_rate": 1.939770272252204e-05, "loss": 1.2981, "step": 655 }, { "epoch": 0.14, "learning_rate": 1.9395371724540936e-05, "loss": 1.2683, "step": 656 }, { "epoch": 0.14, "learning_rate": 1.939303636518461e-05, "loss": 1.3267, "step": 657 }, { "epoch": 0.14, "learning_rate": 1.9390696645537148e-05, "loss": 1.2989, "step": 658 }, { "epoch": 0.14, "learning_rate": 1.9388352566684667e-05, "loss": 1.2844, "step": 659 }, { "epoch": 0.14, "learning_rate": 1.9386004129715288e-05, "loss": 1.2751, "step": 660 }, { "epoch": 0.14, "learning_rate": 1.938365133571917e-05, "loss": 1.2999, "step": 661 }, { "epoch": 0.14, "learning_rate": 1.9381294185788495e-05, "loss": 1.2722, "step": 662 }, { "epoch": 0.14, "learning_rate": 1.9378932681017457e-05, "loss": 1.2778, "step": 663 }, { "epoch": 0.14, "learning_rate": 1.937656682250228e-05, "loss": 1.2701, "step": 664 }, { "epoch": 0.14, "learning_rate": 1.9374196611341212e-05, "loss": 1.2505, "step": 665 }, { "epoch": 0.14, "learning_rate": 1.9371822048634504e-05, "loss": 1.2715, "step": 666 }, { "epoch": 0.14, "learning_rate": 1.936944313548445e-05, "loss": 1.2996, "step": 667 }, { "epoch": 0.14, "learning_rate": 1.9367059872995348e-05, "loss": 1.2985, "step": 668 }, { "epoch": 0.14, "learning_rate": 1.936467226227352e-05, "loss": 1.3211, "step": 669 }, { "epoch": 0.14, "learning_rate": 1.9362280304427305e-05, "loss": 1.2572, "step": 670 }, { "epoch": 0.14, "learning_rate": 1.9359884000567062e-05, "loss": 1.3176, "step": 671 }, { "epoch": 0.14, "learning_rate": 1.9357483351805166e-05, "loss": 1.291, "step": 672 }, { "epoch": 0.14, "learning_rate": 1.935507835925601e-05, "loss": 1.2942, "step": 673 }, { "epoch": 0.14, "learning_rate": 1.9352669024036004e-05, "loss": 1.2979, "step": 674 }, { "epoch": 0.14, "learning_rate": 1.935025534726357e-05, "loss": 1.2556, "step": 675 }, { "epoch": 0.14, "learning_rate": 1.9347837330059147e-05, "loss": 1.271, "step": 676 }, { "epoch": 0.14, "learning_rate": 1.9345414973545193e-05, "loss": 1.323, "step": 677 }, { "epoch": 0.14, "learning_rate": 1.9342988278846175e-05, "loss": 1.3123, "step": 678 }, { "epoch": 0.14, "learning_rate": 1.934055724708858e-05, "loss": 1.2961, "step": 679 }, { "epoch": 0.14, "learning_rate": 1.9338121879400896e-05, "loss": 1.1945, "step": 680 }, { "epoch": 0.14, "learning_rate": 1.9335682176913642e-05, "loss": 1.3124, "step": 681 }, { "epoch": 0.14, "learning_rate": 1.9333238140759333e-05, "loss": 1.2727, "step": 682 }, { "epoch": 0.14, "learning_rate": 1.93307897720725e-05, "loss": 1.3096, "step": 683 }, { "epoch": 0.14, "learning_rate": 1.9328337071989693e-05, "loss": 1.1739, "step": 684 }, { "epoch": 0.14, "learning_rate": 1.9325880041649463e-05, "loss": 1.2653, "step": 685 }, { "epoch": 0.14, "learning_rate": 1.9323418682192375e-05, "loss": 1.3503, "step": 686 }, { "epoch": 0.14, "learning_rate": 1.9320952994761008e-05, "loss": 1.2661, "step": 687 }, { "epoch": 0.14, "learning_rate": 1.931848298049994e-05, "loss": 1.2901, "step": 688 }, { "epoch": 0.14, "learning_rate": 1.9316008640555766e-05, "loss": 1.326, "step": 689 }, { "epoch": 0.15, "learning_rate": 1.9313529976077085e-05, "loss": 1.3126, "step": 690 }, { "epoch": 0.15, "learning_rate": 1.9311046988214506e-05, "loss": 1.2737, "step": 691 }, { "epoch": 0.15, "learning_rate": 1.9308559678120643e-05, "loss": 1.263, "step": 692 }, { "epoch": 0.15, "learning_rate": 1.9306068046950112e-05, "loss": 1.2488, "step": 693 }, { "epoch": 0.15, "learning_rate": 1.9303572095859545e-05, "loss": 1.2719, "step": 694 }, { "epoch": 0.15, "learning_rate": 1.9301071826007575e-05, "loss": 1.2797, "step": 695 }, { "epoch": 0.15, "learning_rate": 1.9298567238554837e-05, "loss": 1.238, "step": 696 }, { "epoch": 0.15, "learning_rate": 1.9296058334663973e-05, "loss": 1.2764, "step": 697 }, { "epoch": 0.15, "learning_rate": 1.9293545115499624e-05, "loss": 1.2832, "step": 698 }, { "epoch": 0.15, "learning_rate": 1.9291027582228445e-05, "loss": 1.2794, "step": 699 }, { "epoch": 0.15, "learning_rate": 1.928850573601908e-05, "loss": 1.2927, "step": 700 }, { "epoch": 0.15, "learning_rate": 1.9285979578042186e-05, "loss": 1.2847, "step": 701 }, { "epoch": 0.15, "learning_rate": 1.9283449109470413e-05, "loss": 1.2869, "step": 702 }, { "epoch": 0.15, "learning_rate": 1.928091433147842e-05, "loss": 1.2855, "step": 703 }, { "epoch": 0.15, "learning_rate": 1.9278375245242862e-05, "loss": 1.2693, "step": 704 }, { "epoch": 0.15, "learning_rate": 1.9275831851942395e-05, "loss": 1.3073, "step": 705 }, { "epoch": 0.15, "learning_rate": 1.9273284152757673e-05, "loss": 1.2933, "step": 706 }, { "epoch": 0.15, "learning_rate": 1.927073214887135e-05, "loss": 1.3085, "step": 707 }, { "epoch": 0.15, "learning_rate": 1.9268175841468078e-05, "loss": 1.2935, "step": 708 }, { "epoch": 0.15, "learning_rate": 1.9265615231734508e-05, "loss": 1.2676, "step": 709 }, { "epoch": 0.15, "learning_rate": 1.9263050320859284e-05, "loss": 1.336, "step": 710 }, { "epoch": 0.15, "learning_rate": 1.926048111003305e-05, "loss": 1.2969, "step": 711 }, { "epoch": 0.15, "learning_rate": 1.9257907600448453e-05, "loss": 1.2652, "step": 712 }, { "epoch": 0.15, "learning_rate": 1.9255329793300116e-05, "loss": 1.2773, "step": 713 }, { "epoch": 0.15, "learning_rate": 1.9252747689784677e-05, "loss": 1.3134, "step": 714 }, { "epoch": 0.15, "learning_rate": 1.925016129110076e-05, "loss": 1.2843, "step": 715 }, { "epoch": 0.15, "learning_rate": 1.924757059844898e-05, "loss": 1.3113, "step": 716 }, { "epoch": 0.15, "learning_rate": 1.924497561303195e-05, "loss": 1.2259, "step": 717 }, { "epoch": 0.15, "learning_rate": 1.924237633605428e-05, "loss": 1.3069, "step": 718 }, { "epoch": 0.15, "learning_rate": 1.9239772768722558e-05, "loss": 1.2945, "step": 719 }, { "epoch": 0.15, "learning_rate": 1.9237164912245374e-05, "loss": 1.3127, "step": 720 }, { "epoch": 0.15, "learning_rate": 1.923455276783331e-05, "loss": 1.2505, "step": 721 }, { "epoch": 0.15, "learning_rate": 1.9231936336698934e-05, "loss": 1.3057, "step": 722 }, { "epoch": 0.15, "learning_rate": 1.9229315620056805e-05, "loss": 1.2989, "step": 723 }, { "epoch": 0.15, "learning_rate": 1.922669061912347e-05, "loss": 1.3121, "step": 724 }, { "epoch": 0.15, "learning_rate": 1.9224061335117474e-05, "loss": 1.3241, "step": 725 }, { "epoch": 0.15, "learning_rate": 1.9221427769259333e-05, "loss": 1.2605, "step": 726 }, { "epoch": 0.15, "learning_rate": 1.9218789922771567e-05, "loss": 1.2475, "step": 727 }, { "epoch": 0.15, "learning_rate": 1.9216147796878675e-05, "loss": 1.242, "step": 728 }, { "epoch": 0.15, "learning_rate": 1.9213501392807146e-05, "loss": 1.2228, "step": 729 }, { "epoch": 0.15, "learning_rate": 1.9210850711785447e-05, "loss": 1.2249, "step": 730 }, { "epoch": 0.15, "learning_rate": 1.9208195755044044e-05, "loss": 1.2255, "step": 731 }, { "epoch": 0.15, "learning_rate": 1.9205536523815372e-05, "loss": 1.3201, "step": 732 }, { "epoch": 0.15, "learning_rate": 1.920287301933387e-05, "loss": 1.2947, "step": 733 }, { "epoch": 0.15, "learning_rate": 1.920020524283594e-05, "loss": 1.2833, "step": 734 }, { "epoch": 0.15, "learning_rate": 1.919753319555998e-05, "loss": 1.2397, "step": 735 }, { "epoch": 0.15, "learning_rate": 1.919485687874636e-05, "loss": 1.2747, "step": 736 }, { "epoch": 0.16, "learning_rate": 1.919217629363745e-05, "loss": 1.2567, "step": 737 }, { "epoch": 0.16, "learning_rate": 1.918949144147758e-05, "loss": 1.2655, "step": 738 }, { "epoch": 0.16, "learning_rate": 1.918680232351308e-05, "loss": 1.309, "step": 739 }, { "epoch": 0.16, "learning_rate": 1.9184108940992242e-05, "loss": 1.25, "step": 740 }, { "epoch": 0.16, "learning_rate": 1.918141129516535e-05, "loss": 1.2891, "step": 741 }, { "epoch": 0.16, "learning_rate": 1.9178709387284668e-05, "loss": 1.2056, "step": 742 }, { "epoch": 0.16, "learning_rate": 1.917600321860443e-05, "loss": 1.2391, "step": 743 }, { "epoch": 0.16, "learning_rate": 1.9173292790380852e-05, "loss": 1.2829, "step": 744 }, { "epoch": 0.16, "learning_rate": 1.9170578103872127e-05, "loss": 1.2763, "step": 745 }, { "epoch": 0.16, "learning_rate": 1.9167859160338425e-05, "loss": 1.2955, "step": 746 }, { "epoch": 0.16, "learning_rate": 1.9165135961041898e-05, "loss": 1.233, "step": 747 }, { "epoch": 0.16, "learning_rate": 1.9162408507246663e-05, "loss": 1.3017, "step": 748 }, { "epoch": 0.16, "learning_rate": 1.9159676800218815e-05, "loss": 1.275, "step": 749 }, { "epoch": 0.16, "learning_rate": 1.9156940841226427e-05, "loss": 1.2425, "step": 750 }, { "epoch": 0.16, "learning_rate": 1.9154200631539546e-05, "loss": 1.2916, "step": 751 }, { "epoch": 0.16, "learning_rate": 1.9151456172430186e-05, "loss": 1.3006, "step": 752 }, { "epoch": 0.16, "learning_rate": 1.9148707465172342e-05, "loss": 1.283, "step": 753 }, { "epoch": 0.16, "learning_rate": 1.914595451104197e-05, "loss": 1.2908, "step": 754 }, { "epoch": 0.16, "learning_rate": 1.9143197311317014e-05, "loss": 1.2687, "step": 755 }, { "epoch": 0.16, "learning_rate": 1.9140435867277373e-05, "loss": 1.2836, "step": 756 }, { "epoch": 0.16, "learning_rate": 1.913767018020492e-05, "loss": 1.2725, "step": 757 }, { "epoch": 0.16, "learning_rate": 1.91349002513835e-05, "loss": 1.2742, "step": 758 }, { "epoch": 0.16, "learning_rate": 1.9132126082098933e-05, "loss": 1.2126, "step": 759 }, { "epoch": 0.16, "learning_rate": 1.9129347673638996e-05, "loss": 1.2747, "step": 760 }, { "epoch": 0.16, "learning_rate": 1.9126565027293443e-05, "loss": 1.2408, "step": 761 }, { "epoch": 0.16, "learning_rate": 1.9123778144353985e-05, "loss": 1.2914, "step": 762 }, { "epoch": 0.16, "learning_rate": 1.9120987026114314e-05, "loss": 1.3012, "step": 763 }, { "epoch": 0.16, "learning_rate": 1.9118191673870073e-05, "loss": 1.271, "step": 764 }, { "epoch": 0.16, "learning_rate": 1.9115392088918884e-05, "loss": 1.3113, "step": 765 }, { "epoch": 0.16, "learning_rate": 1.9112588272560323e-05, "loss": 1.2685, "step": 766 }, { "epoch": 0.16, "learning_rate": 1.910978022609593e-05, "loss": 1.2732, "step": 767 }, { "epoch": 0.16, "learning_rate": 1.9106967950829223e-05, "loss": 1.3063, "step": 768 }, { "epoch": 0.16, "learning_rate": 1.9104151448065672e-05, "loss": 1.2164, "step": 769 }, { "epoch": 0.16, "learning_rate": 1.9101330719112705e-05, "loss": 1.265, "step": 770 }, { "epoch": 0.16, "learning_rate": 1.909850576527972e-05, "loss": 1.2864, "step": 771 }, { "epoch": 0.16, "learning_rate": 1.9095676587878072e-05, "loss": 1.2488, "step": 772 }, { "epoch": 0.16, "learning_rate": 1.9092843188221084e-05, "loss": 1.2341, "step": 773 }, { "epoch": 0.16, "learning_rate": 1.9090005567624028e-05, "loss": 1.3139, "step": 774 }, { "epoch": 0.16, "learning_rate": 1.9087163727404137e-05, "loss": 1.2824, "step": 775 }, { "epoch": 0.16, "learning_rate": 1.908431766888062e-05, "loss": 1.2678, "step": 776 }, { "epoch": 0.16, "learning_rate": 1.9081467393374616e-05, "loss": 1.2986, "step": 777 }, { "epoch": 0.16, "learning_rate": 1.9078612902209246e-05, "loss": 1.2531, "step": 778 }, { "epoch": 0.16, "learning_rate": 1.9075754196709574e-05, "loss": 1.2978, "step": 779 }, { "epoch": 0.16, "learning_rate": 1.9072891278202622e-05, "loss": 1.3015, "step": 780 }, { "epoch": 0.16, "learning_rate": 1.9070024148017375e-05, "loss": 1.2811, "step": 781 }, { "epoch": 0.16, "learning_rate": 1.9067152807484763e-05, "loss": 1.2937, "step": 782 }, { "epoch": 0.16, "learning_rate": 1.9064277257937684e-05, "loss": 1.2894, "step": 783 }, { "epoch": 0.16, "learning_rate": 1.906139750071097e-05, "loss": 1.2562, "step": 784 }, { "epoch": 0.17, "learning_rate": 1.9058513537141427e-05, "loss": 1.2952, "step": 785 }, { "epoch": 0.17, "learning_rate": 1.90556253685678e-05, "loss": 1.2491, "step": 786 }, { "epoch": 0.17, "learning_rate": 1.905273299633079e-05, "loss": 1.2728, "step": 787 }, { "epoch": 0.17, "learning_rate": 1.9049836421773053e-05, "loss": 1.291, "step": 788 }, { "epoch": 0.17, "learning_rate": 1.904693564623919e-05, "loss": 1.3236, "step": 789 }, { "epoch": 0.17, "learning_rate": 1.9044030671075755e-05, "loss": 1.2832, "step": 790 }, { "epoch": 0.17, "learning_rate": 1.904112149763125e-05, "loss": 1.2689, "step": 791 }, { "epoch": 0.17, "learning_rate": 1.9038208127256128e-05, "loss": 1.2551, "step": 792 }, { "epoch": 0.17, "learning_rate": 1.903529056130279e-05, "loss": 1.2309, "step": 793 }, { "epoch": 0.17, "learning_rate": 1.9032368801125578e-05, "loss": 1.2944, "step": 794 }, { "epoch": 0.17, "learning_rate": 1.9029442848080797e-05, "loss": 1.3019, "step": 795 }, { "epoch": 0.17, "learning_rate": 1.9026512703526675e-05, "loss": 1.2714, "step": 796 }, { "epoch": 0.17, "learning_rate": 1.9023578368823412e-05, "loss": 1.2829, "step": 797 }, { "epoch": 0.17, "learning_rate": 1.9020639845333127e-05, "loss": 1.2133, "step": 798 }, { "epoch": 0.17, "learning_rate": 1.901769713441991e-05, "loss": 1.2532, "step": 799 }, { "epoch": 0.17, "learning_rate": 1.9014750237449772e-05, "loss": 1.2657, "step": 800 }, { "epoch": 0.17, "learning_rate": 1.9011799155790674e-05, "loss": 1.2872, "step": 801 }, { "epoch": 0.17, "learning_rate": 1.900884389081253e-05, "loss": 1.284, "step": 802 }, { "epoch": 0.17, "learning_rate": 1.9005884443887182e-05, "loss": 1.2763, "step": 803 }, { "epoch": 0.17, "learning_rate": 1.900292081638842e-05, "loss": 1.2721, "step": 804 }, { "epoch": 0.17, "learning_rate": 1.8999953009691977e-05, "loss": 1.3395, "step": 805 }, { "epoch": 0.17, "learning_rate": 1.899698102517552e-05, "loss": 1.2813, "step": 806 }, { "epoch": 0.17, "learning_rate": 1.8994004864218652e-05, "loss": 1.3151, "step": 807 }, { "epoch": 0.17, "learning_rate": 1.8991024528202933e-05, "loss": 1.311, "step": 808 }, { "epoch": 0.17, "learning_rate": 1.898804001851184e-05, "loss": 1.2997, "step": 809 }, { "epoch": 0.17, "learning_rate": 1.89850513365308e-05, "loss": 1.299, "step": 810 }, { "epoch": 0.17, "learning_rate": 1.898205848364717e-05, "loss": 1.2376, "step": 811 }, { "epoch": 0.17, "learning_rate": 1.897906146125025e-05, "loss": 1.3083, "step": 812 }, { "epoch": 0.17, "learning_rate": 1.897606027073127e-05, "loss": 1.2865, "step": 813 }, { "epoch": 0.17, "learning_rate": 1.8973054913483394e-05, "loss": 1.2709, "step": 814 }, { "epoch": 0.17, "learning_rate": 1.8970045390901727e-05, "loss": 1.2694, "step": 815 }, { "epoch": 0.17, "learning_rate": 1.8967031704383303e-05, "loss": 1.2804, "step": 816 }, { "epoch": 0.17, "learning_rate": 1.8964013855327086e-05, "loss": 1.2881, "step": 817 }, { "epoch": 0.17, "learning_rate": 1.896099184513398e-05, "loss": 1.2451, "step": 818 }, { "epoch": 0.17, "learning_rate": 1.895796567520681e-05, "loss": 1.2943, "step": 819 }, { "epoch": 0.17, "learning_rate": 1.8954935346950346e-05, "loss": 1.2441, "step": 820 }, { "epoch": 0.17, "learning_rate": 1.8951900861771274e-05, "loss": 1.2488, "step": 821 }, { "epoch": 0.17, "learning_rate": 1.894886222107822e-05, "loss": 1.2824, "step": 822 }, { "epoch": 0.17, "learning_rate": 1.8945819426281727e-05, "loss": 1.2706, "step": 823 }, { "epoch": 0.17, "learning_rate": 1.894277247879429e-05, "loss": 1.2365, "step": 824 }, { "epoch": 0.17, "learning_rate": 1.8939721380030302e-05, "loss": 1.3056, "step": 825 }, { "epoch": 0.17, "learning_rate": 1.89366661314061e-05, "loss": 1.2773, "step": 826 }, { "epoch": 0.17, "learning_rate": 1.8933606734339955e-05, "loss": 1.29, "step": 827 }, { "epoch": 0.17, "learning_rate": 1.893054319025204e-05, "loss": 1.2747, "step": 828 }, { "epoch": 0.17, "learning_rate": 1.892747550056447e-05, "loss": 1.2627, "step": 829 }, { "epoch": 0.17, "learning_rate": 1.8924403666701286e-05, "loss": 1.2602, "step": 830 }, { "epoch": 0.17, "learning_rate": 1.8921327690088444e-05, "loss": 1.2818, "step": 831 }, { "epoch": 0.18, "learning_rate": 1.8918247572153822e-05, "loss": 1.2129, "step": 832 }, { "epoch": 0.18, "learning_rate": 1.8915163314327235e-05, "loss": 1.33, "step": 833 }, { "epoch": 0.18, "learning_rate": 1.8912074918040404e-05, "loss": 1.2498, "step": 834 }, { "epoch": 0.18, "learning_rate": 1.8908982384726978e-05, "loss": 1.2589, "step": 835 }, { "epoch": 0.18, "learning_rate": 1.8905885715822526e-05, "loss": 1.2909, "step": 836 }, { "epoch": 0.18, "learning_rate": 1.8902784912764528e-05, "loss": 1.2669, "step": 837 }, { "epoch": 0.18, "learning_rate": 1.8899679976992407e-05, "loss": 1.2766, "step": 838 }, { "epoch": 0.18, "learning_rate": 1.8896570909947477e-05, "loss": 1.3106, "step": 839 }, { "epoch": 0.18, "learning_rate": 1.889345771307298e-05, "loss": 1.3401, "step": 840 }, { "epoch": 0.18, "learning_rate": 1.8890340387814086e-05, "loss": 1.2851, "step": 841 }, { "epoch": 0.18, "learning_rate": 1.8887218935617862e-05, "loss": 1.2543, "step": 842 }, { "epoch": 0.18, "learning_rate": 1.888409335793331e-05, "loss": 1.2481, "step": 843 }, { "epoch": 0.18, "learning_rate": 1.8880963656211325e-05, "loss": 1.2936, "step": 844 }, { "epoch": 0.18, "learning_rate": 1.8877829831904746e-05, "loss": 1.2959, "step": 845 }, { "epoch": 0.18, "learning_rate": 1.8874691886468293e-05, "loss": 1.2696, "step": 846 }, { "epoch": 0.18, "learning_rate": 1.8871549821358624e-05, "loss": 1.3119, "step": 847 }, { "epoch": 0.18, "learning_rate": 1.8868403638034302e-05, "loss": 1.2781, "step": 848 }, { "epoch": 0.18, "learning_rate": 1.886525333795579e-05, "loss": 1.2803, "step": 849 }, { "epoch": 0.18, "learning_rate": 1.8862098922585475e-05, "loss": 1.2656, "step": 850 }, { "epoch": 0.18, "learning_rate": 1.885894039338766e-05, "loss": 1.2584, "step": 851 }, { "epoch": 0.18, "learning_rate": 1.885577775182854e-05, "loss": 1.2337, "step": 852 }, { "epoch": 0.18, "learning_rate": 1.8852610999376235e-05, "loss": 1.2999, "step": 853 }, { "epoch": 0.18, "learning_rate": 1.8849440137500758e-05, "loss": 1.2722, "step": 854 }, { "epoch": 0.18, "learning_rate": 1.884626516767404e-05, "loss": 1.2992, "step": 855 }, { "epoch": 0.18, "learning_rate": 1.8843086091369923e-05, "loss": 1.2548, "step": 856 }, { "epoch": 0.18, "learning_rate": 1.883990291006414e-05, "loss": 1.2023, "step": 857 }, { "epoch": 0.18, "learning_rate": 1.8836715625234348e-05, "loss": 1.291, "step": 858 }, { "epoch": 0.18, "learning_rate": 1.883352423836009e-05, "loss": 1.3294, "step": 859 }, { "epoch": 0.18, "learning_rate": 1.883032875092283e-05, "loss": 1.2806, "step": 860 }, { "epoch": 0.18, "learning_rate": 1.8827129164405928e-05, "loss": 1.2882, "step": 861 }, { "epoch": 0.18, "learning_rate": 1.8823925480294637e-05, "loss": 1.2715, "step": 862 }, { "epoch": 0.18, "learning_rate": 1.8820717700076133e-05, "loss": 1.244, "step": 863 }, { "epoch": 0.18, "learning_rate": 1.881750582523948e-05, "loss": 1.2719, "step": 864 }, { "epoch": 0.18, "learning_rate": 1.8814289857275643e-05, "loss": 1.29, "step": 865 }, { "epoch": 0.18, "learning_rate": 1.8811069797677488e-05, "loss": 1.3332, "step": 866 }, { "epoch": 0.18, "learning_rate": 1.8807845647939787e-05, "loss": 1.2767, "step": 867 }, { "epoch": 0.18, "learning_rate": 1.88046174095592e-05, "loss": 1.2865, "step": 868 }, { "epoch": 0.18, "learning_rate": 1.8801385084034293e-05, "loss": 1.2691, "step": 869 }, { "epoch": 0.18, "learning_rate": 1.8798148672865525e-05, "loss": 1.261, "step": 870 }, { "epoch": 0.18, "learning_rate": 1.8794908177555254e-05, "loss": 1.281, "step": 871 }, { "epoch": 0.18, "learning_rate": 1.8791663599607736e-05, "loss": 1.2703, "step": 872 }, { "epoch": 0.18, "learning_rate": 1.8788414940529114e-05, "loss": 1.2891, "step": 873 }, { "epoch": 0.18, "learning_rate": 1.8785162201827433e-05, "loss": 1.2453, "step": 874 }, { "epoch": 0.18, "learning_rate": 1.878190538501263e-05, "loss": 1.3207, "step": 875 }, { "epoch": 0.18, "learning_rate": 1.877864449159653e-05, "loss": 1.2735, "step": 876 }, { "epoch": 0.18, "learning_rate": 1.8775379523092863e-05, "loss": 1.3772, "step": 877 }, { "epoch": 0.18, "learning_rate": 1.8772110481017238e-05, "loss": 1.2844, "step": 878 }, { "epoch": 0.18, "learning_rate": 1.8768837366887158e-05, "loss": 1.2144, "step": 879 }, { "epoch": 0.19, "learning_rate": 1.876556018222202e-05, "loss": 1.3447, "step": 880 }, { "epoch": 0.19, "learning_rate": 1.876227892854311e-05, "loss": 1.2813, "step": 881 }, { "epoch": 0.19, "learning_rate": 1.8758993607373593e-05, "loss": 1.2637, "step": 882 }, { "epoch": 0.19, "learning_rate": 1.875570422023854e-05, "loss": 1.2826, "step": 883 }, { "epoch": 0.19, "learning_rate": 1.8752410768664897e-05, "loss": 1.262, "step": 884 }, { "epoch": 0.19, "learning_rate": 1.8749113254181498e-05, "loss": 1.258, "step": 885 }, { "epoch": 0.19, "learning_rate": 1.8745811678319064e-05, "loss": 1.3409, "step": 886 }, { "epoch": 0.19, "learning_rate": 1.8742506042610208e-05, "loss": 1.2524, "step": 887 }, { "epoch": 0.19, "learning_rate": 1.873919634858941e-05, "loss": 1.2903, "step": 888 }, { "epoch": 0.19, "learning_rate": 1.8735882597793057e-05, "loss": 1.2677, "step": 889 }, { "epoch": 0.19, "learning_rate": 1.87325647917594e-05, "loss": 1.2754, "step": 890 }, { "epoch": 0.19, "learning_rate": 1.8729242932028585e-05, "loss": 1.2538, "step": 891 }, { "epoch": 0.19, "learning_rate": 1.8725917020142626e-05, "loss": 1.2456, "step": 892 }, { "epoch": 0.19, "learning_rate": 1.8722587057645437e-05, "loss": 1.2911, "step": 893 }, { "epoch": 0.19, "learning_rate": 1.87192530460828e-05, "loss": 1.2573, "step": 894 }, { "epoch": 0.19, "learning_rate": 1.8715914987002373e-05, "loss": 1.2251, "step": 895 }, { "epoch": 0.19, "learning_rate": 1.8712572881953704e-05, "loss": 1.3163, "step": 896 }, { "epoch": 0.19, "learning_rate": 1.8709226732488216e-05, "loss": 1.2118, "step": 897 }, { "epoch": 0.19, "learning_rate": 1.8705876540159196e-05, "loss": 1.2479, "step": 898 }, { "epoch": 0.19, "learning_rate": 1.870252230652183e-05, "loss": 1.2925, "step": 899 }, { "epoch": 0.19, "learning_rate": 1.869916403313316e-05, "loss": 1.234, "step": 900 }, { "epoch": 0.19, "learning_rate": 1.8695801721552122e-05, "loss": 1.2397, "step": 901 }, { "epoch": 0.19, "learning_rate": 1.869243537333951e-05, "loss": 1.2817, "step": 902 }, { "epoch": 0.19, "learning_rate": 1.8689064990058e-05, "loss": 1.2586, "step": 903 }, { "epoch": 0.19, "learning_rate": 1.8685690573272138e-05, "loss": 1.2683, "step": 904 }, { "epoch": 0.19, "learning_rate": 1.8682312124548346e-05, "loss": 1.2352, "step": 905 }, { "epoch": 0.19, "learning_rate": 1.8678929645454916e-05, "loss": 1.2204, "step": 906 }, { "epoch": 0.19, "learning_rate": 1.8675543137562012e-05, "loss": 1.24, "step": 907 }, { "epoch": 0.19, "learning_rate": 1.8672152602441663e-05, "loss": 1.3066, "step": 908 }, { "epoch": 0.19, "learning_rate": 1.8668758041667768e-05, "loss": 1.311, "step": 909 }, { "epoch": 0.19, "learning_rate": 1.8665359456816107e-05, "loss": 1.296, "step": 910 }, { "epoch": 0.19, "learning_rate": 1.866195684946431e-05, "loss": 1.2303, "step": 911 }, { "epoch": 0.19, "learning_rate": 1.865855022119189e-05, "loss": 1.2993, "step": 912 }, { "epoch": 0.19, "learning_rate": 1.865513957358022e-05, "loss": 1.2679, "step": 913 }, { "epoch": 0.19, "learning_rate": 1.8651724908212533e-05, "loss": 1.2811, "step": 914 }, { "epoch": 0.19, "learning_rate": 1.8648306226673933e-05, "loss": 1.2669, "step": 915 }, { "epoch": 0.19, "learning_rate": 1.864488353055139e-05, "loss": 1.2559, "step": 916 }, { "epoch": 0.19, "learning_rate": 1.8641456821433742e-05, "loss": 1.2428, "step": 917 }, { "epoch": 0.19, "learning_rate": 1.863802610091167e-05, "loss": 1.26, "step": 918 }, { "epoch": 0.19, "learning_rate": 1.863459137057774e-05, "loss": 1.2563, "step": 919 }, { "epoch": 0.19, "learning_rate": 1.8631152632026364e-05, "loss": 1.2735, "step": 920 }, { "epoch": 0.19, "learning_rate": 1.8627709886853825e-05, "loss": 1.2709, "step": 921 }, { "epoch": 0.19, "learning_rate": 1.8624263136658262e-05, "loss": 1.2177, "step": 922 }, { "epoch": 0.19, "learning_rate": 1.8620812383039665e-05, "loss": 1.2764, "step": 923 }, { "epoch": 0.19, "learning_rate": 1.86173576275999e-05, "loss": 1.2341, "step": 924 }, { "epoch": 0.19, "learning_rate": 1.8613898871942674e-05, "loss": 1.2408, "step": 925 }, { "epoch": 0.19, "learning_rate": 1.8610436117673557e-05, "loss": 1.2316, "step": 926 }, { "epoch": 0.19, "learning_rate": 1.860696936639998e-05, "loss": 1.2512, "step": 927 }, { "epoch": 0.2, "learning_rate": 1.860349861973122e-05, "loss": 1.2845, "step": 928 }, { "epoch": 0.2, "learning_rate": 1.860002387927842e-05, "loss": 1.2652, "step": 929 }, { "epoch": 0.2, "learning_rate": 1.859654514665456e-05, "loss": 1.2751, "step": 930 }, { "epoch": 0.2, "learning_rate": 1.8593062423474497e-05, "loss": 1.2466, "step": 931 }, { "epoch": 0.2, "learning_rate": 1.8589575711354922e-05, "loss": 1.2792, "step": 932 }, { "epoch": 0.2, "learning_rate": 1.8586085011914376e-05, "loss": 1.2807, "step": 933 }, { "epoch": 0.2, "learning_rate": 1.858259032677327e-05, "loss": 1.2744, "step": 934 }, { "epoch": 0.2, "learning_rate": 1.8579091657553844e-05, "loss": 1.2749, "step": 935 }, { "epoch": 0.2, "learning_rate": 1.85755890058802e-05, "loss": 1.3115, "step": 936 }, { "epoch": 0.2, "learning_rate": 1.8572082373378285e-05, "loss": 1.2884, "step": 937 }, { "epoch": 0.2, "learning_rate": 1.8568571761675893e-05, "loss": 1.2675, "step": 938 }, { "epoch": 0.2, "learning_rate": 1.856505717240267e-05, "loss": 1.2573, "step": 939 }, { "epoch": 0.2, "learning_rate": 1.85615386071901e-05, "loss": 1.236, "step": 940 }, { "epoch": 0.2, "learning_rate": 1.8558016067671517e-05, "loss": 1.2977, "step": 941 }, { "epoch": 0.2, "learning_rate": 1.8554489555482106e-05, "loss": 1.2269, "step": 942 }, { "epoch": 0.2, "learning_rate": 1.8550959072258883e-05, "loss": 1.235, "step": 943 }, { "epoch": 0.2, "learning_rate": 1.8547424619640722e-05, "loss": 1.2867, "step": 944 }, { "epoch": 0.2, "learning_rate": 1.8543886199268328e-05, "loss": 1.2822, "step": 945 }, { "epoch": 0.2, "learning_rate": 1.854034381278425e-05, "loss": 1.2269, "step": 946 }, { "epoch": 0.2, "learning_rate": 1.8536797461832887e-05, "loss": 1.2737, "step": 947 }, { "epoch": 0.2, "learning_rate": 1.8533247148060467e-05, "loss": 1.2491, "step": 948 }, { "epoch": 0.2, "learning_rate": 1.852969287311506e-05, "loss": 1.27, "step": 949 }, { "epoch": 0.2, "learning_rate": 1.8526134638646584e-05, "loss": 1.3033, "step": 950 }, { "epoch": 0.2, "learning_rate": 1.8522572446306784e-05, "loss": 1.2329, "step": 951 }, { "epoch": 0.2, "learning_rate": 1.851900629774924e-05, "loss": 1.302, "step": 952 }, { "epoch": 0.2, "learning_rate": 1.8515436194629388e-05, "loss": 1.2279, "step": 953 }, { "epoch": 0.2, "learning_rate": 1.8511862138604474e-05, "loss": 1.2557, "step": 954 }, { "epoch": 0.2, "learning_rate": 1.8508284131333604e-05, "loss": 1.265, "step": 955 }, { "epoch": 0.2, "learning_rate": 1.8504702174477692e-05, "loss": 1.2808, "step": 956 }, { "epoch": 0.2, "learning_rate": 1.8501116269699507e-05, "loss": 1.277, "step": 957 }, { "epoch": 0.2, "learning_rate": 1.8497526418663637e-05, "loss": 1.2876, "step": 958 }, { "epoch": 0.2, "learning_rate": 1.8493932623036518e-05, "loss": 1.2778, "step": 959 }, { "epoch": 0.2, "learning_rate": 1.8490334884486397e-05, "loss": 1.266, "step": 960 }, { "epoch": 0.2, "learning_rate": 1.8486733204683368e-05, "loss": 1.2415, "step": 961 }, { "epoch": 0.2, "learning_rate": 1.848312758529934e-05, "loss": 1.25, "step": 962 }, { "epoch": 0.2, "learning_rate": 1.847951802800806e-05, "loss": 1.2937, "step": 963 }, { "epoch": 0.2, "learning_rate": 1.8475904534485106e-05, "loss": 1.2748, "step": 964 }, { "epoch": 0.2, "learning_rate": 1.8472287106407874e-05, "loss": 1.3052, "step": 965 }, { "epoch": 0.2, "learning_rate": 1.8468665745455597e-05, "loss": 1.315, "step": 966 }, { "epoch": 0.2, "learning_rate": 1.846504045330932e-05, "loss": 1.2443, "step": 967 }, { "epoch": 0.2, "learning_rate": 1.8461411231651923e-05, "loss": 1.284, "step": 968 }, { "epoch": 0.2, "learning_rate": 1.845777808216811e-05, "loss": 1.2805, "step": 969 }, { "epoch": 0.2, "learning_rate": 1.84541410065444e-05, "loss": 1.2796, "step": 970 }, { "epoch": 0.2, "learning_rate": 1.845050000646915e-05, "loss": 1.2005, "step": 971 }, { "epoch": 0.2, "learning_rate": 1.844685508363252e-05, "loss": 1.3045, "step": 972 }, { "epoch": 0.2, "learning_rate": 1.8443206239726502e-05, "loss": 1.3283, "step": 973 }, { "epoch": 0.2, "learning_rate": 1.8439553476444913e-05, "loss": 1.312, "step": 974 }, { "epoch": 0.21, "learning_rate": 1.8435896795483376e-05, "loss": 1.2747, "step": 975 }, { "epoch": 0.21, "learning_rate": 1.843223619853934e-05, "loss": 1.2876, "step": 976 }, { "epoch": 0.21, "learning_rate": 1.8428571687312074e-05, "loss": 1.2639, "step": 977 }, { "epoch": 0.21, "learning_rate": 1.8424903263502658e-05, "loss": 1.2343, "step": 978 }, { "epoch": 0.21, "learning_rate": 1.8421230928813993e-05, "loss": 1.276, "step": 979 }, { "epoch": 0.21, "learning_rate": 1.8417554684950793e-05, "loss": 1.2461, "step": 980 }, { "epoch": 0.21, "learning_rate": 1.841387453361959e-05, "loss": 1.2503, "step": 981 }, { "epoch": 0.21, "learning_rate": 1.8410190476528723e-05, "loss": 1.2735, "step": 982 }, { "epoch": 0.21, "learning_rate": 1.8406502515388353e-05, "loss": 1.2623, "step": 983 }, { "epoch": 0.21, "learning_rate": 1.8402810651910444e-05, "loss": 1.2395, "step": 984 }, { "epoch": 0.21, "learning_rate": 1.839911488780878e-05, "loss": 1.2624, "step": 985 }, { "epoch": 0.21, "learning_rate": 1.839541522479895e-05, "loss": 1.2845, "step": 986 }, { "epoch": 0.21, "learning_rate": 1.8391711664598356e-05, "loss": 1.2338, "step": 987 }, { "epoch": 0.21, "learning_rate": 1.8388004208926208e-05, "loss": 1.223, "step": 988 }, { "epoch": 0.21, "learning_rate": 1.838429285950352e-05, "loss": 1.28, "step": 989 }, { "epoch": 0.21, "learning_rate": 1.8380577618053123e-05, "loss": 1.3271, "step": 990 }, { "epoch": 0.21, "learning_rate": 1.837685848629965e-05, "loss": 1.2511, "step": 991 }, { "epoch": 0.21, "learning_rate": 1.8373135465969532e-05, "loss": 1.2724, "step": 992 }, { "epoch": 0.21, "learning_rate": 1.8369408558791023e-05, "loss": 1.2525, "step": 993 }, { "epoch": 0.21, "learning_rate": 1.8365677766494158e-05, "loss": 1.2744, "step": 994 }, { "epoch": 0.21, "learning_rate": 1.8361943090810797e-05, "loss": 1.2867, "step": 995 }, { "epoch": 0.21, "learning_rate": 1.8358204533474592e-05, "loss": 1.2303, "step": 996 }, { "epoch": 0.21, "learning_rate": 1.8354462096220996e-05, "loss": 1.2674, "step": 997 }, { "epoch": 0.21, "learning_rate": 1.835071578078727e-05, "loss": 1.3139, "step": 998 }, { "epoch": 0.21, "learning_rate": 1.8346965588912467e-05, "loss": 1.24, "step": 999 }, { "epoch": 0.21, "learning_rate": 1.834321152233745e-05, "loss": 1.2725, "step": 1000 }, { "epoch": 0.21, "learning_rate": 1.8339453582804866e-05, "loss": 1.2648, "step": 1001 }, { "epoch": 0.21, "learning_rate": 1.833569177205917e-05, "loss": 1.2218, "step": 1002 }, { "epoch": 0.21, "learning_rate": 1.833192609184662e-05, "loss": 1.2718, "step": 1003 }, { "epoch": 0.21, "learning_rate": 1.8328156543915253e-05, "loss": 1.2888, "step": 1004 }, { "epoch": 0.21, "learning_rate": 1.832438313001491e-05, "loss": 1.2729, "step": 1005 }, { "epoch": 0.21, "learning_rate": 1.8320605851897233e-05, "loss": 1.2833, "step": 1006 }, { "epoch": 0.21, "learning_rate": 1.831682471131565e-05, "loss": 1.2774, "step": 1007 }, { "epoch": 0.21, "learning_rate": 1.8313039710025384e-05, "loss": 1.2949, "step": 1008 }, { "epoch": 0.21, "learning_rate": 1.830925084978345e-05, "loss": 1.2741, "step": 1009 }, { "epoch": 0.21, "learning_rate": 1.8305458132348655e-05, "loss": 1.2651, "step": 1010 }, { "epoch": 0.21, "learning_rate": 1.8301661559481597e-05, "loss": 1.2525, "step": 1011 }, { "epoch": 0.21, "learning_rate": 1.829786113294466e-05, "loss": 1.2708, "step": 1012 }, { "epoch": 0.21, "learning_rate": 1.829405685450202e-05, "loss": 1.2273, "step": 1013 }, { "epoch": 0.21, "learning_rate": 1.829024872591964e-05, "loss": 1.2799, "step": 1014 }, { "epoch": 0.21, "learning_rate": 1.828643674896528e-05, "loss": 1.3052, "step": 1015 }, { "epoch": 0.21, "learning_rate": 1.8282620925408464e-05, "loss": 1.2057, "step": 1016 }, { "epoch": 0.21, "learning_rate": 1.8278801257020524e-05, "loss": 1.2635, "step": 1017 }, { "epoch": 0.21, "learning_rate": 1.8274977745574566e-05, "loss": 1.2858, "step": 1018 }, { "epoch": 0.21, "learning_rate": 1.8271150392845477e-05, "loss": 1.2317, "step": 1019 }, { "epoch": 0.21, "learning_rate": 1.8267319200609937e-05, "loss": 1.2921, "step": 1020 }, { "epoch": 0.21, "learning_rate": 1.8263484170646404e-05, "loss": 1.2703, "step": 1021 }, { "epoch": 0.21, "learning_rate": 1.8259645304735112e-05, "loss": 1.2704, "step": 1022 }, { "epoch": 0.22, "learning_rate": 1.8255802604658086e-05, "loss": 1.2299, "step": 1023 }, { "epoch": 0.22, "learning_rate": 1.8251956072199124e-05, "loss": 1.2415, "step": 1024 }, { "epoch": 0.22, "learning_rate": 1.82481057091438e-05, "loss": 1.2356, "step": 1025 }, { "epoch": 0.22, "learning_rate": 1.8244251517279474e-05, "loss": 1.2692, "step": 1026 }, { "epoch": 0.22, "learning_rate": 1.8240393498395277e-05, "loss": 1.2863, "step": 1027 }, { "epoch": 0.22, "learning_rate": 1.8236531654282125e-05, "loss": 1.2754, "step": 1028 }, { "epoch": 0.22, "learning_rate": 1.8232665986732695e-05, "loss": 1.2501, "step": 1029 }, { "epoch": 0.22, "learning_rate": 1.8228796497541458e-05, "loss": 1.2264, "step": 1030 }, { "epoch": 0.22, "learning_rate": 1.822492318850464e-05, "loss": 1.2845, "step": 1031 }, { "epoch": 0.22, "learning_rate": 1.8221046061420253e-05, "loss": 1.2804, "step": 1032 }, { "epoch": 0.22, "learning_rate": 1.821716511808808e-05, "loss": 1.2578, "step": 1033 }, { "epoch": 0.22, "learning_rate": 1.8213280360309665e-05, "loss": 1.2427, "step": 1034 }, { "epoch": 0.22, "learning_rate": 1.820939178988834e-05, "loss": 1.2665, "step": 1035 }, { "epoch": 0.22, "learning_rate": 1.8205499408629196e-05, "loss": 1.2717, "step": 1036 }, { "epoch": 0.22, "learning_rate": 1.8201603218339087e-05, "loss": 1.2653, "step": 1037 }, { "epoch": 0.22, "learning_rate": 1.819770322082665e-05, "loss": 1.2713, "step": 1038 }, { "epoch": 0.22, "learning_rate": 1.8193799417902277e-05, "loss": 1.3112, "step": 1039 }, { "epoch": 0.22, "learning_rate": 1.8189891811378138e-05, "loss": 1.2708, "step": 1040 }, { "epoch": 0.22, "learning_rate": 1.8185980403068157e-05, "loss": 1.2848, "step": 1041 }, { "epoch": 0.22, "learning_rate": 1.8182065194788024e-05, "loss": 1.259, "step": 1042 }, { "epoch": 0.22, "learning_rate": 1.8178146188355205e-05, "loss": 1.2569, "step": 1043 }, { "epoch": 0.22, "learning_rate": 1.817422338558892e-05, "loss": 1.2697, "step": 1044 }, { "epoch": 0.22, "learning_rate": 1.817029678831015e-05, "loss": 1.2895, "step": 1045 }, { "epoch": 0.22, "learning_rate": 1.8166366398341634e-05, "loss": 1.2735, "step": 1046 }, { "epoch": 0.22, "learning_rate": 1.816243221750789e-05, "loss": 1.2531, "step": 1047 }, { "epoch": 0.22, "learning_rate": 1.815849424763517e-05, "loss": 1.2589, "step": 1048 }, { "epoch": 0.22, "learning_rate": 1.815455249055151e-05, "loss": 1.2595, "step": 1049 }, { "epoch": 0.22, "learning_rate": 1.815060694808668e-05, "loss": 1.2703, "step": 1050 }, { "epoch": 0.22, "learning_rate": 1.814665762207223e-05, "loss": 1.2634, "step": 1051 }, { "epoch": 0.22, "learning_rate": 1.8142704514341447e-05, "loss": 1.2199, "step": 1052 }, { "epoch": 0.22, "learning_rate": 1.813874762672939e-05, "loss": 1.3005, "step": 1053 }, { "epoch": 0.22, "learning_rate": 1.8134786961072855e-05, "loss": 1.2753, "step": 1054 }, { "epoch": 0.22, "learning_rate": 1.813082251921041e-05, "loss": 1.2754, "step": 1055 }, { "epoch": 0.22, "learning_rate": 1.8126854302982362e-05, "loss": 1.2478, "step": 1056 }, { "epoch": 0.22, "learning_rate": 1.8122882314230778e-05, "loss": 1.2757, "step": 1057 }, { "epoch": 0.22, "learning_rate": 1.8118906554799472e-05, "loss": 1.3, "step": 1058 }, { "epoch": 0.22, "learning_rate": 1.8114927026534014e-05, "loss": 1.2578, "step": 1059 }, { "epoch": 0.22, "learning_rate": 1.8110943731281715e-05, "loss": 1.2732, "step": 1060 }, { "epoch": 0.22, "learning_rate": 1.810695667089164e-05, "loss": 1.2457, "step": 1061 }, { "epoch": 0.22, "learning_rate": 1.81029658472146e-05, "loss": 1.2667, "step": 1062 }, { "epoch": 0.22, "learning_rate": 1.809897126210316e-05, "loss": 1.259, "step": 1063 }, { "epoch": 0.22, "learning_rate": 1.8094972917411616e-05, "loss": 1.2669, "step": 1064 }, { "epoch": 0.22, "learning_rate": 1.8090970814996025e-05, "loss": 1.2524, "step": 1065 }, { "epoch": 0.22, "learning_rate": 1.808696495671418e-05, "loss": 1.2541, "step": 1066 }, { "epoch": 0.22, "learning_rate": 1.8082955344425616e-05, "loss": 1.2653, "step": 1067 }, { "epoch": 0.22, "learning_rate": 1.8078941979991617e-05, "loss": 1.1984, "step": 1068 }, { "epoch": 0.22, "learning_rate": 1.8074924865275205e-05, "loss": 1.2681, "step": 1069 }, { "epoch": 0.23, "learning_rate": 1.807090400214114e-05, "loss": 1.2956, "step": 1070 }, { "epoch": 0.23, "learning_rate": 1.8066879392455932e-05, "loss": 1.2725, "step": 1071 }, { "epoch": 0.23, "learning_rate": 1.8062851038087818e-05, "loss": 1.2823, "step": 1072 }, { "epoch": 0.23, "learning_rate": 1.805881894090678e-05, "loss": 1.2424, "step": 1073 }, { "epoch": 0.23, "learning_rate": 1.8054783102784537e-05, "loss": 1.2742, "step": 1074 }, { "epoch": 0.23, "learning_rate": 1.805074352559454e-05, "loss": 1.273, "step": 1075 }, { "epoch": 0.23, "learning_rate": 1.8046700211211986e-05, "loss": 1.221, "step": 1076 }, { "epoch": 0.23, "learning_rate": 1.8042653161513798e-05, "loss": 1.3004, "step": 1077 }, { "epoch": 0.23, "learning_rate": 1.8038602378378636e-05, "loss": 1.3067, "step": 1078 }, { "epoch": 0.23, "learning_rate": 1.803454786368689e-05, "loss": 1.2845, "step": 1079 }, { "epoch": 0.23, "learning_rate": 1.8030489619320682e-05, "loss": 1.2649, "step": 1080 }, { "epoch": 0.23, "learning_rate": 1.8026427647163873e-05, "loss": 1.2854, "step": 1081 }, { "epoch": 0.23, "learning_rate": 1.8022361949102046e-05, "loss": 1.2322, "step": 1082 }, { "epoch": 0.23, "learning_rate": 1.8018292527022522e-05, "loss": 1.2436, "step": 1083 }, { "epoch": 0.23, "learning_rate": 1.801421938281434e-05, "loss": 1.2654, "step": 1084 }, { "epoch": 0.23, "learning_rate": 1.8010142518368277e-05, "loss": 1.2338, "step": 1085 }, { "epoch": 0.23, "learning_rate": 1.800606193557683e-05, "loss": 1.2854, "step": 1086 }, { "epoch": 0.23, "learning_rate": 1.8001977636334222e-05, "loss": 1.2685, "step": 1087 }, { "epoch": 0.23, "learning_rate": 1.7997889622536408e-05, "loss": 1.2575, "step": 1088 }, { "epoch": 0.23, "learning_rate": 1.7993797896081063e-05, "loss": 1.3194, "step": 1089 }, { "epoch": 0.23, "learning_rate": 1.798970245886758e-05, "loss": 1.2184, "step": 1090 }, { "epoch": 0.23, "learning_rate": 1.7985603312797088e-05, "loss": 1.2788, "step": 1091 }, { "epoch": 0.23, "learning_rate": 1.7981500459772422e-05, "loss": 1.2692, "step": 1092 }, { "epoch": 0.23, "learning_rate": 1.797739390169815e-05, "loss": 1.247, "step": 1093 }, { "epoch": 0.23, "learning_rate": 1.7973283640480554e-05, "loss": 1.2873, "step": 1094 }, { "epoch": 0.23, "learning_rate": 1.7969169678027637e-05, "loss": 1.2719, "step": 1095 }, { "epoch": 0.23, "learning_rate": 1.796505201624912e-05, "loss": 1.2649, "step": 1096 }, { "epoch": 0.23, "learning_rate": 1.796093065705644e-05, "loss": 1.2721, "step": 1097 }, { "epoch": 0.23, "learning_rate": 1.7956805602362748e-05, "loss": 1.2803, "step": 1098 }, { "epoch": 0.23, "learning_rate": 1.795267685408292e-05, "loss": 1.3043, "step": 1099 }, { "epoch": 0.23, "learning_rate": 1.7948544414133534e-05, "loss": 1.2462, "step": 1100 }, { "epoch": 0.23, "learning_rate": 1.794440828443289e-05, "loss": 1.2445, "step": 1101 }, { "epoch": 0.23, "learning_rate": 1.7940268466901e-05, "loss": 1.2799, "step": 1102 }, { "epoch": 0.23, "learning_rate": 1.7936124963459586e-05, "loss": 1.3007, "step": 1103 }, { "epoch": 0.23, "learning_rate": 1.7931977776032085e-05, "loss": 1.245, "step": 1104 }, { "epoch": 0.23, "learning_rate": 1.792782690654363e-05, "loss": 1.194, "step": 1105 }, { "epoch": 0.23, "learning_rate": 1.7923672356921086e-05, "loss": 1.2301, "step": 1106 }, { "epoch": 0.23, "learning_rate": 1.7919514129093007e-05, "loss": 1.2714, "step": 1107 }, { "epoch": 0.23, "learning_rate": 1.791535222498966e-05, "loss": 1.2838, "step": 1108 }, { "epoch": 0.23, "learning_rate": 1.7911186646543028e-05, "loss": 1.2841, "step": 1109 }, { "epoch": 0.23, "learning_rate": 1.790701739568679e-05, "loss": 1.2518, "step": 1110 }, { "epoch": 0.23, "learning_rate": 1.7902844474356324e-05, "loss": 1.2721, "step": 1111 }, { "epoch": 0.23, "learning_rate": 1.789866788448873e-05, "loss": 1.2407, "step": 1112 }, { "epoch": 0.23, "learning_rate": 1.789448762802279e-05, "loss": 1.3033, "step": 1113 }, { "epoch": 0.23, "learning_rate": 1.7890303706899005e-05, "loss": 1.3214, "step": 1114 }, { "epoch": 0.23, "learning_rate": 1.7886116123059574e-05, "loss": 1.2559, "step": 1115 }, { "epoch": 0.23, "learning_rate": 1.7881924878448384e-05, "loss": 1.2475, "step": 1116 }, { "epoch": 0.23, "learning_rate": 1.7877729975011036e-05, "loss": 1.2803, "step": 1117 }, { "epoch": 0.24, "learning_rate": 1.787353141469482e-05, "loss": 1.3183, "step": 1118 }, { "epoch": 0.24, "learning_rate": 1.786932919944873e-05, "loss": 1.3077, "step": 1119 }, { "epoch": 0.24, "learning_rate": 1.7865123331223458e-05, "loss": 1.259, "step": 1120 }, { "epoch": 0.24, "learning_rate": 1.786091381197138e-05, "loss": 1.2875, "step": 1121 }, { "epoch": 0.24, "learning_rate": 1.7856700643646575e-05, "loss": 1.2321, "step": 1122 }, { "epoch": 0.24, "learning_rate": 1.785248382820482e-05, "loss": 1.2292, "step": 1123 }, { "epoch": 0.24, "learning_rate": 1.7848263367603577e-05, "loss": 1.2717, "step": 1124 }, { "epoch": 0.24, "learning_rate": 1.7844039263802e-05, "loss": 1.2085, "step": 1125 }, { "epoch": 0.24, "learning_rate": 1.783981151876095e-05, "loss": 1.2286, "step": 1126 }, { "epoch": 0.24, "learning_rate": 1.7835580134442954e-05, "loss": 1.2909, "step": 1127 }, { "epoch": 0.24, "learning_rate": 1.7831345112812242e-05, "loss": 1.245, "step": 1128 }, { "epoch": 0.24, "learning_rate": 1.782710645583473e-05, "loss": 1.2384, "step": 1129 }, { "epoch": 0.24, "learning_rate": 1.7822864165478037e-05, "loss": 1.2586, "step": 1130 }, { "epoch": 0.24, "learning_rate": 1.781861824371143e-05, "loss": 1.2442, "step": 1131 }, { "epoch": 0.24, "learning_rate": 1.7814368692505907e-05, "loss": 1.2454, "step": 1132 }, { "epoch": 0.24, "learning_rate": 1.781011551383411e-05, "loss": 1.2637, "step": 1133 }, { "epoch": 0.24, "learning_rate": 1.7805858709670404e-05, "loss": 1.2977, "step": 1134 }, { "epoch": 0.24, "learning_rate": 1.78015982819908e-05, "loss": 1.2777, "step": 1135 }, { "epoch": 0.24, "learning_rate": 1.779733423277302e-05, "loss": 1.2959, "step": 1136 }, { "epoch": 0.24, "learning_rate": 1.7793066563996452e-05, "loss": 1.3069, "step": 1137 }, { "epoch": 0.24, "learning_rate": 1.7788795277642165e-05, "loss": 1.2405, "step": 1138 }, { "epoch": 0.24, "learning_rate": 1.7784520375692912e-05, "loss": 1.2809, "step": 1139 }, { "epoch": 0.24, "learning_rate": 1.7780241860133123e-05, "loss": 1.2179, "step": 1140 }, { "epoch": 0.24, "learning_rate": 1.7775959732948904e-05, "loss": 1.2936, "step": 1141 }, { "epoch": 0.24, "learning_rate": 1.777167399612804e-05, "loss": 1.2572, "step": 1142 }, { "epoch": 0.24, "learning_rate": 1.776738465165998e-05, "loss": 1.2497, "step": 1143 }, { "epoch": 0.24, "learning_rate": 1.7763091701535877e-05, "loss": 1.2403, "step": 1144 }, { "epoch": 0.24, "learning_rate": 1.7758795147748526e-05, "loss": 1.3083, "step": 1145 }, { "epoch": 0.24, "learning_rate": 1.7754494992292404e-05, "loss": 1.2564, "step": 1146 }, { "epoch": 0.24, "learning_rate": 1.775019123716367e-05, "loss": 1.2903, "step": 1147 }, { "epoch": 0.24, "learning_rate": 1.7745883884360143e-05, "loss": 1.343, "step": 1148 }, { "epoch": 0.24, "learning_rate": 1.7741572935881323e-05, "loss": 1.2845, "step": 1149 }, { "epoch": 0.24, "learning_rate": 1.7737258393728363e-05, "loss": 1.2807, "step": 1150 }, { "epoch": 0.24, "learning_rate": 1.77329402599041e-05, "loss": 1.2548, "step": 1151 }, { "epoch": 0.24, "learning_rate": 1.772861853641303e-05, "loss": 1.2082, "step": 1152 }, { "epoch": 0.24, "learning_rate": 1.7724293225261317e-05, "loss": 1.2685, "step": 1153 }, { "epoch": 0.24, "learning_rate": 1.7719964328456793e-05, "loss": 1.2608, "step": 1154 }, { "epoch": 0.24, "learning_rate": 1.7715631848008944e-05, "loss": 1.2319, "step": 1155 }, { "epoch": 0.24, "learning_rate": 1.7711295785928936e-05, "loss": 1.2472, "step": 1156 }, { "epoch": 0.24, "learning_rate": 1.7706956144229587e-05, "loss": 1.2511, "step": 1157 }, { "epoch": 0.24, "learning_rate": 1.7702612924925377e-05, "loss": 1.2902, "step": 1158 }, { "epoch": 0.24, "learning_rate": 1.769826613003245e-05, "loss": 1.2135, "step": 1159 }, { "epoch": 0.24, "learning_rate": 1.7693915761568608e-05, "loss": 1.2929, "step": 1160 }, { "epoch": 0.24, "learning_rate": 1.7689561821553313e-05, "loss": 1.2568, "step": 1161 }, { "epoch": 0.24, "learning_rate": 1.7685204312007682e-05, "loss": 1.2856, "step": 1162 }, { "epoch": 0.24, "learning_rate": 1.7680843234954494e-05, "loss": 1.3143, "step": 1163 }, { "epoch": 0.24, "learning_rate": 1.767647859241818e-05, "loss": 1.2799, "step": 1164 }, { "epoch": 0.25, "learning_rate": 1.7672110386424827e-05, "loss": 1.3241, "step": 1165 }, { "epoch": 0.25, "learning_rate": 1.7667738619002175e-05, "loss": 1.269, "step": 1166 }, { "epoch": 0.25, "learning_rate": 1.7663363292179624e-05, "loss": 1.2463, "step": 1167 }, { "epoch": 0.25, "learning_rate": 1.7658984407988216e-05, "loss": 1.3375, "step": 1168 }, { "epoch": 0.25, "learning_rate": 1.765460196846065e-05, "loss": 1.2926, "step": 1169 }, { "epoch": 0.25, "learning_rate": 1.7650215975631275e-05, "loss": 1.2059, "step": 1170 }, { "epoch": 0.25, "learning_rate": 1.7645826431536094e-05, "loss": 1.2889, "step": 1171 }, { "epoch": 0.25, "learning_rate": 1.7641433338212745e-05, "loss": 1.2661, "step": 1172 }, { "epoch": 0.25, "learning_rate": 1.763703669770053e-05, "loss": 1.2984, "step": 1173 }, { "epoch": 0.25, "learning_rate": 1.7632636512040388e-05, "loss": 1.2396, "step": 1174 }, { "epoch": 0.25, "learning_rate": 1.76282327832749e-05, "loss": 1.2637, "step": 1175 }, { "epoch": 0.25, "learning_rate": 1.762382551344831e-05, "loss": 1.256, "step": 1176 }, { "epoch": 0.25, "learning_rate": 1.761941470460648e-05, "loss": 1.2616, "step": 1177 }, { "epoch": 0.25, "learning_rate": 1.7615000358796932e-05, "loss": 1.2521, "step": 1178 }, { "epoch": 0.25, "learning_rate": 1.761058247806883e-05, "loss": 1.2622, "step": 1179 }, { "epoch": 0.25, "learning_rate": 1.760616106447297e-05, "loss": 1.2531, "step": 1180 }, { "epoch": 0.25, "learning_rate": 1.7601736120061797e-05, "loss": 1.2892, "step": 1181 }, { "epoch": 0.25, "learning_rate": 1.7597307646889386e-05, "loss": 1.2603, "step": 1182 }, { "epoch": 0.25, "learning_rate": 1.7592875647011457e-05, "loss": 1.2383, "step": 1183 }, { "epoch": 0.25, "learning_rate": 1.7588440122485367e-05, "loss": 1.2557, "step": 1184 }, { "epoch": 0.25, "learning_rate": 1.7584001075370105e-05, "loss": 1.2671, "step": 1185 }, { "epoch": 0.25, "learning_rate": 1.7579558507726296e-05, "loss": 1.2578, "step": 1186 }, { "epoch": 0.25, "learning_rate": 1.7575112421616203e-05, "loss": 1.216, "step": 1187 }, { "epoch": 0.25, "learning_rate": 1.757066281910372e-05, "loss": 1.2586, "step": 1188 }, { "epoch": 0.25, "learning_rate": 1.756620970225437e-05, "loss": 1.2487, "step": 1189 }, { "epoch": 0.25, "learning_rate": 1.756175307313531e-05, "loss": 1.238, "step": 1190 }, { "epoch": 0.25, "learning_rate": 1.7557292933815335e-05, "loss": 1.246, "step": 1191 }, { "epoch": 0.25, "learning_rate": 1.755282928636486e-05, "loss": 1.237, "step": 1192 }, { "epoch": 0.25, "learning_rate": 1.7548362132855924e-05, "loss": 1.2323, "step": 1193 }, { "epoch": 0.25, "learning_rate": 1.7543891475362204e-05, "loss": 1.2417, "step": 1194 }, { "epoch": 0.25, "learning_rate": 1.7539417315959007e-05, "loss": 1.274, "step": 1195 }, { "epoch": 0.25, "learning_rate": 1.7534939656723245e-05, "loss": 1.2907, "step": 1196 }, { "epoch": 0.25, "learning_rate": 1.7530458499733484e-05, "loss": 1.2362, "step": 1197 }, { "epoch": 0.25, "learning_rate": 1.7525973847069885e-05, "loss": 1.2817, "step": 1198 }, { "epoch": 0.25, "learning_rate": 1.7521485700814255e-05, "loss": 1.2042, "step": 1199 }, { "epoch": 0.25, "learning_rate": 1.751699406305e-05, "loss": 1.239, "step": 1200 }, { "epoch": 0.25, "learning_rate": 1.751249893586217e-05, "loss": 1.2836, "step": 1201 }, { "epoch": 0.25, "learning_rate": 1.7508000321337418e-05, "loss": 1.2574, "step": 1202 }, { "epoch": 0.25, "learning_rate": 1.7503498221564026e-05, "loss": 1.2471, "step": 1203 }, { "epoch": 0.25, "learning_rate": 1.7498992638631883e-05, "loss": 1.2482, "step": 1204 }, { "epoch": 0.25, "learning_rate": 1.749448357463251e-05, "loss": 1.2419, "step": 1205 }, { "epoch": 0.25, "learning_rate": 1.7489971031659033e-05, "loss": 1.2755, "step": 1206 }, { "epoch": 0.25, "learning_rate": 1.748545501180619e-05, "loss": 1.239, "step": 1207 }, { "epoch": 0.25, "learning_rate": 1.7480935517170345e-05, "loss": 1.2713, "step": 1208 }, { "epoch": 0.25, "learning_rate": 1.7476412549849463e-05, "loss": 1.26, "step": 1209 }, { "epoch": 0.25, "learning_rate": 1.7471886111943132e-05, "loss": 1.2552, "step": 1210 }, { "epoch": 0.25, "learning_rate": 1.7467356205552536e-05, "loss": 1.2598, "step": 1211 }, { "epoch": 0.25, "learning_rate": 1.746282283278049e-05, "loss": 1.2497, "step": 1212 }, { "epoch": 0.26, "learning_rate": 1.7458285995731405e-05, "loss": 1.2684, "step": 1213 }, { "epoch": 0.26, "learning_rate": 1.745374569651129e-05, "loss": 1.3255, "step": 1214 }, { "epoch": 0.26, "learning_rate": 1.7449201937227792e-05, "loss": 1.2797, "step": 1215 }, { "epoch": 0.26, "learning_rate": 1.7444654719990128e-05, "loss": 1.2572, "step": 1216 }, { "epoch": 0.26, "learning_rate": 1.7440104046909146e-05, "loss": 1.3067, "step": 1217 }, { "epoch": 0.26, "learning_rate": 1.7435549920097288e-05, "loss": 1.2363, "step": 1218 }, { "epoch": 0.26, "learning_rate": 1.7430992341668603e-05, "loss": 1.2167, "step": 1219 }, { "epoch": 0.26, "learning_rate": 1.7426431313738735e-05, "loss": 1.2234, "step": 1220 }, { "epoch": 0.26, "learning_rate": 1.7421866838424938e-05, "loss": 1.2891, "step": 1221 }, { "epoch": 0.26, "learning_rate": 1.7417298917846058e-05, "loss": 1.2434, "step": 1222 }, { "epoch": 0.26, "learning_rate": 1.7412727554122555e-05, "loss": 1.2796, "step": 1223 }, { "epoch": 0.26, "learning_rate": 1.7408152749376465e-05, "loss": 1.2467, "step": 1224 }, { "epoch": 0.26, "learning_rate": 1.7403574505731446e-05, "loss": 1.2941, "step": 1225 }, { "epoch": 0.26, "learning_rate": 1.739899282531273e-05, "loss": 1.2409, "step": 1226 }, { "epoch": 0.26, "learning_rate": 1.739440771024716e-05, "loss": 1.2811, "step": 1227 }, { "epoch": 0.26, "learning_rate": 1.7389819162663163e-05, "loss": 1.3096, "step": 1228 }, { "epoch": 0.26, "learning_rate": 1.738522718469077e-05, "loss": 1.286, "step": 1229 }, { "epoch": 0.26, "learning_rate": 1.738063177846159e-05, "loss": 1.2678, "step": 1230 }, { "epoch": 0.26, "learning_rate": 1.737603294610884e-05, "loss": 1.2278, "step": 1231 }, { "epoch": 0.26, "learning_rate": 1.7371430689767318e-05, "loss": 1.2899, "step": 1232 }, { "epoch": 0.26, "learning_rate": 1.736682501157341e-05, "loss": 1.2591, "step": 1233 }, { "epoch": 0.26, "learning_rate": 1.7362215913665098e-05, "loss": 1.2466, "step": 1234 }, { "epoch": 0.26, "learning_rate": 1.7357603398181938e-05, "loss": 1.2352, "step": 1235 }, { "epoch": 0.26, "learning_rate": 1.735298746726509e-05, "loss": 1.2782, "step": 1236 }, { "epoch": 0.26, "learning_rate": 1.7348368123057286e-05, "loss": 1.2416, "step": 1237 }, { "epoch": 0.26, "learning_rate": 1.7343745367702848e-05, "loss": 1.2485, "step": 1238 }, { "epoch": 0.26, "learning_rate": 1.733911920334768e-05, "loss": 1.297, "step": 1239 }, { "epoch": 0.26, "learning_rate": 1.733448963213927e-05, "loss": 1.2374, "step": 1240 }, { "epoch": 0.26, "learning_rate": 1.7329856656226685e-05, "loss": 1.2548, "step": 1241 }, { "epoch": 0.26, "learning_rate": 1.732522027776058e-05, "loss": 1.2383, "step": 1242 }, { "epoch": 0.26, "learning_rate": 1.732058049889317e-05, "loss": 1.2463, "step": 1243 }, { "epoch": 0.26, "learning_rate": 1.7315937321778275e-05, "loss": 1.2767, "step": 1244 }, { "epoch": 0.26, "learning_rate": 1.7311290748571273e-05, "loss": 1.2753, "step": 1245 }, { "epoch": 0.26, "learning_rate": 1.730664078142913e-05, "loss": 1.2283, "step": 1246 }, { "epoch": 0.26, "learning_rate": 1.7301987422510377e-05, "loss": 1.2933, "step": 1247 }, { "epoch": 0.26, "learning_rate": 1.7297330673975127e-05, "loss": 1.3265, "step": 1248 }, { "epoch": 0.26, "learning_rate": 1.7292670537985064e-05, "loss": 1.2717, "step": 1249 }, { "epoch": 0.26, "learning_rate": 1.7288007016703446e-05, "loss": 1.2868, "step": 1250 }, { "epoch": 0.26, "learning_rate": 1.72833401122951e-05, "loss": 1.273, "step": 1251 }, { "epoch": 0.26, "learning_rate": 1.727866982692642e-05, "loss": 1.2506, "step": 1252 }, { "epoch": 0.26, "learning_rate": 1.7273996162765384e-05, "loss": 1.2771, "step": 1253 }, { "epoch": 0.26, "learning_rate": 1.7269319121981523e-05, "loss": 1.2099, "step": 1254 }, { "epoch": 0.26, "learning_rate": 1.7264638706745942e-05, "loss": 1.2733, "step": 1255 }, { "epoch": 0.26, "learning_rate": 1.725995491923131e-05, "loss": 1.3026, "step": 1256 }, { "epoch": 0.26, "learning_rate": 1.7255267761611864e-05, "loss": 1.2856, "step": 1257 }, { "epoch": 0.26, "learning_rate": 1.7250577236063408e-05, "loss": 1.2976, "step": 1258 }, { "epoch": 0.26, "learning_rate": 1.7245883344763298e-05, "loss": 1.2452, "step": 1259 }, { "epoch": 0.27, "learning_rate": 1.7241186089890465e-05, "loss": 1.2575, "step": 1260 }, { "epoch": 0.27, "learning_rate": 1.7236485473625397e-05, "loss": 1.2448, "step": 1261 }, { "epoch": 0.27, "learning_rate": 1.723178149815014e-05, "loss": 1.2801, "step": 1262 }, { "epoch": 0.27, "learning_rate": 1.7227074165648303e-05, "loss": 1.2414, "step": 1263 }, { "epoch": 0.27, "learning_rate": 1.722236347830505e-05, "loss": 1.2769, "step": 1264 }, { "epoch": 0.27, "learning_rate": 1.7217649438307105e-05, "loss": 1.292, "step": 1265 }, { "epoch": 0.27, "learning_rate": 1.721293204784275e-05, "loss": 1.3293, "step": 1266 }, { "epoch": 0.27, "learning_rate": 1.7208211309101813e-05, "loss": 1.2611, "step": 1267 }, { "epoch": 0.27, "learning_rate": 1.7203487224275686e-05, "loss": 1.2488, "step": 1268 }, { "epoch": 0.27, "learning_rate": 1.7198759795557315e-05, "loss": 1.2961, "step": 1269 }, { "epoch": 0.27, "learning_rate": 1.719402902514119e-05, "loss": 1.2626, "step": 1270 }, { "epoch": 0.27, "learning_rate": 1.7189294915223357e-05, "loss": 1.2453, "step": 1271 }, { "epoch": 0.27, "learning_rate": 1.718455746800141e-05, "loss": 1.212, "step": 1272 }, { "epoch": 0.27, "learning_rate": 1.71798166856745e-05, "loss": 1.2638, "step": 1273 }, { "epoch": 0.27, "learning_rate": 1.717507257044331e-05, "loss": 1.2552, "step": 1274 }, { "epoch": 0.27, "learning_rate": 1.7170325124510095e-05, "loss": 1.2445, "step": 1275 }, { "epoch": 0.27, "learning_rate": 1.7165574350078632e-05, "loss": 1.2452, "step": 1276 }, { "epoch": 0.27, "learning_rate": 1.716082024935425e-05, "loss": 1.2417, "step": 1277 }, { "epoch": 0.27, "learning_rate": 1.7156062824543832e-05, "loss": 1.2521, "step": 1278 }, { "epoch": 0.27, "learning_rate": 1.715130207785579e-05, "loss": 1.2882, "step": 1279 }, { "epoch": 0.27, "learning_rate": 1.7146538011500095e-05, "loss": 1.241, "step": 1280 }, { "epoch": 0.27, "learning_rate": 1.7141770627688233e-05, "loss": 1.2337, "step": 1281 }, { "epoch": 0.27, "learning_rate": 1.713699992863326e-05, "loss": 1.2926, "step": 1282 }, { "epoch": 0.27, "learning_rate": 1.713222591654975e-05, "loss": 1.2624, "step": 1283 }, { "epoch": 0.27, "learning_rate": 1.7127448593653825e-05, "loss": 1.299, "step": 1284 }, { "epoch": 0.27, "learning_rate": 1.7122667962163137e-05, "loss": 1.2683, "step": 1285 }, { "epoch": 0.27, "learning_rate": 1.7117884024296876e-05, "loss": 1.2432, "step": 1286 }, { "epoch": 0.27, "learning_rate": 1.711309678227577e-05, "loss": 1.2984, "step": 1287 }, { "epoch": 0.27, "learning_rate": 1.7108306238322084e-05, "loss": 1.2486, "step": 1288 }, { "epoch": 0.27, "learning_rate": 1.7103512394659605e-05, "loss": 1.247, "step": 1289 }, { "epoch": 0.27, "learning_rate": 1.709871525351366e-05, "loss": 1.2242, "step": 1290 }, { "epoch": 0.27, "learning_rate": 1.7093914817111103e-05, "loss": 1.2535, "step": 1291 }, { "epoch": 0.27, "learning_rate": 1.708911108768032e-05, "loss": 1.2787, "step": 1292 }, { "epoch": 0.27, "learning_rate": 1.7084304067451222e-05, "loss": 1.2222, "step": 1293 }, { "epoch": 0.27, "learning_rate": 1.7079493758655253e-05, "loss": 1.2548, "step": 1294 }, { "epoch": 0.27, "learning_rate": 1.7074680163525375e-05, "loss": 1.2708, "step": 1295 }, { "epoch": 0.27, "learning_rate": 1.706986328429609e-05, "loss": 1.2649, "step": 1296 }, { "epoch": 0.27, "learning_rate": 1.7065043123203417e-05, "loss": 1.2816, "step": 1297 }, { "epoch": 0.27, "learning_rate": 1.7060219682484883e-05, "loss": 1.2348, "step": 1298 }, { "epoch": 0.27, "learning_rate": 1.705539296437956e-05, "loss": 1.2513, "step": 1299 }, { "epoch": 0.27, "learning_rate": 1.7050562971128036e-05, "loss": 1.2547, "step": 1300 }, { "epoch": 0.27, "learning_rate": 1.7045729704972418e-05, "loss": 1.2244, "step": 1301 }, { "epoch": 0.27, "learning_rate": 1.7040893168156324e-05, "loss": 1.2375, "step": 1302 }, { "epoch": 0.27, "learning_rate": 1.7036053362924896e-05, "loss": 1.2802, "step": 1303 }, { "epoch": 0.27, "learning_rate": 1.70312102915248e-05, "loss": 1.229, "step": 1304 }, { "epoch": 0.27, "learning_rate": 1.7026363956204207e-05, "loss": 1.269, "step": 1305 }, { "epoch": 0.27, "learning_rate": 1.7021514359212813e-05, "loss": 1.2468, "step": 1306 }, { "epoch": 0.27, "learning_rate": 1.701666150280182e-05, "loss": 1.2528, "step": 1307 }, { "epoch": 0.28, "learning_rate": 1.7011805389223947e-05, "loss": 1.2724, "step": 1308 }, { "epoch": 0.28, "learning_rate": 1.7006946020733426e-05, "loss": 1.2466, "step": 1309 }, { "epoch": 0.28, "learning_rate": 1.7002083399586002e-05, "loss": 1.2788, "step": 1310 }, { "epoch": 0.28, "learning_rate": 1.6997217528038913e-05, "loss": 1.2707, "step": 1311 }, { "epoch": 0.28, "learning_rate": 1.6992348408350933e-05, "loss": 1.2261, "step": 1312 }, { "epoch": 0.28, "learning_rate": 1.6987476042782324e-05, "loss": 1.2982, "step": 1313 }, { "epoch": 0.28, "learning_rate": 1.6982600433594867e-05, "loss": 1.3045, "step": 1314 }, { "epoch": 0.28, "learning_rate": 1.6977721583051835e-05, "loss": 1.2778, "step": 1315 }, { "epoch": 0.28, "learning_rate": 1.697283949341801e-05, "loss": 1.2501, "step": 1316 }, { "epoch": 0.28, "learning_rate": 1.6967954166959693e-05, "loss": 1.255, "step": 1317 }, { "epoch": 0.28, "learning_rate": 1.6963065605944668e-05, "loss": 1.2349, "step": 1318 }, { "epoch": 0.28, "learning_rate": 1.6958173812642226e-05, "loss": 1.2585, "step": 1319 }, { "epoch": 0.28, "learning_rate": 1.6953278789323158e-05, "loss": 1.2164, "step": 1320 }, { "epoch": 0.28, "learning_rate": 1.6948380538259765e-05, "loss": 1.2786, "step": 1321 }, { "epoch": 0.28, "learning_rate": 1.694347906172584e-05, "loss": 1.2828, "step": 1322 }, { "epoch": 0.28, "learning_rate": 1.6938574361996653e-05, "loss": 1.29, "step": 1323 }, { "epoch": 0.28, "learning_rate": 1.6933666441349008e-05, "loss": 1.2258, "step": 1324 }, { "epoch": 0.28, "learning_rate": 1.6928755302061175e-05, "loss": 1.2163, "step": 1325 }, { "epoch": 0.28, "learning_rate": 1.6923840946412933e-05, "loss": 1.2573, "step": 1326 }, { "epoch": 0.28, "learning_rate": 1.691892337668554e-05, "loss": 1.2674, "step": 1327 }, { "epoch": 0.28, "learning_rate": 1.6914002595161764e-05, "loss": 1.2678, "step": 1328 }, { "epoch": 0.28, "learning_rate": 1.690907860412585e-05, "loss": 1.2729, "step": 1329 }, { "epoch": 0.28, "learning_rate": 1.6904151405863537e-05, "loss": 1.2508, "step": 1330 }, { "epoch": 0.28, "learning_rate": 1.6899221002662055e-05, "loss": 1.273, "step": 1331 }, { "epoch": 0.28, "learning_rate": 1.689428739681012e-05, "loss": 1.2981, "step": 1332 }, { "epoch": 0.28, "learning_rate": 1.6889350590597935e-05, "loss": 1.2144, "step": 1333 }, { "epoch": 0.28, "learning_rate": 1.6884410586317186e-05, "loss": 1.2589, "step": 1334 }, { "epoch": 0.28, "learning_rate": 1.687946738626105e-05, "loss": 1.2036, "step": 1335 }, { "epoch": 0.28, "learning_rate": 1.687452099272418e-05, "loss": 1.2837, "step": 1336 }, { "epoch": 0.28, "learning_rate": 1.686957140800272e-05, "loss": 1.2803, "step": 1337 }, { "epoch": 0.28, "learning_rate": 1.686461863439428e-05, "loss": 1.2761, "step": 1338 }, { "epoch": 0.28, "learning_rate": 1.685966267419797e-05, "loss": 1.2613, "step": 1339 }, { "epoch": 0.28, "learning_rate": 1.685470352971437e-05, "loss": 1.23, "step": 1340 }, { "epoch": 0.28, "learning_rate": 1.684974120324553e-05, "loss": 1.246, "step": 1341 }, { "epoch": 0.28, "learning_rate": 1.684477569709499e-05, "loss": 1.2683, "step": 1342 }, { "epoch": 0.28, "learning_rate": 1.6839807013567764e-05, "loss": 1.2729, "step": 1343 }, { "epoch": 0.28, "learning_rate": 1.6834835154970332e-05, "loss": 1.2401, "step": 1344 }, { "epoch": 0.28, "learning_rate": 1.6829860123610656e-05, "loss": 1.2089, "step": 1345 }, { "epoch": 0.28, "learning_rate": 1.6824881921798167e-05, "loss": 1.2465, "step": 1346 }, { "epoch": 0.28, "learning_rate": 1.6819900551843776e-05, "loss": 1.2621, "step": 1347 }, { "epoch": 0.28, "learning_rate": 1.681491601605985e-05, "loss": 1.2737, "step": 1348 }, { "epoch": 0.28, "learning_rate": 1.680992831676024e-05, "loss": 1.2802, "step": 1349 }, { "epoch": 0.28, "learning_rate": 1.6804937456260254e-05, "loss": 1.2158, "step": 1350 }, { "epoch": 0.28, "learning_rate": 1.679994343687668e-05, "loss": 1.2689, "step": 1351 }, { "epoch": 0.28, "learning_rate": 1.6794946260927757e-05, "loss": 1.2522, "step": 1352 }, { "epoch": 0.28, "learning_rate": 1.67899459307332e-05, "loss": 1.2509, "step": 1353 }, { "epoch": 0.28, "learning_rate": 1.6784942448614186e-05, "loss": 1.2753, "step": 1354 }, { "epoch": 0.29, "learning_rate": 1.6779935816893354e-05, "loss": 1.1971, "step": 1355 }, { "epoch": 0.29, "learning_rate": 1.6774926037894806e-05, "loss": 1.2217, "step": 1356 }, { "epoch": 0.29, "learning_rate": 1.6769913113944108e-05, "loss": 1.2908, "step": 1357 }, { "epoch": 0.29, "learning_rate": 1.676489704736828e-05, "loss": 1.251, "step": 1358 }, { "epoch": 0.29, "learning_rate": 1.6759877840495797e-05, "loss": 1.2709, "step": 1359 }, { "epoch": 0.29, "learning_rate": 1.675485549565661e-05, "loss": 1.299, "step": 1360 }, { "epoch": 0.29, "learning_rate": 1.6749830015182106e-05, "loss": 1.2288, "step": 1361 }, { "epoch": 0.29, "learning_rate": 1.6744801401405138e-05, "loss": 1.2455, "step": 1362 }, { "epoch": 0.29, "learning_rate": 1.673976965666002e-05, "loss": 1.2311, "step": 1363 }, { "epoch": 0.29, "learning_rate": 1.67347347832825e-05, "loss": 1.249, "step": 1364 }, { "epoch": 0.29, "learning_rate": 1.6729696783609796e-05, "loss": 1.2567, "step": 1365 }, { "epoch": 0.29, "learning_rate": 1.6724655659980567e-05, "loss": 1.2421, "step": 1366 }, { "epoch": 0.29, "learning_rate": 1.6719611414734926e-05, "loss": 1.3094, "step": 1367 }, { "epoch": 0.29, "learning_rate": 1.671456405021444e-05, "loss": 1.2926, "step": 1368 }, { "epoch": 0.29, "learning_rate": 1.6709513568762118e-05, "loss": 1.2201, "step": 1369 }, { "epoch": 0.29, "learning_rate": 1.6704459972722415e-05, "loss": 1.2011, "step": 1370 }, { "epoch": 0.29, "learning_rate": 1.669940326444123e-05, "loss": 1.23, "step": 1371 }, { "epoch": 0.29, "learning_rate": 1.6694343446265916e-05, "loss": 1.2434, "step": 1372 }, { "epoch": 0.29, "learning_rate": 1.6689280520545265e-05, "loss": 1.2495, "step": 1373 }, { "epoch": 0.29, "learning_rate": 1.6684214489629503e-05, "loss": 1.2252, "step": 1374 }, { "epoch": 0.29, "learning_rate": 1.6679145355870316e-05, "loss": 1.2785, "step": 1375 }, { "epoch": 0.29, "learning_rate": 1.6674073121620815e-05, "loss": 1.2161, "step": 1376 }, { "epoch": 0.29, "learning_rate": 1.6668997789235547e-05, "loss": 1.2719, "step": 1377 }, { "epoch": 0.29, "learning_rate": 1.6663919361070515e-05, "loss": 1.274, "step": 1378 }, { "epoch": 0.29, "learning_rate": 1.6658837839483144e-05, "loss": 1.2558, "step": 1379 }, { "epoch": 0.29, "learning_rate": 1.6653753226832304e-05, "loss": 1.2392, "step": 1380 }, { "epoch": 0.29, "learning_rate": 1.6648665525478288e-05, "loss": 1.2799, "step": 1381 }, { "epoch": 0.29, "learning_rate": 1.664357473778284e-05, "loss": 1.2283, "step": 1382 }, { "epoch": 0.29, "learning_rate": 1.663848086610912e-05, "loss": 1.2416, "step": 1383 }, { "epoch": 0.29, "learning_rate": 1.6633383912821735e-05, "loss": 1.2828, "step": 1384 }, { "epoch": 0.29, "learning_rate": 1.6628283880286703e-05, "loss": 1.2026, "step": 1385 }, { "epoch": 0.29, "learning_rate": 1.6623180770871494e-05, "loss": 1.2705, "step": 1386 }, { "epoch": 0.29, "learning_rate": 1.6618074586944987e-05, "loss": 1.2461, "step": 1387 }, { "epoch": 0.29, "learning_rate": 1.6612965330877503e-05, "loss": 1.2935, "step": 1388 }, { "epoch": 0.29, "learning_rate": 1.6607853005040777e-05, "loss": 1.3039, "step": 1389 }, { "epoch": 0.29, "learning_rate": 1.6602737611807975e-05, "loss": 1.2625, "step": 1390 }, { "epoch": 0.29, "learning_rate": 1.6597619153553687e-05, "loss": 1.2747, "step": 1391 }, { "epoch": 0.29, "learning_rate": 1.659249763265393e-05, "loss": 1.2766, "step": 1392 }, { "epoch": 0.29, "learning_rate": 1.658737305148613e-05, "loss": 1.2494, "step": 1393 }, { "epoch": 0.29, "learning_rate": 1.6582245412429146e-05, "loss": 1.266, "step": 1394 }, { "epoch": 0.29, "learning_rate": 1.657711471786325e-05, "loss": 1.2863, "step": 1395 }, { "epoch": 0.29, "learning_rate": 1.6571980970170137e-05, "loss": 1.219, "step": 1396 }, { "epoch": 0.29, "learning_rate": 1.6566844171732914e-05, "loss": 1.2375, "step": 1397 }, { "epoch": 0.29, "learning_rate": 1.6561704324936105e-05, "loss": 1.2575, "step": 1398 }, { "epoch": 0.29, "learning_rate": 1.6556561432165656e-05, "loss": 1.2646, "step": 1399 }, { "epoch": 0.29, "learning_rate": 1.6551415495808916e-05, "loss": 1.2706, "step": 1400 }, { "epoch": 0.29, "learning_rate": 1.654626651825466e-05, "loss": 1.2773, "step": 1401 }, { "epoch": 0.29, "learning_rate": 1.6541114501893057e-05, "loss": 1.2834, "step": 1402 }, { "epoch": 0.3, "learning_rate": 1.6535959449115707e-05, "loss": 1.2107, "step": 1403 }, { "epoch": 0.3, "learning_rate": 1.6530801362315603e-05, "loss": 1.3115, "step": 1404 }, { "epoch": 0.3, "learning_rate": 1.6525640243887158e-05, "loss": 1.2406, "step": 1405 }, { "epoch": 0.3, "learning_rate": 1.652047609622618e-05, "loss": 1.2537, "step": 1406 }, { "epoch": 0.3, "learning_rate": 1.6515308921729896e-05, "loss": 1.2825, "step": 1407 }, { "epoch": 0.3, "learning_rate": 1.6510138722796935e-05, "loss": 1.2467, "step": 1408 }, { "epoch": 0.3, "learning_rate": 1.6504965501827315e-05, "loss": 1.2545, "step": 1409 }, { "epoch": 0.3, "learning_rate": 1.649978926122248e-05, "loss": 1.2785, "step": 1410 }, { "epoch": 0.3, "learning_rate": 1.6494610003385266e-05, "loss": 1.2241, "step": 1411 }, { "epoch": 0.3, "learning_rate": 1.6489427730719903e-05, "loss": 1.2285, "step": 1412 }, { "epoch": 0.3, "learning_rate": 1.648424244563203e-05, "loss": 1.2264, "step": 1413 }, { "epoch": 0.3, "learning_rate": 1.647905415052868e-05, "loss": 1.2404, "step": 1414 }, { "epoch": 0.3, "learning_rate": 1.647386284781828e-05, "loss": 1.2339, "step": 1415 }, { "epoch": 0.3, "learning_rate": 1.646866853991066e-05, "loss": 1.2535, "step": 1416 }, { "epoch": 0.3, "learning_rate": 1.6463471229217036e-05, "loss": 1.2622, "step": 1417 }, { "epoch": 0.3, "learning_rate": 1.6458270918150033e-05, "loss": 1.2926, "step": 1418 }, { "epoch": 0.3, "learning_rate": 1.6453067609123656e-05, "loss": 1.2969, "step": 1419 }, { "epoch": 0.3, "learning_rate": 1.64478613045533e-05, "loss": 1.2377, "step": 1420 }, { "epoch": 0.3, "learning_rate": 1.6442652006855762e-05, "loss": 1.2259, "step": 1421 }, { "epoch": 0.3, "learning_rate": 1.6437439718449222e-05, "loss": 1.2145, "step": 1422 }, { "epoch": 0.3, "learning_rate": 1.6432224441753246e-05, "loss": 1.2707, "step": 1423 }, { "epoch": 0.3, "learning_rate": 1.642700617918879e-05, "loss": 1.2338, "step": 1424 }, { "epoch": 0.3, "learning_rate": 1.64217849331782e-05, "loss": 1.2654, "step": 1425 }, { "epoch": 0.3, "learning_rate": 1.6416560706145193e-05, "loss": 1.2692, "step": 1426 }, { "epoch": 0.3, "learning_rate": 1.641133350051489e-05, "loss": 1.2389, "step": 1427 }, { "epoch": 0.3, "learning_rate": 1.640610331871378e-05, "loss": 1.2052, "step": 1428 }, { "epoch": 0.3, "learning_rate": 1.6400870163169735e-05, "loss": 1.269, "step": 1429 }, { "epoch": 0.3, "learning_rate": 1.6395634036312013e-05, "loss": 1.2535, "step": 1430 }, { "epoch": 0.3, "learning_rate": 1.639039494057125e-05, "loss": 1.2077, "step": 1431 }, { "epoch": 0.3, "learning_rate": 1.6385152878379454e-05, "loss": 1.2412, "step": 1432 }, { "epoch": 0.3, "learning_rate": 1.6379907852170017e-05, "loss": 1.2322, "step": 1433 }, { "epoch": 0.3, "learning_rate": 1.6374659864377707e-05, "loss": 1.227, "step": 1434 }, { "epoch": 0.3, "learning_rate": 1.636940891743866e-05, "loss": 1.2712, "step": 1435 }, { "epoch": 0.3, "learning_rate": 1.6364155013790393e-05, "loss": 1.2988, "step": 1436 }, { "epoch": 0.3, "learning_rate": 1.635889815587179e-05, "loss": 1.216, "step": 1437 }, { "epoch": 0.3, "learning_rate": 1.635363834612311e-05, "loss": 1.2946, "step": 1438 }, { "epoch": 0.3, "learning_rate": 1.634837558698598e-05, "loss": 1.272, "step": 1439 }, { "epoch": 0.3, "learning_rate": 1.6343109880903404e-05, "loss": 1.2329, "step": 1440 }, { "epoch": 0.3, "learning_rate": 1.633784123031974e-05, "loss": 1.1612, "step": 1441 }, { "epoch": 0.3, "learning_rate": 1.6332569637680723e-05, "loss": 1.263, "step": 1442 }, { "epoch": 0.3, "learning_rate": 1.632729510543345e-05, "loss": 1.2427, "step": 1443 }, { "epoch": 0.3, "learning_rate": 1.6322017636026378e-05, "loss": 1.2826, "step": 1444 }, { "epoch": 0.3, "learning_rate": 1.6316737231909344e-05, "loss": 1.2429, "step": 1445 }, { "epoch": 0.3, "learning_rate": 1.6311453895533528e-05, "loss": 1.2528, "step": 1446 }, { "epoch": 0.3, "learning_rate": 1.630616762935148e-05, "loss": 1.2631, "step": 1447 }, { "epoch": 0.3, "learning_rate": 1.6300878435817115e-05, "loss": 1.2791, "step": 1448 }, { "epoch": 0.3, "learning_rate": 1.629558631738569e-05, "loss": 1.2082, "step": 1449 }, { "epoch": 0.3, "learning_rate": 1.6290291276513842e-05, "loss": 1.2589, "step": 1450 }, { "epoch": 0.31, "learning_rate": 1.6284993315659552e-05, "loss": 1.1912, "step": 1451 }, { "epoch": 0.31, "learning_rate": 1.6279692437282152e-05, "loss": 1.2369, "step": 1452 }, { "epoch": 0.31, "learning_rate": 1.627438864384234e-05, "loss": 1.1832, "step": 1453 }, { "epoch": 0.31, "learning_rate": 1.6269081937802158e-05, "loss": 1.2487, "step": 1454 }, { "epoch": 0.31, "learning_rate": 1.6263772321625005e-05, "loss": 1.2471, "step": 1455 }, { "epoch": 0.31, "learning_rate": 1.625845979777563e-05, "loss": 1.2327, "step": 1456 }, { "epoch": 0.31, "learning_rate": 1.6253144368720132e-05, "loss": 1.2639, "step": 1457 }, { "epoch": 0.31, "learning_rate": 1.624782603692596e-05, "loss": 1.2734, "step": 1458 }, { "epoch": 0.31, "learning_rate": 1.62425048048619e-05, "loss": 1.2645, "step": 1459 }, { "epoch": 0.31, "learning_rate": 1.6237180674998102e-05, "loss": 1.3041, "step": 1460 }, { "epoch": 0.31, "learning_rate": 1.6231853649806043e-05, "loss": 1.2569, "step": 1461 }, { "epoch": 0.31, "learning_rate": 1.622652373175856e-05, "loss": 1.2553, "step": 1462 }, { "epoch": 0.31, "learning_rate": 1.6221190923329822e-05, "loss": 1.2494, "step": 1463 }, { "epoch": 0.31, "learning_rate": 1.6215855226995345e-05, "loss": 1.2854, "step": 1464 }, { "epoch": 0.31, "learning_rate": 1.6210516645231983e-05, "loss": 1.2296, "step": 1465 }, { "epoch": 0.31, "learning_rate": 1.6205175180517934e-05, "loss": 1.217, "step": 1466 }, { "epoch": 0.31, "learning_rate": 1.6199830835332724e-05, "loss": 1.2533, "step": 1467 }, { "epoch": 0.31, "learning_rate": 1.6194483612157232e-05, "loss": 1.2513, "step": 1468 }, { "epoch": 0.31, "learning_rate": 1.6189133513473657e-05, "loss": 1.2358, "step": 1469 }, { "epoch": 0.31, "learning_rate": 1.6183780541765537e-05, "loss": 1.2692, "step": 1470 }, { "epoch": 0.31, "learning_rate": 1.6178424699517755e-05, "loss": 1.2663, "step": 1471 }, { "epoch": 0.31, "learning_rate": 1.617306598921651e-05, "loss": 1.2572, "step": 1472 }, { "epoch": 0.31, "learning_rate": 1.6167704413349346e-05, "loss": 1.2325, "step": 1473 }, { "epoch": 0.31, "learning_rate": 1.616233997440513e-05, "loss": 1.2163, "step": 1474 }, { "epoch": 0.31, "learning_rate": 1.6156972674874058e-05, "loss": 1.2627, "step": 1475 }, { "epoch": 0.31, "learning_rate": 1.6151602517247654e-05, "loss": 1.2441, "step": 1476 }, { "epoch": 0.31, "learning_rate": 1.6146229504018777e-05, "loss": 1.2655, "step": 1477 }, { "epoch": 0.31, "learning_rate": 1.6140853637681597e-05, "loss": 1.27, "step": 1478 }, { "epoch": 0.31, "learning_rate": 1.6135474920731614e-05, "loss": 1.2519, "step": 1479 }, { "epoch": 0.31, "learning_rate": 1.613009335566567e-05, "loss": 1.2541, "step": 1480 }, { "epoch": 0.31, "learning_rate": 1.612470894498189e-05, "loss": 1.2338, "step": 1481 }, { "epoch": 0.31, "learning_rate": 1.6119321691179756e-05, "loss": 1.2686, "step": 1482 }, { "epoch": 0.31, "learning_rate": 1.6113931596760058e-05, "loss": 1.275, "step": 1483 }, { "epoch": 0.31, "learning_rate": 1.6108538664224895e-05, "loss": 1.2485, "step": 1484 }, { "epoch": 0.31, "learning_rate": 1.6103142896077696e-05, "loss": 1.2589, "step": 1485 }, { "epoch": 0.31, "learning_rate": 1.60977442948232e-05, "loss": 1.2842, "step": 1486 }, { "epoch": 0.31, "learning_rate": 1.6092342862967463e-05, "loss": 1.2481, "step": 1487 }, { "epoch": 0.31, "learning_rate": 1.6086938603017857e-05, "loss": 1.3273, "step": 1488 }, { "epoch": 0.31, "learning_rate": 1.6081531517483063e-05, "loss": 1.2042, "step": 1489 }, { "epoch": 0.31, "learning_rate": 1.607612160887307e-05, "loss": 1.2554, "step": 1490 }, { "epoch": 0.31, "learning_rate": 1.6070708879699193e-05, "loss": 1.2375, "step": 1491 }, { "epoch": 0.31, "learning_rate": 1.6065293332474036e-05, "loss": 1.226, "step": 1492 }, { "epoch": 0.31, "learning_rate": 1.6059874969711532e-05, "loss": 1.2551, "step": 1493 }, { "epoch": 0.31, "learning_rate": 1.6054453793926897e-05, "loss": 1.2782, "step": 1494 }, { "epoch": 0.31, "learning_rate": 1.6049029807636677e-05, "loss": 1.2441, "step": 1495 }, { "epoch": 0.31, "learning_rate": 1.6043603013358707e-05, "loss": 1.2221, "step": 1496 }, { "epoch": 0.31, "learning_rate": 1.6038173413612136e-05, "loss": 1.2725, "step": 1497 }, { "epoch": 0.32, "learning_rate": 1.6032741010917396e-05, "loss": 1.2347, "step": 1498 }, { "epoch": 0.32, "learning_rate": 1.602730580779625e-05, "loss": 1.2442, "step": 1499 }, { "epoch": 0.32, "learning_rate": 1.6021867806771726e-05, "loss": 1.2411, "step": 1500 }, { "epoch": 0.32, "learning_rate": 1.6016427010368187e-05, "loss": 1.3072, "step": 1501 }, { "epoch": 0.32, "learning_rate": 1.6010983421111266e-05, "loss": 1.2402, "step": 1502 }, { "epoch": 0.32, "learning_rate": 1.6005537041527907e-05, "loss": 1.2449, "step": 1503 }, { "epoch": 0.32, "learning_rate": 1.600008787414634e-05, "loss": 1.2593, "step": 1504 }, { "epoch": 0.32, "learning_rate": 1.59946359214961e-05, "loss": 1.2777, "step": 1505 }, { "epoch": 0.32, "learning_rate": 1.5989181186108003e-05, "loss": 1.2659, "step": 1506 }, { "epoch": 0.32, "learning_rate": 1.5983723670514165e-05, "loss": 1.2932, "step": 1507 }, { "epoch": 0.32, "learning_rate": 1.5978263377247995e-05, "loss": 1.2308, "step": 1508 }, { "epoch": 0.32, "learning_rate": 1.5972800308844175e-05, "loss": 1.2715, "step": 1509 }, { "epoch": 0.32, "learning_rate": 1.5967334467838694e-05, "loss": 1.2309, "step": 1510 }, { "epoch": 0.32, "learning_rate": 1.5961865856768827e-05, "loss": 1.2777, "step": 1511 }, { "epoch": 0.32, "learning_rate": 1.5956394478173113e-05, "loss": 1.2298, "step": 1512 }, { "epoch": 0.32, "learning_rate": 1.5950920334591405e-05, "loss": 1.2379, "step": 1513 }, { "epoch": 0.32, "learning_rate": 1.594544342856482e-05, "loss": 1.2928, "step": 1514 }, { "epoch": 0.32, "learning_rate": 1.5939963762635765e-05, "loss": 1.2734, "step": 1515 }, { "epoch": 0.32, "learning_rate": 1.5934481339347924e-05, "loss": 1.2677, "step": 1516 }, { "epoch": 0.32, "learning_rate": 1.5928996161246262e-05, "loss": 1.2631, "step": 1517 }, { "epoch": 0.32, "learning_rate": 1.5923508230877034e-05, "loss": 1.2628, "step": 1518 }, { "epoch": 0.32, "learning_rate": 1.5918017550787754e-05, "loss": 1.3288, "step": 1519 }, { "epoch": 0.32, "learning_rate": 1.5912524123527222e-05, "loss": 1.2647, "step": 1520 }, { "epoch": 0.32, "learning_rate": 1.590702795164551e-05, "loss": 1.1993, "step": 1521 }, { "epoch": 0.32, "learning_rate": 1.590152903769398e-05, "loss": 1.2999, "step": 1522 }, { "epoch": 0.32, "learning_rate": 1.5896027384225233e-05, "loss": 1.2348, "step": 1523 }, { "epoch": 0.32, "learning_rate": 1.589052299379318e-05, "loss": 1.2865, "step": 1524 }, { "epoch": 0.32, "learning_rate": 1.5885015868952973e-05, "loss": 1.2746, "step": 1525 }, { "epoch": 0.32, "learning_rate": 1.5879506012261052e-05, "loss": 1.2643, "step": 1526 }, { "epoch": 0.32, "learning_rate": 1.5873993426275113e-05, "loss": 1.3119, "step": 1527 }, { "epoch": 0.32, "learning_rate": 1.586847811355413e-05, "loss": 1.2822, "step": 1528 }, { "epoch": 0.32, "learning_rate": 1.586296007665833e-05, "loss": 1.2494, "step": 1529 }, { "epoch": 0.32, "learning_rate": 1.5857439318149216e-05, "loss": 1.2103, "step": 1530 }, { "epoch": 0.32, "learning_rate": 1.5851915840589548e-05, "loss": 1.2588, "step": 1531 }, { "epoch": 0.32, "learning_rate": 1.5846389646543347e-05, "loss": 1.2705, "step": 1532 }, { "epoch": 0.32, "learning_rate": 1.5840860738575904e-05, "loss": 1.2522, "step": 1533 }, { "epoch": 0.32, "learning_rate": 1.5835329119253762e-05, "loss": 1.2885, "step": 1534 }, { "epoch": 0.32, "learning_rate": 1.5829794791144723e-05, "loss": 1.2593, "step": 1535 }, { "epoch": 0.32, "learning_rate": 1.582425775681785e-05, "loss": 1.2769, "step": 1536 }, { "epoch": 0.32, "learning_rate": 1.581871801884345e-05, "loss": 1.2558, "step": 1537 }, { "epoch": 0.32, "learning_rate": 1.5813175579793112e-05, "loss": 1.2334, "step": 1538 }, { "epoch": 0.32, "learning_rate": 1.5807630442239652e-05, "loss": 1.2936, "step": 1539 }, { "epoch": 0.32, "learning_rate": 1.5802082608757148e-05, "loss": 1.2677, "step": 1540 }, { "epoch": 0.32, "learning_rate": 1.5796532081920935e-05, "loss": 1.2442, "step": 1541 }, { "epoch": 0.32, "learning_rate": 1.5790978864307586e-05, "loss": 1.2799, "step": 1542 }, { "epoch": 0.32, "learning_rate": 1.5785422958494938e-05, "loss": 1.2637, "step": 1543 }, { "epoch": 0.32, "learning_rate": 1.5779864367062062e-05, "loss": 1.3032, "step": 1544 }, { "epoch": 0.32, "learning_rate": 1.577430309258929e-05, "loss": 1.2482, "step": 1545 }, { "epoch": 0.33, "learning_rate": 1.5768739137658182e-05, "loss": 1.2338, "step": 1546 }, { "epoch": 0.33, "learning_rate": 1.576317250485156e-05, "loss": 1.2747, "step": 1547 }, { "epoch": 0.33, "learning_rate": 1.5757603196753476e-05, "loss": 1.2693, "step": 1548 }, { "epoch": 0.33, "learning_rate": 1.575203121594923e-05, "loss": 1.2868, "step": 1549 }, { "epoch": 0.33, "learning_rate": 1.5746456565025357e-05, "loss": 1.2612, "step": 1550 }, { "epoch": 0.33, "learning_rate": 1.5740879246569647e-05, "loss": 1.307, "step": 1551 }, { "epoch": 0.33, "learning_rate": 1.5735299263171106e-05, "loss": 1.2456, "step": 1552 }, { "epoch": 0.33, "learning_rate": 1.5729716617419994e-05, "loss": 1.2747, "step": 1553 }, { "epoch": 0.33, "learning_rate": 1.5724131311907796e-05, "loss": 1.2619, "step": 1554 }, { "epoch": 0.33, "learning_rate": 1.5718543349227245e-05, "loss": 1.2539, "step": 1555 }, { "epoch": 0.33, "learning_rate": 1.5712952731972295e-05, "loss": 1.2361, "step": 1556 }, { "epoch": 0.33, "learning_rate": 1.5707359462738134e-05, "loss": 1.2493, "step": 1557 }, { "epoch": 0.33, "learning_rate": 1.5701763544121183e-05, "loss": 1.2305, "step": 1558 }, { "epoch": 0.33, "learning_rate": 1.5696164978719102e-05, "loss": 1.2378, "step": 1559 }, { "epoch": 0.33, "learning_rate": 1.5690563769130762e-05, "loss": 1.2415, "step": 1560 }, { "epoch": 0.33, "learning_rate": 1.5684959917956274e-05, "loss": 1.2491, "step": 1561 }, { "epoch": 0.33, "learning_rate": 1.5679353427796967e-05, "loss": 1.2458, "step": 1562 }, { "epoch": 0.33, "learning_rate": 1.5673744301255412e-05, "loss": 1.2852, "step": 1563 }, { "epoch": 0.33, "learning_rate": 1.566813254093538e-05, "loss": 1.258, "step": 1564 }, { "epoch": 0.33, "learning_rate": 1.566251814944188e-05, "loss": 1.2084, "step": 1565 }, { "epoch": 0.33, "learning_rate": 1.565690112938114e-05, "loss": 1.2275, "step": 1566 }, { "epoch": 0.33, "learning_rate": 1.5651281483360603e-05, "loss": 1.2514, "step": 1567 }, { "epoch": 0.33, "learning_rate": 1.5645659213988938e-05, "loss": 1.212, "step": 1568 }, { "epoch": 0.33, "learning_rate": 1.5640034323876033e-05, "loss": 1.2446, "step": 1569 }, { "epoch": 0.33, "learning_rate": 1.5634406815632975e-05, "loss": 1.2682, "step": 1570 }, { "epoch": 0.33, "learning_rate": 1.5628776691872093e-05, "loss": 1.2614, "step": 1571 }, { "epoch": 0.33, "learning_rate": 1.5623143955206912e-05, "loss": 1.2206, "step": 1572 }, { "epoch": 0.33, "learning_rate": 1.561750860825217e-05, "loss": 1.2441, "step": 1573 }, { "epoch": 0.33, "learning_rate": 1.5611870653623826e-05, "loss": 1.2533, "step": 1574 }, { "epoch": 0.33, "learning_rate": 1.5606230093939042e-05, "loss": 1.2494, "step": 1575 }, { "epoch": 0.33, "learning_rate": 1.5600586931816198e-05, "loss": 1.2693, "step": 1576 }, { "epoch": 0.33, "learning_rate": 1.5594941169874867e-05, "loss": 1.2409, "step": 1577 }, { "epoch": 0.33, "learning_rate": 1.5589292810735845e-05, "loss": 1.2603, "step": 1578 }, { "epoch": 0.33, "learning_rate": 1.5583641857021126e-05, "loss": 1.2471, "step": 1579 }, { "epoch": 0.33, "learning_rate": 1.5577988311353905e-05, "loss": 1.2365, "step": 1580 }, { "epoch": 0.33, "learning_rate": 1.5572332176358586e-05, "loss": 1.2343, "step": 1581 }, { "epoch": 0.33, "learning_rate": 1.5566673454660775e-05, "loss": 1.2434, "step": 1582 }, { "epoch": 0.33, "learning_rate": 1.5561012148887272e-05, "loss": 1.2651, "step": 1583 }, { "epoch": 0.33, "learning_rate": 1.5555348261666094e-05, "loss": 1.2609, "step": 1584 }, { "epoch": 0.33, "learning_rate": 1.5549681795626422e-05, "loss": 1.2116, "step": 1585 }, { "epoch": 0.33, "learning_rate": 1.5544012753398678e-05, "loss": 1.2654, "step": 1586 }, { "epoch": 0.33, "learning_rate": 1.5538341137614447e-05, "loss": 1.2399, "step": 1587 }, { "epoch": 0.33, "learning_rate": 1.553266695090652e-05, "loss": 1.294, "step": 1588 }, { "epoch": 0.33, "learning_rate": 1.5526990195908876e-05, "loss": 1.2514, "step": 1589 }, { "epoch": 0.33, "learning_rate": 1.5521310875256702e-05, "loss": 1.2563, "step": 1590 }, { "epoch": 0.33, "learning_rate": 1.551562899158636e-05, "loss": 1.2182, "step": 1591 }, { "epoch": 0.33, "learning_rate": 1.5509944547535405e-05, "loss": 1.1996, "step": 1592 }, { "epoch": 0.34, "learning_rate": 1.5504257545742585e-05, "loss": 1.2138, "step": 1593 }, { "epoch": 0.34, "learning_rate": 1.549856798884783e-05, "loss": 1.2616, "step": 1594 }, { "epoch": 0.34, "learning_rate": 1.549287587949226e-05, "loss": 1.2444, "step": 1595 }, { "epoch": 0.34, "learning_rate": 1.548718122031818e-05, "loss": 1.221, "step": 1596 }, { "epoch": 0.34, "learning_rate": 1.5481484013969078e-05, "loss": 1.3119, "step": 1597 }, { "epoch": 0.34, "learning_rate": 1.5475784263089617e-05, "loss": 1.2591, "step": 1598 }, { "epoch": 0.34, "learning_rate": 1.5470081970325654e-05, "loss": 1.27, "step": 1599 }, { "epoch": 0.34, "learning_rate": 1.5464377138324222e-05, "loss": 1.2656, "step": 1600 }, { "epoch": 0.34, "learning_rate": 1.5458669769733524e-05, "loss": 1.2036, "step": 1601 }, { "epoch": 0.34, "learning_rate": 1.545295986720295e-05, "loss": 1.2596, "step": 1602 }, { "epoch": 0.34, "learning_rate": 1.5447247433383064e-05, "loss": 1.2785, "step": 1603 }, { "epoch": 0.34, "learning_rate": 1.54415324709256e-05, "loss": 1.2726, "step": 1604 }, { "epoch": 0.34, "learning_rate": 1.543581498248348e-05, "loss": 1.2612, "step": 1605 }, { "epoch": 0.34, "learning_rate": 1.543009497071078e-05, "loss": 1.2743, "step": 1606 }, { "epoch": 0.34, "learning_rate": 1.5424372438262756e-05, "loss": 1.2833, "step": 1607 }, { "epoch": 0.34, "learning_rate": 1.5418647387795838e-05, "loss": 1.2982, "step": 1608 }, { "epoch": 0.34, "learning_rate": 1.541291982196762e-05, "loss": 1.2564, "step": 1609 }, { "epoch": 0.34, "learning_rate": 1.5407189743436864e-05, "loss": 1.2584, "step": 1610 }, { "epoch": 0.34, "learning_rate": 1.5401457154863502e-05, "loss": 1.311, "step": 1611 }, { "epoch": 0.34, "learning_rate": 1.5395722058908623e-05, "loss": 1.2407, "step": 1612 }, { "epoch": 0.34, "learning_rate": 1.5389984458234487e-05, "loss": 1.2759, "step": 1613 }, { "epoch": 0.34, "learning_rate": 1.538424435550452e-05, "loss": 1.2559, "step": 1614 }, { "epoch": 0.34, "learning_rate": 1.5378501753383295e-05, "loss": 1.2685, "step": 1615 }, { "epoch": 0.34, "learning_rate": 1.537275665453656e-05, "loss": 1.2709, "step": 1616 }, { "epoch": 0.34, "learning_rate": 1.536700906163122e-05, "loss": 1.2468, "step": 1617 }, { "epoch": 0.34, "learning_rate": 1.5361258977335325e-05, "loss": 1.2547, "step": 1618 }, { "epoch": 0.34, "learning_rate": 1.5355506404318105e-05, "loss": 1.2151, "step": 1619 }, { "epoch": 0.34, "learning_rate": 1.5349751345249914e-05, "loss": 1.2507, "step": 1620 }, { "epoch": 0.34, "learning_rate": 1.534399380280229e-05, "loss": 1.2683, "step": 1621 }, { "epoch": 0.34, "learning_rate": 1.533823377964791e-05, "loss": 1.2165, "step": 1622 }, { "epoch": 0.34, "learning_rate": 1.5332471278460594e-05, "loss": 1.2187, "step": 1623 }, { "epoch": 0.34, "learning_rate": 1.5326706301915333e-05, "loss": 1.2636, "step": 1624 }, { "epoch": 0.34, "learning_rate": 1.532093885268825e-05, "loss": 1.2566, "step": 1625 }, { "epoch": 0.34, "learning_rate": 1.5315168933456618e-05, "loss": 1.2491, "step": 1626 }, { "epoch": 0.34, "learning_rate": 1.530939654689887e-05, "loss": 1.2175, "step": 1627 }, { "epoch": 0.34, "learning_rate": 1.5303621695694564e-05, "loss": 1.2655, "step": 1628 }, { "epoch": 0.34, "learning_rate": 1.5297844382524427e-05, "loss": 1.2517, "step": 1629 }, { "epoch": 0.34, "learning_rate": 1.5292064610070304e-05, "loss": 1.2816, "step": 1630 }, { "epoch": 0.34, "learning_rate": 1.528628238101519e-05, "loss": 1.2392, "step": 1631 }, { "epoch": 0.34, "learning_rate": 1.528049769804323e-05, "loss": 1.2679, "step": 1632 }, { "epoch": 0.34, "learning_rate": 1.5274710563839698e-05, "loss": 1.292, "step": 1633 }, { "epoch": 0.34, "learning_rate": 1.5268920981091005e-05, "loss": 1.2655, "step": 1634 }, { "epoch": 0.34, "learning_rate": 1.526312895248471e-05, "loss": 1.2652, "step": 1635 }, { "epoch": 0.34, "learning_rate": 1.5257334480709492e-05, "loss": 1.1762, "step": 1636 }, { "epoch": 0.34, "learning_rate": 1.5251537568455176e-05, "loss": 1.2288, "step": 1637 }, { "epoch": 0.34, "learning_rate": 1.5245738218412713e-05, "loss": 1.2782, "step": 1638 }, { "epoch": 0.34, "learning_rate": 1.523993643327419e-05, "loss": 1.2249, "step": 1639 }, { "epoch": 0.34, "learning_rate": 1.5234132215732822e-05, "loss": 1.2799, "step": 1640 }, { "epoch": 0.35, "learning_rate": 1.5228325568482955e-05, "loss": 1.2975, "step": 1641 }, { "epoch": 0.35, "learning_rate": 1.5222516494220063e-05, "loss": 1.2482, "step": 1642 }, { "epoch": 0.35, "learning_rate": 1.5216704995640741e-05, "loss": 1.2409, "step": 1643 }, { "epoch": 0.35, "learning_rate": 1.5210891075442715e-05, "loss": 1.2203, "step": 1644 }, { "epoch": 0.35, "learning_rate": 1.5205074736324835e-05, "loss": 1.1835, "step": 1645 }, { "epoch": 0.35, "learning_rate": 1.5199255980987072e-05, "loss": 1.1603, "step": 1646 }, { "epoch": 0.35, "learning_rate": 1.5193434812130518e-05, "loss": 1.237, "step": 1647 }, { "epoch": 0.35, "learning_rate": 1.5187611232457391e-05, "loss": 1.2292, "step": 1648 }, { "epoch": 0.35, "learning_rate": 1.5181785244671016e-05, "loss": 1.2478, "step": 1649 }, { "epoch": 0.35, "learning_rate": 1.517595685147585e-05, "loss": 1.2448, "step": 1650 }, { "epoch": 0.35, "learning_rate": 1.517012605557746e-05, "loss": 1.2183, "step": 1651 }, { "epoch": 0.35, "learning_rate": 1.5164292859682528e-05, "loss": 1.2517, "step": 1652 }, { "epoch": 0.35, "learning_rate": 1.5158457266498847e-05, "loss": 1.2552, "step": 1653 }, { "epoch": 0.35, "learning_rate": 1.515261927873533e-05, "loss": 1.27, "step": 1654 }, { "epoch": 0.35, "learning_rate": 1.5146778899101999e-05, "loss": 1.2638, "step": 1655 }, { "epoch": 0.35, "learning_rate": 1.5140936130309982e-05, "loss": 1.2703, "step": 1656 }, { "epoch": 0.35, "learning_rate": 1.5135090975071523e-05, "loss": 1.2569, "step": 1657 }, { "epoch": 0.35, "learning_rate": 1.5129243436099965e-05, "loss": 1.2536, "step": 1658 }, { "epoch": 0.35, "learning_rate": 1.5123393516109768e-05, "loss": 1.2933, "step": 1659 }, { "epoch": 0.35, "learning_rate": 1.5117541217816492e-05, "loss": 1.2905, "step": 1660 }, { "epoch": 0.35, "learning_rate": 1.5111686543936794e-05, "loss": 1.2573, "step": 1661 }, { "epoch": 0.35, "learning_rate": 1.510582949718845e-05, "loss": 1.253, "step": 1662 }, { "epoch": 0.35, "learning_rate": 1.509997008029032e-05, "loss": 1.2278, "step": 1663 }, { "epoch": 0.35, "learning_rate": 1.509410829596238e-05, "loss": 1.2558, "step": 1664 }, { "epoch": 0.35, "learning_rate": 1.5088244146925694e-05, "loss": 1.2534, "step": 1665 }, { "epoch": 0.35, "learning_rate": 1.5082377635902428e-05, "loss": 1.2599, "step": 1666 }, { "epoch": 0.35, "learning_rate": 1.5076508765615845e-05, "loss": 1.2759, "step": 1667 }, { "epoch": 0.35, "learning_rate": 1.5070637538790296e-05, "loss": 1.204, "step": 1668 }, { "epoch": 0.35, "learning_rate": 1.5064763958151241e-05, "loss": 1.2795, "step": 1669 }, { "epoch": 0.35, "learning_rate": 1.5058888026425213e-05, "loss": 1.2417, "step": 1670 }, { "epoch": 0.35, "learning_rate": 1.5053009746339857e-05, "loss": 1.2141, "step": 1671 }, { "epoch": 0.35, "learning_rate": 1.504712912062389e-05, "loss": 1.2209, "step": 1672 }, { "epoch": 0.35, "learning_rate": 1.504124615200713e-05, "loss": 1.2436, "step": 1673 }, { "epoch": 0.35, "learning_rate": 1.5035360843220476e-05, "loss": 1.2949, "step": 1674 }, { "epoch": 0.35, "learning_rate": 1.5029473196995916e-05, "loss": 1.2487, "step": 1675 }, { "epoch": 0.35, "learning_rate": 1.5023583216066526e-05, "loss": 1.2288, "step": 1676 }, { "epoch": 0.35, "learning_rate": 1.501769090316646e-05, "loss": 1.2692, "step": 1677 }, { "epoch": 0.35, "learning_rate": 1.5011796261030958e-05, "loss": 1.2302, "step": 1678 }, { "epoch": 0.35, "learning_rate": 1.5005899292396336e-05, "loss": 1.2644, "step": 1679 }, { "epoch": 0.35, "learning_rate": 1.5000000000000002e-05, "loss": 1.255, "step": 1680 }, { "epoch": 0.35, "learning_rate": 1.4994098386580428e-05, "loss": 1.2674, "step": 1681 }, { "epoch": 0.35, "learning_rate": 1.4988194454877174e-05, "loss": 1.2183, "step": 1682 }, { "epoch": 0.35, "learning_rate": 1.498228820763087e-05, "loss": 1.21, "step": 1683 }, { "epoch": 0.35, "learning_rate": 1.497637964758323e-05, "loss": 1.2599, "step": 1684 }, { "epoch": 0.35, "learning_rate": 1.4970468777477027e-05, "loss": 1.2421, "step": 1685 }, { "epoch": 0.35, "learning_rate": 1.4964555600056116e-05, "loss": 1.2388, "step": 1686 }, { "epoch": 0.35, "learning_rate": 1.4958640118065424e-05, "loss": 1.2052, "step": 1687 }, { "epoch": 0.36, "learning_rate": 1.4952722334250944e-05, "loss": 1.2119, "step": 1688 }, { "epoch": 0.36, "learning_rate": 1.4946802251359737e-05, "loss": 1.264, "step": 1689 }, { "epoch": 0.36, "learning_rate": 1.4940879872139938e-05, "loss": 1.2695, "step": 1690 }, { "epoch": 0.36, "learning_rate": 1.493495519934074e-05, "loss": 1.2508, "step": 1691 }, { "epoch": 0.36, "learning_rate": 1.4929028235712402e-05, "loss": 1.2702, "step": 1692 }, { "epoch": 0.36, "learning_rate": 1.492309898400625e-05, "loss": 1.2547, "step": 1693 }, { "epoch": 0.36, "learning_rate": 1.491716744697467e-05, "loss": 1.294, "step": 1694 }, { "epoch": 0.36, "learning_rate": 1.4911233627371106e-05, "loss": 1.27, "step": 1695 }, { "epoch": 0.36, "learning_rate": 1.4905297527950074e-05, "loss": 1.2038, "step": 1696 }, { "epoch": 0.36, "learning_rate": 1.4899359151467127e-05, "loss": 1.2721, "step": 1697 }, { "epoch": 0.36, "learning_rate": 1.4893418500678891e-05, "loss": 1.2426, "step": 1698 }, { "epoch": 0.36, "learning_rate": 1.4887475578343047e-05, "loss": 1.2935, "step": 1699 }, { "epoch": 0.36, "learning_rate": 1.4881530387218325e-05, "loss": 1.237, "step": 1700 }, { "epoch": 0.36, "learning_rate": 1.4875582930064512e-05, "loss": 1.247, "step": 1701 }, { "epoch": 0.36, "learning_rate": 1.4869633209642444e-05, "loss": 1.2663, "step": 1702 }, { "epoch": 0.36, "learning_rate": 1.4863681228714006e-05, "loss": 1.2548, "step": 1703 }, { "epoch": 0.36, "learning_rate": 1.485772699004214e-05, "loss": 1.2439, "step": 1704 }, { "epoch": 0.36, "learning_rate": 1.485177049639083e-05, "loss": 1.2155, "step": 1705 }, { "epoch": 0.36, "learning_rate": 1.4845811750525106e-05, "loss": 1.2187, "step": 1706 }, { "epoch": 0.36, "learning_rate": 1.4839850755211045e-05, "loss": 1.2432, "step": 1707 }, { "epoch": 0.36, "learning_rate": 1.4833887513215772e-05, "loss": 1.2807, "step": 1708 }, { "epoch": 0.36, "learning_rate": 1.482792202730745e-05, "loss": 1.2508, "step": 1709 }, { "epoch": 0.36, "learning_rate": 1.4821954300255286e-05, "loss": 1.2617, "step": 1710 }, { "epoch": 0.36, "learning_rate": 1.4815984334829523e-05, "loss": 1.2807, "step": 1711 }, { "epoch": 0.36, "learning_rate": 1.4810012133801453e-05, "loss": 1.2533, "step": 1712 }, { "epoch": 0.36, "learning_rate": 1.4804037699943396e-05, "loss": 1.2715, "step": 1713 }, { "epoch": 0.36, "learning_rate": 1.4798061036028712e-05, "loss": 1.2169, "step": 1714 }, { "epoch": 0.36, "learning_rate": 1.4792082144831795e-05, "loss": 1.247, "step": 1715 }, { "epoch": 0.36, "learning_rate": 1.4786101029128076e-05, "loss": 1.1844, "step": 1716 }, { "epoch": 0.36, "learning_rate": 1.4780117691694016e-05, "loss": 1.2773, "step": 1717 }, { "epoch": 0.36, "learning_rate": 1.4774132135307109e-05, "loss": 1.2292, "step": 1718 }, { "epoch": 0.36, "learning_rate": 1.4768144362745875e-05, "loss": 1.2334, "step": 1719 }, { "epoch": 0.36, "learning_rate": 1.476215437678987e-05, "loss": 1.2155, "step": 1720 }, { "epoch": 0.36, "learning_rate": 1.4756162180219674e-05, "loss": 1.2399, "step": 1721 }, { "epoch": 0.36, "learning_rate": 1.4750167775816887e-05, "loss": 1.2648, "step": 1722 }, { "epoch": 0.36, "learning_rate": 1.4744171166364138e-05, "loss": 1.2458, "step": 1723 }, { "epoch": 0.36, "learning_rate": 1.4738172354645088e-05, "loss": 1.2182, "step": 1724 }, { "epoch": 0.36, "learning_rate": 1.4732171343444413e-05, "loss": 1.2636, "step": 1725 }, { "epoch": 0.36, "learning_rate": 1.4726168135547807e-05, "loss": 1.2282, "step": 1726 }, { "epoch": 0.36, "learning_rate": 1.4720162733741987e-05, "loss": 1.2849, "step": 1727 }, { "epoch": 0.36, "learning_rate": 1.471415514081469e-05, "loss": 1.263, "step": 1728 }, { "epoch": 0.36, "learning_rate": 1.4708145359554667e-05, "loss": 1.2865, "step": 1729 }, { "epoch": 0.36, "learning_rate": 1.4702133392751688e-05, "loss": 1.2538, "step": 1730 }, { "epoch": 0.36, "learning_rate": 1.4696119243196534e-05, "loss": 1.2507, "step": 1731 }, { "epoch": 0.36, "learning_rate": 1.4690102913681004e-05, "loss": 1.2285, "step": 1732 }, { "epoch": 0.36, "learning_rate": 1.4684084406997903e-05, "loss": 1.2459, "step": 1733 }, { "epoch": 0.36, "learning_rate": 1.4678063725941052e-05, "loss": 1.2444, "step": 1734 }, { "epoch": 0.36, "learning_rate": 1.4672040873305276e-05, "loss": 1.2062, "step": 1735 }, { "epoch": 0.37, "learning_rate": 1.4666015851886415e-05, "loss": 1.1922, "step": 1736 }, { "epoch": 0.37, "learning_rate": 1.465998866448131e-05, "loss": 1.2306, "step": 1737 }, { "epoch": 0.37, "learning_rate": 1.4653959313887813e-05, "loss": 1.2466, "step": 1738 }, { "epoch": 0.37, "learning_rate": 1.4647927802904772e-05, "loss": 1.2147, "step": 1739 }, { "epoch": 0.37, "learning_rate": 1.4641894134332042e-05, "loss": 1.2471, "step": 1740 }, { "epoch": 0.37, "learning_rate": 1.4635858310970483e-05, "loss": 1.2532, "step": 1741 }, { "epoch": 0.37, "learning_rate": 1.4629820335621952e-05, "loss": 1.2784, "step": 1742 }, { "epoch": 0.37, "learning_rate": 1.4623780211089302e-05, "loss": 1.2444, "step": 1743 }, { "epoch": 0.37, "learning_rate": 1.461773794017639e-05, "loss": 1.2384, "step": 1744 }, { "epoch": 0.37, "learning_rate": 1.4611693525688066e-05, "loss": 1.2477, "step": 1745 }, { "epoch": 0.37, "learning_rate": 1.4605646970430174e-05, "loss": 1.3044, "step": 1746 }, { "epoch": 0.37, "learning_rate": 1.4599598277209555e-05, "loss": 1.2452, "step": 1747 }, { "epoch": 0.37, "learning_rate": 1.4593547448834038e-05, "loss": 1.2208, "step": 1748 }, { "epoch": 0.37, "learning_rate": 1.4587494488112449e-05, "loss": 1.1797, "step": 1749 }, { "epoch": 0.37, "learning_rate": 1.4581439397854595e-05, "loss": 1.2356, "step": 1750 }, { "epoch": 0.37, "learning_rate": 1.4575382180871282e-05, "loss": 1.2614, "step": 1751 }, { "epoch": 0.37, "learning_rate": 1.4569322839974298e-05, "loss": 1.2626, "step": 1752 }, { "epoch": 0.37, "learning_rate": 1.4563261377976415e-05, "loss": 1.2332, "step": 1753 }, { "epoch": 0.37, "learning_rate": 1.4557197797691393e-05, "loss": 1.2343, "step": 1754 }, { "epoch": 0.37, "learning_rate": 1.4551132101933974e-05, "loss": 1.2386, "step": 1755 }, { "epoch": 0.37, "learning_rate": 1.4545064293519882e-05, "loss": 1.2667, "step": 1756 }, { "epoch": 0.37, "learning_rate": 1.4538994375265822e-05, "loss": 1.2238, "step": 1757 }, { "epoch": 0.37, "learning_rate": 1.4532922349989478e-05, "loss": 1.2769, "step": 1758 }, { "epoch": 0.37, "learning_rate": 1.4526848220509511e-05, "loss": 1.2391, "step": 1759 }, { "epoch": 0.37, "learning_rate": 1.4520771989645563e-05, "loss": 1.203, "step": 1760 }, { "epoch": 0.37, "learning_rate": 1.451469366021825e-05, "loss": 1.2125, "step": 1761 }, { "epoch": 0.37, "learning_rate": 1.4508613235049153e-05, "loss": 1.2386, "step": 1762 }, { "epoch": 0.37, "learning_rate": 1.4502530716960843e-05, "loss": 1.2237, "step": 1763 }, { "epoch": 0.37, "learning_rate": 1.4496446108776847e-05, "loss": 1.2547, "step": 1764 }, { "epoch": 0.37, "learning_rate": 1.449035941332167e-05, "loss": 1.2116, "step": 1765 }, { "epoch": 0.37, "learning_rate": 1.4484270633420785e-05, "loss": 1.2443, "step": 1766 }, { "epoch": 0.37, "learning_rate": 1.4478179771900634e-05, "loss": 1.2057, "step": 1767 }, { "epoch": 0.37, "learning_rate": 1.447208683158862e-05, "loss": 1.2369, "step": 1768 }, { "epoch": 0.37, "learning_rate": 1.4465991815313109e-05, "loss": 1.2307, "step": 1769 }, { "epoch": 0.37, "learning_rate": 1.4459894725903447e-05, "loss": 1.2445, "step": 1770 }, { "epoch": 0.37, "learning_rate": 1.4453795566189926e-05, "loss": 1.2061, "step": 1771 }, { "epoch": 0.37, "learning_rate": 1.4447694339003801e-05, "loss": 1.2839, "step": 1772 }, { "epoch": 0.37, "learning_rate": 1.4441591047177297e-05, "loss": 1.1503, "step": 1773 }, { "epoch": 0.37, "learning_rate": 1.4435485693543586e-05, "loss": 1.2169, "step": 1774 }, { "epoch": 0.37, "learning_rate": 1.4429378280936803e-05, "loss": 1.2364, "step": 1775 }, { "epoch": 0.37, "learning_rate": 1.4423268812192035e-05, "loss": 1.2413, "step": 1776 }, { "epoch": 0.37, "learning_rate": 1.4417157290145328e-05, "loss": 1.2562, "step": 1777 }, { "epoch": 0.37, "learning_rate": 1.441104371763368e-05, "loss": 1.1979, "step": 1778 }, { "epoch": 0.37, "learning_rate": 1.4404928097495034e-05, "loss": 1.2807, "step": 1779 }, { "epoch": 0.37, "learning_rate": 1.4398810432568297e-05, "loss": 1.2461, "step": 1780 }, { "epoch": 0.37, "learning_rate": 1.4392690725693314e-05, "loss": 1.2702, "step": 1781 }, { "epoch": 0.37, "learning_rate": 1.438656897971088e-05, "loss": 1.2565, "step": 1782 }, { "epoch": 0.38, "learning_rate": 1.4380445197462738e-05, "loss": 1.1919, "step": 1783 }, { "epoch": 0.38, "learning_rate": 1.4374319381791579e-05, "loss": 1.3088, "step": 1784 }, { "epoch": 0.38, "learning_rate": 1.4368191535541032e-05, "loss": 1.2846, "step": 1785 }, { "epoch": 0.38, "learning_rate": 1.4362061661555675e-05, "loss": 1.2547, "step": 1786 }, { "epoch": 0.38, "learning_rate": 1.4355929762681022e-05, "loss": 1.2323, "step": 1787 }, { "epoch": 0.38, "learning_rate": 1.4349795841763529e-05, "loss": 1.2974, "step": 1788 }, { "epoch": 0.38, "learning_rate": 1.4343659901650591e-05, "loss": 1.1894, "step": 1789 }, { "epoch": 0.38, "learning_rate": 1.433752194519054e-05, "loss": 1.2398, "step": 1790 }, { "epoch": 0.38, "learning_rate": 1.4331381975232642e-05, "loss": 1.2934, "step": 1791 }, { "epoch": 0.38, "learning_rate": 1.4325239994627103e-05, "loss": 1.2278, "step": 1792 }, { "epoch": 0.38, "learning_rate": 1.4319096006225057e-05, "loss": 1.2241, "step": 1793 }, { "epoch": 0.38, "learning_rate": 1.431295001287857e-05, "loss": 1.2583, "step": 1794 }, { "epoch": 0.38, "learning_rate": 1.4306802017440644e-05, "loss": 1.253, "step": 1795 }, { "epoch": 0.38, "learning_rate": 1.4300652022765207e-05, "loss": 1.2403, "step": 1796 }, { "epoch": 0.38, "learning_rate": 1.429450003170711e-05, "loss": 1.2345, "step": 1797 }, { "epoch": 0.38, "learning_rate": 1.4288346047122143e-05, "loss": 1.2438, "step": 1798 }, { "epoch": 0.38, "learning_rate": 1.428219007186701e-05, "loss": 1.2347, "step": 1799 }, { "epoch": 0.38, "learning_rate": 1.4276032108799349e-05, "loss": 1.2415, "step": 1800 }, { "epoch": 0.38, "learning_rate": 1.4269872160777704e-05, "loss": 1.2681, "step": 1801 }, { "epoch": 0.38, "learning_rate": 1.426371023066156e-05, "loss": 1.1872, "step": 1802 }, { "epoch": 0.38, "learning_rate": 1.4257546321311313e-05, "loss": 1.24, "step": 1803 }, { "epoch": 0.38, "learning_rate": 1.4251380435588273e-05, "loss": 1.2413, "step": 1804 }, { "epoch": 0.38, "learning_rate": 1.4245212576354683e-05, "loss": 1.3144, "step": 1805 }, { "epoch": 0.38, "learning_rate": 1.423904274647368e-05, "loss": 1.2228, "step": 1806 }, { "epoch": 0.38, "learning_rate": 1.4232870948809336e-05, "loss": 1.2641, "step": 1807 }, { "epoch": 0.38, "learning_rate": 1.4226697186226626e-05, "loss": 1.2877, "step": 1808 }, { "epoch": 0.38, "learning_rate": 1.4220521461591436e-05, "loss": 1.2061, "step": 1809 }, { "epoch": 0.38, "learning_rate": 1.421434377777057e-05, "loss": 1.2323, "step": 1810 }, { "epoch": 0.38, "learning_rate": 1.4208164137631736e-05, "loss": 1.2508, "step": 1811 }, { "epoch": 0.38, "learning_rate": 1.4201982544043553e-05, "loss": 1.2613, "step": 1812 }, { "epoch": 0.38, "learning_rate": 1.4195798999875545e-05, "loss": 1.2688, "step": 1813 }, { "epoch": 0.38, "learning_rate": 1.4189613507998142e-05, "loss": 1.23, "step": 1814 }, { "epoch": 0.38, "learning_rate": 1.4183426071282676e-05, "loss": 1.2387, "step": 1815 }, { "epoch": 0.38, "learning_rate": 1.4177236692601386e-05, "loss": 1.2357, "step": 1816 }, { "epoch": 0.38, "learning_rate": 1.417104537482741e-05, "loss": 1.2378, "step": 1817 }, { "epoch": 0.38, "learning_rate": 1.4164852120834787e-05, "loss": 1.2462, "step": 1818 }, { "epoch": 0.38, "learning_rate": 1.4158656933498448e-05, "loss": 1.2146, "step": 1819 }, { "epoch": 0.38, "learning_rate": 1.415245981569424e-05, "loss": 1.2116, "step": 1820 }, { "epoch": 0.38, "learning_rate": 1.4146260770298883e-05, "loss": 1.1839, "step": 1821 }, { "epoch": 0.38, "learning_rate": 1.414005980019001e-05, "loss": 1.2069, "step": 1822 }, { "epoch": 0.38, "learning_rate": 1.4133856908246135e-05, "loss": 1.2156, "step": 1823 }, { "epoch": 0.38, "learning_rate": 1.4127652097346672e-05, "loss": 1.245, "step": 1824 }, { "epoch": 0.38, "learning_rate": 1.4121445370371922e-05, "loss": 1.2551, "step": 1825 }, { "epoch": 0.38, "learning_rate": 1.4115236730203077e-05, "loss": 1.2876, "step": 1826 }, { "epoch": 0.38, "learning_rate": 1.4109026179722217e-05, "loss": 1.274, "step": 1827 }, { "epoch": 0.38, "learning_rate": 1.4102813721812308e-05, "loss": 1.2782, "step": 1828 }, { "epoch": 0.38, "learning_rate": 1.4096599359357201e-05, "loss": 1.2382, "step": 1829 }, { "epoch": 0.38, "learning_rate": 1.4090383095241631e-05, "loss": 1.2373, "step": 1830 }, { "epoch": 0.39, "learning_rate": 1.4084164932351221e-05, "loss": 1.3111, "step": 1831 }, { "epoch": 0.39, "learning_rate": 1.4077944873572466e-05, "loss": 1.2335, "step": 1832 }, { "epoch": 0.39, "learning_rate": 1.4071722921792749e-05, "loss": 1.2561, "step": 1833 }, { "epoch": 0.39, "learning_rate": 1.4065499079900332e-05, "loss": 1.2657, "step": 1834 }, { "epoch": 0.39, "learning_rate": 1.4059273350784344e-05, "loss": 1.2203, "step": 1835 }, { "epoch": 0.39, "learning_rate": 1.40530457373348e-05, "loss": 1.2464, "step": 1836 }, { "epoch": 0.39, "learning_rate": 1.4046816242442591e-05, "loss": 1.194, "step": 1837 }, { "epoch": 0.39, "learning_rate": 1.4040584868999478e-05, "loss": 1.2771, "step": 1838 }, { "epoch": 0.39, "learning_rate": 1.4034351619898088e-05, "loss": 1.2005, "step": 1839 }, { "epoch": 0.39, "learning_rate": 1.4028116498031925e-05, "loss": 1.2145, "step": 1840 }, { "epoch": 0.39, "learning_rate": 1.4021879506295362e-05, "loss": 1.2711, "step": 1841 }, { "epoch": 0.39, "learning_rate": 1.4015640647583645e-05, "loss": 1.223, "step": 1842 }, { "epoch": 0.39, "learning_rate": 1.4009399924792875e-05, "loss": 1.2326, "step": 1843 }, { "epoch": 0.39, "learning_rate": 1.4003157340820024e-05, "loss": 1.2666, "step": 1844 }, { "epoch": 0.39, "learning_rate": 1.3996912898562933e-05, "loss": 1.1871, "step": 1845 }, { "epoch": 0.39, "learning_rate": 1.3990666600920298e-05, "loss": 1.2284, "step": 1846 }, { "epoch": 0.39, "learning_rate": 1.3984418450791682e-05, "loss": 1.2562, "step": 1847 }, { "epoch": 0.39, "learning_rate": 1.3978168451077501e-05, "loss": 1.2017, "step": 1848 }, { "epoch": 0.39, "learning_rate": 1.3971916604679038e-05, "loss": 1.2425, "step": 1849 }, { "epoch": 0.39, "learning_rate": 1.396566291449843e-05, "loss": 1.2801, "step": 1850 }, { "epoch": 0.39, "learning_rate": 1.3959407383438666e-05, "loss": 1.2141, "step": 1851 }, { "epoch": 0.39, "learning_rate": 1.395315001440359e-05, "loss": 1.2026, "step": 1852 }, { "epoch": 0.39, "learning_rate": 1.3946890810297909e-05, "loss": 1.2526, "step": 1853 }, { "epoch": 0.39, "learning_rate": 1.394062977402717e-05, "loss": 1.2376, "step": 1854 }, { "epoch": 0.39, "learning_rate": 1.3934366908497776e-05, "loss": 1.2552, "step": 1855 }, { "epoch": 0.39, "learning_rate": 1.3928102216616977e-05, "loss": 1.1996, "step": 1856 }, { "epoch": 0.39, "learning_rate": 1.3921835701292874e-05, "loss": 1.2207, "step": 1857 }, { "epoch": 0.39, "learning_rate": 1.391556736543441e-05, "loss": 1.2283, "step": 1858 }, { "epoch": 0.39, "learning_rate": 1.390929721195138e-05, "loss": 1.1814, "step": 1859 }, { "epoch": 0.39, "learning_rate": 1.3903025243754416e-05, "loss": 1.2625, "step": 1860 }, { "epoch": 0.39, "learning_rate": 1.3896751463754994e-05, "loss": 1.2531, "step": 1861 }, { "epoch": 0.39, "learning_rate": 1.3890475874865432e-05, "loss": 1.3048, "step": 1862 }, { "epoch": 0.39, "learning_rate": 1.3884198479998892e-05, "loss": 1.262, "step": 1863 }, { "epoch": 0.39, "learning_rate": 1.3877919282069361e-05, "loss": 1.216, "step": 1864 }, { "epoch": 0.39, "learning_rate": 1.3871638283991677e-05, "loss": 1.2087, "step": 1865 }, { "epoch": 0.39, "learning_rate": 1.3865355488681511e-05, "loss": 1.2767, "step": 1866 }, { "epoch": 0.39, "learning_rate": 1.3859070899055365e-05, "loss": 1.2435, "step": 1867 }, { "epoch": 0.39, "learning_rate": 1.3852784518030568e-05, "loss": 1.2363, "step": 1868 }, { "epoch": 0.39, "learning_rate": 1.3846496348525294e-05, "loss": 1.2685, "step": 1869 }, { "epoch": 0.39, "learning_rate": 1.3840206393458537e-05, "loss": 1.218, "step": 1870 }, { "epoch": 0.39, "learning_rate": 1.3833914655750127e-05, "loss": 1.2248, "step": 1871 }, { "epoch": 0.39, "learning_rate": 1.3827621138320714e-05, "loss": 1.216, "step": 1872 }, { "epoch": 0.39, "learning_rate": 1.3821325844091779e-05, "loss": 1.2227, "step": 1873 }, { "epoch": 0.39, "learning_rate": 1.381502877598563e-05, "loss": 1.2601, "step": 1874 }, { "epoch": 0.39, "learning_rate": 1.3808729936925396e-05, "loss": 1.2542, "step": 1875 }, { "epoch": 0.39, "learning_rate": 1.380242932983502e-05, "loss": 1.2299, "step": 1876 }, { "epoch": 0.39, "learning_rate": 1.3796126957639278e-05, "loss": 1.2099, "step": 1877 }, { "epoch": 0.4, "learning_rate": 1.3789822823263761e-05, "loss": 1.2468, "step": 1878 }, { "epoch": 0.4, "learning_rate": 1.3783516929634877e-05, "loss": 1.221, "step": 1879 }, { "epoch": 0.4, "learning_rate": 1.377720927967985e-05, "loss": 1.229, "step": 1880 }, { "epoch": 0.4, "learning_rate": 1.377089987632672e-05, "loss": 1.2259, "step": 1881 }, { "epoch": 0.4, "learning_rate": 1.376458872250434e-05, "loss": 1.2359, "step": 1882 }, { "epoch": 0.4, "learning_rate": 1.3758275821142382e-05, "loss": 1.2066, "step": 1883 }, { "epoch": 0.4, "learning_rate": 1.3751961175171323e-05, "loss": 1.2403, "step": 1884 }, { "epoch": 0.4, "learning_rate": 1.3745644787522447e-05, "loss": 1.2531, "step": 1885 }, { "epoch": 0.4, "learning_rate": 1.3739326661127854e-05, "loss": 1.1893, "step": 1886 }, { "epoch": 0.4, "learning_rate": 1.3733006798920444e-05, "loss": 1.2447, "step": 1887 }, { "epoch": 0.4, "learning_rate": 1.3726685203833934e-05, "loss": 1.2592, "step": 1888 }, { "epoch": 0.4, "learning_rate": 1.372036187880283e-05, "loss": 1.2624, "step": 1889 }, { "epoch": 0.4, "learning_rate": 1.371403682676245e-05, "loss": 1.2594, "step": 1890 }, { "epoch": 0.4, "learning_rate": 1.3707710050648918e-05, "loss": 1.2059, "step": 1891 }, { "epoch": 0.4, "learning_rate": 1.3701381553399147e-05, "loss": 1.2846, "step": 1892 }, { "epoch": 0.4, "learning_rate": 1.3695051337950855e-05, "loss": 1.2781, "step": 1893 }, { "epoch": 0.4, "learning_rate": 1.3688719407242559e-05, "loss": 1.2018, "step": 1894 }, { "epoch": 0.4, "learning_rate": 1.3682385764213573e-05, "loss": 1.2652, "step": 1895 }, { "epoch": 0.4, "learning_rate": 1.3676050411804003e-05, "loss": 1.234, "step": 1896 }, { "epoch": 0.4, "learning_rate": 1.3669713352954746e-05, "loss": 1.2791, "step": 1897 }, { "epoch": 0.4, "learning_rate": 1.3663374590607496e-05, "loss": 1.2487, "step": 1898 }, { "epoch": 0.4, "learning_rate": 1.3657034127704733e-05, "loss": 1.236, "step": 1899 }, { "epoch": 0.4, "learning_rate": 1.3650691967189738e-05, "loss": 1.207, "step": 1900 }, { "epoch": 0.4, "learning_rate": 1.3644348112006562e-05, "loss": 1.2648, "step": 1901 }, { "epoch": 0.4, "learning_rate": 1.3638002565100054e-05, "loss": 1.2349, "step": 1902 }, { "epoch": 0.4, "learning_rate": 1.3631655329415849e-05, "loss": 1.2081, "step": 1903 }, { "epoch": 0.4, "learning_rate": 1.3625306407900365e-05, "loss": 1.2413, "step": 1904 }, { "epoch": 0.4, "learning_rate": 1.3618955803500793e-05, "loss": 1.2243, "step": 1905 }, { "epoch": 0.4, "learning_rate": 1.361260351916512e-05, "loss": 1.2618, "step": 1906 }, { "epoch": 0.4, "learning_rate": 1.3606249557842102e-05, "loss": 1.2005, "step": 1907 }, { "epoch": 0.4, "learning_rate": 1.3599893922481285e-05, "loss": 1.1871, "step": 1908 }, { "epoch": 0.4, "learning_rate": 1.3593536616032974e-05, "loss": 1.2035, "step": 1909 }, { "epoch": 0.4, "learning_rate": 1.3587177641448266e-05, "loss": 1.2453, "step": 1910 }, { "epoch": 0.4, "learning_rate": 1.3580817001679028e-05, "loss": 1.1957, "step": 1911 }, { "epoch": 0.4, "learning_rate": 1.3574454699677893e-05, "loss": 1.2336, "step": 1912 }, { "epoch": 0.4, "learning_rate": 1.3568090738398276e-05, "loss": 1.2173, "step": 1913 }, { "epoch": 0.4, "learning_rate": 1.3561725120794355e-05, "loss": 1.1871, "step": 1914 }, { "epoch": 0.4, "learning_rate": 1.355535784982108e-05, "loss": 1.2804, "step": 1915 }, { "epoch": 0.4, "learning_rate": 1.3548988928434168e-05, "loss": 1.2525, "step": 1916 }, { "epoch": 0.4, "learning_rate": 1.35426183595901e-05, "loss": 1.2747, "step": 1917 }, { "epoch": 0.4, "learning_rate": 1.3536246146246125e-05, "loss": 1.2454, "step": 1918 }, { "epoch": 0.4, "learning_rate": 1.3529872291360258e-05, "loss": 1.2312, "step": 1919 }, { "epoch": 0.4, "learning_rate": 1.3523496797891266e-05, "loss": 1.2429, "step": 1920 }, { "epoch": 0.4, "learning_rate": 1.3517119668798689e-05, "loss": 1.2555, "step": 1921 }, { "epoch": 0.4, "learning_rate": 1.3510740907042811e-05, "loss": 1.1561, "step": 1922 }, { "epoch": 0.4, "learning_rate": 1.3504360515584689e-05, "loss": 1.2469, "step": 1923 }, { "epoch": 0.4, "learning_rate": 1.349797849738613e-05, "loss": 1.2306, "step": 1924 }, { "epoch": 0.4, "learning_rate": 1.3491594855409698e-05, "loss": 1.2411, "step": 1925 }, { "epoch": 0.41, "learning_rate": 1.3485209592618704e-05, "loss": 1.3025, "step": 1926 }, { "epoch": 0.41, "learning_rate": 1.3478822711977219e-05, "loss": 1.2182, "step": 1927 }, { "epoch": 0.41, "learning_rate": 1.3472434216450065e-05, "loss": 1.2563, "step": 1928 }, { "epoch": 0.41, "learning_rate": 1.3466044109002811e-05, "loss": 1.2293, "step": 1929 }, { "epoch": 0.41, "learning_rate": 1.3459652392601771e-05, "loss": 1.2552, "step": 1930 }, { "epoch": 0.41, "learning_rate": 1.3453259070214013e-05, "loss": 1.2228, "step": 1931 }, { "epoch": 0.41, "learning_rate": 1.3446864144807346e-05, "loss": 1.2321, "step": 1932 }, { "epoch": 0.41, "learning_rate": 1.3440467619350324e-05, "loss": 1.253, "step": 1933 }, { "epoch": 0.41, "learning_rate": 1.3434069496812243e-05, "loss": 1.2957, "step": 1934 }, { "epoch": 0.41, "learning_rate": 1.3427669780163142e-05, "loss": 1.2214, "step": 1935 }, { "epoch": 0.41, "learning_rate": 1.3421268472373799e-05, "loss": 1.288, "step": 1936 }, { "epoch": 0.41, "learning_rate": 1.341486557641573e-05, "loss": 1.2114, "step": 1937 }, { "epoch": 0.41, "learning_rate": 1.3408461095261193e-05, "loss": 1.2359, "step": 1938 }, { "epoch": 0.41, "learning_rate": 1.3402055031883172e-05, "loss": 1.2537, "step": 1939 }, { "epoch": 0.41, "learning_rate": 1.3395647389255395e-05, "loss": 1.2011, "step": 1940 }, { "epoch": 0.41, "learning_rate": 1.3389238170352318e-05, "loss": 1.184, "step": 1941 }, { "epoch": 0.41, "learning_rate": 1.3382827378149132e-05, "loss": 1.1913, "step": 1942 }, { "epoch": 0.41, "learning_rate": 1.3376415015621755e-05, "loss": 1.2432, "step": 1943 }, { "epoch": 0.41, "learning_rate": 1.3370001085746838e-05, "loss": 1.2478, "step": 1944 }, { "epoch": 0.41, "learning_rate": 1.3363585591501751e-05, "loss": 1.2359, "step": 1945 }, { "epoch": 0.41, "learning_rate": 1.3357168535864603e-05, "loss": 1.2541, "step": 1946 }, { "epoch": 0.41, "learning_rate": 1.3350749921814221e-05, "loss": 1.2391, "step": 1947 }, { "epoch": 0.41, "learning_rate": 1.334432975233015e-05, "loss": 1.2278, "step": 1948 }, { "epoch": 0.41, "learning_rate": 1.3337908030392664e-05, "loss": 1.2695, "step": 1949 }, { "epoch": 0.41, "learning_rate": 1.333148475898276e-05, "loss": 1.2371, "step": 1950 }, { "epoch": 0.41, "learning_rate": 1.3325059941082146e-05, "loss": 1.2265, "step": 1951 }, { "epoch": 0.41, "learning_rate": 1.3318633579673255e-05, "loss": 1.258, "step": 1952 }, { "epoch": 0.41, "learning_rate": 1.3312205677739231e-05, "loss": 1.231, "step": 1953 }, { "epoch": 0.41, "learning_rate": 1.330577623826394e-05, "loss": 1.2546, "step": 1954 }, { "epoch": 0.41, "learning_rate": 1.3299345264231958e-05, "loss": 1.2267, "step": 1955 }, { "epoch": 0.41, "learning_rate": 1.3292912758628566e-05, "loss": 1.1987, "step": 1956 }, { "epoch": 0.41, "learning_rate": 1.3286478724439774e-05, "loss": 1.2366, "step": 1957 }, { "epoch": 0.41, "learning_rate": 1.328004316465228e-05, "loss": 1.2729, "step": 1958 }, { "epoch": 0.41, "learning_rate": 1.3273606082253505e-05, "loss": 1.2179, "step": 1959 }, { "epoch": 0.41, "learning_rate": 1.3267167480231576e-05, "loss": 1.1986, "step": 1960 }, { "epoch": 0.41, "learning_rate": 1.3260727361575315e-05, "loss": 1.2504, "step": 1961 }, { "epoch": 0.41, "learning_rate": 1.325428572927426e-05, "loss": 1.2327, "step": 1962 }, { "epoch": 0.41, "learning_rate": 1.3247842586318646e-05, "loss": 1.2942, "step": 1963 }, { "epoch": 0.41, "learning_rate": 1.3241397935699408e-05, "loss": 1.2383, "step": 1964 }, { "epoch": 0.41, "learning_rate": 1.3234951780408186e-05, "loss": 1.257, "step": 1965 }, { "epoch": 0.41, "learning_rate": 1.322850412343731e-05, "loss": 1.2247, "step": 1966 }, { "epoch": 0.41, "learning_rate": 1.3222054967779816e-05, "loss": 1.2415, "step": 1967 }, { "epoch": 0.41, "learning_rate": 1.321560431642943e-05, "loss": 1.2638, "step": 1968 }, { "epoch": 0.41, "learning_rate": 1.3209152172380582e-05, "loss": 1.2546, "step": 1969 }, { "epoch": 0.41, "learning_rate": 1.3202698538628376e-05, "loss": 1.187, "step": 1970 }, { "epoch": 0.41, "learning_rate": 1.3196243418168628e-05, "loss": 1.2319, "step": 1971 }, { "epoch": 0.41, "learning_rate": 1.318978681399783e-05, "loss": 1.2246, "step": 1972 }, { "epoch": 0.41, "learning_rate": 1.3183328729113165e-05, "loss": 1.2566, "step": 1973 }, { "epoch": 0.42, "learning_rate": 1.3176869166512511e-05, "loss": 1.215, "step": 1974 }, { "epoch": 0.42, "learning_rate": 1.3170408129194429e-05, "loss": 1.2757, "step": 1975 }, { "epoch": 0.42, "learning_rate": 1.3163945620158156e-05, "loss": 1.2397, "step": 1976 }, { "epoch": 0.42, "learning_rate": 1.3157481642403623e-05, "loss": 1.211, "step": 1977 }, { "epoch": 0.42, "learning_rate": 1.3151016198931434e-05, "loss": 1.2087, "step": 1978 }, { "epoch": 0.42, "learning_rate": 1.3144549292742886e-05, "loss": 1.2318, "step": 1979 }, { "epoch": 0.42, "learning_rate": 1.3138080926839938e-05, "loss": 1.2188, "step": 1980 }, { "epoch": 0.42, "learning_rate": 1.313161110422524e-05, "loss": 1.2199, "step": 1981 }, { "epoch": 0.42, "learning_rate": 1.3125139827902116e-05, "loss": 1.2509, "step": 1982 }, { "epoch": 0.42, "learning_rate": 1.3118667100874563e-05, "loss": 1.2816, "step": 1983 }, { "epoch": 0.42, "learning_rate": 1.3112192926147242e-05, "loss": 1.2865, "step": 1984 }, { "epoch": 0.42, "learning_rate": 1.3105717306725503e-05, "loss": 1.2337, "step": 1985 }, { "epoch": 0.42, "learning_rate": 1.3099240245615356e-05, "loss": 1.2298, "step": 1986 }, { "epoch": 0.42, "learning_rate": 1.3092761745823485e-05, "loss": 1.2267, "step": 1987 }, { "epoch": 0.42, "learning_rate": 1.3086281810357237e-05, "loss": 1.2493, "step": 1988 }, { "epoch": 0.42, "learning_rate": 1.3079800442224629e-05, "loss": 1.2173, "step": 1989 }, { "epoch": 0.42, "learning_rate": 1.3073317644434342e-05, "loss": 1.209, "step": 1990 }, { "epoch": 0.42, "learning_rate": 1.3066833419995722e-05, "loss": 1.222, "step": 1991 }, { "epoch": 0.42, "learning_rate": 1.3060347771918775e-05, "loss": 1.189, "step": 1992 }, { "epoch": 0.42, "learning_rate": 1.3053860703214169e-05, "loss": 1.2622, "step": 1993 }, { "epoch": 0.42, "learning_rate": 1.3047372216893233e-05, "loss": 1.2491, "step": 1994 }, { "epoch": 0.42, "learning_rate": 1.3040882315967948e-05, "loss": 1.2505, "step": 1995 }, { "epoch": 0.42, "learning_rate": 1.3034391003450965e-05, "loss": 1.2467, "step": 1996 }, { "epoch": 0.42, "learning_rate": 1.3027898282355576e-05, "loss": 1.2059, "step": 1997 }, { "epoch": 0.42, "learning_rate": 1.3021404155695728e-05, "loss": 1.2166, "step": 1998 }, { "epoch": 0.42, "learning_rate": 1.3014908626486032e-05, "loss": 1.2185, "step": 1999 }, { "epoch": 0.42, "learning_rate": 1.3008411697741741e-05, "loss": 1.2177, "step": 2000 }, { "epoch": 0.42, "learning_rate": 1.3001913372478755e-05, "loss": 1.2474, "step": 2001 }, { "epoch": 0.42, "learning_rate": 1.2995413653713632e-05, "loss": 1.2898, "step": 2002 }, { "epoch": 0.42, "learning_rate": 1.2988912544463568e-05, "loss": 1.1655, "step": 2003 }, { "epoch": 0.42, "learning_rate": 1.2982410047746412e-05, "loss": 1.2328, "step": 2004 }, { "epoch": 0.42, "learning_rate": 1.2975906166580649e-05, "loss": 1.2412, "step": 2005 }, { "epoch": 0.42, "learning_rate": 1.2969400903985415e-05, "loss": 1.2226, "step": 2006 }, { "epoch": 0.42, "learning_rate": 1.2962894262980482e-05, "loss": 1.2531, "step": 2007 }, { "epoch": 0.42, "learning_rate": 1.2956386246586262e-05, "loss": 1.2572, "step": 2008 }, { "epoch": 0.42, "learning_rate": 1.2949876857823807e-05, "loss": 1.2665, "step": 2009 }, { "epoch": 0.42, "learning_rate": 1.2943366099714803e-05, "loss": 1.1844, "step": 2010 }, { "epoch": 0.42, "learning_rate": 1.2936853975281575e-05, "loss": 1.2135, "step": 2011 }, { "epoch": 0.42, "learning_rate": 1.2930340487547089e-05, "loss": 1.2274, "step": 2012 }, { "epoch": 0.42, "learning_rate": 1.2923825639534924e-05, "loss": 1.2418, "step": 2013 }, { "epoch": 0.42, "learning_rate": 1.2917309434269312e-05, "loss": 1.2567, "step": 2014 }, { "epoch": 0.42, "learning_rate": 1.29107918747751e-05, "loss": 1.2205, "step": 2015 }, { "epoch": 0.42, "learning_rate": 1.2904272964077774e-05, "loss": 1.2894, "step": 2016 }, { "epoch": 0.42, "learning_rate": 1.2897752705203439e-05, "loss": 1.222, "step": 2017 }, { "epoch": 0.42, "learning_rate": 1.2891231101178832e-05, "loss": 1.2646, "step": 2018 }, { "epoch": 0.42, "learning_rate": 1.2884708155031311e-05, "loss": 1.3026, "step": 2019 }, { "epoch": 0.42, "learning_rate": 1.2878183869788857e-05, "loss": 1.22, "step": 2020 }, { "epoch": 0.43, "learning_rate": 1.2871658248480077e-05, "loss": 1.2554, "step": 2021 }, { "epoch": 0.43, "learning_rate": 1.286513129413419e-05, "loss": 1.2054, "step": 2022 }, { "epoch": 0.43, "learning_rate": 1.285860300978104e-05, "loss": 1.2441, "step": 2023 }, { "epoch": 0.43, "learning_rate": 1.285207339845109e-05, "loss": 1.2402, "step": 2024 }, { "epoch": 0.43, "learning_rate": 1.2845542463175413e-05, "loss": 1.2012, "step": 2025 }, { "epoch": 0.43, "learning_rate": 1.28390102069857e-05, "loss": 1.2411, "step": 2026 }, { "epoch": 0.43, "learning_rate": 1.2832476632914252e-05, "loss": 1.1981, "step": 2027 }, { "epoch": 0.43, "learning_rate": 1.282594174399399e-05, "loss": 1.2081, "step": 2028 }, { "epoch": 0.43, "learning_rate": 1.2819405543258436e-05, "loss": 1.2018, "step": 2029 }, { "epoch": 0.43, "learning_rate": 1.2812868033741725e-05, "loss": 1.1814, "step": 2030 }, { "epoch": 0.43, "learning_rate": 1.2806329218478602e-05, "loss": 1.2487, "step": 2031 }, { "epoch": 0.43, "learning_rate": 1.2799789100504412e-05, "loss": 1.2736, "step": 2032 }, { "epoch": 0.43, "learning_rate": 1.2793247682855113e-05, "loss": 1.2327, "step": 2033 }, { "epoch": 0.43, "learning_rate": 1.2786704968567254e-05, "loss": 1.2283, "step": 2034 }, { "epoch": 0.43, "learning_rate": 1.2780160960677998e-05, "loss": 1.2609, "step": 2035 }, { "epoch": 0.43, "learning_rate": 1.2773615662225105e-05, "loss": 1.2426, "step": 2036 }, { "epoch": 0.43, "learning_rate": 1.2767069076246931e-05, "loss": 1.1839, "step": 2037 }, { "epoch": 0.43, "learning_rate": 1.2760521205782433e-05, "loss": 1.2459, "step": 2038 }, { "epoch": 0.43, "learning_rate": 1.2753972053871157e-05, "loss": 1.1819, "step": 2039 }, { "epoch": 0.43, "learning_rate": 1.274742162355326e-05, "loss": 1.2597, "step": 2040 }, { "epoch": 0.43, "learning_rate": 1.2740869917869475e-05, "loss": 1.2527, "step": 2041 }, { "epoch": 0.43, "learning_rate": 1.2734316939861136e-05, "loss": 1.2188, "step": 2042 }, { "epoch": 0.43, "learning_rate": 1.2727762692570165e-05, "loss": 1.2574, "step": 2043 }, { "epoch": 0.43, "learning_rate": 1.2721207179039074e-05, "loss": 1.2054, "step": 2044 }, { "epoch": 0.43, "learning_rate": 1.2714650402310967e-05, "loss": 1.224, "step": 2045 }, { "epoch": 0.43, "learning_rate": 1.2708092365429525e-05, "loss": 1.2591, "step": 2046 }, { "epoch": 0.43, "learning_rate": 1.2701533071439024e-05, "loss": 1.2346, "step": 2047 }, { "epoch": 0.43, "learning_rate": 1.269497252338431e-05, "loss": 1.249, "step": 2048 }, { "epoch": 0.43, "learning_rate": 1.268841072431083e-05, "loss": 1.2404, "step": 2049 }, { "epoch": 0.43, "learning_rate": 1.26818476772646e-05, "loss": 1.2378, "step": 2050 }, { "epoch": 0.43, "learning_rate": 1.2675283385292212e-05, "loss": 1.2457, "step": 2051 }, { "epoch": 0.43, "learning_rate": 1.2668717851440842e-05, "loss": 1.2045, "step": 2052 }, { "epoch": 0.43, "learning_rate": 1.2662151078758243e-05, "loss": 1.198, "step": 2053 }, { "epoch": 0.43, "learning_rate": 1.265558307029274e-05, "loss": 1.2633, "step": 2054 }, { "epoch": 0.43, "learning_rate": 1.2649013829093236e-05, "loss": 1.2095, "step": 2055 }, { "epoch": 0.43, "learning_rate": 1.26424433582092e-05, "loss": 1.2257, "step": 2056 }, { "epoch": 0.43, "learning_rate": 1.2635871660690677e-05, "loss": 1.2036, "step": 2057 }, { "epoch": 0.43, "learning_rate": 1.262929873958828e-05, "loss": 1.2208, "step": 2058 }, { "epoch": 0.43, "learning_rate": 1.262272459795319e-05, "loss": 1.2279, "step": 2059 }, { "epoch": 0.43, "learning_rate": 1.2616149238837148e-05, "loss": 1.2155, "step": 2060 }, { "epoch": 0.43, "learning_rate": 1.2609572665292473e-05, "loss": 1.2233, "step": 2061 }, { "epoch": 0.43, "learning_rate": 1.2602994880372042e-05, "loss": 1.1768, "step": 2062 }, { "epoch": 0.43, "learning_rate": 1.2596415887129288e-05, "loss": 1.1935, "step": 2063 }, { "epoch": 0.43, "learning_rate": 1.2589835688618211e-05, "loss": 1.242, "step": 2064 }, { "epoch": 0.43, "learning_rate": 1.258325428789337e-05, "loss": 1.2122, "step": 2065 }, { "epoch": 0.43, "learning_rate": 1.2576671688009886e-05, "loss": 1.26, "step": 2066 }, { "epoch": 0.43, "learning_rate": 1.2570087892023429e-05, "loss": 1.1888, "step": 2067 }, { "epoch": 0.43, "learning_rate": 1.2563502902990225e-05, "loss": 1.2602, "step": 2068 }, { "epoch": 0.44, "learning_rate": 1.2556916723967063e-05, "loss": 1.2556, "step": 2069 }, { "epoch": 0.44, "learning_rate": 1.2550329358011276e-05, "loss": 1.2443, "step": 2070 }, { "epoch": 0.44, "learning_rate": 1.2543740808180746e-05, "loss": 1.2207, "step": 2071 }, { "epoch": 0.44, "learning_rate": 1.2537151077533911e-05, "loss": 1.2046, "step": 2072 }, { "epoch": 0.44, "learning_rate": 1.2530560169129756e-05, "loss": 1.2333, "step": 2073 }, { "epoch": 0.44, "learning_rate": 1.2523968086027807e-05, "loss": 1.2313, "step": 2074 }, { "epoch": 0.44, "learning_rate": 1.2517374831288147e-05, "loss": 1.2776, "step": 2075 }, { "epoch": 0.44, "learning_rate": 1.2510780407971389e-05, "loss": 1.2209, "step": 2076 }, { "epoch": 0.44, "learning_rate": 1.2504184819138691e-05, "loss": 1.2538, "step": 2077 }, { "epoch": 0.44, "learning_rate": 1.2497588067851767e-05, "loss": 1.2212, "step": 2078 }, { "epoch": 0.44, "learning_rate": 1.2490990157172856e-05, "loss": 1.2899, "step": 2079 }, { "epoch": 0.44, "learning_rate": 1.2484391090164735e-05, "loss": 1.2187, "step": 2080 }, { "epoch": 0.44, "learning_rate": 1.2477790869890725e-05, "loss": 1.2536, "step": 2081 }, { "epoch": 0.44, "learning_rate": 1.2471189499414675e-05, "loss": 1.2583, "step": 2082 }, { "epoch": 0.44, "learning_rate": 1.2464586981800979e-05, "loss": 1.2338, "step": 2083 }, { "epoch": 0.44, "learning_rate": 1.2457983320114549e-05, "loss": 1.207, "step": 2084 }, { "epoch": 0.44, "learning_rate": 1.2451378517420837e-05, "loss": 1.1923, "step": 2085 }, { "epoch": 0.44, "learning_rate": 1.2444772576785828e-05, "loss": 1.254, "step": 2086 }, { "epoch": 0.44, "learning_rate": 1.2438165501276028e-05, "loss": 1.1831, "step": 2087 }, { "epoch": 0.44, "learning_rate": 1.2431557293958467e-05, "loss": 1.2155, "step": 2088 }, { "epoch": 0.44, "learning_rate": 1.242494795790071e-05, "loss": 1.2366, "step": 2089 }, { "epoch": 0.44, "learning_rate": 1.2418337496170842e-05, "loss": 1.1961, "step": 2090 }, { "epoch": 0.44, "learning_rate": 1.2411725911837471e-05, "loss": 1.2096, "step": 2091 }, { "epoch": 0.44, "learning_rate": 1.240511320796972e-05, "loss": 1.2091, "step": 2092 }, { "epoch": 0.44, "learning_rate": 1.2398499387637241e-05, "loss": 1.2404, "step": 2093 }, { "epoch": 0.44, "learning_rate": 1.2391884453910201e-05, "loss": 1.2518, "step": 2094 }, { "epoch": 0.44, "learning_rate": 1.2385268409859281e-05, "loss": 1.2387, "step": 2095 }, { "epoch": 0.44, "learning_rate": 1.237865125855568e-05, "loss": 1.2532, "step": 2096 }, { "epoch": 0.44, "learning_rate": 1.2372033003071109e-05, "loss": 1.2234, "step": 2097 }, { "epoch": 0.44, "learning_rate": 1.2365413646477796e-05, "loss": 1.2032, "step": 2098 }, { "epoch": 0.44, "learning_rate": 1.2358793191848473e-05, "loss": 1.2055, "step": 2099 }, { "epoch": 0.44, "learning_rate": 1.235217164225639e-05, "loss": 1.2413, "step": 2100 }, { "epoch": 0.44, "learning_rate": 1.2345549000775294e-05, "loss": 1.199, "step": 2101 }, { "epoch": 0.44, "learning_rate": 1.2338925270479453e-05, "loss": 1.1907, "step": 2102 }, { "epoch": 0.44, "learning_rate": 1.2332300454443636e-05, "loss": 1.2582, "step": 2103 }, { "epoch": 0.44, "learning_rate": 1.2325674555743106e-05, "loss": 1.2538, "step": 2104 }, { "epoch": 0.44, "learning_rate": 1.2319047577453638e-05, "loss": 1.2633, "step": 2105 }, { "epoch": 0.44, "learning_rate": 1.2312419522651507e-05, "loss": 1.246, "step": 2106 }, { "epoch": 0.44, "learning_rate": 1.2305790394413486e-05, "loss": 1.2331, "step": 2107 }, { "epoch": 0.44, "learning_rate": 1.2299160195816851e-05, "loss": 1.2373, "step": 2108 }, { "epoch": 0.44, "learning_rate": 1.2292528929939366e-05, "loss": 1.2626, "step": 2109 }, { "epoch": 0.44, "learning_rate": 1.2285896599859297e-05, "loss": 1.2567, "step": 2110 }, { "epoch": 0.44, "learning_rate": 1.2279263208655403e-05, "loss": 1.2622, "step": 2111 }, { "epoch": 0.44, "learning_rate": 1.2272628759406933e-05, "loss": 1.1758, "step": 2112 }, { "epoch": 0.44, "learning_rate": 1.2265993255193634e-05, "loss": 1.2079, "step": 2113 }, { "epoch": 0.44, "learning_rate": 1.2259356699095732e-05, "loss": 1.2283, "step": 2114 }, { "epoch": 0.44, "learning_rate": 1.225271909419395e-05, "loss": 1.1975, "step": 2115 }, { "epoch": 0.45, "learning_rate": 1.2246080443569499e-05, "loss": 1.2383, "step": 2116 }, { "epoch": 0.45, "learning_rate": 1.2239440750304064e-05, "loss": 1.2132, "step": 2117 }, { "epoch": 0.45, "learning_rate": 1.2232800017479827e-05, "loss": 1.2312, "step": 2118 }, { "epoch": 0.45, "learning_rate": 1.2226158248179442e-05, "loss": 1.2741, "step": 2119 }, { "epoch": 0.45, "learning_rate": 1.2219515445486055e-05, "loss": 1.1969, "step": 2120 }, { "epoch": 0.45, "learning_rate": 1.2212871612483283e-05, "loss": 1.2258, "step": 2121 }, { "epoch": 0.45, "learning_rate": 1.2206226752255223e-05, "loss": 1.176, "step": 2122 }, { "epoch": 0.45, "learning_rate": 1.219958086788645e-05, "loss": 1.2883, "step": 2123 }, { "epoch": 0.45, "learning_rate": 1.2192933962462018e-05, "loss": 1.2664, "step": 2124 }, { "epoch": 0.45, "learning_rate": 1.218628603906745e-05, "loss": 1.2192, "step": 2125 }, { "epoch": 0.45, "learning_rate": 1.2179637100788739e-05, "loss": 1.3025, "step": 2126 }, { "epoch": 0.45, "learning_rate": 1.2172987150712358e-05, "loss": 1.2189, "step": 2127 }, { "epoch": 0.45, "learning_rate": 1.216633619192524e-05, "loss": 1.205, "step": 2128 }, { "epoch": 0.45, "learning_rate": 1.2159684227514799e-05, "loss": 1.2166, "step": 2129 }, { "epoch": 0.45, "learning_rate": 1.21530312605689e-05, "loss": 1.2478, "step": 2130 }, { "epoch": 0.45, "learning_rate": 1.2146377294175884e-05, "loss": 1.1973, "step": 2131 }, { "epoch": 0.45, "learning_rate": 1.213972233142455e-05, "loss": 1.2675, "step": 2132 }, { "epoch": 0.45, "learning_rate": 1.213306637540417e-05, "loss": 1.2533, "step": 2133 }, { "epoch": 0.45, "learning_rate": 1.2126409429204462e-05, "loss": 1.2567, "step": 2134 }, { "epoch": 0.45, "learning_rate": 1.2119751495915617e-05, "loss": 1.2399, "step": 2135 }, { "epoch": 0.45, "learning_rate": 1.2113092578628273e-05, "loss": 1.1579, "step": 2136 }, { "epoch": 0.45, "learning_rate": 1.2106432680433536e-05, "loss": 1.2012, "step": 2137 }, { "epoch": 0.45, "learning_rate": 1.2099771804422958e-05, "loss": 1.2218, "step": 2138 }, { "epoch": 0.45, "learning_rate": 1.2093109953688548e-05, "loss": 1.2294, "step": 2139 }, { "epoch": 0.45, "learning_rate": 1.2086447131322772e-05, "loss": 1.2759, "step": 2140 }, { "epoch": 0.45, "learning_rate": 1.2079783340418543e-05, "loss": 1.2352, "step": 2141 }, { "epoch": 0.45, "learning_rate": 1.207311858406922e-05, "loss": 1.2237, "step": 2142 }, { "epoch": 0.45, "learning_rate": 1.2066452865368614e-05, "loss": 1.2559, "step": 2143 }, { "epoch": 0.45, "learning_rate": 1.2059786187410984e-05, "loss": 1.2276, "step": 2144 }, { "epoch": 0.45, "learning_rate": 1.2053118553291036e-05, "loss": 1.2116, "step": 2145 }, { "epoch": 0.45, "learning_rate": 1.2046449966103912e-05, "loss": 1.2364, "step": 2146 }, { "epoch": 0.45, "learning_rate": 1.2039780428945202e-05, "loss": 1.2033, "step": 2147 }, { "epoch": 0.45, "learning_rate": 1.203310994491094e-05, "loss": 1.235, "step": 2148 }, { "epoch": 0.45, "learning_rate": 1.202643851709759e-05, "loss": 1.2381, "step": 2149 }, { "epoch": 0.45, "learning_rate": 1.2019766148602063e-05, "loss": 1.2033, "step": 2150 }, { "epoch": 0.45, "learning_rate": 1.2013092842521702e-05, "loss": 1.1869, "step": 2151 }, { "epoch": 0.45, "learning_rate": 1.2006418601954291e-05, "loss": 1.2331, "step": 2152 }, { "epoch": 0.45, "learning_rate": 1.1999743429998036e-05, "loss": 1.2547, "step": 2153 }, { "epoch": 0.45, "learning_rate": 1.1993067329751593e-05, "loss": 1.2047, "step": 2154 }, { "epoch": 0.45, "learning_rate": 1.1986390304314029e-05, "loss": 1.2845, "step": 2155 }, { "epoch": 0.45, "learning_rate": 1.1979712356784854e-05, "loss": 1.2363, "step": 2156 }, { "epoch": 0.45, "learning_rate": 1.1973033490264e-05, "loss": 1.2491, "step": 2157 }, { "epoch": 0.45, "learning_rate": 1.1966353707851837e-05, "loss": 1.2234, "step": 2158 }, { "epoch": 0.45, "learning_rate": 1.195967301264914e-05, "loss": 1.1823, "step": 2159 }, { "epoch": 0.45, "learning_rate": 1.1952991407757122e-05, "loss": 1.22, "step": 2160 }, { "epoch": 0.45, "learning_rate": 1.1946308896277416e-05, "loss": 1.222, "step": 2161 }, { "epoch": 0.45, "learning_rate": 1.1939625481312076e-05, "loss": 1.2639, "step": 2162 }, { "epoch": 0.45, "learning_rate": 1.1932941165963574e-05, "loss": 1.2088, "step": 2163 }, { "epoch": 0.46, "learning_rate": 1.19262559533348e-05, "loss": 1.2414, "step": 2164 }, { "epoch": 0.46, "learning_rate": 1.1919569846529058e-05, "loss": 1.2218, "step": 2165 }, { "epoch": 0.46, "learning_rate": 1.1912882848650075e-05, "loss": 1.2365, "step": 2166 }, { "epoch": 0.46, "learning_rate": 1.1906194962801982e-05, "loss": 1.1877, "step": 2167 }, { "epoch": 0.46, "learning_rate": 1.1899506192089327e-05, "loss": 1.2313, "step": 2168 }, { "epoch": 0.46, "learning_rate": 1.1892816539617072e-05, "loss": 1.2446, "step": 2169 }, { "epoch": 0.46, "learning_rate": 1.1886126008490584e-05, "loss": 1.2384, "step": 2170 }, { "epoch": 0.46, "learning_rate": 1.1879434601815633e-05, "loss": 1.2709, "step": 2171 }, { "epoch": 0.46, "learning_rate": 1.1872742322698408e-05, "loss": 1.213, "step": 2172 }, { "epoch": 0.46, "learning_rate": 1.186604917424549e-05, "loss": 1.1935, "step": 2173 }, { "epoch": 0.46, "learning_rate": 1.1859355159563877e-05, "loss": 1.2465, "step": 2174 }, { "epoch": 0.46, "learning_rate": 1.1852660281760954e-05, "loss": 1.2484, "step": 2175 }, { "epoch": 0.46, "learning_rate": 1.1845964543944521e-05, "loss": 1.2646, "step": 2176 }, { "epoch": 0.46, "learning_rate": 1.1839267949222766e-05, "loss": 1.2185, "step": 2177 }, { "epoch": 0.46, "learning_rate": 1.1832570500704281e-05, "loss": 1.2333, "step": 2178 }, { "epoch": 0.46, "learning_rate": 1.1825872201498057e-05, "loss": 1.2085, "step": 2179 }, { "epoch": 0.46, "learning_rate": 1.1819173054713466e-05, "loss": 1.2642, "step": 2180 }, { "epoch": 0.46, "learning_rate": 1.181247306346029e-05, "loss": 1.2726, "step": 2181 }, { "epoch": 0.46, "learning_rate": 1.1805772230848696e-05, "loss": 1.2457, "step": 2182 }, { "epoch": 0.46, "learning_rate": 1.1799070559989241e-05, "loss": 1.2663, "step": 2183 }, { "epoch": 0.46, "learning_rate": 1.1792368053992868e-05, "loss": 1.2411, "step": 2184 }, { "epoch": 0.46, "learning_rate": 1.1785664715970915e-05, "loss": 1.1873, "step": 2185 }, { "epoch": 0.46, "learning_rate": 1.17789605490351e-05, "loss": 1.2806, "step": 2186 }, { "epoch": 0.46, "learning_rate": 1.1772255556297534e-05, "loss": 1.2386, "step": 2187 }, { "epoch": 0.46, "learning_rate": 1.1765549740870701e-05, "loss": 1.1898, "step": 2188 }, { "epoch": 0.46, "learning_rate": 1.1758843105867474e-05, "loss": 1.2627, "step": 2189 }, { "epoch": 0.46, "learning_rate": 1.1752135654401102e-05, "loss": 1.2364, "step": 2190 }, { "epoch": 0.46, "learning_rate": 1.1745427389585218e-05, "loss": 1.2493, "step": 2191 }, { "epoch": 0.46, "learning_rate": 1.1738718314533826e-05, "loss": 1.2347, "step": 2192 }, { "epoch": 0.46, "learning_rate": 1.1732008432361313e-05, "loss": 1.2108, "step": 2193 }, { "epoch": 0.46, "learning_rate": 1.1725297746182434e-05, "loss": 1.2398, "step": 2194 }, { "epoch": 0.46, "learning_rate": 1.1718586259112325e-05, "loss": 1.1633, "step": 2195 }, { "epoch": 0.46, "learning_rate": 1.1711873974266488e-05, "loss": 1.2683, "step": 2196 }, { "epoch": 0.46, "learning_rate": 1.1705160894760796e-05, "loss": 1.1923, "step": 2197 }, { "epoch": 0.46, "learning_rate": 1.169844702371149e-05, "loss": 1.2334, "step": 2198 }, { "epoch": 0.46, "learning_rate": 1.1691732364235185e-05, "loss": 1.1687, "step": 2199 }, { "epoch": 0.46, "learning_rate": 1.1685016919448851e-05, "loss": 1.1943, "step": 2200 }, { "epoch": 0.46, "learning_rate": 1.1678300692469832e-05, "loss": 1.2447, "step": 2201 }, { "epoch": 0.46, "learning_rate": 1.1671583686415833e-05, "loss": 1.255, "step": 2202 }, { "epoch": 0.46, "learning_rate": 1.1664865904404918e-05, "loss": 1.253, "step": 2203 }, { "epoch": 0.46, "learning_rate": 1.1658147349555515e-05, "loss": 1.211, "step": 2204 }, { "epoch": 0.46, "learning_rate": 1.1651428024986405e-05, "loss": 1.2338, "step": 2205 }, { "epoch": 0.46, "learning_rate": 1.164470793381673e-05, "loss": 1.2317, "step": 2206 }, { "epoch": 0.46, "learning_rate": 1.1637987079165988e-05, "loss": 1.233, "step": 2207 }, { "epoch": 0.46, "learning_rate": 1.1631265464154034e-05, "loss": 1.1903, "step": 2208 }, { "epoch": 0.46, "learning_rate": 1.1624543091901069e-05, "loss": 1.2574, "step": 2209 }, { "epoch": 0.46, "learning_rate": 1.161781996552765e-05, "loss": 1.2247, "step": 2210 }, { "epoch": 0.47, "learning_rate": 1.1611096088154683e-05, "loss": 1.2617, "step": 2211 }, { "epoch": 0.47, "learning_rate": 1.1604371462903425e-05, "loss": 1.2349, "step": 2212 }, { "epoch": 0.47, "learning_rate": 1.1597646092895479e-05, "loss": 1.256, "step": 2213 }, { "epoch": 0.47, "learning_rate": 1.159091998125279e-05, "loss": 1.1816, "step": 2214 }, { "epoch": 0.47, "learning_rate": 1.1584193131097653e-05, "loss": 1.2385, "step": 2215 }, { "epoch": 0.47, "learning_rate": 1.1577465545552705e-05, "loss": 1.1915, "step": 2216 }, { "epoch": 0.47, "learning_rate": 1.1570737227740915e-05, "loss": 1.2108, "step": 2217 }, { "epoch": 0.47, "learning_rate": 1.1564008180785606e-05, "loss": 1.2748, "step": 2218 }, { "epoch": 0.47, "learning_rate": 1.1557278407810431e-05, "loss": 1.2557, "step": 2219 }, { "epoch": 0.47, "learning_rate": 1.1550547911939382e-05, "loss": 1.2675, "step": 2220 }, { "epoch": 0.47, "learning_rate": 1.1543816696296785e-05, "loss": 1.2256, "step": 2221 }, { "epoch": 0.47, "learning_rate": 1.1537084764007303e-05, "loss": 1.2327, "step": 2222 }, { "epoch": 0.47, "learning_rate": 1.153035211819593e-05, "loss": 1.2148, "step": 2223 }, { "epoch": 0.47, "learning_rate": 1.1523618761987992e-05, "loss": 1.2238, "step": 2224 }, { "epoch": 0.47, "learning_rate": 1.1516884698509143e-05, "loss": 1.1561, "step": 2225 }, { "epoch": 0.47, "learning_rate": 1.1510149930885366e-05, "loss": 1.1706, "step": 2226 }, { "epoch": 0.47, "learning_rate": 1.1503414462242975e-05, "loss": 1.2067, "step": 2227 }, { "epoch": 0.47, "learning_rate": 1.1496678295708602e-05, "loss": 1.2221, "step": 2228 }, { "epoch": 0.47, "learning_rate": 1.1489941434409211e-05, "loss": 1.2104, "step": 2229 }, { "epoch": 0.47, "learning_rate": 1.148320388147208e-05, "loss": 1.2622, "step": 2230 }, { "epoch": 0.47, "learning_rate": 1.1476465640024814e-05, "loss": 1.2486, "step": 2231 }, { "epoch": 0.47, "learning_rate": 1.1469726713195335e-05, "loss": 1.2606, "step": 2232 }, { "epoch": 0.47, "learning_rate": 1.1462987104111889e-05, "loss": 1.24, "step": 2233 }, { "epoch": 0.47, "learning_rate": 1.1456246815903028e-05, "loss": 1.2074, "step": 2234 }, { "epoch": 0.47, "learning_rate": 1.1449505851697627e-05, "loss": 1.2446, "step": 2235 }, { "epoch": 0.47, "learning_rate": 1.144276421462487e-05, "loss": 1.1775, "step": 2236 }, { "epoch": 0.47, "learning_rate": 1.1436021907814262e-05, "loss": 1.1936, "step": 2237 }, { "epoch": 0.47, "learning_rate": 1.1429278934395611e-05, "loss": 1.2507, "step": 2238 }, { "epoch": 0.47, "learning_rate": 1.1422535297499035e-05, "loss": 1.2608, "step": 2239 }, { "epoch": 0.47, "learning_rate": 1.1415791000254966e-05, "loss": 1.2139, "step": 2240 }, { "epoch": 0.47, "learning_rate": 1.1409046045794133e-05, "loss": 1.2234, "step": 2241 }, { "epoch": 0.47, "learning_rate": 1.1402300437247575e-05, "loss": 1.237, "step": 2242 }, { "epoch": 0.47, "learning_rate": 1.1395554177746639e-05, "loss": 1.2107, "step": 2243 }, { "epoch": 0.47, "learning_rate": 1.1388807270422967e-05, "loss": 1.2551, "step": 2244 }, { "epoch": 0.47, "learning_rate": 1.1382059718408508e-05, "loss": 1.2251, "step": 2245 }, { "epoch": 0.47, "learning_rate": 1.1375311524835501e-05, "loss": 1.243, "step": 2246 }, { "epoch": 0.47, "learning_rate": 1.1368562692836496e-05, "loss": 1.2353, "step": 2247 }, { "epoch": 0.47, "learning_rate": 1.1361813225544322e-05, "loss": 1.2806, "step": 2248 }, { "epoch": 0.47, "learning_rate": 1.1355063126092128e-05, "loss": 1.2503, "step": 2249 }, { "epoch": 0.47, "learning_rate": 1.1348312397613331e-05, "loss": 1.1957, "step": 2250 }, { "epoch": 0.47, "learning_rate": 1.1341561043241649e-05, "loss": 1.1852, "step": 2251 }, { "epoch": 0.47, "learning_rate": 1.13348090661111e-05, "loss": 1.2423, "step": 2252 }, { "epoch": 0.47, "learning_rate": 1.1328056469355976e-05, "loss": 1.2417, "step": 2253 }, { "epoch": 0.47, "learning_rate": 1.132130325611087e-05, "loss": 1.2562, "step": 2254 }, { "epoch": 0.47, "learning_rate": 1.131454942951065e-05, "loss": 1.2483, "step": 2255 }, { "epoch": 0.47, "learning_rate": 1.1307794992690475e-05, "loss": 1.2151, "step": 2256 }, { "epoch": 0.47, "learning_rate": 1.1301039948785788e-05, "loss": 1.2412, "step": 2257 }, { "epoch": 0.47, "learning_rate": 1.129428430093231e-05, "loss": 1.2204, "step": 2258 }, { "epoch": 0.48, "learning_rate": 1.1287528052266044e-05, "loss": 1.2151, "step": 2259 }, { "epoch": 0.48, "learning_rate": 1.1280771205923269e-05, "loss": 1.2553, "step": 2260 }, { "epoch": 0.48, "learning_rate": 1.127401376504055e-05, "loss": 1.2185, "step": 2261 }, { "epoch": 0.48, "learning_rate": 1.1267255732754725e-05, "loss": 1.2726, "step": 2262 }, { "epoch": 0.48, "learning_rate": 1.1260497112202895e-05, "loss": 1.2617, "step": 2263 }, { "epoch": 0.48, "learning_rate": 1.1253737906522448e-05, "loss": 1.2474, "step": 2264 }, { "epoch": 0.48, "learning_rate": 1.1246978118851037e-05, "loss": 1.228, "step": 2265 }, { "epoch": 0.48, "learning_rate": 1.1240217752326593e-05, "loss": 1.2241, "step": 2266 }, { "epoch": 0.48, "learning_rate": 1.12334568100873e-05, "loss": 1.2369, "step": 2267 }, { "epoch": 0.48, "learning_rate": 1.1226695295271623e-05, "loss": 1.2121, "step": 2268 }, { "epoch": 0.48, "learning_rate": 1.121993321101829e-05, "loss": 1.2239, "step": 2269 }, { "epoch": 0.48, "learning_rate": 1.1213170560466291e-05, "loss": 1.2323, "step": 2270 }, { "epoch": 0.48, "learning_rate": 1.120640734675488e-05, "loss": 1.217, "step": 2271 }, { "epoch": 0.48, "learning_rate": 1.1199643573023569e-05, "loss": 1.2093, "step": 2272 }, { "epoch": 0.48, "learning_rate": 1.1192879242412137e-05, "loss": 1.2233, "step": 2273 }, { "epoch": 0.48, "learning_rate": 1.1186114358060623e-05, "loss": 1.2539, "step": 2274 }, { "epoch": 0.48, "learning_rate": 1.1179348923109306e-05, "loss": 1.2331, "step": 2275 }, { "epoch": 0.48, "learning_rate": 1.1172582940698738e-05, "loss": 1.2549, "step": 2276 }, { "epoch": 0.48, "learning_rate": 1.1165816413969719e-05, "loss": 1.2275, "step": 2277 }, { "epoch": 0.48, "learning_rate": 1.1159049346063305e-05, "loss": 1.2079, "step": 2278 }, { "epoch": 0.48, "learning_rate": 1.1152281740120796e-05, "loss": 1.2075, "step": 2279 }, { "epoch": 0.48, "learning_rate": 1.1145513599283751e-05, "loss": 1.2717, "step": 2280 }, { "epoch": 0.48, "learning_rate": 1.1138744926693966e-05, "loss": 1.2313, "step": 2281 }, { "epoch": 0.48, "learning_rate": 1.1131975725493493e-05, "loss": 1.2212, "step": 2282 }, { "epoch": 0.48, "learning_rate": 1.1125205998824633e-05, "loss": 1.2142, "step": 2283 }, { "epoch": 0.48, "learning_rate": 1.1118435749829912e-05, "loss": 1.2641, "step": 2284 }, { "epoch": 0.48, "learning_rate": 1.1111664981652121e-05, "loss": 1.2223, "step": 2285 }, { "epoch": 0.48, "learning_rate": 1.1104893697434281e-05, "loss": 1.2224, "step": 2286 }, { "epoch": 0.48, "learning_rate": 1.1098121900319652e-05, "loss": 1.2304, "step": 2287 }, { "epoch": 0.48, "learning_rate": 1.1091349593451731e-05, "loss": 1.189, "step": 2288 }, { "epoch": 0.48, "learning_rate": 1.1084576779974257e-05, "loss": 1.2627, "step": 2289 }, { "epoch": 0.48, "learning_rate": 1.10778034630312e-05, "loss": 1.2502, "step": 2290 }, { "epoch": 0.48, "learning_rate": 1.1071029645766771e-05, "loss": 1.2134, "step": 2291 }, { "epoch": 0.48, "learning_rate": 1.1064255331325398e-05, "loss": 1.203, "step": 2292 }, { "epoch": 0.48, "learning_rate": 1.1057480522851755e-05, "loss": 1.2574, "step": 2293 }, { "epoch": 0.48, "learning_rate": 1.105070522349074e-05, "loss": 1.2372, "step": 2294 }, { "epoch": 0.48, "learning_rate": 1.1043929436387477e-05, "loss": 1.2313, "step": 2295 }, { "epoch": 0.48, "learning_rate": 1.1037153164687318e-05, "loss": 1.2713, "step": 2296 }, { "epoch": 0.48, "learning_rate": 1.103037641153584e-05, "loss": 1.242, "step": 2297 }, { "epoch": 0.48, "learning_rate": 1.1023599180078842e-05, "loss": 1.2337, "step": 2298 }, { "epoch": 0.48, "learning_rate": 1.1016821473462349e-05, "loss": 1.2369, "step": 2299 }, { "epoch": 0.48, "learning_rate": 1.1010043294832601e-05, "loss": 1.1817, "step": 2300 }, { "epoch": 0.48, "learning_rate": 1.1003264647336063e-05, "loss": 1.2304, "step": 2301 }, { "epoch": 0.48, "learning_rate": 1.0996485534119413e-05, "loss": 1.2192, "step": 2302 }, { "epoch": 0.48, "learning_rate": 1.0989705958329546e-05, "loss": 1.2225, "step": 2303 }, { "epoch": 0.48, "learning_rate": 1.0982925923113575e-05, "loss": 1.179, "step": 2304 }, { "epoch": 0.48, "learning_rate": 1.097614543161882e-05, "loss": 1.221, "step": 2305 }, { "epoch": 0.49, "learning_rate": 1.096936448699282e-05, "loss": 1.2022, "step": 2306 }, { "epoch": 0.49, "learning_rate": 1.0962583092383318e-05, "loss": 1.2553, "step": 2307 }, { "epoch": 0.49, "learning_rate": 1.0955801250938273e-05, "loss": 1.1969, "step": 2308 }, { "epoch": 0.49, "learning_rate": 1.0949018965805844e-05, "loss": 1.2394, "step": 2309 }, { "epoch": 0.49, "learning_rate": 1.0942236240134403e-05, "loss": 1.1972, "step": 2310 }, { "epoch": 0.49, "learning_rate": 1.0935453077072518e-05, "loss": 1.1939, "step": 2311 }, { "epoch": 0.49, "learning_rate": 1.092866947976897e-05, "loss": 1.1986, "step": 2312 }, { "epoch": 0.49, "learning_rate": 1.0921885451372731e-05, "loss": 1.2462, "step": 2313 }, { "epoch": 0.49, "learning_rate": 1.0915100995032983e-05, "loss": 1.1965, "step": 2314 }, { "epoch": 0.49, "learning_rate": 1.0908316113899098e-05, "loss": 1.257, "step": 2315 }, { "epoch": 0.49, "learning_rate": 1.0901530811120655e-05, "loss": 1.2211, "step": 2316 }, { "epoch": 0.49, "learning_rate": 1.0894745089847416e-05, "loss": 1.2031, "step": 2317 }, { "epoch": 0.49, "learning_rate": 1.0887958953229349e-05, "loss": 1.2305, "step": 2318 }, { "epoch": 0.49, "learning_rate": 1.088117240441661e-05, "loss": 1.2185, "step": 2319 }, { "epoch": 0.49, "learning_rate": 1.0874385446559545e-05, "loss": 1.2173, "step": 2320 }, { "epoch": 0.49, "learning_rate": 1.0867598082808692e-05, "loss": 1.1425, "step": 2321 }, { "epoch": 0.49, "learning_rate": 1.0860810316314776e-05, "loss": 1.2497, "step": 2322 }, { "epoch": 0.49, "learning_rate": 1.0854022150228711e-05, "loss": 1.185, "step": 2323 }, { "epoch": 0.49, "learning_rate": 1.0847233587701598e-05, "loss": 1.282, "step": 2324 }, { "epoch": 0.49, "learning_rate": 1.0840444631884713e-05, "loss": 1.2064, "step": 2325 }, { "epoch": 0.49, "learning_rate": 1.0833655285929525e-05, "loss": 1.2397, "step": 2326 }, { "epoch": 0.49, "learning_rate": 1.0826865552987678e-05, "loss": 1.2595, "step": 2327 }, { "epoch": 0.49, "learning_rate": 1.0820075436210999e-05, "loss": 1.199, "step": 2328 }, { "epoch": 0.49, "learning_rate": 1.0813284938751492e-05, "loss": 1.2532, "step": 2329 }, { "epoch": 0.49, "learning_rate": 1.0806494063761334e-05, "loss": 1.2012, "step": 2330 }, { "epoch": 0.49, "learning_rate": 1.0799702814392886e-05, "loss": 1.2006, "step": 2331 }, { "epoch": 0.49, "learning_rate": 1.0792911193798675e-05, "loss": 1.2514, "step": 2332 }, { "epoch": 0.49, "learning_rate": 1.07861192051314e-05, "loss": 1.2349, "step": 2333 }, { "epoch": 0.49, "learning_rate": 1.0779326851543938e-05, "loss": 1.1896, "step": 2334 }, { "epoch": 0.49, "learning_rate": 1.0772534136189332e-05, "loss": 1.2683, "step": 2335 }, { "epoch": 0.49, "learning_rate": 1.0765741062220787e-05, "loss": 1.22, "step": 2336 }, { "epoch": 0.49, "learning_rate": 1.0758947632791686e-05, "loss": 1.2016, "step": 2337 }, { "epoch": 0.49, "learning_rate": 1.0752153851055563e-05, "loss": 1.2566, "step": 2338 }, { "epoch": 0.49, "learning_rate": 1.0745359720166126e-05, "loss": 1.2166, "step": 2339 }, { "epoch": 0.49, "learning_rate": 1.0738565243277244e-05, "loss": 1.2248, "step": 2340 }, { "epoch": 0.49, "learning_rate": 1.0731770423542946e-05, "loss": 1.2725, "step": 2341 }, { "epoch": 0.49, "learning_rate": 1.0724975264117414e-05, "loss": 1.268, "step": 2342 }, { "epoch": 0.49, "learning_rate": 1.0718179768154996e-05, "loss": 1.2346, "step": 2343 }, { "epoch": 0.49, "learning_rate": 1.0711383938810191e-05, "loss": 1.21, "step": 2344 }, { "epoch": 0.49, "learning_rate": 1.0704587779237654e-05, "loss": 1.2667, "step": 2345 }, { "epoch": 0.49, "learning_rate": 1.0697791292592198e-05, "loss": 1.223, "step": 2346 }, { "epoch": 0.49, "learning_rate": 1.069099448202878e-05, "loss": 1.2229, "step": 2347 }, { "epoch": 0.49, "learning_rate": 1.0684197350702511e-05, "loss": 1.1876, "step": 2348 }, { "epoch": 0.49, "learning_rate": 1.0677399901768656e-05, "loss": 1.199, "step": 2349 }, { "epoch": 0.49, "learning_rate": 1.0670602138382615e-05, "loss": 1.1973, "step": 2350 }, { "epoch": 0.49, "learning_rate": 1.0663804063699948e-05, "loss": 1.2212, "step": 2351 }, { "epoch": 0.49, "learning_rate": 1.0657005680876344e-05, "loss": 1.2383, "step": 2352 }, { "epoch": 0.49, "learning_rate": 1.0650206993067656e-05, "loss": 1.2149, "step": 2353 }, { "epoch": 0.5, "learning_rate": 1.0643408003429856e-05, "loss": 1.2641, "step": 2354 }, { "epoch": 0.5, "learning_rate": 1.0636608715119073e-05, "loss": 1.2528, "step": 2355 }, { "epoch": 0.5, "learning_rate": 1.0629809131291561e-05, "loss": 1.209, "step": 2356 }, { "epoch": 0.5, "learning_rate": 1.0623009255103726e-05, "loss": 1.2289, "step": 2357 }, { "epoch": 0.5, "learning_rate": 1.0616209089712102e-05, "loss": 1.1957, "step": 2358 }, { "epoch": 0.5, "learning_rate": 1.060940863827335e-05, "loss": 1.1742, "step": 2359 }, { "epoch": 0.5, "learning_rate": 1.060260790394428e-05, "loss": 1.2415, "step": 2360 }, { "epoch": 0.5, "learning_rate": 1.0595806889881817e-05, "loss": 1.1835, "step": 2361 }, { "epoch": 0.5, "learning_rate": 1.0589005599243032e-05, "loss": 1.1945, "step": 2362 }, { "epoch": 0.5, "learning_rate": 1.0582204035185106e-05, "loss": 1.2266, "step": 2363 }, { "epoch": 0.5, "learning_rate": 1.0575402200865362e-05, "loss": 1.2158, "step": 2364 }, { "epoch": 0.5, "learning_rate": 1.0568600099441242e-05, "loss": 1.1867, "step": 2365 }, { "epoch": 0.5, "learning_rate": 1.0561797734070315e-05, "loss": 1.2088, "step": 2366 }, { "epoch": 0.5, "learning_rate": 1.0554995107910266e-05, "loss": 1.2201, "step": 2367 }, { "epoch": 0.5, "learning_rate": 1.0548192224118912e-05, "loss": 1.241, "step": 2368 }, { "epoch": 0.5, "learning_rate": 1.0541389085854177e-05, "loss": 1.1846, "step": 2369 }, { "epoch": 0.5, "learning_rate": 1.0534585696274113e-05, "loss": 1.2283, "step": 2370 }, { "epoch": 0.5, "learning_rate": 1.0527782058536884e-05, "loss": 1.2134, "step": 2371 }, { "epoch": 0.5, "learning_rate": 1.0520978175800771e-05, "loss": 1.2017, "step": 2372 }, { "epoch": 0.5, "learning_rate": 1.0514174051224169e-05, "loss": 1.1547, "step": 2373 }, { "epoch": 0.5, "learning_rate": 1.0507369687965583e-05, "loss": 1.2351, "step": 2374 }, { "epoch": 0.5, "learning_rate": 1.0500565089183628e-05, "loss": 1.2305, "step": 2375 }, { "epoch": 0.5, "learning_rate": 1.049376025803703e-05, "loss": 1.2251, "step": 2376 }, { "epoch": 0.5, "learning_rate": 1.048695519768463e-05, "loss": 1.3089, "step": 2377 }, { "epoch": 0.5, "learning_rate": 1.0480149911285364e-05, "loss": 1.2303, "step": 2378 }, { "epoch": 0.5, "learning_rate": 1.0473344401998275e-05, "loss": 1.2631, "step": 2379 }, { "epoch": 0.5, "learning_rate": 1.0466538672982516e-05, "loss": 1.2673, "step": 2380 }, { "epoch": 0.5, "learning_rate": 1.0459732727397336e-05, "loss": 1.2353, "step": 2381 }, { "epoch": 0.5, "learning_rate": 1.0452926568402087e-05, "loss": 1.2388, "step": 2382 }, { "epoch": 0.5, "learning_rate": 1.044612019915622e-05, "loss": 1.2381, "step": 2383 }, { "epoch": 0.5, "learning_rate": 1.0439313622819285e-05, "loss": 1.2566, "step": 2384 }, { "epoch": 0.5, "learning_rate": 1.0432506842550922e-05, "loss": 1.25, "step": 2385 }, { "epoch": 0.5, "learning_rate": 1.0425699861510875e-05, "loss": 1.2588, "step": 2386 }, { "epoch": 0.5, "learning_rate": 1.0418892682858974e-05, "loss": 1.2392, "step": 2387 }, { "epoch": 0.5, "learning_rate": 1.0412085309755141e-05, "loss": 1.2244, "step": 2388 }, { "epoch": 0.5, "learning_rate": 1.0405277745359394e-05, "loss": 1.2105, "step": 2389 }, { "epoch": 0.5, "learning_rate": 1.0398469992831833e-05, "loss": 1.2292, "step": 2390 }, { "epoch": 0.5, "learning_rate": 1.0391662055332653e-05, "loss": 1.2237, "step": 2391 }, { "epoch": 0.5, "learning_rate": 1.0384853936022121e-05, "loss": 1.2302, "step": 2392 }, { "epoch": 0.5, "learning_rate": 1.0378045638060606e-05, "loss": 1.272, "step": 2393 }, { "epoch": 0.5, "learning_rate": 1.0371237164608547e-05, "loss": 1.2702, "step": 2394 }, { "epoch": 0.5, "learning_rate": 1.0364428518826473e-05, "loss": 1.2416, "step": 2395 }, { "epoch": 0.5, "learning_rate": 1.0357619703874986e-05, "loss": 1.2597, "step": 2396 }, { "epoch": 0.5, "learning_rate": 1.035081072291477e-05, "loss": 1.2303, "step": 2397 }, { "epoch": 0.5, "learning_rate": 1.0344001579106586e-05, "loss": 1.284, "step": 2398 }, { "epoch": 0.5, "learning_rate": 1.0337192275611275e-05, "loss": 1.2511, "step": 2399 }, { "epoch": 0.5, "learning_rate": 1.0330382815589737e-05, "loss": 1.2032, "step": 2400 }, { "epoch": 0.51, "learning_rate": 1.0323573202202962e-05, "loss": 1.2033, "step": 2401 }, { "epoch": 0.51, "learning_rate": 1.0316763438612003e-05, "loss": 1.2255, "step": 2402 }, { "epoch": 0.51, "learning_rate": 1.0309953527977985e-05, "loss": 1.1898, "step": 2403 }, { "epoch": 0.51, "learning_rate": 1.0303143473462096e-05, "loss": 1.246, "step": 2404 }, { "epoch": 0.51, "learning_rate": 1.0296333278225599e-05, "loss": 1.2466, "step": 2405 }, { "epoch": 0.51, "learning_rate": 1.0289522945429818e-05, "loss": 1.2269, "step": 2406 }, { "epoch": 0.51, "learning_rate": 1.0282712478236139e-05, "loss": 1.2831, "step": 2407 }, { "epoch": 0.51, "learning_rate": 1.0275901879806014e-05, "loss": 1.2472, "step": 2408 }, { "epoch": 0.51, "learning_rate": 1.0269091153300954e-05, "loss": 1.2122, "step": 2409 }, { "epoch": 0.51, "learning_rate": 1.026228030188253e-05, "loss": 1.2544, "step": 2410 }, { "epoch": 0.51, "learning_rate": 1.025546932871237e-05, "loss": 1.1934, "step": 2411 }, { "epoch": 0.51, "learning_rate": 1.0248658236952162e-05, "loss": 1.2475, "step": 2412 }, { "epoch": 0.51, "learning_rate": 1.0241847029763644e-05, "loss": 1.2237, "step": 2413 }, { "epoch": 0.51, "learning_rate": 1.0235035710308611e-05, "loss": 1.2194, "step": 2414 }, { "epoch": 0.51, "learning_rate": 1.022822428174891e-05, "loss": 1.2196, "step": 2415 }, { "epoch": 0.51, "learning_rate": 1.0221412747246438e-05, "loss": 1.2166, "step": 2416 }, { "epoch": 0.51, "learning_rate": 1.0214601109963137e-05, "loss": 1.2111, "step": 2417 }, { "epoch": 0.51, "learning_rate": 1.0207789373061004e-05, "loss": 1.2607, "step": 2418 }, { "epoch": 0.51, "learning_rate": 1.0200977539702077e-05, "loss": 1.2253, "step": 2419 }, { "epoch": 0.51, "learning_rate": 1.0194165613048443e-05, "loss": 1.2612, "step": 2420 }, { "epoch": 0.51, "learning_rate": 1.0187353596262228e-05, "loss": 1.1862, "step": 2421 }, { "epoch": 0.51, "learning_rate": 1.0180541492505605e-05, "loss": 1.2951, "step": 2422 }, { "epoch": 0.51, "learning_rate": 1.017372930494078e-05, "loss": 1.1637, "step": 2423 }, { "epoch": 0.51, "learning_rate": 1.0166917036730009e-05, "loss": 1.2037, "step": 2424 }, { "epoch": 0.51, "learning_rate": 1.016010469103557e-05, "loss": 1.2007, "step": 2425 }, { "epoch": 0.51, "learning_rate": 1.015329227101979e-05, "loss": 1.1965, "step": 2426 }, { "epoch": 0.51, "learning_rate": 1.0146479779845024e-05, "loss": 1.2523, "step": 2427 }, { "epoch": 0.51, "learning_rate": 1.0139667220673665e-05, "loss": 1.2139, "step": 2428 }, { "epoch": 0.51, "learning_rate": 1.0132854596668133e-05, "loss": 1.2634, "step": 2429 }, { "epoch": 0.51, "learning_rate": 1.012604191099088e-05, "loss": 1.2344, "step": 2430 }, { "epoch": 0.51, "learning_rate": 1.0119229166804383e-05, "loss": 1.2879, "step": 2431 }, { "epoch": 0.51, "learning_rate": 1.0112416367271157e-05, "loss": 1.1895, "step": 2432 }, { "epoch": 0.51, "learning_rate": 1.0105603515553729e-05, "loss": 1.2212, "step": 2433 }, { "epoch": 0.51, "learning_rate": 1.0098790614814658e-05, "loss": 1.1932, "step": 2434 }, { "epoch": 0.51, "learning_rate": 1.0091977668216524e-05, "loss": 1.2289, "step": 2435 }, { "epoch": 0.51, "learning_rate": 1.008516467892193e-05, "loss": 1.2199, "step": 2436 }, { "epoch": 0.51, "learning_rate": 1.0078351650093499e-05, "loss": 1.2511, "step": 2437 }, { "epoch": 0.51, "learning_rate": 1.0071538584893862e-05, "loss": 1.2096, "step": 2438 }, { "epoch": 0.51, "learning_rate": 1.0064725486485684e-05, "loss": 1.231, "step": 2439 }, { "epoch": 0.51, "learning_rate": 1.0057912358031637e-05, "loss": 1.2033, "step": 2440 }, { "epoch": 0.51, "learning_rate": 1.0051099202694403e-05, "loss": 1.2364, "step": 2441 }, { "epoch": 0.51, "learning_rate": 1.004428602363668e-05, "loss": 1.1945, "step": 2442 }, { "epoch": 0.51, "learning_rate": 1.0037472824021176e-05, "loss": 1.2333, "step": 2443 }, { "epoch": 0.51, "learning_rate": 1.0030659607010618e-05, "loss": 1.1726, "step": 2444 }, { "epoch": 0.51, "learning_rate": 1.0023846375767727e-05, "loss": 1.2269, "step": 2445 }, { "epoch": 0.51, "learning_rate": 1.0017033133455238e-05, "loss": 1.1918, "step": 2446 }, { "epoch": 0.51, "learning_rate": 1.0010219883235887e-05, "loss": 1.2373, "step": 2447 }, { "epoch": 0.51, "learning_rate": 1.000340662827242e-05, "loss": 1.1908, "step": 2448 }, { "epoch": 0.52, "learning_rate": 9.99659337172758e-06, "loss": 1.2223, "step": 2449 }, { "epoch": 0.52, "learning_rate": 9.989780116764114e-06, "loss": 1.1779, "step": 2450 }, { "epoch": 0.52, "learning_rate": 9.982966866544767e-06, "loss": 1.2027, "step": 2451 }, { "epoch": 0.52, "learning_rate": 9.976153624232275e-06, "loss": 1.168, "step": 2452 }, { "epoch": 0.52, "learning_rate": 9.969340392989385e-06, "loss": 1.2472, "step": 2453 }, { "epoch": 0.52, "learning_rate": 9.962527175978824e-06, "loss": 1.2546, "step": 2454 }, { "epoch": 0.52, "learning_rate": 9.955713976363322e-06, "loss": 1.2341, "step": 2455 }, { "epoch": 0.52, "learning_rate": 9.948900797305602e-06, "loss": 1.1793, "step": 2456 }, { "epoch": 0.52, "learning_rate": 9.942087641968365e-06, "loss": 1.1833, "step": 2457 }, { "epoch": 0.52, "learning_rate": 9.935274513514317e-06, "loss": 1.2088, "step": 2458 }, { "epoch": 0.52, "learning_rate": 9.928461415106143e-06, "loss": 1.2319, "step": 2459 }, { "epoch": 0.52, "learning_rate": 9.921648349906505e-06, "loss": 1.1651, "step": 2460 }, { "epoch": 0.52, "learning_rate": 9.914835321078074e-06, "loss": 1.2286, "step": 2461 }, { "epoch": 0.52, "learning_rate": 9.908022331783475e-06, "loss": 1.2334, "step": 2462 }, { "epoch": 0.52, "learning_rate": 9.901209385185345e-06, "loss": 1.1385, "step": 2463 }, { "epoch": 0.52, "learning_rate": 9.894396484446276e-06, "loss": 1.224, "step": 2464 }, { "epoch": 0.52, "learning_rate": 9.887583632728845e-06, "loss": 1.2489, "step": 2465 }, { "epoch": 0.52, "learning_rate": 9.88077083319562e-06, "loss": 1.17, "step": 2466 }, { "epoch": 0.52, "learning_rate": 9.873958089009123e-06, "loss": 1.2075, "step": 2467 }, { "epoch": 0.52, "learning_rate": 9.867145403331869e-06, "loss": 1.2173, "step": 2468 }, { "epoch": 0.52, "learning_rate": 9.860332779326338e-06, "loss": 1.2064, "step": 2469 }, { "epoch": 0.52, "learning_rate": 9.853520220154976e-06, "loss": 1.2335, "step": 2470 }, { "epoch": 0.52, "learning_rate": 9.846707728980213e-06, "loss": 1.2241, "step": 2471 }, { "epoch": 0.52, "learning_rate": 9.839895308964436e-06, "loss": 1.194, "step": 2472 }, { "epoch": 0.52, "learning_rate": 9.833082963269994e-06, "loss": 1.2139, "step": 2473 }, { "epoch": 0.52, "learning_rate": 9.826270695059222e-06, "loss": 1.2209, "step": 2474 }, { "epoch": 0.52, "learning_rate": 9.819458507494395e-06, "loss": 1.1931, "step": 2475 }, { "epoch": 0.52, "learning_rate": 9.812646403737774e-06, "loss": 1.19, "step": 2476 }, { "epoch": 0.52, "learning_rate": 9.805834386951562e-06, "loss": 1.2181, "step": 2477 }, { "epoch": 0.52, "learning_rate": 9.799022460297926e-06, "loss": 1.1677, "step": 2478 }, { "epoch": 0.52, "learning_rate": 9.792210626939e-06, "loss": 1.2698, "step": 2479 }, { "epoch": 0.52, "learning_rate": 9.785398890036868e-06, "loss": 1.2253, "step": 2480 }, { "epoch": 0.52, "learning_rate": 9.778587252753565e-06, "loss": 1.2588, "step": 2481 }, { "epoch": 0.52, "learning_rate": 9.771775718251093e-06, "loss": 1.2323, "step": 2482 }, { "epoch": 0.52, "learning_rate": 9.764964289691389e-06, "loss": 1.2086, "step": 2483 }, { "epoch": 0.52, "learning_rate": 9.758152970236357e-06, "loss": 1.2505, "step": 2484 }, { "epoch": 0.52, "learning_rate": 9.751341763047841e-06, "loss": 1.2708, "step": 2485 }, { "epoch": 0.52, "learning_rate": 9.74453067128763e-06, "loss": 1.2137, "step": 2486 }, { "epoch": 0.52, "learning_rate": 9.737719698117474e-06, "loss": 1.2244, "step": 2487 }, { "epoch": 0.52, "learning_rate": 9.730908846699052e-06, "loss": 1.2694, "step": 2488 }, { "epoch": 0.52, "learning_rate": 9.72409812019399e-06, "loss": 1.1824, "step": 2489 }, { "epoch": 0.52, "learning_rate": 9.717287521763865e-06, "loss": 1.1987, "step": 2490 }, { "epoch": 0.52, "learning_rate": 9.710477054570186e-06, "loss": 1.1975, "step": 2491 }, { "epoch": 0.52, "learning_rate": 9.703666721774403e-06, "loss": 1.267, "step": 2492 }, { "epoch": 0.52, "learning_rate": 9.696856526537908e-06, "loss": 1.1996, "step": 2493 }, { "epoch": 0.52, "learning_rate": 9.690046472022018e-06, "loss": 1.2505, "step": 2494 }, { "epoch": 0.52, "learning_rate": 9.683236561388e-06, "loss": 1.1913, "step": 2495 }, { "epoch": 0.53, "learning_rate": 9.676426797797038e-06, "loss": 1.2085, "step": 2496 }, { "epoch": 0.53, "learning_rate": 9.669617184410267e-06, "loss": 1.2295, "step": 2497 }, { "epoch": 0.53, "learning_rate": 9.66280772438873e-06, "loss": 1.1922, "step": 2498 }, { "epoch": 0.53, "learning_rate": 9.655998420893412e-06, "loss": 1.2286, "step": 2499 }, { "epoch": 0.53, "learning_rate": 9.649189277085232e-06, "loss": 1.1887, "step": 2500 }, { "epoch": 0.53, "learning_rate": 9.642380296125019e-06, "loss": 1.2253, "step": 2501 }, { "epoch": 0.53, "learning_rate": 9.635571481173529e-06, "loss": 1.2251, "step": 2502 }, { "epoch": 0.53, "learning_rate": 9.628762835391455e-06, "loss": 1.1981, "step": 2503 }, { "epoch": 0.53, "learning_rate": 9.621954361939395e-06, "loss": 1.2436, "step": 2504 }, { "epoch": 0.53, "learning_rate": 9.61514606397788e-06, "loss": 1.2033, "step": 2505 }, { "epoch": 0.53, "learning_rate": 9.608337944667352e-06, "loss": 1.2431, "step": 2506 }, { "epoch": 0.53, "learning_rate": 9.601530007168167e-06, "loss": 1.1899, "step": 2507 }, { "epoch": 0.53, "learning_rate": 9.594722254640608e-06, "loss": 1.2517, "step": 2508 }, { "epoch": 0.53, "learning_rate": 9.587914690244864e-06, "loss": 1.1841, "step": 2509 }, { "epoch": 0.53, "learning_rate": 9.581107317141027e-06, "loss": 1.2179, "step": 2510 }, { "epoch": 0.53, "learning_rate": 9.574300138489127e-06, "loss": 1.2149, "step": 2511 }, { "epoch": 0.53, "learning_rate": 9.567493157449078e-06, "loss": 1.2527, "step": 2512 }, { "epoch": 0.53, "learning_rate": 9.560686377180719e-06, "loss": 1.2253, "step": 2513 }, { "epoch": 0.53, "learning_rate": 9.553879800843781e-06, "loss": 1.1917, "step": 2514 }, { "epoch": 0.53, "learning_rate": 9.547073431597915e-06, "loss": 1.2176, "step": 2515 }, { "epoch": 0.53, "learning_rate": 9.540267272602667e-06, "loss": 1.2293, "step": 2516 }, { "epoch": 0.53, "learning_rate": 9.533461327017487e-06, "loss": 1.2237, "step": 2517 }, { "epoch": 0.53, "learning_rate": 9.526655598001729e-06, "loss": 1.256, "step": 2518 }, { "epoch": 0.53, "learning_rate": 9.519850088714641e-06, "loss": 1.1881, "step": 2519 }, { "epoch": 0.53, "learning_rate": 9.513044802315371e-06, "loss": 1.2663, "step": 2520 }, { "epoch": 0.53, "learning_rate": 9.506239741962971e-06, "loss": 1.1681, "step": 2521 }, { "epoch": 0.53, "learning_rate": 9.499434910816379e-06, "loss": 1.2508, "step": 2522 }, { "epoch": 0.53, "learning_rate": 9.49263031203442e-06, "loss": 1.2559, "step": 2523 }, { "epoch": 0.53, "learning_rate": 9.485825948775836e-06, "loss": 1.2464, "step": 2524 }, { "epoch": 0.53, "learning_rate": 9.479021824199229e-06, "loss": 1.219, "step": 2525 }, { "epoch": 0.53, "learning_rate": 9.472217941463118e-06, "loss": 1.2477, "step": 2526 }, { "epoch": 0.53, "learning_rate": 9.46541430372589e-06, "loss": 1.1372, "step": 2527 }, { "epoch": 0.53, "learning_rate": 9.458610914145826e-06, "loss": 1.2356, "step": 2528 }, { "epoch": 0.53, "learning_rate": 9.451807775881091e-06, "loss": 1.2531, "step": 2529 }, { "epoch": 0.53, "learning_rate": 9.445004892089738e-06, "loss": 1.2386, "step": 2530 }, { "epoch": 0.53, "learning_rate": 9.438202265929688e-06, "loss": 1.2734, "step": 2531 }, { "epoch": 0.53, "learning_rate": 9.431399900558761e-06, "loss": 1.2366, "step": 2532 }, { "epoch": 0.53, "learning_rate": 9.42459779913464e-06, "loss": 1.2334, "step": 2533 }, { "epoch": 0.53, "learning_rate": 9.417795964814897e-06, "loss": 1.2326, "step": 2534 }, { "epoch": 0.53, "learning_rate": 9.410994400756975e-06, "loss": 1.1743, "step": 2535 }, { "epoch": 0.53, "learning_rate": 9.404193110118183e-06, "loss": 1.2306, "step": 2536 }, { "epoch": 0.53, "learning_rate": 9.397392096055724e-06, "loss": 1.247, "step": 2537 }, { "epoch": 0.53, "learning_rate": 9.390591361726652e-06, "loss": 1.2326, "step": 2538 }, { "epoch": 0.53, "learning_rate": 9.383790910287902e-06, "loss": 1.1915, "step": 2539 }, { "epoch": 0.53, "learning_rate": 9.376990744896276e-06, "loss": 1.2038, "step": 2540 }, { "epoch": 0.53, "learning_rate": 9.37019086870844e-06, "loss": 1.2227, "step": 2541 }, { "epoch": 0.53, "learning_rate": 9.363391284880932e-06, "loss": 1.1806, "step": 2542 }, { "epoch": 0.53, "learning_rate": 9.35659199657015e-06, "loss": 1.2223, "step": 2543 }, { "epoch": 0.54, "learning_rate": 9.349793006932347e-06, "loss": 1.2226, "step": 2544 }, { "epoch": 0.54, "learning_rate": 9.342994319123658e-06, "loss": 1.2456, "step": 2545 }, { "epoch": 0.54, "learning_rate": 9.336195936300054e-06, "loss": 1.1272, "step": 2546 }, { "epoch": 0.54, "learning_rate": 9.329397861617387e-06, "loss": 1.1981, "step": 2547 }, { "epoch": 0.54, "learning_rate": 9.322600098231349e-06, "loss": 1.169, "step": 2548 }, { "epoch": 0.54, "learning_rate": 9.315802649297489e-06, "loss": 1.2386, "step": 2549 }, { "epoch": 0.54, "learning_rate": 9.309005517971222e-06, "loss": 1.2277, "step": 2550 }, { "epoch": 0.54, "learning_rate": 9.302208707407805e-06, "loss": 1.2272, "step": 2551 }, { "epoch": 0.54, "learning_rate": 9.295412220762348e-06, "loss": 1.196, "step": 2552 }, { "epoch": 0.54, "learning_rate": 9.288616061189812e-06, "loss": 1.2258, "step": 2553 }, { "epoch": 0.54, "learning_rate": 9.281820231845008e-06, "loss": 1.2002, "step": 2554 }, { "epoch": 0.54, "learning_rate": 9.275024735882588e-06, "loss": 1.1992, "step": 2555 }, { "epoch": 0.54, "learning_rate": 9.26822957645706e-06, "loss": 1.2596, "step": 2556 }, { "epoch": 0.54, "learning_rate": 9.261434756722757e-06, "loss": 1.2529, "step": 2557 }, { "epoch": 0.54, "learning_rate": 9.254640279833876e-06, "loss": 1.219, "step": 2558 }, { "epoch": 0.54, "learning_rate": 9.247846148944442e-06, "loss": 1.2136, "step": 2559 }, { "epoch": 0.54, "learning_rate": 9.24105236720832e-06, "loss": 1.2217, "step": 2560 }, { "epoch": 0.54, "learning_rate": 9.234258937779216e-06, "loss": 1.2573, "step": 2561 }, { "epoch": 0.54, "learning_rate": 9.22746586381067e-06, "loss": 1.2715, "step": 2562 }, { "epoch": 0.54, "learning_rate": 9.220673148456064e-06, "loss": 1.2174, "step": 2563 }, { "epoch": 0.54, "learning_rate": 9.213880794868601e-06, "loss": 1.2118, "step": 2564 }, { "epoch": 0.54, "learning_rate": 9.207088806201328e-06, "loss": 1.2737, "step": 2565 }, { "epoch": 0.54, "learning_rate": 9.200297185607117e-06, "loss": 1.2422, "step": 2566 }, { "epoch": 0.54, "learning_rate": 9.193505936238669e-06, "loss": 1.2037, "step": 2567 }, { "epoch": 0.54, "learning_rate": 9.186715061248511e-06, "loss": 1.2489, "step": 2568 }, { "epoch": 0.54, "learning_rate": 9.179924563789005e-06, "loss": 1.2198, "step": 2569 }, { "epoch": 0.54, "learning_rate": 9.173134447012322e-06, "loss": 1.2462, "step": 2570 }, { "epoch": 0.54, "learning_rate": 9.166344714070478e-06, "loss": 1.2093, "step": 2571 }, { "epoch": 0.54, "learning_rate": 9.15955536811529e-06, "loss": 1.2295, "step": 2572 }, { "epoch": 0.54, "learning_rate": 9.152766412298405e-06, "loss": 1.2453, "step": 2573 }, { "epoch": 0.54, "learning_rate": 9.145977849771292e-06, "loss": 1.2249, "step": 2574 }, { "epoch": 0.54, "learning_rate": 9.139189683685224e-06, "loss": 1.2275, "step": 2575 }, { "epoch": 0.54, "learning_rate": 9.132401917191311e-06, "loss": 1.2516, "step": 2576 }, { "epoch": 0.54, "learning_rate": 9.125614553440458e-06, "loss": 1.2454, "step": 2577 }, { "epoch": 0.54, "learning_rate": 9.118827595583392e-06, "loss": 1.2541, "step": 2578 }, { "epoch": 0.54, "learning_rate": 9.112041046770653e-06, "loss": 1.2008, "step": 2579 }, { "epoch": 0.54, "learning_rate": 9.105254910152587e-06, "loss": 1.2035, "step": 2580 }, { "epoch": 0.54, "learning_rate": 9.098469188879348e-06, "loss": 1.197, "step": 2581 }, { "epoch": 0.54, "learning_rate": 9.091683886100905e-06, "loss": 1.1965, "step": 2582 }, { "epoch": 0.54, "learning_rate": 9.084899004967019e-06, "loss": 1.2451, "step": 2583 }, { "epoch": 0.54, "learning_rate": 9.07811454862727e-06, "loss": 1.2859, "step": 2584 }, { "epoch": 0.54, "learning_rate": 9.071330520231034e-06, "loss": 1.2214, "step": 2585 }, { "epoch": 0.54, "learning_rate": 9.064546922927482e-06, "loss": 1.2259, "step": 2586 }, { "epoch": 0.54, "learning_rate": 9.0577637598656e-06, "loss": 1.2108, "step": 2587 }, { "epoch": 0.54, "learning_rate": 9.050981034194158e-06, "loss": 1.1767, "step": 2588 }, { "epoch": 0.54, "learning_rate": 9.04419874906173e-06, "loss": 1.2149, "step": 2589 }, { "epoch": 0.54, "learning_rate": 9.037416907616685e-06, "loss": 1.2189, "step": 2590 }, { "epoch": 0.54, "learning_rate": 9.030635513007182e-06, "loss": 1.2228, "step": 2591 }, { "epoch": 0.55, "learning_rate": 9.023854568381184e-06, "loss": 1.1984, "step": 2592 }, { "epoch": 0.55, "learning_rate": 9.017074076886432e-06, "loss": 1.1838, "step": 2593 }, { "epoch": 0.55, "learning_rate": 9.010294041670456e-06, "loss": 1.2448, "step": 2594 }, { "epoch": 0.55, "learning_rate": 9.003514465880592e-06, "loss": 1.2261, "step": 2595 }, { "epoch": 0.55, "learning_rate": 8.996735352663938e-06, "loss": 1.2322, "step": 2596 }, { "epoch": 0.55, "learning_rate": 8.9899567051674e-06, "loss": 1.1869, "step": 2597 }, { "epoch": 0.55, "learning_rate": 8.983178526537656e-06, "loss": 1.2563, "step": 2598 }, { "epoch": 0.55, "learning_rate": 8.97640081992116e-06, "loss": 1.2242, "step": 2599 }, { "epoch": 0.55, "learning_rate": 8.969623588464164e-06, "loss": 1.2283, "step": 2600 }, { "epoch": 0.55, "learning_rate": 8.962846835312685e-06, "loss": 1.2285, "step": 2601 }, { "epoch": 0.55, "learning_rate": 8.956070563612525e-06, "loss": 1.2205, "step": 2602 }, { "epoch": 0.55, "learning_rate": 8.949294776509261e-06, "loss": 1.2399, "step": 2603 }, { "epoch": 0.55, "learning_rate": 8.942519477148245e-06, "loss": 1.1809, "step": 2604 }, { "epoch": 0.55, "learning_rate": 8.935744668674604e-06, "loss": 1.2304, "step": 2605 }, { "epoch": 0.55, "learning_rate": 8.928970354233236e-06, "loss": 1.1932, "step": 2606 }, { "epoch": 0.55, "learning_rate": 8.9221965369688e-06, "loss": 1.227, "step": 2607 }, { "epoch": 0.55, "learning_rate": 8.915423220025747e-06, "loss": 1.1954, "step": 2608 }, { "epoch": 0.55, "learning_rate": 8.908650406548274e-06, "loss": 1.2214, "step": 2609 }, { "epoch": 0.55, "learning_rate": 8.901878099680352e-06, "loss": 1.2101, "step": 2610 }, { "epoch": 0.55, "learning_rate": 8.895106302565724e-06, "loss": 1.2212, "step": 2611 }, { "epoch": 0.55, "learning_rate": 8.888335018347879e-06, "loss": 1.2644, "step": 2612 }, { "epoch": 0.55, "learning_rate": 8.88156425017009e-06, "loss": 1.2389, "step": 2613 }, { "epoch": 0.55, "learning_rate": 8.874794001175374e-06, "loss": 1.2448, "step": 2614 }, { "epoch": 0.55, "learning_rate": 8.868024274506506e-06, "loss": 1.2393, "step": 2615 }, { "epoch": 0.55, "learning_rate": 8.861255073306037e-06, "loss": 1.1556, "step": 2616 }, { "epoch": 0.55, "learning_rate": 8.854486400716256e-06, "loss": 1.1791, "step": 2617 }, { "epoch": 0.55, "learning_rate": 8.847718259879206e-06, "loss": 1.2088, "step": 2618 }, { "epoch": 0.55, "learning_rate": 8.8409506539367e-06, "loss": 1.1806, "step": 2619 }, { "epoch": 0.55, "learning_rate": 8.834183586030281e-06, "loss": 1.2505, "step": 2620 }, { "epoch": 0.55, "learning_rate": 8.827417059301266e-06, "loss": 1.2754, "step": 2621 }, { "epoch": 0.55, "learning_rate": 8.820651076890701e-06, "loss": 1.2274, "step": 2622 }, { "epoch": 0.55, "learning_rate": 8.813885641939382e-06, "loss": 1.2165, "step": 2623 }, { "epoch": 0.55, "learning_rate": 8.807120757587864e-06, "loss": 1.2553, "step": 2624 }, { "epoch": 0.55, "learning_rate": 8.800356426976431e-06, "loss": 1.2243, "step": 2625 }, { "epoch": 0.55, "learning_rate": 8.793592653245122e-06, "loss": 1.1955, "step": 2626 }, { "epoch": 0.55, "learning_rate": 8.786829439533712e-06, "loss": 1.2402, "step": 2627 }, { "epoch": 0.55, "learning_rate": 8.780066788981711e-06, "loss": 1.2093, "step": 2628 }, { "epoch": 0.55, "learning_rate": 8.77330470472838e-06, "loss": 1.2009, "step": 2629 }, { "epoch": 0.55, "learning_rate": 8.766543189912706e-06, "loss": 1.1832, "step": 2630 }, { "epoch": 0.55, "learning_rate": 8.75978224767341e-06, "loss": 1.2115, "step": 2631 }, { "epoch": 0.55, "learning_rate": 8.753021881148966e-06, "loss": 1.2496, "step": 2632 }, { "epoch": 0.55, "learning_rate": 8.746262093477554e-06, "loss": 1.2159, "step": 2633 }, { "epoch": 0.55, "learning_rate": 8.739502887797108e-06, "loss": 1.2509, "step": 2634 }, { "epoch": 0.55, "learning_rate": 8.73274426724528e-06, "loss": 1.2146, "step": 2635 }, { "epoch": 0.55, "learning_rate": 8.72598623495945e-06, "loss": 1.225, "step": 2636 }, { "epoch": 0.55, "learning_rate": 8.719228794076733e-06, "loss": 1.1484, "step": 2637 }, { "epoch": 0.55, "learning_rate": 8.71247194773396e-06, "loss": 1.1622, "step": 2638 }, { "epoch": 0.56, "learning_rate": 8.705715699067692e-06, "loss": 1.228, "step": 2639 }, { "epoch": 0.56, "learning_rate": 8.698960051214216e-06, "loss": 1.1872, "step": 2640 }, { "epoch": 0.56, "learning_rate": 8.692205007309525e-06, "loss": 1.2548, "step": 2641 }, { "epoch": 0.56, "learning_rate": 8.685450570489353e-06, "loss": 1.2366, "step": 2642 }, { "epoch": 0.56, "learning_rate": 8.678696743889135e-06, "loss": 1.2267, "step": 2643 }, { "epoch": 0.56, "learning_rate": 8.671943530644025e-06, "loss": 1.2908, "step": 2644 }, { "epoch": 0.56, "learning_rate": 8.665190933888904e-06, "loss": 1.173, "step": 2645 }, { "epoch": 0.56, "learning_rate": 8.658438956758355e-06, "loss": 1.2183, "step": 2646 }, { "epoch": 0.56, "learning_rate": 8.651687602386674e-06, "loss": 1.1995, "step": 2647 }, { "epoch": 0.56, "learning_rate": 8.644936873907877e-06, "loss": 1.1884, "step": 2648 }, { "epoch": 0.56, "learning_rate": 8.638186774455678e-06, "loss": 1.1741, "step": 2649 }, { "epoch": 0.56, "learning_rate": 8.631437307163508e-06, "loss": 1.1794, "step": 2650 }, { "epoch": 0.56, "learning_rate": 8.624688475164502e-06, "loss": 1.1828, "step": 2651 }, { "epoch": 0.56, "learning_rate": 8.617940281591494e-06, "loss": 1.1328, "step": 2652 }, { "epoch": 0.56, "learning_rate": 8.611192729577036e-06, "loss": 1.2695, "step": 2653 }, { "epoch": 0.56, "learning_rate": 8.60444582225336e-06, "loss": 1.1538, "step": 2654 }, { "epoch": 0.56, "learning_rate": 8.597699562752427e-06, "loss": 1.2559, "step": 2655 }, { "epoch": 0.56, "learning_rate": 8.590953954205872e-06, "loss": 1.1956, "step": 2656 }, { "epoch": 0.56, "learning_rate": 8.584208999745036e-06, "loss": 1.2334, "step": 2657 }, { "epoch": 0.56, "learning_rate": 8.577464702500968e-06, "loss": 1.1937, "step": 2658 }, { "epoch": 0.56, "learning_rate": 8.570721065604394e-06, "loss": 1.2482, "step": 2659 }, { "epoch": 0.56, "learning_rate": 8.56397809218574e-06, "loss": 1.2383, "step": 2660 }, { "epoch": 0.56, "learning_rate": 8.557235785375133e-06, "loss": 1.2145, "step": 2661 }, { "epoch": 0.56, "learning_rate": 8.550494148302378e-06, "loss": 1.1907, "step": 2662 }, { "epoch": 0.56, "learning_rate": 8.543753184096976e-06, "loss": 1.1843, "step": 2663 }, { "epoch": 0.56, "learning_rate": 8.537012895888116e-06, "loss": 1.218, "step": 2664 }, { "epoch": 0.56, "learning_rate": 8.530273286804665e-06, "loss": 1.2512, "step": 2665 }, { "epoch": 0.56, "learning_rate": 8.52353435997519e-06, "loss": 1.2178, "step": 2666 }, { "epoch": 0.56, "learning_rate": 8.516796118527924e-06, "loss": 1.2324, "step": 2667 }, { "epoch": 0.56, "learning_rate": 8.510058565590792e-06, "loss": 1.2863, "step": 2668 }, { "epoch": 0.56, "learning_rate": 8.5033217042914e-06, "loss": 1.2398, "step": 2669 }, { "epoch": 0.56, "learning_rate": 8.496585537757027e-06, "loss": 1.2119, "step": 2670 }, { "epoch": 0.56, "learning_rate": 8.489850069114636e-06, "loss": 1.2241, "step": 2671 }, { "epoch": 0.56, "learning_rate": 8.483115301490864e-06, "loss": 1.2214, "step": 2672 }, { "epoch": 0.56, "learning_rate": 8.476381238012012e-06, "loss": 1.1735, "step": 2673 }, { "epoch": 0.56, "learning_rate": 8.469647881804073e-06, "loss": 1.1531, "step": 2674 }, { "epoch": 0.56, "learning_rate": 8.462915235992699e-06, "loss": 1.2491, "step": 2675 }, { "epoch": 0.56, "learning_rate": 8.456183303703217e-06, "loss": 1.2501, "step": 2676 }, { "epoch": 0.56, "learning_rate": 8.449452088060623e-06, "loss": 1.2046, "step": 2677 }, { "epoch": 0.56, "learning_rate": 8.44272159218957e-06, "loss": 1.2664, "step": 2678 }, { "epoch": 0.56, "learning_rate": 8.435991819214397e-06, "loss": 1.2805, "step": 2679 }, { "epoch": 0.56, "learning_rate": 8.42926277225909e-06, "loss": 1.2273, "step": 2680 }, { "epoch": 0.56, "learning_rate": 8.422534454447299e-06, "loss": 1.2179, "step": 2681 }, { "epoch": 0.56, "learning_rate": 8.41580686890235e-06, "loss": 1.2344, "step": 2682 }, { "epoch": 0.56, "learning_rate": 8.40908001874721e-06, "loss": 1.211, "step": 2683 }, { "epoch": 0.56, "learning_rate": 8.402353907104523e-06, "loss": 1.1908, "step": 2684 }, { "epoch": 0.56, "learning_rate": 8.395628537096578e-06, "loss": 1.2562, "step": 2685 }, { "epoch": 0.56, "learning_rate": 8.38890391184532e-06, "loss": 1.2229, "step": 2686 }, { "epoch": 0.57, "learning_rate": 8.382180034472353e-06, "loss": 1.1851, "step": 2687 }, { "epoch": 0.57, "learning_rate": 8.375456908098936e-06, "loss": 1.2684, "step": 2688 }, { "epoch": 0.57, "learning_rate": 8.368734535845968e-06, "loss": 1.2503, "step": 2689 }, { "epoch": 0.57, "learning_rate": 8.362012920834013e-06, "loss": 1.2217, "step": 2690 }, { "epoch": 0.57, "learning_rate": 8.355292066183271e-06, "loss": 1.2432, "step": 2691 }, { "epoch": 0.57, "learning_rate": 8.348571975013597e-06, "loss": 1.1877, "step": 2692 }, { "epoch": 0.57, "learning_rate": 8.341852650444489e-06, "loss": 1.2358, "step": 2693 }, { "epoch": 0.57, "learning_rate": 8.33513409559508e-06, "loss": 1.2771, "step": 2694 }, { "epoch": 0.57, "learning_rate": 8.328416313584169e-06, "loss": 1.2575, "step": 2695 }, { "epoch": 0.57, "learning_rate": 8.321699307530173e-06, "loss": 1.2217, "step": 2696 }, { "epoch": 0.57, "learning_rate": 8.314983080551152e-06, "loss": 1.1771, "step": 2697 }, { "epoch": 0.57, "learning_rate": 8.30826763576482e-06, "loss": 1.2408, "step": 2698 }, { "epoch": 0.57, "learning_rate": 8.301552976288512e-06, "loss": 1.2487, "step": 2699 }, { "epoch": 0.57, "learning_rate": 8.294839105239208e-06, "loss": 1.2644, "step": 2700 }, { "epoch": 0.57, "learning_rate": 8.288126025733515e-06, "loss": 1.1981, "step": 2701 }, { "epoch": 0.57, "learning_rate": 8.281413740887676e-06, "loss": 1.2385, "step": 2702 }, { "epoch": 0.57, "learning_rate": 8.274702253817567e-06, "loss": 1.2154, "step": 2703 }, { "epoch": 0.57, "learning_rate": 8.267991567638689e-06, "loss": 1.2853, "step": 2704 }, { "epoch": 0.57, "learning_rate": 8.261281685466178e-06, "loss": 1.2097, "step": 2705 }, { "epoch": 0.57, "learning_rate": 8.254572610414787e-06, "loss": 1.2273, "step": 2706 }, { "epoch": 0.57, "learning_rate": 8.2478643455989e-06, "loss": 1.2086, "step": 2707 }, { "epoch": 0.57, "learning_rate": 8.241156894132529e-06, "loss": 1.1881, "step": 2708 }, { "epoch": 0.57, "learning_rate": 8.2344502591293e-06, "loss": 1.2345, "step": 2709 }, { "epoch": 0.57, "learning_rate": 8.227744443702468e-06, "loss": 1.227, "step": 2710 }, { "epoch": 0.57, "learning_rate": 8.221039450964901e-06, "loss": 1.246, "step": 2711 }, { "epoch": 0.57, "learning_rate": 8.214335284029088e-06, "loss": 1.2204, "step": 2712 }, { "epoch": 0.57, "learning_rate": 8.207631946007135e-06, "loss": 1.1863, "step": 2713 }, { "epoch": 0.57, "learning_rate": 8.200929440010766e-06, "loss": 1.2416, "step": 2714 }, { "epoch": 0.57, "learning_rate": 8.194227769151306e-06, "loss": 1.2216, "step": 2715 }, { "epoch": 0.57, "learning_rate": 8.187526936539712e-06, "loss": 1.2361, "step": 2716 }, { "epoch": 0.57, "learning_rate": 8.180826945286539e-06, "loss": 1.2384, "step": 2717 }, { "epoch": 0.57, "learning_rate": 8.174127798501948e-06, "loss": 1.2386, "step": 2718 }, { "epoch": 0.57, "learning_rate": 8.16742949929572e-06, "loss": 1.1699, "step": 2719 }, { "epoch": 0.57, "learning_rate": 8.160732050777235e-06, "loss": 1.1396, "step": 2720 }, { "epoch": 0.57, "learning_rate": 8.154035456055482e-06, "loss": 1.195, "step": 2721 }, { "epoch": 0.57, "learning_rate": 8.147339718239047e-06, "loss": 1.2063, "step": 2722 }, { "epoch": 0.57, "learning_rate": 8.140644840436126e-06, "loss": 1.1577, "step": 2723 }, { "epoch": 0.57, "learning_rate": 8.133950825754511e-06, "loss": 1.2097, "step": 2724 }, { "epoch": 0.57, "learning_rate": 8.127257677301597e-06, "loss": 1.2012, "step": 2725 }, { "epoch": 0.57, "learning_rate": 8.120565398184368e-06, "loss": 1.2409, "step": 2726 }, { "epoch": 0.57, "learning_rate": 8.113873991509422e-06, "loss": 1.2502, "step": 2727 }, { "epoch": 0.57, "learning_rate": 8.107183460382928e-06, "loss": 1.2378, "step": 2728 }, { "epoch": 0.57, "learning_rate": 8.100493807910675e-06, "loss": 1.2136, "step": 2729 }, { "epoch": 0.57, "learning_rate": 8.093805037198022e-06, "loss": 1.2082, "step": 2730 }, { "epoch": 0.57, "learning_rate": 8.087117151349928e-06, "loss": 1.2376, "step": 2731 }, { "epoch": 0.57, "learning_rate": 8.080430153470945e-06, "loss": 1.2115, "step": 2732 }, { "epoch": 0.57, "learning_rate": 8.073744046665202e-06, "loss": 1.2417, "step": 2733 }, { "epoch": 0.58, "learning_rate": 8.067058834036429e-06, "loss": 1.2313, "step": 2734 }, { "epoch": 0.58, "learning_rate": 8.060374518687925e-06, "loss": 1.2377, "step": 2735 }, { "epoch": 0.58, "learning_rate": 8.053691103722585e-06, "loss": 1.2084, "step": 2736 }, { "epoch": 0.58, "learning_rate": 8.047008592242882e-06, "loss": 1.216, "step": 2737 }, { "epoch": 0.58, "learning_rate": 8.040326987350866e-06, "loss": 1.2322, "step": 2738 }, { "epoch": 0.58, "learning_rate": 8.033646292148167e-06, "loss": 1.2534, "step": 2739 }, { "epoch": 0.58, "learning_rate": 8.026966509736001e-06, "loss": 1.2011, "step": 2740 }, { "epoch": 0.58, "learning_rate": 8.020287643215147e-06, "loss": 1.2152, "step": 2741 }, { "epoch": 0.58, "learning_rate": 8.013609695685975e-06, "loss": 1.1857, "step": 2742 }, { "epoch": 0.58, "learning_rate": 8.006932670248413e-06, "loss": 1.2392, "step": 2743 }, { "epoch": 0.58, "learning_rate": 8.000256570001964e-06, "loss": 1.2247, "step": 2744 }, { "epoch": 0.58, "learning_rate": 7.993581398045714e-06, "loss": 1.2522, "step": 2745 }, { "epoch": 0.58, "learning_rate": 7.986907157478301e-06, "loss": 1.2336, "step": 2746 }, { "epoch": 0.58, "learning_rate": 7.98023385139794e-06, "loss": 1.2523, "step": 2747 }, { "epoch": 0.58, "learning_rate": 7.973561482902414e-06, "loss": 1.2711, "step": 2748 }, { "epoch": 0.58, "learning_rate": 7.966890055089064e-06, "loss": 1.238, "step": 2749 }, { "epoch": 0.58, "learning_rate": 7.9602195710548e-06, "loss": 1.2048, "step": 2750 }, { "epoch": 0.58, "learning_rate": 7.953550033896093e-06, "loss": 1.2273, "step": 2751 }, { "epoch": 0.58, "learning_rate": 7.946881446708966e-06, "loss": 1.2129, "step": 2752 }, { "epoch": 0.58, "learning_rate": 7.940213812589018e-06, "loss": 1.2219, "step": 2753 }, { "epoch": 0.58, "learning_rate": 7.933547134631386e-06, "loss": 1.2327, "step": 2754 }, { "epoch": 0.58, "learning_rate": 7.926881415930784e-06, "loss": 1.2143, "step": 2755 }, { "epoch": 0.58, "learning_rate": 7.920216659581462e-06, "loss": 1.2426, "step": 2756 }, { "epoch": 0.58, "learning_rate": 7.913552868677228e-06, "loss": 1.2561, "step": 2757 }, { "epoch": 0.58, "learning_rate": 7.906890046311453e-06, "loss": 1.2146, "step": 2758 }, { "epoch": 0.58, "learning_rate": 7.900228195577046e-06, "loss": 1.2221, "step": 2759 }, { "epoch": 0.58, "learning_rate": 7.893567319566468e-06, "loss": 1.2554, "step": 2760 }, { "epoch": 0.58, "learning_rate": 7.88690742137173e-06, "loss": 1.2215, "step": 2761 }, { "epoch": 0.58, "learning_rate": 7.880248504084385e-06, "loss": 1.221, "step": 2762 }, { "epoch": 0.58, "learning_rate": 7.87359057079554e-06, "loss": 1.2506, "step": 2763 }, { "epoch": 0.58, "learning_rate": 7.866933624595833e-06, "loss": 1.1696, "step": 2764 }, { "epoch": 0.58, "learning_rate": 7.86027766857545e-06, "loss": 1.2242, "step": 2765 }, { "epoch": 0.58, "learning_rate": 7.85362270582412e-06, "loss": 1.249, "step": 2766 }, { "epoch": 0.58, "learning_rate": 7.846968739431105e-06, "loss": 1.2258, "step": 2767 }, { "epoch": 0.58, "learning_rate": 7.840315772485203e-06, "loss": 1.2131, "step": 2768 }, { "epoch": 0.58, "learning_rate": 7.833663808074761e-06, "loss": 1.21, "step": 2769 }, { "epoch": 0.58, "learning_rate": 7.827012849287643e-06, "loss": 1.2674, "step": 2770 }, { "epoch": 0.58, "learning_rate": 7.820362899211263e-06, "loss": 1.1952, "step": 2771 }, { "epoch": 0.58, "learning_rate": 7.813713960932553e-06, "loss": 1.157, "step": 2772 }, { "epoch": 0.58, "learning_rate": 7.807066037537983e-06, "loss": 1.2472, "step": 2773 }, { "epoch": 0.58, "learning_rate": 7.800419132113551e-06, "loss": 1.2524, "step": 2774 }, { "epoch": 0.58, "learning_rate": 7.793773247744782e-06, "loss": 1.2224, "step": 2775 }, { "epoch": 0.58, "learning_rate": 7.78712838751672e-06, "loss": 1.1804, "step": 2776 }, { "epoch": 0.58, "learning_rate": 7.780484554513948e-06, "loss": 1.2094, "step": 2777 }, { "epoch": 0.58, "learning_rate": 7.773841751820558e-06, "loss": 1.2597, "step": 2778 }, { "epoch": 0.58, "learning_rate": 7.767199982520177e-06, "loss": 1.2218, "step": 2779 }, { "epoch": 0.58, "learning_rate": 7.76055924969594e-06, "loss": 1.2089, "step": 2780 }, { "epoch": 0.58, "learning_rate": 7.753919556430504e-06, "loss": 1.2424, "step": 2781 }, { "epoch": 0.59, "learning_rate": 7.747280905806051e-06, "loss": 1.2711, "step": 2782 }, { "epoch": 0.59, "learning_rate": 7.74064330090427e-06, "loss": 1.1915, "step": 2783 }, { "epoch": 0.59, "learning_rate": 7.734006744806371e-06, "loss": 1.1905, "step": 2784 }, { "epoch": 0.59, "learning_rate": 7.727371240593069e-06, "loss": 1.2748, "step": 2785 }, { "epoch": 0.59, "learning_rate": 7.720736791344599e-06, "loss": 1.2488, "step": 2786 }, { "epoch": 0.59, "learning_rate": 7.714103400140707e-06, "loss": 1.223, "step": 2787 }, { "epoch": 0.59, "learning_rate": 7.707471070060639e-06, "loss": 1.2325, "step": 2788 }, { "epoch": 0.59, "learning_rate": 7.70083980418315e-06, "loss": 1.2495, "step": 2789 }, { "epoch": 0.59, "learning_rate": 7.694209605586515e-06, "loss": 1.244, "step": 2790 }, { "epoch": 0.59, "learning_rate": 7.687580477348493e-06, "loss": 1.1885, "step": 2791 }, { "epoch": 0.59, "learning_rate": 7.680952422546365e-06, "loss": 1.2381, "step": 2792 }, { "epoch": 0.59, "learning_rate": 7.674325444256899e-06, "loss": 1.1696, "step": 2793 }, { "epoch": 0.59, "learning_rate": 7.667699545556366e-06, "loss": 1.1912, "step": 2794 }, { "epoch": 0.59, "learning_rate": 7.661074729520549e-06, "loss": 1.2232, "step": 2795 }, { "epoch": 0.59, "learning_rate": 7.654450999224708e-06, "loss": 1.192, "step": 2796 }, { "epoch": 0.59, "learning_rate": 7.647828357743613e-06, "loss": 1.2395, "step": 2797 }, { "epoch": 0.59, "learning_rate": 7.641206808151532e-06, "loss": 1.2369, "step": 2798 }, { "epoch": 0.59, "learning_rate": 7.634586353522206e-06, "loss": 1.1683, "step": 2799 }, { "epoch": 0.59, "learning_rate": 7.6279669969288925e-06, "loss": 1.2448, "step": 2800 }, { "epoch": 0.59, "learning_rate": 7.6213487414443246e-06, "loss": 1.209, "step": 2801 }, { "epoch": 0.59, "learning_rate": 7.614731590140721e-06, "loss": 1.2022, "step": 2802 }, { "epoch": 0.59, "learning_rate": 7.608115546089802e-06, "loss": 1.2089, "step": 2803 }, { "epoch": 0.59, "learning_rate": 7.601500612362762e-06, "loss": 1.1891, "step": 2804 }, { "epoch": 0.59, "learning_rate": 7.594886792030282e-06, "loss": 1.2565, "step": 2805 }, { "epoch": 0.59, "learning_rate": 7.588274088162535e-06, "loss": 1.2589, "step": 2806 }, { "epoch": 0.59, "learning_rate": 7.5816625038291595e-06, "loss": 1.233, "step": 2807 }, { "epoch": 0.59, "learning_rate": 7.5750520420992934e-06, "loss": 1.1788, "step": 2808 }, { "epoch": 0.59, "learning_rate": 7.568442706041537e-06, "loss": 1.2205, "step": 2809 }, { "epoch": 0.59, "learning_rate": 7.561834498723974e-06, "loss": 1.2346, "step": 2810 }, { "epoch": 0.59, "learning_rate": 7.555227423214174e-06, "loss": 1.221, "step": 2811 }, { "epoch": 0.59, "learning_rate": 7.548621482579161e-06, "loss": 1.2044, "step": 2812 }, { "epoch": 0.59, "learning_rate": 7.5420166798854535e-06, "loss": 1.1918, "step": 2813 }, { "epoch": 0.59, "learning_rate": 7.535413018199026e-06, "loss": 1.2197, "step": 2814 }, { "epoch": 0.59, "learning_rate": 7.528810500585325e-06, "loss": 1.1393, "step": 2815 }, { "epoch": 0.59, "learning_rate": 7.5222091301092796e-06, "loss": 1.2521, "step": 2816 }, { "epoch": 0.59, "learning_rate": 7.51560890983527e-06, "loss": 1.2612, "step": 2817 }, { "epoch": 0.59, "learning_rate": 7.5090098428271475e-06, "loss": 1.2324, "step": 2818 }, { "epoch": 0.59, "learning_rate": 7.502411932148236e-06, "loss": 1.196, "step": 2819 }, { "epoch": 0.59, "learning_rate": 7.4958151808613096e-06, "loss": 1.1821, "step": 2820 }, { "epoch": 0.59, "learning_rate": 7.489219592028616e-06, "loss": 1.157, "step": 2821 }, { "epoch": 0.59, "learning_rate": 7.482625168711859e-06, "loss": 1.1813, "step": 2822 }, { "epoch": 0.59, "learning_rate": 7.476031913972192e-06, "loss": 1.2496, "step": 2823 }, { "epoch": 0.59, "learning_rate": 7.4694398308702474e-06, "loss": 1.1898, "step": 2824 }, { "epoch": 0.59, "learning_rate": 7.462848922466093e-06, "loss": 1.2026, "step": 2825 }, { "epoch": 0.59, "learning_rate": 7.456259191819256e-06, "loss": 1.2085, "step": 2826 }, { "epoch": 0.59, "learning_rate": 7.4496706419887286e-06, "loss": 1.1943, "step": 2827 }, { "epoch": 0.59, "learning_rate": 7.443083276032937e-06, "loss": 1.2433, "step": 2828 }, { "epoch": 0.6, "learning_rate": 7.436497097009776e-06, "loss": 1.2379, "step": 2829 }, { "epoch": 0.6, "learning_rate": 7.429912107976577e-06, "loss": 1.198, "step": 2830 }, { "epoch": 0.6, "learning_rate": 7.423328311990117e-06, "loss": 1.1739, "step": 2831 }, { "epoch": 0.6, "learning_rate": 7.416745712106634e-06, "loss": 1.1514, "step": 2832 }, { "epoch": 0.6, "learning_rate": 7.410164311381792e-06, "loss": 1.2251, "step": 2833 }, { "epoch": 0.6, "learning_rate": 7.4035841128707165e-06, "loss": 1.2246, "step": 2834 }, { "epoch": 0.6, "learning_rate": 7.397005119627962e-06, "loss": 1.2199, "step": 2835 }, { "epoch": 0.6, "learning_rate": 7.3904273347075265e-06, "loss": 1.2195, "step": 2836 }, { "epoch": 0.6, "learning_rate": 7.383850761162854e-06, "loss": 1.2858, "step": 2837 }, { "epoch": 0.6, "learning_rate": 7.377275402046816e-06, "loss": 1.1994, "step": 2838 }, { "epoch": 0.6, "learning_rate": 7.370701260411721e-06, "loss": 1.1785, "step": 2839 }, { "epoch": 0.6, "learning_rate": 7.364128339309326e-06, "loss": 1.2237, "step": 2840 }, { "epoch": 0.6, "learning_rate": 7.3575566417908e-06, "loss": 1.2255, "step": 2841 }, { "epoch": 0.6, "learning_rate": 7.350986170906767e-06, "loss": 1.1816, "step": 2842 }, { "epoch": 0.6, "learning_rate": 7.344416929707264e-06, "loss": 1.1532, "step": 2843 }, { "epoch": 0.6, "learning_rate": 7.33784892124176e-06, "loss": 1.2111, "step": 2844 }, { "epoch": 0.6, "learning_rate": 7.3312821485591625e-06, "loss": 1.1997, "step": 2845 }, { "epoch": 0.6, "learning_rate": 7.324716614707794e-06, "loss": 1.1626, "step": 2846 }, { "epoch": 0.6, "learning_rate": 7.3181523227354035e-06, "loss": 1.2038, "step": 2847 }, { "epoch": 0.6, "learning_rate": 7.311589275689173e-06, "loss": 1.19, "step": 2848 }, { "epoch": 0.6, "learning_rate": 7.305027476615689e-06, "loss": 1.2548, "step": 2849 }, { "epoch": 0.6, "learning_rate": 7.2984669285609814e-06, "loss": 1.2457, "step": 2850 }, { "epoch": 0.6, "learning_rate": 7.291907634570478e-06, "loss": 1.2466, "step": 2851 }, { "epoch": 0.6, "learning_rate": 7.285349597689036e-06, "loss": 1.1999, "step": 2852 }, { "epoch": 0.6, "learning_rate": 7.2787928209609295e-06, "loss": 1.2142, "step": 2853 }, { "epoch": 0.6, "learning_rate": 7.272237307429842e-06, "loss": 1.2338, "step": 2854 }, { "epoch": 0.6, "learning_rate": 7.265683060138868e-06, "loss": 1.2176, "step": 2855 }, { "epoch": 0.6, "learning_rate": 7.259130082130529e-06, "loss": 1.231, "step": 2856 }, { "epoch": 0.6, "learning_rate": 7.252578376446744e-06, "loss": 1.2381, "step": 2857 }, { "epoch": 0.6, "learning_rate": 7.246027946128844e-06, "loss": 1.2526, "step": 2858 }, { "epoch": 0.6, "learning_rate": 7.239478794217574e-06, "loss": 1.1789, "step": 2859 }, { "epoch": 0.6, "learning_rate": 7.232930923753071e-06, "loss": 1.2488, "step": 2860 }, { "epoch": 0.6, "learning_rate": 7.226384337774899e-06, "loss": 1.2138, "step": 2861 }, { "epoch": 0.6, "learning_rate": 7.219839039322002e-06, "loss": 1.2159, "step": 2862 }, { "epoch": 0.6, "learning_rate": 7.213295031432749e-06, "loss": 1.1975, "step": 2863 }, { "epoch": 0.6, "learning_rate": 7.2067523171448925e-06, "loss": 1.2696, "step": 2864 }, { "epoch": 0.6, "learning_rate": 7.200210899495589e-06, "loss": 1.1878, "step": 2865 }, { "epoch": 0.6, "learning_rate": 7.1936707815214005e-06, "loss": 1.2306, "step": 2866 }, { "epoch": 0.6, "learning_rate": 7.1871319662582784e-06, "loss": 1.2865, "step": 2867 }, { "epoch": 0.6, "learning_rate": 7.180594456741566e-06, "loss": 1.2226, "step": 2868 }, { "epoch": 0.6, "learning_rate": 7.174058256006012e-06, "loss": 1.1867, "step": 2869 }, { "epoch": 0.6, "learning_rate": 7.1675233670857495e-06, "loss": 1.2356, "step": 2870 }, { "epoch": 0.6, "learning_rate": 7.160989793014303e-06, "loss": 1.2373, "step": 2871 }, { "epoch": 0.6, "learning_rate": 7.154457536824591e-06, "loss": 1.2265, "step": 2872 }, { "epoch": 0.6, "learning_rate": 7.147926601548911e-06, "loss": 1.2428, "step": 2873 }, { "epoch": 0.6, "learning_rate": 7.141396990218962e-06, "loss": 1.2116, "step": 2874 }, { "epoch": 0.6, "learning_rate": 7.134868705865814e-06, "loss": 1.2569, "step": 2875 }, { "epoch": 0.6, "learning_rate": 7.128341751519925e-06, "loss": 1.1784, "step": 2876 }, { "epoch": 0.61, "learning_rate": 7.121816130211146e-06, "loss": 1.2038, "step": 2877 }, { "epoch": 0.61, "learning_rate": 7.1152918449686905e-06, "loss": 1.2258, "step": 2878 }, { "epoch": 0.61, "learning_rate": 7.108768898821171e-06, "loss": 1.1609, "step": 2879 }, { "epoch": 0.61, "learning_rate": 7.1022472947965635e-06, "loss": 1.2228, "step": 2880 }, { "epoch": 0.61, "learning_rate": 7.095727035922229e-06, "loss": 1.2173, "step": 2881 }, { "epoch": 0.61, "learning_rate": 7.089208125224902e-06, "loss": 1.2147, "step": 2882 }, { "epoch": 0.61, "learning_rate": 7.082690565730694e-06, "loss": 1.1845, "step": 2883 }, { "epoch": 0.61, "learning_rate": 7.076174360465078e-06, "loss": 1.2341, "step": 2884 }, { "epoch": 0.61, "learning_rate": 7.069659512452918e-06, "loss": 1.225, "step": 2885 }, { "epoch": 0.61, "learning_rate": 7.063146024718425e-06, "loss": 1.2453, "step": 2886 }, { "epoch": 0.61, "learning_rate": 7.056633900285201e-06, "loss": 1.2099, "step": 2887 }, { "epoch": 0.61, "learning_rate": 7.0501231421762e-06, "loss": 1.1801, "step": 2888 }, { "epoch": 0.61, "learning_rate": 7.043613753413741e-06, "loss": 1.1979, "step": 2889 }, { "epoch": 0.61, "learning_rate": 7.037105737019523e-06, "loss": 1.1947, "step": 2890 }, { "epoch": 0.61, "learning_rate": 7.030599096014585e-06, "loss": 1.2247, "step": 2891 }, { "epoch": 0.61, "learning_rate": 7.0240938334193535e-06, "loss": 1.2448, "step": 2892 }, { "epoch": 0.61, "learning_rate": 7.017589952253592e-06, "loss": 1.2719, "step": 2893 }, { "epoch": 0.61, "learning_rate": 7.011087455536434e-06, "loss": 1.2045, "step": 2894 }, { "epoch": 0.61, "learning_rate": 7.004586346286372e-06, "loss": 1.2328, "step": 2895 }, { "epoch": 0.61, "learning_rate": 6.99808662752125e-06, "loss": 1.2214, "step": 2896 }, { "epoch": 0.61, "learning_rate": 6.991588302258263e-06, "loss": 1.1805, "step": 2897 }, { "epoch": 0.61, "learning_rate": 6.985091373513972e-06, "loss": 1.2505, "step": 2898 }, { "epoch": 0.61, "learning_rate": 6.978595844304272e-06, "loss": 1.1989, "step": 2899 }, { "epoch": 0.61, "learning_rate": 6.9721017176444285e-06, "loss": 1.2064, "step": 2900 }, { "epoch": 0.61, "learning_rate": 6.965608996549038e-06, "loss": 1.2458, "step": 2901 }, { "epoch": 0.61, "learning_rate": 6.959117684032051e-06, "loss": 1.1587, "step": 2902 }, { "epoch": 0.61, "learning_rate": 6.952627783106771e-06, "loss": 1.2308, "step": 2903 }, { "epoch": 0.61, "learning_rate": 6.9461392967858345e-06, "loss": 1.1975, "step": 2904 }, { "epoch": 0.61, "learning_rate": 6.9396522280812285e-06, "loss": 1.1856, "step": 2905 }, { "epoch": 0.61, "learning_rate": 6.933166580004281e-06, "loss": 1.2194, "step": 2906 }, { "epoch": 0.61, "learning_rate": 6.926682355565662e-06, "loss": 1.1429, "step": 2907 }, { "epoch": 0.61, "learning_rate": 6.920199557775374e-06, "loss": 1.1824, "step": 2908 }, { "epoch": 0.61, "learning_rate": 6.913718189642769e-06, "loss": 1.2006, "step": 2909 }, { "epoch": 0.61, "learning_rate": 6.907238254176518e-06, "loss": 1.2383, "step": 2910 }, { "epoch": 0.61, "learning_rate": 6.900759754384647e-06, "loss": 1.2117, "step": 2911 }, { "epoch": 0.61, "learning_rate": 6.894282693274498e-06, "loss": 1.1849, "step": 2912 }, { "epoch": 0.61, "learning_rate": 6.887807073852761e-06, "loss": 1.1902, "step": 2913 }, { "epoch": 0.61, "learning_rate": 6.8813328991254436e-06, "loss": 1.2074, "step": 2914 }, { "epoch": 0.61, "learning_rate": 6.874860172097884e-06, "loss": 1.1943, "step": 2915 }, { "epoch": 0.61, "learning_rate": 6.86838889577476e-06, "loss": 1.2185, "step": 2916 }, { "epoch": 0.61, "learning_rate": 6.861919073160064e-06, "loss": 1.2411, "step": 2917 }, { "epoch": 0.61, "learning_rate": 6.8554507072571176e-06, "loss": 1.2144, "step": 2918 }, { "epoch": 0.61, "learning_rate": 6.848983801068567e-06, "loss": 1.2496, "step": 2919 }, { "epoch": 0.61, "learning_rate": 6.84251835759638e-06, "loss": 1.1743, "step": 2920 }, { "epoch": 0.61, "learning_rate": 6.836054379841847e-06, "loss": 1.2584, "step": 2921 }, { "epoch": 0.61, "learning_rate": 6.829591870805577e-06, "loss": 1.1596, "step": 2922 }, { "epoch": 0.61, "learning_rate": 6.823130833487488e-06, "loss": 1.2381, "step": 2923 }, { "epoch": 0.62, "learning_rate": 6.816671270886836e-06, "loss": 1.2029, "step": 2924 }, { "epoch": 0.62, "learning_rate": 6.810213186002177e-06, "loss": 1.2435, "step": 2925 }, { "epoch": 0.62, "learning_rate": 6.803756581831375e-06, "loss": 1.195, "step": 2926 }, { "epoch": 0.62, "learning_rate": 6.797301461371626e-06, "loss": 1.2335, "step": 2927 }, { "epoch": 0.62, "learning_rate": 6.79084782761942e-06, "loss": 1.2196, "step": 2928 }, { "epoch": 0.62, "learning_rate": 6.78439568357057e-06, "loss": 1.2181, "step": 2929 }, { "epoch": 0.62, "learning_rate": 6.777945032220187e-06, "loss": 1.2107, "step": 2930 }, { "epoch": 0.62, "learning_rate": 6.7714958765626935e-06, "loss": 1.2177, "step": 2931 }, { "epoch": 0.62, "learning_rate": 6.765048219591819e-06, "loss": 1.2295, "step": 2932 }, { "epoch": 0.62, "learning_rate": 6.758602064300597e-06, "loss": 1.2609, "step": 2933 }, { "epoch": 0.62, "learning_rate": 6.7521574136813574e-06, "loss": 1.2134, "step": 2934 }, { "epoch": 0.62, "learning_rate": 6.745714270725744e-06, "loss": 1.217, "step": 2935 }, { "epoch": 0.62, "learning_rate": 6.739272638424686e-06, "loss": 1.1834, "step": 2936 }, { "epoch": 0.62, "learning_rate": 6.732832519768428e-06, "loss": 1.175, "step": 2937 }, { "epoch": 0.62, "learning_rate": 6.726393917746499e-06, "loss": 1.222, "step": 2938 }, { "epoch": 0.62, "learning_rate": 6.7199568353477225e-06, "loss": 1.2228, "step": 2939 }, { "epoch": 0.62, "learning_rate": 6.7135212755602305e-06, "loss": 1.2031, "step": 2940 }, { "epoch": 0.62, "learning_rate": 6.707087241371433e-06, "loss": 1.2004, "step": 2941 }, { "epoch": 0.62, "learning_rate": 6.700654735768046e-06, "loss": 1.2568, "step": 2942 }, { "epoch": 0.62, "learning_rate": 6.694223761736061e-06, "loss": 1.2064, "step": 2943 }, { "epoch": 0.62, "learning_rate": 6.68779432226077e-06, "loss": 1.1779, "step": 2944 }, { "epoch": 0.62, "learning_rate": 6.681366420326747e-06, "loss": 1.2357, "step": 2945 }, { "epoch": 0.62, "learning_rate": 6.674940058917858e-06, "loss": 1.2483, "step": 2946 }, { "epoch": 0.62, "learning_rate": 6.6685152410172425e-06, "loss": 1.189, "step": 2947 }, { "epoch": 0.62, "learning_rate": 6.66209196960734e-06, "loss": 1.2249, "step": 2948 }, { "epoch": 0.62, "learning_rate": 6.655670247669852e-06, "loss": 1.2474, "step": 2949 }, { "epoch": 0.62, "learning_rate": 6.649250078185783e-06, "loss": 1.2165, "step": 2950 }, { "epoch": 0.62, "learning_rate": 6.6428314641354e-06, "loss": 1.2045, "step": 2951 }, { "epoch": 0.62, "learning_rate": 6.636414408498249e-06, "loss": 1.14, "step": 2952 }, { "epoch": 0.62, "learning_rate": 6.629998914253166e-06, "loss": 1.1931, "step": 2953 }, { "epoch": 0.62, "learning_rate": 6.623584984378247e-06, "loss": 1.2232, "step": 2954 }, { "epoch": 0.62, "learning_rate": 6.617172621850871e-06, "loss": 1.2336, "step": 2955 }, { "epoch": 0.62, "learning_rate": 6.610761829647685e-06, "loss": 1.2568, "step": 2956 }, { "epoch": 0.62, "learning_rate": 6.604352610744605e-06, "loss": 1.2216, "step": 2957 }, { "epoch": 0.62, "learning_rate": 6.597944968116831e-06, "loss": 1.1896, "step": 2958 }, { "epoch": 0.62, "learning_rate": 6.591538904738813e-06, "loss": 1.2167, "step": 2959 }, { "epoch": 0.62, "learning_rate": 6.5851344235842705e-06, "loss": 1.1983, "step": 2960 }, { "epoch": 0.62, "learning_rate": 6.578731527626205e-06, "loss": 1.2602, "step": 2961 }, { "epoch": 0.62, "learning_rate": 6.572330219836864e-06, "loss": 1.226, "step": 2962 }, { "epoch": 0.62, "learning_rate": 6.5659305031877596e-06, "loss": 1.2398, "step": 2963 }, { "epoch": 0.62, "learning_rate": 6.55953238064968e-06, "loss": 1.2197, "step": 2964 }, { "epoch": 0.62, "learning_rate": 6.553135855192656e-06, "loss": 1.219, "step": 2965 }, { "epoch": 0.62, "learning_rate": 6.546740929785991e-06, "loss": 1.1968, "step": 2966 }, { "epoch": 0.62, "learning_rate": 6.540347607398232e-06, "loss": 1.2587, "step": 2967 }, { "epoch": 0.62, "learning_rate": 6.533955890997192e-06, "loss": 1.2304, "step": 2968 }, { "epoch": 0.62, "learning_rate": 6.527565783549936e-06, "loss": 1.2326, "step": 2969 }, { "epoch": 0.62, "learning_rate": 6.521177288022779e-06, "loss": 1.2432, "step": 2970 }, { "epoch": 0.62, "learning_rate": 6.514790407381298e-06, "loss": 1.2807, "step": 2971 }, { "epoch": 0.63, "learning_rate": 6.508405144590308e-06, "loss": 1.2201, "step": 2972 }, { "epoch": 0.63, "learning_rate": 6.502021502613871e-06, "loss": 1.2352, "step": 2973 }, { "epoch": 0.63, "learning_rate": 6.495639484415313e-06, "loss": 1.2044, "step": 2974 }, { "epoch": 0.63, "learning_rate": 6.489259092957194e-06, "loss": 1.182, "step": 2975 }, { "epoch": 0.63, "learning_rate": 6.482880331201315e-06, "loss": 1.247, "step": 2976 }, { "epoch": 0.63, "learning_rate": 6.476503202108737e-06, "loss": 1.2475, "step": 2977 }, { "epoch": 0.63, "learning_rate": 6.470127708639743e-06, "loss": 1.1863, "step": 2978 }, { "epoch": 0.63, "learning_rate": 6.4637538537538755e-06, "loss": 1.2118, "step": 2979 }, { "epoch": 0.63, "learning_rate": 6.457381640409902e-06, "loss": 1.2316, "step": 2980 }, { "epoch": 0.63, "learning_rate": 6.451011071565834e-06, "loss": 1.2422, "step": 2981 }, { "epoch": 0.63, "learning_rate": 6.444642150178924e-06, "loss": 1.145, "step": 2982 }, { "epoch": 0.63, "learning_rate": 6.43827487920565e-06, "loss": 1.2354, "step": 2983 }, { "epoch": 0.63, "learning_rate": 6.431909261601726e-06, "loss": 1.2225, "step": 2984 }, { "epoch": 0.63, "learning_rate": 6.4255453003221115e-06, "loss": 1.1833, "step": 2985 }, { "epoch": 0.63, "learning_rate": 6.4191829983209745e-06, "loss": 1.1715, "step": 2986 }, { "epoch": 0.63, "learning_rate": 6.412822358551736e-06, "loss": 1.2003, "step": 2987 }, { "epoch": 0.63, "learning_rate": 6.40646338396703e-06, "loss": 1.2017, "step": 2988 }, { "epoch": 0.63, "learning_rate": 6.400106077518719e-06, "loss": 1.1904, "step": 2989 }, { "epoch": 0.63, "learning_rate": 6.3937504421579e-06, "loss": 1.1668, "step": 2990 }, { "epoch": 0.63, "learning_rate": 6.387396480834882e-06, "loss": 1.2039, "step": 2991 }, { "epoch": 0.63, "learning_rate": 6.381044196499208e-06, "loss": 1.2074, "step": 2992 }, { "epoch": 0.63, "learning_rate": 6.374693592099641e-06, "loss": 1.2127, "step": 2993 }, { "epoch": 0.63, "learning_rate": 6.368344670584152e-06, "loss": 1.1974, "step": 2994 }, { "epoch": 0.63, "learning_rate": 6.361997434899949e-06, "loss": 1.26, "step": 2995 }, { "epoch": 0.63, "learning_rate": 6.355651887993444e-06, "loss": 1.2111, "step": 2996 }, { "epoch": 0.63, "learning_rate": 6.349308032810264e-06, "loss": 1.2129, "step": 2997 }, { "epoch": 0.63, "learning_rate": 6.342965872295268e-06, "loss": 1.2407, "step": 2998 }, { "epoch": 0.63, "learning_rate": 6.336625409392505e-06, "loss": 1.2022, "step": 2999 }, { "epoch": 0.63, "learning_rate": 6.330286647045256e-06, "loss": 1.1693, "step": 3000 }, { "epoch": 0.63, "learning_rate": 6.323949588196001e-06, "loss": 1.2543, "step": 3001 }, { "epoch": 0.63, "learning_rate": 6.317614235786426e-06, "loss": 1.1746, "step": 3002 }, { "epoch": 0.63, "learning_rate": 6.311280592757443e-06, "loss": 1.2498, "step": 3003 }, { "epoch": 0.63, "learning_rate": 6.304948662049148e-06, "loss": 1.2539, "step": 3004 }, { "epoch": 0.63, "learning_rate": 6.298618446600856e-06, "loss": 1.193, "step": 3005 }, { "epoch": 0.63, "learning_rate": 6.292289949351087e-06, "loss": 1.2542, "step": 3006 }, { "epoch": 0.63, "learning_rate": 6.28596317323755e-06, "loss": 1.2055, "step": 3007 }, { "epoch": 0.63, "learning_rate": 6.279638121197174e-06, "loss": 1.2234, "step": 3008 }, { "epoch": 0.63, "learning_rate": 6.27331479616607e-06, "loss": 1.2285, "step": 3009 }, { "epoch": 0.63, "learning_rate": 6.266993201079556e-06, "loss": 1.1784, "step": 3010 }, { "epoch": 0.63, "learning_rate": 6.26067333887215e-06, "loss": 1.2281, "step": 3011 }, { "epoch": 0.63, "learning_rate": 6.254355212477558e-06, "loss": 1.1934, "step": 3012 }, { "epoch": 0.63, "learning_rate": 6.24803882482868e-06, "loss": 1.1855, "step": 3013 }, { "epoch": 0.63, "learning_rate": 6.241724178857621e-06, "loss": 1.2149, "step": 3014 }, { "epoch": 0.63, "learning_rate": 6.235411277495661e-06, "loss": 1.1939, "step": 3015 }, { "epoch": 0.63, "learning_rate": 6.229100123673284e-06, "loss": 1.2343, "step": 3016 }, { "epoch": 0.63, "learning_rate": 6.222790720320156e-06, "loss": 1.2421, "step": 3017 }, { "epoch": 0.63, "learning_rate": 6.216483070365125e-06, "loss": 1.1586, "step": 3018 }, { "epoch": 0.64, "learning_rate": 6.2101771767362415e-06, "loss": 1.1996, "step": 3019 }, { "epoch": 0.64, "learning_rate": 6.203873042360722e-06, "loss": 1.2479, "step": 3020 }, { "epoch": 0.64, "learning_rate": 6.197570670164983e-06, "loss": 1.223, "step": 3021 }, { "epoch": 0.64, "learning_rate": 6.191270063074609e-06, "loss": 1.2058, "step": 3022 }, { "epoch": 0.64, "learning_rate": 6.184971224014369e-06, "loss": 1.222, "step": 3023 }, { "epoch": 0.64, "learning_rate": 6.178674155908222e-06, "loss": 1.2572, "step": 3024 }, { "epoch": 0.64, "learning_rate": 6.1723788616792915e-06, "loss": 1.17, "step": 3025 }, { "epoch": 0.64, "learning_rate": 6.166085344249877e-06, "loss": 1.1811, "step": 3026 }, { "epoch": 0.64, "learning_rate": 6.1597936065414656e-06, "loss": 1.2178, "step": 3027 }, { "epoch": 0.64, "learning_rate": 6.15350365147471e-06, "loss": 1.1974, "step": 3028 }, { "epoch": 0.64, "learning_rate": 6.147215481969435e-06, "loss": 1.2251, "step": 3029 }, { "epoch": 0.64, "learning_rate": 6.140929100944642e-06, "loss": 1.2427, "step": 3030 }, { "epoch": 0.64, "learning_rate": 6.13464451131849e-06, "loss": 1.1918, "step": 3031 }, { "epoch": 0.64, "learning_rate": 6.128361716008324e-06, "loss": 1.2252, "step": 3032 }, { "epoch": 0.64, "learning_rate": 6.122080717930643e-06, "loss": 1.1888, "step": 3033 }, { "epoch": 0.64, "learning_rate": 6.1158015200011125e-06, "loss": 1.2284, "step": 3034 }, { "epoch": 0.64, "learning_rate": 6.1095241251345705e-06, "loss": 1.1936, "step": 3035 }, { "epoch": 0.64, "learning_rate": 6.103248536245007e-06, "loss": 1.2315, "step": 3036 }, { "epoch": 0.64, "learning_rate": 6.096974756245586e-06, "loss": 1.1174, "step": 3037 }, { "epoch": 0.64, "learning_rate": 6.0907027880486234e-06, "loss": 1.2012, "step": 3038 }, { "epoch": 0.64, "learning_rate": 6.084432634565591e-06, "loss": 1.2137, "step": 3039 }, { "epoch": 0.64, "learning_rate": 6.078164298707129e-06, "loss": 1.2666, "step": 3040 }, { "epoch": 0.64, "learning_rate": 6.071897783383027e-06, "loss": 1.1903, "step": 3041 }, { "epoch": 0.64, "learning_rate": 6.065633091502228e-06, "loss": 1.1512, "step": 3042 }, { "epoch": 0.64, "learning_rate": 6.059370225972834e-06, "loss": 1.2477, "step": 3043 }, { "epoch": 0.64, "learning_rate": 6.053109189702094e-06, "loss": 1.184, "step": 3044 }, { "epoch": 0.64, "learning_rate": 6.046849985596413e-06, "loss": 1.1997, "step": 3045 }, { "epoch": 0.64, "learning_rate": 6.04059261656134e-06, "loss": 1.2294, "step": 3046 }, { "epoch": 0.64, "learning_rate": 6.034337085501573e-06, "loss": 1.1917, "step": 3047 }, { "epoch": 0.64, "learning_rate": 6.028083395320965e-06, "loss": 1.2441, "step": 3048 }, { "epoch": 0.64, "learning_rate": 6.0218315489224995e-06, "loss": 1.1781, "step": 3049 }, { "epoch": 0.64, "learning_rate": 6.015581549208322e-06, "loss": 1.2078, "step": 3050 }, { "epoch": 0.64, "learning_rate": 6.009333399079704e-06, "loss": 1.1876, "step": 3051 }, { "epoch": 0.64, "learning_rate": 6.0030871014370706e-06, "loss": 1.1966, "step": 3052 }, { "epoch": 0.64, "learning_rate": 5.996842659179978e-06, "loss": 1.2481, "step": 3053 }, { "epoch": 0.64, "learning_rate": 5.990600075207131e-06, "loss": 1.2253, "step": 3054 }, { "epoch": 0.64, "learning_rate": 5.984359352416358e-06, "loss": 1.22, "step": 3055 }, { "epoch": 0.64, "learning_rate": 5.97812049370464e-06, "loss": 1.2463, "step": 3056 }, { "epoch": 0.64, "learning_rate": 5.971883501968076e-06, "loss": 1.2834, "step": 3057 }, { "epoch": 0.64, "learning_rate": 5.965648380101916e-06, "loss": 1.1579, "step": 3058 }, { "epoch": 0.64, "learning_rate": 5.959415131000528e-06, "loss": 1.1992, "step": 3059 }, { "epoch": 0.64, "learning_rate": 5.953183757557408e-06, "loss": 1.1829, "step": 3060 }, { "epoch": 0.64, "learning_rate": 5.946954262665201e-06, "loss": 1.1723, "step": 3061 }, { "epoch": 0.64, "learning_rate": 5.9407266492156615e-06, "loss": 1.2411, "step": 3062 }, { "epoch": 0.64, "learning_rate": 5.9345009200996725e-06, "loss": 1.1977, "step": 3063 }, { "epoch": 0.64, "learning_rate": 5.928277078207252e-06, "loss": 1.2492, "step": 3064 }, { "epoch": 0.64, "learning_rate": 5.922055126427536e-06, "loss": 1.2124, "step": 3065 }, { "epoch": 0.64, "learning_rate": 5.915835067648782e-06, "loss": 1.1868, "step": 3066 }, { "epoch": 0.65, "learning_rate": 5.909616904758373e-06, "loss": 1.2158, "step": 3067 }, { "epoch": 0.65, "learning_rate": 5.903400640642802e-06, "loss": 1.2549, "step": 3068 }, { "epoch": 0.65, "learning_rate": 5.897186278187696e-06, "loss": 1.1493, "step": 3069 }, { "epoch": 0.65, "learning_rate": 5.890973820277785e-06, "loss": 1.2042, "step": 3070 }, { "epoch": 0.65, "learning_rate": 5.884763269796925e-06, "loss": 1.2192, "step": 3071 }, { "epoch": 0.65, "learning_rate": 5.878554629628081e-06, "loss": 1.2324, "step": 3072 }, { "epoch": 0.65, "learning_rate": 5.87234790265333e-06, "loss": 1.2288, "step": 3073 }, { "epoch": 0.65, "learning_rate": 5.866143091753869e-06, "loss": 1.2315, "step": 3074 }, { "epoch": 0.65, "learning_rate": 5.859940199809993e-06, "loss": 1.2292, "step": 3075 }, { "epoch": 0.65, "learning_rate": 5.853739229701118e-06, "loss": 1.1841, "step": 3076 }, { "epoch": 0.65, "learning_rate": 5.847540184305763e-06, "loss": 1.1981, "step": 3077 }, { "epoch": 0.65, "learning_rate": 5.841343066501552e-06, "loss": 1.2447, "step": 3078 }, { "epoch": 0.65, "learning_rate": 5.83514787916522e-06, "loss": 1.2262, "step": 3079 }, { "epoch": 0.65, "learning_rate": 5.828954625172597e-06, "loss": 1.2246, "step": 3080 }, { "epoch": 0.65, "learning_rate": 5.8227633073986175e-06, "loss": 1.189, "step": 3081 }, { "epoch": 0.65, "learning_rate": 5.816573928717329e-06, "loss": 1.2027, "step": 3082 }, { "epoch": 0.65, "learning_rate": 5.810386492001862e-06, "loss": 1.164, "step": 3083 }, { "epoch": 0.65, "learning_rate": 5.804201000124454e-06, "loss": 1.1821, "step": 3084 }, { "epoch": 0.65, "learning_rate": 5.798017455956448e-06, "loss": 1.2239, "step": 3085 }, { "epoch": 0.65, "learning_rate": 5.791835862368263e-06, "loss": 1.2298, "step": 3086 }, { "epoch": 0.65, "learning_rate": 5.78565622222943e-06, "loss": 1.2411, "step": 3087 }, { "epoch": 0.65, "learning_rate": 5.779478538408567e-06, "loss": 1.2214, "step": 3088 }, { "epoch": 0.65, "learning_rate": 5.773302813773376e-06, "loss": 1.1955, "step": 3089 }, { "epoch": 0.65, "learning_rate": 5.767129051190666e-06, "loss": 1.2155, "step": 3090 }, { "epoch": 0.65, "learning_rate": 5.760957253526323e-06, "loss": 1.2108, "step": 3091 }, { "epoch": 0.65, "learning_rate": 5.754787423645321e-06, "loss": 1.2181, "step": 3092 }, { "epoch": 0.65, "learning_rate": 5.748619564411728e-06, "loss": 1.229, "step": 3093 }, { "epoch": 0.65, "learning_rate": 5.742453678688689e-06, "loss": 1.2002, "step": 3094 }, { "epoch": 0.65, "learning_rate": 5.736289769338441e-06, "loss": 1.1808, "step": 3095 }, { "epoch": 0.65, "learning_rate": 5.730127839222299e-06, "loss": 1.2078, "step": 3096 }, { "epoch": 0.65, "learning_rate": 5.723967891200655e-06, "loss": 1.2322, "step": 3097 }, { "epoch": 0.65, "learning_rate": 5.717809928132991e-06, "loss": 1.206, "step": 3098 }, { "epoch": 0.65, "learning_rate": 5.711653952877857e-06, "loss": 1.2103, "step": 3099 }, { "epoch": 0.65, "learning_rate": 5.705499968292891e-06, "loss": 1.2083, "step": 3100 }, { "epoch": 0.65, "learning_rate": 5.699347977234799e-06, "loss": 1.1926, "step": 3101 }, { "epoch": 0.65, "learning_rate": 5.693197982559357e-06, "loss": 1.1987, "step": 3102 }, { "epoch": 0.65, "learning_rate": 5.687049987121433e-06, "loss": 1.159, "step": 3103 }, { "epoch": 0.65, "learning_rate": 5.68090399377495e-06, "loss": 1.2415, "step": 3104 }, { "epoch": 0.65, "learning_rate": 5.674760005372901e-06, "loss": 1.2167, "step": 3105 }, { "epoch": 0.65, "learning_rate": 5.668618024767363e-06, "loss": 1.1981, "step": 3106 }, { "epoch": 0.65, "learning_rate": 5.6624780548094634e-06, "loss": 1.2142, "step": 3107 }, { "epoch": 0.65, "learning_rate": 5.656340098349409e-06, "loss": 1.1859, "step": 3108 }, { "epoch": 0.65, "learning_rate": 5.650204158236472e-06, "loss": 1.1691, "step": 3109 }, { "epoch": 0.65, "learning_rate": 5.644070237318977e-06, "loss": 1.2411, "step": 3110 }, { "epoch": 0.65, "learning_rate": 5.637938338444325e-06, "loss": 1.2056, "step": 3111 }, { "epoch": 0.65, "learning_rate": 5.63180846445897e-06, "loss": 1.1899, "step": 3112 }, { "epoch": 0.65, "learning_rate": 5.625680618208423e-06, "loss": 1.1748, "step": 3113 }, { "epoch": 0.65, "learning_rate": 5.619554802537263e-06, "loss": 1.2547, "step": 3114 }, { "epoch": 0.66, "learning_rate": 5.613431020289122e-06, "loss": 1.218, "step": 3115 }, { "epoch": 0.66, "learning_rate": 5.6073092743066895e-06, "loss": 1.2229, "step": 3116 }, { "epoch": 0.66, "learning_rate": 5.601189567431706e-06, "loss": 1.1751, "step": 3117 }, { "epoch": 0.66, "learning_rate": 5.595071902504967e-06, "loss": 1.2489, "step": 3118 }, { "epoch": 0.66, "learning_rate": 5.588956282366325e-06, "loss": 1.2466, "step": 3119 }, { "epoch": 0.66, "learning_rate": 5.5828427098546765e-06, "loss": 1.2465, "step": 3120 }, { "epoch": 0.66, "learning_rate": 5.576731187807967e-06, "loss": 1.2001, "step": 3121 }, { "epoch": 0.66, "learning_rate": 5.5706217190632016e-06, "loss": 1.1892, "step": 3122 }, { "epoch": 0.66, "learning_rate": 5.564514306456416e-06, "loss": 1.2393, "step": 3123 }, { "epoch": 0.66, "learning_rate": 5.558408952822706e-06, "loss": 1.205, "step": 3124 }, { "epoch": 0.66, "learning_rate": 5.552305660996202e-06, "loss": 1.1475, "step": 3125 }, { "epoch": 0.66, "learning_rate": 5.5462044338100776e-06, "loss": 1.2169, "step": 3126 }, { "epoch": 0.66, "learning_rate": 5.540105274096557e-06, "loss": 1.2189, "step": 3127 }, { "epoch": 0.66, "learning_rate": 5.534008184686892e-06, "loss": 1.2327, "step": 3128 }, { "epoch": 0.66, "learning_rate": 5.527913168411387e-06, "loss": 1.2138, "step": 3129 }, { "epoch": 0.66, "learning_rate": 5.5218202280993725e-06, "loss": 1.2372, "step": 3130 }, { "epoch": 0.66, "learning_rate": 5.515729366579217e-06, "loss": 1.2267, "step": 3131 }, { "epoch": 0.66, "learning_rate": 5.50964058667833e-06, "loss": 1.2101, "step": 3132 }, { "epoch": 0.66, "learning_rate": 5.503553891223155e-06, "loss": 1.2681, "step": 3133 }, { "epoch": 0.66, "learning_rate": 5.4974692830391574e-06, "loss": 1.2789, "step": 3134 }, { "epoch": 0.66, "learning_rate": 5.491386764950848e-06, "loss": 1.1927, "step": 3135 }, { "epoch": 0.66, "learning_rate": 5.4853063397817505e-06, "loss": 1.2472, "step": 3136 }, { "epoch": 0.66, "learning_rate": 5.479228010354437e-06, "loss": 1.1985, "step": 3137 }, { "epoch": 0.66, "learning_rate": 5.473151779490491e-06, "loss": 1.1778, "step": 3138 }, { "epoch": 0.66, "learning_rate": 5.467077650010524e-06, "loss": 1.1847, "step": 3139 }, { "epoch": 0.66, "learning_rate": 5.461005624734181e-06, "loss": 1.2247, "step": 3140 }, { "epoch": 0.66, "learning_rate": 5.454935706480123e-06, "loss": 1.241, "step": 3141 }, { "epoch": 0.66, "learning_rate": 5.448867898066028e-06, "loss": 1.1868, "step": 3142 }, { "epoch": 0.66, "learning_rate": 5.4428022023086105e-06, "loss": 1.2011, "step": 3143 }, { "epoch": 0.66, "learning_rate": 5.436738622023585e-06, "loss": 1.1786, "step": 3144 }, { "epoch": 0.66, "learning_rate": 5.430677160025705e-06, "loss": 1.2148, "step": 3145 }, { "epoch": 0.66, "learning_rate": 5.424617819128721e-06, "loss": 1.2114, "step": 3146 }, { "epoch": 0.66, "learning_rate": 5.418560602145406e-06, "loss": 1.1941, "step": 3147 }, { "epoch": 0.66, "learning_rate": 5.4125055118875555e-06, "loss": 1.2257, "step": 3148 }, { "epoch": 0.66, "learning_rate": 5.406452551165964e-06, "loss": 1.212, "step": 3149 }, { "epoch": 0.66, "learning_rate": 5.400401722790448e-06, "loss": 1.237, "step": 3150 }, { "epoch": 0.66, "learning_rate": 5.39435302956983e-06, "loss": 1.195, "step": 3151 }, { "epoch": 0.66, "learning_rate": 5.388306474311937e-06, "loss": 1.1723, "step": 3152 }, { "epoch": 0.66, "learning_rate": 5.382262059823614e-06, "loss": 1.2133, "step": 3153 }, { "epoch": 0.66, "learning_rate": 5.376219788910699e-06, "loss": 1.1739, "step": 3154 }, { "epoch": 0.66, "learning_rate": 5.370179664378053e-06, "loss": 1.2327, "step": 3155 }, { "epoch": 0.66, "learning_rate": 5.364141689029518e-06, "loss": 1.1883, "step": 3156 }, { "epoch": 0.66, "learning_rate": 5.358105865667961e-06, "loss": 1.1653, "step": 3157 }, { "epoch": 0.66, "learning_rate": 5.352072197095229e-06, "loss": 1.1739, "step": 3158 }, { "epoch": 0.66, "learning_rate": 5.346040686112189e-06, "loss": 1.2234, "step": 3159 }, { "epoch": 0.66, "learning_rate": 5.340011335518688e-06, "loss": 1.2334, "step": 3160 }, { "epoch": 0.66, "learning_rate": 5.333984148113584e-06, "loss": 1.1988, "step": 3161 }, { "epoch": 0.67, "learning_rate": 5.327959126694726e-06, "loss": 1.2138, "step": 3162 }, { "epoch": 0.67, "learning_rate": 5.32193627405895e-06, "loss": 1.1715, "step": 3163 }, { "epoch": 0.67, "learning_rate": 5.3159155930021e-06, "loss": 1.233, "step": 3164 }, { "epoch": 0.67, "learning_rate": 5.309897086318997e-06, "loss": 1.2147, "step": 3165 }, { "epoch": 0.67, "learning_rate": 5.303880756803468e-06, "loss": 1.218, "step": 3166 }, { "epoch": 0.67, "learning_rate": 5.297866607248316e-06, "loss": 1.2059, "step": 3167 }, { "epoch": 0.67, "learning_rate": 5.291854640445334e-06, "loss": 1.1946, "step": 3168 }, { "epoch": 0.67, "learning_rate": 5.285844859185314e-06, "loss": 1.2402, "step": 3169 }, { "epoch": 0.67, "learning_rate": 5.2798372662580165e-06, "loss": 1.2258, "step": 3170 }, { "epoch": 0.67, "learning_rate": 5.273831864452196e-06, "loss": 1.2293, "step": 3171 }, { "epoch": 0.67, "learning_rate": 5.267828656555591e-06, "loss": 1.1812, "step": 3172 }, { "epoch": 0.67, "learning_rate": 5.261827645354912e-06, "loss": 1.2468, "step": 3173 }, { "epoch": 0.67, "learning_rate": 5.2558288336358655e-06, "loss": 1.2193, "step": 3174 }, { "epoch": 0.67, "learning_rate": 5.249832224183123e-06, "loss": 1.2099, "step": 3175 }, { "epoch": 0.67, "learning_rate": 5.243837819780332e-06, "loss": 1.2543, "step": 3176 }, { "epoch": 0.67, "learning_rate": 5.237845623210134e-06, "loss": 1.2168, "step": 3177 }, { "epoch": 0.67, "learning_rate": 5.2318556372541266e-06, "loss": 1.1777, "step": 3178 }, { "epoch": 0.67, "learning_rate": 5.225867864692895e-06, "loss": 1.1885, "step": 3179 }, { "epoch": 0.67, "learning_rate": 5.219882308305986e-06, "loss": 1.1687, "step": 3180 }, { "epoch": 0.67, "learning_rate": 5.213898970871927e-06, "loss": 1.2095, "step": 3181 }, { "epoch": 0.67, "learning_rate": 5.207917855168205e-06, "loss": 1.1876, "step": 3182 }, { "epoch": 0.67, "learning_rate": 5.201938963971291e-06, "loss": 1.1922, "step": 3183 }, { "epoch": 0.67, "learning_rate": 5.1959623000566054e-06, "loss": 1.1176, "step": 3184 }, { "epoch": 0.67, "learning_rate": 5.189987866198548e-06, "loss": 1.1952, "step": 3185 }, { "epoch": 0.67, "learning_rate": 5.184015665170474e-06, "loss": 1.1402, "step": 3186 }, { "epoch": 0.67, "learning_rate": 5.178045699744716e-06, "loss": 1.2019, "step": 3187 }, { "epoch": 0.67, "learning_rate": 5.172077972692553e-06, "loss": 1.1745, "step": 3188 }, { "epoch": 0.67, "learning_rate": 5.166112486784229e-06, "loss": 1.1418, "step": 3189 }, { "epoch": 0.67, "learning_rate": 5.160149244788956e-06, "loss": 1.1728, "step": 3190 }, { "epoch": 0.67, "learning_rate": 5.1541882494749e-06, "loss": 1.2075, "step": 3191 }, { "epoch": 0.67, "learning_rate": 5.148229503609173e-06, "loss": 1.1742, "step": 3192 }, { "epoch": 0.67, "learning_rate": 5.1422730099578635e-06, "loss": 1.1708, "step": 3193 }, { "epoch": 0.67, "learning_rate": 5.136318771285994e-06, "loss": 1.1931, "step": 3194 }, { "epoch": 0.67, "learning_rate": 5.13036679035756e-06, "loss": 1.2345, "step": 3195 }, { "epoch": 0.67, "learning_rate": 5.124417069935492e-06, "loss": 1.1765, "step": 3196 }, { "epoch": 0.67, "learning_rate": 5.1184696127816755e-06, "loss": 1.2139, "step": 3197 }, { "epoch": 0.67, "learning_rate": 5.112524421656957e-06, "loss": 1.1739, "step": 3198 }, { "epoch": 0.67, "learning_rate": 5.106581499321115e-06, "loss": 1.115, "step": 3199 }, { "epoch": 0.67, "learning_rate": 5.1006408485328785e-06, "loss": 1.1678, "step": 3200 }, { "epoch": 0.67, "learning_rate": 5.094702472049934e-06, "loss": 1.233, "step": 3201 }, { "epoch": 0.67, "learning_rate": 5.088766372628895e-06, "loss": 1.1644, "step": 3202 }, { "epoch": 0.67, "learning_rate": 5.082832553025335e-06, "loss": 1.1827, "step": 3203 }, { "epoch": 0.67, "learning_rate": 5.076901015993752e-06, "loss": 1.218, "step": 3204 }, { "epoch": 0.67, "learning_rate": 5.070971764287602e-06, "loss": 1.1886, "step": 3205 }, { "epoch": 0.67, "learning_rate": 5.065044800659262e-06, "loss": 1.2179, "step": 3206 }, { "epoch": 0.67, "learning_rate": 5.059120127860064e-06, "loss": 1.1822, "step": 3207 }, { "epoch": 0.67, "learning_rate": 5.053197748640262e-06, "loss": 1.2001, "step": 3208 }, { "epoch": 0.67, "learning_rate": 5.047277665749059e-06, "loss": 1.2227, "step": 3209 }, { "epoch": 0.68, "learning_rate": 5.0413598819345765e-06, "loss": 1.1708, "step": 3210 }, { "epoch": 0.68, "learning_rate": 5.035444399943886e-06, "loss": 1.229, "step": 3211 }, { "epoch": 0.68, "learning_rate": 5.029531222522978e-06, "loss": 1.2395, "step": 3212 }, { "epoch": 0.68, "learning_rate": 5.023620352416772e-06, "loss": 1.1835, "step": 3213 }, { "epoch": 0.68, "learning_rate": 5.017711792369131e-06, "loss": 1.2211, "step": 3214 }, { "epoch": 0.68, "learning_rate": 5.011805545122827e-06, "loss": 1.2334, "step": 3215 }, { "epoch": 0.68, "learning_rate": 5.005901613419574e-06, "loss": 1.2535, "step": 3216 }, { "epoch": 0.68, "learning_rate": 5.000000000000003e-06, "loss": 1.2346, "step": 3217 }, { "epoch": 0.68, "learning_rate": 4.994100707603664e-06, "loss": 1.1689, "step": 3218 }, { "epoch": 0.68, "learning_rate": 4.988203738969046e-06, "loss": 1.1838, "step": 3219 }, { "epoch": 0.68, "learning_rate": 4.982309096833545e-06, "loss": 1.163, "step": 3220 }, { "epoch": 0.68, "learning_rate": 4.976416783933477e-06, "loss": 1.2149, "step": 3221 }, { "epoch": 0.68, "learning_rate": 4.970526803004088e-06, "loss": 1.169, "step": 3222 }, { "epoch": 0.68, "learning_rate": 4.964639156779527e-06, "loss": 1.2104, "step": 3223 }, { "epoch": 0.68, "learning_rate": 4.958753847992876e-06, "loss": 1.2044, "step": 3224 }, { "epoch": 0.68, "learning_rate": 4.952870879376117e-06, "loss": 1.2497, "step": 3225 }, { "epoch": 0.68, "learning_rate": 4.946990253660147e-06, "loss": 1.2313, "step": 3226 }, { "epoch": 0.68, "learning_rate": 4.941111973574792e-06, "loss": 1.1895, "step": 3227 }, { "epoch": 0.68, "learning_rate": 4.935236041848763e-06, "loss": 1.2152, "step": 3228 }, { "epoch": 0.68, "learning_rate": 4.929362461209707e-06, "loss": 1.2076, "step": 3229 }, { "epoch": 0.68, "learning_rate": 4.923491234384158e-06, "loss": 1.2119, "step": 3230 }, { "epoch": 0.68, "learning_rate": 4.917622364097575e-06, "loss": 1.1836, "step": 3231 }, { "epoch": 0.68, "learning_rate": 4.911755853074306e-06, "loss": 1.2053, "step": 3232 }, { "epoch": 0.68, "learning_rate": 4.905891704037622e-06, "loss": 1.2079, "step": 3233 }, { "epoch": 0.68, "learning_rate": 4.900029919709679e-06, "loss": 1.2152, "step": 3234 }, { "epoch": 0.68, "learning_rate": 4.894170502811554e-06, "loss": 1.1925, "step": 3235 }, { "epoch": 0.68, "learning_rate": 4.888313456063206e-06, "loss": 1.1874, "step": 3236 }, { "epoch": 0.68, "learning_rate": 4.882458782183512e-06, "loss": 1.187, "step": 3237 }, { "epoch": 0.68, "learning_rate": 4.876606483890236e-06, "loss": 1.2238, "step": 3238 }, { "epoch": 0.68, "learning_rate": 4.870756563900036e-06, "loss": 1.2227, "step": 3239 }, { "epoch": 0.68, "learning_rate": 4.8649090249284814e-06, "loss": 1.2214, "step": 3240 }, { "epoch": 0.68, "learning_rate": 4.859063869690022e-06, "loss": 1.1817, "step": 3241 }, { "epoch": 0.68, "learning_rate": 4.8532211008980034e-06, "loss": 1.2022, "step": 3242 }, { "epoch": 0.68, "learning_rate": 4.847380721264673e-06, "loss": 1.1509, "step": 3243 }, { "epoch": 0.68, "learning_rate": 4.841542733501153e-06, "loss": 1.1784, "step": 3244 }, { "epoch": 0.68, "learning_rate": 4.835707140317475e-06, "loss": 1.1811, "step": 3245 }, { "epoch": 0.68, "learning_rate": 4.829873944422544e-06, "loss": 1.243, "step": 3246 }, { "epoch": 0.68, "learning_rate": 4.824043148524151e-06, "loss": 1.2465, "step": 3247 }, { "epoch": 0.68, "learning_rate": 4.818214755328987e-06, "loss": 1.1869, "step": 3248 }, { "epoch": 0.68, "learning_rate": 4.812388767542616e-06, "loss": 1.1304, "step": 3249 }, { "epoch": 0.68, "learning_rate": 4.806565187869485e-06, "loss": 1.1869, "step": 3250 }, { "epoch": 0.68, "learning_rate": 4.800744019012934e-06, "loss": 1.1719, "step": 3251 }, { "epoch": 0.68, "learning_rate": 4.7949252636751686e-06, "loss": 1.1838, "step": 3252 }, { "epoch": 0.68, "learning_rate": 4.7891089245572894e-06, "loss": 1.2304, "step": 3253 }, { "epoch": 0.68, "learning_rate": 4.783295004359262e-06, "loss": 1.1878, "step": 3254 }, { "epoch": 0.68, "learning_rate": 4.777483505779937e-06, "loss": 1.1349, "step": 3255 }, { "epoch": 0.68, "learning_rate": 4.771674431517045e-06, "loss": 1.1989, "step": 3256 }, { "epoch": 0.69, "learning_rate": 4.765867784267175e-06, "loss": 1.2321, "step": 3257 }, { "epoch": 0.69, "learning_rate": 4.7600635667258095e-06, "loss": 1.1425, "step": 3258 }, { "epoch": 0.69, "learning_rate": 4.754261781587289e-06, "loss": 1.1738, "step": 3259 }, { "epoch": 0.69, "learning_rate": 4.748462431544826e-06, "loss": 1.2152, "step": 3260 }, { "epoch": 0.69, "learning_rate": 4.742665519290511e-06, "loss": 1.1649, "step": 3261 }, { "epoch": 0.69, "learning_rate": 4.736871047515294e-06, "loss": 1.2424, "step": 3262 }, { "epoch": 0.69, "learning_rate": 4.731079018908995e-06, "loss": 1.2125, "step": 3263 }, { "epoch": 0.69, "learning_rate": 4.725289436160307e-06, "loss": 1.1832, "step": 3264 }, { "epoch": 0.69, "learning_rate": 4.719502301956771e-06, "loss": 1.2298, "step": 3265 }, { "epoch": 0.69, "learning_rate": 4.713717618984812e-06, "loss": 1.1979, "step": 3266 }, { "epoch": 0.69, "learning_rate": 4.707935389929702e-06, "loss": 1.2252, "step": 3267 }, { "epoch": 0.69, "learning_rate": 4.702155617475574e-06, "loss": 1.2119, "step": 3268 }, { "epoch": 0.69, "learning_rate": 4.6963783043054355e-06, "loss": 1.2078, "step": 3269 }, { "epoch": 0.69, "learning_rate": 4.690603453101134e-06, "loss": 1.2345, "step": 3270 }, { "epoch": 0.69, "learning_rate": 4.684831066543384e-06, "loss": 1.1925, "step": 3271 }, { "epoch": 0.69, "learning_rate": 4.679061147311757e-06, "loss": 1.2033, "step": 3272 }, { "epoch": 0.69, "learning_rate": 4.67329369808467e-06, "loss": 1.1672, "step": 3273 }, { "epoch": 0.69, "learning_rate": 4.6675287215394085e-06, "loss": 1.1886, "step": 3274 }, { "epoch": 0.69, "learning_rate": 4.661766220352098e-06, "loss": 1.2154, "step": 3275 }, { "epoch": 0.69, "learning_rate": 4.656006197197711e-06, "loss": 1.2095, "step": 3276 }, { "epoch": 0.69, "learning_rate": 4.650248654750088e-06, "loss": 1.1918, "step": 3277 }, { "epoch": 0.69, "learning_rate": 4.6444935956818996e-06, "loss": 1.216, "step": 3278 }, { "epoch": 0.69, "learning_rate": 4.638741022664672e-06, "loss": 1.1688, "step": 3279 }, { "epoch": 0.69, "learning_rate": 4.632990938368783e-06, "loss": 1.2094, "step": 3280 }, { "epoch": 0.69, "learning_rate": 4.627243345463439e-06, "loss": 1.2081, "step": 3281 }, { "epoch": 0.69, "learning_rate": 4.621498246616707e-06, "loss": 1.2293, "step": 3282 }, { "epoch": 0.69, "learning_rate": 4.615755644495485e-06, "loss": 1.1965, "step": 3283 }, { "epoch": 0.69, "learning_rate": 4.610015541765513e-06, "loss": 1.2477, "step": 3284 }, { "epoch": 0.69, "learning_rate": 4.604277941091381e-06, "loss": 1.196, "step": 3285 }, { "epoch": 0.69, "learning_rate": 4.5985428451365e-06, "loss": 1.2108, "step": 3286 }, { "epoch": 0.69, "learning_rate": 4.592810256563137e-06, "loss": 1.185, "step": 3287 }, { "epoch": 0.69, "learning_rate": 4.5870801780323825e-06, "loss": 1.2028, "step": 3288 }, { "epoch": 0.69, "learning_rate": 4.581352612204163e-06, "loss": 1.1833, "step": 3289 }, { "epoch": 0.69, "learning_rate": 4.575627561737247e-06, "loss": 1.2275, "step": 3290 }, { "epoch": 0.69, "learning_rate": 4.569905029289226e-06, "loss": 1.2106, "step": 3291 }, { "epoch": 0.69, "learning_rate": 4.564185017516523e-06, "loss": 1.1812, "step": 3292 }, { "epoch": 0.69, "learning_rate": 4.558467529074401e-06, "loss": 1.213, "step": 3293 }, { "epoch": 0.69, "learning_rate": 4.552752566616938e-06, "loss": 1.2027, "step": 3294 }, { "epoch": 0.69, "learning_rate": 4.547040132797054e-06, "loss": 1.1977, "step": 3295 }, { "epoch": 0.69, "learning_rate": 4.541330230266481e-06, "loss": 1.207, "step": 3296 }, { "epoch": 0.69, "learning_rate": 4.535622861675782e-06, "loss": 1.2534, "step": 3297 }, { "epoch": 0.69, "learning_rate": 4.529918029674348e-06, "loss": 1.2247, "step": 3298 }, { "epoch": 0.69, "learning_rate": 4.524215736910388e-06, "loss": 1.229, "step": 3299 }, { "epoch": 0.69, "learning_rate": 4.518515986030927e-06, "loss": 1.1774, "step": 3300 }, { "epoch": 0.69, "learning_rate": 4.512818779681825e-06, "loss": 1.205, "step": 3301 }, { "epoch": 0.69, "learning_rate": 4.507124120507742e-06, "loss": 1.218, "step": 3302 }, { "epoch": 0.69, "learning_rate": 4.50143201115217e-06, "loss": 1.2227, "step": 3303 }, { "epoch": 0.69, "learning_rate": 4.495742454257418e-06, "loss": 1.1996, "step": 3304 }, { "epoch": 0.7, "learning_rate": 4.490055452464594e-06, "loss": 1.2313, "step": 3305 }, { "epoch": 0.7, "learning_rate": 4.484371008413641e-06, "loss": 1.1805, "step": 3306 }, { "epoch": 0.7, "learning_rate": 4.478689124743296e-06, "loss": 1.1972, "step": 3307 }, { "epoch": 0.7, "learning_rate": 4.473009804091123e-06, "loss": 1.2529, "step": 3308 }, { "epoch": 0.7, "learning_rate": 4.467333049093485e-06, "loss": 1.1926, "step": 3309 }, { "epoch": 0.7, "learning_rate": 4.461658862385555e-06, "loss": 1.206, "step": 3310 }, { "epoch": 0.7, "learning_rate": 4.455987246601325e-06, "loss": 1.2429, "step": 3311 }, { "epoch": 0.7, "learning_rate": 4.450318204373578e-06, "loss": 1.2533, "step": 3312 }, { "epoch": 0.7, "learning_rate": 4.4446517383339114e-06, "loss": 1.2103, "step": 3313 }, { "epoch": 0.7, "learning_rate": 4.4389878511127285e-06, "loss": 1.2597, "step": 3314 }, { "epoch": 0.7, "learning_rate": 4.433326545339226e-06, "loss": 1.2085, "step": 3315 }, { "epoch": 0.7, "learning_rate": 4.427667823641416e-06, "loss": 1.2277, "step": 3316 }, { "epoch": 0.7, "learning_rate": 4.422011688646099e-06, "loss": 1.2134, "step": 3317 }, { "epoch": 0.7, "learning_rate": 4.416358142978877e-06, "loss": 1.259, "step": 3318 }, { "epoch": 0.7, "learning_rate": 4.410707189264157e-06, "loss": 1.2302, "step": 3319 }, { "epoch": 0.7, "learning_rate": 4.405058830125138e-06, "loss": 1.2256, "step": 3320 }, { "epoch": 0.7, "learning_rate": 4.399413068183806e-06, "loss": 1.2123, "step": 3321 }, { "epoch": 0.7, "learning_rate": 4.393769906060962e-06, "loss": 1.2146, "step": 3322 }, { "epoch": 0.7, "learning_rate": 4.388129346376177e-06, "loss": 1.1925, "step": 3323 }, { "epoch": 0.7, "learning_rate": 4.382491391747835e-06, "loss": 1.2128, "step": 3324 }, { "epoch": 0.7, "learning_rate": 4.376856044793092e-06, "loss": 1.2477, "step": 3325 }, { "epoch": 0.7, "learning_rate": 4.37122330812791e-06, "loss": 1.1929, "step": 3326 }, { "epoch": 0.7, "learning_rate": 4.3655931843670245e-06, "loss": 1.2068, "step": 3327 }, { "epoch": 0.7, "learning_rate": 4.359965676123972e-06, "loss": 1.2015, "step": 3328 }, { "epoch": 0.7, "learning_rate": 4.354340786011061e-06, "loss": 1.2146, "step": 3329 }, { "epoch": 0.7, "learning_rate": 4.348718516639398e-06, "loss": 1.2087, "step": 3330 }, { "epoch": 0.7, "learning_rate": 4.343098870618861e-06, "loss": 1.1939, "step": 3331 }, { "epoch": 0.7, "learning_rate": 4.3374818505581214e-06, "loss": 1.179, "step": 3332 }, { "epoch": 0.7, "learning_rate": 4.331867459064623e-06, "loss": 1.1638, "step": 3333 }, { "epoch": 0.7, "learning_rate": 4.326255698744591e-06, "loss": 1.1877, "step": 3334 }, { "epoch": 0.7, "learning_rate": 4.3206465722030335e-06, "loss": 1.2288, "step": 3335 }, { "epoch": 0.7, "learning_rate": 4.315040082043728e-06, "loss": 1.2332, "step": 3336 }, { "epoch": 0.7, "learning_rate": 4.309436230869241e-06, "loss": 1.1785, "step": 3337 }, { "epoch": 0.7, "learning_rate": 4.303835021280903e-06, "loss": 1.2199, "step": 3338 }, { "epoch": 0.7, "learning_rate": 4.2982364558788166e-06, "loss": 1.2149, "step": 3339 }, { "epoch": 0.7, "learning_rate": 4.29264053726187e-06, "loss": 1.2048, "step": 3340 }, { "epoch": 0.7, "learning_rate": 4.287047268027711e-06, "loss": 1.2204, "step": 3341 }, { "epoch": 0.7, "learning_rate": 4.281456650772756e-06, "loss": 1.2403, "step": 3342 }, { "epoch": 0.7, "learning_rate": 4.275868688092205e-06, "loss": 1.246, "step": 3343 }, { "epoch": 0.7, "learning_rate": 4.270283382580008e-06, "loss": 1.196, "step": 3344 }, { "epoch": 0.7, "learning_rate": 4.264700736828897e-06, "loss": 1.1998, "step": 3345 }, { "epoch": 0.7, "learning_rate": 4.2591207534303595e-06, "loss": 1.2125, "step": 3346 }, { "epoch": 0.7, "learning_rate": 4.253543434974643e-06, "loss": 1.2241, "step": 3347 }, { "epoch": 0.7, "learning_rate": 4.2479687840507755e-06, "loss": 1.2478, "step": 3348 }, { "epoch": 0.7, "learning_rate": 4.242396803246527e-06, "loss": 1.1881, "step": 3349 }, { "epoch": 0.7, "learning_rate": 4.2368274951484435e-06, "loss": 1.2493, "step": 3350 }, { "epoch": 0.7, "learning_rate": 4.231260862341819e-06, "loss": 1.1738, "step": 3351 }, { "epoch": 0.71, "learning_rate": 4.2256969074107135e-06, "loss": 1.1896, "step": 3352 }, { "epoch": 0.71, "learning_rate": 4.2201356329379375e-06, "loss": 1.2117, "step": 3353 }, { "epoch": 0.71, "learning_rate": 4.214577041505065e-06, "loss": 1.1789, "step": 3354 }, { "epoch": 0.71, "learning_rate": 4.209021135692414e-06, "loss": 1.2078, "step": 3355 }, { "epoch": 0.71, "learning_rate": 4.203467918079069e-06, "loss": 1.1695, "step": 3356 }, { "epoch": 0.71, "learning_rate": 4.197917391242855e-06, "loss": 1.1846, "step": 3357 }, { "epoch": 0.71, "learning_rate": 4.19236955776035e-06, "loss": 1.2243, "step": 3358 }, { "epoch": 0.71, "learning_rate": 4.186824420206891e-06, "loss": 1.2152, "step": 3359 }, { "epoch": 0.71, "learning_rate": 4.181281981156548e-06, "loss": 1.2199, "step": 3360 }, { "epoch": 0.71, "learning_rate": 4.175742243182155e-06, "loss": 1.1547, "step": 3361 }, { "epoch": 0.71, "learning_rate": 4.170205208855281e-06, "loss": 1.1954, "step": 3362 }, { "epoch": 0.71, "learning_rate": 4.16467088074624e-06, "loss": 1.1979, "step": 3363 }, { "epoch": 0.71, "learning_rate": 4.159139261424099e-06, "loss": 1.177, "step": 3364 }, { "epoch": 0.71, "learning_rate": 4.153610353456654e-06, "loss": 1.2152, "step": 3365 }, { "epoch": 0.71, "learning_rate": 4.148084159410457e-06, "loss": 1.1966, "step": 3366 }, { "epoch": 0.71, "learning_rate": 4.14256068185079e-06, "loss": 1.2413, "step": 3367 }, { "epoch": 0.71, "learning_rate": 4.137039923341673e-06, "loss": 1.2117, "step": 3368 }, { "epoch": 0.71, "learning_rate": 4.131521886445876e-06, "loss": 1.2269, "step": 3369 }, { "epoch": 0.71, "learning_rate": 4.1260065737248925e-06, "loss": 1.1673, "step": 3370 }, { "epoch": 0.71, "learning_rate": 4.120493987738952e-06, "loss": 1.1687, "step": 3371 }, { "epoch": 0.71, "learning_rate": 4.114984131047032e-06, "loss": 1.1611, "step": 3372 }, { "epoch": 0.71, "learning_rate": 4.1094770062068234e-06, "loss": 1.1765, "step": 3373 }, { "epoch": 0.71, "learning_rate": 4.1039726157747695e-06, "loss": 1.1757, "step": 3374 }, { "epoch": 0.71, "learning_rate": 4.098470962306025e-06, "loss": 1.277, "step": 3375 }, { "epoch": 0.71, "learning_rate": 4.092972048354491e-06, "loss": 1.2029, "step": 3376 }, { "epoch": 0.71, "learning_rate": 4.08747587647278e-06, "loss": 1.1939, "step": 3377 }, { "epoch": 0.71, "learning_rate": 4.08198244921225e-06, "loss": 1.1843, "step": 3378 }, { "epoch": 0.71, "learning_rate": 4.076491769122966e-06, "loss": 1.1883, "step": 3379 }, { "epoch": 0.71, "learning_rate": 4.071003838753738e-06, "loss": 1.2305, "step": 3380 }, { "epoch": 0.71, "learning_rate": 4.065518660652077e-06, "loss": 1.2122, "step": 3381 }, { "epoch": 0.71, "learning_rate": 4.060036237364238e-06, "loss": 1.1985, "step": 3382 }, { "epoch": 0.71, "learning_rate": 4.054556571435185e-06, "loss": 1.2219, "step": 3383 }, { "epoch": 0.71, "learning_rate": 4.049079665408596e-06, "loss": 1.2059, "step": 3384 }, { "epoch": 0.71, "learning_rate": 4.0436055218268885e-06, "loss": 1.212, "step": 3385 }, { "epoch": 0.71, "learning_rate": 4.038134143231176e-06, "loss": 1.2444, "step": 3386 }, { "epoch": 0.71, "learning_rate": 4.032665532161306e-06, "loss": 1.2167, "step": 3387 }, { "epoch": 0.71, "learning_rate": 4.027199691155829e-06, "loss": 1.1242, "step": 3388 }, { "epoch": 0.71, "learning_rate": 4.021736622752009e-06, "loss": 1.1822, "step": 3389 }, { "epoch": 0.71, "learning_rate": 4.016276329485836e-06, "loss": 1.2159, "step": 3390 }, { "epoch": 0.71, "learning_rate": 4.010818813892e-06, "loss": 1.2081, "step": 3391 }, { "epoch": 0.71, "learning_rate": 4.005364078503903e-06, "loss": 1.1286, "step": 3392 }, { "epoch": 0.71, "learning_rate": 3.999912125853662e-06, "loss": 1.2108, "step": 3393 }, { "epoch": 0.71, "learning_rate": 3.994462958472095e-06, "loss": 1.2124, "step": 3394 }, { "epoch": 0.71, "learning_rate": 3.989016578888737e-06, "loss": 1.2168, "step": 3395 }, { "epoch": 0.71, "learning_rate": 3.983572989631818e-06, "loss": 1.2208, "step": 3396 }, { "epoch": 0.71, "learning_rate": 3.978132193228276e-06, "loss": 1.1738, "step": 3397 }, { "epoch": 0.71, "learning_rate": 3.972694192203759e-06, "loss": 1.1873, "step": 3398 }, { "epoch": 0.71, "learning_rate": 3.9672589890826055e-06, "loss": 1.2007, "step": 3399 }, { "epoch": 0.72, "learning_rate": 3.96182658638787e-06, "loss": 1.224, "step": 3400 }, { "epoch": 0.72, "learning_rate": 3.956396986641292e-06, "loss": 1.1984, "step": 3401 }, { "epoch": 0.72, "learning_rate": 3.950970192363325e-06, "loss": 1.2207, "step": 3402 }, { "epoch": 0.72, "learning_rate": 3.945546206073102e-06, "loss": 1.2073, "step": 3403 }, { "epoch": 0.72, "learning_rate": 3.940125030288472e-06, "loss": 1.1918, "step": 3404 }, { "epoch": 0.72, "learning_rate": 3.934706667525963e-06, "loss": 1.1716, "step": 3405 }, { "epoch": 0.72, "learning_rate": 3.92929112030081e-06, "loss": 1.2186, "step": 3406 }, { "epoch": 0.72, "learning_rate": 3.923878391126932e-06, "loss": 1.1253, "step": 3407 }, { "epoch": 0.72, "learning_rate": 3.9184684825169405e-06, "loss": 1.1967, "step": 3408 }, { "epoch": 0.72, "learning_rate": 3.9130613969821466e-06, "loss": 1.2113, "step": 3409 }, { "epoch": 0.72, "learning_rate": 3.907657137032537e-06, "loss": 1.1779, "step": 3410 }, { "epoch": 0.72, "learning_rate": 3.902255705176802e-06, "loss": 1.2116, "step": 3411 }, { "epoch": 0.72, "learning_rate": 3.896857103922308e-06, "loss": 1.2221, "step": 3412 }, { "epoch": 0.72, "learning_rate": 3.891461335775106e-06, "loss": 1.1611, "step": 3413 }, { "epoch": 0.72, "learning_rate": 3.886068403239946e-06, "loss": 1.1306, "step": 3414 }, { "epoch": 0.72, "learning_rate": 3.880678308820243e-06, "loss": 1.1699, "step": 3415 }, { "epoch": 0.72, "learning_rate": 3.875291055018112e-06, "loss": 1.1648, "step": 3416 }, { "epoch": 0.72, "learning_rate": 3.869906644334338e-06, "loss": 1.2162, "step": 3417 }, { "epoch": 0.72, "learning_rate": 3.864525079268385e-06, "loss": 1.1816, "step": 3418 }, { "epoch": 0.72, "learning_rate": 3.859146362318408e-06, "loss": 1.212, "step": 3419 }, { "epoch": 0.72, "learning_rate": 3.85377049598123e-06, "loss": 1.1987, "step": 3420 }, { "epoch": 0.72, "learning_rate": 3.848397482752348e-06, "loss": 1.1871, "step": 3421 }, { "epoch": 0.72, "learning_rate": 3.843027325125946e-06, "loss": 1.228, "step": 3422 }, { "epoch": 0.72, "learning_rate": 3.837660025594873e-06, "loss": 1.1365, "step": 3423 }, { "epoch": 0.72, "learning_rate": 3.832295586650657e-06, "loss": 1.1889, "step": 3424 }, { "epoch": 0.72, "learning_rate": 3.82693401078349e-06, "loss": 1.2241, "step": 3425 }, { "epoch": 0.72, "learning_rate": 3.821575300482245e-06, "loss": 1.2139, "step": 3426 }, { "epoch": 0.72, "learning_rate": 3.816219458234464e-06, "loss": 1.2155, "step": 3427 }, { "epoch": 0.72, "learning_rate": 3.810866486526348e-06, "loss": 1.1602, "step": 3428 }, { "epoch": 0.72, "learning_rate": 3.8055163878427703e-06, "loss": 1.1668, "step": 3429 }, { "epoch": 0.72, "learning_rate": 3.800169164667277e-06, "loss": 1.1848, "step": 3430 }, { "epoch": 0.72, "learning_rate": 3.7948248194820668e-06, "loss": 1.2027, "step": 3431 }, { "epoch": 0.72, "learning_rate": 3.7894833547680178e-06, "loss": 1.2217, "step": 3432 }, { "epoch": 0.72, "learning_rate": 3.784144773004658e-06, "loss": 1.2113, "step": 3433 }, { "epoch": 0.72, "learning_rate": 3.778809076670179e-06, "loss": 1.1861, "step": 3434 }, { "epoch": 0.72, "learning_rate": 3.773476268241444e-06, "loss": 1.1845, "step": 3435 }, { "epoch": 0.72, "learning_rate": 3.768146350193962e-06, "loss": 1.2505, "step": 3436 }, { "epoch": 0.72, "learning_rate": 3.7628193250019028e-06, "loss": 1.1435, "step": 3437 }, { "epoch": 0.72, "learning_rate": 3.7574951951381033e-06, "loss": 1.2124, "step": 3438 }, { "epoch": 0.72, "learning_rate": 3.752173963074043e-06, "loss": 1.2045, "step": 3439 }, { "epoch": 0.72, "learning_rate": 3.7468556312798687e-06, "loss": 1.2337, "step": 3440 }, { "epoch": 0.72, "learning_rate": 3.741540202224372e-06, "loss": 1.2458, "step": 3441 }, { "epoch": 0.72, "learning_rate": 3.736227678374996e-06, "loss": 1.1979, "step": 3442 }, { "epoch": 0.72, "learning_rate": 3.730918062197846e-06, "loss": 1.2221, "step": 3443 }, { "epoch": 0.72, "learning_rate": 3.7256113561576625e-06, "loss": 1.1851, "step": 3444 }, { "epoch": 0.72, "learning_rate": 3.720307562717851e-06, "loss": 1.169, "step": 3445 }, { "epoch": 0.72, "learning_rate": 3.715006684340454e-06, "loss": 1.2114, "step": 3446 }, { "epoch": 0.73, "learning_rate": 3.7097087234861594e-06, "loss": 1.2098, "step": 3447 }, { "epoch": 0.73, "learning_rate": 3.7044136826143127e-06, "loss": 1.1078, "step": 3448 }, { "epoch": 0.73, "learning_rate": 3.6991215641828903e-06, "loss": 1.1498, "step": 3449 }, { "epoch": 0.73, "learning_rate": 3.69383237064852e-06, "loss": 1.2399, "step": 3450 }, { "epoch": 0.73, "learning_rate": 3.688546104466475e-06, "loss": 1.2267, "step": 3451 }, { "epoch": 0.73, "learning_rate": 3.6832627680906562e-06, "loss": 1.2016, "step": 3452 }, { "epoch": 0.73, "learning_rate": 3.677982363973621e-06, "loss": 1.2144, "step": 3453 }, { "epoch": 0.73, "learning_rate": 3.6727048945665543e-06, "loss": 1.2198, "step": 3454 }, { "epoch": 0.73, "learning_rate": 3.6674303623192777e-06, "loss": 1.2129, "step": 3455 }, { "epoch": 0.73, "learning_rate": 3.6621587696802607e-06, "loss": 1.2296, "step": 3456 }, { "epoch": 0.73, "learning_rate": 3.6568901190965977e-06, "loss": 1.2157, "step": 3457 }, { "epoch": 0.73, "learning_rate": 3.651624413014018e-06, "loss": 1.162, "step": 3458 }, { "epoch": 0.73, "learning_rate": 3.646361653876892e-06, "loss": 1.2215, "step": 3459 }, { "epoch": 0.73, "learning_rate": 3.641101844128211e-06, "loss": 1.1808, "step": 3460 }, { "epoch": 0.73, "learning_rate": 3.63584498620961e-06, "loss": 1.2189, "step": 3461 }, { "epoch": 0.73, "learning_rate": 3.6305910825613445e-06, "loss": 1.2352, "step": 3462 }, { "epoch": 0.73, "learning_rate": 3.6253401356222963e-06, "loss": 1.2313, "step": 3463 }, { "epoch": 0.73, "learning_rate": 3.6200921478299854e-06, "loss": 1.215, "step": 3464 }, { "epoch": 0.73, "learning_rate": 3.6148471216205484e-06, "loss": 1.1556, "step": 3465 }, { "epoch": 0.73, "learning_rate": 3.6096050594287545e-06, "loss": 1.1797, "step": 3466 }, { "epoch": 0.73, "learning_rate": 3.604365963687991e-06, "loss": 1.2059, "step": 3467 }, { "epoch": 0.73, "learning_rate": 3.5991298368302674e-06, "loss": 1.2063, "step": 3468 }, { "epoch": 0.73, "learning_rate": 3.5938966812862253e-06, "loss": 1.2308, "step": 3469 }, { "epoch": 0.73, "learning_rate": 3.588666499485115e-06, "loss": 1.1757, "step": 3470 }, { "epoch": 0.73, "learning_rate": 3.583439293854809e-06, "loss": 1.1944, "step": 3471 }, { "epoch": 0.73, "learning_rate": 3.5782150668218063e-06, "loss": 1.2188, "step": 3472 }, { "epoch": 0.73, "learning_rate": 3.5729938208112115e-06, "loss": 1.1731, "step": 3473 }, { "epoch": 0.73, "learning_rate": 3.5677755582467543e-06, "loss": 1.1497, "step": 3474 }, { "epoch": 0.73, "learning_rate": 3.562560281550781e-06, "loss": 1.168, "step": 3475 }, { "epoch": 0.73, "learning_rate": 3.557347993144238e-06, "loss": 1.1684, "step": 3476 }, { "epoch": 0.73, "learning_rate": 3.552138695446702e-06, "loss": 1.208, "step": 3477 }, { "epoch": 0.73, "learning_rate": 3.5469323908763507e-06, "loss": 1.1925, "step": 3478 }, { "epoch": 0.73, "learning_rate": 3.5417290818499694e-06, "loss": 1.159, "step": 3479 }, { "epoch": 0.73, "learning_rate": 3.5365287707829667e-06, "loss": 1.2173, "step": 3480 }, { "epoch": 0.73, "learning_rate": 3.5313314600893444e-06, "loss": 1.1626, "step": 3481 }, { "epoch": 0.73, "learning_rate": 3.5261371521817247e-06, "loss": 1.2088, "step": 3482 }, { "epoch": 0.73, "learning_rate": 3.5209458494713245e-06, "loss": 1.1874, "step": 3483 }, { "epoch": 0.73, "learning_rate": 3.51575755436797e-06, "loss": 1.2273, "step": 3484 }, { "epoch": 0.73, "learning_rate": 3.510572269280097e-06, "loss": 1.1543, "step": 3485 }, { "epoch": 0.73, "learning_rate": 3.5053899966147355e-06, "loss": 1.1756, "step": 3486 }, { "epoch": 0.73, "learning_rate": 3.500210738777519e-06, "loss": 1.2191, "step": 3487 }, { "epoch": 0.73, "learning_rate": 3.4950344981726867e-06, "loss": 1.194, "step": 3488 }, { "epoch": 0.73, "learning_rate": 3.4898612772030692e-06, "loss": 1.18, "step": 3489 }, { "epoch": 0.73, "learning_rate": 3.484691078270106e-06, "loss": 1.241, "step": 3490 }, { "epoch": 0.73, "learning_rate": 3.4795239037738238e-06, "loss": 1.2119, "step": 3491 }, { "epoch": 0.73, "learning_rate": 3.4743597561128462e-06, "loss": 1.1945, "step": 3492 }, { "epoch": 0.73, "learning_rate": 3.4691986376844e-06, "loss": 1.1929, "step": 3493 }, { "epoch": 0.73, "learning_rate": 3.4640405508842945e-06, "loss": 1.228, "step": 3494 }, { "epoch": 0.74, "learning_rate": 3.4588854981069453e-06, "loss": 1.2099, "step": 3495 }, { "epoch": 0.74, "learning_rate": 3.453733481745346e-06, "loss": 1.2067, "step": 3496 }, { "epoch": 0.74, "learning_rate": 3.448584504191086e-06, "loss": 1.2174, "step": 3497 }, { "epoch": 0.74, "learning_rate": 3.443438567834345e-06, "loss": 1.2487, "step": 3498 }, { "epoch": 0.74, "learning_rate": 3.4382956750638975e-06, "loss": 1.1865, "step": 3499 }, { "epoch": 0.74, "learning_rate": 3.4331558282670884e-06, "loss": 1.191, "step": 3500 }, { "epoch": 0.74, "learning_rate": 3.4280190298298656e-06, "loss": 1.2378, "step": 3501 }, { "epoch": 0.74, "learning_rate": 3.4228852821367497e-06, "loss": 1.1822, "step": 3502 }, { "epoch": 0.74, "learning_rate": 3.4177545875708564e-06, "loss": 1.1996, "step": 3503 }, { "epoch": 0.74, "learning_rate": 3.412626948513873e-06, "loss": 1.2286, "step": 3504 }, { "epoch": 0.74, "learning_rate": 3.407502367346073e-06, "loss": 1.2269, "step": 3505 }, { "epoch": 0.74, "learning_rate": 3.4023808464463135e-06, "loss": 1.1793, "step": 3506 }, { "epoch": 0.74, "learning_rate": 3.3972623881920296e-06, "loss": 1.158, "step": 3507 }, { "epoch": 0.74, "learning_rate": 3.392146994959227e-06, "loss": 1.213, "step": 3508 }, { "epoch": 0.74, "learning_rate": 3.387034669122501e-06, "loss": 1.1926, "step": 3509 }, { "epoch": 0.74, "learning_rate": 3.3819254130550126e-06, "loss": 1.2413, "step": 3510 }, { "epoch": 0.74, "learning_rate": 3.3768192291285087e-06, "loss": 1.1744, "step": 3511 }, { "epoch": 0.74, "learning_rate": 3.3717161197132995e-06, "loss": 1.1786, "step": 3512 }, { "epoch": 0.74, "learning_rate": 3.3666160871782684e-06, "loss": 1.1463, "step": 3513 }, { "epoch": 0.74, "learning_rate": 3.3615191338908815e-06, "loss": 1.2072, "step": 3514 }, { "epoch": 0.74, "learning_rate": 3.3564252622171645e-06, "loss": 1.2487, "step": 3515 }, { "epoch": 0.74, "learning_rate": 3.3513344745217138e-06, "loss": 1.1823, "step": 3516 }, { "epoch": 0.74, "learning_rate": 3.346246773167702e-06, "loss": 1.1733, "step": 3517 }, { "epoch": 0.74, "learning_rate": 3.3411621605168576e-06, "loss": 1.1809, "step": 3518 }, { "epoch": 0.74, "learning_rate": 3.3360806389294886e-06, "loss": 1.2514, "step": 3519 }, { "epoch": 0.74, "learning_rate": 3.3310022107644546e-06, "loss": 1.1202, "step": 3520 }, { "epoch": 0.74, "learning_rate": 3.325926878379191e-06, "loss": 1.2036, "step": 3521 }, { "epoch": 0.74, "learning_rate": 3.320854644129684e-06, "loss": 1.2519, "step": 3522 }, { "epoch": 0.74, "learning_rate": 3.3157855103704963e-06, "loss": 1.2315, "step": 3523 }, { "epoch": 0.74, "learning_rate": 3.310719479454736e-06, "loss": 1.1939, "step": 3524 }, { "epoch": 0.74, "learning_rate": 3.3056565537340847e-06, "loss": 1.1745, "step": 3525 }, { "epoch": 0.74, "learning_rate": 3.3005967355587698e-06, "loss": 1.2338, "step": 3526 }, { "epoch": 0.74, "learning_rate": 3.2955400272775884e-06, "loss": 1.2035, "step": 3527 }, { "epoch": 0.74, "learning_rate": 3.2904864312378846e-06, "loss": 1.1906, "step": 3528 }, { "epoch": 0.74, "learning_rate": 3.2854359497855593e-06, "loss": 1.2038, "step": 3529 }, { "epoch": 0.74, "learning_rate": 3.280388585265075e-06, "loss": 1.176, "step": 3530 }, { "epoch": 0.74, "learning_rate": 3.2753443400194353e-06, "loss": 1.1859, "step": 3531 }, { "epoch": 0.74, "learning_rate": 3.2703032163902082e-06, "loss": 1.226, "step": 3532 }, { "epoch": 0.74, "learning_rate": 3.2652652167175037e-06, "loss": 1.1943, "step": 3533 }, { "epoch": 0.74, "learning_rate": 3.260230343339982e-06, "loss": 1.1896, "step": 3534 }, { "epoch": 0.74, "learning_rate": 3.255198598594862e-06, "loss": 1.1839, "step": 3535 }, { "epoch": 0.74, "learning_rate": 3.250169984817897e-06, "loss": 1.1086, "step": 3536 }, { "epoch": 0.74, "learning_rate": 3.245144504343393e-06, "loss": 1.2, "step": 3537 }, { "epoch": 0.74, "learning_rate": 3.2401221595042054e-06, "loss": 1.2323, "step": 3538 }, { "epoch": 0.74, "learning_rate": 3.2351029526317247e-06, "loss": 1.1737, "step": 3539 }, { "epoch": 0.74, "learning_rate": 3.2300868860558956e-06, "loss": 1.2281, "step": 3540 }, { "epoch": 0.74, "learning_rate": 3.2250739621051973e-06, "loss": 1.1536, "step": 3541 }, { "epoch": 0.75, "learning_rate": 3.2200641831066483e-06, "loss": 1.1968, "step": 3542 }, { "epoch": 0.75, "learning_rate": 3.2150575513858186e-06, "loss": 1.1695, "step": 3543 }, { "epoch": 0.75, "learning_rate": 3.2100540692668036e-06, "loss": 1.1862, "step": 3544 }, { "epoch": 0.75, "learning_rate": 3.205053739072248e-06, "loss": 1.2428, "step": 3545 }, { "epoch": 0.75, "learning_rate": 3.2000565631233227e-06, "loss": 1.2155, "step": 3546 }, { "epoch": 0.75, "learning_rate": 3.1950625437397464e-06, "loss": 1.2251, "step": 3547 }, { "epoch": 0.75, "learning_rate": 3.1900716832397603e-06, "loss": 1.2112, "step": 3548 }, { "epoch": 0.75, "learning_rate": 3.1850839839401513e-06, "loss": 1.147, "step": 3549 }, { "epoch": 0.75, "learning_rate": 3.1800994481562254e-06, "loss": 1.2035, "step": 3550 }, { "epoch": 0.75, "learning_rate": 3.175118078201833e-06, "loss": 1.2427, "step": 3551 }, { "epoch": 0.75, "learning_rate": 3.1701398763893452e-06, "loss": 1.1532, "step": 3552 }, { "epoch": 0.75, "learning_rate": 3.165164845029671e-06, "loss": 1.262, "step": 3553 }, { "epoch": 0.75, "learning_rate": 3.16019298643224e-06, "loss": 1.179, "step": 3554 }, { "epoch": 0.75, "learning_rate": 3.15522430290501e-06, "loss": 1.2075, "step": 3555 }, { "epoch": 0.75, "learning_rate": 3.150258796754472e-06, "loss": 1.1571, "step": 3556 }, { "epoch": 0.75, "learning_rate": 3.1452964702856346e-06, "loss": 1.2466, "step": 3557 }, { "epoch": 0.75, "learning_rate": 3.1403373258020297e-06, "loss": 1.1974, "step": 3558 }, { "epoch": 0.75, "learning_rate": 3.135381365605721e-06, "loss": 1.1682, "step": 3559 }, { "epoch": 0.75, "learning_rate": 3.130428591997282e-06, "loss": 1.2244, "step": 3560 }, { "epoch": 0.75, "learning_rate": 3.12547900727582e-06, "loss": 1.1991, "step": 3561 }, { "epoch": 0.75, "learning_rate": 3.1205326137389525e-06, "loss": 1.2087, "step": 3562 }, { "epoch": 0.75, "learning_rate": 3.1155894136828135e-06, "loss": 1.212, "step": 3563 }, { "epoch": 0.75, "learning_rate": 3.110649409402068e-06, "loss": 1.2075, "step": 3564 }, { "epoch": 0.75, "learning_rate": 3.1057126031898843e-06, "loss": 1.1866, "step": 3565 }, { "epoch": 0.75, "learning_rate": 3.1007789973379475e-06, "loss": 1.2148, "step": 3566 }, { "epoch": 0.75, "learning_rate": 3.0958485941364667e-06, "loss": 1.2075, "step": 3567 }, { "epoch": 0.75, "learning_rate": 3.0909213958741534e-06, "loss": 1.2137, "step": 3568 }, { "epoch": 0.75, "learning_rate": 3.08599740483824e-06, "loss": 1.1751, "step": 3569 }, { "epoch": 0.75, "learning_rate": 3.0810766233144606e-06, "loss": 1.1811, "step": 3570 }, { "epoch": 0.75, "learning_rate": 3.076159053587072e-06, "loss": 1.2021, "step": 3571 }, { "epoch": 0.75, "learning_rate": 3.0712446979388257e-06, "loss": 1.2123, "step": 3572 }, { "epoch": 0.75, "learning_rate": 3.0663335586509945e-06, "loss": 1.1874, "step": 3573 }, { "epoch": 0.75, "learning_rate": 3.061425638003346e-06, "loss": 1.1831, "step": 3574 }, { "epoch": 0.75, "learning_rate": 3.0565209382741666e-06, "loss": 1.1845, "step": 3575 }, { "epoch": 0.75, "learning_rate": 3.051619461740234e-06, "loss": 1.2103, "step": 3576 }, { "epoch": 0.75, "learning_rate": 3.0467212106768416e-06, "loss": 1.1768, "step": 3577 }, { "epoch": 0.75, "learning_rate": 3.0418261873577804e-06, "loss": 1.2222, "step": 3578 }, { "epoch": 0.75, "learning_rate": 3.0369343940553364e-06, "loss": 1.2388, "step": 3579 }, { "epoch": 0.75, "learning_rate": 3.0320458330403104e-06, "loss": 1.199, "step": 3580 }, { "epoch": 0.75, "learning_rate": 3.027160506581989e-06, "loss": 1.2091, "step": 3581 }, { "epoch": 0.75, "learning_rate": 3.022278416948169e-06, "loss": 1.2103, "step": 3582 }, { "epoch": 0.75, "learning_rate": 3.017399566405137e-06, "loss": 1.1745, "step": 3583 }, { "epoch": 0.75, "learning_rate": 3.0125239572176747e-06, "loss": 1.2345, "step": 3584 }, { "epoch": 0.75, "learning_rate": 3.0076515916490677e-06, "loss": 1.1921, "step": 3585 }, { "epoch": 0.75, "learning_rate": 3.0027824719610897e-06, "loss": 1.2236, "step": 3586 }, { "epoch": 0.75, "learning_rate": 2.9979166004140037e-06, "loss": 1.2279, "step": 3587 }, { "epoch": 0.75, "learning_rate": 2.9930539792665767e-06, "loss": 1.1768, "step": 3588 }, { "epoch": 0.75, "learning_rate": 2.988194610776055e-06, "loss": 1.2086, "step": 3589 }, { "epoch": 0.76, "learning_rate": 2.9833384971981838e-06, "loss": 1.263, "step": 3590 }, { "epoch": 0.76, "learning_rate": 2.978485640787193e-06, "loss": 1.217, "step": 3591 }, { "epoch": 0.76, "learning_rate": 2.9736360437957957e-06, "loss": 1.2033, "step": 3592 }, { "epoch": 0.76, "learning_rate": 2.9687897084752048e-06, "loss": 1.1977, "step": 3593 }, { "epoch": 0.76, "learning_rate": 2.963946637075107e-06, "loss": 1.1465, "step": 3594 }, { "epoch": 0.76, "learning_rate": 2.959106831843681e-06, "loss": 1.2425, "step": 3595 }, { "epoch": 0.76, "learning_rate": 2.954270295027584e-06, "loss": 1.1699, "step": 3596 }, { "epoch": 0.76, "learning_rate": 2.949437028871963e-06, "loss": 1.1733, "step": 3597 }, { "epoch": 0.76, "learning_rate": 2.944607035620438e-06, "loss": 1.23, "step": 3598 }, { "epoch": 0.76, "learning_rate": 2.939780317515121e-06, "loss": 1.2471, "step": 3599 }, { "epoch": 0.76, "learning_rate": 2.9349568767965876e-06, "loss": 1.2316, "step": 3600 }, { "epoch": 0.76, "learning_rate": 2.9301367157039106e-06, "loss": 1.1644, "step": 3601 }, { "epoch": 0.76, "learning_rate": 2.9253198364746228e-06, "loss": 1.2101, "step": 3602 }, { "epoch": 0.76, "learning_rate": 2.9205062413447504e-06, "loss": 1.1979, "step": 3603 }, { "epoch": 0.76, "learning_rate": 2.915695932548782e-06, "loss": 1.146, "step": 3604 }, { "epoch": 0.76, "learning_rate": 2.9108889123196825e-06, "loss": 1.2145, "step": 3605 }, { "epoch": 0.76, "learning_rate": 2.9060851828889003e-06, "loss": 1.2367, "step": 3606 }, { "epoch": 0.76, "learning_rate": 2.9012847464863435e-06, "loss": 1.2398, "step": 3607 }, { "epoch": 0.76, "learning_rate": 2.8964876053403967e-06, "loss": 1.2033, "step": 3608 }, { "epoch": 0.76, "learning_rate": 2.8916937616779193e-06, "loss": 1.1844, "step": 3609 }, { "epoch": 0.76, "learning_rate": 2.8869032177242297e-06, "loss": 1.1953, "step": 3610 }, { "epoch": 0.76, "learning_rate": 2.882115975703128e-06, "loss": 1.1874, "step": 3611 }, { "epoch": 0.76, "learning_rate": 2.87733203783687e-06, "loss": 1.1534, "step": 3612 }, { "epoch": 0.76, "learning_rate": 2.872551406346179e-06, "loss": 1.15, "step": 3613 }, { "epoch": 0.76, "learning_rate": 2.867774083450253e-06, "loss": 1.1738, "step": 3614 }, { "epoch": 0.76, "learning_rate": 2.863000071366745e-06, "loss": 1.1934, "step": 3615 }, { "epoch": 0.76, "learning_rate": 2.858229372311768e-06, "loss": 1.1713, "step": 3616 }, { "epoch": 0.76, "learning_rate": 2.8534619884999125e-06, "loss": 1.1899, "step": 3617 }, { "epoch": 0.76, "learning_rate": 2.8486979221442114e-06, "loss": 1.1919, "step": 3618 }, { "epoch": 0.76, "learning_rate": 2.843937175456173e-06, "loss": 1.237, "step": 3619 }, { "epoch": 0.76, "learning_rate": 2.839179750645752e-06, "loss": 1.189, "step": 3620 }, { "epoch": 0.76, "learning_rate": 2.8344256499213695e-06, "loss": 1.2109, "step": 3621 }, { "epoch": 0.76, "learning_rate": 2.829674875489907e-06, "loss": 1.2044, "step": 3622 }, { "epoch": 0.76, "learning_rate": 2.8249274295566863e-06, "loss": 1.2078, "step": 3623 }, { "epoch": 0.76, "learning_rate": 2.8201833143255033e-06, "loss": 1.18, "step": 3624 }, { "epoch": 0.76, "learning_rate": 2.8154425319985934e-06, "loss": 1.2058, "step": 3625 }, { "epoch": 0.76, "learning_rate": 2.810705084776646e-06, "loss": 1.2002, "step": 3626 }, { "epoch": 0.76, "learning_rate": 2.8059709748588138e-06, "loss": 1.1596, "step": 3627 }, { "epoch": 0.76, "learning_rate": 2.8012402044426888e-06, "loss": 1.2148, "step": 3628 }, { "epoch": 0.76, "learning_rate": 2.796512775724315e-06, "loss": 1.2012, "step": 3629 }, { "epoch": 0.76, "learning_rate": 2.7917886908981907e-06, "loss": 1.2323, "step": 3630 }, { "epoch": 0.76, "learning_rate": 2.787067952157252e-06, "loss": 1.2345, "step": 3631 }, { "epoch": 0.76, "learning_rate": 2.7823505616928958e-06, "loss": 1.1643, "step": 3632 }, { "epoch": 0.76, "learning_rate": 2.777636521694953e-06, "loss": 1.172, "step": 3633 }, { "epoch": 0.76, "learning_rate": 2.7729258343516983e-06, "loss": 1.2206, "step": 3634 }, { "epoch": 0.76, "learning_rate": 2.768218501849862e-06, "loss": 1.2091, "step": 3635 }, { "epoch": 0.76, "learning_rate": 2.763514526374608e-06, "loss": 1.1488, "step": 3636 }, { "epoch": 0.76, "learning_rate": 2.758813910109538e-06, "loss": 1.2242, "step": 3637 }, { "epoch": 0.77, "learning_rate": 2.754116655236706e-06, "loss": 1.2327, "step": 3638 }, { "epoch": 0.77, "learning_rate": 2.7494227639365965e-06, "loss": 1.2466, "step": 3639 }, { "epoch": 0.77, "learning_rate": 2.744732238388138e-06, "loss": 1.1793, "step": 3640 }, { "epoch": 0.77, "learning_rate": 2.740045080768694e-06, "loss": 1.1475, "step": 3641 }, { "epoch": 0.77, "learning_rate": 2.735361293254061e-06, "loss": 1.2315, "step": 3642 }, { "epoch": 0.77, "learning_rate": 2.7306808780184813e-06, "loss": 1.1917, "step": 3643 }, { "epoch": 0.77, "learning_rate": 2.7260038372346176e-06, "loss": 1.2297, "step": 3644 }, { "epoch": 0.77, "learning_rate": 2.7213301730735785e-06, "loss": 1.2082, "step": 3645 }, { "epoch": 0.77, "learning_rate": 2.7166598877049034e-06, "loss": 1.2438, "step": 3646 }, { "epoch": 0.77, "learning_rate": 2.7119929832965552e-06, "loss": 1.2089, "step": 3647 }, { "epoch": 0.77, "learning_rate": 2.7073294620149373e-06, "loss": 1.2283, "step": 3648 }, { "epoch": 0.77, "learning_rate": 2.7026693260248747e-06, "loss": 1.2024, "step": 3649 }, { "epoch": 0.77, "learning_rate": 2.6980125774896236e-06, "loss": 1.1458, "step": 3650 }, { "epoch": 0.77, "learning_rate": 2.6933592185708714e-06, "loss": 1.1485, "step": 3651 }, { "epoch": 0.77, "learning_rate": 2.688709251428725e-06, "loss": 1.1958, "step": 3652 }, { "epoch": 0.77, "learning_rate": 2.684062678221726e-06, "loss": 1.2027, "step": 3653 }, { "epoch": 0.77, "learning_rate": 2.6794195011068325e-06, "loss": 1.1884, "step": 3654 }, { "epoch": 0.77, "learning_rate": 2.6747797222394246e-06, "loss": 1.2245, "step": 3655 }, { "epoch": 0.77, "learning_rate": 2.6701433437733172e-06, "loss": 1.1767, "step": 3656 }, { "epoch": 0.77, "learning_rate": 2.6655103678607343e-06, "loss": 1.2229, "step": 3657 }, { "epoch": 0.77, "learning_rate": 2.660880796652323e-06, "loss": 1.214, "step": 3658 }, { "epoch": 0.77, "learning_rate": 2.656254632297156e-06, "loss": 1.1185, "step": 3659 }, { "epoch": 0.77, "learning_rate": 2.6516318769427165e-06, "loss": 1.1927, "step": 3660 }, { "epoch": 0.77, "learning_rate": 2.6470125327349137e-06, "loss": 1.2071, "step": 3661 }, { "epoch": 0.77, "learning_rate": 2.642396601818066e-06, "loss": 1.2482, "step": 3662 }, { "epoch": 0.77, "learning_rate": 2.6377840863349067e-06, "loss": 1.1574, "step": 3663 }, { "epoch": 0.77, "learning_rate": 2.6331749884265933e-06, "loss": 1.2081, "step": 3664 }, { "epoch": 0.77, "learning_rate": 2.628569310232687e-06, "loss": 1.1404, "step": 3665 }, { "epoch": 0.77, "learning_rate": 2.623967053891162e-06, "loss": 1.179, "step": 3666 }, { "epoch": 0.77, "learning_rate": 2.619368221538414e-06, "loss": 1.2396, "step": 3667 }, { "epoch": 0.77, "learning_rate": 2.6147728153092357e-06, "loss": 1.1752, "step": 3668 }, { "epoch": 0.77, "learning_rate": 2.6101808373368386e-06, "loss": 1.216, "step": 3669 }, { "epoch": 0.77, "learning_rate": 2.605592289752844e-06, "loss": 1.2167, "step": 3670 }, { "epoch": 0.77, "learning_rate": 2.6010071746872712e-06, "loss": 1.1802, "step": 3671 }, { "epoch": 0.77, "learning_rate": 2.5964254942685575e-06, "loss": 1.2397, "step": 3672 }, { "epoch": 0.77, "learning_rate": 2.591847250623536e-06, "loss": 1.1924, "step": 3673 }, { "epoch": 0.77, "learning_rate": 2.587272445877448e-06, "loss": 1.1859, "step": 3674 }, { "epoch": 0.77, "learning_rate": 2.582701082153942e-06, "loss": 1.1595, "step": 3675 }, { "epoch": 0.77, "learning_rate": 2.578133161575065e-06, "loss": 1.2142, "step": 3676 }, { "epoch": 0.77, "learning_rate": 2.5735686862612685e-06, "loss": 1.1714, "step": 3677 }, { "epoch": 0.77, "learning_rate": 2.5690076583314016e-06, "loss": 1.1848, "step": 3678 }, { "epoch": 0.77, "learning_rate": 2.5644500799027127e-06, "loss": 1.2376, "step": 3679 }, { "epoch": 0.77, "learning_rate": 2.5598959530908564e-06, "loss": 1.2144, "step": 3680 }, { "epoch": 0.77, "learning_rate": 2.555345280009872e-06, "loss": 1.2353, "step": 3681 }, { "epoch": 0.77, "learning_rate": 2.5507980627722117e-06, "loss": 1.1717, "step": 3682 }, { "epoch": 0.77, "learning_rate": 2.54625430348871e-06, "loss": 1.2048, "step": 3683 }, { "epoch": 0.77, "learning_rate": 2.5417140042685985e-06, "loss": 1.2223, "step": 3684 }, { "epoch": 0.78, "learning_rate": 2.537177167219511e-06, "loss": 1.2407, "step": 3685 }, { "epoch": 0.78, "learning_rate": 2.5326437944474667e-06, "loss": 1.1444, "step": 3686 }, { "epoch": 0.78, "learning_rate": 2.5281138880568736e-06, "loss": 1.1983, "step": 3687 }, { "epoch": 0.78, "learning_rate": 2.5235874501505418e-06, "loss": 1.2007, "step": 3688 }, { "epoch": 0.78, "learning_rate": 2.5190644828296585e-06, "loss": 1.2141, "step": 3689 }, { "epoch": 0.78, "learning_rate": 2.5145449881938134e-06, "loss": 1.2152, "step": 3690 }, { "epoch": 0.78, "learning_rate": 2.5100289683409695e-06, "loss": 1.2001, "step": 3691 }, { "epoch": 0.78, "learning_rate": 2.5055164253674913e-06, "loss": 1.199, "step": 3692 }, { "epoch": 0.78, "learning_rate": 2.501007361368115e-06, "loss": 1.1673, "step": 3693 }, { "epoch": 0.78, "learning_rate": 2.496501778435977e-06, "loss": 1.1929, "step": 3694 }, { "epoch": 0.78, "learning_rate": 2.491999678662582e-06, "loss": 1.1376, "step": 3695 }, { "epoch": 0.78, "learning_rate": 2.487501064137833e-06, "loss": 1.1782, "step": 3696 }, { "epoch": 0.78, "learning_rate": 2.4830059369500005e-06, "loss": 1.1941, "step": 3697 }, { "epoch": 0.78, "learning_rate": 2.478514299185749e-06, "loss": 1.2004, "step": 3698 }, { "epoch": 0.78, "learning_rate": 2.4740261529301167e-06, "loss": 1.165, "step": 3699 }, { "epoch": 0.78, "learning_rate": 2.469541500266518e-06, "loss": 1.1687, "step": 3700 }, { "epoch": 0.78, "learning_rate": 2.465060343276755e-06, "loss": 1.1634, "step": 3701 }, { "epoch": 0.78, "learning_rate": 2.460582684040995e-06, "loss": 1.2312, "step": 3702 }, { "epoch": 0.78, "learning_rate": 2.456108524637797e-06, "loss": 1.1746, "step": 3703 }, { "epoch": 0.78, "learning_rate": 2.451637867144081e-06, "loss": 1.1557, "step": 3704 }, { "epoch": 0.78, "learning_rate": 2.4471707136351443e-06, "loss": 1.1815, "step": 3705 }, { "epoch": 0.78, "learning_rate": 2.4427070661846675e-06, "loss": 1.1933, "step": 3706 }, { "epoch": 0.78, "learning_rate": 2.4382469268646923e-06, "loss": 1.1745, "step": 3707 }, { "epoch": 0.78, "learning_rate": 2.433790297745634e-06, "loss": 1.1607, "step": 3708 }, { "epoch": 0.78, "learning_rate": 2.4293371808962853e-06, "loss": 1.2123, "step": 3709 }, { "epoch": 0.78, "learning_rate": 2.424887578383799e-06, "loss": 1.199, "step": 3710 }, { "epoch": 0.78, "learning_rate": 2.4204414922737076e-06, "loss": 1.2266, "step": 3711 }, { "epoch": 0.78, "learning_rate": 2.4159989246299e-06, "loss": 1.23, "step": 3712 }, { "epoch": 0.78, "learning_rate": 2.4115598775146356e-06, "loss": 1.1974, "step": 3713 }, { "epoch": 0.78, "learning_rate": 2.407124352988546e-06, "loss": 1.1546, "step": 3714 }, { "epoch": 0.78, "learning_rate": 2.402692353110616e-06, "loss": 1.1744, "step": 3715 }, { "epoch": 0.78, "learning_rate": 2.3982638799382075e-06, "loss": 1.1629, "step": 3716 }, { "epoch": 0.78, "learning_rate": 2.393838935527031e-06, "loss": 1.2127, "step": 3717 }, { "epoch": 0.78, "learning_rate": 2.3894175219311733e-06, "loss": 1.2365, "step": 3718 }, { "epoch": 0.78, "learning_rate": 2.3849996412030684e-06, "loss": 1.2084, "step": 3719 }, { "epoch": 0.78, "learning_rate": 2.3805852953935237e-06, "loss": 1.2333, "step": 3720 }, { "epoch": 0.78, "learning_rate": 2.3761744865516922e-06, "loss": 1.1932, "step": 3721 }, { "epoch": 0.78, "learning_rate": 2.3717672167250996e-06, "loss": 1.2029, "step": 3722 }, { "epoch": 0.78, "learning_rate": 2.3673634879596165e-06, "loss": 1.1948, "step": 3723 }, { "epoch": 0.78, "learning_rate": 2.362963302299471e-06, "loss": 1.1797, "step": 3724 }, { "epoch": 0.78, "learning_rate": 2.3585666617872572e-06, "loss": 1.2142, "step": 3725 }, { "epoch": 0.78, "learning_rate": 2.354173568463909e-06, "loss": 1.1959, "step": 3726 }, { "epoch": 0.78, "learning_rate": 2.3497840243687266e-06, "loss": 1.2549, "step": 3727 }, { "epoch": 0.78, "learning_rate": 2.3453980315393543e-06, "loss": 1.1652, "step": 3728 }, { "epoch": 0.78, "learning_rate": 2.3410155920117872e-06, "loss": 1.1904, "step": 3729 }, { "epoch": 0.78, "learning_rate": 2.33663670782038e-06, "loss": 1.1504, "step": 3730 }, { "epoch": 0.78, "learning_rate": 2.3322613809978258e-06, "loss": 1.1872, "step": 3731 }, { "epoch": 0.78, "learning_rate": 2.327889613575176e-06, "loss": 1.2065, "step": 3732 }, { "epoch": 0.79, "learning_rate": 2.3235214075818245e-06, "loss": 1.1708, "step": 3733 }, { "epoch": 0.79, "learning_rate": 2.319156765045508e-06, "loss": 1.2141, "step": 3734 }, { "epoch": 0.79, "learning_rate": 2.3147956879923207e-06, "loss": 1.2047, "step": 3735 }, { "epoch": 0.79, "learning_rate": 2.3104381784466923e-06, "loss": 1.1536, "step": 3736 }, { "epoch": 0.79, "learning_rate": 2.3060842384313943e-06, "loss": 1.2315, "step": 3737 }, { "epoch": 0.79, "learning_rate": 2.301733869967554e-06, "loss": 1.177, "step": 3738 }, { "epoch": 0.79, "learning_rate": 2.2973870750746253e-06, "loss": 1.182, "step": 3739 }, { "epoch": 0.79, "learning_rate": 2.293043855770416e-06, "loss": 1.1685, "step": 3740 }, { "epoch": 0.79, "learning_rate": 2.288704214071065e-06, "loss": 1.2094, "step": 3741 }, { "epoch": 0.79, "learning_rate": 2.2843681519910576e-06, "loss": 1.1718, "step": 3742 }, { "epoch": 0.79, "learning_rate": 2.2800356715432104e-06, "loss": 1.1934, "step": 3743 }, { "epoch": 0.79, "learning_rate": 2.275706774738685e-06, "loss": 1.1833, "step": 3744 }, { "epoch": 0.79, "learning_rate": 2.2713814635869703e-06, "loss": 1.1852, "step": 3745 }, { "epoch": 0.79, "learning_rate": 2.2670597400959006e-06, "loss": 1.1304, "step": 3746 }, { "epoch": 0.79, "learning_rate": 2.2627416062716366e-06, "loss": 1.1853, "step": 3747 }, { "epoch": 0.79, "learning_rate": 2.258427064118679e-06, "loss": 1.1759, "step": 3748 }, { "epoch": 0.79, "learning_rate": 2.2541161156398583e-06, "loss": 1.2181, "step": 3749 }, { "epoch": 0.79, "learning_rate": 2.249808762836332e-06, "loss": 1.1789, "step": 3750 }, { "epoch": 0.79, "learning_rate": 2.245505007707599e-06, "loss": 1.1937, "step": 3751 }, { "epoch": 0.79, "learning_rate": 2.2412048522514807e-06, "loss": 1.1233, "step": 3752 }, { "epoch": 0.79, "learning_rate": 2.2369082984641253e-06, "loss": 1.1885, "step": 3753 }, { "epoch": 0.79, "learning_rate": 2.232615348340019e-06, "loss": 1.1698, "step": 3754 }, { "epoch": 0.79, "learning_rate": 2.2283260038719644e-06, "loss": 1.2135, "step": 3755 }, { "epoch": 0.79, "learning_rate": 2.224040267051101e-06, "loss": 1.2094, "step": 3756 }, { "epoch": 0.79, "learning_rate": 2.219758139866882e-06, "loss": 1.1965, "step": 3757 }, { "epoch": 0.79, "learning_rate": 2.215479624307092e-06, "loss": 1.1929, "step": 3758 }, { "epoch": 0.79, "learning_rate": 2.21120472235784e-06, "loss": 1.2048, "step": 3759 }, { "epoch": 0.79, "learning_rate": 2.206933436003552e-06, "loss": 1.1444, "step": 3760 }, { "epoch": 0.79, "learning_rate": 2.2026657672269834e-06, "loss": 1.171, "step": 3761 }, { "epoch": 0.79, "learning_rate": 2.198401718009203e-06, "loss": 1.2006, "step": 3762 }, { "epoch": 0.79, "learning_rate": 2.1941412903296012e-06, "loss": 1.1406, "step": 3763 }, { "epoch": 0.79, "learning_rate": 2.189884486165892e-06, "loss": 1.2505, "step": 3764 }, { "epoch": 0.79, "learning_rate": 2.185631307494098e-06, "loss": 1.2205, "step": 3765 }, { "epoch": 0.79, "learning_rate": 2.181381756288572e-06, "loss": 1.196, "step": 3766 }, { "epoch": 0.79, "learning_rate": 2.177135834521967e-06, "loss": 1.147, "step": 3767 }, { "epoch": 0.79, "learning_rate": 2.1728935441652687e-06, "loss": 1.2046, "step": 3768 }, { "epoch": 0.79, "learning_rate": 2.16865488718776e-06, "loss": 1.2141, "step": 3769 }, { "epoch": 0.79, "learning_rate": 2.1644198655570503e-06, "loss": 1.2388, "step": 3770 }, { "epoch": 0.79, "learning_rate": 2.1601884812390527e-06, "loss": 1.1992, "step": 3771 }, { "epoch": 0.79, "learning_rate": 2.1559607361979995e-06, "loss": 1.2121, "step": 3772 }, { "epoch": 0.79, "learning_rate": 2.1517366323964283e-06, "loss": 1.2155, "step": 3773 }, { "epoch": 0.79, "learning_rate": 2.147516171795183e-06, "loss": 1.1684, "step": 3774 }, { "epoch": 0.79, "learning_rate": 2.143299356353429e-06, "loss": 1.1862, "step": 3775 }, { "epoch": 0.79, "learning_rate": 2.139086188028623e-06, "loss": 1.2169, "step": 3776 }, { "epoch": 0.79, "learning_rate": 2.1348766687765456e-06, "loss": 1.2497, "step": 3777 }, { "epoch": 0.79, "learning_rate": 2.1306708005512712e-06, "loss": 1.2046, "step": 3778 }, { "epoch": 0.79, "learning_rate": 2.12646858530518e-06, "loss": 1.2153, "step": 3779 }, { "epoch": 0.8, "learning_rate": 2.122270024988967e-06, "loss": 1.1817, "step": 3780 }, { "epoch": 0.8, "learning_rate": 2.118075121551617e-06, "loss": 1.1842, "step": 3781 }, { "epoch": 0.8, "learning_rate": 2.1138838769404304e-06, "loss": 1.2151, "step": 3782 }, { "epoch": 0.8, "learning_rate": 2.1096962931009967e-06, "loss": 1.1823, "step": 3783 }, { "epoch": 0.8, "learning_rate": 2.105512371977212e-06, "loss": 1.1818, "step": 3784 }, { "epoch": 0.8, "learning_rate": 2.1013321155112753e-06, "loss": 1.1817, "step": 3785 }, { "epoch": 0.8, "learning_rate": 2.0971555256436793e-06, "loss": 1.1712, "step": 3786 }, { "epoch": 0.8, "learning_rate": 2.092982604313214e-06, "loss": 1.2387, "step": 3787 }, { "epoch": 0.8, "learning_rate": 2.088813353456974e-06, "loss": 1.2082, "step": 3788 }, { "epoch": 0.8, "learning_rate": 2.08464777501034e-06, "loss": 1.2128, "step": 3789 }, { "epoch": 0.8, "learning_rate": 2.0804858709069974e-06, "loss": 1.1685, "step": 3790 }, { "epoch": 0.8, "learning_rate": 2.076327643078917e-06, "loss": 1.2128, "step": 3791 }, { "epoch": 0.8, "learning_rate": 2.072173093456369e-06, "loss": 1.2243, "step": 3792 }, { "epoch": 0.8, "learning_rate": 2.0680222239679183e-06, "loss": 1.2618, "step": 3793 }, { "epoch": 0.8, "learning_rate": 2.0638750365404146e-06, "loss": 1.1742, "step": 3794 }, { "epoch": 0.8, "learning_rate": 2.0597315330989985e-06, "loss": 1.2211, "step": 3795 }, { "epoch": 0.8, "learning_rate": 2.0555917155671114e-06, "loss": 1.1918, "step": 3796 }, { "epoch": 0.8, "learning_rate": 2.0514555858664663e-06, "loss": 1.168, "step": 3797 }, { "epoch": 0.8, "learning_rate": 2.047323145917083e-06, "loss": 1.1922, "step": 3798 }, { "epoch": 0.8, "learning_rate": 2.043194397637254e-06, "loss": 1.1459, "step": 3799 }, { "epoch": 0.8, "learning_rate": 2.0390693429435626e-06, "loss": 1.1672, "step": 3800 }, { "epoch": 0.8, "learning_rate": 2.0349479837508823e-06, "loss": 1.1468, "step": 3801 }, { "epoch": 0.8, "learning_rate": 2.0308303219723657e-06, "loss": 1.1931, "step": 3802 }, { "epoch": 0.8, "learning_rate": 2.0267163595194474e-06, "loss": 1.1771, "step": 3803 }, { "epoch": 0.8, "learning_rate": 2.0226060983018527e-06, "loss": 1.1921, "step": 3804 }, { "epoch": 0.8, "learning_rate": 2.01849954022758e-06, "loss": 1.2163, "step": 3805 }, { "epoch": 0.8, "learning_rate": 2.0143966872029166e-06, "loss": 1.1964, "step": 3806 }, { "epoch": 0.8, "learning_rate": 2.0102975411324234e-06, "loss": 1.2116, "step": 3807 }, { "epoch": 0.8, "learning_rate": 2.0062021039189406e-06, "loss": 1.1958, "step": 3808 }, { "epoch": 0.8, "learning_rate": 2.0021103774635952e-06, "loss": 1.2219, "step": 3809 }, { "epoch": 0.8, "learning_rate": 1.998022363665779e-06, "loss": 1.1857, "step": 3810 }, { "epoch": 0.8, "learning_rate": 1.9939380644231744e-06, "loss": 1.2178, "step": 3811 }, { "epoch": 0.8, "learning_rate": 1.9898574816317274e-06, "loss": 1.2131, "step": 3812 }, { "epoch": 0.8, "learning_rate": 1.985780617185662e-06, "loss": 1.1978, "step": 3813 }, { "epoch": 0.8, "learning_rate": 1.981707472977481e-06, "loss": 1.1981, "step": 3814 }, { "epoch": 0.8, "learning_rate": 1.977638050897954e-06, "loss": 1.2096, "step": 3815 }, { "epoch": 0.8, "learning_rate": 1.9735723528361284e-06, "loss": 1.1814, "step": 3816 }, { "epoch": 0.8, "learning_rate": 1.9695103806793203e-06, "loss": 1.1944, "step": 3817 }, { "epoch": 0.8, "learning_rate": 1.965452136313113e-06, "loss": 1.2358, "step": 3818 }, { "epoch": 0.8, "learning_rate": 1.961397621621367e-06, "loss": 1.1932, "step": 3819 }, { "epoch": 0.8, "learning_rate": 1.957346838486204e-06, "loss": 1.1853, "step": 3820 }, { "epoch": 0.8, "learning_rate": 1.9532997887880135e-06, "loss": 1.1593, "step": 3821 }, { "epoch": 0.8, "learning_rate": 1.949256474405461e-06, "loss": 1.213, "step": 3822 }, { "epoch": 0.8, "learning_rate": 1.945216897215467e-06, "loss": 1.2074, "step": 3823 }, { "epoch": 0.8, "learning_rate": 1.9411810590932225e-06, "loss": 1.235, "step": 3824 }, { "epoch": 0.8, "learning_rate": 1.937148961912185e-06, "loss": 1.1643, "step": 3825 }, { "epoch": 0.8, "learning_rate": 1.93312060754407e-06, "loss": 1.2014, "step": 3826 }, { "epoch": 0.8, "learning_rate": 1.929095997858861e-06, "loss": 1.2363, "step": 3827 }, { "epoch": 0.81, "learning_rate": 1.9250751347247987e-06, "loss": 1.2117, "step": 3828 }, { "epoch": 0.81, "learning_rate": 1.921058020008385e-06, "loss": 1.2127, "step": 3829 }, { "epoch": 0.81, "learning_rate": 1.917044655574387e-06, "loss": 1.1971, "step": 3830 }, { "epoch": 0.81, "learning_rate": 1.9130350432858257e-06, "loss": 1.195, "step": 3831 }, { "epoch": 0.81, "learning_rate": 1.909029185003978e-06, "loss": 1.1724, "step": 3832 }, { "epoch": 0.81, "learning_rate": 1.9050270825883876e-06, "loss": 1.1878, "step": 3833 }, { "epoch": 0.81, "learning_rate": 1.901028737896844e-06, "loss": 1.127, "step": 3834 }, { "epoch": 0.81, "learning_rate": 1.897034152785402e-06, "loss": 1.1972, "step": 3835 }, { "epoch": 0.81, "learning_rate": 1.893043329108365e-06, "loss": 1.1826, "step": 3836 }, { "epoch": 0.81, "learning_rate": 1.889056268718289e-06, "loss": 1.1981, "step": 3837 }, { "epoch": 0.81, "learning_rate": 1.8850729734659912e-06, "loss": 1.217, "step": 3838 }, { "epoch": 0.81, "learning_rate": 1.8810934452005292e-06, "loss": 1.1801, "step": 3839 }, { "epoch": 0.81, "learning_rate": 1.8771176857692231e-06, "loss": 1.1734, "step": 3840 }, { "epoch": 0.81, "learning_rate": 1.8731456970176399e-06, "loss": 1.2372, "step": 3841 }, { "epoch": 0.81, "learning_rate": 1.8691774807895913e-06, "loss": 1.2418, "step": 3842 }, { "epoch": 0.81, "learning_rate": 1.8652130389271462e-06, "loss": 1.2095, "step": 3843 }, { "epoch": 0.81, "learning_rate": 1.8612523732706145e-06, "loss": 1.1728, "step": 3844 }, { "epoch": 0.81, "learning_rate": 1.8572954856585535e-06, "loss": 1.2014, "step": 3845 }, { "epoch": 0.81, "learning_rate": 1.8533423779277725e-06, "loss": 1.2529, "step": 3846 }, { "epoch": 0.81, "learning_rate": 1.8493930519133195e-06, "loss": 1.2426, "step": 3847 }, { "epoch": 0.81, "learning_rate": 1.8454475094484936e-06, "loss": 1.2042, "step": 3848 }, { "epoch": 0.81, "learning_rate": 1.8415057523648317e-06, "loss": 1.1928, "step": 3849 }, { "epoch": 0.81, "learning_rate": 1.8375677824921134e-06, "loss": 1.132, "step": 3850 }, { "epoch": 0.81, "learning_rate": 1.8336336016583667e-06, "loss": 1.218, "step": 3851 }, { "epoch": 0.81, "learning_rate": 1.8297032116898562e-06, "loss": 1.1923, "step": 3852 }, { "epoch": 0.81, "learning_rate": 1.8257766144110823e-06, "loss": 1.2383, "step": 3853 }, { "epoch": 0.81, "learning_rate": 1.821853811644796e-06, "loss": 1.2258, "step": 3854 }, { "epoch": 0.81, "learning_rate": 1.817934805211976e-06, "loss": 1.1901, "step": 3855 }, { "epoch": 0.81, "learning_rate": 1.8140195969318475e-06, "loss": 1.1739, "step": 3856 }, { "epoch": 0.81, "learning_rate": 1.8101081886218663e-06, "loss": 1.2065, "step": 3857 }, { "epoch": 0.81, "learning_rate": 1.8062005820977236e-06, "loss": 1.2324, "step": 3858 }, { "epoch": 0.81, "learning_rate": 1.8022967791733527e-06, "loss": 1.1995, "step": 3859 }, { "epoch": 0.81, "learning_rate": 1.7983967816609138e-06, "loss": 1.208, "step": 3860 }, { "epoch": 0.81, "learning_rate": 1.794500591370808e-06, "loss": 1.195, "step": 3861 }, { "epoch": 0.81, "learning_rate": 1.7906082101116618e-06, "loss": 1.0981, "step": 3862 }, { "epoch": 0.81, "learning_rate": 1.7867196396903352e-06, "loss": 1.1499, "step": 3863 }, { "epoch": 0.81, "learning_rate": 1.7828348819119223e-06, "loss": 1.1857, "step": 3864 }, { "epoch": 0.81, "learning_rate": 1.7789539385797494e-06, "loss": 1.1976, "step": 3865 }, { "epoch": 0.81, "learning_rate": 1.775076811495362e-06, "loss": 1.1903, "step": 3866 }, { "epoch": 0.81, "learning_rate": 1.771203502458546e-06, "loss": 1.1611, "step": 3867 }, { "epoch": 0.81, "learning_rate": 1.7673340132673056e-06, "loss": 1.1891, "step": 3868 }, { "epoch": 0.81, "learning_rate": 1.7634683457178792e-06, "loss": 1.2145, "step": 3869 }, { "epoch": 0.81, "learning_rate": 1.7596065016047259e-06, "loss": 1.2172, "step": 3870 }, { "epoch": 0.81, "learning_rate": 1.7557484827205285e-06, "loss": 1.1607, "step": 3871 }, { "epoch": 0.81, "learning_rate": 1.751894290856203e-06, "loss": 1.2015, "step": 3872 }, { "epoch": 0.81, "learning_rate": 1.7480439278008798e-06, "loss": 1.178, "step": 3873 }, { "epoch": 0.81, "learning_rate": 1.7441973953419145e-06, "loss": 1.1868, "step": 3874 }, { "epoch": 0.82, "learning_rate": 1.7403546952648886e-06, "loss": 1.1974, "step": 3875 }, { "epoch": 0.82, "learning_rate": 1.7365158293535966e-06, "loss": 1.1536, "step": 3876 }, { "epoch": 0.82, "learning_rate": 1.732680799390064e-06, "loss": 1.2006, "step": 3877 }, { "epoch": 0.82, "learning_rate": 1.7288496071545258e-06, "loss": 1.249, "step": 3878 }, { "epoch": 0.82, "learning_rate": 1.725022254425438e-06, "loss": 1.218, "step": 3879 }, { "epoch": 0.82, "learning_rate": 1.7211987429794797e-06, "loss": 1.1568, "step": 3880 }, { "epoch": 0.82, "learning_rate": 1.7173790745915398e-06, "loss": 1.1712, "step": 3881 }, { "epoch": 0.82, "learning_rate": 1.7135632510347245e-06, "loss": 1.2313, "step": 3882 }, { "epoch": 0.82, "learning_rate": 1.7097512740803612e-06, "loss": 1.1883, "step": 3883 }, { "epoch": 0.82, "learning_rate": 1.7059431454979825e-06, "loss": 1.2191, "step": 3884 }, { "epoch": 0.82, "learning_rate": 1.7021388670553452e-06, "loss": 1.2129, "step": 3885 }, { "epoch": 0.82, "learning_rate": 1.6983384405184066e-06, "loss": 1.2359, "step": 3886 }, { "epoch": 0.82, "learning_rate": 1.6945418676513482e-06, "loss": 1.2245, "step": 3887 }, { "epoch": 0.82, "learning_rate": 1.6907491502165507e-06, "loss": 1.2039, "step": 3888 }, { "epoch": 0.82, "learning_rate": 1.6869602899746185e-06, "loss": 1.2092, "step": 3889 }, { "epoch": 0.82, "learning_rate": 1.6831752886843511e-06, "loss": 1.1874, "step": 3890 }, { "epoch": 0.82, "learning_rate": 1.6793941481027687e-06, "loss": 1.2121, "step": 3891 }, { "epoch": 0.82, "learning_rate": 1.6756168699850906e-06, "loss": 1.1988, "step": 3892 }, { "epoch": 0.82, "learning_rate": 1.6718434560847506e-06, "loss": 1.1957, "step": 3893 }, { "epoch": 0.82, "learning_rate": 1.6680739081533837e-06, "loss": 1.1587, "step": 3894 }, { "epoch": 0.82, "learning_rate": 1.6643082279408285e-06, "loss": 1.1831, "step": 3895 }, { "epoch": 0.82, "learning_rate": 1.6605464171951358e-06, "loss": 1.1619, "step": 3896 }, { "epoch": 0.82, "learning_rate": 1.6567884776625508e-06, "loss": 1.2275, "step": 3897 }, { "epoch": 0.82, "learning_rate": 1.6530344110875318e-06, "loss": 1.162, "step": 3898 }, { "epoch": 0.82, "learning_rate": 1.6492842192127324e-06, "loss": 1.1662, "step": 3899 }, { "epoch": 0.82, "learning_rate": 1.6455379037790043e-06, "loss": 1.2019, "step": 3900 }, { "epoch": 0.82, "learning_rate": 1.641795466525411e-06, "loss": 1.2002, "step": 3901 }, { "epoch": 0.82, "learning_rate": 1.6380569091892063e-06, "loss": 1.1895, "step": 3902 }, { "epoch": 0.82, "learning_rate": 1.6343222335058439e-06, "loss": 1.1417, "step": 3903 }, { "epoch": 0.82, "learning_rate": 1.630591441208983e-06, "loss": 1.2148, "step": 3904 }, { "epoch": 0.82, "learning_rate": 1.626864534030469e-06, "loss": 1.2151, "step": 3905 }, { "epoch": 0.82, "learning_rate": 1.6231415137003536e-06, "loss": 1.2026, "step": 3906 }, { "epoch": 0.82, "learning_rate": 1.6194223819468792e-06, "loss": 1.2143, "step": 3907 }, { "epoch": 0.82, "learning_rate": 1.6157071404964809e-06, "loss": 1.1959, "step": 3908 }, { "epoch": 0.82, "learning_rate": 1.611995791073796e-06, "loss": 1.187, "step": 3909 }, { "epoch": 0.82, "learning_rate": 1.6082883354016455e-06, "loss": 1.1651, "step": 3910 }, { "epoch": 0.82, "learning_rate": 1.6045847752010524e-06, "loss": 1.1909, "step": 3911 }, { "epoch": 0.82, "learning_rate": 1.6008851121912217e-06, "loss": 1.2223, "step": 3912 }, { "epoch": 0.82, "learning_rate": 1.5971893480895583e-06, "loss": 1.2162, "step": 3913 }, { "epoch": 0.82, "learning_rate": 1.5934974846116491e-06, "loss": 1.1705, "step": 3914 }, { "epoch": 0.82, "learning_rate": 1.5898095234712785e-06, "loss": 1.2017, "step": 3915 }, { "epoch": 0.82, "learning_rate": 1.5861254663804116e-06, "loss": 1.1451, "step": 3916 }, { "epoch": 0.82, "learning_rate": 1.582445315049208e-06, "loss": 1.2007, "step": 3917 }, { "epoch": 0.82, "learning_rate": 1.5787690711860071e-06, "loss": 1.1926, "step": 3918 }, { "epoch": 0.82, "learning_rate": 1.5750967364973434e-06, "loss": 1.1966, "step": 3919 }, { "epoch": 0.82, "learning_rate": 1.571428312687928e-06, "loss": 1.2434, "step": 3920 }, { "epoch": 0.82, "learning_rate": 1.56776380146066e-06, "loss": 1.1577, "step": 3921 }, { "epoch": 0.82, "learning_rate": 1.5641032045166261e-06, "loss": 1.1309, "step": 3922 }, { "epoch": 0.83, "learning_rate": 1.5604465235550893e-06, "loss": 1.1874, "step": 3923 }, { "epoch": 0.83, "learning_rate": 1.5567937602734973e-06, "loss": 1.1706, "step": 3924 }, { "epoch": 0.83, "learning_rate": 1.5531449163674828e-06, "loss": 1.1722, "step": 3925 }, { "epoch": 0.83, "learning_rate": 1.549499993530853e-06, "loss": 1.1847, "step": 3926 }, { "epoch": 0.83, "learning_rate": 1.5458589934556011e-06, "loss": 1.1981, "step": 3927 }, { "epoch": 0.83, "learning_rate": 1.5422219178318954e-06, "loss": 1.1572, "step": 3928 }, { "epoch": 0.83, "learning_rate": 1.5385887683480794e-06, "loss": 1.2346, "step": 3929 }, { "epoch": 0.83, "learning_rate": 1.5349595466906841e-06, "loss": 1.202, "step": 3930 }, { "epoch": 0.83, "learning_rate": 1.5313342545444088e-06, "loss": 1.1859, "step": 3931 }, { "epoch": 0.83, "learning_rate": 1.5277128935921271e-06, "loss": 1.207, "step": 3932 }, { "epoch": 0.83, "learning_rate": 1.5240954655148977e-06, "loss": 1.1981, "step": 3933 }, { "epoch": 0.83, "learning_rate": 1.5204819719919417e-06, "loss": 1.2481, "step": 3934 }, { "epoch": 0.83, "learning_rate": 1.5168724147006654e-06, "loss": 1.1975, "step": 3935 }, { "epoch": 0.83, "learning_rate": 1.5132667953166359e-06, "loss": 1.1442, "step": 3936 }, { "epoch": 0.83, "learning_rate": 1.5096651155136045e-06, "loss": 1.1702, "step": 3937 }, { "epoch": 0.83, "learning_rate": 1.5060673769634825e-06, "loss": 1.1837, "step": 3938 }, { "epoch": 0.83, "learning_rate": 1.5024735813363623e-06, "loss": 1.2122, "step": 3939 }, { "epoch": 0.83, "learning_rate": 1.4988837303004955e-06, "loss": 1.2174, "step": 3940 }, { "epoch": 0.83, "learning_rate": 1.4952978255223105e-06, "loss": 1.1503, "step": 3941 }, { "epoch": 0.83, "learning_rate": 1.4917158686663992e-06, "loss": 1.2332, "step": 3942 }, { "epoch": 0.83, "learning_rate": 1.4881378613955254e-06, "loss": 1.1688, "step": 3943 }, { "epoch": 0.83, "learning_rate": 1.4845638053706146e-06, "loss": 1.1657, "step": 3944 }, { "epoch": 0.83, "learning_rate": 1.480993702250758e-06, "loss": 1.2206, "step": 3945 }, { "epoch": 0.83, "learning_rate": 1.4774275536932193e-06, "loss": 1.1624, "step": 3946 }, { "epoch": 0.83, "learning_rate": 1.473865361353416e-06, "loss": 1.188, "step": 3947 }, { "epoch": 0.83, "learning_rate": 1.47030712688494e-06, "loss": 1.2176, "step": 3948 }, { "epoch": 0.83, "learning_rate": 1.4667528519395358e-06, "loss": 1.1635, "step": 3949 }, { "epoch": 0.83, "learning_rate": 1.4632025381671134e-06, "loss": 1.2314, "step": 3950 }, { "epoch": 0.83, "learning_rate": 1.45965618721575e-06, "loss": 1.2061, "step": 3951 }, { "epoch": 0.83, "learning_rate": 1.4561138007316755e-06, "loss": 1.2459, "step": 3952 }, { "epoch": 0.83, "learning_rate": 1.4525753803592802e-06, "loss": 1.2212, "step": 3953 }, { "epoch": 0.83, "learning_rate": 1.4490409277411188e-06, "loss": 1.2045, "step": 3954 }, { "epoch": 0.83, "learning_rate": 1.4455104445178958e-06, "loss": 1.187, "step": 3955 }, { "epoch": 0.83, "learning_rate": 1.4419839323284845e-06, "loss": 1.2546, "step": 3956 }, { "epoch": 0.83, "learning_rate": 1.4384613928099045e-06, "loss": 1.2038, "step": 3957 }, { "epoch": 0.83, "learning_rate": 1.4349428275973332e-06, "loss": 1.1527, "step": 3958 }, { "epoch": 0.83, "learning_rate": 1.4314282383241097e-06, "loss": 1.2257, "step": 3959 }, { "epoch": 0.83, "learning_rate": 1.4279176266217166e-06, "loss": 1.1868, "step": 3960 }, { "epoch": 0.83, "learning_rate": 1.4244109941198026e-06, "loss": 1.2094, "step": 3961 }, { "epoch": 0.83, "learning_rate": 1.4209083424461579e-06, "loss": 1.1362, "step": 3962 }, { "epoch": 0.83, "learning_rate": 1.417409673226734e-06, "loss": 1.2134, "step": 3963 }, { "epoch": 0.83, "learning_rate": 1.413914988085624e-06, "loss": 1.2065, "step": 3964 }, { "epoch": 0.83, "learning_rate": 1.4104242886450826e-06, "loss": 1.2131, "step": 3965 }, { "epoch": 0.83, "learning_rate": 1.4069375765255032e-06, "loss": 1.177, "step": 3966 }, { "epoch": 0.83, "learning_rate": 1.4034548533454406e-06, "loss": 1.1709, "step": 3967 }, { "epoch": 0.83, "learning_rate": 1.3999761207215835e-06, "loss": 1.2813, "step": 3968 }, { "epoch": 0.83, "learning_rate": 1.3965013802687822e-06, "loss": 1.1929, "step": 3969 }, { "epoch": 0.84, "learning_rate": 1.3930306336000243e-06, "loss": 1.1966, "step": 3970 }, { "epoch": 0.84, "learning_rate": 1.3895638823264447e-06, "loss": 1.2151, "step": 3971 }, { "epoch": 0.84, "learning_rate": 1.38610112805733e-06, "loss": 1.2046, "step": 3972 }, { "epoch": 0.84, "learning_rate": 1.3826423724001037e-06, "loss": 1.1875, "step": 3973 }, { "epoch": 0.84, "learning_rate": 1.3791876169603357e-06, "loss": 1.2008, "step": 3974 }, { "epoch": 0.84, "learning_rate": 1.3757368633417422e-06, "loss": 1.1387, "step": 3975 }, { "epoch": 0.84, "learning_rate": 1.3722901131461752e-06, "loss": 1.1658, "step": 3976 }, { "epoch": 0.84, "learning_rate": 1.368847367973638e-06, "loss": 1.1637, "step": 3977 }, { "epoch": 0.84, "learning_rate": 1.3654086294222645e-06, "loss": 1.2312, "step": 3978 }, { "epoch": 0.84, "learning_rate": 1.3619738990883325e-06, "loss": 1.1583, "step": 3979 }, { "epoch": 0.84, "learning_rate": 1.3585431785662628e-06, "loss": 1.2291, "step": 3980 }, { "epoch": 0.84, "learning_rate": 1.3551164694486108e-06, "loss": 1.2093, "step": 3981 }, { "epoch": 0.84, "learning_rate": 1.3516937733260682e-06, "loss": 1.2226, "step": 3982 }, { "epoch": 0.84, "learning_rate": 1.348275091787471e-06, "loss": 1.1658, "step": 3983 }, { "epoch": 0.84, "learning_rate": 1.3448604264197828e-06, "loss": 1.147, "step": 3984 }, { "epoch": 0.84, "learning_rate": 1.3414497788081115e-06, "loss": 1.1884, "step": 3985 }, { "epoch": 0.84, "learning_rate": 1.3380431505356905e-06, "loss": 1.1864, "step": 3986 }, { "epoch": 0.84, "learning_rate": 1.334640543183895e-06, "loss": 1.1939, "step": 3987 }, { "epoch": 0.84, "learning_rate": 1.3312419583322334e-06, "loss": 1.1946, "step": 3988 }, { "epoch": 0.84, "learning_rate": 1.3278473975583418e-06, "loss": 1.1402, "step": 3989 }, { "epoch": 0.84, "learning_rate": 1.3244568624379906e-06, "loss": 1.1948, "step": 3990 }, { "epoch": 0.84, "learning_rate": 1.3210703545450854e-06, "loss": 1.1932, "step": 3991 }, { "epoch": 0.84, "learning_rate": 1.317687875451653e-06, "loss": 1.1932, "step": 3992 }, { "epoch": 0.84, "learning_rate": 1.3143094267278634e-06, "loss": 1.2459, "step": 3993 }, { "epoch": 0.84, "learning_rate": 1.3109350099420026e-06, "loss": 1.2041, "step": 3994 }, { "epoch": 0.84, "learning_rate": 1.3075646266604913e-06, "loss": 1.2257, "step": 3995 }, { "epoch": 0.84, "learning_rate": 1.3041982784478802e-06, "loss": 1.2293, "step": 3996 }, { "epoch": 0.84, "learning_rate": 1.300835966866839e-06, "loss": 1.2408, "step": 3997 }, { "epoch": 0.84, "learning_rate": 1.297477693478173e-06, "loss": 1.2128, "step": 3998 }, { "epoch": 0.84, "learning_rate": 1.294123459840807e-06, "loss": 1.2078, "step": 3999 }, { "epoch": 0.84, "learning_rate": 1.2907732675117878e-06, "loss": 1.1928, "step": 4000 }, { "epoch": 0.84, "learning_rate": 1.2874271180462972e-06, "loss": 1.2412, "step": 4001 }, { "epoch": 0.84, "learning_rate": 1.284085012997629e-06, "loss": 1.2247, "step": 4002 }, { "epoch": 0.84, "learning_rate": 1.2807469539172025e-06, "loss": 1.2133, "step": 4003 }, { "epoch": 0.84, "learning_rate": 1.2774129423545633e-06, "loss": 1.2661, "step": 4004 }, { "epoch": 0.84, "learning_rate": 1.274082979857374e-06, "loss": 1.1526, "step": 4005 }, { "epoch": 0.84, "learning_rate": 1.27075706797142e-06, "loss": 1.1939, "step": 4006 }, { "epoch": 0.84, "learning_rate": 1.267435208240604e-06, "loss": 1.1854, "step": 4007 }, { "epoch": 0.84, "learning_rate": 1.2641174022069457e-06, "loss": 1.1639, "step": 4008 }, { "epoch": 0.84, "learning_rate": 1.2608036514105915e-06, "loss": 1.1939, "step": 4009 }, { "epoch": 0.84, "learning_rate": 1.257493957389796e-06, "loss": 1.1628, "step": 4010 }, { "epoch": 0.84, "learning_rate": 1.2541883216809347e-06, "loss": 1.2032, "step": 4011 }, { "epoch": 0.84, "learning_rate": 1.2508867458185037e-06, "loss": 1.2141, "step": 4012 }, { "epoch": 0.84, "learning_rate": 1.2475892313351035e-06, "loss": 1.21, "step": 4013 }, { "epoch": 0.84, "learning_rate": 1.2442957797614607e-06, "loss": 1.1933, "step": 4014 }, { "epoch": 0.84, "learning_rate": 1.2410063926264081e-06, "loss": 1.1904, "step": 4015 }, { "epoch": 0.84, "learning_rate": 1.2377210714568944e-06, "loss": 1.0871, "step": 4016 }, { "epoch": 0.84, "learning_rate": 1.2344398177779825e-06, "loss": 1.1941, "step": 4017 }, { "epoch": 0.85, "learning_rate": 1.2311626331128435e-06, "loss": 1.1615, "step": 4018 }, { "epoch": 0.85, "learning_rate": 1.2278895189827645e-06, "loss": 1.1756, "step": 4019 }, { "epoch": 0.85, "learning_rate": 1.2246204769071401e-06, "loss": 1.2144, "step": 4020 }, { "epoch": 0.85, "learning_rate": 1.22135550840347e-06, "loss": 1.2115, "step": 4021 }, { "epoch": 0.85, "learning_rate": 1.218094614987374e-06, "loss": 1.1769, "step": 4022 }, { "epoch": 0.85, "learning_rate": 1.2148377981725712e-06, "loss": 1.2208, "step": 4023 }, { "epoch": 0.85, "learning_rate": 1.211585059470889e-06, "loss": 1.1729, "step": 4024 }, { "epoch": 0.85, "learning_rate": 1.2083364003922682e-06, "loss": 1.1744, "step": 4025 }, { "epoch": 0.85, "learning_rate": 1.2050918224447462e-06, "loss": 1.259, "step": 4026 }, { "epoch": 0.85, "learning_rate": 1.2018513271344777e-06, "loss": 1.2286, "step": 4027 }, { "epoch": 0.85, "learning_rate": 1.1986149159657112e-06, "loss": 1.2344, "step": 4028 }, { "epoch": 0.85, "learning_rate": 1.1953825904408033e-06, "loss": 1.1907, "step": 4029 }, { "epoch": 0.85, "learning_rate": 1.1921543520602175e-06, "loss": 1.222, "step": 4030 }, { "epoch": 0.85, "learning_rate": 1.1889302023225157e-06, "loss": 1.1753, "step": 4031 }, { "epoch": 0.85, "learning_rate": 1.185710142724361e-06, "loss": 1.1029, "step": 4032 }, { "epoch": 0.85, "learning_rate": 1.1824941747605246e-06, "loss": 1.2088, "step": 4033 }, { "epoch": 0.85, "learning_rate": 1.1792822999238685e-06, "loss": 1.2327, "step": 4034 }, { "epoch": 0.85, "learning_rate": 1.1760745197053636e-06, "loss": 1.165, "step": 4035 }, { "epoch": 0.85, "learning_rate": 1.1728708355940765e-06, "loss": 1.1921, "step": 4036 }, { "epoch": 0.85, "learning_rate": 1.16967124907717e-06, "loss": 1.1999, "step": 4037 }, { "epoch": 0.85, "learning_rate": 1.1664757616399103e-06, "loss": 1.2246, "step": 4038 }, { "epoch": 0.85, "learning_rate": 1.1632843747656553e-06, "loss": 1.2008, "step": 4039 }, { "epoch": 0.85, "learning_rate": 1.1600970899358588e-06, "loss": 1.2134, "step": 4040 }, { "epoch": 0.85, "learning_rate": 1.1569139086300796e-06, "loss": 1.1715, "step": 4041 }, { "epoch": 0.85, "learning_rate": 1.153734832325959e-06, "loss": 1.154, "step": 4042 }, { "epoch": 0.85, "learning_rate": 1.1505598624992443e-06, "loss": 1.2171, "step": 4043 }, { "epoch": 0.85, "learning_rate": 1.1473890006237698e-06, "loss": 1.2363, "step": 4044 }, { "epoch": 0.85, "learning_rate": 1.1442222481714594e-06, "loss": 1.1435, "step": 4045 }, { "epoch": 0.85, "learning_rate": 1.1410596066123415e-06, "loss": 1.1344, "step": 4046 }, { "epoch": 0.85, "learning_rate": 1.1379010774145239e-06, "loss": 1.1923, "step": 4047 }, { "epoch": 0.85, "learning_rate": 1.1347466620442138e-06, "loss": 1.2105, "step": 4048 }, { "epoch": 0.85, "learning_rate": 1.1315963619657045e-06, "loss": 1.0918, "step": 4049 }, { "epoch": 0.85, "learning_rate": 1.1284501786413771e-06, "loss": 1.2176, "step": 4050 }, { "epoch": 0.85, "learning_rate": 1.125308113531709e-06, "loss": 1.1841, "step": 4051 }, { "epoch": 0.85, "learning_rate": 1.122170168095259e-06, "loss": 1.2123, "step": 4052 }, { "epoch": 0.85, "learning_rate": 1.1190363437886754e-06, "loss": 1.1927, "step": 4053 }, { "epoch": 0.85, "learning_rate": 1.1159066420666953e-06, "loss": 1.1974, "step": 4054 }, { "epoch": 0.85, "learning_rate": 1.1127810643821403e-06, "loss": 1.2465, "step": 4055 }, { "epoch": 0.85, "learning_rate": 1.1096596121859193e-06, "loss": 1.1625, "step": 4056 }, { "epoch": 0.85, "learning_rate": 1.1065422869270238e-06, "loss": 1.1992, "step": 4057 }, { "epoch": 0.85, "learning_rate": 1.1034290900525279e-06, "loss": 1.2052, "step": 4058 }, { "epoch": 0.85, "learning_rate": 1.1003200230075962e-06, "loss": 1.1972, "step": 4059 }, { "epoch": 0.85, "learning_rate": 1.0972150872354714e-06, "loss": 1.2491, "step": 4060 }, { "epoch": 0.85, "learning_rate": 1.0941142841774777e-06, "loss": 1.1566, "step": 4061 }, { "epoch": 0.85, "learning_rate": 1.0910176152730246e-06, "loss": 1.2082, "step": 4062 }, { "epoch": 0.85, "learning_rate": 1.0879250819595954e-06, "loss": 1.1595, "step": 4063 }, { "epoch": 0.85, "learning_rate": 1.0848366856727654e-06, "loss": 1.1812, "step": 4064 }, { "epoch": 0.86, "learning_rate": 1.0817524278461777e-06, "loss": 1.182, "step": 4065 }, { "epoch": 0.86, "learning_rate": 1.0786723099115593e-06, "loss": 1.1167, "step": 4066 }, { "epoch": 0.86, "learning_rate": 1.0755963332987162e-06, "loss": 1.208, "step": 4067 }, { "epoch": 0.86, "learning_rate": 1.0725244994355322e-06, "loss": 1.1421, "step": 4068 }, { "epoch": 0.86, "learning_rate": 1.069456809747964e-06, "loss": 1.1609, "step": 4069 }, { "epoch": 0.86, "learning_rate": 1.0663932656600507e-06, "loss": 1.1657, "step": 4070 }, { "epoch": 0.86, "learning_rate": 1.0633338685938998e-06, "loss": 1.1521, "step": 4071 }, { "epoch": 0.86, "learning_rate": 1.0602786199697025e-06, "loss": 1.1659, "step": 4072 }, { "epoch": 0.86, "learning_rate": 1.057227521205716e-06, "loss": 1.1884, "step": 4073 }, { "epoch": 0.86, "learning_rate": 1.054180573718273e-06, "loss": 1.1947, "step": 4074 }, { "epoch": 0.86, "learning_rate": 1.0511377789217857e-06, "loss": 1.1954, "step": 4075 }, { "epoch": 0.86, "learning_rate": 1.0480991382287298e-06, "loss": 1.1698, "step": 4076 }, { "epoch": 0.86, "learning_rate": 1.0450646530496577e-06, "loss": 1.1907, "step": 4077 }, { "epoch": 0.86, "learning_rate": 1.0420343247931919e-06, "loss": 1.1676, "step": 4078 }, { "epoch": 0.86, "learning_rate": 1.0390081548660235e-06, "loss": 1.2364, "step": 4079 }, { "epoch": 0.86, "learning_rate": 1.0359861446729158e-06, "loss": 1.2122, "step": 4080 }, { "epoch": 0.86, "learning_rate": 1.0329682956166986e-06, "loss": 1.172, "step": 4081 }, { "epoch": 0.86, "learning_rate": 1.0299546090982737e-06, "loss": 1.1158, "step": 4082 }, { "epoch": 0.86, "learning_rate": 1.0269450865166064e-06, "loss": 1.2256, "step": 4083 }, { "epoch": 0.86, "learning_rate": 1.0239397292687325e-06, "loss": 1.2452, "step": 4084 }, { "epoch": 0.86, "learning_rate": 1.0209385387497517e-06, "loss": 1.1574, "step": 4085 }, { "epoch": 0.86, "learning_rate": 1.0179415163528316e-06, "loss": 1.1934, "step": 4086 }, { "epoch": 0.86, "learning_rate": 1.0149486634692019e-06, "loss": 1.2259, "step": 4087 }, { "epoch": 0.86, "learning_rate": 1.011959981488162e-06, "loss": 1.1985, "step": 4088 }, { "epoch": 0.86, "learning_rate": 1.0089754717970701e-06, "loss": 1.1387, "step": 4089 }, { "epoch": 0.86, "learning_rate": 1.005995135781348e-06, "loss": 1.1987, "step": 4090 }, { "epoch": 0.86, "learning_rate": 1.0030189748244855e-06, "loss": 1.1845, "step": 4091 }, { "epoch": 0.86, "learning_rate": 1.0000469903080256e-06, "loss": 1.2134, "step": 4092 }, { "epoch": 0.86, "learning_rate": 9.970791836115812e-07, "loss": 1.2177, "step": 4093 }, { "epoch": 0.86, "learning_rate": 9.941155561128212e-07, "loss": 1.1821, "step": 4094 }, { "epoch": 0.86, "learning_rate": 9.911561091874722e-07, "loss": 1.2077, "step": 4095 }, { "epoch": 0.86, "learning_rate": 9.882008442093273e-07, "loss": 1.1286, "step": 4096 }, { "epoch": 0.86, "learning_rate": 9.852497625502311e-07, "loss": 1.2122, "step": 4097 }, { "epoch": 0.86, "learning_rate": 9.823028655800926e-07, "loss": 1.1877, "step": 4098 }, { "epoch": 0.86, "learning_rate": 9.79360154666873e-07, "loss": 1.1626, "step": 4099 }, { "epoch": 0.86, "learning_rate": 9.764216311765906e-07, "loss": 1.1581, "step": 4100 }, { "epoch": 0.86, "learning_rate": 9.734872964733256e-07, "loss": 1.1873, "step": 4101 }, { "epoch": 0.86, "learning_rate": 9.705571519192082e-07, "loss": 1.1532, "step": 4102 }, { "epoch": 0.86, "learning_rate": 9.676311988744225e-07, "loss": 1.1854, "step": 4103 }, { "epoch": 0.86, "learning_rate": 9.647094386972144e-07, "loss": 1.1371, "step": 4104 }, { "epoch": 0.86, "learning_rate": 9.617918727438734e-07, "loss": 1.1658, "step": 4105 }, { "epoch": 0.86, "learning_rate": 9.58878502368753e-07, "loss": 1.2125, "step": 4106 }, { "epoch": 0.86, "learning_rate": 9.559693289242467e-07, "loss": 1.1843, "step": 4107 }, { "epoch": 0.86, "learning_rate": 9.53064353760812e-07, "loss": 1.1797, "step": 4108 }, { "epoch": 0.86, "learning_rate": 9.50163578226948e-07, "loss": 1.1864, "step": 4109 }, { "epoch": 0.86, "learning_rate": 9.47267003669211e-07, "loss": 1.1852, "step": 4110 }, { "epoch": 0.86, "learning_rate": 9.443746314322011e-07, "loss": 1.1322, "step": 4111 }, { "epoch": 0.86, "learning_rate": 9.414864628585752e-07, "loss": 1.1891, "step": 4112 }, { "epoch": 0.87, "learning_rate": 9.386024992890308e-07, "loss": 1.1751, "step": 4113 }, { "epoch": 0.87, "learning_rate": 9.357227420623205e-07, "loss": 1.1715, "step": 4114 }, { "epoch": 0.87, "learning_rate": 9.328471925152383e-07, "loss": 1.2139, "step": 4115 }, { "epoch": 0.87, "learning_rate": 9.299758519826274e-07, "loss": 1.1825, "step": 4116 }, { "epoch": 0.87, "learning_rate": 9.271087217973807e-07, "loss": 1.1529, "step": 4117 }, { "epoch": 0.87, "learning_rate": 9.242458032904311e-07, "loss": 1.1935, "step": 4118 }, { "epoch": 0.87, "learning_rate": 9.213870977907569e-07, "loss": 1.1995, "step": 4119 }, { "epoch": 0.87, "learning_rate": 9.185326066253853e-07, "loss": 1.1617, "step": 4120 }, { "epoch": 0.87, "learning_rate": 9.156823311193819e-07, "loss": 1.2363, "step": 4121 }, { "epoch": 0.87, "learning_rate": 9.128362725958617e-07, "loss": 1.2144, "step": 4122 }, { "epoch": 0.87, "learning_rate": 9.099944323759768e-07, "loss": 1.162, "step": 4123 }, { "epoch": 0.87, "learning_rate": 9.071568117789187e-07, "loss": 1.1757, "step": 4124 }, { "epoch": 0.87, "learning_rate": 9.043234121219291e-07, "loss": 1.1529, "step": 4125 }, { "epoch": 0.87, "learning_rate": 9.014942347202816e-07, "loss": 1.272, "step": 4126 }, { "epoch": 0.87, "learning_rate": 8.986692808872976e-07, "loss": 1.1423, "step": 4127 }, { "epoch": 0.87, "learning_rate": 8.958485519343307e-07, "loss": 1.1533, "step": 4128 }, { "epoch": 0.87, "learning_rate": 8.930320491707766e-07, "loss": 1.2178, "step": 4129 }, { "epoch": 0.87, "learning_rate": 8.902197739040708e-07, "loss": 1.2008, "step": 4130 }, { "epoch": 0.87, "learning_rate": 8.87411727439681e-07, "loss": 1.1672, "step": 4131 }, { "epoch": 0.87, "learning_rate": 8.846079110811189e-07, "loss": 1.1401, "step": 4132 }, { "epoch": 0.87, "learning_rate": 8.81808326129927e-07, "loss": 1.1852, "step": 4133 }, { "epoch": 0.87, "learning_rate": 8.79012973885689e-07, "loss": 1.1892, "step": 4134 }, { "epoch": 0.87, "learning_rate": 8.762218556460145e-07, "loss": 1.165, "step": 4135 }, { "epoch": 0.87, "learning_rate": 8.734349727065606e-07, "loss": 1.1881, "step": 4136 }, { "epoch": 0.87, "learning_rate": 8.706523263610045e-07, "loss": 1.1729, "step": 4137 }, { "epoch": 0.87, "learning_rate": 8.678739179010687e-07, "loss": 1.1997, "step": 4138 }, { "epoch": 0.87, "learning_rate": 8.650997486165014e-07, "loss": 1.1865, "step": 4139 }, { "epoch": 0.87, "learning_rate": 8.623298197950835e-07, "loss": 1.1725, "step": 4140 }, { "epoch": 0.87, "learning_rate": 8.595641327226323e-07, "loss": 1.1425, "step": 4141 }, { "epoch": 0.87, "learning_rate": 8.568026886829883e-07, "loss": 1.1762, "step": 4142 }, { "epoch": 0.87, "learning_rate": 8.540454889580307e-07, "loss": 1.2378, "step": 4143 }, { "epoch": 0.87, "learning_rate": 8.512925348276624e-07, "loss": 1.2013, "step": 4144 }, { "epoch": 0.87, "learning_rate": 8.485438275698154e-07, "loss": 1.2131, "step": 4145 }, { "epoch": 0.87, "learning_rate": 8.457993684604571e-07, "loss": 1.1839, "step": 4146 }, { "epoch": 0.87, "learning_rate": 8.430591587735748e-07, "loss": 1.2354, "step": 4147 }, { "epoch": 0.87, "learning_rate": 8.403231997811867e-07, "loss": 1.2172, "step": 4148 }, { "epoch": 0.87, "learning_rate": 8.375914927533402e-07, "loss": 1.1618, "step": 4149 }, { "epoch": 0.87, "learning_rate": 8.34864038958103e-07, "loss": 1.1751, "step": 4150 }, { "epoch": 0.87, "learning_rate": 8.321408396615749e-07, "loss": 1.1766, "step": 4151 }, { "epoch": 0.87, "learning_rate": 8.294218961278766e-07, "loss": 1.1712, "step": 4152 }, { "epoch": 0.87, "learning_rate": 8.267072096191519e-07, "loss": 1.2005, "step": 4153 }, { "epoch": 0.87, "learning_rate": 8.239967813955741e-07, "loss": 1.1957, "step": 4154 }, { "epoch": 0.87, "learning_rate": 8.212906127153353e-07, "loss": 1.193, "step": 4155 }, { "epoch": 0.87, "learning_rate": 8.18588704834652e-07, "loss": 1.1534, "step": 4156 }, { "epoch": 0.87, "learning_rate": 8.158910590077607e-07, "loss": 1.1982, "step": 4157 }, { "epoch": 0.87, "learning_rate": 8.131976764869232e-07, "loss": 1.2308, "step": 4158 }, { "epoch": 0.87, "learning_rate": 8.105085585224215e-07, "loss": 1.1898, "step": 4159 }, { "epoch": 0.88, "learning_rate": 8.078237063625539e-07, "loss": 1.2135, "step": 4160 }, { "epoch": 0.88, "learning_rate": 8.051431212536409e-07, "loss": 1.2004, "step": 4161 }, { "epoch": 0.88, "learning_rate": 8.024668044400253e-07, "loss": 1.1784, "step": 4162 }, { "epoch": 0.88, "learning_rate": 7.997947571640618e-07, "loss": 1.1915, "step": 4163 }, { "epoch": 0.88, "learning_rate": 7.971269806661319e-07, "loss": 1.1505, "step": 4164 }, { "epoch": 0.88, "learning_rate": 7.944634761846271e-07, "loss": 1.1697, "step": 4165 }, { "epoch": 0.88, "learning_rate": 7.918042449559582e-07, "loss": 1.1594, "step": 4166 }, { "epoch": 0.88, "learning_rate": 7.891492882145546e-07, "loss": 1.1817, "step": 4167 }, { "epoch": 0.88, "learning_rate": 7.864986071928582e-07, "loss": 1.194, "step": 4168 }, { "epoch": 0.88, "learning_rate": 7.83852203121327e-07, "loss": 1.1432, "step": 4169 }, { "epoch": 0.88, "learning_rate": 7.812100772284358e-07, "loss": 1.2137, "step": 4170 }, { "epoch": 0.88, "learning_rate": 7.785722307406685e-07, "loss": 1.1719, "step": 4171 }, { "epoch": 0.88, "learning_rate": 7.759386648825306e-07, "loss": 1.18, "step": 4172 }, { "epoch": 0.88, "learning_rate": 7.733093808765324e-07, "loss": 1.2036, "step": 4173 }, { "epoch": 0.88, "learning_rate": 7.706843799431985e-07, "loss": 1.0885, "step": 4174 }, { "epoch": 0.88, "learning_rate": 7.680636633010696e-07, "loss": 1.1889, "step": 4175 }, { "epoch": 0.88, "learning_rate": 7.654472321666928e-07, "loss": 1.2232, "step": 4176 }, { "epoch": 0.88, "learning_rate": 7.628350877546287e-07, "loss": 1.1961, "step": 4177 }, { "epoch": 0.88, "learning_rate": 7.602272312774461e-07, "loss": 1.2214, "step": 4178 }, { "epoch": 0.88, "learning_rate": 7.576236639457224e-07, "loss": 1.1932, "step": 4179 }, { "epoch": 0.88, "learning_rate": 7.550243869680496e-07, "loss": 1.2248, "step": 4180 }, { "epoch": 0.88, "learning_rate": 7.524294015510203e-07, "loss": 1.1875, "step": 4181 }, { "epoch": 0.88, "learning_rate": 7.498387088992409e-07, "loss": 1.1802, "step": 4182 }, { "epoch": 0.88, "learning_rate": 7.47252310215324e-07, "loss": 1.2129, "step": 4183 }, { "epoch": 0.88, "learning_rate": 7.446702066998845e-07, "loss": 1.1944, "step": 4184 }, { "epoch": 0.88, "learning_rate": 7.420923995515505e-07, "loss": 1.1712, "step": 4185 }, { "epoch": 0.88, "learning_rate": 7.395188899669504e-07, "loss": 1.2224, "step": 4186 }, { "epoch": 0.88, "learning_rate": 7.369496791407172e-07, "loss": 1.1869, "step": 4187 }, { "epoch": 0.88, "learning_rate": 7.343847682654948e-07, "loss": 1.2278, "step": 4188 }, { "epoch": 0.88, "learning_rate": 7.318241585319252e-07, "loss": 1.1768, "step": 4189 }, { "epoch": 0.88, "learning_rate": 7.292678511286521e-07, "loss": 1.236, "step": 4190 }, { "epoch": 0.88, "learning_rate": 7.267158472423297e-07, "loss": 1.2211, "step": 4191 }, { "epoch": 0.88, "learning_rate": 7.241681480576057e-07, "loss": 1.1488, "step": 4192 }, { "epoch": 0.88, "learning_rate": 7.216247547571398e-07, "loss": 1.1833, "step": 4193 }, { "epoch": 0.88, "learning_rate": 7.19085668521582e-07, "loss": 1.2162, "step": 4194 }, { "epoch": 0.88, "learning_rate": 7.165508905295882e-07, "loss": 1.2154, "step": 4195 }, { "epoch": 0.88, "learning_rate": 7.140204219578184e-07, "loss": 1.1821, "step": 4196 }, { "epoch": 0.88, "learning_rate": 7.114942639809231e-07, "loss": 1.1568, "step": 4197 }, { "epoch": 0.88, "learning_rate": 7.089724177715584e-07, "loss": 1.2174, "step": 4198 }, { "epoch": 0.88, "learning_rate": 7.064548845003772e-07, "loss": 1.2099, "step": 4199 }, { "epoch": 0.88, "learning_rate": 7.039416653360287e-07, "loss": 1.1703, "step": 4200 }, { "epoch": 0.88, "learning_rate": 7.014327614451643e-07, "loss": 1.1897, "step": 4201 }, { "epoch": 0.88, "learning_rate": 6.989281739924269e-07, "loss": 1.2042, "step": 4202 }, { "epoch": 0.88, "learning_rate": 6.964279041404553e-07, "loss": 1.1722, "step": 4203 }, { "epoch": 0.88, "learning_rate": 6.939319530498911e-07, "loss": 1.2145, "step": 4204 }, { "epoch": 0.88, "learning_rate": 6.914403218793608e-07, "loss": 1.2711, "step": 4205 }, { "epoch": 0.88, "learning_rate": 6.889530117854959e-07, "loss": 1.2113, "step": 4206 }, { "epoch": 0.88, "learning_rate": 6.864700239229161e-07, "loss": 1.1632, "step": 4207 }, { "epoch": 0.89, "learning_rate": 6.839913594442338e-07, "loss": 1.2141, "step": 4208 }, { "epoch": 0.89, "learning_rate": 6.815170195000597e-07, "loss": 1.1035, "step": 4209 }, { "epoch": 0.89, "learning_rate": 6.79047005238993e-07, "loss": 1.2049, "step": 4210 }, { "epoch": 0.89, "learning_rate": 6.76581317807623e-07, "loss": 1.1912, "step": 4211 }, { "epoch": 0.89, "learning_rate": 6.741199583505376e-07, "loss": 1.2191, "step": 4212 }, { "epoch": 0.89, "learning_rate": 6.716629280103082e-07, "loss": 1.1969, "step": 4213 }, { "epoch": 0.89, "learning_rate": 6.692102279275014e-07, "loss": 1.1934, "step": 4214 }, { "epoch": 0.89, "learning_rate": 6.66761859240671e-07, "loss": 1.1344, "step": 4215 }, { "epoch": 0.89, "learning_rate": 6.643178230863601e-07, "loss": 1.1555, "step": 4216 }, { "epoch": 0.89, "learning_rate": 6.618781205991043e-07, "loss": 1.2119, "step": 4217 }, { "epoch": 0.89, "learning_rate": 6.594427529114234e-07, "loss": 1.1761, "step": 4218 }, { "epoch": 0.89, "learning_rate": 6.570117211538252e-07, "loss": 1.1794, "step": 4219 }, { "epoch": 0.89, "learning_rate": 6.545850264548093e-07, "loss": 1.2367, "step": 4220 }, { "epoch": 0.89, "learning_rate": 6.521626699408545e-07, "loss": 1.2114, "step": 4221 }, { "epoch": 0.89, "learning_rate": 6.497446527364327e-07, "loss": 1.2483, "step": 4222 }, { "epoch": 0.89, "learning_rate": 6.473309759639989e-07, "loss": 1.2648, "step": 4223 }, { "epoch": 0.89, "learning_rate": 6.449216407439906e-07, "loss": 1.1697, "step": 4224 }, { "epoch": 0.89, "learning_rate": 6.425166481948353e-07, "loss": 1.2364, "step": 4225 }, { "epoch": 0.89, "learning_rate": 6.401159994329409e-07, "loss": 1.2212, "step": 4226 }, { "epoch": 0.89, "learning_rate": 6.377196955726972e-07, "loss": 1.1803, "step": 4227 }, { "epoch": 0.89, "learning_rate": 6.353277377264832e-07, "loss": 1.2002, "step": 4228 }, { "epoch": 0.89, "learning_rate": 6.329401270046542e-07, "loss": 1.2294, "step": 4229 }, { "epoch": 0.89, "learning_rate": 6.305568645155502e-07, "loss": 1.1865, "step": 4230 }, { "epoch": 0.89, "learning_rate": 6.281779513654962e-07, "loss": 1.1615, "step": 4231 }, { "epoch": 0.89, "learning_rate": 6.258033886587911e-07, "loss": 1.2259, "step": 4232 }, { "epoch": 0.89, "learning_rate": 6.234331774977198e-07, "loss": 1.186, "step": 4233 }, { "epoch": 0.89, "learning_rate": 6.210673189825434e-07, "loss": 1.1805, "step": 4234 }, { "epoch": 0.89, "learning_rate": 6.187058142115076e-07, "loss": 1.2045, "step": 4235 }, { "epoch": 0.89, "learning_rate": 6.163486642808314e-07, "loss": 1.2345, "step": 4236 }, { "epoch": 0.89, "learning_rate": 6.139958702847149e-07, "loss": 1.1986, "step": 4237 }, { "epoch": 0.89, "learning_rate": 6.116474333153366e-07, "loss": 1.1751, "step": 4238 }, { "epoch": 0.89, "learning_rate": 6.093033544628535e-07, "loss": 1.228, "step": 4239 }, { "epoch": 0.89, "learning_rate": 6.069636348153929e-07, "loss": 1.2146, "step": 4240 }, { "epoch": 0.89, "learning_rate": 6.046282754590693e-07, "loss": 1.1851, "step": 4241 }, { "epoch": 0.89, "learning_rate": 6.022972774779634e-07, "loss": 1.211, "step": 4242 }, { "epoch": 0.89, "learning_rate": 5.999706419541374e-07, "loss": 1.1422, "step": 4243 }, { "epoch": 0.89, "learning_rate": 5.976483699676261e-07, "loss": 1.1983, "step": 4244 }, { "epoch": 0.89, "learning_rate": 5.953304625964373e-07, "loss": 1.1586, "step": 4245 }, { "epoch": 0.89, "learning_rate": 5.93016920916556e-07, "loss": 1.2048, "step": 4246 }, { "epoch": 0.89, "learning_rate": 5.907077460019394e-07, "loss": 1.2428, "step": 4247 }, { "epoch": 0.89, "learning_rate": 5.88402938924515e-07, "loss": 1.1557, "step": 4248 }, { "epoch": 0.89, "learning_rate": 5.861025007541876e-07, "loss": 1.2349, "step": 4249 }, { "epoch": 0.89, "learning_rate": 5.838064325588289e-07, "loss": 1.1516, "step": 4250 }, { "epoch": 0.89, "learning_rate": 5.81514735404286e-07, "loss": 1.1716, "step": 4251 }, { "epoch": 0.89, "learning_rate": 5.792274103543749e-07, "loss": 1.1798, "step": 4252 }, { "epoch": 0.89, "learning_rate": 5.769444584708827e-07, "loss": 1.1828, "step": 4253 }, { "epoch": 0.89, "learning_rate": 5.746658808135663e-07, "loss": 1.1956, "step": 4254 }, { "epoch": 0.89, "learning_rate": 5.723916784401518e-07, "loss": 1.2163, "step": 4255 }, { "epoch": 0.9, "learning_rate": 5.701218524063335e-07, "loss": 1.1594, "step": 4256 }, { "epoch": 0.9, "learning_rate": 5.678564037657786e-07, "loss": 1.1896, "step": 4257 }, { "epoch": 0.9, "learning_rate": 5.655953335701137e-07, "loss": 1.2155, "step": 4258 }, { "epoch": 0.9, "learning_rate": 5.633386428689435e-07, "loss": 1.1531, "step": 4259 }, { "epoch": 0.9, "learning_rate": 5.610863327098315e-07, "loss": 1.2123, "step": 4260 }, { "epoch": 0.9, "learning_rate": 5.588384041383089e-07, "loss": 1.165, "step": 4261 }, { "epoch": 0.9, "learning_rate": 5.565948581978786e-07, "loss": 1.124, "step": 4262 }, { "epoch": 0.9, "learning_rate": 5.543556959300011e-07, "loss": 1.162, "step": 4263 }, { "epoch": 0.9, "learning_rate": 5.521209183741094e-07, "loss": 1.2594, "step": 4264 }, { "epoch": 0.9, "learning_rate": 5.498905265675958e-07, "loss": 1.2058, "step": 4265 }, { "epoch": 0.9, "learning_rate": 5.476645215458177e-07, "loss": 1.188, "step": 4266 }, { "epoch": 0.9, "learning_rate": 5.454429043420994e-07, "loss": 1.2028, "step": 4267 }, { "epoch": 0.9, "learning_rate": 5.432256759877242e-07, "loss": 1.1766, "step": 4268 }, { "epoch": 0.9, "learning_rate": 5.410128375119395e-07, "loss": 1.1951, "step": 4269 }, { "epoch": 0.9, "learning_rate": 5.38804389941956e-07, "loss": 1.2075, "step": 4270 }, { "epoch": 0.9, "learning_rate": 5.366003343029446e-07, "loss": 1.2007, "step": 4271 }, { "epoch": 0.9, "learning_rate": 5.344006716180406e-07, "loss": 1.2066, "step": 4272 }, { "epoch": 0.9, "learning_rate": 5.322054029083346e-07, "loss": 1.205, "step": 4273 }, { "epoch": 0.9, "learning_rate": 5.300145291928816e-07, "loss": 1.2, "step": 4274 }, { "epoch": 0.9, "learning_rate": 5.278280514886957e-07, "loss": 1.1501, "step": 4275 }, { "epoch": 0.9, "learning_rate": 5.256459708107498e-07, "loss": 1.1901, "step": 4276 }, { "epoch": 0.9, "learning_rate": 5.234682881719766e-07, "loss": 1.1709, "step": 4277 }, { "epoch": 0.9, "learning_rate": 5.212950045832643e-07, "loss": 1.2556, "step": 4278 }, { "epoch": 0.9, "learning_rate": 5.191261210534637e-07, "loss": 1.2111, "step": 4279 }, { "epoch": 0.9, "learning_rate": 5.169616385893795e-07, "loss": 1.1411, "step": 4280 }, { "epoch": 0.9, "learning_rate": 5.148015581957755e-07, "loss": 1.1999, "step": 4281 }, { "epoch": 0.9, "learning_rate": 5.126458808753687e-07, "loss": 1.219, "step": 4282 }, { "epoch": 0.9, "learning_rate": 5.104946076288375e-07, "loss": 1.209, "step": 4283 }, { "epoch": 0.9, "learning_rate": 5.083477394548097e-07, "loss": 1.1616, "step": 4284 }, { "epoch": 0.9, "learning_rate": 5.062052773498761e-07, "loss": 1.1961, "step": 4285 }, { "epoch": 0.9, "learning_rate": 5.04067222308573e-07, "loss": 1.1954, "step": 4286 }, { "epoch": 0.9, "learning_rate": 5.019335753233967e-07, "loss": 1.1835, "step": 4287 }, { "epoch": 0.9, "learning_rate": 4.998043373847983e-07, "loss": 1.1867, "step": 4288 }, { "epoch": 0.9, "learning_rate": 4.976795094811781e-07, "loss": 1.1907, "step": 4289 }, { "epoch": 0.9, "learning_rate": 4.955590925988896e-07, "loss": 1.1596, "step": 4290 }, { "epoch": 0.9, "learning_rate": 4.934430877222429e-07, "loss": 1.1642, "step": 4291 }, { "epoch": 0.9, "learning_rate": 4.913314958334958e-07, "loss": 1.1908, "step": 4292 }, { "epoch": 0.9, "learning_rate": 4.892243179128597e-07, "loss": 1.1929, "step": 4293 }, { "epoch": 0.9, "learning_rate": 4.871215549384966e-07, "loss": 1.2491, "step": 4294 }, { "epoch": 0.9, "learning_rate": 4.850232078865169e-07, "loss": 1.1501, "step": 4295 }, { "epoch": 0.9, "learning_rate": 4.829292777309869e-07, "loss": 1.2153, "step": 4296 }, { "epoch": 0.9, "learning_rate": 4.808397654439156e-07, "loss": 1.1866, "step": 4297 }, { "epoch": 0.9, "learning_rate": 4.787546719952629e-07, "loss": 1.1913, "step": 4298 }, { "epoch": 0.9, "learning_rate": 4.7667399835294337e-07, "loss": 1.2039, "step": 4299 }, { "epoch": 0.9, "learning_rate": 4.745977454828121e-07, "loss": 1.1517, "step": 4300 }, { "epoch": 0.9, "learning_rate": 4.725259143486771e-07, "loss": 1.2409, "step": 4301 }, { "epoch": 0.9, "learning_rate": 4.7045850591229013e-07, "loss": 1.1992, "step": 4302 }, { "epoch": 0.91, "learning_rate": 4.683955211333535e-07, "loss": 1.201, "step": 4303 }, { "epoch": 0.91, "learning_rate": 4.663369609695123e-07, "loss": 1.1866, "step": 4304 }, { "epoch": 0.91, "learning_rate": 4.642828263763632e-07, "loss": 1.1894, "step": 4305 }, { "epoch": 0.91, "learning_rate": 4.622331183074413e-07, "loss": 1.2135, "step": 4306 }, { "epoch": 0.91, "learning_rate": 4.6018783771423325e-07, "loss": 1.222, "step": 4307 }, { "epoch": 0.91, "learning_rate": 4.581469855461662e-07, "loss": 1.196, "step": 4308 }, { "epoch": 0.91, "learning_rate": 4.561105627506146e-07, "loss": 1.1435, "step": 4309 }, { "epoch": 0.91, "learning_rate": 4.540785702728956e-07, "loss": 1.1532, "step": 4310 }, { "epoch": 0.91, "learning_rate": 4.5205100905626573e-07, "loss": 1.2194, "step": 4311 }, { "epoch": 0.91, "learning_rate": 4.500278800419333e-07, "loss": 1.2106, "step": 4312 }, { "epoch": 0.91, "learning_rate": 4.4800918416904036e-07, "loss": 1.2001, "step": 4313 }, { "epoch": 0.91, "learning_rate": 4.4599492237467846e-07, "loss": 1.1899, "step": 4314 }, { "epoch": 0.91, "learning_rate": 4.4398509559387517e-07, "loss": 1.2249, "step": 4315 }, { "epoch": 0.91, "learning_rate": 4.4197970475959975e-07, "loss": 1.149, "step": 4316 }, { "epoch": 0.91, "learning_rate": 4.3997875080276866e-07, "loss": 1.1904, "step": 4317 }, { "epoch": 0.91, "learning_rate": 4.3798223465223e-07, "loss": 1.1745, "step": 4318 }, { "epoch": 0.91, "learning_rate": 4.359901572347758e-07, "loss": 1.1924, "step": 4319 }, { "epoch": 0.91, "learning_rate": 4.340025194751407e-07, "loss": 1.2341, "step": 4320 }, { "epoch": 0.91, "learning_rate": 4.3201932229599233e-07, "loss": 1.1997, "step": 4321 }, { "epoch": 0.91, "learning_rate": 4.3004056661794303e-07, "loss": 1.1642, "step": 4322 }, { "epoch": 0.91, "learning_rate": 4.280662533595381e-07, "loss": 1.1706, "step": 4323 }, { "epoch": 0.91, "learning_rate": 4.260963834372622e-07, "loss": 1.1332, "step": 4324 }, { "epoch": 0.91, "learning_rate": 4.241309577655406e-07, "loss": 1.1846, "step": 4325 }, { "epoch": 0.91, "learning_rate": 4.2216997725673113e-07, "loss": 1.1127, "step": 4326 }, { "epoch": 0.91, "learning_rate": 4.202134428211313e-07, "loss": 1.1716, "step": 4327 }, { "epoch": 0.91, "learning_rate": 4.1826135536697235e-07, "loss": 1.1375, "step": 4328 }, { "epoch": 0.91, "learning_rate": 4.1631371580042377e-07, "loss": 1.212, "step": 4329 }, { "epoch": 0.91, "learning_rate": 4.1437052502558693e-07, "loss": 1.1416, "step": 4330 }, { "epoch": 0.91, "learning_rate": 4.1243178394450246e-07, "loss": 1.1719, "step": 4331 }, { "epoch": 0.91, "learning_rate": 4.104974934571404e-07, "loss": 1.1838, "step": 4332 }, { "epoch": 0.91, "learning_rate": 4.085676544614103e-07, "loss": 1.1517, "step": 4333 }, { "epoch": 0.91, "learning_rate": 4.0664226785314896e-07, "loss": 1.1752, "step": 4334 }, { "epoch": 0.91, "learning_rate": 4.047213345261336e-07, "loss": 1.2252, "step": 4335 }, { "epoch": 0.91, "learning_rate": 4.028048553720676e-07, "loss": 1.2244, "step": 4336 }, { "epoch": 0.91, "learning_rate": 4.0089283128059045e-07, "loss": 1.167, "step": 4337 }, { "epoch": 0.91, "learning_rate": 3.989852631392721e-07, "loss": 1.1609, "step": 4338 }, { "epoch": 0.91, "learning_rate": 3.9708215183361544e-07, "loss": 1.185, "step": 4339 }, { "epoch": 0.91, "learning_rate": 3.951834982470526e-07, "loss": 1.187, "step": 4340 }, { "epoch": 0.91, "learning_rate": 3.932893032609486e-07, "loss": 1.2098, "step": 4341 }, { "epoch": 0.91, "learning_rate": 3.9139956775459454e-07, "loss": 1.1939, "step": 4342 }, { "epoch": 0.91, "learning_rate": 3.8951429260521865e-07, "loss": 1.208, "step": 4343 }, { "epoch": 0.91, "learning_rate": 3.876334786879732e-07, "loss": 1.2207, "step": 4344 }, { "epoch": 0.91, "learning_rate": 3.8575712687593857e-07, "loss": 1.1352, "step": 4345 }, { "epoch": 0.91, "learning_rate": 3.838852380401281e-07, "loss": 1.1641, "step": 4346 }, { "epoch": 0.91, "learning_rate": 3.8201781304948005e-07, "loss": 1.1797, "step": 4347 }, { "epoch": 0.91, "learning_rate": 3.801548527708621e-07, "loss": 1.1772, "step": 4348 }, { "epoch": 0.91, "learning_rate": 3.7829635806907015e-07, "loss": 1.2148, "step": 4349 }, { "epoch": 0.91, "learning_rate": 3.7644232980682427e-07, "loss": 1.1916, "step": 4350 }, { "epoch": 0.92, "learning_rate": 3.745927688447759e-07, "loss": 1.2283, "step": 4351 }, { "epoch": 0.92, "learning_rate": 3.7274767604149497e-07, "loss": 1.1948, "step": 4352 }, { "epoch": 0.92, "learning_rate": 3.7090705225348634e-07, "loss": 1.2056, "step": 4353 }, { "epoch": 0.92, "learning_rate": 3.690708983351754e-07, "loss": 1.2004, "step": 4354 }, { "epoch": 0.92, "learning_rate": 3.672392151389137e-07, "loss": 1.227, "step": 4355 }, { "epoch": 0.92, "learning_rate": 3.6541200351497443e-07, "loss": 1.181, "step": 4356 }, { "epoch": 0.92, "learning_rate": 3.635892643115602e-07, "loss": 1.1955, "step": 4357 }, { "epoch": 0.92, "learning_rate": 3.6177099837479414e-07, "loss": 1.1876, "step": 4358 }, { "epoch": 0.92, "learning_rate": 3.5995720654872444e-07, "loss": 1.222, "step": 4359 }, { "epoch": 0.92, "learning_rate": 3.58147889675321e-07, "loss": 1.2125, "step": 4360 }, { "epoch": 0.92, "learning_rate": 3.5634304859447633e-07, "loss": 1.197, "step": 4361 }, { "epoch": 0.92, "learning_rate": 3.5454268414400803e-07, "loss": 1.2042, "step": 4362 }, { "epoch": 0.92, "learning_rate": 3.527467971596521e-07, "loss": 1.246, "step": 4363 }, { "epoch": 0.92, "learning_rate": 3.509553884750683e-07, "loss": 1.2277, "step": 4364 }, { "epoch": 0.92, "learning_rate": 3.4916845892183693e-07, "loss": 1.2676, "step": 4365 }, { "epoch": 0.92, "learning_rate": 3.473860093294579e-07, "loss": 1.1952, "step": 4366 }, { "epoch": 0.92, "learning_rate": 3.456080405253548e-07, "loss": 1.1777, "step": 4367 }, { "epoch": 0.92, "learning_rate": 3.4383455333486746e-07, "loss": 1.2415, "step": 4368 }, { "epoch": 0.92, "learning_rate": 3.4206554858125517e-07, "loss": 1.2322, "step": 4369 }, { "epoch": 0.92, "learning_rate": 3.403010270857021e-07, "loss": 1.1925, "step": 4370 }, { "epoch": 0.92, "learning_rate": 3.385409896673042e-07, "loss": 1.183, "step": 4371 }, { "epoch": 0.92, "learning_rate": 3.367854371430812e-07, "loss": 1.192, "step": 4372 }, { "epoch": 0.92, "learning_rate": 3.3503437032796795e-07, "loss": 1.1691, "step": 4373 }, { "epoch": 0.92, "learning_rate": 3.3328779003481524e-07, "loss": 1.2083, "step": 4374 }, { "epoch": 0.92, "learning_rate": 3.31545697074398e-07, "loss": 1.1429, "step": 4375 }, { "epoch": 0.92, "learning_rate": 3.298080922554003e-07, "loss": 1.1743, "step": 4376 }, { "epoch": 0.92, "learning_rate": 3.280749763844293e-07, "loss": 1.1516, "step": 4377 }, { "epoch": 0.92, "learning_rate": 3.263463502660036e-07, "loss": 1.1472, "step": 4378 }, { "epoch": 0.92, "learning_rate": 3.246222147025602e-07, "loss": 1.2035, "step": 4379 }, { "epoch": 0.92, "learning_rate": 3.2290257049445105e-07, "loss": 1.1674, "step": 4380 }, { "epoch": 0.92, "learning_rate": 3.2118741843994217e-07, "loss": 1.1977, "step": 4381 }, { "epoch": 0.92, "learning_rate": 3.194767593352155e-07, "loss": 1.1642, "step": 4382 }, { "epoch": 0.92, "learning_rate": 3.1777059397436693e-07, "loss": 1.1991, "step": 4383 }, { "epoch": 0.92, "learning_rate": 3.160689231494074e-07, "loss": 1.1916, "step": 4384 }, { "epoch": 0.92, "learning_rate": 3.1437174765025725e-07, "loss": 1.2177, "step": 4385 }, { "epoch": 0.92, "learning_rate": 3.126790682647574e-07, "loss": 1.2191, "step": 4386 }, { "epoch": 0.92, "learning_rate": 3.1099088577865366e-07, "loss": 1.1692, "step": 4387 }, { "epoch": 0.92, "learning_rate": 3.093072009756115e-07, "loss": 1.2326, "step": 4388 }, { "epoch": 0.92, "learning_rate": 3.076280146372046e-07, "loss": 1.1845, "step": 4389 }, { "epoch": 0.92, "learning_rate": 3.0595332754291605e-07, "loss": 1.1977, "step": 4390 }, { "epoch": 0.92, "learning_rate": 3.042831404701463e-07, "loss": 1.2271, "step": 4391 }, { "epoch": 0.92, "learning_rate": 3.0261745419420287e-07, "loss": 1.1843, "step": 4392 }, { "epoch": 0.92, "learning_rate": 3.0095626948830616e-07, "loss": 1.1915, "step": 4393 }, { "epoch": 0.92, "learning_rate": 2.992995871235849e-07, "loss": 1.1961, "step": 4394 }, { "epoch": 0.92, "learning_rate": 2.976474078690783e-07, "loss": 1.1425, "step": 4395 }, { "epoch": 0.92, "learning_rate": 2.959997324917374e-07, "loss": 1.1978, "step": 4396 }, { "epoch": 0.92, "learning_rate": 2.9435656175641923e-07, "loss": 1.1986, "step": 4397 }, { "epoch": 0.93, "learning_rate": 2.927178964258903e-07, "loss": 1.1996, "step": 4398 }, { "epoch": 0.93, "learning_rate": 2.9108373726082997e-07, "loss": 1.2028, "step": 4399 }, { "epoch": 0.93, "learning_rate": 2.8945408501981907e-07, "loss": 1.1199, "step": 4400 }, { "epoch": 0.93, "learning_rate": 2.878289404593504e-07, "loss": 1.2014, "step": 4401 }, { "epoch": 0.93, "learning_rate": 2.862083043338271e-07, "loss": 1.1997, "step": 4402 }, { "epoch": 0.93, "learning_rate": 2.8459217739555067e-07, "loss": 1.214, "step": 4403 }, { "epoch": 0.93, "learning_rate": 2.8298056039474e-07, "loss": 1.1647, "step": 4404 }, { "epoch": 0.93, "learning_rate": 2.813734540795121e-07, "loss": 1.2053, "step": 4405 }, { "epoch": 0.93, "learning_rate": 2.7977085919589253e-07, "loss": 1.1985, "step": 4406 }, { "epoch": 0.93, "learning_rate": 2.7817277648781503e-07, "loss": 1.1653, "step": 4407 }, { "epoch": 0.93, "learning_rate": 2.7657920669711625e-07, "loss": 1.2132, "step": 4408 }, { "epoch": 0.93, "learning_rate": 2.749901505635388e-07, "loss": 1.2283, "step": 4409 }, { "epoch": 0.93, "learning_rate": 2.7340560882473164e-07, "loss": 1.22, "step": 4410 }, { "epoch": 0.93, "learning_rate": 2.71825582216243e-07, "loss": 1.2112, "step": 4411 }, { "epoch": 0.93, "learning_rate": 2.702500714715317e-07, "loss": 1.1587, "step": 4412 }, { "epoch": 0.93, "learning_rate": 2.6867907732195387e-07, "loss": 1.1735, "step": 4413 }, { "epoch": 0.93, "learning_rate": 2.671126004967761e-07, "loss": 1.1843, "step": 4414 }, { "epoch": 0.93, "learning_rate": 2.6555064172316236e-07, "loss": 1.1704, "step": 4415 }, { "epoch": 0.93, "learning_rate": 2.639932017261793e-07, "loss": 1.1978, "step": 4416 }, { "epoch": 0.93, "learning_rate": 2.6244028122880184e-07, "loss": 1.1949, "step": 4417 }, { "epoch": 0.93, "learning_rate": 2.608918809519001e-07, "loss": 1.2074, "step": 4418 }, { "epoch": 0.93, "learning_rate": 2.5934800161424913e-07, "loss": 1.1677, "step": 4419 }, { "epoch": 0.93, "learning_rate": 2.5780864393252446e-07, "loss": 1.1719, "step": 4420 }, { "epoch": 0.93, "learning_rate": 2.562738086213046e-07, "loss": 1.1745, "step": 4421 }, { "epoch": 0.93, "learning_rate": 2.5474349639306616e-07, "loss": 1.164, "step": 4422 }, { "epoch": 0.93, "learning_rate": 2.532177079581877e-07, "loss": 1.1666, "step": 4423 }, { "epoch": 0.93, "learning_rate": 2.5169644402494584e-07, "loss": 1.2549, "step": 4424 }, { "epoch": 0.93, "learning_rate": 2.501797052995192e-07, "loss": 1.1619, "step": 4425 }, { "epoch": 0.93, "learning_rate": 2.4866749248598666e-07, "loss": 1.1363, "step": 4426 }, { "epoch": 0.93, "learning_rate": 2.471598062863223e-07, "loss": 1.2035, "step": 4427 }, { "epoch": 0.93, "learning_rate": 2.4565664740040185e-07, "loss": 1.1488, "step": 4428 }, { "epoch": 0.93, "learning_rate": 2.4415801652599824e-07, "loss": 1.215, "step": 4429 }, { "epoch": 0.93, "learning_rate": 2.426639143587839e-07, "loss": 1.1794, "step": 4430 }, { "epoch": 0.93, "learning_rate": 2.4117434159232734e-07, "loss": 1.2041, "step": 4431 }, { "epoch": 0.93, "learning_rate": 2.3968929891809435e-07, "loss": 1.2171, "step": 4432 }, { "epoch": 0.93, "learning_rate": 2.3820878702545126e-07, "loss": 1.2038, "step": 4433 }, { "epoch": 0.93, "learning_rate": 2.3673280660165498e-07, "loss": 1.1862, "step": 4434 }, { "epoch": 0.93, "learning_rate": 2.3526135833186527e-07, "loss": 1.1605, "step": 4435 }, { "epoch": 0.93, "learning_rate": 2.3379444289913344e-07, "loss": 1.2001, "step": 4436 }, { "epoch": 0.93, "learning_rate": 2.323320609844093e-07, "loss": 1.2252, "step": 4437 }, { "epoch": 0.93, "learning_rate": 2.308742132665376e-07, "loss": 1.1836, "step": 4438 }, { "epoch": 0.93, "learning_rate": 2.2942090042225806e-07, "loss": 1.2106, "step": 4439 }, { "epoch": 0.93, "learning_rate": 2.2797212312620444e-07, "loss": 1.192, "step": 4440 }, { "epoch": 0.93, "learning_rate": 2.2652788205090647e-07, "loss": 1.1893, "step": 4441 }, { "epoch": 0.93, "learning_rate": 2.2508817786678683e-07, "loss": 1.2241, "step": 4442 }, { "epoch": 0.93, "learning_rate": 2.2365301124216422e-07, "loss": 1.173, "step": 4443 }, { "epoch": 0.93, "learning_rate": 2.2222238284324904e-07, "loss": 1.2129, "step": 4444 }, { "epoch": 0.93, "learning_rate": 2.2079629333414454e-07, "loss": 1.2008, "step": 4445 }, { "epoch": 0.94, "learning_rate": 2.1937474337685116e-07, "loss": 1.2002, "step": 4446 }, { "epoch": 0.94, "learning_rate": 2.1795773363125438e-07, "loss": 1.2026, "step": 4447 }, { "epoch": 0.94, "learning_rate": 2.1654526475514138e-07, "loss": 1.1824, "step": 4448 }, { "epoch": 0.94, "learning_rate": 2.151373374041832e-07, "loss": 1.1315, "step": 4449 }, { "epoch": 0.94, "learning_rate": 2.1373395223194926e-07, "loss": 1.1951, "step": 4450 }, { "epoch": 0.94, "learning_rate": 2.1233510988989403e-07, "loss": 1.1502, "step": 4451 }, { "epoch": 0.94, "learning_rate": 2.109408110273703e-07, "loss": 1.1791, "step": 4452 }, { "epoch": 0.94, "learning_rate": 2.0955105629161477e-07, "loss": 1.1786, "step": 4453 }, { "epoch": 0.94, "learning_rate": 2.0816584632776137e-07, "loss": 1.1789, "step": 4454 }, { "epoch": 0.94, "learning_rate": 2.06785181778828e-07, "loss": 1.1984, "step": 4455 }, { "epoch": 0.94, "learning_rate": 2.054090632857264e-07, "loss": 1.2041, "step": 4456 }, { "epoch": 0.94, "learning_rate": 2.0403749148725893e-07, "loss": 1.1668, "step": 4457 }, { "epoch": 0.94, "learning_rate": 2.0267046702011296e-07, "loss": 1.1438, "step": 4458 }, { "epoch": 0.94, "learning_rate": 2.0130799051887085e-07, "loss": 1.1537, "step": 4459 }, { "epoch": 0.94, "learning_rate": 1.999500626159967e-07, "loss": 1.1744, "step": 4460 }, { "epoch": 0.94, "learning_rate": 1.985966839418496e-07, "loss": 1.2459, "step": 4461 }, { "epoch": 0.94, "learning_rate": 1.972478551246737e-07, "loss": 1.1753, "step": 4462 }, { "epoch": 0.94, "learning_rate": 1.9590357679060036e-07, "loss": 1.1879, "step": 4463 }, { "epoch": 0.94, "learning_rate": 1.9456384956365149e-07, "loss": 1.2132, "step": 4464 }, { "epoch": 0.94, "learning_rate": 1.932286740657341e-07, "loss": 1.1865, "step": 4465 }, { "epoch": 0.94, "learning_rate": 1.9189805091664125e-07, "loss": 1.2021, "step": 4466 }, { "epoch": 0.94, "learning_rate": 1.905719807340578e-07, "loss": 1.1983, "step": 4467 }, { "epoch": 0.94, "learning_rate": 1.8925046413354908e-07, "loss": 1.2146, "step": 4468 }, { "epoch": 0.94, "learning_rate": 1.8793350172856995e-07, "loss": 1.1827, "step": 4469 }, { "epoch": 0.94, "learning_rate": 1.8662109413046026e-07, "loss": 1.1592, "step": 4470 }, { "epoch": 0.94, "learning_rate": 1.8531324194844602e-07, "loss": 1.2237, "step": 4471 }, { "epoch": 0.94, "learning_rate": 1.8400994578963826e-07, "loss": 1.2455, "step": 4472 }, { "epoch": 0.94, "learning_rate": 1.8271120625903306e-07, "loss": 1.1478, "step": 4473 }, { "epoch": 0.94, "learning_rate": 1.8141702395951145e-07, "loss": 1.2118, "step": 4474 }, { "epoch": 0.94, "learning_rate": 1.8012739949183844e-07, "loss": 1.2526, "step": 4475 }, { "epoch": 0.94, "learning_rate": 1.7884233345466518e-07, "loss": 1.1727, "step": 4476 }, { "epoch": 0.94, "learning_rate": 1.7756182644452225e-07, "loss": 1.2218, "step": 4477 }, { "epoch": 0.94, "learning_rate": 1.7628587905583083e-07, "loss": 1.1937, "step": 4478 }, { "epoch": 0.94, "learning_rate": 1.7501449188088826e-07, "loss": 1.191, "step": 4479 }, { "epoch": 0.94, "learning_rate": 1.737476655098802e-07, "loss": 1.1767, "step": 4480 }, { "epoch": 0.94, "learning_rate": 1.7248540053087403e-07, "loss": 1.1923, "step": 4481 }, { "epoch": 0.94, "learning_rate": 1.712276975298166e-07, "loss": 1.2354, "step": 4482 }, { "epoch": 0.94, "learning_rate": 1.6997455709054088e-07, "loss": 1.175, "step": 4483 }, { "epoch": 0.94, "learning_rate": 1.6872597979476045e-07, "loss": 1.1807, "step": 4484 }, { "epoch": 0.94, "learning_rate": 1.6748196622207057e-07, "loss": 1.1912, "step": 4485 }, { "epoch": 0.94, "learning_rate": 1.662425169499482e-07, "loss": 1.1976, "step": 4486 }, { "epoch": 0.94, "learning_rate": 1.6500763255375195e-07, "loss": 1.1662, "step": 4487 }, { "epoch": 0.94, "learning_rate": 1.6377731360672e-07, "loss": 1.1536, "step": 4488 }, { "epoch": 0.94, "learning_rate": 1.6255156067997325e-07, "loss": 1.2202, "step": 4489 }, { "epoch": 0.94, "learning_rate": 1.6133037434250988e-07, "loss": 1.201, "step": 4490 }, { "epoch": 0.94, "learning_rate": 1.6011375516121198e-07, "loss": 1.1867, "step": 4491 }, { "epoch": 0.94, "learning_rate": 1.589017037008389e-07, "loss": 1.2107, "step": 4492 }, { "epoch": 0.95, "learning_rate": 1.5769422052403172e-07, "loss": 1.1894, "step": 4493 }, { "epoch": 0.95, "learning_rate": 1.5649130619130982e-07, "loss": 1.1917, "step": 4494 }, { "epoch": 0.95, "learning_rate": 1.5529296126106985e-07, "loss": 1.1679, "step": 4495 }, { "epoch": 0.95, "learning_rate": 1.5409918628959018e-07, "loss": 1.21, "step": 4496 }, { "epoch": 0.95, "learning_rate": 1.5290998183102645e-07, "loss": 1.1536, "step": 4497 }, { "epoch": 0.95, "learning_rate": 1.517253484374137e-07, "loss": 1.2062, "step": 4498 }, { "epoch": 0.95, "learning_rate": 1.505452866586632e-07, "loss": 1.2063, "step": 4499 }, { "epoch": 0.95, "learning_rate": 1.493697970425656e-07, "loss": 1.2096, "step": 4500 }, { "epoch": 0.95, "learning_rate": 1.4819888013478889e-07, "loss": 1.2305, "step": 4501 }, { "epoch": 0.95, "learning_rate": 1.4703253647887828e-07, "loss": 1.201, "step": 4502 }, { "epoch": 0.95, "learning_rate": 1.45870766616254e-07, "loss": 1.1818, "step": 4503 }, { "epoch": 0.95, "learning_rate": 1.4471357108621796e-07, "loss": 1.1495, "step": 4504 }, { "epoch": 0.95, "learning_rate": 1.4356095042594386e-07, "loss": 1.1922, "step": 4505 }, { "epoch": 0.95, "learning_rate": 1.4241290517048366e-07, "loss": 1.1937, "step": 4506 }, { "epoch": 0.95, "learning_rate": 1.4126943585276554e-07, "loss": 1.1724, "step": 4507 }, { "epoch": 0.95, "learning_rate": 1.4013054300359373e-07, "loss": 1.1856, "step": 4508 }, { "epoch": 0.95, "learning_rate": 1.3899622715164763e-07, "loss": 1.1819, "step": 4509 }, { "epoch": 0.95, "learning_rate": 1.3786648882348042e-07, "loss": 1.2089, "step": 4510 }, { "epoch": 0.95, "learning_rate": 1.3674132854352372e-07, "loss": 1.1511, "step": 4511 }, { "epoch": 0.95, "learning_rate": 1.3562074683408089e-07, "loss": 1.198, "step": 4512 }, { "epoch": 0.95, "learning_rate": 1.3450474421533245e-07, "loss": 1.1766, "step": 4513 }, { "epoch": 0.95, "learning_rate": 1.3339332120532956e-07, "loss": 1.2217, "step": 4514 }, { "epoch": 0.95, "learning_rate": 1.3228647832000286e-07, "loss": 1.2053, "step": 4515 }, { "epoch": 0.95, "learning_rate": 1.3118421607315245e-07, "loss": 1.1891, "step": 4516 }, { "epoch": 0.95, "learning_rate": 1.3008653497645463e-07, "loss": 1.2175, "step": 4517 }, { "epoch": 0.95, "learning_rate": 1.289934355394562e-07, "loss": 1.1982, "step": 4518 }, { "epoch": 0.95, "learning_rate": 1.2790491826958017e-07, "loss": 1.1744, "step": 4519 }, { "epoch": 0.95, "learning_rate": 1.2682098367212238e-07, "loss": 1.1833, "step": 4520 }, { "epoch": 0.95, "learning_rate": 1.2574163225024694e-07, "loss": 1.225, "step": 4521 }, { "epoch": 0.95, "learning_rate": 1.2466686450499866e-07, "loss": 1.1871, "step": 4522 }, { "epoch": 0.95, "learning_rate": 1.235966809352851e-07, "loss": 1.1937, "step": 4523 }, { "epoch": 0.95, "learning_rate": 1.2253108203789332e-07, "loss": 1.1745, "step": 4524 }, { "epoch": 0.95, "learning_rate": 1.2147006830747764e-07, "loss": 1.1747, "step": 4525 }, { "epoch": 0.95, "learning_rate": 1.2041364023656743e-07, "loss": 1.197, "step": 4526 }, { "epoch": 0.95, "learning_rate": 1.1936179831555928e-07, "loss": 1.1924, "step": 4527 }, { "epoch": 0.95, "learning_rate": 1.1831454303272482e-07, "loss": 1.1764, "step": 4528 }, { "epoch": 0.95, "learning_rate": 1.17271874874203e-07, "loss": 1.2488, "step": 4529 }, { "epoch": 0.95, "learning_rate": 1.1623379432400772e-07, "loss": 1.1827, "step": 4530 }, { "epoch": 0.95, "learning_rate": 1.152003018640202e-07, "loss": 1.2022, "step": 4531 }, { "epoch": 0.95, "learning_rate": 1.1417139797399001e-07, "loss": 1.1814, "step": 4532 }, { "epoch": 0.95, "learning_rate": 1.1314708313154177e-07, "loss": 1.1283, "step": 4533 }, { "epoch": 0.95, "learning_rate": 1.1212735781216733e-07, "loss": 1.1654, "step": 4534 }, { "epoch": 0.95, "learning_rate": 1.111122224892247e-07, "loss": 1.1442, "step": 4535 }, { "epoch": 0.95, "learning_rate": 1.1010167763394808e-07, "loss": 1.1938, "step": 4536 }, { "epoch": 0.95, "learning_rate": 1.0909572371543442e-07, "loss": 1.1808, "step": 4537 }, { "epoch": 0.95, "learning_rate": 1.0809436120065464e-07, "loss": 1.1971, "step": 4538 }, { "epoch": 0.95, "learning_rate": 1.0709759055444469e-07, "loss": 1.2019, "step": 4539 }, { "epoch": 0.95, "learning_rate": 1.0610541223950887e-07, "loss": 1.2026, "step": 4540 }, { "epoch": 0.96, "learning_rate": 1.0511782671642213e-07, "loss": 1.2094, "step": 4541 }, { "epoch": 0.96, "learning_rate": 1.041348344436277e-07, "loss": 1.16, "step": 4542 }, { "epoch": 0.96, "learning_rate": 1.0315643587743284e-07, "loss": 1.182, "step": 4543 }, { "epoch": 0.96, "learning_rate": 1.0218263147201535e-07, "loss": 1.2133, "step": 4544 }, { "epoch": 0.96, "learning_rate": 1.0121342167942028e-07, "loss": 1.2141, "step": 4545 }, { "epoch": 0.96, "learning_rate": 1.0024880694955886e-07, "loss": 1.2487, "step": 4546 }, { "epoch": 0.96, "learning_rate": 9.928878773020956e-08, "loss": 1.1818, "step": 4547 }, { "epoch": 0.96, "learning_rate": 9.833336446701814e-08, "loss": 1.1677, "step": 4548 }, { "epoch": 0.96, "learning_rate": 9.738253760349759e-08, "loss": 1.197, "step": 4549 }, { "epoch": 0.96, "learning_rate": 9.643630758102485e-08, "loss": 1.1955, "step": 4550 }, { "epoch": 0.96, "learning_rate": 9.549467483884412e-08, "loss": 1.2368, "step": 4551 }, { "epoch": 0.96, "learning_rate": 9.455763981406685e-08, "loss": 1.1799, "step": 4552 }, { "epoch": 0.96, "learning_rate": 9.362520294166733e-08, "loss": 1.1604, "step": 4553 }, { "epoch": 0.96, "learning_rate": 9.26973646544882e-08, "loss": 1.1793, "step": 4554 }, { "epoch": 0.96, "learning_rate": 9.177412538323716e-08, "loss": 1.163, "step": 4555 }, { "epoch": 0.96, "learning_rate": 9.08554855564836e-08, "loss": 1.1487, "step": 4556 }, { "epoch": 0.96, "learning_rate": 8.994144560066754e-08, "loss": 1.1879, "step": 4557 }, { "epoch": 0.96, "learning_rate": 8.903200594008732e-08, "loss": 1.1827, "step": 4558 }, { "epoch": 0.96, "learning_rate": 8.812716699691193e-08, "loss": 1.2039, "step": 4559 }, { "epoch": 0.96, "learning_rate": 8.722692919117093e-08, "loss": 1.1794, "step": 4560 }, { "epoch": 0.96, "learning_rate": 8.633129294075781e-08, "loss": 1.1401, "step": 4561 }, { "epoch": 0.96, "learning_rate": 8.544025866143224e-08, "loss": 1.16, "step": 4562 }, { "epoch": 0.96, "learning_rate": 8.455382676681667e-08, "loss": 1.2203, "step": 4563 }, { "epoch": 0.96, "learning_rate": 8.367199766839529e-08, "loss": 1.2053, "step": 4564 }, { "epoch": 0.96, "learning_rate": 8.279477177551843e-08, "loss": 1.184, "step": 4565 }, { "epoch": 0.96, "learning_rate": 8.192214949539812e-08, "loss": 1.1826, "step": 4566 }, { "epoch": 0.96, "learning_rate": 8.105413123311034e-08, "loss": 1.1476, "step": 4567 }, { "epoch": 0.96, "learning_rate": 8.01907173915928e-08, "loss": 1.1669, "step": 4568 }, { "epoch": 0.96, "learning_rate": 7.933190837164595e-08, "loss": 1.2033, "step": 4569 }, { "epoch": 0.96, "learning_rate": 7.847770457193316e-08, "loss": 1.1935, "step": 4570 }, { "epoch": 0.96, "learning_rate": 7.762810638898054e-08, "loss": 1.1534, "step": 4571 }, { "epoch": 0.96, "learning_rate": 7.678311421717267e-08, "loss": 1.1559, "step": 4572 }, { "epoch": 0.96, "learning_rate": 7.59427284487635e-08, "loss": 1.1683, "step": 4573 }, { "epoch": 0.96, "learning_rate": 7.510694947385876e-08, "loss": 1.1875, "step": 4574 }, { "epoch": 0.96, "learning_rate": 7.427577768043592e-08, "loss": 1.2023, "step": 4575 }, { "epoch": 0.96, "learning_rate": 7.344921345432521e-08, "loss": 1.1547, "step": 4576 }, { "epoch": 0.96, "learning_rate": 7.262725717922304e-08, "loss": 1.2013, "step": 4577 }, { "epoch": 0.96, "learning_rate": 7.18099092366853e-08, "loss": 1.1547, "step": 4578 }, { "epoch": 0.96, "learning_rate": 7.099717000612849e-08, "loss": 1.1746, "step": 4579 }, { "epoch": 0.96, "learning_rate": 7.018903986483083e-08, "loss": 1.248, "step": 4580 }, { "epoch": 0.96, "learning_rate": 6.93855191879278e-08, "loss": 1.1925, "step": 4581 }, { "epoch": 0.96, "learning_rate": 6.858660834841991e-08, "loss": 1.1956, "step": 4582 }, { "epoch": 0.96, "learning_rate": 6.779230771716384e-08, "loss": 1.1766, "step": 4583 }, { "epoch": 0.96, "learning_rate": 6.700261766287796e-08, "loss": 1.2142, "step": 4584 }, { "epoch": 0.96, "learning_rate": 6.62175385521402e-08, "loss": 1.1779, "step": 4585 }, { "epoch": 0.96, "learning_rate": 6.543707074938787e-08, "loss": 1.1874, "step": 4586 }, { "epoch": 0.96, "learning_rate": 6.466121461691787e-08, "loss": 1.2028, "step": 4587 }, { "epoch": 0.97, "learning_rate": 6.388997051488655e-08, "loss": 1.2415, "step": 4588 }, { "epoch": 0.97, "learning_rate": 6.31233388013075e-08, "loss": 1.1779, "step": 4589 }, { "epoch": 0.97, "learning_rate": 6.236131983205607e-08, "loss": 1.2063, "step": 4590 }, { "epoch": 0.97, "learning_rate": 6.160391396086596e-08, "loss": 1.1625, "step": 4591 }, { "epoch": 0.97, "learning_rate": 6.085112153932592e-08, "loss": 1.1981, "step": 4592 }, { "epoch": 0.97, "learning_rate": 6.010294291688756e-08, "loss": 1.1537, "step": 4593 }, { "epoch": 0.97, "learning_rate": 5.93593784408586e-08, "loss": 1.1748, "step": 4594 }, { "epoch": 0.97, "learning_rate": 5.862042845640403e-08, "loss": 1.1629, "step": 4595 }, { "epoch": 0.97, "learning_rate": 5.788609330654948e-08, "loss": 1.1757, "step": 4596 }, { "epoch": 0.97, "learning_rate": 5.7156373332175606e-08, "loss": 1.1105, "step": 4597 }, { "epoch": 0.97, "learning_rate": 5.643126887202144e-08, "loss": 1.1964, "step": 4598 }, { "epoch": 0.97, "learning_rate": 5.57107802626855e-08, "loss": 1.1862, "step": 4599 }, { "epoch": 0.97, "learning_rate": 5.499490783862138e-08, "loss": 1.2479, "step": 4600 }, { "epoch": 0.97, "learning_rate": 5.4283651932139916e-08, "loss": 1.1887, "step": 4601 }, { "epoch": 0.97, "learning_rate": 5.357701287340922e-08, "loss": 1.2016, "step": 4602 }, { "epoch": 0.97, "learning_rate": 5.2874990990454676e-08, "loss": 1.228, "step": 4603 }, { "epoch": 0.97, "learning_rate": 5.217758660915784e-08, "loss": 1.1616, "step": 4604 }, { "epoch": 0.97, "learning_rate": 5.148480005325862e-08, "loss": 1.1614, "step": 4605 }, { "epoch": 0.97, "learning_rate": 5.079663164434867e-08, "loss": 1.156, "step": 4606 }, { "epoch": 0.97, "learning_rate": 5.0113081701882446e-08, "loss": 1.1798, "step": 4607 }, { "epoch": 0.97, "learning_rate": 4.943415054316392e-08, "loss": 1.1767, "step": 4608 }, { "epoch": 0.97, "learning_rate": 4.8759838483358745e-08, "loss": 1.2385, "step": 4609 }, { "epoch": 0.97, "learning_rate": 4.809014583548433e-08, "loss": 1.2363, "step": 4610 }, { "epoch": 0.97, "learning_rate": 4.742507291041532e-08, "loss": 1.1528, "step": 4611 }, { "epoch": 0.97, "learning_rate": 4.6764620016881426e-08, "loss": 1.177, "step": 4612 }, { "epoch": 0.97, "learning_rate": 4.610878746146852e-08, "loss": 1.1776, "step": 4613 }, { "epoch": 0.97, "learning_rate": 4.545757554861641e-08, "loss": 1.1956, "step": 4614 }, { "epoch": 0.97, "learning_rate": 4.481098458062105e-08, "loss": 1.1699, "step": 4615 }, { "epoch": 0.97, "learning_rate": 4.416901485763236e-08, "loss": 1.1898, "step": 4616 }, { "epoch": 0.97, "learning_rate": 4.3531666677657514e-08, "loss": 1.2006, "step": 4617 }, { "epoch": 0.97, "learning_rate": 4.289894033655429e-08, "loss": 1.2532, "step": 4618 }, { "epoch": 0.97, "learning_rate": 4.227083612803884e-08, "loss": 1.1686, "step": 4619 }, { "epoch": 0.97, "learning_rate": 4.164735434367795e-08, "loss": 1.216, "step": 4620 }, { "epoch": 0.97, "learning_rate": 4.102849527289787e-08, "loss": 1.1995, "step": 4621 }, { "epoch": 0.97, "learning_rate": 4.041425920297326e-08, "loss": 1.1909, "step": 4622 }, { "epoch": 0.97, "learning_rate": 3.980464641903714e-08, "loss": 1.2362, "step": 4623 }, { "epoch": 0.97, "learning_rate": 3.919965720407315e-08, "loss": 1.1618, "step": 4624 }, { "epoch": 0.97, "learning_rate": 3.859929183892108e-08, "loss": 1.2006, "step": 4625 }, { "epoch": 0.97, "learning_rate": 3.800355060227246e-08, "loss": 1.1717, "step": 4626 }, { "epoch": 0.97, "learning_rate": 3.741243377067383e-08, "loss": 1.2523, "step": 4627 }, { "epoch": 0.97, "learning_rate": 3.6825941618524594e-08, "loss": 1.2083, "step": 4628 }, { "epoch": 0.97, "learning_rate": 3.624407441807587e-08, "loss": 1.2225, "step": 4629 }, { "epoch": 0.97, "learning_rate": 3.566683243943381e-08, "loss": 1.1788, "step": 4630 }, { "epoch": 0.97, "learning_rate": 3.5094215950557396e-08, "loss": 1.204, "step": 4631 }, { "epoch": 0.97, "learning_rate": 3.452622521725735e-08, "loss": 1.2447, "step": 4632 }, { "epoch": 0.97, "learning_rate": 3.3962860503198305e-08, "loss": 1.2525, "step": 4633 }, { "epoch": 0.97, "learning_rate": 3.3404122069895516e-08, "loss": 1.2321, "step": 4634 }, { "epoch": 0.97, "learning_rate": 3.285001017671818e-08, "loss": 1.1813, "step": 4635 }, { "epoch": 0.98, "learning_rate": 3.230052508088943e-08, "loss": 1.1765, "step": 4636 }, { "epoch": 0.98, "learning_rate": 3.175566703747968e-08, "loss": 1.2351, "step": 4637 }, { "epoch": 0.98, "learning_rate": 3.12154362994177e-08, "loss": 1.163, "step": 4638 }, { "epoch": 0.98, "learning_rate": 3.0679833117479576e-08, "loss": 1.2038, "step": 4639 }, { "epoch": 0.98, "learning_rate": 3.014885774029419e-08, "loss": 1.1715, "step": 4640 }, { "epoch": 0.98, "learning_rate": 2.9622510414343275e-08, "loss": 1.1805, "step": 4641 }, { "epoch": 0.98, "learning_rate": 2.910079138396027e-08, "loss": 1.1962, "step": 4642 }, { "epoch": 0.98, "learning_rate": 2.8583700891330335e-08, "loss": 1.1678, "step": 4643 }, { "epoch": 0.98, "learning_rate": 2.807123917648702e-08, "loss": 1.1614, "step": 4644 }, { "epoch": 0.98, "learning_rate": 2.756340647731892e-08, "loss": 1.2261, "step": 4645 }, { "epoch": 0.98, "learning_rate": 2.7060203029564137e-08, "loss": 1.1618, "step": 4646 }, { "epoch": 0.98, "learning_rate": 2.656162906681137e-08, "loss": 1.2048, "step": 4647 }, { "epoch": 0.98, "learning_rate": 2.606768482050215e-08, "loss": 1.1575, "step": 4648 }, { "epoch": 0.98, "learning_rate": 2.55783705199264e-08, "loss": 1.1752, "step": 4649 }, { "epoch": 0.98, "learning_rate": 2.5093686392226867e-08, "loss": 1.1789, "step": 4650 }, { "epoch": 0.98, "learning_rate": 2.4613632662396914e-08, "loss": 1.168, "step": 4651 }, { "epoch": 0.98, "learning_rate": 2.413820955327828e-08, "loss": 1.1799, "step": 4652 }, { "epoch": 0.98, "learning_rate": 2.3667417285564433e-08, "loss": 1.194, "step": 4653 }, { "epoch": 0.98, "learning_rate": 2.320125607780166e-08, "loss": 1.1576, "step": 4654 }, { "epoch": 0.98, "learning_rate": 2.2739726146381315e-08, "loss": 1.1718, "step": 4655 }, { "epoch": 0.98, "learning_rate": 2.228282770554979e-08, "loss": 1.1587, "step": 4656 }, { "epoch": 0.98, "learning_rate": 2.183056096740077e-08, "loss": 1.15, "step": 4657 }, { "epoch": 0.98, "learning_rate": 2.1382926141877426e-08, "loss": 1.1721, "step": 4658 }, { "epoch": 0.98, "learning_rate": 2.0939923436775755e-08, "loss": 1.1405, "step": 4659 }, { "epoch": 0.98, "learning_rate": 2.0501553057739043e-08, "loss": 1.2103, "step": 4660 }, { "epoch": 0.98, "learning_rate": 2.0067815208260066e-08, "loss": 1.184, "step": 4661 }, { "epoch": 0.98, "learning_rate": 1.963871008968332e-08, "loss": 1.2166, "step": 4662 }, { "epoch": 0.98, "learning_rate": 1.9214237901199472e-08, "loss": 1.1771, "step": 4663 }, { "epoch": 0.98, "learning_rate": 1.8794398839853123e-08, "loss": 1.1911, "step": 4664 }, { "epoch": 0.98, "learning_rate": 1.8379193100532822e-08, "loss": 1.2095, "step": 4665 }, { "epoch": 0.98, "learning_rate": 1.7968620875981058e-08, "loss": 1.2029, "step": 4666 }, { "epoch": 0.98, "learning_rate": 1.7562682356786488e-08, "loss": 1.1936, "step": 4667 }, { "epoch": 0.98, "learning_rate": 1.7161377731387263e-08, "loss": 1.1, "step": 4668 }, { "epoch": 0.98, "learning_rate": 1.6764707186071037e-08, "loss": 1.25, "step": 4669 }, { "epoch": 0.98, "learning_rate": 1.6372670904974962e-08, "loss": 1.2072, "step": 4670 }, { "epoch": 0.98, "learning_rate": 1.5985269070083463e-08, "loss": 1.1765, "step": 4671 }, { "epoch": 0.98, "learning_rate": 1.5602501861230467e-08, "loss": 1.1828, "step": 4672 }, { "epoch": 0.98, "learning_rate": 1.5224369456098287e-08, "loss": 1.2103, "step": 4673 }, { "epoch": 0.98, "learning_rate": 1.4850872030217623e-08, "loss": 1.1717, "step": 4674 }, { "epoch": 0.98, "learning_rate": 1.4482009756967563e-08, "loss": 1.1878, "step": 4675 }, { "epoch": 0.98, "learning_rate": 1.4117782807575586e-08, "loss": 1.1393, "step": 4676 }, { "epoch": 0.98, "learning_rate": 1.3758191351117556e-08, "loss": 1.167, "step": 4677 }, { "epoch": 0.98, "learning_rate": 1.3403235554518834e-08, "loss": 1.1962, "step": 4678 }, { "epoch": 0.98, "learning_rate": 1.3052915582549842e-08, "loss": 1.1948, "step": 4679 }, { "epoch": 0.98, "learning_rate": 1.2707231597831604e-08, "loss": 1.1806, "step": 4680 }, { "epoch": 0.98, "learning_rate": 1.236618376083132e-08, "loss": 1.1725, "step": 4681 }, { "epoch": 0.98, "learning_rate": 1.2029772229865678e-08, "loss": 1.1324, "step": 4682 }, { "epoch": 0.99, "learning_rate": 1.1697997161099761e-08, "loss": 1.1387, "step": 4683 }, { "epoch": 0.99, "learning_rate": 1.1370858708542598e-08, "loss": 1.2094, "step": 4684 }, { "epoch": 0.99, "learning_rate": 1.1048357024054934e-08, "loss": 1.1951, "step": 4685 }, { "epoch": 0.99, "learning_rate": 1.0730492257342573e-08, "loss": 1.2451, "step": 4686 }, { "epoch": 0.99, "learning_rate": 1.0417264555960815e-08, "loss": 1.1628, "step": 4687 }, { "epoch": 0.99, "learning_rate": 1.010867406531113e-08, "loss": 1.1806, "step": 4688 }, { "epoch": 0.99, "learning_rate": 9.804720928643375e-09, "loss": 1.1792, "step": 4689 }, { "epoch": 0.99, "learning_rate": 9.505405287051351e-09, "loss": 1.166, "step": 4690 }, { "epoch": 0.99, "learning_rate": 9.21072727948169e-09, "loss": 1.2074, "step": 4691 }, { "epoch": 0.99, "learning_rate": 8.920687042723864e-09, "loss": 1.2013, "step": 4692 }, { "epoch": 0.99, "learning_rate": 8.635284711416835e-09, "loss": 1.2151, "step": 4693 }, { "epoch": 0.99, "learning_rate": 8.354520418044631e-09, "loss": 1.1813, "step": 4694 }, { "epoch": 0.99, "learning_rate": 8.078394292939662e-09, "loss": 1.2125, "step": 4695 }, { "epoch": 0.99, "learning_rate": 7.806906464281617e-09, "loss": 1.1952, "step": 4696 }, { "epoch": 0.99, "learning_rate": 7.54005705809524e-09, "loss": 1.2475, "step": 4697 }, { "epoch": 0.99, "learning_rate": 7.2778461982547784e-09, "loss": 1.1781, "step": 4698 }, { "epoch": 0.99, "learning_rate": 7.020274006478422e-09, "loss": 1.1653, "step": 4699 }, { "epoch": 0.99, "learning_rate": 6.767340602332751e-09, "loss": 1.1838, "step": 4700 }, { "epoch": 0.99, "learning_rate": 6.5190461032305085e-09, "loss": 1.1988, "step": 4701 }, { "epoch": 0.99, "learning_rate": 6.275390624431721e-09, "loss": 1.1958, "step": 4702 }, { "epoch": 0.99, "learning_rate": 6.036374279041468e-09, "loss": 1.1996, "step": 4703 }, { "epoch": 0.99, "learning_rate": 5.801997178013219e-09, "loss": 1.1707, "step": 4704 }, { "epoch": 0.99, "learning_rate": 5.572259430145499e-09, "loss": 1.1874, "step": 4705 }, { "epoch": 0.99, "learning_rate": 5.347161142083001e-09, "loss": 1.1825, "step": 4706 }, { "epoch": 0.99, "learning_rate": 5.126702418318808e-09, "loss": 1.1682, "step": 4707 }, { "epoch": 0.99, "learning_rate": 4.910883361189944e-09, "loss": 1.2087, "step": 4708 }, { "epoch": 0.99, "learning_rate": 4.699704070880717e-09, "loss": 1.1498, "step": 4709 }, { "epoch": 0.99, "learning_rate": 4.4931646454216e-09, "loss": 1.2258, "step": 4710 }, { "epoch": 0.99, "learning_rate": 4.29126518068923e-09, "loss": 1.2365, "step": 4711 }, { "epoch": 0.99, "learning_rate": 4.094005770406417e-09, "loss": 1.2283, "step": 4712 }, { "epoch": 0.99, "learning_rate": 3.901386506141025e-09, "loss": 1.2064, "step": 4713 }, { "epoch": 0.99, "learning_rate": 3.7134074773093054e-09, "loss": 1.1932, "step": 4714 }, { "epoch": 0.99, "learning_rate": 3.5300687711703473e-09, "loss": 1.2108, "step": 4715 }, { "epoch": 0.99, "learning_rate": 3.351370472831628e-09, "loss": 1.119, "step": 4716 }, { "epoch": 0.99, "learning_rate": 3.17731266524679e-09, "loss": 1.1932, "step": 4717 }, { "epoch": 0.99, "learning_rate": 3.0078954292123152e-09, "loss": 1.2385, "step": 4718 }, { "epoch": 0.99, "learning_rate": 2.8431188433730715e-09, "loss": 1.172, "step": 4719 }, { "epoch": 0.99, "learning_rate": 2.682982984220095e-09, "loss": 1.1843, "step": 4720 }, { "epoch": 0.99, "learning_rate": 2.527487926088368e-09, "loss": 1.213, "step": 4721 }, { "epoch": 0.99, "learning_rate": 2.37663374116015e-09, "loss": 1.214, "step": 4722 }, { "epoch": 0.99, "learning_rate": 2.2304204994616498e-09, "loss": 1.2028, "step": 4723 }, { "epoch": 0.99, "learning_rate": 2.08884826886524e-09, "loss": 1.2237, "step": 4724 }, { "epoch": 0.99, "learning_rate": 1.951917115091684e-09, "loss": 1.207, "step": 4725 }, { "epoch": 0.99, "learning_rate": 1.8196271017034694e-09, "loss": 1.2264, "step": 4726 }, { "epoch": 0.99, "learning_rate": 1.691978290109253e-09, "loss": 1.1568, "step": 4727 }, { "epoch": 0.99, "learning_rate": 1.5689707395671883e-09, "loss": 1.1969, "step": 4728 }, { "epoch": 0.99, "learning_rate": 1.4506045071749352e-09, "loss": 1.2022, "step": 4729 }, { "epoch": 0.99, "learning_rate": 1.3368796478807622e-09, "loss": 1.127, "step": 4730 }, { "epoch": 1.0, "learning_rate": 1.2277962144746636e-09, "loss": 1.1954, "step": 4731 }, { "epoch": 1.0, "learning_rate": 1.1233542575950218e-09, "loss": 1.183, "step": 4732 }, { "epoch": 1.0, "learning_rate": 1.023553825723056e-09, "loss": 1.2511, "step": 4733 }, { "epoch": 1.0, "learning_rate": 9.283949651883728e-10, "loss": 1.1954, "step": 4734 }, { "epoch": 1.0, "learning_rate": 8.378777201611954e-10, "loss": 1.2071, "step": 4735 }, { "epoch": 1.0, "learning_rate": 7.520021326634652e-10, "loss": 1.1647, "step": 4736 }, { "epoch": 1.0, "learning_rate": 6.707682425566297e-10, "loss": 1.1826, "step": 4737 }, { "epoch": 1.0, "learning_rate": 5.941760875516345e-10, "loss": 1.191, "step": 4738 }, { "epoch": 1.0, "learning_rate": 5.222257032011513e-10, "loss": 1.22, "step": 4739 }, { "epoch": 1.0, "learning_rate": 4.5491712290512966e-10, "loss": 1.2068, "step": 4740 }, { "epoch": 1.0, "learning_rate": 3.9225037790968647e-10, "loss": 1.201, "step": 4741 }, { "epoch": 1.0, "learning_rate": 3.3422549730377507e-10, "loss": 1.1673, "step": 4742 }, { "epoch": 1.0, "learning_rate": 2.808425080247368e-10, "loss": 1.2017, "step": 4743 }, { "epoch": 1.0, "learning_rate": 2.3210143485052904e-10, "loss": 1.1728, "step": 4744 }, { "epoch": 1.0, "learning_rate": 1.8800230040860735e-10, "loss": 1.1879, "step": 4745 }, { "epoch": 1.0, "learning_rate": 1.48545125170374e-10, "loss": 1.1823, "step": 4746 }, { "epoch": 1.0, "learning_rate": 1.1372992745117828e-10, "loss": 1.125, "step": 4747 }, { "epoch": 1.0, "learning_rate": 8.355672341253673e-11, "loss": 1.228, "step": 4748 }, { "epoch": 1.0, "learning_rate": 5.802552706102305e-11, "loss": 1.1718, "step": 4749 }, { "epoch": 1.0, "learning_rate": 3.7136350248268005e-11, "loss": 1.1835, "step": 4750 }, { "epoch": 1.0, "learning_rate": 2.0889202670959508e-11, "loss": 1.2091, "step": 4751 }, { "epoch": 1.0, "learning_rate": 9.284091871952783e-12, "loss": 1.177, "step": 4752 }, { "epoch": 1.0, "learning_rate": 2.3210232369397233e-12, "loss": 1.1736, "step": 4753 }, { "epoch": 1.0, "learning_rate": 0.0, "loss": 1.1231, "step": 4754 }, { "epoch": 1.0, "step": 4754, "total_flos": 1.1050762967760398e+20, "train_loss": 1.2421564723846539, "train_runtime": 293812.8553, "train_samples_per_second": 2.071, "train_steps_per_second": 0.016 } ], "logging_steps": 1.0, "max_steps": 4754, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 2000, "total_flos": 1.1050762967760398e+20, "train_batch_size": 2, "trial_name": null, "trial_params": null }