{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.6329094049301305, "eval_steps": 500, "global_step": 2600, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.6785, "step": 1 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.8173, "step": 2 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.8519, "step": 3 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.8161, "step": 4 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.9623, "step": 5 }, { "epoch": 0.0, "learning_rate": 3.125e-07, "loss": 1.9781, "step": 6 }, { "epoch": 0.0, "learning_rate": 6.25e-07, "loss": 1.8395, "step": 7 }, { "epoch": 0.01, "learning_rate": 9.375000000000001e-07, "loss": 1.8538, "step": 8 }, { "epoch": 0.01, "learning_rate": 1.25e-06, "loss": 1.8738, "step": 9 }, { "epoch": 0.01, "learning_rate": 1.25e-06, "loss": 1.7725, "step": 10 }, { "epoch": 0.01, "learning_rate": 1.5625e-06, "loss": 1.7717, "step": 11 }, { "epoch": 0.01, "learning_rate": 1.5625e-06, "loss": 1.3481, "step": 12 }, { "epoch": 0.01, "learning_rate": 1.8750000000000003e-06, "loss": 1.3672, "step": 13 }, { "epoch": 0.01, "learning_rate": 2.1875000000000002e-06, "loss": 1.1213, "step": 14 }, { "epoch": 0.01, "learning_rate": 2.5e-06, "loss": 1.0117, "step": 15 }, { "epoch": 0.01, "learning_rate": 2.8125e-06, "loss": 0.8643, "step": 16 }, { "epoch": 0.01, "learning_rate": 3.125e-06, "loss": 0.826, "step": 17 }, { "epoch": 0.01, "learning_rate": 3.4375e-06, "loss": 0.7619, "step": 18 }, { "epoch": 0.01, "learning_rate": 3.7500000000000005e-06, "loss": 0.7784, "step": 19 }, { "epoch": 0.01, "learning_rate": 4.0625000000000005e-06, "loss": 0.7272, "step": 20 }, { "epoch": 0.01, "learning_rate": 4.3750000000000005e-06, "loss": 0.7252, "step": 21 }, { "epoch": 0.01, "learning_rate": 4.6875000000000004e-06, "loss": 0.6816, "step": 22 }, { "epoch": 0.01, "learning_rate": 5e-06, "loss": 0.6981, "step": 23 }, { "epoch": 0.02, "learning_rate": 5.3125e-06, "loss": 0.6461, "step": 24 }, { "epoch": 0.02, "learning_rate": 5.625e-06, "loss": 0.5715, "step": 25 }, { "epoch": 0.02, "learning_rate": 5.9375e-06, "loss": 0.6193, "step": 26 }, { "epoch": 0.02, "learning_rate": 6.25e-06, "loss": 0.606, "step": 27 }, { "epoch": 0.02, "learning_rate": 6.5625e-06, "loss": 0.6024, "step": 28 }, { "epoch": 0.02, "learning_rate": 6.875e-06, "loss": 0.6012, "step": 29 }, { "epoch": 0.02, "learning_rate": 7.1875e-06, "loss": 0.613, "step": 30 }, { "epoch": 0.02, "learning_rate": 7.500000000000001e-06, "loss": 0.6226, "step": 31 }, { "epoch": 0.02, "learning_rate": 7.8125e-06, "loss": 0.6528, "step": 32 }, { "epoch": 0.02, "learning_rate": 8.125000000000001e-06, "loss": 0.5886, "step": 33 }, { "epoch": 0.02, "learning_rate": 8.4375e-06, "loss": 0.5643, "step": 34 }, { "epoch": 0.02, "learning_rate": 8.750000000000001e-06, "loss": 0.5814, "step": 35 }, { "epoch": 0.02, "learning_rate": 9.0625e-06, "loss": 0.5491, "step": 36 }, { "epoch": 0.02, "learning_rate": 9.375000000000001e-06, "loss": 0.5851, "step": 37 }, { "epoch": 0.02, "learning_rate": 9.6875e-06, "loss": 0.5716, "step": 38 }, { "epoch": 0.02, "learning_rate": 1e-05, "loss": 0.5773, "step": 39 }, { "epoch": 0.03, "learning_rate": 9.999997516482067e-06, "loss": 0.5746, "step": 40 }, { "epoch": 0.03, "learning_rate": 9.99999006593073e-06, "loss": 0.5307, "step": 41 }, { "epoch": 0.03, "learning_rate": 9.999977648353398e-06, "loss": 0.5376, "step": 42 }, { "epoch": 0.03, "learning_rate": 9.9999602637624e-06, "loss": 0.5464, "step": 43 }, { "epoch": 0.03, "learning_rate": 9.999937912175009e-06, "loss": 0.5671, "step": 44 }, { "epoch": 0.03, "learning_rate": 9.999910593613427e-06, "loss": 0.5556, "step": 45 }, { "epoch": 0.03, "learning_rate": 9.999878308104794e-06, "loss": 0.5888, "step": 46 }, { "epoch": 0.03, "learning_rate": 9.999841055681184e-06, "loss": 0.5263, "step": 47 }, { "epoch": 0.03, "learning_rate": 9.999798836379602e-06, "loss": 0.5179, "step": 48 }, { "epoch": 0.03, "learning_rate": 9.999751650241989e-06, "loss": 0.5093, "step": 49 }, { "epoch": 0.03, "learning_rate": 9.999699497315221e-06, "loss": 0.5054, "step": 50 }, { "epoch": 0.03, "learning_rate": 9.999642377651106e-06, "loss": 0.5352, "step": 51 }, { "epoch": 0.03, "learning_rate": 9.999580291306388e-06, "loss": 0.5508, "step": 52 }, { "epoch": 0.03, "learning_rate": 9.999513238342742e-06, "loss": 0.5124, "step": 53 }, { "epoch": 0.03, "learning_rate": 9.999441218826783e-06, "loss": 0.5289, "step": 54 }, { "epoch": 0.03, "learning_rate": 9.999364232830053e-06, "loss": 0.5413, "step": 55 }, { "epoch": 0.04, "learning_rate": 9.99928228042903e-06, "loss": 0.5443, "step": 56 }, { "epoch": 0.04, "learning_rate": 9.999195361705126e-06, "loss": 0.5719, "step": 57 }, { "epoch": 0.04, "learning_rate": 9.999103476744687e-06, "loss": 0.5293, "step": 58 }, { "epoch": 0.04, "learning_rate": 9.999006625638994e-06, "loss": 0.5248, "step": 59 }, { "epoch": 0.04, "learning_rate": 9.998904808484257e-06, "loss": 0.5064, "step": 60 }, { "epoch": 0.04, "learning_rate": 9.998798025381626e-06, "loss": 0.5314, "step": 61 }, { "epoch": 0.04, "learning_rate": 9.998686276437174e-06, "loss": 0.5468, "step": 62 }, { "epoch": 0.04, "learning_rate": 9.99856956176192e-06, "loss": 0.5297, "step": 63 }, { "epoch": 0.04, "learning_rate": 9.998447881471802e-06, "loss": 0.5323, "step": 64 }, { "epoch": 0.04, "learning_rate": 9.998321235687703e-06, "loss": 0.5214, "step": 65 }, { "epoch": 0.04, "learning_rate": 9.998189624535432e-06, "loss": 0.5443, "step": 66 }, { "epoch": 0.04, "learning_rate": 9.998053048145735e-06, "loss": 0.5355, "step": 67 }, { "epoch": 0.04, "learning_rate": 9.997911506654283e-06, "loss": 0.4714, "step": 68 }, { "epoch": 0.04, "learning_rate": 9.997765000201689e-06, "loss": 0.5, "step": 69 }, { "epoch": 0.04, "learning_rate": 9.997613528933493e-06, "loss": 0.5075, "step": 70 }, { "epoch": 0.04, "learning_rate": 9.997457093000165e-06, "loss": 0.4432, "step": 71 }, { "epoch": 0.05, "learning_rate": 9.997295692557109e-06, "loss": 0.4635, "step": 72 }, { "epoch": 0.05, "learning_rate": 9.997129327764668e-06, "loss": 0.5223, "step": 73 }, { "epoch": 0.05, "learning_rate": 9.996957998788102e-06, "loss": 0.5161, "step": 74 }, { "epoch": 0.05, "learning_rate": 9.996781705797615e-06, "loss": 0.4285, "step": 75 }, { "epoch": 0.05, "learning_rate": 9.996600448968337e-06, "loss": 0.4788, "step": 76 }, { "epoch": 0.05, "learning_rate": 9.996414228480327e-06, "loss": 0.4414, "step": 77 }, { "epoch": 0.05, "learning_rate": 9.996223044518582e-06, "loss": 0.5264, "step": 78 }, { "epoch": 0.05, "learning_rate": 9.996026897273024e-06, "loss": 0.4764, "step": 79 }, { "epoch": 0.05, "learning_rate": 9.995825786938506e-06, "loss": 0.4739, "step": 80 }, { "epoch": 0.05, "learning_rate": 9.995619713714815e-06, "loss": 0.4916, "step": 81 }, { "epoch": 0.05, "learning_rate": 9.995408677806662e-06, "loss": 0.503, "step": 82 }, { "epoch": 0.05, "learning_rate": 9.995192679423692e-06, "loss": 0.5006, "step": 83 }, { "epoch": 0.05, "learning_rate": 9.994971718780484e-06, "loss": 0.4586, "step": 84 }, { "epoch": 0.05, "learning_rate": 9.994745796096537e-06, "loss": 0.5031, "step": 85 }, { "epoch": 0.05, "learning_rate": 9.994514911596288e-06, "loss": 0.4747, "step": 86 }, { "epoch": 0.05, "learning_rate": 9.994279065509094e-06, "loss": 0.4787, "step": 87 }, { "epoch": 0.06, "learning_rate": 9.99403825806925e-06, "loss": 0.4675, "step": 88 }, { "epoch": 0.06, "learning_rate": 9.993792489515977e-06, "loss": 0.4971, "step": 89 }, { "epoch": 0.06, "learning_rate": 9.99354176009342e-06, "loss": 0.4658, "step": 90 }, { "epoch": 0.06, "learning_rate": 9.993286070050656e-06, "loss": 0.4619, "step": 91 }, { "epoch": 0.06, "learning_rate": 9.993025419641692e-06, "loss": 0.4556, "step": 92 }, { "epoch": 0.06, "learning_rate": 9.992759809125456e-06, "loss": 0.4838, "step": 93 }, { "epoch": 0.06, "learning_rate": 9.992489238765812e-06, "loss": 0.4907, "step": 94 }, { "epoch": 0.06, "learning_rate": 9.992213708831542e-06, "loss": 0.476, "step": 95 }, { "epoch": 0.06, "learning_rate": 9.991933219596363e-06, "loss": 0.4988, "step": 96 }, { "epoch": 0.06, "learning_rate": 9.991647771338915e-06, "loss": 0.4686, "step": 97 }, { "epoch": 0.06, "learning_rate": 9.99135736434276e-06, "loss": 0.4952, "step": 98 }, { "epoch": 0.06, "learning_rate": 9.991061998896395e-06, "loss": 0.4864, "step": 99 }, { "epoch": 0.06, "learning_rate": 9.990761675293235e-06, "loss": 0.4574, "step": 100 }, { "epoch": 0.06, "learning_rate": 9.990456393831626e-06, "loss": 0.4194, "step": 101 }, { "epoch": 0.06, "learning_rate": 9.990146154814834e-06, "loss": 0.5, "step": 102 }, { "epoch": 0.06, "learning_rate": 9.989830958551058e-06, "loss": 0.4835, "step": 103 }, { "epoch": 0.07, "learning_rate": 9.989510805353411e-06, "loss": 0.4773, "step": 104 }, { "epoch": 0.07, "learning_rate": 9.989185695539937e-06, "loss": 0.4607, "step": 105 }, { "epoch": 0.07, "learning_rate": 9.988855629433602e-06, "loss": 0.4755, "step": 106 }, { "epoch": 0.07, "learning_rate": 9.988520607362297e-06, "loss": 0.4556, "step": 107 }, { "epoch": 0.07, "learning_rate": 9.988180629658837e-06, "loss": 0.4509, "step": 108 }, { "epoch": 0.07, "learning_rate": 9.987835696660955e-06, "loss": 0.4994, "step": 109 }, { "epoch": 0.07, "learning_rate": 9.987485808711311e-06, "loss": 0.4651, "step": 110 }, { "epoch": 0.07, "learning_rate": 9.987130966157486e-06, "loss": 0.4638, "step": 111 }, { "epoch": 0.07, "learning_rate": 9.986771169351983e-06, "loss": 0.4551, "step": 112 }, { "epoch": 0.07, "learning_rate": 9.986406418652227e-06, "loss": 0.4722, "step": 113 }, { "epoch": 0.07, "learning_rate": 9.986036714420567e-06, "loss": 0.4616, "step": 114 }, { "epoch": 0.07, "learning_rate": 9.985662057024267e-06, "loss": 0.5015, "step": 115 }, { "epoch": 0.07, "learning_rate": 9.98528244683551e-06, "loss": 0.4371, "step": 116 }, { "epoch": 0.07, "learning_rate": 9.984897884231412e-06, "loss": 0.4672, "step": 117 }, { "epoch": 0.07, "learning_rate": 9.984508369593995e-06, "loss": 0.4731, "step": 118 }, { "epoch": 0.07, "learning_rate": 9.984113903310206e-06, "loss": 0.4834, "step": 119 }, { "epoch": 0.08, "learning_rate": 9.983714485771912e-06, "loss": 0.4393, "step": 120 }, { "epoch": 0.08, "learning_rate": 9.983310117375897e-06, "loss": 0.421, "step": 121 }, { "epoch": 0.08, "learning_rate": 9.982900798523862e-06, "loss": 0.4593, "step": 122 }, { "epoch": 0.08, "learning_rate": 9.98248652962243e-06, "loss": 0.4296, "step": 123 }, { "epoch": 0.08, "learning_rate": 9.982067311083137e-06, "loss": 0.4162, "step": 124 }, { "epoch": 0.08, "learning_rate": 9.981643143322437e-06, "loss": 0.496, "step": 125 }, { "epoch": 0.08, "learning_rate": 9.981214026761703e-06, "loss": 0.4415, "step": 126 }, { "epoch": 0.08, "learning_rate": 9.98077996182722e-06, "loss": 0.5109, "step": 127 }, { "epoch": 0.08, "learning_rate": 9.980340948950194e-06, "loss": 0.4666, "step": 128 }, { "epoch": 0.08, "learning_rate": 9.979896988566742e-06, "loss": 0.4529, "step": 129 }, { "epoch": 0.08, "learning_rate": 9.979448081117897e-06, "loss": 0.5027, "step": 130 }, { "epoch": 0.08, "learning_rate": 9.97899422704961e-06, "loss": 0.4485, "step": 131 }, { "epoch": 0.08, "learning_rate": 9.978535426812738e-06, "loss": 0.4089, "step": 132 }, { "epoch": 0.08, "learning_rate": 9.978071680863061e-06, "loss": 0.4242, "step": 133 }, { "epoch": 0.08, "learning_rate": 9.977602989661264e-06, "loss": 0.4296, "step": 134 }, { "epoch": 0.08, "learning_rate": 9.977129353672951e-06, "loss": 0.4329, "step": 135 }, { "epoch": 0.09, "learning_rate": 9.976650773368634e-06, "loss": 0.4391, "step": 136 }, { "epoch": 0.09, "learning_rate": 9.976167249223737e-06, "loss": 0.4464, "step": 137 }, { "epoch": 0.09, "learning_rate": 9.975678781718599e-06, "loss": 0.4338, "step": 138 }, { "epoch": 0.09, "learning_rate": 9.975185371338464e-06, "loss": 0.4687, "step": 139 }, { "epoch": 0.09, "learning_rate": 9.974687018573492e-06, "loss": 0.4463, "step": 140 }, { "epoch": 0.09, "learning_rate": 9.974183723918752e-06, "loss": 0.4751, "step": 141 }, { "epoch": 0.09, "learning_rate": 9.973675487874214e-06, "loss": 0.4319, "step": 142 }, { "epoch": 0.09, "learning_rate": 9.973162310944769e-06, "loss": 0.4496, "step": 143 }, { "epoch": 0.09, "learning_rate": 9.97264419364021e-06, "loss": 0.4124, "step": 144 }, { "epoch": 0.09, "learning_rate": 9.972121136475234e-06, "loss": 0.4491, "step": 145 }, { "epoch": 0.09, "learning_rate": 9.971593139969454e-06, "loss": 0.4247, "step": 146 }, { "epoch": 0.09, "learning_rate": 9.971060204647384e-06, "loss": 0.4121, "step": 147 }, { "epoch": 0.09, "learning_rate": 9.970522331038449e-06, "loss": 0.3973, "step": 148 }, { "epoch": 0.09, "learning_rate": 9.969979519676973e-06, "loss": 0.4493, "step": 149 }, { "epoch": 0.09, "learning_rate": 9.96943177110219e-06, "loss": 0.4328, "step": 150 }, { "epoch": 0.09, "learning_rate": 9.968879085858234e-06, "loss": 0.3766, "step": 151 }, { "epoch": 0.1, "learning_rate": 9.968321464494152e-06, "loss": 0.3669, "step": 152 }, { "epoch": 0.1, "learning_rate": 9.967758907563885e-06, "loss": 0.4442, "step": 153 }, { "epoch": 0.1, "learning_rate": 9.967191415626283e-06, "loss": 0.4076, "step": 154 }, { "epoch": 0.1, "learning_rate": 9.966618989245096e-06, "loss": 0.4205, "step": 155 }, { "epoch": 0.1, "learning_rate": 9.966041628988977e-06, "loss": 0.3831, "step": 156 }, { "epoch": 0.1, "learning_rate": 9.96545933543148e-06, "loss": 0.4252, "step": 157 }, { "epoch": 0.1, "learning_rate": 9.964872109151058e-06, "loss": 0.4758, "step": 158 }, { "epoch": 0.1, "learning_rate": 9.964279950731066e-06, "loss": 0.425, "step": 159 }, { "epoch": 0.1, "learning_rate": 9.96368286075976e-06, "loss": 0.484, "step": 160 }, { "epoch": 0.1, "learning_rate": 9.963080839830294e-06, "loss": 0.4269, "step": 161 }, { "epoch": 0.1, "learning_rate": 9.962473888540717e-06, "loss": 0.4072, "step": 162 }, { "epoch": 0.1, "learning_rate": 9.961862007493981e-06, "loss": 0.3922, "step": 163 }, { "epoch": 0.1, "learning_rate": 9.961245197297931e-06, "loss": 0.3935, "step": 164 }, { "epoch": 0.1, "learning_rate": 9.960623458565313e-06, "loss": 0.4195, "step": 165 }, { "epoch": 0.1, "learning_rate": 9.959996791913766e-06, "loss": 0.4105, "step": 166 }, { "epoch": 0.1, "learning_rate": 9.959365197965824e-06, "loss": 0.4035, "step": 167 }, { "epoch": 0.11, "learning_rate": 9.958728677348919e-06, "loss": 0.4195, "step": 168 }, { "epoch": 0.11, "learning_rate": 9.958087230695374e-06, "loss": 0.4059, "step": 169 }, { "epoch": 0.11, "learning_rate": 9.957440858642406e-06, "loss": 0.4078, "step": 170 }, { "epoch": 0.11, "learning_rate": 9.956789561832127e-06, "loss": 0.4163, "step": 171 }, { "epoch": 0.11, "learning_rate": 9.956133340911538e-06, "loss": 0.4355, "step": 172 }, { "epoch": 0.11, "learning_rate": 9.955472196532537e-06, "loss": 0.4146, "step": 173 }, { "epoch": 0.11, "learning_rate": 9.954806129351905e-06, "loss": 0.3807, "step": 174 }, { "epoch": 0.11, "learning_rate": 9.954135140031322e-06, "loss": 0.4081, "step": 175 }, { "epoch": 0.11, "learning_rate": 9.95345922923735e-06, "loss": 0.4189, "step": 176 }, { "epoch": 0.11, "learning_rate": 9.952778397641447e-06, "loss": 0.4468, "step": 177 }, { "epoch": 0.11, "learning_rate": 9.952092645919955e-06, "loss": 0.4059, "step": 178 }, { "epoch": 0.11, "learning_rate": 9.951401974754103e-06, "loss": 0.4354, "step": 179 }, { "epoch": 0.11, "learning_rate": 9.950706384830012e-06, "loss": 0.3806, "step": 180 }, { "epoch": 0.11, "learning_rate": 9.950005876838681e-06, "loss": 0.3779, "step": 181 }, { "epoch": 0.11, "learning_rate": 9.949300451476004e-06, "loss": 0.3968, "step": 182 }, { "epoch": 0.11, "learning_rate": 9.948590109442755e-06, "loss": 0.408, "step": 183 }, { "epoch": 0.12, "learning_rate": 9.94787485144459e-06, "loss": 0.4403, "step": 184 }, { "epoch": 0.12, "learning_rate": 9.947154678192055e-06, "loss": 0.3764, "step": 185 }, { "epoch": 0.12, "learning_rate": 9.946429590400573e-06, "loss": 0.4093, "step": 186 }, { "epoch": 0.12, "learning_rate": 9.945699588790455e-06, "loss": 0.44, "step": 187 }, { "epoch": 0.12, "learning_rate": 9.944964674086884e-06, "loss": 0.3524, "step": 188 }, { "epoch": 0.12, "learning_rate": 9.944224847019933e-06, "loss": 0.3876, "step": 189 }, { "epoch": 0.12, "learning_rate": 9.943480108324551e-06, "loss": 0.3764, "step": 190 }, { "epoch": 0.12, "learning_rate": 9.942730458740568e-06, "loss": 0.3845, "step": 191 }, { "epoch": 0.12, "learning_rate": 9.941975899012688e-06, "loss": 0.3673, "step": 192 }, { "epoch": 0.12, "learning_rate": 9.941216429890501e-06, "loss": 0.4156, "step": 193 }, { "epoch": 0.12, "learning_rate": 9.940452052128464e-06, "loss": 0.398, "step": 194 }, { "epoch": 0.12, "learning_rate": 9.93968276648592e-06, "loss": 0.3948, "step": 195 }, { "epoch": 0.12, "learning_rate": 9.938908573727078e-06, "loss": 0.4053, "step": 196 }, { "epoch": 0.12, "learning_rate": 9.93812947462103e-06, "loss": 0.3382, "step": 197 }, { "epoch": 0.12, "learning_rate": 9.937345469941737e-06, "loss": 0.4168, "step": 198 }, { "epoch": 0.12, "learning_rate": 9.936556560468037e-06, "loss": 0.3929, "step": 199 }, { "epoch": 0.13, "learning_rate": 9.935762746983636e-06, "loss": 0.4287, "step": 200 }, { "epoch": 0.13, "learning_rate": 9.934964030277117e-06, "loss": 0.3609, "step": 201 }, { "epoch": 0.13, "learning_rate": 9.934160411141926e-06, "loss": 0.4088, "step": 202 }, { "epoch": 0.13, "learning_rate": 9.933351890376389e-06, "loss": 0.4093, "step": 203 }, { "epoch": 0.13, "learning_rate": 9.932538468783692e-06, "loss": 0.4293, "step": 204 }, { "epoch": 0.13, "learning_rate": 9.931720147171896e-06, "loss": 0.3853, "step": 205 }, { "epoch": 0.13, "learning_rate": 9.93089692635393e-06, "loss": 0.4042, "step": 206 }, { "epoch": 0.13, "learning_rate": 9.930068807147585e-06, "loss": 0.4361, "step": 207 }, { "epoch": 0.13, "learning_rate": 9.92923579037552e-06, "loss": 0.4101, "step": 208 }, { "epoch": 0.13, "learning_rate": 9.92839787686526e-06, "loss": 0.3964, "step": 209 }, { "epoch": 0.13, "learning_rate": 9.927555067449194e-06, "loss": 0.3836, "step": 210 }, { "epoch": 0.13, "learning_rate": 9.926707362964579e-06, "loss": 0.3744, "step": 211 }, { "epoch": 0.13, "learning_rate": 9.925854764253524e-06, "loss": 0.3407, "step": 212 }, { "epoch": 0.13, "learning_rate": 9.92499727216301e-06, "loss": 0.4, "step": 213 }, { "epoch": 0.13, "learning_rate": 9.924134887544878e-06, "loss": 0.3627, "step": 214 }, { "epoch": 0.14, "learning_rate": 9.923267611255824e-06, "loss": 0.4092, "step": 215 }, { "epoch": 0.14, "learning_rate": 9.922395444157406e-06, "loss": 0.364, "step": 216 }, { "epoch": 0.14, "learning_rate": 9.921518387116045e-06, "loss": 0.3623, "step": 217 }, { "epoch": 0.14, "learning_rate": 9.920636441003013e-06, "loss": 0.3927, "step": 218 }, { "epoch": 0.14, "learning_rate": 9.91974960669444e-06, "loss": 0.3637, "step": 219 }, { "epoch": 0.14, "learning_rate": 9.918857885071316e-06, "loss": 0.3802, "step": 220 }, { "epoch": 0.14, "learning_rate": 9.917961277019485e-06, "loss": 0.3547, "step": 221 }, { "epoch": 0.14, "learning_rate": 9.91705978342964e-06, "loss": 0.391, "step": 222 }, { "epoch": 0.14, "learning_rate": 9.916153405197333e-06, "loss": 0.402, "step": 223 }, { "epoch": 0.14, "learning_rate": 9.915242143222967e-06, "loss": 0.3652, "step": 224 }, { "epoch": 0.14, "learning_rate": 9.914325998411797e-06, "loss": 0.3991, "step": 225 }, { "epoch": 0.14, "learning_rate": 9.913404971673925e-06, "loss": 0.407, "step": 226 }, { "epoch": 0.14, "learning_rate": 9.912479063924309e-06, "loss": 0.3634, "step": 227 }, { "epoch": 0.14, "learning_rate": 9.91154827608275e-06, "loss": 0.4086, "step": 228 }, { "epoch": 0.14, "learning_rate": 9.9106126090739e-06, "loss": 0.3804, "step": 229 }, { "epoch": 0.14, "learning_rate": 9.909672063827257e-06, "loss": 0.3997, "step": 230 }, { "epoch": 0.15, "learning_rate": 9.908726641277167e-06, "loss": 0.3684, "step": 231 }, { "epoch": 0.15, "learning_rate": 9.907776342362817e-06, "loss": 0.3782, "step": 232 }, { "epoch": 0.15, "learning_rate": 9.906821168028244e-06, "loss": 0.3708, "step": 233 }, { "epoch": 0.15, "learning_rate": 9.905861119222321e-06, "loss": 0.396, "step": 234 }, { "epoch": 0.15, "learning_rate": 9.904896196898771e-06, "loss": 0.412, "step": 235 }, { "epoch": 0.15, "learning_rate": 9.903926402016153e-06, "loss": 0.4072, "step": 236 }, { "epoch": 0.15, "learning_rate": 9.902951735537867e-06, "loss": 0.379, "step": 237 }, { "epoch": 0.15, "learning_rate": 9.901972198432157e-06, "loss": 0.3951, "step": 238 }, { "epoch": 0.15, "learning_rate": 9.9009877916721e-06, "loss": 0.3666, "step": 239 }, { "epoch": 0.15, "learning_rate": 9.899998516235612e-06, "loss": 0.3667, "step": 240 }, { "epoch": 0.15, "learning_rate": 9.899004373105447e-06, "loss": 0.4278, "step": 241 }, { "epoch": 0.15, "learning_rate": 9.898005363269196e-06, "loss": 0.3795, "step": 242 }, { "epoch": 0.15, "learning_rate": 9.89700148771928e-06, "loss": 0.3462, "step": 243 }, { "epoch": 0.15, "learning_rate": 9.895992747452957e-06, "loss": 0.3601, "step": 244 }, { "epoch": 0.15, "learning_rate": 9.894979143472316e-06, "loss": 0.3967, "step": 245 }, { "epoch": 0.15, "learning_rate": 9.89396067678428e-06, "loss": 0.3593, "step": 246 }, { "epoch": 0.16, "learning_rate": 9.8929373484006e-06, "loss": 0.3585, "step": 247 }, { "epoch": 0.16, "learning_rate": 9.89190915933786e-06, "loss": 0.3811, "step": 248 }, { "epoch": 0.16, "learning_rate": 9.890876110617466e-06, "loss": 0.3162, "step": 249 }, { "epoch": 0.16, "learning_rate": 9.88983820326566e-06, "loss": 0.4086, "step": 250 }, { "epoch": 0.16, "learning_rate": 9.888795438313506e-06, "loss": 0.3511, "step": 251 }, { "epoch": 0.16, "learning_rate": 9.88774781679689e-06, "loss": 0.3837, "step": 252 }, { "epoch": 0.16, "learning_rate": 9.886695339756534e-06, "loss": 0.4009, "step": 253 }, { "epoch": 0.16, "learning_rate": 9.88563800823797e-06, "loss": 0.3712, "step": 254 }, { "epoch": 0.16, "learning_rate": 9.884575823291561e-06, "loss": 0.4292, "step": 255 }, { "epoch": 0.16, "learning_rate": 9.883508785972488e-06, "loss": 0.3568, "step": 256 }, { "epoch": 0.16, "learning_rate": 9.882436897340757e-06, "loss": 0.3482, "step": 257 }, { "epoch": 0.16, "learning_rate": 9.881360158461186e-06, "loss": 0.3464, "step": 258 }, { "epoch": 0.16, "learning_rate": 9.880278570403419e-06, "loss": 0.3448, "step": 259 }, { "epoch": 0.16, "learning_rate": 9.879192134241908e-06, "loss": 0.3548, "step": 260 }, { "epoch": 0.16, "learning_rate": 9.878100851055931e-06, "loss": 0.3547, "step": 261 }, { "epoch": 0.16, "learning_rate": 9.877004721929576e-06, "loss": 0.3592, "step": 262 }, { "epoch": 0.17, "learning_rate": 9.875903747951742e-06, "loss": 0.3841, "step": 263 }, { "epoch": 0.17, "learning_rate": 9.874797930216149e-06, "loss": 0.354, "step": 264 }, { "epoch": 0.17, "learning_rate": 9.873687269821322e-06, "loss": 0.3598, "step": 265 }, { "epoch": 0.17, "learning_rate": 9.8725717678706e-06, "loss": 0.3182, "step": 266 }, { "epoch": 0.17, "learning_rate": 9.871451425472128e-06, "loss": 0.3462, "step": 267 }, { "epoch": 0.17, "learning_rate": 9.870326243738866e-06, "loss": 0.3896, "step": 268 }, { "epoch": 0.17, "learning_rate": 9.869196223788574e-06, "loss": 0.3294, "step": 269 }, { "epoch": 0.17, "learning_rate": 9.868061366743824e-06, "loss": 0.3356, "step": 270 }, { "epoch": 0.17, "learning_rate": 9.866921673731991e-06, "loss": 0.4046, "step": 271 }, { "epoch": 0.17, "learning_rate": 9.865777145885255e-06, "loss": 0.384, "step": 272 }, { "epoch": 0.17, "learning_rate": 9.864627784340596e-06, "loss": 0.3748, "step": 273 }, { "epoch": 0.17, "learning_rate": 9.863473590239799e-06, "loss": 0.3602, "step": 274 }, { "epoch": 0.17, "learning_rate": 9.86231456472945e-06, "loss": 0.3643, "step": 275 }, { "epoch": 0.17, "learning_rate": 9.861150708960932e-06, "loss": 0.358, "step": 276 }, { "epoch": 0.17, "learning_rate": 9.859982024090426e-06, "loss": 0.3561, "step": 277 }, { "epoch": 0.17, "learning_rate": 9.858808511278914e-06, "loss": 0.348, "step": 278 }, { "epoch": 0.18, "learning_rate": 9.857630171692175e-06, "loss": 0.3607, "step": 279 }, { "epoch": 0.18, "learning_rate": 9.856447006500774e-06, "loss": 0.3783, "step": 280 }, { "epoch": 0.18, "learning_rate": 9.855259016880079e-06, "loss": 0.3923, "step": 281 }, { "epoch": 0.18, "learning_rate": 9.854066204010247e-06, "loss": 0.4031, "step": 282 }, { "epoch": 0.18, "learning_rate": 9.852868569076225e-06, "loss": 0.353, "step": 283 }, { "epoch": 0.18, "learning_rate": 9.851666113267758e-06, "loss": 0.3803, "step": 284 }, { "epoch": 0.18, "learning_rate": 9.850458837779366e-06, "loss": 0.3352, "step": 285 }, { "epoch": 0.18, "learning_rate": 9.849246743810371e-06, "loss": 0.3584, "step": 286 }, { "epoch": 0.18, "learning_rate": 9.848029832564875e-06, "loss": 0.3436, "step": 287 }, { "epoch": 0.18, "learning_rate": 9.846808105251764e-06, "loss": 0.3445, "step": 288 }, { "epoch": 0.18, "learning_rate": 9.845581563084714e-06, "loss": 0.3367, "step": 289 }, { "epoch": 0.18, "learning_rate": 9.844350207282178e-06, "loss": 0.323, "step": 290 }, { "epoch": 0.18, "learning_rate": 9.843114039067395e-06, "loss": 0.408, "step": 291 }, { "epoch": 0.18, "learning_rate": 9.841873059668381e-06, "loss": 0.3212, "step": 292 }, { "epoch": 0.18, "learning_rate": 9.840627270317938e-06, "loss": 0.3396, "step": 293 }, { "epoch": 0.18, "learning_rate": 9.83937667225364e-06, "loss": 0.3286, "step": 294 }, { "epoch": 0.19, "learning_rate": 9.83812126671784e-06, "loss": 0.351, "step": 295 }, { "epoch": 0.19, "learning_rate": 9.836861054957667e-06, "loss": 0.352, "step": 296 }, { "epoch": 0.19, "learning_rate": 9.835596038225023e-06, "loss": 0.3616, "step": 297 }, { "epoch": 0.19, "learning_rate": 9.834326217776587e-06, "loss": 0.3649, "step": 298 }, { "epoch": 0.19, "learning_rate": 9.833051594873805e-06, "loss": 0.3448, "step": 299 }, { "epoch": 0.19, "learning_rate": 9.8317721707829e-06, "loss": 0.3769, "step": 300 }, { "epoch": 0.19, "learning_rate": 9.830487946774861e-06, "loss": 0.352, "step": 301 }, { "epoch": 0.19, "learning_rate": 9.829198924125441e-06, "loss": 0.3543, "step": 302 }, { "epoch": 0.19, "learning_rate": 9.827905104115167e-06, "loss": 0.4017, "step": 303 }, { "epoch": 0.19, "learning_rate": 9.826606488029332e-06, "loss": 0.3504, "step": 304 }, { "epoch": 0.19, "learning_rate": 9.825303077157983e-06, "loss": 0.352, "step": 305 }, { "epoch": 0.19, "learning_rate": 9.823994872795945e-06, "loss": 0.3666, "step": 306 }, { "epoch": 0.19, "learning_rate": 9.822681876242797e-06, "loss": 0.3174, "step": 307 }, { "epoch": 0.19, "learning_rate": 9.821364088802874e-06, "loss": 0.3034, "step": 308 }, { "epoch": 0.19, "learning_rate": 9.82004151178528e-06, "loss": 0.352, "step": 309 }, { "epoch": 0.19, "learning_rate": 9.818714146503872e-06, "loss": 0.3076, "step": 310 }, { "epoch": 0.2, "learning_rate": 9.81738199427726e-06, "loss": 0.388, "step": 311 }, { "epoch": 0.2, "learning_rate": 9.81604505642882e-06, "loss": 0.3096, "step": 312 }, { "epoch": 0.2, "learning_rate": 9.814703334286673e-06, "loss": 0.3593, "step": 313 }, { "epoch": 0.2, "learning_rate": 9.813356829183693e-06, "loss": 0.3491, "step": 314 }, { "epoch": 0.2, "learning_rate": 9.81200554245751e-06, "loss": 0.339, "step": 315 }, { "epoch": 0.2, "learning_rate": 9.810649475450503e-06, "loss": 0.341, "step": 316 }, { "epoch": 0.2, "learning_rate": 9.809288629509797e-06, "loss": 0.3789, "step": 317 }, { "epoch": 0.2, "learning_rate": 9.807923005987265e-06, "loss": 0.336, "step": 318 }, { "epoch": 0.2, "learning_rate": 9.80655260623953e-06, "loss": 0.2954, "step": 319 }, { "epoch": 0.2, "learning_rate": 9.805177431627957e-06, "loss": 0.345, "step": 320 }, { "epoch": 0.2, "learning_rate": 9.80379748351865e-06, "loss": 0.3766, "step": 321 }, { "epoch": 0.2, "learning_rate": 9.802412763282465e-06, "loss": 0.3233, "step": 322 }, { "epoch": 0.2, "learning_rate": 9.80102327229499e-06, "loss": 0.3255, "step": 323 }, { "epoch": 0.2, "learning_rate": 9.799629011936554e-06, "loss": 0.335, "step": 324 }, { "epoch": 0.2, "learning_rate": 9.798229983592229e-06, "loss": 0.3623, "step": 325 }, { "epoch": 0.2, "learning_rate": 9.796826188651815e-06, "loss": 0.3363, "step": 326 }, { "epoch": 0.21, "learning_rate": 9.795417628509857e-06, "loss": 0.3097, "step": 327 }, { "epoch": 0.21, "learning_rate": 9.794004304565627e-06, "loss": 0.291, "step": 328 }, { "epoch": 0.21, "learning_rate": 9.79258621822313e-06, "loss": 0.3407, "step": 329 }, { "epoch": 0.21, "learning_rate": 9.791163370891102e-06, "loss": 0.3642, "step": 330 }, { "epoch": 0.21, "learning_rate": 9.789735763983013e-06, "loss": 0.3644, "step": 331 }, { "epoch": 0.21, "learning_rate": 9.788303398917057e-06, "loss": 0.3504, "step": 332 }, { "epoch": 0.21, "learning_rate": 9.786866277116154e-06, "loss": 0.3632, "step": 333 }, { "epoch": 0.21, "learning_rate": 9.785424400007952e-06, "loss": 0.3494, "step": 334 }, { "epoch": 0.21, "learning_rate": 9.783977769024821e-06, "loss": 0.3421, "step": 335 }, { "epoch": 0.21, "learning_rate": 9.78252638560386e-06, "loss": 0.3497, "step": 336 }, { "epoch": 0.21, "learning_rate": 9.781070251186877e-06, "loss": 0.3171, "step": 337 }, { "epoch": 0.21, "learning_rate": 9.779609367220409e-06, "loss": 0.3424, "step": 338 }, { "epoch": 0.21, "learning_rate": 9.778143735155709e-06, "loss": 0.3439, "step": 339 }, { "epoch": 0.21, "learning_rate": 9.776673356448746e-06, "loss": 0.3313, "step": 340 }, { "epoch": 0.21, "learning_rate": 9.775198232560204e-06, "loss": 0.3577, "step": 341 }, { "epoch": 0.21, "learning_rate": 9.773718364955484e-06, "loss": 0.3221, "step": 342 }, { "epoch": 0.22, "learning_rate": 9.772233755104695e-06, "loss": 0.3194, "step": 343 }, { "epoch": 0.22, "learning_rate": 9.770744404482659e-06, "loss": 0.3341, "step": 344 }, { "epoch": 0.22, "learning_rate": 9.769250314568907e-06, "loss": 0.354, "step": 345 }, { "epoch": 0.22, "learning_rate": 9.767751486847681e-06, "loss": 0.3176, "step": 346 }, { "epoch": 0.22, "learning_rate": 9.766247922807927e-06, "loss": 0.3179, "step": 347 }, { "epoch": 0.22, "learning_rate": 9.764739623943295e-06, "loss": 0.3452, "step": 348 }, { "epoch": 0.22, "learning_rate": 9.76322659175214e-06, "loss": 0.3289, "step": 349 }, { "epoch": 0.22, "learning_rate": 9.76170882773752e-06, "loss": 0.3524, "step": 350 }, { "epoch": 0.22, "learning_rate": 9.76018633340719e-06, "loss": 0.3628, "step": 351 }, { "epoch": 0.22, "learning_rate": 9.75865911027361e-06, "loss": 0.2923, "step": 352 }, { "epoch": 0.22, "learning_rate": 9.757127159853934e-06, "loss": 0.3264, "step": 353 }, { "epoch": 0.22, "learning_rate": 9.75559048367001e-06, "loss": 0.3246, "step": 354 }, { "epoch": 0.22, "learning_rate": 9.754049083248384e-06, "loss": 0.3164, "step": 355 }, { "epoch": 0.22, "learning_rate": 9.752502960120298e-06, "loss": 0.3056, "step": 356 }, { "epoch": 0.22, "learning_rate": 9.750952115821677e-06, "loss": 0.3218, "step": 357 }, { "epoch": 0.22, "learning_rate": 9.749396551893144e-06, "loss": 0.3012, "step": 358 }, { "epoch": 0.23, "learning_rate": 9.747836269880005e-06, "loss": 0.3214, "step": 359 }, { "epoch": 0.23, "learning_rate": 9.746271271332255e-06, "loss": 0.3344, "step": 360 }, { "epoch": 0.23, "learning_rate": 9.744701557804578e-06, "loss": 0.3282, "step": 361 }, { "epoch": 0.23, "learning_rate": 9.743127130856338e-06, "loss": 0.3542, "step": 362 }, { "epoch": 0.23, "learning_rate": 9.74154799205158e-06, "loss": 0.3447, "step": 363 }, { "epoch": 0.23, "learning_rate": 9.739964142959032e-06, "loss": 0.3302, "step": 364 }, { "epoch": 0.23, "learning_rate": 9.738375585152103e-06, "loss": 0.3464, "step": 365 }, { "epoch": 0.23, "learning_rate": 9.736782320208875e-06, "loss": 0.3102, "step": 366 }, { "epoch": 0.23, "learning_rate": 9.73518434971211e-06, "loss": 0.321, "step": 367 }, { "epoch": 0.23, "learning_rate": 9.733581675249245e-06, "loss": 0.2966, "step": 368 }, { "epoch": 0.23, "learning_rate": 9.731974298412387e-06, "loss": 0.3368, "step": 369 }, { "epoch": 0.23, "learning_rate": 9.730362220798315e-06, "loss": 0.314, "step": 370 }, { "epoch": 0.23, "learning_rate": 9.728745444008481e-06, "loss": 0.34, "step": 371 }, { "epoch": 0.23, "learning_rate": 9.727123969648998e-06, "loss": 0.3521, "step": 372 }, { "epoch": 0.23, "learning_rate": 9.725497799330654e-06, "loss": 0.3258, "step": 373 }, { "epoch": 0.23, "learning_rate": 9.7238669346689e-06, "loss": 0.3533, "step": 374 }, { "epoch": 0.24, "learning_rate": 9.722231377283841e-06, "loss": 0.3175, "step": 375 }, { "epoch": 0.24, "learning_rate": 9.72059112880026e-06, "loss": 0.3175, "step": 376 }, { "epoch": 0.24, "learning_rate": 9.718946190847587e-06, "loss": 0.3492, "step": 377 }, { "epoch": 0.24, "learning_rate": 9.717296565059918e-06, "loss": 0.3029, "step": 378 }, { "epoch": 0.24, "learning_rate": 9.715642253075998e-06, "loss": 0.3457, "step": 379 }, { "epoch": 0.24, "learning_rate": 9.713983256539238e-06, "loss": 0.3227, "step": 380 }, { "epoch": 0.24, "learning_rate": 9.712319577097693e-06, "loss": 0.311, "step": 381 }, { "epoch": 0.24, "learning_rate": 9.710651216404076e-06, "loss": 0.3054, "step": 382 }, { "epoch": 0.24, "learning_rate": 9.70897817611575e-06, "loss": 0.3558, "step": 383 }, { "epoch": 0.24, "learning_rate": 9.707300457894723e-06, "loss": 0.3299, "step": 384 }, { "epoch": 0.24, "learning_rate": 9.705618063407653e-06, "loss": 0.3091, "step": 385 }, { "epoch": 0.24, "learning_rate": 9.703930994325842e-06, "loss": 0.3639, "step": 386 }, { "epoch": 0.24, "learning_rate": 9.702239252325237e-06, "loss": 0.3128, "step": 387 }, { "epoch": 0.24, "learning_rate": 9.700542839086426e-06, "loss": 0.3543, "step": 388 }, { "epoch": 0.24, "learning_rate": 9.698841756294641e-06, "loss": 0.2852, "step": 389 }, { "epoch": 0.24, "learning_rate": 9.697136005639746e-06, "loss": 0.3029, "step": 390 }, { "epoch": 0.25, "learning_rate": 9.695425588816248e-06, "loss": 0.3236, "step": 391 }, { "epoch": 0.25, "learning_rate": 9.693710507523288e-06, "loss": 0.3257, "step": 392 }, { "epoch": 0.25, "learning_rate": 9.691990763464638e-06, "loss": 0.2911, "step": 393 }, { "epoch": 0.25, "learning_rate": 9.690266358348706e-06, "loss": 0.3343, "step": 394 }, { "epoch": 0.25, "learning_rate": 9.688537293888525e-06, "loss": 0.3303, "step": 395 }, { "epoch": 0.25, "learning_rate": 9.686803571801766e-06, "loss": 0.3332, "step": 396 }, { "epoch": 0.25, "learning_rate": 9.685065193810716e-06, "loss": 0.3159, "step": 397 }, { "epoch": 0.25, "learning_rate": 9.683322161642292e-06, "loss": 0.3344, "step": 398 }, { "epoch": 0.25, "learning_rate": 9.681574477028039e-06, "loss": 0.3084, "step": 399 }, { "epoch": 0.25, "learning_rate": 9.679822141704116e-06, "loss": 0.3278, "step": 400 }, { "epoch": 0.25, "learning_rate": 9.678065157411304e-06, "loss": 0.3392, "step": 401 }, { "epoch": 0.25, "learning_rate": 9.67630352589501e-06, "loss": 0.3176, "step": 402 }, { "epoch": 0.25, "learning_rate": 9.674537248905244e-06, "loss": 0.3025, "step": 403 }, { "epoch": 0.25, "learning_rate": 9.672766328196644e-06, "loss": 0.3252, "step": 404 }, { "epoch": 0.25, "learning_rate": 9.670990765528453e-06, "loss": 0.3337, "step": 405 }, { "epoch": 0.25, "learning_rate": 9.669210562664527e-06, "loss": 0.3039, "step": 406 }, { "epoch": 0.26, "learning_rate": 9.667425721373333e-06, "loss": 0.2994, "step": 407 }, { "epoch": 0.26, "learning_rate": 9.665636243427944e-06, "loss": 0.3202, "step": 408 }, { "epoch": 0.26, "learning_rate": 9.663842130606042e-06, "loss": 0.2981, "step": 409 }, { "epoch": 0.26, "learning_rate": 9.662043384689912e-06, "loss": 0.2577, "step": 410 }, { "epoch": 0.26, "learning_rate": 9.66024000746644e-06, "loss": 0.3168, "step": 411 }, { "epoch": 0.26, "learning_rate": 9.658432000727113e-06, "loss": 0.3115, "step": 412 }, { "epoch": 0.26, "learning_rate": 9.65661936626802e-06, "loss": 0.3473, "step": 413 }, { "epoch": 0.26, "learning_rate": 9.654802105889842e-06, "loss": 0.3284, "step": 414 }, { "epoch": 0.26, "learning_rate": 9.65298022139786e-06, "loss": 0.3092, "step": 415 }, { "epoch": 0.26, "learning_rate": 9.651153714601951e-06, "loss": 0.3061, "step": 416 }, { "epoch": 0.26, "learning_rate": 9.649322587316573e-06, "loss": 0.3345, "step": 417 }, { "epoch": 0.26, "learning_rate": 9.647486841360786e-06, "loss": 0.3008, "step": 418 }, { "epoch": 0.26, "learning_rate": 9.64564647855823e-06, "loss": 0.2709, "step": 419 }, { "epoch": 0.26, "learning_rate": 9.643801500737137e-06, "loss": 0.2634, "step": 420 }, { "epoch": 0.26, "learning_rate": 9.641951909730322e-06, "loss": 0.3617, "step": 421 }, { "epoch": 0.27, "learning_rate": 9.64009770737518e-06, "loss": 0.3339, "step": 422 }, { "epoch": 0.27, "learning_rate": 9.638238895513687e-06, "loss": 0.3407, "step": 423 }, { "epoch": 0.27, "learning_rate": 9.636375475992404e-06, "loss": 0.329, "step": 424 }, { "epoch": 0.27, "learning_rate": 9.634507450662463e-06, "loss": 0.2944, "step": 425 }, { "epoch": 0.27, "learning_rate": 9.632634821379575e-06, "loss": 0.3318, "step": 426 }, { "epoch": 0.27, "learning_rate": 9.630757590004023e-06, "loss": 0.2944, "step": 427 }, { "epoch": 0.27, "learning_rate": 9.628875758400662e-06, "loss": 0.3467, "step": 428 }, { "epoch": 0.27, "learning_rate": 9.626989328438915e-06, "loss": 0.3268, "step": 429 }, { "epoch": 0.27, "learning_rate": 9.625098301992779e-06, "loss": 0.3411, "step": 430 }, { "epoch": 0.27, "learning_rate": 9.623202680940811e-06, "loss": 0.3116, "step": 431 }, { "epoch": 0.27, "learning_rate": 9.621302467166134e-06, "loss": 0.2948, "step": 432 }, { "epoch": 0.27, "learning_rate": 9.619397662556434e-06, "loss": 0.2803, "step": 433 }, { "epoch": 0.27, "learning_rate": 9.61748826900396e-06, "loss": 0.2734, "step": 434 }, { "epoch": 0.27, "learning_rate": 9.615574288405515e-06, "loss": 0.2909, "step": 435 }, { "epoch": 0.27, "learning_rate": 9.613655722662462e-06, "loss": 0.3217, "step": 436 }, { "epoch": 0.27, "learning_rate": 9.611732573680716e-06, "loss": 0.3327, "step": 437 }, { "epoch": 0.28, "learning_rate": 9.60980484337075e-06, "loss": 0.3011, "step": 438 }, { "epoch": 0.28, "learning_rate": 9.607872533647584e-06, "loss": 0.2987, "step": 439 }, { "epoch": 0.28, "learning_rate": 9.605935646430788e-06, "loss": 0.29, "step": 440 }, { "epoch": 0.28, "learning_rate": 9.603994183644478e-06, "loss": 0.2837, "step": 441 }, { "epoch": 0.28, "learning_rate": 9.60204814721732e-06, "loss": 0.3354, "step": 442 }, { "epoch": 0.28, "learning_rate": 9.60009753908252e-06, "loss": 0.3071, "step": 443 }, { "epoch": 0.28, "learning_rate": 9.598142361177824e-06, "loss": 0.3099, "step": 444 }, { "epoch": 0.28, "learning_rate": 9.596182615445522e-06, "loss": 0.31, "step": 445 }, { "epoch": 0.28, "learning_rate": 9.594218303832439e-06, "loss": 0.2743, "step": 446 }, { "epoch": 0.28, "learning_rate": 9.592249428289935e-06, "loss": 0.2577, "step": 447 }, { "epoch": 0.28, "learning_rate": 9.590275990773907e-06, "loss": 0.3064, "step": 448 }, { "epoch": 0.28, "learning_rate": 9.588297993244779e-06, "loss": 0.307, "step": 449 }, { "epoch": 0.28, "learning_rate": 9.586315437667512e-06, "loss": 0.3044, "step": 450 }, { "epoch": 0.28, "learning_rate": 9.584328326011588e-06, "loss": 0.3023, "step": 451 }, { "epoch": 0.28, "learning_rate": 9.58233666025102e-06, "loss": 0.2951, "step": 452 }, { "epoch": 0.28, "learning_rate": 9.58034044236434e-06, "loss": 0.2505, "step": 453 }, { "epoch": 0.29, "learning_rate": 9.578339674334608e-06, "loss": 0.307, "step": 454 }, { "epoch": 0.29, "learning_rate": 9.5763343581494e-06, "loss": 0.3296, "step": 455 }, { "epoch": 0.29, "learning_rate": 9.574324495800812e-06, "loss": 0.287, "step": 456 }, { "epoch": 0.29, "learning_rate": 9.572310089285455e-06, "loss": 0.3218, "step": 457 }, { "epoch": 0.29, "learning_rate": 9.570291140604455e-06, "loss": 0.3028, "step": 458 }, { "epoch": 0.29, "learning_rate": 9.568267651763453e-06, "loss": 0.3035, "step": 459 }, { "epoch": 0.29, "learning_rate": 9.566239624772593e-06, "loss": 0.3165, "step": 460 }, { "epoch": 0.29, "learning_rate": 9.564207061646533e-06, "loss": 0.3131, "step": 461 }, { "epoch": 0.29, "learning_rate": 9.562169964404436e-06, "loss": 0.3258, "step": 462 }, { "epoch": 0.29, "learning_rate": 9.560128335069971e-06, "loss": 0.2786, "step": 463 }, { "epoch": 0.29, "learning_rate": 9.558082175671304e-06, "loss": 0.2729, "step": 464 }, { "epoch": 0.29, "learning_rate": 9.556031488241107e-06, "loss": 0.3237, "step": 465 }, { "epoch": 0.29, "learning_rate": 9.553976274816544e-06, "loss": 0.2798, "step": 466 }, { "epoch": 0.29, "learning_rate": 9.551916537439282e-06, "loss": 0.3119, "step": 467 }, { "epoch": 0.29, "learning_rate": 9.549852278155481e-06, "loss": 0.3163, "step": 468 }, { "epoch": 0.29, "learning_rate": 9.547783499015785e-06, "loss": 0.3231, "step": 469 }, { "epoch": 0.3, "learning_rate": 9.545710202075337e-06, "loss": 0.298, "step": 470 }, { "epoch": 0.3, "learning_rate": 9.543632389393767e-06, "loss": 0.2678, "step": 471 }, { "epoch": 0.3, "learning_rate": 9.541550063035188e-06, "loss": 0.2852, "step": 472 }, { "epoch": 0.3, "learning_rate": 9.539463225068195e-06, "loss": 0.333, "step": 473 }, { "epoch": 0.3, "learning_rate": 9.537371877565871e-06, "loss": 0.3141, "step": 474 }, { "epoch": 0.3, "learning_rate": 9.535276022605772e-06, "loss": 0.2992, "step": 475 }, { "epoch": 0.3, "learning_rate": 9.53317566226994e-06, "loss": 0.2571, "step": 476 }, { "epoch": 0.3, "learning_rate": 9.531070798644886e-06, "loss": 0.2459, "step": 477 }, { "epoch": 0.3, "learning_rate": 9.528961433821598e-06, "loss": 0.2946, "step": 478 }, { "epoch": 0.3, "learning_rate": 9.526847569895529e-06, "loss": 0.3441, "step": 479 }, { "epoch": 0.3, "learning_rate": 9.524729208966613e-06, "loss": 0.2801, "step": 480 }, { "epoch": 0.3, "learning_rate": 9.522606353139243e-06, "loss": 0.3003, "step": 481 }, { "epoch": 0.3, "learning_rate": 9.520479004522276e-06, "loss": 0.2524, "step": 482 }, { "epoch": 0.3, "learning_rate": 9.518347165229038e-06, "loss": 0.2923, "step": 483 }, { "epoch": 0.3, "learning_rate": 9.516210837377315e-06, "loss": 0.3268, "step": 484 }, { "epoch": 0.3, "learning_rate": 9.514070023089348e-06, "loss": 0.2731, "step": 485 }, { "epoch": 0.31, "learning_rate": 9.511924724491837e-06, "loss": 0.2975, "step": 486 }, { "epoch": 0.31, "learning_rate": 9.50977494371594e-06, "loss": 0.3059, "step": 487 }, { "epoch": 0.31, "learning_rate": 9.50762068289726e-06, "loss": 0.2807, "step": 488 }, { "epoch": 0.31, "learning_rate": 9.505461944175858e-06, "loss": 0.2945, "step": 489 }, { "epoch": 0.31, "learning_rate": 9.503298729696242e-06, "loss": 0.2821, "step": 490 }, { "epoch": 0.31, "learning_rate": 9.501131041607364e-06, "loss": 0.3065, "step": 491 }, { "epoch": 0.31, "learning_rate": 9.498958882062617e-06, "loss": 0.2692, "step": 492 }, { "epoch": 0.31, "learning_rate": 9.496782253219843e-06, "loss": 0.3029, "step": 493 }, { "epoch": 0.31, "learning_rate": 9.494601157241323e-06, "loss": 0.292, "step": 494 }, { "epoch": 0.31, "learning_rate": 9.49241559629377e-06, "loss": 0.3208, "step": 495 }, { "epoch": 0.31, "learning_rate": 9.490225572548335e-06, "loss": 0.3062, "step": 496 }, { "epoch": 0.31, "learning_rate": 9.488031088180607e-06, "loss": 0.2788, "step": 497 }, { "epoch": 0.31, "learning_rate": 9.4858321453706e-06, "loss": 0.2928, "step": 498 }, { "epoch": 0.31, "learning_rate": 9.483628746302761e-06, "loss": 0.2932, "step": 499 }, { "epoch": 0.31, "learning_rate": 9.481420893165962e-06, "loss": 0.295, "step": 500 }, { "epoch": 0.31, "learning_rate": 9.4792085881535e-06, "loss": 0.3107, "step": 501 }, { "epoch": 0.32, "learning_rate": 9.476991833463091e-06, "loss": 0.2725, "step": 502 }, { "epoch": 0.32, "learning_rate": 9.474770631296882e-06, "loss": 0.2736, "step": 503 }, { "epoch": 0.32, "learning_rate": 9.472544983861426e-06, "loss": 0.2568, "step": 504 }, { "epoch": 0.32, "learning_rate": 9.4703148933677e-06, "loss": 0.2901, "step": 505 }, { "epoch": 0.32, "learning_rate": 9.468080362031091e-06, "loss": 0.312, "step": 506 }, { "epoch": 0.32, "learning_rate": 9.465841392071396e-06, "loss": 0.2723, "step": 507 }, { "epoch": 0.32, "learning_rate": 9.463597985712827e-06, "loss": 0.3092, "step": 508 }, { "epoch": 0.32, "learning_rate": 9.461350145183999e-06, "loss": 0.2844, "step": 509 }, { "epoch": 0.32, "learning_rate": 9.459097872717933e-06, "loss": 0.3094, "step": 510 }, { "epoch": 0.32, "learning_rate": 9.456841170552054e-06, "loss": 0.2933, "step": 511 }, { "epoch": 0.32, "learning_rate": 9.454580040928183e-06, "loss": 0.2935, "step": 512 }, { "epoch": 0.32, "learning_rate": 9.452314486092543e-06, "loss": 0.2966, "step": 513 }, { "epoch": 0.32, "learning_rate": 9.450044508295755e-06, "loss": 0.279, "step": 514 }, { "epoch": 0.32, "learning_rate": 9.447770109792828e-06, "loss": 0.3039, "step": 515 }, { "epoch": 0.32, "learning_rate": 9.44549129284317e-06, "loss": 0.3017, "step": 516 }, { "epoch": 0.32, "learning_rate": 9.44320805971057e-06, "loss": 0.3409, "step": 517 }, { "epoch": 0.33, "learning_rate": 9.440920412663208e-06, "loss": 0.2959, "step": 518 }, { "epoch": 0.33, "learning_rate": 9.438628353973654e-06, "loss": 0.278, "step": 519 }, { "epoch": 0.33, "learning_rate": 9.43633188591885e-06, "loss": 0.2707, "step": 520 }, { "epoch": 0.33, "learning_rate": 9.434031010780127e-06, "loss": 0.2821, "step": 521 }, { "epoch": 0.33, "learning_rate": 9.431725730843188e-06, "loss": 0.2921, "step": 522 }, { "epoch": 0.33, "learning_rate": 9.42941604839812e-06, "loss": 0.2685, "step": 523 }, { "epoch": 0.33, "learning_rate": 9.427101965739371e-06, "loss": 0.2652, "step": 524 }, { "epoch": 0.33, "learning_rate": 9.424783485165775e-06, "loss": 0.2869, "step": 525 }, { "epoch": 0.33, "learning_rate": 9.42246060898052e-06, "loss": 0.3254, "step": 526 }, { "epoch": 0.33, "learning_rate": 9.420133339491171e-06, "loss": 0.2696, "step": 527 }, { "epoch": 0.33, "learning_rate": 9.417801679009657e-06, "loss": 0.3066, "step": 528 }, { "epoch": 0.33, "learning_rate": 9.41546562985226e-06, "loss": 0.3021, "step": 529 }, { "epoch": 0.33, "learning_rate": 9.413125194339634e-06, "loss": 0.3017, "step": 530 }, { "epoch": 0.33, "learning_rate": 9.410780374796782e-06, "loss": 0.2839, "step": 531 }, { "epoch": 0.33, "learning_rate": 9.408431173553062e-06, "loss": 0.2341, "step": 532 }, { "epoch": 0.33, "learning_rate": 9.406077592942193e-06, "loss": 0.2968, "step": 533 }, { "epoch": 0.34, "learning_rate": 9.403719635302235e-06, "loss": 0.3122, "step": 534 }, { "epoch": 0.34, "learning_rate": 9.4013573029756e-06, "loss": 0.3122, "step": 535 }, { "epoch": 0.34, "learning_rate": 9.398990598309048e-06, "loss": 0.2734, "step": 536 }, { "epoch": 0.34, "learning_rate": 9.396619523653679e-06, "loss": 0.2848, "step": 537 }, { "epoch": 0.34, "learning_rate": 9.394244081364934e-06, "loss": 0.3145, "step": 538 }, { "epoch": 0.34, "learning_rate": 9.391864273802596e-06, "loss": 0.2398, "step": 539 }, { "epoch": 0.34, "learning_rate": 9.389480103330783e-06, "loss": 0.3414, "step": 540 }, { "epoch": 0.34, "learning_rate": 9.387091572317947e-06, "loss": 0.3315, "step": 541 }, { "epoch": 0.34, "learning_rate": 9.384698683136874e-06, "loss": 0.288, "step": 542 }, { "epoch": 0.34, "learning_rate": 9.382301438164673e-06, "loss": 0.2751, "step": 543 }, { "epoch": 0.34, "learning_rate": 9.379899839782786e-06, "loss": 0.2734, "step": 544 }, { "epoch": 0.34, "learning_rate": 9.37749389037698e-06, "loss": 0.2301, "step": 545 }, { "epoch": 0.34, "learning_rate": 9.375083592337341e-06, "loss": 0.2781, "step": 546 }, { "epoch": 0.34, "learning_rate": 9.372668948058276e-06, "loss": 0.2474, "step": 547 }, { "epoch": 0.34, "learning_rate": 9.370249959938508e-06, "loss": 0.2913, "step": 548 }, { "epoch": 0.34, "learning_rate": 9.367826630381082e-06, "loss": 0.272, "step": 549 }, { "epoch": 0.35, "learning_rate": 9.365398961793348e-06, "loss": 0.2872, "step": 550 }, { "epoch": 0.35, "learning_rate": 9.36296695658697e-06, "loss": 0.2537, "step": 551 }, { "epoch": 0.35, "learning_rate": 9.36053061717792e-06, "loss": 0.2631, "step": 552 }, { "epoch": 0.35, "learning_rate": 9.358089945986471e-06, "loss": 0.3095, "step": 553 }, { "epoch": 0.35, "learning_rate": 9.355644945437208e-06, "loss": 0.2459, "step": 554 }, { "epoch": 0.35, "learning_rate": 9.353195617959011e-06, "loss": 0.2643, "step": 555 }, { "epoch": 0.35, "learning_rate": 9.35074196598506e-06, "loss": 0.3032, "step": 556 }, { "epoch": 0.35, "learning_rate": 9.348283991952827e-06, "loss": 0.2572, "step": 557 }, { "epoch": 0.35, "learning_rate": 9.345821698304086e-06, "loss": 0.2613, "step": 558 }, { "epoch": 0.35, "learning_rate": 9.343355087484893e-06, "loss": 0.2597, "step": 559 }, { "epoch": 0.35, "learning_rate": 9.340884161945598e-06, "loss": 0.2608, "step": 560 }, { "epoch": 0.35, "learning_rate": 9.338408924140839e-06, "loss": 0.3131, "step": 561 }, { "epoch": 0.35, "learning_rate": 9.33592937652953e-06, "loss": 0.2676, "step": 562 }, { "epoch": 0.35, "learning_rate": 9.333445521574878e-06, "loss": 0.2769, "step": 563 }, { "epoch": 0.35, "learning_rate": 9.330957361744355e-06, "loss": 0.295, "step": 564 }, { "epoch": 0.35, "learning_rate": 9.328464899509722e-06, "loss": 0.2827, "step": 565 }, { "epoch": 0.36, "learning_rate": 9.325968137347007e-06, "loss": 0.2667, "step": 566 }, { "epoch": 0.36, "learning_rate": 9.323467077736513e-06, "loss": 0.2719, "step": 567 }, { "epoch": 0.36, "learning_rate": 9.320961723162808e-06, "loss": 0.3023, "step": 568 }, { "epoch": 0.36, "learning_rate": 9.318452076114727e-06, "loss": 0.2873, "step": 569 }, { "epoch": 0.36, "learning_rate": 9.315938139085379e-06, "loss": 0.2735, "step": 570 }, { "epoch": 0.36, "learning_rate": 9.313419914572122e-06, "loss": 0.2593, "step": 571 }, { "epoch": 0.36, "learning_rate": 9.310897405076578e-06, "loss": 0.2787, "step": 572 }, { "epoch": 0.36, "learning_rate": 9.308370613104628e-06, "loss": 0.2967, "step": 573 }, { "epoch": 0.36, "learning_rate": 9.305839541166404e-06, "loss": 0.2991, "step": 574 }, { "epoch": 0.36, "learning_rate": 9.303304191776291e-06, "loss": 0.2558, "step": 575 }, { "epoch": 0.36, "learning_rate": 9.300764567452925e-06, "loss": 0.2796, "step": 576 }, { "epoch": 0.36, "learning_rate": 9.298220670719185e-06, "loss": 0.2559, "step": 577 }, { "epoch": 0.36, "learning_rate": 9.295672504102197e-06, "loss": 0.3041, "step": 578 }, { "epoch": 0.36, "learning_rate": 9.293120070133327e-06, "loss": 0.2884, "step": 579 }, { "epoch": 0.36, "learning_rate": 9.290563371348185e-06, "loss": 0.2465, "step": 580 }, { "epoch": 0.36, "learning_rate": 9.288002410286609e-06, "loss": 0.2557, "step": 581 }, { "epoch": 0.37, "learning_rate": 9.285437189492679e-06, "loss": 0.2604, "step": 582 }, { "epoch": 0.37, "learning_rate": 9.282867711514703e-06, "loss": 0.2838, "step": 583 }, { "epoch": 0.37, "learning_rate": 9.280293978905217e-06, "loss": 0.2633, "step": 584 }, { "epoch": 0.37, "learning_rate": 9.277715994220989e-06, "loss": 0.2402, "step": 585 }, { "epoch": 0.37, "learning_rate": 9.275133760023005e-06, "loss": 0.2422, "step": 586 }, { "epoch": 0.37, "learning_rate": 9.272547278876475e-06, "loss": 0.281, "step": 587 }, { "epoch": 0.37, "learning_rate": 9.269956553350831e-06, "loss": 0.2667, "step": 588 }, { "epoch": 0.37, "learning_rate": 9.267361586019714e-06, "loss": 0.2879, "step": 589 }, { "epoch": 0.37, "learning_rate": 9.264762379460986e-06, "loss": 0.272, "step": 590 }, { "epoch": 0.37, "learning_rate": 9.262158936256717e-06, "loss": 0.2732, "step": 591 }, { "epoch": 0.37, "learning_rate": 9.259551258993185e-06, "loss": 0.2853, "step": 592 }, { "epoch": 0.37, "learning_rate": 9.256939350260876e-06, "loss": 0.2483, "step": 593 }, { "epoch": 0.37, "learning_rate": 9.254323212654479e-06, "loss": 0.2836, "step": 594 }, { "epoch": 0.37, "learning_rate": 9.251702848772884e-06, "loss": 0.2815, "step": 595 }, { "epoch": 0.37, "learning_rate": 9.249078261219178e-06, "loss": 0.2459, "step": 596 }, { "epoch": 0.37, "learning_rate": 9.246449452600648e-06, "loss": 0.2612, "step": 597 }, { "epoch": 0.38, "learning_rate": 9.243816425528766e-06, "loss": 0.2601, "step": 598 }, { "epoch": 0.38, "learning_rate": 9.241179182619207e-06, "loss": 0.2706, "step": 599 }, { "epoch": 0.38, "learning_rate": 9.238537726491822e-06, "loss": 0.2834, "step": 600 }, { "epoch": 0.38, "learning_rate": 9.235892059770655e-06, "loss": 0.2547, "step": 601 }, { "epoch": 0.38, "learning_rate": 9.233242185083927e-06, "loss": 0.2638, "step": 602 }, { "epoch": 0.38, "learning_rate": 9.230588105064047e-06, "loss": 0.2518, "step": 603 }, { "epoch": 0.38, "learning_rate": 9.227929822347593e-06, "loss": 0.239, "step": 604 }, { "epoch": 0.38, "learning_rate": 9.225267339575325e-06, "loss": 0.2742, "step": 605 }, { "epoch": 0.38, "learning_rate": 9.222600659392172e-06, "loss": 0.2224, "step": 606 }, { "epoch": 0.38, "learning_rate": 9.219929784447232e-06, "loss": 0.2303, "step": 607 }, { "epoch": 0.38, "learning_rate": 9.217254717393772e-06, "loss": 0.282, "step": 608 }, { "epoch": 0.38, "learning_rate": 9.214575460889224e-06, "loss": 0.2435, "step": 609 }, { "epoch": 0.38, "learning_rate": 9.211892017595177e-06, "loss": 0.262, "step": 610 }, { "epoch": 0.38, "learning_rate": 9.209204390177388e-06, "loss": 0.2994, "step": 611 }, { "epoch": 0.38, "learning_rate": 9.206512581305762e-06, "loss": 0.2344, "step": 612 }, { "epoch": 0.38, "learning_rate": 9.203816593654361e-06, "loss": 0.2594, "step": 613 }, { "epoch": 0.39, "learning_rate": 9.2011164299014e-06, "loss": 0.2624, "step": 614 }, { "epoch": 0.39, "learning_rate": 9.19841209272924e-06, "loss": 0.3107, "step": 615 }, { "epoch": 0.39, "learning_rate": 9.195703584824389e-06, "loss": 0.2488, "step": 616 }, { "epoch": 0.39, "learning_rate": 9.192990908877499e-06, "loss": 0.2671, "step": 617 }, { "epoch": 0.39, "learning_rate": 9.190274067583363e-06, "loss": 0.2744, "step": 618 }, { "epoch": 0.39, "learning_rate": 9.187553063640905e-06, "loss": 0.2621, "step": 619 }, { "epoch": 0.39, "learning_rate": 9.184827899753197e-06, "loss": 0.2509, "step": 620 }, { "epoch": 0.39, "learning_rate": 9.182098578627431e-06, "loss": 0.3117, "step": 621 }, { "epoch": 0.39, "learning_rate": 9.17936510297494e-06, "loss": 0.2422, "step": 622 }, { "epoch": 0.39, "learning_rate": 9.176627475511171e-06, "loss": 0.2497, "step": 623 }, { "epoch": 0.39, "learning_rate": 9.173885698955707e-06, "loss": 0.2502, "step": 624 }, { "epoch": 0.39, "learning_rate": 9.17113977603225e-06, "loss": 0.292, "step": 625 }, { "epoch": 0.39, "learning_rate": 9.168389709468615e-06, "loss": 0.2733, "step": 626 }, { "epoch": 0.39, "learning_rate": 9.16563550199674e-06, "loss": 0.2611, "step": 627 }, { "epoch": 0.39, "learning_rate": 9.162877156352677e-06, "loss": 0.267, "step": 628 }, { "epoch": 0.4, "learning_rate": 9.160114675276582e-06, "loss": 0.2738, "step": 629 }, { "epoch": 0.4, "learning_rate": 9.157348061512728e-06, "loss": 0.2594, "step": 630 }, { "epoch": 0.4, "learning_rate": 9.154577317809483e-06, "loss": 0.2613, "step": 631 }, { "epoch": 0.4, "learning_rate": 9.15180244691933e-06, "loss": 0.2628, "step": 632 }, { "epoch": 0.4, "learning_rate": 9.149023451598843e-06, "loss": 0.2499, "step": 633 }, { "epoch": 0.4, "learning_rate": 9.146240334608694e-06, "loss": 0.257, "step": 634 }, { "epoch": 0.4, "learning_rate": 9.143453098713655e-06, "loss": 0.2818, "step": 635 }, { "epoch": 0.4, "learning_rate": 9.140661746682583e-06, "loss": 0.2422, "step": 636 }, { "epoch": 0.4, "learning_rate": 9.137866281288428e-06, "loss": 0.2625, "step": 637 }, { "epoch": 0.4, "learning_rate": 9.135066705308227e-06, "loss": 0.2597, "step": 638 }, { "epoch": 0.4, "learning_rate": 9.132263021523096e-06, "loss": 0.2523, "step": 639 }, { "epoch": 0.4, "learning_rate": 9.129455232718238e-06, "loss": 0.2331, "step": 640 }, { "epoch": 0.4, "learning_rate": 9.126643341682928e-06, "loss": 0.2361, "step": 641 }, { "epoch": 0.4, "learning_rate": 9.123827351210519e-06, "loss": 0.2709, "step": 642 }, { "epoch": 0.4, "learning_rate": 9.121007264098438e-06, "loss": 0.247, "step": 643 }, { "epoch": 0.4, "learning_rate": 9.118183083148178e-06, "loss": 0.3083, "step": 644 }, { "epoch": 0.41, "learning_rate": 9.115354811165298e-06, "loss": 0.2604, "step": 645 }, { "epoch": 0.41, "learning_rate": 9.112522450959431e-06, "loss": 0.2661, "step": 646 }, { "epoch": 0.41, "learning_rate": 9.109686005344258e-06, "loss": 0.2556, "step": 647 }, { "epoch": 0.41, "learning_rate": 9.106845477137525e-06, "loss": 0.2449, "step": 648 }, { "epoch": 0.41, "learning_rate": 9.104000869161035e-06, "loss": 0.2665, "step": 649 }, { "epoch": 0.41, "learning_rate": 9.101152184240642e-06, "loss": 0.2392, "step": 650 }, { "epoch": 0.41, "learning_rate": 9.098299425206248e-06, "loss": 0.2532, "step": 651 }, { "epoch": 0.41, "learning_rate": 9.095442594891806e-06, "loss": 0.2376, "step": 652 }, { "epoch": 0.41, "learning_rate": 9.092581696135311e-06, "loss": 0.249, "step": 653 }, { "epoch": 0.41, "learning_rate": 9.089716731778801e-06, "loss": 0.2483, "step": 654 }, { "epoch": 0.41, "learning_rate": 9.086847704668352e-06, "loss": 0.2563, "step": 655 }, { "epoch": 0.41, "learning_rate": 9.083974617654074e-06, "loss": 0.2241, "step": 656 }, { "epoch": 0.41, "learning_rate": 9.081097473590114e-06, "loss": 0.2933, "step": 657 }, { "epoch": 0.41, "learning_rate": 9.078216275334649e-06, "loss": 0.2313, "step": 658 }, { "epoch": 0.41, "learning_rate": 9.07533102574988e-06, "loss": 0.2224, "step": 659 }, { "epoch": 0.41, "learning_rate": 9.072441727702036e-06, "loss": 0.2612, "step": 660 }, { "epoch": 0.42, "learning_rate": 9.069548384061364e-06, "loss": 0.2534, "step": 661 }, { "epoch": 0.42, "learning_rate": 9.066650997702135e-06, "loss": 0.2749, "step": 662 }, { "epoch": 0.42, "learning_rate": 9.063749571502633e-06, "loss": 0.2764, "step": 663 }, { "epoch": 0.42, "learning_rate": 9.060844108345155e-06, "loss": 0.2729, "step": 664 }, { "epoch": 0.42, "learning_rate": 9.057934611116008e-06, "loss": 0.2691, "step": 665 }, { "epoch": 0.42, "learning_rate": 9.055021082705508e-06, "loss": 0.2654, "step": 666 }, { "epoch": 0.42, "learning_rate": 9.052103526007976e-06, "loss": 0.2529, "step": 667 }, { "epoch": 0.42, "learning_rate": 9.049181943921734e-06, "loss": 0.273, "step": 668 }, { "epoch": 0.42, "learning_rate": 9.0462563393491e-06, "loss": 0.274, "step": 669 }, { "epoch": 0.42, "learning_rate": 9.043326715196392e-06, "loss": 0.2623, "step": 670 }, { "epoch": 0.42, "learning_rate": 9.040393074373921e-06, "loss": 0.234, "step": 671 }, { "epoch": 0.42, "learning_rate": 9.037455419795986e-06, "loss": 0.2513, "step": 672 }, { "epoch": 0.42, "learning_rate": 9.034513754380874e-06, "loss": 0.2653, "step": 673 }, { "epoch": 0.42, "learning_rate": 9.031568081050857e-06, "loss": 0.2287, "step": 674 }, { "epoch": 0.42, "learning_rate": 9.028618402732185e-06, "loss": 0.2368, "step": 675 }, { "epoch": 0.42, "learning_rate": 9.025664722355094e-06, "loss": 0.2588, "step": 676 }, { "epoch": 0.43, "learning_rate": 9.022707042853789e-06, "loss": 0.2234, "step": 677 }, { "epoch": 0.43, "learning_rate": 9.01974536716645e-06, "loss": 0.2483, "step": 678 }, { "epoch": 0.43, "learning_rate": 9.016779698235227e-06, "loss": 0.2673, "step": 679 }, { "epoch": 0.43, "learning_rate": 9.013810039006237e-06, "loss": 0.2529, "step": 680 }, { "epoch": 0.43, "learning_rate": 9.01083639242956e-06, "loss": 0.2837, "step": 681 }, { "epoch": 0.43, "learning_rate": 9.00785876145924e-06, "loss": 0.2394, "step": 682 }, { "epoch": 0.43, "learning_rate": 9.004877149053277e-06, "loss": 0.257, "step": 683 }, { "epoch": 0.43, "learning_rate": 9.001891558173623e-06, "loss": 0.2563, "step": 684 }, { "epoch": 0.43, "learning_rate": 8.998901991786186e-06, "loss": 0.2234, "step": 685 }, { "epoch": 0.43, "learning_rate": 8.995908452860827e-06, "loss": 0.2596, "step": 686 }, { "epoch": 0.43, "learning_rate": 8.992910944371343e-06, "loss": 0.2553, "step": 687 }, { "epoch": 0.43, "learning_rate": 8.989909469295484e-06, "loss": 0.2584, "step": 688 }, { "epoch": 0.43, "learning_rate": 8.986904030614936e-06, "loss": 0.2599, "step": 689 }, { "epoch": 0.43, "learning_rate": 8.983894631315323e-06, "loss": 0.2348, "step": 690 }, { "epoch": 0.43, "learning_rate": 8.980881274386207e-06, "loss": 0.2746, "step": 691 }, { "epoch": 0.43, "learning_rate": 8.977863962821073e-06, "loss": 0.2363, "step": 692 }, { "epoch": 0.44, "learning_rate": 8.974842699617345e-06, "loss": 0.2273, "step": 693 }, { "epoch": 0.44, "learning_rate": 8.971817487776365e-06, "loss": 0.2858, "step": 694 }, { "epoch": 0.44, "learning_rate": 8.9687883303034e-06, "loss": 0.2692, "step": 695 }, { "epoch": 0.44, "learning_rate": 8.965755230207636e-06, "loss": 0.2356, "step": 696 }, { "epoch": 0.44, "learning_rate": 8.962718190502178e-06, "loss": 0.2828, "step": 697 }, { "epoch": 0.44, "learning_rate": 8.959677214204043e-06, "loss": 0.2906, "step": 698 }, { "epoch": 0.44, "learning_rate": 8.956632304334158e-06, "loss": 0.2388, "step": 699 }, { "epoch": 0.44, "learning_rate": 8.953583463917359e-06, "loss": 0.2629, "step": 700 }, { "epoch": 0.44, "learning_rate": 8.950530695982386e-06, "loss": 0.2635, "step": 701 }, { "epoch": 0.44, "learning_rate": 8.94747400356188e-06, "loss": 0.275, "step": 702 }, { "epoch": 0.44, "learning_rate": 8.94441338969238e-06, "loss": 0.256, "step": 703 }, { "epoch": 0.44, "learning_rate": 8.941348857414324e-06, "loss": 0.2334, "step": 704 }, { "epoch": 0.44, "learning_rate": 8.938280409772038e-06, "loss": 0.2364, "step": 705 }, { "epoch": 0.44, "learning_rate": 8.935208049813744e-06, "loss": 0.2405, "step": 706 }, { "epoch": 0.44, "learning_rate": 8.932131780591542e-06, "loss": 0.2564, "step": 707 }, { "epoch": 0.44, "learning_rate": 8.92905160516142e-06, "loss": 0.2353, "step": 708 }, { "epoch": 0.45, "learning_rate": 8.925967526583252e-06, "loss": 0.2579, "step": 709 }, { "epoch": 0.45, "learning_rate": 8.922879547920777e-06, "loss": 0.2597, "step": 710 }, { "epoch": 0.45, "learning_rate": 8.919787672241619e-06, "loss": 0.2856, "step": 711 }, { "epoch": 0.45, "learning_rate": 8.916691902617268e-06, "loss": 0.2447, "step": 712 }, { "epoch": 0.45, "learning_rate": 8.913592242123084e-06, "loss": 0.2752, "step": 713 }, { "epoch": 0.45, "learning_rate": 8.910488693838292e-06, "loss": 0.307, "step": 714 }, { "epoch": 0.45, "learning_rate": 8.90738126084598e-06, "loss": 0.2319, "step": 715 }, { "epoch": 0.45, "learning_rate": 8.904269946233093e-06, "loss": 0.2488, "step": 716 }, { "epoch": 0.45, "learning_rate": 8.901154753090434e-06, "loss": 0.2345, "step": 717 }, { "epoch": 0.45, "learning_rate": 8.898035684512655e-06, "loss": 0.2418, "step": 718 }, { "epoch": 0.45, "learning_rate": 8.894912743598269e-06, "loss": 0.243, "step": 719 }, { "epoch": 0.45, "learning_rate": 8.891785933449621e-06, "loss": 0.2474, "step": 720 }, { "epoch": 0.45, "learning_rate": 8.888655257172908e-06, "loss": 0.2628, "step": 721 }, { "epoch": 0.45, "learning_rate": 8.885520717878166e-06, "loss": 0.2503, "step": 722 }, { "epoch": 0.45, "learning_rate": 8.882382318679272e-06, "loss": 0.2554, "step": 723 }, { "epoch": 0.45, "learning_rate": 8.87924006269393e-06, "loss": 0.2367, "step": 724 }, { "epoch": 0.46, "learning_rate": 8.876093953043683e-06, "loss": 0.2385, "step": 725 }, { "epoch": 0.46, "learning_rate": 8.872943992853898e-06, "loss": 0.2278, "step": 726 }, { "epoch": 0.46, "learning_rate": 8.869790185253766e-06, "loss": 0.279, "step": 727 }, { "epoch": 0.46, "learning_rate": 8.866632533376303e-06, "loss": 0.2622, "step": 728 }, { "epoch": 0.46, "learning_rate": 8.863471040358347e-06, "loss": 0.2527, "step": 729 }, { "epoch": 0.46, "learning_rate": 8.860305709340541e-06, "loss": 0.249, "step": 730 }, { "epoch": 0.46, "learning_rate": 8.857136543467353e-06, "loss": 0.2712, "step": 731 }, { "epoch": 0.46, "learning_rate": 8.853963545887052e-06, "loss": 0.2407, "step": 732 }, { "epoch": 0.46, "learning_rate": 8.850786719751719e-06, "loss": 0.2471, "step": 733 }, { "epoch": 0.46, "learning_rate": 8.847606068217235e-06, "loss": 0.2283, "step": 734 }, { "epoch": 0.46, "learning_rate": 8.84442159444328e-06, "loss": 0.2344, "step": 735 }, { "epoch": 0.46, "learning_rate": 8.841233301593334e-06, "loss": 0.2516, "step": 736 }, { "epoch": 0.46, "learning_rate": 8.838041192834672e-06, "loss": 0.2129, "step": 737 }, { "epoch": 0.46, "learning_rate": 8.834845271338357e-06, "loss": 0.2312, "step": 738 }, { "epoch": 0.46, "learning_rate": 8.831645540279239e-06, "loss": 0.2316, "step": 739 }, { "epoch": 0.46, "learning_rate": 8.828442002835956e-06, "loss": 0.2597, "step": 740 }, { "epoch": 0.47, "learning_rate": 8.825234662190922e-06, "loss": 0.2432, "step": 741 }, { "epoch": 0.47, "learning_rate": 8.822023521530335e-06, "loss": 0.2234, "step": 742 }, { "epoch": 0.47, "learning_rate": 8.818808584044163e-06, "loss": 0.2325, "step": 743 }, { "epoch": 0.47, "learning_rate": 8.81558985292615e-06, "loss": 0.2521, "step": 744 }, { "epoch": 0.47, "learning_rate": 8.812367331373806e-06, "loss": 0.2473, "step": 745 }, { "epoch": 0.47, "learning_rate": 8.809141022588406e-06, "loss": 0.2857, "step": 746 }, { "epoch": 0.47, "learning_rate": 8.805910929774989e-06, "loss": 0.2755, "step": 747 }, { "epoch": 0.47, "learning_rate": 8.80267705614235e-06, "loss": 0.2159, "step": 748 }, { "epoch": 0.47, "learning_rate": 8.799439404903047e-06, "loss": 0.2505, "step": 749 }, { "epoch": 0.47, "learning_rate": 8.796197979273383e-06, "loss": 0.2731, "step": 750 }, { "epoch": 0.47, "learning_rate": 8.792952782473415e-06, "loss": 0.2657, "step": 751 }, { "epoch": 0.47, "learning_rate": 8.789703817726942e-06, "loss": 0.2211, "step": 752 }, { "epoch": 0.47, "learning_rate": 8.786451088261511e-06, "loss": 0.264, "step": 753 }, { "epoch": 0.47, "learning_rate": 8.783194597308408e-06, "loss": 0.2726, "step": 754 }, { "epoch": 0.47, "learning_rate": 8.779934348102652e-06, "loss": 0.2216, "step": 755 }, { "epoch": 0.47, "learning_rate": 8.776670343882998e-06, "loss": 0.2757, "step": 756 }, { "epoch": 0.48, "learning_rate": 8.773402587891933e-06, "loss": 0.2269, "step": 757 }, { "epoch": 0.48, "learning_rate": 8.770131083375668e-06, "loss": 0.2452, "step": 758 }, { "epoch": 0.48, "learning_rate": 8.76685583358414e-06, "loss": 0.2228, "step": 759 }, { "epoch": 0.48, "learning_rate": 8.763576841771004e-06, "loss": 0.2585, "step": 760 }, { "epoch": 0.48, "learning_rate": 8.760294111193635e-06, "loss": 0.2072, "step": 761 }, { "epoch": 0.48, "learning_rate": 8.757007645113122e-06, "loss": 0.2318, "step": 762 }, { "epoch": 0.48, "learning_rate": 8.753717446794262e-06, "loss": 0.2278, "step": 763 }, { "epoch": 0.48, "learning_rate": 8.750423519505564e-06, "loss": 0.2163, "step": 764 }, { "epoch": 0.48, "learning_rate": 8.747125866519236e-06, "loss": 0.2314, "step": 765 }, { "epoch": 0.48, "learning_rate": 8.743824491111195e-06, "loss": 0.2311, "step": 766 }, { "epoch": 0.48, "learning_rate": 8.740519396561045e-06, "loss": 0.2581, "step": 767 }, { "epoch": 0.48, "learning_rate": 8.737210586152094e-06, "loss": 0.2583, "step": 768 }, { "epoch": 0.48, "learning_rate": 8.733898063171338e-06, "loss": 0.2541, "step": 769 }, { "epoch": 0.48, "learning_rate": 8.730581830909459e-06, "loss": 0.261, "step": 770 }, { "epoch": 0.48, "learning_rate": 8.72726189266083e-06, "loss": 0.227, "step": 771 }, { "epoch": 0.48, "learning_rate": 8.723938251723498e-06, "loss": 0.2302, "step": 772 }, { "epoch": 0.49, "learning_rate": 8.720610911399193e-06, "loss": 0.2364, "step": 773 }, { "epoch": 0.49, "learning_rate": 8.717279874993316e-06, "loss": 0.2643, "step": 774 }, { "epoch": 0.49, "learning_rate": 8.713945145814948e-06, "loss": 0.2143, "step": 775 }, { "epoch": 0.49, "learning_rate": 8.710606727176829e-06, "loss": 0.2209, "step": 776 }, { "epoch": 0.49, "learning_rate": 8.707264622395366e-06, "loss": 0.224, "step": 777 }, { "epoch": 0.49, "learning_rate": 8.703918834790637e-06, "loss": 0.2126, "step": 778 }, { "epoch": 0.49, "learning_rate": 8.700569367686365e-06, "loss": 0.2536, "step": 779 }, { "epoch": 0.49, "learning_rate": 8.697216224409937e-06, "loss": 0.2244, "step": 780 }, { "epoch": 0.49, "learning_rate": 8.693859408292387e-06, "loss": 0.2541, "step": 781 }, { "epoch": 0.49, "learning_rate": 8.690498922668406e-06, "loss": 0.2698, "step": 782 }, { "epoch": 0.49, "learning_rate": 8.68713477087632e-06, "loss": 0.2282, "step": 783 }, { "epoch": 0.49, "learning_rate": 8.683766956258102e-06, "loss": 0.2403, "step": 784 }, { "epoch": 0.49, "learning_rate": 8.680395482159364e-06, "loss": 0.242, "step": 785 }, { "epoch": 0.49, "learning_rate": 8.677020351929353e-06, "loss": 0.2136, "step": 786 }, { "epoch": 0.49, "learning_rate": 8.673641568920944e-06, "loss": 0.258, "step": 787 }, { "epoch": 0.49, "learning_rate": 8.670259136490652e-06, "loss": 0.2493, "step": 788 }, { "epoch": 0.5, "learning_rate": 8.666873057998602e-06, "loss": 0.2571, "step": 789 }, { "epoch": 0.5, "learning_rate": 8.663483336808551e-06, "loss": 0.2574, "step": 790 }, { "epoch": 0.5, "learning_rate": 8.660089976287875e-06, "loss": 0.2383, "step": 791 }, { "epoch": 0.5, "learning_rate": 8.65669297980756e-06, "loss": 0.2378, "step": 792 }, { "epoch": 0.5, "learning_rate": 8.653292350742207e-06, "loss": 0.2407, "step": 793 }, { "epoch": 0.5, "learning_rate": 8.649888092470025e-06, "loss": 0.2501, "step": 794 }, { "epoch": 0.5, "learning_rate": 8.646480208372829e-06, "loss": 0.2197, "step": 795 }, { "epoch": 0.5, "learning_rate": 8.643068701836034e-06, "loss": 0.2429, "step": 796 }, { "epoch": 0.5, "learning_rate": 8.639653576248658e-06, "loss": 0.2283, "step": 797 }, { "epoch": 0.5, "learning_rate": 8.636234835003313e-06, "loss": 0.2458, "step": 798 }, { "epoch": 0.5, "learning_rate": 8.632812481496195e-06, "loss": 0.2499, "step": 799 }, { "epoch": 0.5, "learning_rate": 8.629386519127097e-06, "loss": 0.2332, "step": 800 }, { "epoch": 0.5, "learning_rate": 8.625956951299398e-06, "loss": 0.2174, "step": 801 }, { "epoch": 0.5, "learning_rate": 8.62252378142005e-06, "loss": 0.235, "step": 802 }, { "epoch": 0.5, "learning_rate": 8.619087012899591e-06, "loss": 0.2268, "step": 803 }, { "epoch": 0.5, "learning_rate": 8.615646649152132e-06, "loss": 0.2203, "step": 804 }, { "epoch": 0.51, "learning_rate": 8.612202693595356e-06, "loss": 0.2396, "step": 805 }, { "epoch": 0.51, "learning_rate": 8.60875514965051e-06, "loss": 0.2385, "step": 806 }, { "epoch": 0.51, "learning_rate": 8.60530402074241e-06, "loss": 0.2826, "step": 807 }, { "epoch": 0.51, "learning_rate": 8.601849310299435e-06, "loss": 0.2469, "step": 808 }, { "epoch": 0.51, "learning_rate": 8.598391021753515e-06, "loss": 0.2082, "step": 809 }, { "epoch": 0.51, "learning_rate": 8.59492915854014e-06, "loss": 0.2456, "step": 810 }, { "epoch": 0.51, "learning_rate": 8.591463724098352e-06, "loss": 0.2235, "step": 811 }, { "epoch": 0.51, "learning_rate": 8.587994721870736e-06, "loss": 0.253, "step": 812 }, { "epoch": 0.51, "learning_rate": 8.584522155303426e-06, "loss": 0.2379, "step": 813 }, { "epoch": 0.51, "learning_rate": 8.58104602784609e-06, "loss": 0.2355, "step": 814 }, { "epoch": 0.51, "learning_rate": 8.577566342951944e-06, "loss": 0.2445, "step": 815 }, { "epoch": 0.51, "learning_rate": 8.574083104077727e-06, "loss": 0.2138, "step": 816 }, { "epoch": 0.51, "learning_rate": 8.570596314683715e-06, "loss": 0.2816, "step": 817 }, { "epoch": 0.51, "learning_rate": 8.567105978233712e-06, "loss": 0.2298, "step": 818 }, { "epoch": 0.51, "learning_rate": 8.56361209819504e-06, "loss": 0.2196, "step": 819 }, { "epoch": 0.51, "learning_rate": 8.560114678038547e-06, "loss": 0.2301, "step": 820 }, { "epoch": 0.52, "learning_rate": 8.556613721238591e-06, "loss": 0.2432, "step": 821 }, { "epoch": 0.52, "learning_rate": 8.553109231273053e-06, "loss": 0.2252, "step": 822 }, { "epoch": 0.52, "learning_rate": 8.549601211623316e-06, "loss": 0.1991, "step": 823 }, { "epoch": 0.52, "learning_rate": 8.546089665774272e-06, "loss": 0.2533, "step": 824 }, { "epoch": 0.52, "learning_rate": 8.542574597214314e-06, "loss": 0.2578, "step": 825 }, { "epoch": 0.52, "learning_rate": 8.53905600943534e-06, "loss": 0.234, "step": 826 }, { "epoch": 0.52, "learning_rate": 8.535533905932739e-06, "loss": 0.2546, "step": 827 }, { "epoch": 0.52, "learning_rate": 8.532008290205391e-06, "loss": 0.1968, "step": 828 }, { "epoch": 0.52, "learning_rate": 8.528479165755671e-06, "loss": 0.2464, "step": 829 }, { "epoch": 0.52, "learning_rate": 8.524946536089437e-06, "loss": 0.2554, "step": 830 }, { "epoch": 0.52, "learning_rate": 8.521410404716029e-06, "loss": 0.2131, "step": 831 }, { "epoch": 0.52, "learning_rate": 8.517870775148261e-06, "loss": 0.2474, "step": 832 }, { "epoch": 0.52, "learning_rate": 8.514327650902429e-06, "loss": 0.2207, "step": 833 }, { "epoch": 0.52, "learning_rate": 8.510781035498298e-06, "loss": 0.2763, "step": 834 }, { "epoch": 0.52, "learning_rate": 8.507230932459102e-06, "loss": 0.1969, "step": 835 }, { "epoch": 0.53, "learning_rate": 8.503677345311541e-06, "loss": 0.2588, "step": 836 }, { "epoch": 0.53, "learning_rate": 8.500120277585768e-06, "loss": 0.222, "step": 837 }, { "epoch": 0.53, "learning_rate": 8.496559732815403e-06, "loss": 0.223, "step": 838 }, { "epoch": 0.53, "learning_rate": 8.492995714537519e-06, "loss": 0.2484, "step": 839 }, { "epoch": 0.53, "learning_rate": 8.489428226292633e-06, "loss": 0.2321, "step": 840 }, { "epoch": 0.53, "learning_rate": 8.485857271624715e-06, "loss": 0.2335, "step": 841 }, { "epoch": 0.53, "learning_rate": 8.482282854081177e-06, "loss": 0.2308, "step": 842 }, { "epoch": 0.53, "learning_rate": 8.478704977212873e-06, "loss": 0.2304, "step": 843 }, { "epoch": 0.53, "learning_rate": 8.475123644574088e-06, "loss": 0.2149, "step": 844 }, { "epoch": 0.53, "learning_rate": 8.471538859722545e-06, "loss": 0.2196, "step": 845 }, { "epoch": 0.53, "learning_rate": 8.467950626219397e-06, "loss": 0.2348, "step": 846 }, { "epoch": 0.53, "learning_rate": 8.46435894762922e-06, "loss": 0.2682, "step": 847 }, { "epoch": 0.53, "learning_rate": 8.46076382752001e-06, "loss": 0.2564, "step": 848 }, { "epoch": 0.53, "learning_rate": 8.457165269463188e-06, "loss": 0.2567, "step": 849 }, { "epoch": 0.53, "learning_rate": 8.453563277033588e-06, "loss": 0.2378, "step": 850 }, { "epoch": 0.53, "learning_rate": 8.449957853809453e-06, "loss": 0.2672, "step": 851 }, { "epoch": 0.54, "learning_rate": 8.446349003372438e-06, "loss": 0.2224, "step": 852 }, { "epoch": 0.54, "learning_rate": 8.442736729307602e-06, "loss": 0.1859, "step": 853 }, { "epoch": 0.54, "learning_rate": 8.439121035203398e-06, "loss": 0.2365, "step": 854 }, { "epoch": 0.54, "learning_rate": 8.43550192465169e-06, "loss": 0.2295, "step": 855 }, { "epoch": 0.54, "learning_rate": 8.431879401247726e-06, "loss": 0.2076, "step": 856 }, { "epoch": 0.54, "learning_rate": 8.428253468590144e-06, "loss": 0.2248, "step": 857 }, { "epoch": 0.54, "learning_rate": 8.424624130280973e-06, "loss": 0.2577, "step": 858 }, { "epoch": 0.54, "learning_rate": 8.420991389925627e-06, "loss": 0.2177, "step": 859 }, { "epoch": 0.54, "learning_rate": 8.41735525113289e-06, "loss": 0.2309, "step": 860 }, { "epoch": 0.54, "learning_rate": 8.413715717514935e-06, "loss": 0.2392, "step": 861 }, { "epoch": 0.54, "learning_rate": 8.410072792687294e-06, "loss": 0.2202, "step": 862 }, { "epoch": 0.54, "learning_rate": 8.406426480268881e-06, "loss": 0.2351, "step": 863 }, { "epoch": 0.54, "learning_rate": 8.402776783881965e-06, "loss": 0.2152, "step": 864 }, { "epoch": 0.54, "learning_rate": 8.399123707152182e-06, "loss": 0.209, "step": 865 }, { "epoch": 0.54, "learning_rate": 8.395467253708525e-06, "loss": 0.2125, "step": 866 }, { "epoch": 0.54, "learning_rate": 8.39180742718334e-06, "loss": 0.2442, "step": 867 }, { "epoch": 0.55, "learning_rate": 8.388144231212324e-06, "loss": 0.2552, "step": 868 }, { "epoch": 0.55, "learning_rate": 8.384477669434526e-06, "loss": 0.2625, "step": 869 }, { "epoch": 0.55, "learning_rate": 8.38080774549233e-06, "loss": 0.2288, "step": 870 }, { "epoch": 0.55, "learning_rate": 8.377134463031468e-06, "loss": 0.1981, "step": 871 }, { "epoch": 0.55, "learning_rate": 8.373457825701003e-06, "loss": 0.2581, "step": 872 }, { "epoch": 0.55, "learning_rate": 8.369777837153337e-06, "loss": 0.2388, "step": 873 }, { "epoch": 0.55, "learning_rate": 8.366094501044193e-06, "loss": 0.2513, "step": 874 }, { "epoch": 0.55, "learning_rate": 8.362407821032623e-06, "loss": 0.2266, "step": 875 }, { "epoch": 0.55, "learning_rate": 8.358717800781004e-06, "loss": 0.2535, "step": 876 }, { "epoch": 0.55, "learning_rate": 8.355024443955029e-06, "loss": 0.2283, "step": 877 }, { "epoch": 0.55, "learning_rate": 8.351327754223701e-06, "loss": 0.2318, "step": 878 }, { "epoch": 0.55, "learning_rate": 8.347627735259344e-06, "loss": 0.2088, "step": 879 }, { "epoch": 0.55, "learning_rate": 8.343924390737578e-06, "loss": 0.2082, "step": 880 }, { "epoch": 0.55, "learning_rate": 8.340217724337337e-06, "loss": 0.1668, "step": 881 }, { "epoch": 0.55, "learning_rate": 8.336507739740844e-06, "loss": 0.2595, "step": 882 }, { "epoch": 0.55, "learning_rate": 8.33279444063363e-06, "loss": 0.2465, "step": 883 }, { "epoch": 0.56, "learning_rate": 8.329077830704508e-06, "loss": 0.2536, "step": 884 }, { "epoch": 0.56, "learning_rate": 8.325357913645589e-06, "loss": 0.231, "step": 885 }, { "epoch": 0.56, "learning_rate": 8.321634693152263e-06, "loss": 0.2102, "step": 886 }, { "epoch": 0.56, "learning_rate": 8.317908172923207e-06, "loss": 0.2274, "step": 887 }, { "epoch": 0.56, "learning_rate": 8.314178356660369e-06, "loss": 0.2045, "step": 888 }, { "epoch": 0.56, "learning_rate": 8.310445248068976e-06, "loss": 0.2444, "step": 889 }, { "epoch": 0.56, "learning_rate": 8.306708850857528e-06, "loss": 0.2271, "step": 890 }, { "epoch": 0.56, "learning_rate": 8.302969168737784e-06, "loss": 0.2426, "step": 891 }, { "epoch": 0.56, "learning_rate": 8.299226205424774e-06, "loss": 0.2752, "step": 892 }, { "epoch": 0.56, "learning_rate": 8.295479964636787e-06, "loss": 0.2644, "step": 893 }, { "epoch": 0.56, "learning_rate": 8.29173045009536e-06, "loss": 0.229, "step": 894 }, { "epoch": 0.56, "learning_rate": 8.287977665525292e-06, "loss": 0.2124, "step": 895 }, { "epoch": 0.56, "learning_rate": 8.284221614654624e-06, "loss": 0.2309, "step": 896 }, { "epoch": 0.56, "learning_rate": 8.280462301214644e-06, "loss": 0.2296, "step": 897 }, { "epoch": 0.56, "learning_rate": 8.27669972893988e-06, "loss": 0.2014, "step": 898 }, { "epoch": 0.56, "learning_rate": 8.272933901568102e-06, "loss": 0.1838, "step": 899 }, { "epoch": 0.57, "learning_rate": 8.269164822840306e-06, "loss": 0.2237, "step": 900 }, { "epoch": 0.57, "learning_rate": 8.265392496500724e-06, "loss": 0.2612, "step": 901 }, { "epoch": 0.57, "learning_rate": 8.261616926296811e-06, "loss": 0.2144, "step": 902 }, { "epoch": 0.57, "learning_rate": 8.257838115979244e-06, "loss": 0.26, "step": 903 }, { "epoch": 0.57, "learning_rate": 8.254056069301926e-06, "loss": 0.2177, "step": 904 }, { "epoch": 0.57, "learning_rate": 8.25027079002196e-06, "loss": 0.2142, "step": 905 }, { "epoch": 0.57, "learning_rate": 8.24648228189968e-06, "loss": 0.2218, "step": 906 }, { "epoch": 0.57, "learning_rate": 8.242690548698611e-06, "loss": 0.2672, "step": 907 }, { "epoch": 0.57, "learning_rate": 8.238895594185489e-06, "loss": 0.2221, "step": 908 }, { "epoch": 0.57, "learning_rate": 8.235097422130249e-06, "loss": 0.2073, "step": 909 }, { "epoch": 0.57, "learning_rate": 8.231296036306022e-06, "loss": 0.228, "step": 910 }, { "epoch": 0.57, "learning_rate": 8.227491440489134e-06, "loss": 0.2129, "step": 911 }, { "epoch": 0.57, "learning_rate": 8.223683638459094e-06, "loss": 0.2168, "step": 912 }, { "epoch": 0.57, "learning_rate": 8.219872633998605e-06, "loss": 0.2633, "step": 913 }, { "epoch": 0.57, "learning_rate": 8.21605843089354e-06, "loss": 0.2217, "step": 914 }, { "epoch": 0.57, "learning_rate": 8.212241032932962e-06, "loss": 0.2507, "step": 915 }, { "epoch": 0.58, "learning_rate": 8.208420443909097e-06, "loss": 0.1949, "step": 916 }, { "epoch": 0.58, "learning_rate": 8.204596667617349e-06, "loss": 0.2298, "step": 917 }, { "epoch": 0.58, "learning_rate": 8.20076970785628e-06, "loss": 0.1981, "step": 918 }, { "epoch": 0.58, "learning_rate": 8.196939568427624e-06, "loss": 0.2126, "step": 919 }, { "epoch": 0.58, "learning_rate": 8.193106253136267e-06, "loss": 0.2512, "step": 920 }, { "epoch": 0.58, "learning_rate": 8.189269765790251e-06, "loss": 0.1981, "step": 921 }, { "epoch": 0.58, "learning_rate": 8.185430110200774e-06, "loss": 0.2074, "step": 922 }, { "epoch": 0.58, "learning_rate": 8.18158729018217e-06, "loss": 0.2426, "step": 923 }, { "epoch": 0.58, "learning_rate": 8.177741309551933e-06, "loss": 0.2417, "step": 924 }, { "epoch": 0.58, "learning_rate": 8.173892172130683e-06, "loss": 0.2341, "step": 925 }, { "epoch": 0.58, "learning_rate": 8.170039881742179e-06, "loss": 0.2215, "step": 926 }, { "epoch": 0.58, "learning_rate": 8.166184442213314e-06, "loss": 0.2199, "step": 927 }, { "epoch": 0.58, "learning_rate": 8.162325857374113e-06, "loss": 0.2422, "step": 928 }, { "epoch": 0.58, "learning_rate": 8.158464131057721e-06, "loss": 0.1998, "step": 929 }, { "epoch": 0.58, "learning_rate": 8.154599267100404e-06, "loss": 0.218, "step": 930 }, { "epoch": 0.58, "learning_rate": 8.150731269341544e-06, "loss": 0.2249, "step": 931 }, { "epoch": 0.59, "learning_rate": 8.146860141623638e-06, "loss": 0.2664, "step": 932 }, { "epoch": 0.59, "learning_rate": 8.142985887792293e-06, "loss": 0.2096, "step": 933 }, { "epoch": 0.59, "learning_rate": 8.139108511696224e-06, "loss": 0.2263, "step": 934 }, { "epoch": 0.59, "learning_rate": 8.135228017187238e-06, "loss": 0.2507, "step": 935 }, { "epoch": 0.59, "learning_rate": 8.13134440812025e-06, "loss": 0.2083, "step": 936 }, { "epoch": 0.59, "learning_rate": 8.127457688353263e-06, "loss": 0.1922, "step": 937 }, { "epoch": 0.59, "learning_rate": 8.123567861747374e-06, "loss": 0.2083, "step": 938 }, { "epoch": 0.59, "learning_rate": 8.119674932166763e-06, "loss": 0.2594, "step": 939 }, { "epoch": 0.59, "learning_rate": 8.115778903478695e-06, "loss": 0.2192, "step": 940 }, { "epoch": 0.59, "learning_rate": 8.111879779553514e-06, "loss": 0.1964, "step": 941 }, { "epoch": 0.59, "learning_rate": 8.107977564264636e-06, "loss": 0.2033, "step": 942 }, { "epoch": 0.59, "learning_rate": 8.10407226148855e-06, "loss": 0.2505, "step": 943 }, { "epoch": 0.59, "learning_rate": 8.100163875104815e-06, "loss": 0.2106, "step": 944 }, { "epoch": 0.59, "learning_rate": 8.096252408996043e-06, "loss": 0.2111, "step": 945 }, { "epoch": 0.59, "learning_rate": 8.092337867047919e-06, "loss": 0.2184, "step": 946 }, { "epoch": 0.59, "learning_rate": 8.088420253149173e-06, "loss": 0.22, "step": 947 }, { "epoch": 0.6, "learning_rate": 8.084499571191593e-06, "loss": 0.2492, "step": 948 }, { "epoch": 0.6, "learning_rate": 8.080575825070013e-06, "loss": 0.2107, "step": 949 }, { "epoch": 0.6, "learning_rate": 8.076649018682308e-06, "loss": 0.2089, "step": 950 }, { "epoch": 0.6, "learning_rate": 8.0727191559294e-06, "loss": 0.1996, "step": 951 }, { "epoch": 0.6, "learning_rate": 8.068786240715237e-06, "loss": 0.2133, "step": 952 }, { "epoch": 0.6, "learning_rate": 8.06485027694681e-06, "loss": 0.2465, "step": 953 }, { "epoch": 0.6, "learning_rate": 8.060911268534131e-06, "loss": 0.2065, "step": 954 }, { "epoch": 0.6, "learning_rate": 8.05696921939024e-06, "loss": 0.245, "step": 955 }, { "epoch": 0.6, "learning_rate": 8.053024133431197e-06, "loss": 0.236, "step": 956 }, { "epoch": 0.6, "learning_rate": 8.04907601457608e-06, "loss": 0.1974, "step": 957 }, { "epoch": 0.6, "learning_rate": 8.045124866746976e-06, "loss": 0.2316, "step": 958 }, { "epoch": 0.6, "learning_rate": 8.041170693868985e-06, "loss": 0.184, "step": 959 }, { "epoch": 0.6, "learning_rate": 8.037213499870211e-06, "loss": 0.2293, "step": 960 }, { "epoch": 0.6, "learning_rate": 8.033253288681759e-06, "loss": 0.2469, "step": 961 }, { "epoch": 0.6, "learning_rate": 8.029290064237729e-06, "loss": 0.2437, "step": 962 }, { "epoch": 0.6, "learning_rate": 8.02532383047522e-06, "loss": 0.215, "step": 963 }, { "epoch": 0.61, "learning_rate": 8.021354591334314e-06, "loss": 0.224, "step": 964 }, { "epoch": 0.61, "learning_rate": 8.017382350758085e-06, "loss": 0.2287, "step": 965 }, { "epoch": 0.61, "learning_rate": 8.013407112692581e-06, "loss": 0.2276, "step": 966 }, { "epoch": 0.61, "learning_rate": 8.009428881086836e-06, "loss": 0.2308, "step": 967 }, { "epoch": 0.61, "learning_rate": 8.00544765989285e-06, "loss": 0.2195, "step": 968 }, { "epoch": 0.61, "learning_rate": 8.001463453065601e-06, "loss": 0.2034, "step": 969 }, { "epoch": 0.61, "learning_rate": 7.997476264563025e-06, "loss": 0.2148, "step": 970 }, { "epoch": 0.61, "learning_rate": 7.993486098346026e-06, "loss": 0.212, "step": 971 }, { "epoch": 0.61, "learning_rate": 7.989492958378462e-06, "loss": 0.2001, "step": 972 }, { "epoch": 0.61, "learning_rate": 7.985496848627147e-06, "loss": 0.1906, "step": 973 }, { "epoch": 0.61, "learning_rate": 7.981497773061848e-06, "loss": 0.2259, "step": 974 }, { "epoch": 0.61, "learning_rate": 7.977495735655271e-06, "loss": 0.2168, "step": 975 }, { "epoch": 0.61, "learning_rate": 7.973490740383073e-06, "loss": 0.2481, "step": 976 }, { "epoch": 0.61, "learning_rate": 7.969482791223842e-06, "loss": 0.2222, "step": 977 }, { "epoch": 0.61, "learning_rate": 7.965471892159104e-06, "loss": 0.225, "step": 978 }, { "epoch": 0.61, "learning_rate": 7.961458047173316e-06, "loss": 0.2223, "step": 979 }, { "epoch": 0.62, "learning_rate": 7.957441260253858e-06, "loss": 0.1949, "step": 980 }, { "epoch": 0.62, "learning_rate": 7.953421535391039e-06, "loss": 0.2203, "step": 981 }, { "epoch": 0.62, "learning_rate": 7.949398876578078e-06, "loss": 0.2166, "step": 982 }, { "epoch": 0.62, "learning_rate": 7.945373287811116e-06, "loss": 0.203, "step": 983 }, { "epoch": 0.62, "learning_rate": 7.9413447730892e-06, "loss": 0.2342, "step": 984 }, { "epoch": 0.62, "learning_rate": 7.937313336414287e-06, "loss": 0.2032, "step": 985 }, { "epoch": 0.62, "learning_rate": 7.933278981791234e-06, "loss": 0.2453, "step": 986 }, { "epoch": 0.62, "learning_rate": 7.9292417132278e-06, "loss": 0.1915, "step": 987 }, { "epoch": 0.62, "learning_rate": 7.925201534734635e-06, "loss": 0.1815, "step": 988 }, { "epoch": 0.62, "learning_rate": 7.92115845032528e-06, "loss": 0.2128, "step": 989 }, { "epoch": 0.62, "learning_rate": 7.917112464016165e-06, "loss": 0.2225, "step": 990 }, { "epoch": 0.62, "learning_rate": 7.913063579826601e-06, "loss": 0.221, "step": 991 }, { "epoch": 0.62, "learning_rate": 7.90901180177878e-06, "loss": 0.2229, "step": 992 }, { "epoch": 0.62, "learning_rate": 7.904957133897771e-06, "loss": 0.2177, "step": 993 }, { "epoch": 0.62, "learning_rate": 7.900899580211502e-06, "loss": 0.2144, "step": 994 }, { "epoch": 0.62, "learning_rate": 7.896839144750782e-06, "loss": 0.2352, "step": 995 }, { "epoch": 0.63, "learning_rate": 7.892775831549273e-06, "loss": 0.2378, "step": 996 }, { "epoch": 0.63, "learning_rate": 7.888709644643504e-06, "loss": 0.2247, "step": 997 }, { "epoch": 0.63, "learning_rate": 7.88464058807285e-06, "loss": 0.2123, "step": 998 }, { "epoch": 0.63, "learning_rate": 7.880568665879542e-06, "loss": 0.2081, "step": 999 }, { "epoch": 0.63, "learning_rate": 7.876493882108659e-06, "loss": 0.2308, "step": 1000 }, { "epoch": 0.63, "learning_rate": 7.872416240808118e-06, "loss": 0.2602, "step": 1001 }, { "epoch": 0.63, "learning_rate": 7.868335746028678e-06, "loss": 0.2267, "step": 1002 }, { "epoch": 0.63, "learning_rate": 7.864252401823928e-06, "loss": 0.2136, "step": 1003 }, { "epoch": 0.63, "learning_rate": 7.8601662122503e-06, "loss": 0.2091, "step": 1004 }, { "epoch": 0.63, "learning_rate": 7.856077181367036e-06, "loss": 0.2394, "step": 1005 }, { "epoch": 0.63, "learning_rate": 7.851985313236212e-06, "loss": 0.1956, "step": 1006 }, { "epoch": 0.63, "learning_rate": 7.847890611922721e-06, "loss": 0.2305, "step": 1007 }, { "epoch": 0.63, "learning_rate": 7.843793081494265e-06, "loss": 0.2031, "step": 1008 }, { "epoch": 0.63, "learning_rate": 7.839692726021362e-06, "loss": 0.2525, "step": 1009 }, { "epoch": 0.63, "learning_rate": 7.835589549577334e-06, "loss": 0.2226, "step": 1010 }, { "epoch": 0.63, "learning_rate": 7.831483556238306e-06, "loss": 0.2015, "step": 1011 }, { "epoch": 0.64, "learning_rate": 7.8273747500832e-06, "loss": 0.2264, "step": 1012 }, { "epoch": 0.64, "learning_rate": 7.823263135193737e-06, "loss": 0.2169, "step": 1013 }, { "epoch": 0.64, "learning_rate": 7.819148715654423e-06, "loss": 0.1996, "step": 1014 }, { "epoch": 0.64, "learning_rate": 7.81503149555255e-06, "loss": 0.2309, "step": 1015 }, { "epoch": 0.64, "learning_rate": 7.810911478978196e-06, "loss": 0.2355, "step": 1016 }, { "epoch": 0.64, "learning_rate": 7.806788670024212e-06, "loss": 0.1969, "step": 1017 }, { "epoch": 0.64, "learning_rate": 7.80266307278623e-06, "loss": 0.2338, "step": 1018 }, { "epoch": 0.64, "learning_rate": 7.798534691362647e-06, "loss": 0.219, "step": 1019 }, { "epoch": 0.64, "learning_rate": 7.794403529854626e-06, "loss": 0.2368, "step": 1020 }, { "epoch": 0.64, "learning_rate": 7.790269592366094e-06, "loss": 0.2291, "step": 1021 }, { "epoch": 0.64, "learning_rate": 7.78613288300373e-06, "loss": 0.2187, "step": 1022 }, { "epoch": 0.64, "learning_rate": 7.781993405876974e-06, "loss": 0.237, "step": 1023 }, { "epoch": 0.64, "learning_rate": 7.777851165098012e-06, "loss": 0.2165, "step": 1024 }, { "epoch": 0.64, "learning_rate": 7.773706164781776e-06, "loss": 0.2, "step": 1025 }, { "epoch": 0.64, "learning_rate": 7.769558409045937e-06, "loss": 0.2087, "step": 1026 }, { "epoch": 0.64, "learning_rate": 7.76540790201091e-06, "loss": 0.2025, "step": 1027 }, { "epoch": 0.65, "learning_rate": 7.761254647799831e-06, "loss": 0.2106, "step": 1028 }, { "epoch": 0.65, "learning_rate": 7.75709865053858e-06, "loss": 0.2501, "step": 1029 }, { "epoch": 0.65, "learning_rate": 7.752939914355748e-06, "loss": 0.2211, "step": 1030 }, { "epoch": 0.65, "learning_rate": 7.748778443382658e-06, "loss": 0.1828, "step": 1031 }, { "epoch": 0.65, "learning_rate": 7.744614241753346e-06, "loss": 0.2233, "step": 1032 }, { "epoch": 0.65, "learning_rate": 7.740447313604555e-06, "loss": 0.2265, "step": 1033 }, { "epoch": 0.65, "learning_rate": 7.736277663075744e-06, "loss": 0.2077, "step": 1034 }, { "epoch": 0.65, "learning_rate": 7.732105294309074e-06, "loss": 0.1973, "step": 1035 }, { "epoch": 0.65, "learning_rate": 7.727930211449407e-06, "loss": 0.2351, "step": 1036 }, { "epoch": 0.65, "learning_rate": 7.723752418644295e-06, "loss": 0.2482, "step": 1037 }, { "epoch": 0.65, "learning_rate": 7.719571920043994e-06, "loss": 0.2157, "step": 1038 }, { "epoch": 0.65, "learning_rate": 7.715388719801437e-06, "loss": 0.2173, "step": 1039 }, { "epoch": 0.65, "learning_rate": 7.71120282207225e-06, "loss": 0.2108, "step": 1040 }, { "epoch": 0.65, "learning_rate": 7.707014231014727e-06, "loss": 0.2, "step": 1041 }, { "epoch": 0.65, "learning_rate": 7.70282295078985e-06, "loss": 0.2048, "step": 1042 }, { "epoch": 0.66, "learning_rate": 7.698628985561263e-06, "loss": 0.1611, "step": 1043 }, { "epoch": 0.66, "learning_rate": 7.694432339495284e-06, "loss": 0.1908, "step": 1044 }, { "epoch": 0.66, "learning_rate": 7.690233016760891e-06, "loss": 0.227, "step": 1045 }, { "epoch": 0.66, "learning_rate": 7.686031021529718e-06, "loss": 0.2243, "step": 1046 }, { "epoch": 0.66, "learning_rate": 7.68182635797606e-06, "loss": 0.2082, "step": 1047 }, { "epoch": 0.66, "learning_rate": 7.677619030276863e-06, "loss": 0.2169, "step": 1048 }, { "epoch": 0.66, "learning_rate": 7.673409042611712e-06, "loss": 0.2151, "step": 1049 }, { "epoch": 0.66, "learning_rate": 7.669196399162837e-06, "loss": 0.2012, "step": 1050 }, { "epoch": 0.66, "learning_rate": 7.664981104115115e-06, "loss": 0.198, "step": 1051 }, { "epoch": 0.66, "learning_rate": 7.660763161656045e-06, "loss": 0.1962, "step": 1052 }, { "epoch": 0.66, "learning_rate": 7.656542575975764e-06, "loss": 0.2087, "step": 1053 }, { "epoch": 0.66, "learning_rate": 7.652319351267028e-06, "loss": 0.2291, "step": 1054 }, { "epoch": 0.66, "learning_rate": 7.648093491725224e-06, "loss": 0.1942, "step": 1055 }, { "epoch": 0.66, "learning_rate": 7.64386500154835e-06, "loss": 0.2444, "step": 1056 }, { "epoch": 0.66, "learning_rate": 7.639633884937017e-06, "loss": 0.1964, "step": 1057 }, { "epoch": 0.66, "learning_rate": 7.635400146094445e-06, "loss": 0.1839, "step": 1058 }, { "epoch": 0.67, "learning_rate": 7.631163789226463e-06, "loss": 0.1979, "step": 1059 }, { "epoch": 0.67, "learning_rate": 7.626924818541499e-06, "loss": 0.228, "step": 1060 }, { "epoch": 0.67, "learning_rate": 7.6226832382505755e-06, "loss": 0.2062, "step": 1061 }, { "epoch": 0.67, "learning_rate": 7.618439052567308e-06, "loss": 0.2006, "step": 1062 }, { "epoch": 0.67, "learning_rate": 7.6141922657079045e-06, "loss": 0.2267, "step": 1063 }, { "epoch": 0.67, "learning_rate": 7.6099428818911504e-06, "loss": 0.1952, "step": 1064 }, { "epoch": 0.67, "learning_rate": 7.605690905338413e-06, "loss": 0.2433, "step": 1065 }, { "epoch": 0.67, "learning_rate": 7.601436340273638e-06, "loss": 0.219, "step": 1066 }, { "epoch": 0.67, "learning_rate": 7.597179190923343e-06, "loss": 0.2098, "step": 1067 }, { "epoch": 0.67, "learning_rate": 7.592919461516608e-06, "loss": 0.2277, "step": 1068 }, { "epoch": 0.67, "learning_rate": 7.588657156285077e-06, "loss": 0.2295, "step": 1069 }, { "epoch": 0.67, "learning_rate": 7.58439227946296e-06, "loss": 0.2178, "step": 1070 }, { "epoch": 0.67, "learning_rate": 7.580124835287013e-06, "loss": 0.2123, "step": 1071 }, { "epoch": 0.67, "learning_rate": 7.575854827996545e-06, "loss": 0.2139, "step": 1072 }, { "epoch": 0.67, "learning_rate": 7.571582261833412e-06, "loss": 0.2244, "step": 1073 }, { "epoch": 0.67, "learning_rate": 7.567307141042013e-06, "loss": 0.2474, "step": 1074 }, { "epoch": 0.68, "learning_rate": 7.563029469869283e-06, "loss": 0.2079, "step": 1075 }, { "epoch": 0.68, "learning_rate": 7.558749252564692e-06, "loss": 0.2018, "step": 1076 }, { "epoch": 0.68, "learning_rate": 7.55446649338024e-06, "loss": 0.1974, "step": 1077 }, { "epoch": 0.68, "learning_rate": 7.5501811965704455e-06, "loss": 0.2195, "step": 1078 }, { "epoch": 0.68, "learning_rate": 7.545893366392358e-06, "loss": 0.21, "step": 1079 }, { "epoch": 0.68, "learning_rate": 7.5416030071055356e-06, "loss": 0.2402, "step": 1080 }, { "epoch": 0.68, "learning_rate": 7.537310122972056e-06, "loss": 0.2317, "step": 1081 }, { "epoch": 0.68, "learning_rate": 7.533014718256496e-06, "loss": 0.225, "step": 1082 }, { "epoch": 0.68, "learning_rate": 7.5287167972259455e-06, "loss": 0.2193, "step": 1083 }, { "epoch": 0.68, "learning_rate": 7.524416364149989e-06, "loss": 0.2211, "step": 1084 }, { "epoch": 0.68, "learning_rate": 7.520113423300706e-06, "loss": 0.2019, "step": 1085 }, { "epoch": 0.68, "learning_rate": 7.5158079789526715e-06, "loss": 0.2163, "step": 1086 }, { "epoch": 0.68, "learning_rate": 7.511500035382943e-06, "loss": 0.1851, "step": 1087 }, { "epoch": 0.68, "learning_rate": 7.507189596871061e-06, "loss": 0.2435, "step": 1088 }, { "epoch": 0.68, "learning_rate": 7.50287666769905e-06, "loss": 0.1838, "step": 1089 }, { "epoch": 0.68, "learning_rate": 7.498561252151401e-06, "loss": 0.2062, "step": 1090 }, { "epoch": 0.69, "learning_rate": 7.494243354515083e-06, "loss": 0.196, "step": 1091 }, { "epoch": 0.69, "learning_rate": 7.4899229790795225e-06, "loss": 0.2331, "step": 1092 }, { "epoch": 0.69, "learning_rate": 7.485600130136611e-06, "loss": 0.2431, "step": 1093 }, { "epoch": 0.69, "learning_rate": 7.481274811980701e-06, "loss": 0.1973, "step": 1094 }, { "epoch": 0.69, "learning_rate": 7.476947028908595e-06, "loss": 0.2126, "step": 1095 }, { "epoch": 0.69, "learning_rate": 7.4726167852195395e-06, "loss": 0.1839, "step": 1096 }, { "epoch": 0.69, "learning_rate": 7.468284085215233e-06, "loss": 0.2007, "step": 1097 }, { "epoch": 0.69, "learning_rate": 7.463948933199809e-06, "loss": 0.185, "step": 1098 }, { "epoch": 0.69, "learning_rate": 7.45961133347984e-06, "loss": 0.2131, "step": 1099 }, { "epoch": 0.69, "learning_rate": 7.455271290364328e-06, "loss": 0.1851, "step": 1100 }, { "epoch": 0.69, "learning_rate": 7.450928808164703e-06, "loss": 0.1896, "step": 1101 }, { "epoch": 0.69, "learning_rate": 7.446583891194818e-06, "loss": 0.1958, "step": 1102 }, { "epoch": 0.69, "learning_rate": 7.442236543770945e-06, "loss": 0.1807, "step": 1103 }, { "epoch": 0.69, "learning_rate": 7.437886770211769e-06, "loss": 0.1989, "step": 1104 }, { "epoch": 0.69, "learning_rate": 7.433534574838387e-06, "loss": 0.2294, "step": 1105 }, { "epoch": 0.69, "learning_rate": 7.429179961974302e-06, "loss": 0.1967, "step": 1106 }, { "epoch": 0.7, "learning_rate": 7.424822935945416e-06, "loss": 0.1932, "step": 1107 }, { "epoch": 0.7, "learning_rate": 7.420463501080031e-06, "loss": 0.1843, "step": 1108 }, { "epoch": 0.7, "learning_rate": 7.4161016617088395e-06, "loss": 0.2137, "step": 1109 }, { "epoch": 0.7, "learning_rate": 7.411737422164926e-06, "loss": 0.2614, "step": 1110 }, { "epoch": 0.7, "learning_rate": 7.407370786783757e-06, "loss": 0.1743, "step": 1111 }, { "epoch": 0.7, "learning_rate": 7.403001759903181e-06, "loss": 0.2036, "step": 1112 }, { "epoch": 0.7, "learning_rate": 7.3986303458634155e-06, "loss": 0.2109, "step": 1113 }, { "epoch": 0.7, "learning_rate": 7.394256549007057e-06, "loss": 0.1969, "step": 1114 }, { "epoch": 0.7, "learning_rate": 7.3898803736790706e-06, "loss": 0.2003, "step": 1115 }, { "epoch": 0.7, "learning_rate": 7.385501824226774e-06, "loss": 0.1964, "step": 1116 }, { "epoch": 0.7, "learning_rate": 7.381120904999852e-06, "loss": 0.2068, "step": 1117 }, { "epoch": 0.7, "learning_rate": 7.376737620350345e-06, "loss": 0.1848, "step": 1118 }, { "epoch": 0.7, "learning_rate": 7.372351974632634e-06, "loss": 0.1936, "step": 1119 }, { "epoch": 0.7, "learning_rate": 7.3679639722034545e-06, "loss": 0.2087, "step": 1120 }, { "epoch": 0.7, "learning_rate": 7.363573617421876e-06, "loss": 0.1987, "step": 1121 }, { "epoch": 0.7, "learning_rate": 7.359180914649313e-06, "loss": 0.2222, "step": 1122 }, { "epoch": 0.71, "learning_rate": 7.354785868249504e-06, "loss": 0.2139, "step": 1123 }, { "epoch": 0.71, "learning_rate": 7.350388482588522e-06, "loss": 0.2296, "step": 1124 }, { "epoch": 0.71, "learning_rate": 7.345988762034759e-06, "loss": 0.2131, "step": 1125 }, { "epoch": 0.71, "learning_rate": 7.341586710958932e-06, "loss": 0.1835, "step": 1126 }, { "epoch": 0.71, "learning_rate": 7.33718233373407e-06, "loss": 0.1869, "step": 1127 }, { "epoch": 0.71, "learning_rate": 7.332775634735511e-06, "loss": 0.2278, "step": 1128 }, { "epoch": 0.71, "learning_rate": 7.328366618340901e-06, "loss": 0.2081, "step": 1129 }, { "epoch": 0.71, "learning_rate": 7.323955288930189e-06, "loss": 0.2009, "step": 1130 }, { "epoch": 0.71, "learning_rate": 7.319541650885623e-06, "loss": 0.1965, "step": 1131 }, { "epoch": 0.71, "learning_rate": 7.315125708591742e-06, "loss": 0.1706, "step": 1132 }, { "epoch": 0.71, "learning_rate": 7.310707466435373e-06, "loss": 0.2596, "step": 1133 }, { "epoch": 0.71, "learning_rate": 7.306286928805632e-06, "loss": 0.1849, "step": 1134 }, { "epoch": 0.71, "learning_rate": 7.3018641000939115e-06, "loss": 0.1963, "step": 1135 }, { "epoch": 0.71, "learning_rate": 7.297438984693881e-06, "loss": 0.2111, "step": 1136 }, { "epoch": 0.71, "learning_rate": 7.293011587001481e-06, "loss": 0.2109, "step": 1137 }, { "epoch": 0.71, "learning_rate": 7.288581911414923e-06, "loss": 0.1733, "step": 1138 }, { "epoch": 0.72, "learning_rate": 7.2841499623346755e-06, "loss": 0.2048, "step": 1139 }, { "epoch": 0.72, "learning_rate": 7.279715744163472e-06, "loss": 0.1959, "step": 1140 }, { "epoch": 0.72, "learning_rate": 7.275279261306292e-06, "loss": 0.2281, "step": 1141 }, { "epoch": 0.72, "learning_rate": 7.270840518170372e-06, "loss": 0.1882, "step": 1142 }, { "epoch": 0.72, "learning_rate": 7.266399519165193e-06, "loss": 0.2059, "step": 1143 }, { "epoch": 0.72, "learning_rate": 7.261956268702473e-06, "loss": 0.1889, "step": 1144 }, { "epoch": 0.72, "learning_rate": 7.2575107711961675e-06, "loss": 0.1913, "step": 1145 }, { "epoch": 0.72, "learning_rate": 7.253063031062468e-06, "loss": 0.1901, "step": 1146 }, { "epoch": 0.72, "learning_rate": 7.248613052719793e-06, "loss": 0.1844, "step": 1147 }, { "epoch": 0.72, "learning_rate": 7.244160840588779e-06, "loss": 0.2089, "step": 1148 }, { "epoch": 0.72, "learning_rate": 7.239706399092288e-06, "loss": 0.2245, "step": 1149 }, { "epoch": 0.72, "learning_rate": 7.235249732655393e-06, "loss": 0.1781, "step": 1150 }, { "epoch": 0.72, "learning_rate": 7.2307908457053786e-06, "loss": 0.2003, "step": 1151 }, { "epoch": 0.72, "learning_rate": 7.226329742671736e-06, "loss": 0.2143, "step": 1152 }, { "epoch": 0.72, "learning_rate": 7.221866427986154e-06, "loss": 0.2085, "step": 1153 }, { "epoch": 0.72, "learning_rate": 7.217400906082526e-06, "loss": 0.2089, "step": 1154 }, { "epoch": 0.73, "learning_rate": 7.2129331813969315e-06, "loss": 0.2439, "step": 1155 }, { "epoch": 0.73, "learning_rate": 7.208463258367637e-06, "loss": 0.1852, "step": 1156 }, { "epoch": 0.73, "learning_rate": 7.203991141435101e-06, "loss": 0.1931, "step": 1157 }, { "epoch": 0.73, "learning_rate": 7.199516835041956e-06, "loss": 0.1933, "step": 1158 }, { "epoch": 0.73, "learning_rate": 7.195040343633006e-06, "loss": 0.1904, "step": 1159 }, { "epoch": 0.73, "learning_rate": 7.190561671655234e-06, "loss": 0.1908, "step": 1160 }, { "epoch": 0.73, "learning_rate": 7.1860808235577815e-06, "loss": 0.2162, "step": 1161 }, { "epoch": 0.73, "learning_rate": 7.1815978037919595e-06, "loss": 0.1939, "step": 1162 }, { "epoch": 0.73, "learning_rate": 7.177112616811228e-06, "loss": 0.2168, "step": 1163 }, { "epoch": 0.73, "learning_rate": 7.172625267071204e-06, "loss": 0.2017, "step": 1164 }, { "epoch": 0.73, "learning_rate": 7.168135759029655e-06, "loss": 0.1945, "step": 1165 }, { "epoch": 0.73, "learning_rate": 7.163644097146489e-06, "loss": 0.173, "step": 1166 }, { "epoch": 0.73, "learning_rate": 7.159150285883757e-06, "loss": 0.1868, "step": 1167 }, { "epoch": 0.73, "learning_rate": 7.154654329705641e-06, "loss": 0.2017, "step": 1168 }, { "epoch": 0.73, "learning_rate": 7.150156233078459e-06, "loss": 0.1903, "step": 1169 }, { "epoch": 0.73, "learning_rate": 7.14565600047065e-06, "loss": 0.1939, "step": 1170 }, { "epoch": 0.74, "learning_rate": 7.141153636352778e-06, "loss": 0.1928, "step": 1171 }, { "epoch": 0.74, "learning_rate": 7.136649145197522e-06, "loss": 0.2033, "step": 1172 }, { "epoch": 0.74, "learning_rate": 7.132142531479679e-06, "loss": 0.1915, "step": 1173 }, { "epoch": 0.74, "learning_rate": 7.1276337996761505e-06, "loss": 0.1906, "step": 1174 }, { "epoch": 0.74, "learning_rate": 7.123122954265942e-06, "loss": 0.2015, "step": 1175 }, { "epoch": 0.74, "learning_rate": 7.118609999730161e-06, "loss": 0.1993, "step": 1176 }, { "epoch": 0.74, "learning_rate": 7.114094940552008e-06, "loss": 0.1881, "step": 1177 }, { "epoch": 0.74, "learning_rate": 7.1095777812167755e-06, "loss": 0.1962, "step": 1178 }, { "epoch": 0.74, "learning_rate": 7.105058526211842e-06, "loss": 0.1888, "step": 1179 }, { "epoch": 0.74, "learning_rate": 7.100537180026666e-06, "loss": 0.201, "step": 1180 }, { "epoch": 0.74, "learning_rate": 7.096013747152789e-06, "loss": 0.1861, "step": 1181 }, { "epoch": 0.74, "learning_rate": 7.0914882320838184e-06, "loss": 0.2056, "step": 1182 }, { "epoch": 0.74, "learning_rate": 7.086960639315437e-06, "loss": 0.2079, "step": 1183 }, { "epoch": 0.74, "learning_rate": 7.0824309733453825e-06, "loss": 0.1981, "step": 1184 }, { "epoch": 0.74, "learning_rate": 7.077899238673461e-06, "loss": 0.1759, "step": 1185 }, { "epoch": 0.74, "learning_rate": 7.073365439801532e-06, "loss": 0.2134, "step": 1186 }, { "epoch": 0.75, "learning_rate": 7.0688295812334995e-06, "loss": 0.2325, "step": 1187 }, { "epoch": 0.75, "learning_rate": 7.064291667475319e-06, "loss": 0.2412, "step": 1188 }, { "epoch": 0.75, "learning_rate": 7.05975170303499e-06, "loss": 0.166, "step": 1189 }, { "epoch": 0.75, "learning_rate": 7.055209692422543e-06, "loss": 0.2047, "step": 1190 }, { "epoch": 0.75, "learning_rate": 7.050665640150045e-06, "loss": 0.223, "step": 1191 }, { "epoch": 0.75, "learning_rate": 7.046119550731587e-06, "loss": 0.2314, "step": 1192 }, { "epoch": 0.75, "learning_rate": 7.04157142868329e-06, "loss": 0.179, "step": 1193 }, { "epoch": 0.75, "learning_rate": 7.037021278523292e-06, "loss": 0.2115, "step": 1194 }, { "epoch": 0.75, "learning_rate": 7.032469104771742e-06, "loss": 0.1978, "step": 1195 }, { "epoch": 0.75, "learning_rate": 7.027914911950802e-06, "loss": 0.2051, "step": 1196 }, { "epoch": 0.75, "learning_rate": 7.023358704584644e-06, "loss": 0.2021, "step": 1197 }, { "epoch": 0.75, "learning_rate": 7.018800487199432e-06, "loss": 0.214, "step": 1198 }, { "epoch": 0.75, "learning_rate": 7.0142402643233346e-06, "loss": 0.1923, "step": 1199 }, { "epoch": 0.75, "learning_rate": 7.0096780404865086e-06, "loss": 0.21, "step": 1200 }, { "epoch": 0.75, "learning_rate": 7.005113820221101e-06, "loss": 0.1955, "step": 1201 }, { "epoch": 0.75, "learning_rate": 7.00054760806124e-06, "loss": 0.1938, "step": 1202 }, { "epoch": 0.76, "learning_rate": 6.995979408543034e-06, "loss": 0.1775, "step": 1203 }, { "epoch": 0.76, "learning_rate": 6.991409226204566e-06, "loss": 0.2004, "step": 1204 }, { "epoch": 0.76, "learning_rate": 6.9868370655858875e-06, "loss": 0.1888, "step": 1205 }, { "epoch": 0.76, "learning_rate": 6.982262931229014e-06, "loss": 0.1684, "step": 1206 }, { "epoch": 0.76, "learning_rate": 6.977686827677926e-06, "loss": 0.2061, "step": 1207 }, { "epoch": 0.76, "learning_rate": 6.973108759478555e-06, "loss": 0.1916, "step": 1208 }, { "epoch": 0.76, "learning_rate": 6.968528731178789e-06, "loss": 0.2231, "step": 1209 }, { "epoch": 0.76, "learning_rate": 6.96394674732846e-06, "loss": 0.2116, "step": 1210 }, { "epoch": 0.76, "learning_rate": 6.959362812479342e-06, "loss": 0.191, "step": 1211 }, { "epoch": 0.76, "learning_rate": 6.954776931185152e-06, "loss": 0.2025, "step": 1212 }, { "epoch": 0.76, "learning_rate": 6.950189108001536e-06, "loss": 0.1862, "step": 1213 }, { "epoch": 0.76, "learning_rate": 6.945599347486069e-06, "loss": 0.1954, "step": 1214 }, { "epoch": 0.76, "learning_rate": 6.941007654198254e-06, "loss": 0.2013, "step": 1215 }, { "epoch": 0.76, "learning_rate": 6.9364140326995114e-06, "loss": 0.1961, "step": 1216 }, { "epoch": 0.76, "learning_rate": 6.931818487553177e-06, "loss": 0.2112, "step": 1217 }, { "epoch": 0.76, "learning_rate": 6.927221023324501e-06, "loss": 0.2011, "step": 1218 }, { "epoch": 0.77, "learning_rate": 6.9226216445806315e-06, "loss": 0.193, "step": 1219 }, { "epoch": 0.77, "learning_rate": 6.918020355890631e-06, "loss": 0.1975, "step": 1220 }, { "epoch": 0.77, "learning_rate": 6.913417161825449e-06, "loss": 0.1954, "step": 1221 }, { "epoch": 0.77, "learning_rate": 6.908812066957933e-06, "loss": 0.178, "step": 1222 }, { "epoch": 0.77, "learning_rate": 6.904205075862816e-06, "loss": 0.2132, "step": 1223 }, { "epoch": 0.77, "learning_rate": 6.899596193116717e-06, "loss": 0.2215, "step": 1224 }, { "epoch": 0.77, "learning_rate": 6.894985423298134e-06, "loss": 0.1878, "step": 1225 }, { "epoch": 0.77, "learning_rate": 6.890372770987436e-06, "loss": 0.2205, "step": 1226 }, { "epoch": 0.77, "learning_rate": 6.885758240766867e-06, "loss": 0.2242, "step": 1227 }, { "epoch": 0.77, "learning_rate": 6.881141837220533e-06, "loss": 0.2085, "step": 1228 }, { "epoch": 0.77, "learning_rate": 6.876523564934404e-06, "loss": 0.1871, "step": 1229 }, { "epoch": 0.77, "learning_rate": 6.871903428496305e-06, "loss": 0.1793, "step": 1230 }, { "epoch": 0.77, "learning_rate": 6.867281432495911e-06, "loss": 0.1854, "step": 1231 }, { "epoch": 0.77, "learning_rate": 6.862657581524748e-06, "loss": 0.1912, "step": 1232 }, { "epoch": 0.77, "learning_rate": 6.858031880176179e-06, "loss": 0.2219, "step": 1233 }, { "epoch": 0.78, "learning_rate": 6.8534043330454135e-06, "loss": 0.2271, "step": 1234 }, { "epoch": 0.78, "learning_rate": 6.848774944729487e-06, "loss": 0.1973, "step": 1235 }, { "epoch": 0.78, "learning_rate": 6.844143719827268e-06, "loss": 0.1961, "step": 1236 }, { "epoch": 0.78, "learning_rate": 6.8395106629394504e-06, "loss": 0.2185, "step": 1237 }, { "epoch": 0.78, "learning_rate": 6.834875778668542e-06, "loss": 0.2134, "step": 1238 }, { "epoch": 0.78, "learning_rate": 6.830239071618874e-06, "loss": 0.1853, "step": 1239 }, { "epoch": 0.78, "learning_rate": 6.825600546396582e-06, "loss": 0.2119, "step": 1240 }, { "epoch": 0.78, "learning_rate": 6.820960207609612e-06, "loss": 0.2069, "step": 1241 }, { "epoch": 0.78, "learning_rate": 6.816318059867708e-06, "loss": 0.1793, "step": 1242 }, { "epoch": 0.78, "learning_rate": 6.811674107782413e-06, "loss": 0.164, "step": 1243 }, { "epoch": 0.78, "learning_rate": 6.807028355967065e-06, "loss": 0.2042, "step": 1244 }, { "epoch": 0.78, "learning_rate": 6.802380809036784e-06, "loss": 0.195, "step": 1245 }, { "epoch": 0.78, "learning_rate": 6.797731471608479e-06, "loss": 0.1947, "step": 1246 }, { "epoch": 0.78, "learning_rate": 6.793080348300834e-06, "loss": 0.192, "step": 1247 }, { "epoch": 0.78, "learning_rate": 6.788427443734307e-06, "loss": 0.2065, "step": 1248 }, { "epoch": 0.78, "learning_rate": 6.783772762531128e-06, "loss": 0.2059, "step": 1249 }, { "epoch": 0.79, "learning_rate": 6.779116309315291e-06, "loss": 0.1947, "step": 1250 }, { "epoch": 0.79, "learning_rate": 6.7744580887125475e-06, "loss": 0.1809, "step": 1251 }, { "epoch": 0.79, "learning_rate": 6.769798105350413e-06, "loss": 0.2045, "step": 1252 }, { "epoch": 0.79, "learning_rate": 6.765136363858143e-06, "loss": 0.2107, "step": 1253 }, { "epoch": 0.79, "learning_rate": 6.760472868866748e-06, "loss": 0.2247, "step": 1254 }, { "epoch": 0.79, "learning_rate": 6.755807625008974e-06, "loss": 0.2216, "step": 1255 }, { "epoch": 0.79, "learning_rate": 6.751140636919311e-06, "loss": 0.1676, "step": 1256 }, { "epoch": 0.79, "learning_rate": 6.746471909233978e-06, "loss": 0.1884, "step": 1257 }, { "epoch": 0.79, "learning_rate": 6.741801446590921e-06, "loss": 0.2162, "step": 1258 }, { "epoch": 0.79, "learning_rate": 6.737129253629812e-06, "loss": 0.1947, "step": 1259 }, { "epoch": 0.79, "learning_rate": 6.732455334992042e-06, "loss": 0.1892, "step": 1260 }, { "epoch": 0.79, "learning_rate": 6.727779695320713e-06, "loss": 0.1971, "step": 1261 }, { "epoch": 0.79, "learning_rate": 6.723102339260641e-06, "loss": 0.2072, "step": 1262 }, { "epoch": 0.79, "learning_rate": 6.718423271458343e-06, "loss": 0.1942, "step": 1263 }, { "epoch": 0.79, "learning_rate": 6.713742496562041e-06, "loss": 0.2055, "step": 1264 }, { "epoch": 0.79, "learning_rate": 6.70906001922165e-06, "loss": 0.1796, "step": 1265 }, { "epoch": 0.8, "learning_rate": 6.704375844088776e-06, "loss": 0.1942, "step": 1266 }, { "epoch": 0.8, "learning_rate": 6.69968997581671e-06, "loss": 0.2084, "step": 1267 }, { "epoch": 0.8, "learning_rate": 6.6950024190604306e-06, "loss": 0.2081, "step": 1268 }, { "epoch": 0.8, "learning_rate": 6.690313178476588e-06, "loss": 0.233, "step": 1269 }, { "epoch": 0.8, "learning_rate": 6.6856222587235085e-06, "loss": 0.2385, "step": 1270 }, { "epoch": 0.8, "learning_rate": 6.680929664461184e-06, "loss": 0.2199, "step": 1271 }, { "epoch": 0.8, "learning_rate": 6.676235400351274e-06, "loss": 0.1834, "step": 1272 }, { "epoch": 0.8, "learning_rate": 6.671539471057093e-06, "loss": 0.2041, "step": 1273 }, { "epoch": 0.8, "learning_rate": 6.66684188124361e-06, "loss": 0.1978, "step": 1274 }, { "epoch": 0.8, "learning_rate": 6.662142635577444e-06, "loss": 0.1995, "step": 1275 }, { "epoch": 0.8, "learning_rate": 6.657441738726862e-06, "loss": 0.214, "step": 1276 }, { "epoch": 0.8, "learning_rate": 6.652739195361766e-06, "loss": 0.1871, "step": 1277 }, { "epoch": 0.8, "learning_rate": 6.648035010153697e-06, "loss": 0.1991, "step": 1278 }, { "epoch": 0.8, "learning_rate": 6.643329187775827e-06, "loss": 0.1822, "step": 1279 }, { "epoch": 0.8, "learning_rate": 6.638621732902952e-06, "loss": 0.1821, "step": 1280 }, { "epoch": 0.8, "learning_rate": 6.633912650211496e-06, "loss": 0.1969, "step": 1281 }, { "epoch": 0.81, "learning_rate": 6.629201944379489e-06, "loss": 0.2232, "step": 1282 }, { "epoch": 0.81, "learning_rate": 6.624489620086584e-06, "loss": 0.2069, "step": 1283 }, { "epoch": 0.81, "learning_rate": 6.619775682014038e-06, "loss": 0.1858, "step": 1284 }, { "epoch": 0.81, "learning_rate": 6.615060134844709e-06, "loss": 0.1776, "step": 1285 }, { "epoch": 0.81, "learning_rate": 6.610342983263056e-06, "loss": 0.1725, "step": 1286 }, { "epoch": 0.81, "learning_rate": 6.6056242319551315e-06, "loss": 0.1762, "step": 1287 }, { "epoch": 0.81, "learning_rate": 6.600903885608578e-06, "loss": 0.1719, "step": 1288 }, { "epoch": 0.81, "learning_rate": 6.59618194891262e-06, "loss": 0.2414, "step": 1289 }, { "epoch": 0.81, "learning_rate": 6.591458426558062e-06, "loss": 0.1903, "step": 1290 }, { "epoch": 0.81, "learning_rate": 6.586733323237288e-06, "loss": 0.205, "step": 1291 }, { "epoch": 0.81, "learning_rate": 6.582006643644248e-06, "loss": 0.2033, "step": 1292 }, { "epoch": 0.81, "learning_rate": 6.577278392474459e-06, "loss": 0.1813, "step": 1293 }, { "epoch": 0.81, "learning_rate": 6.572548574425001e-06, "loss": 0.208, "step": 1294 }, { "epoch": 0.81, "learning_rate": 6.567817194194508e-06, "loss": 0.1791, "step": 1295 }, { "epoch": 0.81, "learning_rate": 6.563084256483167e-06, "loss": 0.1756, "step": 1296 }, { "epoch": 0.81, "learning_rate": 6.558349765992713e-06, "loss": 0.1885, "step": 1297 }, { "epoch": 0.82, "learning_rate": 6.553613727426421e-06, "loss": 0.1547, "step": 1298 }, { "epoch": 0.82, "learning_rate": 6.548876145489109e-06, "loss": 0.1917, "step": 1299 }, { "epoch": 0.82, "learning_rate": 6.544137024887122e-06, "loss": 0.1681, "step": 1300 }, { "epoch": 0.82, "learning_rate": 6.539396370328339e-06, "loss": 0.2118, "step": 1301 }, { "epoch": 0.82, "learning_rate": 6.534654186522158e-06, "loss": 0.2035, "step": 1302 }, { "epoch": 0.82, "learning_rate": 6.529910478179499e-06, "loss": 0.1988, "step": 1303 }, { "epoch": 0.82, "learning_rate": 6.525165250012796e-06, "loss": 0.1497, "step": 1304 }, { "epoch": 0.82, "learning_rate": 6.520418506735992e-06, "loss": 0.1942, "step": 1305 }, { "epoch": 0.82, "learning_rate": 6.515670253064537e-06, "loss": 0.2214, "step": 1306 }, { "epoch": 0.82, "learning_rate": 6.510920493715381e-06, "loss": 0.2001, "step": 1307 }, { "epoch": 0.82, "learning_rate": 6.506169233406965e-06, "loss": 0.1585, "step": 1308 }, { "epoch": 0.82, "learning_rate": 6.501416476859231e-06, "loss": 0.1754, "step": 1309 }, { "epoch": 0.82, "learning_rate": 6.496662228793597e-06, "loss": 0.1866, "step": 1310 }, { "epoch": 0.82, "learning_rate": 6.491906493932968e-06, "loss": 0.1938, "step": 1311 }, { "epoch": 0.82, "learning_rate": 6.4871492770017254e-06, "loss": 0.2204, "step": 1312 }, { "epoch": 0.82, "learning_rate": 6.482390582725724e-06, "loss": 0.1798, "step": 1313 }, { "epoch": 0.83, "learning_rate": 6.477630415832282e-06, "loss": 0.1893, "step": 1314 }, { "epoch": 0.83, "learning_rate": 6.472868781050186e-06, "loss": 0.1737, "step": 1315 }, { "epoch": 0.83, "learning_rate": 6.468105683109679e-06, "loss": 0.2152, "step": 1316 }, { "epoch": 0.83, "learning_rate": 6.463341126742453e-06, "loss": 0.2183, "step": 1317 }, { "epoch": 0.83, "learning_rate": 6.458575116681653e-06, "loss": 0.2001, "step": 1318 }, { "epoch": 0.83, "learning_rate": 6.45380765766187e-06, "loss": 0.168, "step": 1319 }, { "epoch": 0.83, "learning_rate": 6.4490387544191316e-06, "loss": 0.1774, "step": 1320 }, { "epoch": 0.83, "learning_rate": 6.4442684116908995e-06, "loss": 0.2131, "step": 1321 }, { "epoch": 0.83, "learning_rate": 6.439496634216065e-06, "loss": 0.17, "step": 1322 }, { "epoch": 0.83, "learning_rate": 6.434723426734949e-06, "loss": 0.1453, "step": 1323 }, { "epoch": 0.83, "learning_rate": 6.42994879398929e-06, "loss": 0.1989, "step": 1324 }, { "epoch": 0.83, "learning_rate": 6.42517274072224e-06, "loss": 0.1897, "step": 1325 }, { "epoch": 0.83, "learning_rate": 6.420395271678365e-06, "loss": 0.1674, "step": 1326 }, { "epoch": 0.83, "learning_rate": 6.415616391603639e-06, "loss": 0.1926, "step": 1327 }, { "epoch": 0.83, "learning_rate": 6.4108361052454326e-06, "loss": 0.1993, "step": 1328 }, { "epoch": 0.83, "learning_rate": 6.40605441735252e-06, "loss": 0.2081, "step": 1329 }, { "epoch": 0.84, "learning_rate": 6.401271332675062e-06, "loss": 0.1975, "step": 1330 }, { "epoch": 0.84, "learning_rate": 6.396486855964611e-06, "loss": 0.1813, "step": 1331 }, { "epoch": 0.84, "learning_rate": 6.391700991974099e-06, "loss": 0.1992, "step": 1332 }, { "epoch": 0.84, "learning_rate": 6.386913745457839e-06, "loss": 0.1999, "step": 1333 }, { "epoch": 0.84, "learning_rate": 6.3821251211715125e-06, "loss": 0.1876, "step": 1334 }, { "epoch": 0.84, "learning_rate": 6.377335123872177e-06, "loss": 0.1782, "step": 1335 }, { "epoch": 0.84, "learning_rate": 6.372543758318251e-06, "loss": 0.1673, "step": 1336 }, { "epoch": 0.84, "learning_rate": 6.367751029269507e-06, "loss": 0.204, "step": 1337 }, { "epoch": 0.84, "learning_rate": 6.36295694148708e-06, "loss": 0.1906, "step": 1338 }, { "epoch": 0.84, "learning_rate": 6.35816149973345e-06, "loss": 0.2026, "step": 1339 }, { "epoch": 0.84, "learning_rate": 6.353364708772443e-06, "loss": 0.1838, "step": 1340 }, { "epoch": 0.84, "learning_rate": 6.348566573369226e-06, "loss": 0.1651, "step": 1341 }, { "epoch": 0.84, "learning_rate": 6.3437670982902985e-06, "loss": 0.2057, "step": 1342 }, { "epoch": 0.84, "learning_rate": 6.338966288303499e-06, "loss": 0.1808, "step": 1343 }, { "epoch": 0.84, "learning_rate": 6.334164148177982e-06, "loss": 0.2065, "step": 1344 }, { "epoch": 0.84, "learning_rate": 6.329360682684231e-06, "loss": 0.1856, "step": 1345 }, { "epoch": 0.85, "learning_rate": 6.32455589659404e-06, "loss": 0.1763, "step": 1346 }, { "epoch": 0.85, "learning_rate": 6.3197497946805205e-06, "loss": 0.2002, "step": 1347 }, { "epoch": 0.85, "learning_rate": 6.314942381718088e-06, "loss": 0.1867, "step": 1348 }, { "epoch": 0.85, "learning_rate": 6.310133662482459e-06, "loss": 0.1663, "step": 1349 }, { "epoch": 0.85, "learning_rate": 6.305323641750652e-06, "loss": 0.1565, "step": 1350 }, { "epoch": 0.85, "learning_rate": 6.300512324300975e-06, "loss": 0.1833, "step": 1351 }, { "epoch": 0.85, "learning_rate": 6.295699714913027e-06, "loss": 0.2337, "step": 1352 }, { "epoch": 0.85, "learning_rate": 6.290885818367685e-06, "loss": 0.1966, "step": 1353 }, { "epoch": 0.85, "learning_rate": 6.286070639447113e-06, "loss": 0.2031, "step": 1354 }, { "epoch": 0.85, "learning_rate": 6.281254182934742e-06, "loss": 0.1943, "step": 1355 }, { "epoch": 0.85, "learning_rate": 6.2764364536152735e-06, "loss": 0.1709, "step": 1356 }, { "epoch": 0.85, "learning_rate": 6.271617456274676e-06, "loss": 0.2088, "step": 1357 }, { "epoch": 0.85, "learning_rate": 6.2667971957001735e-06, "loss": 0.1901, "step": 1358 }, { "epoch": 0.85, "learning_rate": 6.261975676680252e-06, "loss": 0.1716, "step": 1359 }, { "epoch": 0.85, "learning_rate": 6.2571529040046395e-06, "loss": 0.2058, "step": 1360 }, { "epoch": 0.85, "learning_rate": 6.252328882464313e-06, "loss": 0.1759, "step": 1361 }, { "epoch": 0.86, "learning_rate": 6.247503616851491e-06, "loss": 0.2058, "step": 1362 }, { "epoch": 0.86, "learning_rate": 6.242677111959627e-06, "loss": 0.194, "step": 1363 }, { "epoch": 0.86, "learning_rate": 6.237849372583406e-06, "loss": 0.2008, "step": 1364 }, { "epoch": 0.86, "learning_rate": 6.233020403518739e-06, "loss": 0.1633, "step": 1365 }, { "epoch": 0.86, "learning_rate": 6.228190209562758e-06, "loss": 0.1974, "step": 1366 }, { "epoch": 0.86, "learning_rate": 6.223358795513812e-06, "loss": 0.1605, "step": 1367 }, { "epoch": 0.86, "learning_rate": 6.218526166171462e-06, "loss": 0.2006, "step": 1368 }, { "epoch": 0.86, "learning_rate": 6.213692326336477e-06, "loss": 0.1771, "step": 1369 }, { "epoch": 0.86, "learning_rate": 6.208857280810828e-06, "loss": 0.2039, "step": 1370 }, { "epoch": 0.86, "learning_rate": 6.204021034397686e-06, "loss": 0.1981, "step": 1371 }, { "epoch": 0.86, "learning_rate": 6.199183591901411e-06, "loss": 0.1863, "step": 1372 }, { "epoch": 0.86, "learning_rate": 6.194344958127553e-06, "loss": 0.2024, "step": 1373 }, { "epoch": 0.86, "learning_rate": 6.1895051378828464e-06, "loss": 0.2008, "step": 1374 }, { "epoch": 0.86, "learning_rate": 6.184664135975202e-06, "loss": 0.161, "step": 1375 }, { "epoch": 0.86, "learning_rate": 6.179821957213708e-06, "loss": 0.1655, "step": 1376 }, { "epoch": 0.86, "learning_rate": 6.1749786064086145e-06, "loss": 0.1596, "step": 1377 }, { "epoch": 0.87, "learning_rate": 6.170134088371348e-06, "loss": 0.1835, "step": 1378 }, { "epoch": 0.87, "learning_rate": 6.165288407914482e-06, "loss": 0.1768, "step": 1379 }, { "epoch": 0.87, "learning_rate": 6.160441569851756e-06, "loss": 0.1882, "step": 1380 }, { "epoch": 0.87, "learning_rate": 6.1555935789980456e-06, "loss": 0.1795, "step": 1381 }, { "epoch": 0.87, "learning_rate": 6.150744440169386e-06, "loss": 0.1813, "step": 1382 }, { "epoch": 0.87, "learning_rate": 6.145894158182945e-06, "loss": 0.1706, "step": 1383 }, { "epoch": 0.87, "learning_rate": 6.141042737857026e-06, "loss": 0.2343, "step": 1384 }, { "epoch": 0.87, "learning_rate": 6.136190184011066e-06, "loss": 0.2124, "step": 1385 }, { "epoch": 0.87, "learning_rate": 6.131336501465627e-06, "loss": 0.1687, "step": 1386 }, { "epoch": 0.87, "learning_rate": 6.126481695042392e-06, "loss": 0.1679, "step": 1387 }, { "epoch": 0.87, "learning_rate": 6.12162576956416e-06, "loss": 0.2063, "step": 1388 }, { "epoch": 0.87, "learning_rate": 6.116768729854842e-06, "loss": 0.2061, "step": 1389 }, { "epoch": 0.87, "learning_rate": 6.111910580739456e-06, "loss": 0.1844, "step": 1390 }, { "epoch": 0.87, "learning_rate": 6.107051327044124e-06, "loss": 0.1997, "step": 1391 }, { "epoch": 0.87, "learning_rate": 6.102190973596061e-06, "loss": 0.1811, "step": 1392 }, { "epoch": 0.87, "learning_rate": 6.097329525223579e-06, "loss": 0.206, "step": 1393 }, { "epoch": 0.88, "learning_rate": 6.092466986756076e-06, "loss": 0.1768, "step": 1394 }, { "epoch": 0.88, "learning_rate": 6.087603363024029e-06, "loss": 0.1698, "step": 1395 }, { "epoch": 0.88, "learning_rate": 6.082738658859001e-06, "loss": 0.1577, "step": 1396 }, { "epoch": 0.88, "learning_rate": 6.07787287909362e-06, "loss": 0.2018, "step": 1397 }, { "epoch": 0.88, "learning_rate": 6.07300602856159e-06, "loss": 0.1597, "step": 1398 }, { "epoch": 0.88, "learning_rate": 6.0681381120976745e-06, "loss": 0.1758, "step": 1399 }, { "epoch": 0.88, "learning_rate": 6.063269134537696e-06, "loss": 0.1832, "step": 1400 }, { "epoch": 0.88, "learning_rate": 6.0583991007185304e-06, "loss": 0.1384, "step": 1401 }, { "epoch": 0.88, "learning_rate": 6.053528015478107e-06, "loss": 0.1809, "step": 1402 }, { "epoch": 0.88, "learning_rate": 6.0486558836553955e-06, "loss": 0.1953, "step": 1403 }, { "epoch": 0.88, "learning_rate": 6.043782710090405e-06, "loss": 0.1842, "step": 1404 }, { "epoch": 0.88, "learning_rate": 6.038908499624183e-06, "loss": 0.1692, "step": 1405 }, { "epoch": 0.88, "learning_rate": 6.0340332570988055e-06, "loss": 0.1564, "step": 1406 }, { "epoch": 0.88, "learning_rate": 6.029156987357373e-06, "loss": 0.1758, "step": 1407 }, { "epoch": 0.88, "learning_rate": 6.024279695244006e-06, "loss": 0.187, "step": 1408 }, { "epoch": 0.88, "learning_rate": 6.019401385603843e-06, "loss": 0.2047, "step": 1409 }, { "epoch": 0.89, "learning_rate": 6.014522063283029e-06, "loss": 0.1641, "step": 1410 }, { "epoch": 0.89, "learning_rate": 6.009641733128721e-06, "loss": 0.1595, "step": 1411 }, { "epoch": 0.89, "learning_rate": 6.004760399989072e-06, "loss": 0.1654, "step": 1412 }, { "epoch": 0.89, "learning_rate": 5.999878068713233e-06, "loss": 0.1854, "step": 1413 }, { "epoch": 0.89, "learning_rate": 5.994994744151348e-06, "loss": 0.1944, "step": 1414 }, { "epoch": 0.89, "learning_rate": 5.990110431154549e-06, "loss": 0.2333, "step": 1415 }, { "epoch": 0.89, "learning_rate": 5.985225134574941e-06, "loss": 0.1859, "step": 1416 }, { "epoch": 0.89, "learning_rate": 5.98033885926562e-06, "loss": 0.1659, "step": 1417 }, { "epoch": 0.89, "learning_rate": 5.975451610080643e-06, "loss": 0.1853, "step": 1418 }, { "epoch": 0.89, "learning_rate": 5.9705633918750375e-06, "loss": 0.1776, "step": 1419 }, { "epoch": 0.89, "learning_rate": 5.965674209504797e-06, "loss": 0.2098, "step": 1420 }, { "epoch": 0.89, "learning_rate": 5.960784067826869e-06, "loss": 0.1731, "step": 1421 }, { "epoch": 0.89, "learning_rate": 5.955892971699158e-06, "loss": 0.141, "step": 1422 }, { "epoch": 0.89, "learning_rate": 5.951000925980509e-06, "loss": 0.2262, "step": 1423 }, { "epoch": 0.89, "learning_rate": 5.946107935530718e-06, "loss": 0.1696, "step": 1424 }, { "epoch": 0.89, "learning_rate": 5.941214005210518e-06, "loss": 0.1611, "step": 1425 }, { "epoch": 0.9, "learning_rate": 5.9363191398815736e-06, "loss": 0.1912, "step": 1426 }, { "epoch": 0.9, "learning_rate": 5.931423344406478e-06, "loss": 0.1653, "step": 1427 }, { "epoch": 0.9, "learning_rate": 5.9265266236487515e-06, "loss": 0.1781, "step": 1428 }, { "epoch": 0.9, "learning_rate": 5.92162898247283e-06, "loss": 0.1915, "step": 1429 }, { "epoch": 0.9, "learning_rate": 5.916730425744066e-06, "loss": 0.1781, "step": 1430 }, { "epoch": 0.9, "learning_rate": 5.9118309583287205e-06, "loss": 0.2017, "step": 1431 }, { "epoch": 0.9, "learning_rate": 5.906930585093959e-06, "loss": 0.1794, "step": 1432 }, { "epoch": 0.9, "learning_rate": 5.902029310907849e-06, "loss": 0.1735, "step": 1433 }, { "epoch": 0.9, "learning_rate": 5.897127140639352e-06, "loss": 0.193, "step": 1434 }, { "epoch": 0.9, "learning_rate": 5.892224079158316e-06, "loss": 0.1792, "step": 1435 }, { "epoch": 0.9, "learning_rate": 5.887320131335481e-06, "loss": 0.172, "step": 1436 }, { "epoch": 0.9, "learning_rate": 5.882415302042461e-06, "loss": 0.1752, "step": 1437 }, { "epoch": 0.9, "learning_rate": 5.877509596151752e-06, "loss": 0.1563, "step": 1438 }, { "epoch": 0.9, "learning_rate": 5.872603018536713e-06, "loss": 0.1465, "step": 1439 }, { "epoch": 0.9, "learning_rate": 5.867695574071574e-06, "loss": 0.1736, "step": 1440 }, { "epoch": 0.91, "learning_rate": 5.862787267631431e-06, "loss": 0.1574, "step": 1441 }, { "epoch": 0.91, "learning_rate": 5.857878104092225e-06, "loss": 0.1772, "step": 1442 }, { "epoch": 0.91, "learning_rate": 5.852968088330758e-06, "loss": 0.1712, "step": 1443 }, { "epoch": 0.91, "learning_rate": 5.848057225224671e-06, "loss": 0.188, "step": 1444 }, { "epoch": 0.91, "learning_rate": 5.843145519652455e-06, "loss": 0.1693, "step": 1445 }, { "epoch": 0.91, "learning_rate": 5.83823297649343e-06, "loss": 0.1847, "step": 1446 }, { "epoch": 0.91, "learning_rate": 5.8333196006277536e-06, "loss": 0.1947, "step": 1447 }, { "epoch": 0.91, "learning_rate": 5.828405396936406e-06, "loss": 0.204, "step": 1448 }, { "epoch": 0.91, "learning_rate": 5.823490370301197e-06, "loss": 0.1761, "step": 1449 }, { "epoch": 0.91, "learning_rate": 5.818574525604748e-06, "loss": 0.168, "step": 1450 }, { "epoch": 0.91, "learning_rate": 5.81365786773049e-06, "loss": 0.1607, "step": 1451 }, { "epoch": 0.91, "learning_rate": 5.808740401562669e-06, "loss": 0.1994, "step": 1452 }, { "epoch": 0.91, "learning_rate": 5.8038221319863344e-06, "loss": 0.158, "step": 1453 }, { "epoch": 0.91, "learning_rate": 5.798903063887326e-06, "loss": 0.2012, "step": 1454 }, { "epoch": 0.91, "learning_rate": 5.793983202152283e-06, "loss": 0.1914, "step": 1455 }, { "epoch": 0.91, "learning_rate": 5.78906255166863e-06, "loss": 0.1563, "step": 1456 }, { "epoch": 0.92, "learning_rate": 5.78414111732458e-06, "loss": 0.1998, "step": 1457 }, { "epoch": 0.92, "learning_rate": 5.779218904009119e-06, "loss": 0.1874, "step": 1458 }, { "epoch": 0.92, "learning_rate": 5.7742959166120075e-06, "loss": 0.1906, "step": 1459 }, { "epoch": 0.92, "learning_rate": 5.769372160023777e-06, "loss": 0.1874, "step": 1460 }, { "epoch": 0.92, "learning_rate": 5.764447639135725e-06, "loss": 0.1712, "step": 1461 }, { "epoch": 0.92, "learning_rate": 5.759522358839905e-06, "loss": 0.1814, "step": 1462 }, { "epoch": 0.92, "learning_rate": 5.754596324029125e-06, "loss": 0.1899, "step": 1463 }, { "epoch": 0.92, "learning_rate": 5.749669539596943e-06, "loss": 0.1762, "step": 1464 }, { "epoch": 0.92, "learning_rate": 5.744742010437663e-06, "loss": 0.1909, "step": 1465 }, { "epoch": 0.92, "learning_rate": 5.739813741446327e-06, "loss": 0.1408, "step": 1466 }, { "epoch": 0.92, "learning_rate": 5.734884737518714e-06, "loss": 0.1612, "step": 1467 }, { "epoch": 0.92, "learning_rate": 5.729955003551328e-06, "loss": 0.1873, "step": 1468 }, { "epoch": 0.92, "learning_rate": 5.725024544441408e-06, "loss": 0.1754, "step": 1469 }, { "epoch": 0.92, "learning_rate": 5.720093365086904e-06, "loss": 0.1702, "step": 1470 }, { "epoch": 0.92, "learning_rate": 5.715161470386485e-06, "loss": 0.2225, "step": 1471 }, { "epoch": 0.92, "learning_rate": 5.710228865239531e-06, "loss": 0.1474, "step": 1472 }, { "epoch": 0.93, "learning_rate": 5.705295554546129e-06, "loss": 0.1966, "step": 1473 }, { "epoch": 0.93, "learning_rate": 5.700361543207061e-06, "loss": 0.2182, "step": 1474 }, { "epoch": 0.93, "learning_rate": 5.695426836123814e-06, "loss": 0.1915, "step": 1475 }, { "epoch": 0.93, "learning_rate": 5.690491438198558e-06, "loss": 0.183, "step": 1476 }, { "epoch": 0.93, "learning_rate": 5.6855553543341535e-06, "loss": 0.1562, "step": 1477 }, { "epoch": 0.93, "learning_rate": 5.680618589434144e-06, "loss": 0.1814, "step": 1478 }, { "epoch": 0.93, "learning_rate": 5.675681148402743e-06, "loss": 0.205, "step": 1479 }, { "epoch": 0.93, "learning_rate": 5.670743036144843e-06, "loss": 0.2009, "step": 1480 }, { "epoch": 0.93, "learning_rate": 5.665804257566001e-06, "loss": 0.1628, "step": 1481 }, { "epoch": 0.93, "learning_rate": 5.660864817572431e-06, "loss": 0.1555, "step": 1482 }, { "epoch": 0.93, "learning_rate": 5.655924721071012e-06, "loss": 0.1738, "step": 1483 }, { "epoch": 0.93, "learning_rate": 5.650983972969269e-06, "loss": 0.1736, "step": 1484 }, { "epoch": 0.93, "learning_rate": 5.646042578175378e-06, "loss": 0.1807, "step": 1485 }, { "epoch": 0.93, "learning_rate": 5.641100541598155e-06, "loss": 0.186, "step": 1486 }, { "epoch": 0.93, "learning_rate": 5.636157868147054e-06, "loss": 0.1828, "step": 1487 }, { "epoch": 0.93, "learning_rate": 5.631214562732164e-06, "loss": 0.188, "step": 1488 }, { "epoch": 0.94, "learning_rate": 5.626270630264199e-06, "loss": 0.1701, "step": 1489 }, { "epoch": 0.94, "learning_rate": 5.621326075654499e-06, "loss": 0.1659, "step": 1490 }, { "epoch": 0.94, "learning_rate": 5.616380903815016e-06, "loss": 0.1674, "step": 1491 }, { "epoch": 0.94, "learning_rate": 5.611435119658323e-06, "loss": 0.1632, "step": 1492 }, { "epoch": 0.94, "learning_rate": 5.606488728097595e-06, "loss": 0.1525, "step": 1493 }, { "epoch": 0.94, "learning_rate": 5.601541734046615e-06, "loss": 0.1876, "step": 1494 }, { "epoch": 0.94, "learning_rate": 5.596594142419759e-06, "loss": 0.1718, "step": 1495 }, { "epoch": 0.94, "learning_rate": 5.591645958132003e-06, "loss": 0.1881, "step": 1496 }, { "epoch": 0.94, "learning_rate": 5.58669718609891e-06, "loss": 0.1748, "step": 1497 }, { "epoch": 0.94, "learning_rate": 5.581747831236621e-06, "loss": 0.1965, "step": 1498 }, { "epoch": 0.94, "learning_rate": 5.576797898461865e-06, "loss": 0.1684, "step": 1499 }, { "epoch": 0.94, "learning_rate": 5.571847392691938e-06, "loss": 0.1649, "step": 1500 }, { "epoch": 0.94, "learning_rate": 5.56689631884471e-06, "loss": 0.1858, "step": 1501 }, { "epoch": 0.94, "learning_rate": 5.5619446818386116e-06, "loss": 0.1967, "step": 1502 }, { "epoch": 0.94, "learning_rate": 5.556992486592634e-06, "loss": 0.1796, "step": 1503 }, { "epoch": 0.94, "learning_rate": 5.552039738026324e-06, "loss": 0.1986, "step": 1504 }, { "epoch": 0.95, "learning_rate": 5.547086441059781e-06, "loss": 0.1747, "step": 1505 }, { "epoch": 0.95, "learning_rate": 5.542132600613642e-06, "loss": 0.1802, "step": 1506 }, { "epoch": 0.95, "learning_rate": 5.537178221609088e-06, "loss": 0.1553, "step": 1507 }, { "epoch": 0.95, "learning_rate": 5.532223308967835e-06, "loss": 0.1759, "step": 1508 }, { "epoch": 0.95, "learning_rate": 5.527267867612129e-06, "loss": 0.1561, "step": 1509 }, { "epoch": 0.95, "learning_rate": 5.5223119024647405e-06, "loss": 0.1877, "step": 1510 }, { "epoch": 0.95, "learning_rate": 5.517355418448961e-06, "loss": 0.1674, "step": 1511 }, { "epoch": 0.95, "learning_rate": 5.512398420488598e-06, "loss": 0.1456, "step": 1512 }, { "epoch": 0.95, "learning_rate": 5.5074409135079695e-06, "loss": 0.1669, "step": 1513 }, { "epoch": 0.95, "learning_rate": 5.502482902431897e-06, "loss": 0.1865, "step": 1514 }, { "epoch": 0.95, "learning_rate": 5.4975243921857025e-06, "loss": 0.16, "step": 1515 }, { "epoch": 0.95, "learning_rate": 5.49256538769521e-06, "loss": 0.1489, "step": 1516 }, { "epoch": 0.95, "learning_rate": 5.487605893886727e-06, "loss": 0.1579, "step": 1517 }, { "epoch": 0.95, "learning_rate": 5.482645915687052e-06, "loss": 0.1508, "step": 1518 }, { "epoch": 0.95, "learning_rate": 5.47768545802346e-06, "loss": 0.1656, "step": 1519 }, { "epoch": 0.95, "learning_rate": 5.472724525823708e-06, "loss": 0.164, "step": 1520 }, { "epoch": 0.96, "learning_rate": 5.467763124016023e-06, "loss": 0.1832, "step": 1521 }, { "epoch": 0.96, "learning_rate": 5.4628012575290934e-06, "loss": 0.1593, "step": 1522 }, { "epoch": 0.96, "learning_rate": 5.457838931292074e-06, "loss": 0.1532, "step": 1523 }, { "epoch": 0.96, "learning_rate": 5.452876150234577e-06, "loss": 0.1832, "step": 1524 }, { "epoch": 0.96, "learning_rate": 5.447912919286663e-06, "loss": 0.1715, "step": 1525 }, { "epoch": 0.96, "learning_rate": 5.442949243378843e-06, "loss": 0.1625, "step": 1526 }, { "epoch": 0.96, "learning_rate": 5.437985127442065e-06, "loss": 0.1449, "step": 1527 }, { "epoch": 0.96, "learning_rate": 5.4330205764077214e-06, "loss": 0.1512, "step": 1528 }, { "epoch": 0.96, "learning_rate": 5.428055595207631e-06, "loss": 0.1772, "step": 1529 }, { "epoch": 0.96, "learning_rate": 5.423090188774041e-06, "loss": 0.1372, "step": 1530 }, { "epoch": 0.96, "learning_rate": 5.418124362039621e-06, "loss": 0.1677, "step": 1531 }, { "epoch": 0.96, "learning_rate": 5.413158119937462e-06, "loss": 0.1694, "step": 1532 }, { "epoch": 0.96, "learning_rate": 5.4081914674010614e-06, "loss": 0.1689, "step": 1533 }, { "epoch": 0.96, "learning_rate": 5.40322440936433e-06, "loss": 0.205, "step": 1534 }, { "epoch": 0.96, "learning_rate": 5.398256950761578e-06, "loss": 0.182, "step": 1535 }, { "epoch": 0.96, "learning_rate": 5.393289096527514e-06, "loss": 0.192, "step": 1536 }, { "epoch": 0.97, "learning_rate": 5.38832085159724e-06, "loss": 0.1669, "step": 1537 }, { "epoch": 0.97, "learning_rate": 5.383352220906246e-06, "loss": 0.1734, "step": 1538 }, { "epoch": 0.97, "learning_rate": 5.3783832093904055e-06, "loss": 0.1893, "step": 1539 }, { "epoch": 0.97, "learning_rate": 5.373413821985971e-06, "loss": 0.1754, "step": 1540 }, { "epoch": 0.97, "learning_rate": 5.368444063629566e-06, "loss": 0.1527, "step": 1541 }, { "epoch": 0.97, "learning_rate": 5.363473939258187e-06, "loss": 0.1762, "step": 1542 }, { "epoch": 0.97, "learning_rate": 5.3585034538091885e-06, "loss": 0.182, "step": 1543 }, { "epoch": 0.97, "learning_rate": 5.353532612220288e-06, "loss": 0.1826, "step": 1544 }, { "epoch": 0.97, "learning_rate": 5.348561419429553e-06, "loss": 0.1854, "step": 1545 }, { "epoch": 0.97, "learning_rate": 5.343589880375404e-06, "loss": 0.1722, "step": 1546 }, { "epoch": 0.97, "learning_rate": 5.338617999996603e-06, "loss": 0.1589, "step": 1547 }, { "epoch": 0.97, "learning_rate": 5.333645783232252e-06, "loss": 0.1512, "step": 1548 }, { "epoch": 0.97, "learning_rate": 5.328673235021787e-06, "loss": 0.1658, "step": 1549 }, { "epoch": 0.97, "learning_rate": 5.323700360304971e-06, "loss": 0.1731, "step": 1550 }, { "epoch": 0.97, "learning_rate": 5.318727164021896e-06, "loss": 0.1555, "step": 1551 }, { "epoch": 0.97, "learning_rate": 5.313753651112969e-06, "loss": 0.1516, "step": 1552 }, { "epoch": 0.98, "learning_rate": 5.308779826518916e-06, "loss": 0.1689, "step": 1553 }, { "epoch": 0.98, "learning_rate": 5.303805695180767e-06, "loss": 0.1844, "step": 1554 }, { "epoch": 0.98, "learning_rate": 5.298831262039861e-06, "loss": 0.1877, "step": 1555 }, { "epoch": 0.98, "learning_rate": 5.293856532037837e-06, "loss": 0.1953, "step": 1556 }, { "epoch": 0.98, "learning_rate": 5.288881510116625e-06, "loss": 0.1572, "step": 1557 }, { "epoch": 0.98, "learning_rate": 5.2839062012184475e-06, "loss": 0.1568, "step": 1558 }, { "epoch": 0.98, "learning_rate": 5.278930610285813e-06, "loss": 0.1744, "step": 1559 }, { "epoch": 0.98, "learning_rate": 5.2739547422615114e-06, "loss": 0.1769, "step": 1560 }, { "epoch": 0.98, "learning_rate": 5.268978602088602e-06, "loss": 0.1854, "step": 1561 }, { "epoch": 0.98, "learning_rate": 5.26400219471042e-06, "loss": 0.157, "step": 1562 }, { "epoch": 0.98, "learning_rate": 5.259025525070565e-06, "loss": 0.1328, "step": 1563 }, { "epoch": 0.98, "learning_rate": 5.254048598112894e-06, "loss": 0.1638, "step": 1564 }, { "epoch": 0.98, "learning_rate": 5.249071418781524e-06, "loss": 0.1571, "step": 1565 }, { "epoch": 0.98, "learning_rate": 5.244093992020818e-06, "loss": 0.1667, "step": 1566 }, { "epoch": 0.98, "learning_rate": 5.239116322775392e-06, "loss": 0.1635, "step": 1567 }, { "epoch": 0.98, "learning_rate": 5.234138415990093e-06, "loss": 0.1534, "step": 1568 }, { "epoch": 0.99, "learning_rate": 5.229160276610015e-06, "loss": 0.1671, "step": 1569 }, { "epoch": 0.99, "learning_rate": 5.224181909580473e-06, "loss": 0.1541, "step": 1570 }, { "epoch": 0.99, "learning_rate": 5.219203319847012e-06, "loss": 0.2015, "step": 1571 }, { "epoch": 0.99, "learning_rate": 5.214224512355402e-06, "loss": 0.145, "step": 1572 }, { "epoch": 0.99, "learning_rate": 5.209245492051625e-06, "loss": 0.1655, "step": 1573 }, { "epoch": 0.99, "learning_rate": 5.204266263881874e-06, "loss": 0.1989, "step": 1574 }, { "epoch": 0.99, "learning_rate": 5.199286832792553e-06, "loss": 0.1411, "step": 1575 }, { "epoch": 0.99, "learning_rate": 5.194307203730261e-06, "loss": 0.1621, "step": 1576 }, { "epoch": 0.99, "learning_rate": 5.189327381641801e-06, "loss": 0.169, "step": 1577 }, { "epoch": 0.99, "learning_rate": 5.18434737147416e-06, "loss": 0.1693, "step": 1578 }, { "epoch": 0.99, "learning_rate": 5.179367178174518e-06, "loss": 0.1421, "step": 1579 }, { "epoch": 0.99, "learning_rate": 5.174386806690235e-06, "loss": 0.2041, "step": 1580 }, { "epoch": 0.99, "learning_rate": 5.169406261968847e-06, "loss": 0.1906, "step": 1581 }, { "epoch": 0.99, "learning_rate": 5.164425548958062e-06, "loss": 0.1487, "step": 1582 }, { "epoch": 0.99, "learning_rate": 5.159444672605759e-06, "loss": 0.1816, "step": 1583 }, { "epoch": 0.99, "learning_rate": 5.154463637859974e-06, "loss": 0.1779, "step": 1584 }, { "epoch": 1.0, "learning_rate": 5.149482449668903e-06, "loss": 0.1803, "step": 1585 }, { "epoch": 1.0, "learning_rate": 5.144501112980894e-06, "loss": 0.1743, "step": 1586 }, { "epoch": 1.0, "learning_rate": 5.139519632744443e-06, "loss": 0.1533, "step": 1587 }, { "epoch": 1.0, "learning_rate": 5.134538013908188e-06, "loss": 0.1944, "step": 1588 }, { "epoch": 1.0, "learning_rate": 5.129556261420904e-06, "loss": 0.1461, "step": 1589 }, { "epoch": 1.0, "learning_rate": 5.124574380231501e-06, "loss": 0.1533, "step": 1590 }, { "epoch": 1.0, "learning_rate": 5.119592375289015e-06, "loss": 0.1759, "step": 1591 }, { "epoch": 1.0, "learning_rate": 5.114610251542607e-06, "loss": 0.1487, "step": 1592 }, { "epoch": 1.0, "learning_rate": 5.109628013941551e-06, "loss": 0.1763, "step": 1593 }, { "epoch": 1.0, "learning_rate": 5.104645667435239e-06, "loss": 0.1652, "step": 1594 }, { "epoch": 1.0, "learning_rate": 5.099663216973172e-06, "loss": 0.1611, "step": 1595 }, { "epoch": 1.0, "learning_rate": 5.09468066750495e-06, "loss": 0.1586, "step": 1596 }, { "epoch": 1.0, "learning_rate": 5.089698023980273e-06, "loss": 0.1593, "step": 1597 }, { "epoch": 1.0, "learning_rate": 5.084715291348936e-06, "loss": 0.1676, "step": 1598 }, { "epoch": 1.0, "learning_rate": 5.079732474560821e-06, "loss": 0.1938, "step": 1599 }, { "epoch": 1.0, "learning_rate": 5.074749578565894e-06, "loss": 0.2122, "step": 1600 }, { "epoch": 1.01, "learning_rate": 5.069766608314199e-06, "loss": 0.1714, "step": 1601 }, { "epoch": 1.01, "learning_rate": 5.064783568755855e-06, "loss": 0.1569, "step": 1602 }, { "epoch": 1.01, "learning_rate": 5.059800464841048e-06, "loss": 0.1703, "step": 1603 }, { "epoch": 1.01, "learning_rate": 5.054817301520031e-06, "loss": 0.1554, "step": 1604 }, { "epoch": 1.01, "learning_rate": 5.0498340837431134e-06, "loss": 0.1546, "step": 1605 }, { "epoch": 1.01, "learning_rate": 5.044850816460661e-06, "loss": 0.145, "step": 1606 }, { "epoch": 1.01, "learning_rate": 5.039867504623084e-06, "loss": 0.1689, "step": 1607 }, { "epoch": 1.01, "learning_rate": 5.034884153180843e-06, "loss": 0.1923, "step": 1608 }, { "epoch": 1.01, "learning_rate": 5.029900767084434e-06, "loss": 0.1599, "step": 1609 }, { "epoch": 1.01, "learning_rate": 5.024917351284386e-06, "loss": 0.1878, "step": 1610 }, { "epoch": 1.01, "learning_rate": 5.019933910731264e-06, "loss": 0.1753, "step": 1611 }, { "epoch": 1.01, "learning_rate": 5.014950450375654e-06, "loss": 0.1645, "step": 1612 }, { "epoch": 1.01, "learning_rate": 5.009966975168157e-06, "loss": 0.1962, "step": 1613 }, { "epoch": 1.01, "learning_rate": 5.004983490059396e-06, "loss": 0.1863, "step": 1614 }, { "epoch": 1.01, "learning_rate": 5e-06, "loss": 0.188, "step": 1615 }, { "epoch": 1.01, "learning_rate": 4.995016509940605e-06, "loss": 0.1619, "step": 1616 }, { "epoch": 1.02, "learning_rate": 4.990033024831845e-06, "loss": 0.1838, "step": 1617 }, { "epoch": 1.02, "learning_rate": 4.985049549624347e-06, "loss": 0.1495, "step": 1618 }, { "epoch": 1.02, "learning_rate": 4.980066089268736e-06, "loss": 0.1536, "step": 1619 }, { "epoch": 1.02, "learning_rate": 4.975082648715615e-06, "loss": 0.1465, "step": 1620 }, { "epoch": 1.02, "learning_rate": 4.970099232915569e-06, "loss": 0.1649, "step": 1621 }, { "epoch": 1.02, "learning_rate": 4.965115846819158e-06, "loss": 0.1544, "step": 1622 }, { "epoch": 1.02, "learning_rate": 4.960132495376919e-06, "loss": 0.1502, "step": 1623 }, { "epoch": 1.02, "learning_rate": 4.955149183539342e-06, "loss": 0.1556, "step": 1624 }, { "epoch": 1.02, "learning_rate": 4.950165916256887e-06, "loss": 0.1994, "step": 1625 }, { "epoch": 1.02, "learning_rate": 4.94518269847997e-06, "loss": 0.1882, "step": 1626 }, { "epoch": 1.02, "learning_rate": 4.940199535158954e-06, "loss": 0.1463, "step": 1627 }, { "epoch": 1.02, "learning_rate": 4.935216431244148e-06, "loss": 0.1514, "step": 1628 }, { "epoch": 1.02, "learning_rate": 4.930233391685802e-06, "loss": 0.1776, "step": 1629 }, { "epoch": 1.02, "learning_rate": 4.9252504214341075e-06, "loss": 0.1897, "step": 1630 }, { "epoch": 1.02, "learning_rate": 4.92026752543918e-06, "loss": 0.1416, "step": 1631 }, { "epoch": 1.02, "learning_rate": 4.915284708651065e-06, "loss": 0.16, "step": 1632 }, { "epoch": 1.03, "learning_rate": 4.910301976019728e-06, "loss": 0.156, "step": 1633 }, { "epoch": 1.03, "learning_rate": 4.905319332495051e-06, "loss": 0.1478, "step": 1634 }, { "epoch": 1.03, "learning_rate": 4.9003367830268285e-06, "loss": 0.1366, "step": 1635 }, { "epoch": 1.03, "learning_rate": 4.895354332564761e-06, "loss": 0.183, "step": 1636 }, { "epoch": 1.03, "learning_rate": 4.890371986058452e-06, "loss": 0.2028, "step": 1637 }, { "epoch": 1.03, "learning_rate": 4.885389748457396e-06, "loss": 0.156, "step": 1638 }, { "epoch": 1.03, "learning_rate": 4.880407624710986e-06, "loss": 0.1849, "step": 1639 }, { "epoch": 1.03, "learning_rate": 4.875425619768501e-06, "loss": 0.1632, "step": 1640 }, { "epoch": 1.03, "learning_rate": 4.870443738579097e-06, "loss": 0.1662, "step": 1641 }, { "epoch": 1.03, "learning_rate": 4.865461986091814e-06, "loss": 0.1459, "step": 1642 }, { "epoch": 1.03, "learning_rate": 4.860480367255559e-06, "loss": 0.1523, "step": 1643 }, { "epoch": 1.03, "learning_rate": 4.855498887019108e-06, "loss": 0.1357, "step": 1644 }, { "epoch": 1.03, "learning_rate": 4.8505175503310975e-06, "loss": 0.1587, "step": 1645 }, { "epoch": 1.03, "learning_rate": 4.845536362140026e-06, "loss": 0.1721, "step": 1646 }, { "epoch": 1.03, "learning_rate": 4.8405553273942415e-06, "loss": 0.1596, "step": 1647 }, { "epoch": 1.04, "learning_rate": 4.835574451041938e-06, "loss": 0.1714, "step": 1648 }, { "epoch": 1.04, "learning_rate": 4.830593738031154e-06, "loss": 0.1552, "step": 1649 }, { "epoch": 1.04, "learning_rate": 4.825613193309765e-06, "loss": 0.1811, "step": 1650 }, { "epoch": 1.04, "learning_rate": 4.820632821825484e-06, "loss": 0.1445, "step": 1651 }, { "epoch": 1.04, "learning_rate": 4.8156526285258415e-06, "loss": 0.1888, "step": 1652 }, { "epoch": 1.04, "learning_rate": 4.810672618358202e-06, "loss": 0.1734, "step": 1653 }, { "epoch": 1.04, "learning_rate": 4.80569279626974e-06, "loss": 0.1785, "step": 1654 }, { "epoch": 1.04, "learning_rate": 4.800713167207449e-06, "loss": 0.1491, "step": 1655 }, { "epoch": 1.04, "learning_rate": 4.795733736118127e-06, "loss": 0.1667, "step": 1656 }, { "epoch": 1.04, "learning_rate": 4.7907545079483755e-06, "loss": 0.1412, "step": 1657 }, { "epoch": 1.04, "learning_rate": 4.7857754876445985e-06, "loss": 0.1676, "step": 1658 }, { "epoch": 1.04, "learning_rate": 4.780796680152988e-06, "loss": 0.1705, "step": 1659 }, { "epoch": 1.04, "learning_rate": 4.77581809041953e-06, "loss": 0.1781, "step": 1660 }, { "epoch": 1.04, "learning_rate": 4.770839723389986e-06, "loss": 0.1752, "step": 1661 }, { "epoch": 1.04, "learning_rate": 4.765861584009907e-06, "loss": 0.1713, "step": 1662 }, { "epoch": 1.04, "learning_rate": 4.760883677224609e-06, "loss": 0.1318, "step": 1663 }, { "epoch": 1.05, "learning_rate": 4.755906007979182e-06, "loss": 0.2033, "step": 1664 }, { "epoch": 1.05, "learning_rate": 4.7509285812184785e-06, "loss": 0.1712, "step": 1665 }, { "epoch": 1.05, "learning_rate": 4.745951401887108e-06, "loss": 0.164, "step": 1666 }, { "epoch": 1.05, "learning_rate": 4.740974474929438e-06, "loss": 0.1778, "step": 1667 }, { "epoch": 1.05, "learning_rate": 4.735997805289581e-06, "loss": 0.1617, "step": 1668 }, { "epoch": 1.05, "learning_rate": 4.7310213979113995e-06, "loss": 0.137, "step": 1669 }, { "epoch": 1.05, "learning_rate": 4.726045257738491e-06, "loss": 0.1666, "step": 1670 }, { "epoch": 1.05, "learning_rate": 4.721069389714188e-06, "loss": 0.1735, "step": 1671 }, { "epoch": 1.05, "learning_rate": 4.716093798781555e-06, "loss": 0.1466, "step": 1672 }, { "epoch": 1.05, "learning_rate": 4.7111184898833776e-06, "loss": 0.1714, "step": 1673 }, { "epoch": 1.05, "learning_rate": 4.706143467962165e-06, "loss": 0.1864, "step": 1674 }, { "epoch": 1.05, "learning_rate": 4.70116873796014e-06, "loss": 0.1459, "step": 1675 }, { "epoch": 1.05, "learning_rate": 4.696194304819235e-06, "loss": 0.1713, "step": 1676 }, { "epoch": 1.05, "learning_rate": 4.6912201734810845e-06, "loss": 0.1734, "step": 1677 }, { "epoch": 1.05, "learning_rate": 4.686246348887031e-06, "loss": 0.146, "step": 1678 }, { "epoch": 1.05, "learning_rate": 4.6812728359781064e-06, "loss": 0.1813, "step": 1679 }, { "epoch": 1.06, "learning_rate": 4.676299639695031e-06, "loss": 0.1576, "step": 1680 }, { "epoch": 1.06, "learning_rate": 4.671326764978216e-06, "loss": 0.159, "step": 1681 }, { "epoch": 1.06, "learning_rate": 4.66635421676775e-06, "loss": 0.1417, "step": 1682 }, { "epoch": 1.06, "learning_rate": 4.661382000003399e-06, "loss": 0.1991, "step": 1683 }, { "epoch": 1.06, "learning_rate": 4.656410119624597e-06, "loss": 0.1579, "step": 1684 }, { "epoch": 1.06, "learning_rate": 4.6514385805704486e-06, "loss": 0.1545, "step": 1685 }, { "epoch": 1.06, "learning_rate": 4.646467387779714e-06, "loss": 0.1411, "step": 1686 }, { "epoch": 1.06, "learning_rate": 4.641496546190813e-06, "loss": 0.1393, "step": 1687 }, { "epoch": 1.06, "learning_rate": 4.636526060741814e-06, "loss": 0.1391, "step": 1688 }, { "epoch": 1.06, "learning_rate": 4.631555936370433e-06, "loss": 0.147, "step": 1689 }, { "epoch": 1.06, "learning_rate": 4.626586178014029e-06, "loss": 0.1704, "step": 1690 }, { "epoch": 1.06, "learning_rate": 4.621616790609595e-06, "loss": 0.1453, "step": 1691 }, { "epoch": 1.06, "learning_rate": 4.616647779093755e-06, "loss": 0.1624, "step": 1692 }, { "epoch": 1.06, "learning_rate": 4.611679148402763e-06, "loss": 0.1338, "step": 1693 }, { "epoch": 1.06, "learning_rate": 4.606710903472489e-06, "loss": 0.179, "step": 1694 }, { "epoch": 1.06, "learning_rate": 4.601743049238425e-06, "loss": 0.139, "step": 1695 }, { "epoch": 1.07, "learning_rate": 4.596775590635671e-06, "loss": 0.1632, "step": 1696 }, { "epoch": 1.07, "learning_rate": 4.591808532598939e-06, "loss": 0.1807, "step": 1697 }, { "epoch": 1.07, "learning_rate": 4.58684188006254e-06, "loss": 0.1482, "step": 1698 }, { "epoch": 1.07, "learning_rate": 4.581875637960381e-06, "loss": 0.1804, "step": 1699 }, { "epoch": 1.07, "learning_rate": 4.576909811225961e-06, "loss": 0.1498, "step": 1700 }, { "epoch": 1.07, "learning_rate": 4.5719444047923705e-06, "loss": 0.1705, "step": 1701 }, { "epoch": 1.07, "learning_rate": 4.566979423592279e-06, "loss": 0.1613, "step": 1702 }, { "epoch": 1.07, "learning_rate": 4.562014872557936e-06, "loss": 0.1426, "step": 1703 }, { "epoch": 1.07, "learning_rate": 4.5570507566211585e-06, "loss": 0.172, "step": 1704 }, { "epoch": 1.07, "learning_rate": 4.552087080713337e-06, "loss": 0.1644, "step": 1705 }, { "epoch": 1.07, "learning_rate": 4.547123849765423e-06, "loss": 0.1696, "step": 1706 }, { "epoch": 1.07, "learning_rate": 4.542161068707927e-06, "loss": 0.1687, "step": 1707 }, { "epoch": 1.07, "learning_rate": 4.537198742470908e-06, "loss": 0.1524, "step": 1708 }, { "epoch": 1.07, "learning_rate": 4.532236875983979e-06, "loss": 0.1607, "step": 1709 }, { "epoch": 1.07, "learning_rate": 4.527275474176293e-06, "loss": 0.1576, "step": 1710 }, { "epoch": 1.07, "learning_rate": 4.522314541976541e-06, "loss": 0.1635, "step": 1711 }, { "epoch": 1.08, "learning_rate": 4.51735408431295e-06, "loss": 0.1716, "step": 1712 }, { "epoch": 1.08, "learning_rate": 4.512394106113274e-06, "loss": 0.1482, "step": 1713 }, { "epoch": 1.08, "learning_rate": 4.507434612304792e-06, "loss": 0.1355, "step": 1714 }, { "epoch": 1.08, "learning_rate": 4.502475607814298e-06, "loss": 0.1788, "step": 1715 }, { "epoch": 1.08, "learning_rate": 4.497517097568104e-06, "loss": 0.1448, "step": 1716 }, { "epoch": 1.08, "learning_rate": 4.492559086492031e-06, "loss": 0.1351, "step": 1717 }, { "epoch": 1.08, "learning_rate": 4.487601579511402e-06, "loss": 0.1458, "step": 1718 }, { "epoch": 1.08, "learning_rate": 4.48264458155104e-06, "loss": 0.1661, "step": 1719 }, { "epoch": 1.08, "learning_rate": 4.47768809753526e-06, "loss": 0.1686, "step": 1720 }, { "epoch": 1.08, "learning_rate": 4.472732132387874e-06, "loss": 0.1831, "step": 1721 }, { "epoch": 1.08, "learning_rate": 4.467776691032168e-06, "loss": 0.1877, "step": 1722 }, { "epoch": 1.08, "learning_rate": 4.462821778390914e-06, "loss": 0.1624, "step": 1723 }, { "epoch": 1.08, "learning_rate": 4.457867399386361e-06, "loss": 0.1747, "step": 1724 }, { "epoch": 1.08, "learning_rate": 4.4529135589402206e-06, "loss": 0.1587, "step": 1725 }, { "epoch": 1.08, "learning_rate": 4.447960261973676e-06, "loss": 0.1589, "step": 1726 }, { "epoch": 1.08, "learning_rate": 4.443007513407368e-06, "loss": 0.1507, "step": 1727 }, { "epoch": 1.09, "learning_rate": 4.43805531816139e-06, "loss": 0.1618, "step": 1728 }, { "epoch": 1.09, "learning_rate": 4.433103681155292e-06, "loss": 0.1544, "step": 1729 }, { "epoch": 1.09, "learning_rate": 4.4281526073080624e-06, "loss": 0.1306, "step": 1730 }, { "epoch": 1.09, "learning_rate": 4.423202101538136e-06, "loss": 0.1697, "step": 1731 }, { "epoch": 1.09, "learning_rate": 4.418252168763379e-06, "loss": 0.1537, "step": 1732 }, { "epoch": 1.09, "learning_rate": 4.413302813901091e-06, "loss": 0.1354, "step": 1733 }, { "epoch": 1.09, "learning_rate": 4.408354041867997e-06, "loss": 0.1283, "step": 1734 }, { "epoch": 1.09, "learning_rate": 4.403405857580243e-06, "loss": 0.1559, "step": 1735 }, { "epoch": 1.09, "learning_rate": 4.398458265953388e-06, "loss": 0.1734, "step": 1736 }, { "epoch": 1.09, "learning_rate": 4.393511271902407e-06, "loss": 0.1548, "step": 1737 }, { "epoch": 1.09, "learning_rate": 4.38856488034168e-06, "loss": 0.1628, "step": 1738 }, { "epoch": 1.09, "learning_rate": 4.383619096184985e-06, "loss": 0.144, "step": 1739 }, { "epoch": 1.09, "learning_rate": 4.378673924345503e-06, "loss": 0.1596, "step": 1740 }, { "epoch": 1.09, "learning_rate": 4.373729369735802e-06, "loss": 0.1674, "step": 1741 }, { "epoch": 1.09, "learning_rate": 4.368785437267837e-06, "loss": 0.1526, "step": 1742 }, { "epoch": 1.09, "learning_rate": 4.363842131852948e-06, "loss": 0.1888, "step": 1743 }, { "epoch": 1.1, "learning_rate": 4.358899458401846e-06, "loss": 0.1524, "step": 1744 }, { "epoch": 1.1, "learning_rate": 4.353957421824624e-06, "loss": 0.1305, "step": 1745 }, { "epoch": 1.1, "learning_rate": 4.349016027030732e-06, "loss": 0.1389, "step": 1746 }, { "epoch": 1.1, "learning_rate": 4.344075278928989e-06, "loss": 0.1755, "step": 1747 }, { "epoch": 1.1, "learning_rate": 4.339135182427568e-06, "loss": 0.1365, "step": 1748 }, { "epoch": 1.1, "learning_rate": 4.334195742434002e-06, "loss": 0.1674, "step": 1749 }, { "epoch": 1.1, "learning_rate": 4.329256963855159e-06, "loss": 0.148, "step": 1750 }, { "epoch": 1.1, "learning_rate": 4.3243188515972575e-06, "loss": 0.1376, "step": 1751 }, { "epoch": 1.1, "learning_rate": 4.3193814105658585e-06, "loss": 0.1163, "step": 1752 }, { "epoch": 1.1, "learning_rate": 4.314444645665848e-06, "loss": 0.1626, "step": 1753 }, { "epoch": 1.1, "learning_rate": 4.309508561801444e-06, "loss": 0.1838, "step": 1754 }, { "epoch": 1.1, "learning_rate": 4.304573163876188e-06, "loss": 0.1838, "step": 1755 }, { "epoch": 1.1, "learning_rate": 4.29963845679294e-06, "loss": 0.1361, "step": 1756 }, { "epoch": 1.1, "learning_rate": 4.294704445453873e-06, "loss": 0.1824, "step": 1757 }, { "epoch": 1.1, "learning_rate": 4.28977113476047e-06, "loss": 0.1527, "step": 1758 }, { "epoch": 1.1, "learning_rate": 4.2848385296135165e-06, "loss": 0.1608, "step": 1759 }, { "epoch": 1.11, "learning_rate": 4.279906634913097e-06, "loss": 0.1689, "step": 1760 }, { "epoch": 1.11, "learning_rate": 4.274975455558593e-06, "loss": 0.1907, "step": 1761 }, { "epoch": 1.11, "learning_rate": 4.270044996448672e-06, "loss": 0.1385, "step": 1762 }, { "epoch": 1.11, "learning_rate": 4.265115262481289e-06, "loss": 0.1626, "step": 1763 }, { "epoch": 1.11, "learning_rate": 4.260186258553676e-06, "loss": 0.1167, "step": 1764 }, { "epoch": 1.11, "learning_rate": 4.25525798956234e-06, "loss": 0.157, "step": 1765 }, { "epoch": 1.11, "learning_rate": 4.2503304604030595e-06, "loss": 0.1427, "step": 1766 }, { "epoch": 1.11, "learning_rate": 4.245403675970877e-06, "loss": 0.1372, "step": 1767 }, { "epoch": 1.11, "learning_rate": 4.240477641160096e-06, "loss": 0.1526, "step": 1768 }, { "epoch": 1.11, "learning_rate": 4.235552360864276e-06, "loss": 0.1409, "step": 1769 }, { "epoch": 1.11, "learning_rate": 4.230627839976224e-06, "loss": 0.1509, "step": 1770 }, { "epoch": 1.11, "learning_rate": 4.225704083387993e-06, "loss": 0.1611, "step": 1771 }, { "epoch": 1.11, "learning_rate": 4.220781095990883e-06, "loss": 0.1631, "step": 1772 }, { "epoch": 1.11, "learning_rate": 4.215858882675421e-06, "loss": 0.1515, "step": 1773 }, { "epoch": 1.11, "learning_rate": 4.21093744833137e-06, "loss": 0.1697, "step": 1774 }, { "epoch": 1.11, "learning_rate": 4.206016797847718e-06, "loss": 0.1665, "step": 1775 }, { "epoch": 1.12, "learning_rate": 4.201096936112675e-06, "loss": 0.1521, "step": 1776 }, { "epoch": 1.12, "learning_rate": 4.196177868013669e-06, "loss": 0.1299, "step": 1777 }, { "epoch": 1.12, "learning_rate": 4.191259598437332e-06, "loss": 0.1381, "step": 1778 }, { "epoch": 1.12, "learning_rate": 4.186342132269512e-06, "loss": 0.1782, "step": 1779 }, { "epoch": 1.12, "learning_rate": 4.181425474395255e-06, "loss": 0.1714, "step": 1780 }, { "epoch": 1.12, "learning_rate": 4.1765096296988046e-06, "loss": 0.1429, "step": 1781 }, { "epoch": 1.12, "learning_rate": 4.171594603063594e-06, "loss": 0.1802, "step": 1782 }, { "epoch": 1.12, "learning_rate": 4.166680399372248e-06, "loss": 0.1609, "step": 1783 }, { "epoch": 1.12, "learning_rate": 4.161767023506571e-06, "loss": 0.1578, "step": 1784 }, { "epoch": 1.12, "learning_rate": 4.156854480347547e-06, "loss": 0.1414, "step": 1785 }, { "epoch": 1.12, "learning_rate": 4.15194277477533e-06, "loss": 0.1441, "step": 1786 }, { "epoch": 1.12, "learning_rate": 4.147031911669243e-06, "loss": 0.1272, "step": 1787 }, { "epoch": 1.12, "learning_rate": 4.1421218959077746e-06, "loss": 0.1483, "step": 1788 }, { "epoch": 1.12, "learning_rate": 4.1372127323685694e-06, "loss": 0.1745, "step": 1789 }, { "epoch": 1.12, "learning_rate": 4.132304425928425e-06, "loss": 0.1437, "step": 1790 }, { "epoch": 1.12, "learning_rate": 4.127396981463289e-06, "loss": 0.1583, "step": 1791 }, { "epoch": 1.13, "learning_rate": 4.122490403848252e-06, "loss": 0.1598, "step": 1792 }, { "epoch": 1.13, "learning_rate": 4.11758469795754e-06, "loss": 0.1387, "step": 1793 }, { "epoch": 1.13, "learning_rate": 4.11267986866452e-06, "loss": 0.1538, "step": 1794 }, { "epoch": 1.13, "learning_rate": 4.1077759208416845e-06, "loss": 0.1461, "step": 1795 }, { "epoch": 1.13, "learning_rate": 4.10287285936065e-06, "loss": 0.151, "step": 1796 }, { "epoch": 1.13, "learning_rate": 4.0979706890921515e-06, "loss": 0.1627, "step": 1797 }, { "epoch": 1.13, "learning_rate": 4.093069414906042e-06, "loss": 0.1542, "step": 1798 }, { "epoch": 1.13, "learning_rate": 4.08816904167128e-06, "loss": 0.1631, "step": 1799 }, { "epoch": 1.13, "learning_rate": 4.0832695742559356e-06, "loss": 0.1408, "step": 1800 }, { "epoch": 1.13, "learning_rate": 4.078371017527171e-06, "loss": 0.1731, "step": 1801 }, { "epoch": 1.13, "learning_rate": 4.07347337635125e-06, "loss": 0.1795, "step": 1802 }, { "epoch": 1.13, "learning_rate": 4.068576655593522e-06, "loss": 0.1643, "step": 1803 }, { "epoch": 1.13, "learning_rate": 4.063680860118427e-06, "loss": 0.1477, "step": 1804 }, { "epoch": 1.13, "learning_rate": 4.0587859947894845e-06, "loss": 0.1628, "step": 1805 }, { "epoch": 1.13, "learning_rate": 4.053892064469283e-06, "loss": 0.1428, "step": 1806 }, { "epoch": 1.13, "learning_rate": 4.048999074019493e-06, "loss": 0.1578, "step": 1807 }, { "epoch": 1.14, "learning_rate": 4.044107028300845e-06, "loss": 0.1419, "step": 1808 }, { "epoch": 1.14, "learning_rate": 4.039215932173133e-06, "loss": 0.166, "step": 1809 }, { "epoch": 1.14, "learning_rate": 4.034325790495204e-06, "loss": 0.1766, "step": 1810 }, { "epoch": 1.14, "learning_rate": 4.029436608124963e-06, "loss": 0.1586, "step": 1811 }, { "epoch": 1.14, "learning_rate": 4.02454838991936e-06, "loss": 0.1591, "step": 1812 }, { "epoch": 1.14, "learning_rate": 4.0196611407343825e-06, "loss": 0.1452, "step": 1813 }, { "epoch": 1.14, "learning_rate": 4.01477486542506e-06, "loss": 0.1378, "step": 1814 }, { "epoch": 1.14, "learning_rate": 4.009889568845453e-06, "loss": 0.1135, "step": 1815 }, { "epoch": 1.14, "learning_rate": 4.0050052558486515e-06, "loss": 0.1507, "step": 1816 }, { "epoch": 1.14, "learning_rate": 4.000121931286768e-06, "loss": 0.1806, "step": 1817 }, { "epoch": 1.14, "learning_rate": 3.995239600010929e-06, "loss": 0.1643, "step": 1818 }, { "epoch": 1.14, "learning_rate": 3.990358266871279e-06, "loss": 0.1415, "step": 1819 }, { "epoch": 1.14, "learning_rate": 3.985477936716973e-06, "loss": 0.1398, "step": 1820 }, { "epoch": 1.14, "learning_rate": 3.98059861439616e-06, "loss": 0.1544, "step": 1821 }, { "epoch": 1.14, "learning_rate": 3.975720304755995e-06, "loss": 0.1355, "step": 1822 }, { "epoch": 1.14, "learning_rate": 3.9708430126426286e-06, "loss": 0.1408, "step": 1823 }, { "epoch": 1.15, "learning_rate": 3.965966742901195e-06, "loss": 0.1389, "step": 1824 }, { "epoch": 1.15, "learning_rate": 3.961091500375818e-06, "loss": 0.1268, "step": 1825 }, { "epoch": 1.15, "learning_rate": 3.9562172899095955e-06, "loss": 0.1163, "step": 1826 }, { "epoch": 1.15, "learning_rate": 3.951344116344606e-06, "loss": 0.165, "step": 1827 }, { "epoch": 1.15, "learning_rate": 3.946471984521894e-06, "loss": 0.1453, "step": 1828 }, { "epoch": 1.15, "learning_rate": 3.94160089928147e-06, "loss": 0.1584, "step": 1829 }, { "epoch": 1.15, "learning_rate": 3.936730865462306e-06, "loss": 0.1294, "step": 1830 }, { "epoch": 1.15, "learning_rate": 3.9318618879023255e-06, "loss": 0.1627, "step": 1831 }, { "epoch": 1.15, "learning_rate": 3.92699397143841e-06, "loss": 0.1226, "step": 1832 }, { "epoch": 1.15, "learning_rate": 3.92212712090638e-06, "loss": 0.1669, "step": 1833 }, { "epoch": 1.15, "learning_rate": 3.917261341141001e-06, "loss": 0.1619, "step": 1834 }, { "epoch": 1.15, "learning_rate": 3.912396636975973e-06, "loss": 0.1334, "step": 1835 }, { "epoch": 1.15, "learning_rate": 3.907533013243927e-06, "loss": 0.1571, "step": 1836 }, { "epoch": 1.15, "learning_rate": 3.902670474776423e-06, "loss": 0.1453, "step": 1837 }, { "epoch": 1.15, "learning_rate": 3.8978090264039395e-06, "loss": 0.1752, "step": 1838 }, { "epoch": 1.15, "learning_rate": 3.8929486729558775e-06, "loss": 0.1856, "step": 1839 }, { "epoch": 1.16, "learning_rate": 3.888089419260545e-06, "loss": 0.1678, "step": 1840 }, { "epoch": 1.16, "learning_rate": 3.88323127014516e-06, "loss": 0.1484, "step": 1841 }, { "epoch": 1.16, "learning_rate": 3.878374230435841e-06, "loss": 0.1686, "step": 1842 }, { "epoch": 1.16, "learning_rate": 3.873518304957609e-06, "loss": 0.158, "step": 1843 }, { "epoch": 1.16, "learning_rate": 3.868663498534374e-06, "loss": 0.1339, "step": 1844 }, { "epoch": 1.16, "learning_rate": 3.8638098159889346e-06, "loss": 0.144, "step": 1845 }, { "epoch": 1.16, "learning_rate": 3.858957262142975e-06, "loss": 0.1357, "step": 1846 }, { "epoch": 1.16, "learning_rate": 3.854105841817056e-06, "loss": 0.1661, "step": 1847 }, { "epoch": 1.16, "learning_rate": 3.849255559830617e-06, "loss": 0.1609, "step": 1848 }, { "epoch": 1.16, "learning_rate": 3.844406421001955e-06, "loss": 0.1621, "step": 1849 }, { "epoch": 1.16, "learning_rate": 3.839558430148247e-06, "loss": 0.1441, "step": 1850 }, { "epoch": 1.16, "learning_rate": 3.834711592085519e-06, "loss": 0.139, "step": 1851 }, { "epoch": 1.16, "learning_rate": 3.829865911628654e-06, "loss": 0.136, "step": 1852 }, { "epoch": 1.16, "learning_rate": 3.825021393591386e-06, "loss": 0.1457, "step": 1853 }, { "epoch": 1.16, "learning_rate": 3.820178042786295e-06, "loss": 0.1812, "step": 1854 }, { "epoch": 1.17, "learning_rate": 3.8153358640248e-06, "loss": 0.145, "step": 1855 }, { "epoch": 1.17, "learning_rate": 3.8104948621171556e-06, "loss": 0.1381, "step": 1856 }, { "epoch": 1.17, "learning_rate": 3.8056550418724487e-06, "loss": 0.1585, "step": 1857 }, { "epoch": 1.17, "learning_rate": 3.8008164080985895e-06, "loss": 0.1181, "step": 1858 }, { "epoch": 1.17, "learning_rate": 3.7959789656023145e-06, "loss": 0.1305, "step": 1859 }, { "epoch": 1.17, "learning_rate": 3.791142719189172e-06, "loss": 0.1198, "step": 1860 }, { "epoch": 1.17, "learning_rate": 3.786307673663524e-06, "loss": 0.1737, "step": 1861 }, { "epoch": 1.17, "learning_rate": 3.7814738338285405e-06, "loss": 0.154, "step": 1862 }, { "epoch": 1.17, "learning_rate": 3.776641204486191e-06, "loss": 0.1683, "step": 1863 }, { "epoch": 1.17, "learning_rate": 3.7718097904372445e-06, "loss": 0.1658, "step": 1864 }, { "epoch": 1.17, "learning_rate": 3.7669795964812623e-06, "loss": 0.1543, "step": 1865 }, { "epoch": 1.17, "learning_rate": 3.762150627416595e-06, "loss": 0.1763, "step": 1866 }, { "epoch": 1.17, "learning_rate": 3.7573228880403734e-06, "loss": 0.1247, "step": 1867 }, { "epoch": 1.17, "learning_rate": 3.75249638314851e-06, "loss": 0.1259, "step": 1868 }, { "epoch": 1.17, "learning_rate": 3.7476711175356885e-06, "loss": 0.1308, "step": 1869 }, { "epoch": 1.17, "learning_rate": 3.742847095995362e-06, "loss": 0.1766, "step": 1870 }, { "epoch": 1.18, "learning_rate": 3.738024323319749e-06, "loss": 0.1598, "step": 1871 }, { "epoch": 1.18, "learning_rate": 3.7332028042998265e-06, "loss": 0.1347, "step": 1872 }, { "epoch": 1.18, "learning_rate": 3.728382543725326e-06, "loss": 0.1563, "step": 1873 }, { "epoch": 1.18, "learning_rate": 3.7235635463847265e-06, "loss": 0.1272, "step": 1874 }, { "epoch": 1.18, "learning_rate": 3.718745817065259e-06, "loss": 0.1251, "step": 1875 }, { "epoch": 1.18, "learning_rate": 3.7139293605528892e-06, "loss": 0.1706, "step": 1876 }, { "epoch": 1.18, "learning_rate": 3.7091141816323155e-06, "loss": 0.1552, "step": 1877 }, { "epoch": 1.18, "learning_rate": 3.7043002850869754e-06, "loss": 0.1472, "step": 1878 }, { "epoch": 1.18, "learning_rate": 3.699487675699027e-06, "loss": 0.1232, "step": 1879 }, { "epoch": 1.18, "learning_rate": 3.6946763582493504e-06, "loss": 0.1551, "step": 1880 }, { "epoch": 1.18, "learning_rate": 3.6898663375175424e-06, "loss": 0.16, "step": 1881 }, { "epoch": 1.18, "learning_rate": 3.685057618281914e-06, "loss": 0.157, "step": 1882 }, { "epoch": 1.18, "learning_rate": 3.680250205319481e-06, "loss": 0.1523, "step": 1883 }, { "epoch": 1.18, "learning_rate": 3.6754441034059613e-06, "loss": 0.1398, "step": 1884 }, { "epoch": 1.18, "learning_rate": 3.6706393173157708e-06, "loss": 0.1384, "step": 1885 }, { "epoch": 1.18, "learning_rate": 3.6658358518220178e-06, "loss": 0.1493, "step": 1886 }, { "epoch": 1.19, "learning_rate": 3.661033711696501e-06, "loss": 0.1596, "step": 1887 }, { "epoch": 1.19, "learning_rate": 3.656232901709701e-06, "loss": 0.1332, "step": 1888 }, { "epoch": 1.19, "learning_rate": 3.651433426630776e-06, "loss": 0.1558, "step": 1889 }, { "epoch": 1.19, "learning_rate": 3.64663529122756e-06, "loss": 0.157, "step": 1890 }, { "epoch": 1.19, "learning_rate": 3.641838500266553e-06, "loss": 0.1513, "step": 1891 }, { "epoch": 1.19, "learning_rate": 3.6370430585129223e-06, "loss": 0.1575, "step": 1892 }, { "epoch": 1.19, "learning_rate": 3.632248970730494e-06, "loss": 0.149, "step": 1893 }, { "epoch": 1.19, "learning_rate": 3.6274562416817514e-06, "loss": 0.1553, "step": 1894 }, { "epoch": 1.19, "learning_rate": 3.6226648761278238e-06, "loss": 0.1417, "step": 1895 }, { "epoch": 1.19, "learning_rate": 3.617874878828489e-06, "loss": 0.1636, "step": 1896 }, { "epoch": 1.19, "learning_rate": 3.613086254542163e-06, "loss": 0.1572, "step": 1897 }, { "epoch": 1.19, "learning_rate": 3.608299008025902e-06, "loss": 0.1385, "step": 1898 }, { "epoch": 1.19, "learning_rate": 3.60351314403539e-06, "loss": 0.129, "step": 1899 }, { "epoch": 1.19, "learning_rate": 3.598728667324939e-06, "loss": 0.1577, "step": 1900 }, { "epoch": 1.19, "learning_rate": 3.5939455826474813e-06, "loss": 0.1414, "step": 1901 }, { "epoch": 1.19, "learning_rate": 3.589163894754567e-06, "loss": 0.1576, "step": 1902 }, { "epoch": 1.2, "learning_rate": 3.5843836083963625e-06, "loss": 0.1825, "step": 1903 }, { "epoch": 1.2, "learning_rate": 3.5796047283216373e-06, "loss": 0.1586, "step": 1904 }, { "epoch": 1.2, "learning_rate": 3.574827259277763e-06, "loss": 0.1733, "step": 1905 }, { "epoch": 1.2, "learning_rate": 3.570051206010713e-06, "loss": 0.1456, "step": 1906 }, { "epoch": 1.2, "learning_rate": 3.5652765732650523e-06, "loss": 0.1302, "step": 1907 }, { "epoch": 1.2, "learning_rate": 3.5605033657839365e-06, "loss": 0.1307, "step": 1908 }, { "epoch": 1.2, "learning_rate": 3.5557315883091026e-06, "loss": 0.1342, "step": 1909 }, { "epoch": 1.2, "learning_rate": 3.5509612455808697e-06, "loss": 0.1457, "step": 1910 }, { "epoch": 1.2, "learning_rate": 3.5461923423381313e-06, "loss": 0.1496, "step": 1911 }, { "epoch": 1.2, "learning_rate": 3.5414248833183483e-06, "loss": 0.1691, "step": 1912 }, { "epoch": 1.2, "learning_rate": 3.5366588732575486e-06, "loss": 0.1373, "step": 1913 }, { "epoch": 1.2, "learning_rate": 3.531894316890323e-06, "loss": 0.1324, "step": 1914 }, { "epoch": 1.2, "learning_rate": 3.527131218949814e-06, "loss": 0.1326, "step": 1915 }, { "epoch": 1.2, "learning_rate": 3.522369584167718e-06, "loss": 0.1758, "step": 1916 }, { "epoch": 1.2, "learning_rate": 3.517609417274277e-06, "loss": 0.1564, "step": 1917 }, { "epoch": 1.2, "learning_rate": 3.5128507229982766e-06, "loss": 0.1255, "step": 1918 }, { "epoch": 1.21, "learning_rate": 3.5080935060670345e-06, "loss": 0.1432, "step": 1919 }, { "epoch": 1.21, "learning_rate": 3.5033377712064053e-06, "loss": 0.1312, "step": 1920 }, { "epoch": 1.21, "learning_rate": 3.4985835231407713e-06, "loss": 0.1393, "step": 1921 }, { "epoch": 1.21, "learning_rate": 3.493830766593036e-06, "loss": 0.1503, "step": 1922 }, { "epoch": 1.21, "learning_rate": 3.4890795062846215e-06, "loss": 0.1516, "step": 1923 }, { "epoch": 1.21, "learning_rate": 3.484329746935465e-06, "loss": 0.1374, "step": 1924 }, { "epoch": 1.21, "learning_rate": 3.4795814932640094e-06, "loss": 0.1439, "step": 1925 }, { "epoch": 1.21, "learning_rate": 3.474834749987206e-06, "loss": 0.1352, "step": 1926 }, { "epoch": 1.21, "learning_rate": 3.4700895218205026e-06, "loss": 0.153, "step": 1927 }, { "epoch": 1.21, "learning_rate": 3.465345813477844e-06, "loss": 0.1337, "step": 1928 }, { "epoch": 1.21, "learning_rate": 3.4606036296716614e-06, "loss": 0.1562, "step": 1929 }, { "epoch": 1.21, "learning_rate": 3.455862975112878e-06, "loss": 0.1552, "step": 1930 }, { "epoch": 1.21, "learning_rate": 3.4511238545108915e-06, "loss": 0.1275, "step": 1931 }, { "epoch": 1.21, "learning_rate": 3.44638627257358e-06, "loss": 0.1635, "step": 1932 }, { "epoch": 1.21, "learning_rate": 3.44165023400729e-06, "loss": 0.1313, "step": 1933 }, { "epoch": 1.21, "learning_rate": 3.4369157435168355e-06, "loss": 0.1472, "step": 1934 }, { "epoch": 1.22, "learning_rate": 3.432182805805495e-06, "loss": 0.1493, "step": 1935 }, { "epoch": 1.22, "learning_rate": 3.4274514255750008e-06, "loss": 0.152, "step": 1936 }, { "epoch": 1.22, "learning_rate": 3.422721607525542e-06, "loss": 0.1136, "step": 1937 }, { "epoch": 1.22, "learning_rate": 3.4179933563557536e-06, "loss": 0.1341, "step": 1938 }, { "epoch": 1.22, "learning_rate": 3.4132666767627136e-06, "loss": 0.1464, "step": 1939 }, { "epoch": 1.22, "learning_rate": 3.40854157344194e-06, "loss": 0.1487, "step": 1940 }, { "epoch": 1.22, "learning_rate": 3.403818051087382e-06, "loss": 0.1337, "step": 1941 }, { "epoch": 1.22, "learning_rate": 3.3990961143914235e-06, "loss": 0.1717, "step": 1942 }, { "epoch": 1.22, "learning_rate": 3.3943757680448697e-06, "loss": 0.164, "step": 1943 }, { "epoch": 1.22, "learning_rate": 3.389657016736945e-06, "loss": 0.1722, "step": 1944 }, { "epoch": 1.22, "learning_rate": 3.3849398651552917e-06, "loss": 0.1321, "step": 1945 }, { "epoch": 1.22, "learning_rate": 3.3802243179859653e-06, "loss": 0.1441, "step": 1946 }, { "epoch": 1.22, "learning_rate": 3.375510379913418e-06, "loss": 0.1343, "step": 1947 }, { "epoch": 1.22, "learning_rate": 3.370798055620513e-06, "loss": 0.1481, "step": 1948 }, { "epoch": 1.22, "learning_rate": 3.3660873497885072e-06, "loss": 0.159, "step": 1949 }, { "epoch": 1.22, "learning_rate": 3.361378267097049e-06, "loss": 0.1398, "step": 1950 }, { "epoch": 1.23, "learning_rate": 3.3566708122241753e-06, "loss": 0.1422, "step": 1951 }, { "epoch": 1.23, "learning_rate": 3.3519649898463043e-06, "loss": 0.1457, "step": 1952 }, { "epoch": 1.23, "learning_rate": 3.3472608046382353e-06, "loss": 0.1803, "step": 1953 }, { "epoch": 1.23, "learning_rate": 3.34255826127314e-06, "loss": 0.1154, "step": 1954 }, { "epoch": 1.23, "learning_rate": 3.337857364422557e-06, "loss": 0.1505, "step": 1955 }, { "epoch": 1.23, "learning_rate": 3.333158118756392e-06, "loss": 0.1363, "step": 1956 }, { "epoch": 1.23, "learning_rate": 3.3284605289429077e-06, "loss": 0.1688, "step": 1957 }, { "epoch": 1.23, "learning_rate": 3.3237645996487267e-06, "loss": 0.1513, "step": 1958 }, { "epoch": 1.23, "learning_rate": 3.3190703355388165e-06, "loss": 0.1538, "step": 1959 }, { "epoch": 1.23, "learning_rate": 3.3143777412764944e-06, "loss": 0.1242, "step": 1960 }, { "epoch": 1.23, "learning_rate": 3.3096868215234145e-06, "loss": 0.1527, "step": 1961 }, { "epoch": 1.23, "learning_rate": 3.304997580939573e-06, "loss": 0.1432, "step": 1962 }, { "epoch": 1.23, "learning_rate": 3.300310024183293e-06, "loss": 0.1508, "step": 1963 }, { "epoch": 1.23, "learning_rate": 3.295624155911227e-06, "loss": 0.1497, "step": 1964 }, { "epoch": 1.23, "learning_rate": 3.2909399807783516e-06, "loss": 0.1562, "step": 1965 }, { "epoch": 1.23, "learning_rate": 3.2862575034379595e-06, "loss": 0.1423, "step": 1966 }, { "epoch": 1.24, "learning_rate": 3.2815767285416576e-06, "loss": 0.173, "step": 1967 }, { "epoch": 1.24, "learning_rate": 3.2768976607393597e-06, "loss": 0.1294, "step": 1968 }, { "epoch": 1.24, "learning_rate": 3.2722203046792875e-06, "loss": 0.1582, "step": 1969 }, { "epoch": 1.24, "learning_rate": 3.2675446650079593e-06, "loss": 0.1183, "step": 1970 }, { "epoch": 1.24, "learning_rate": 3.262870746370189e-06, "loss": 0.1494, "step": 1971 }, { "epoch": 1.24, "learning_rate": 3.2581985534090798e-06, "loss": 0.14, "step": 1972 }, { "epoch": 1.24, "learning_rate": 3.2535280907660226e-06, "loss": 0.1647, "step": 1973 }, { "epoch": 1.24, "learning_rate": 3.248859363080691e-06, "loss": 0.1126, "step": 1974 }, { "epoch": 1.24, "learning_rate": 3.244192374991027e-06, "loss": 0.1396, "step": 1975 }, { "epoch": 1.24, "learning_rate": 3.2395271311332543e-06, "loss": 0.1302, "step": 1976 }, { "epoch": 1.24, "learning_rate": 3.2348636361418585e-06, "loss": 0.1595, "step": 1977 }, { "epoch": 1.24, "learning_rate": 3.230201894649589e-06, "loss": 0.1453, "step": 1978 }, { "epoch": 1.24, "learning_rate": 3.2255419112874533e-06, "loss": 0.1384, "step": 1979 }, { "epoch": 1.24, "learning_rate": 3.220883690684711e-06, "loss": 0.1284, "step": 1980 }, { "epoch": 1.24, "learning_rate": 3.2162272374688736e-06, "loss": 0.1473, "step": 1981 }, { "epoch": 1.24, "learning_rate": 3.211572556265695e-06, "loss": 0.148, "step": 1982 }, { "epoch": 1.25, "learning_rate": 3.2069196516991685e-06, "loss": 0.1505, "step": 1983 }, { "epoch": 1.25, "learning_rate": 3.202268528391521e-06, "loss": 0.1311, "step": 1984 }, { "epoch": 1.25, "learning_rate": 3.197619190963216e-06, "loss": 0.1486, "step": 1985 }, { "epoch": 1.25, "learning_rate": 3.1929716440329354e-06, "loss": 0.1215, "step": 1986 }, { "epoch": 1.25, "learning_rate": 3.188325892217587e-06, "loss": 0.1551, "step": 1987 }, { "epoch": 1.25, "learning_rate": 3.1836819401322937e-06, "loss": 0.174, "step": 1988 }, { "epoch": 1.25, "learning_rate": 3.1790397923903906e-06, "loss": 0.1387, "step": 1989 }, { "epoch": 1.25, "learning_rate": 3.17439945360342e-06, "loss": 0.1404, "step": 1990 }, { "epoch": 1.25, "learning_rate": 3.169760928381127e-06, "loss": 0.1585, "step": 1991 }, { "epoch": 1.25, "learning_rate": 3.1651242213314593e-06, "loss": 0.1274, "step": 1992 }, { "epoch": 1.25, "learning_rate": 3.1604893370605517e-06, "loss": 0.1454, "step": 1993 }, { "epoch": 1.25, "learning_rate": 3.1558562801727333e-06, "loss": 0.137, "step": 1994 }, { "epoch": 1.25, "learning_rate": 3.1512250552705147e-06, "loss": 0.1238, "step": 1995 }, { "epoch": 1.25, "learning_rate": 3.1465956669545877e-06, "loss": 0.1655, "step": 1996 }, { "epoch": 1.25, "learning_rate": 3.1419681198238215e-06, "loss": 0.1167, "step": 1997 }, { "epoch": 1.25, "learning_rate": 3.1373424184752545e-06, "loss": 0.1289, "step": 1998 }, { "epoch": 1.26, "learning_rate": 3.1327185675040907e-06, "loss": 0.1301, "step": 1999 }, { "epoch": 1.26, "learning_rate": 3.128096571503696e-06, "loss": 0.1352, "step": 2000 }, { "epoch": 1.26, "learning_rate": 3.1234764350655968e-06, "loss": 0.1164, "step": 2001 }, { "epoch": 1.26, "learning_rate": 3.1188581627794693e-06, "loss": 0.1555, "step": 2002 }, { "epoch": 1.26, "learning_rate": 3.1142417592331354e-06, "loss": 0.1454, "step": 2003 }, { "epoch": 1.26, "learning_rate": 3.109627229012566e-06, "loss": 0.1378, "step": 2004 }, { "epoch": 1.26, "learning_rate": 3.105014576701869e-06, "loss": 0.137, "step": 2005 }, { "epoch": 1.26, "learning_rate": 3.1004038068832847e-06, "loss": 0.1431, "step": 2006 }, { "epoch": 1.26, "learning_rate": 3.0957949241371845e-06, "loss": 0.1731, "step": 2007 }, { "epoch": 1.26, "learning_rate": 3.0911879330420678e-06, "loss": 0.1559, "step": 2008 }, { "epoch": 1.26, "learning_rate": 3.0865828381745515e-06, "loss": 0.1488, "step": 2009 }, { "epoch": 1.26, "learning_rate": 3.0819796441093703e-06, "loss": 0.1334, "step": 2010 }, { "epoch": 1.26, "learning_rate": 3.0773783554193693e-06, "loss": 0.169, "step": 2011 }, { "epoch": 1.26, "learning_rate": 3.072778976675501e-06, "loss": 0.143, "step": 2012 }, { "epoch": 1.26, "learning_rate": 3.0681815124468234e-06, "loss": 0.1291, "step": 2013 }, { "epoch": 1.26, "learning_rate": 3.06358596730049e-06, "loss": 0.1348, "step": 2014 }, { "epoch": 1.27, "learning_rate": 3.0589923458017467e-06, "loss": 0.1281, "step": 2015 }, { "epoch": 1.27, "learning_rate": 3.0544006525139315e-06, "loss": 0.1611, "step": 2016 }, { "epoch": 1.27, "learning_rate": 3.049810891998467e-06, "loss": 0.1369, "step": 2017 }, { "epoch": 1.27, "learning_rate": 3.04522306881485e-06, "loss": 0.1167, "step": 2018 }, { "epoch": 1.27, "learning_rate": 3.0406371875206585e-06, "loss": 0.1379, "step": 2019 }, { "epoch": 1.27, "learning_rate": 3.0360532526715424e-06, "loss": 0.1352, "step": 2020 }, { "epoch": 1.27, "learning_rate": 3.0314712688212133e-06, "loss": 0.145, "step": 2021 }, { "epoch": 1.27, "learning_rate": 3.026891240521447e-06, "loss": 0.1601, "step": 2022 }, { "epoch": 1.27, "learning_rate": 3.0223131723220756e-06, "loss": 0.12, "step": 2023 }, { "epoch": 1.27, "learning_rate": 3.017737068770987e-06, "loss": 0.119, "step": 2024 }, { "epoch": 1.27, "learning_rate": 3.013162934414114e-06, "loss": 0.1512, "step": 2025 }, { "epoch": 1.27, "learning_rate": 3.0085907737954344e-06, "loss": 0.1252, "step": 2026 }, { "epoch": 1.27, "learning_rate": 3.0040205914569664e-06, "loss": 0.1398, "step": 2027 }, { "epoch": 1.27, "learning_rate": 2.9994523919387603e-06, "loss": 0.138, "step": 2028 }, { "epoch": 1.27, "learning_rate": 2.9948861797788996e-06, "loss": 0.1339, "step": 2029 }, { "epoch": 1.27, "learning_rate": 2.9903219595134927e-06, "loss": 0.1735, "step": 2030 }, { "epoch": 1.28, "learning_rate": 2.9857597356766675e-06, "loss": 0.14, "step": 2031 }, { "epoch": 1.28, "learning_rate": 2.98119951280057e-06, "loss": 0.1467, "step": 2032 }, { "epoch": 1.28, "learning_rate": 2.976641295415359e-06, "loss": 0.1711, "step": 2033 }, { "epoch": 1.28, "learning_rate": 2.972085088049199e-06, "loss": 0.1299, "step": 2034 }, { "epoch": 1.28, "learning_rate": 2.9675308952282596e-06, "loss": 0.1252, "step": 2035 }, { "epoch": 1.28, "learning_rate": 2.9629787214767098e-06, "loss": 0.1591, "step": 2036 }, { "epoch": 1.28, "learning_rate": 2.9584285713167106e-06, "loss": 0.1185, "step": 2037 }, { "epoch": 1.28, "learning_rate": 2.953880449268414e-06, "loss": 0.1106, "step": 2038 }, { "epoch": 1.28, "learning_rate": 2.949334359849957e-06, "loss": 0.1323, "step": 2039 }, { "epoch": 1.28, "learning_rate": 2.9447903075774573e-06, "loss": 0.1418, "step": 2040 }, { "epoch": 1.28, "learning_rate": 2.9402482969650103e-06, "loss": 0.1399, "step": 2041 }, { "epoch": 1.28, "learning_rate": 2.9357083325246804e-06, "loss": 0.1597, "step": 2042 }, { "epoch": 1.28, "learning_rate": 2.9311704187665018e-06, "loss": 0.1325, "step": 2043 }, { "epoch": 1.28, "learning_rate": 2.926634560198469e-06, "loss": 0.1431, "step": 2044 }, { "epoch": 1.28, "learning_rate": 2.9221007613265406e-06, "loss": 0.149, "step": 2045 }, { "epoch": 1.28, "learning_rate": 2.9175690266546196e-06, "loss": 0.1418, "step": 2046 }, { "epoch": 1.29, "learning_rate": 2.913039360684565e-06, "loss": 0.1436, "step": 2047 }, { "epoch": 1.29, "learning_rate": 2.9085117679161832e-06, "loss": 0.1293, "step": 2048 }, { "epoch": 1.29, "learning_rate": 2.903986252847213e-06, "loss": 0.1403, "step": 2049 }, { "epoch": 1.29, "learning_rate": 2.8994628199733343e-06, "loss": 0.159, "step": 2050 }, { "epoch": 1.29, "learning_rate": 2.8949414737881603e-06, "loss": 0.1474, "step": 2051 }, { "epoch": 1.29, "learning_rate": 2.8904222187832253e-06, "loss": 0.1366, "step": 2052 }, { "epoch": 1.29, "learning_rate": 2.885905059447992e-06, "loss": 0.1407, "step": 2053 }, { "epoch": 1.29, "learning_rate": 2.8813900002698403e-06, "loss": 0.1237, "step": 2054 }, { "epoch": 1.29, "learning_rate": 2.876877045734058e-06, "loss": 0.1438, "step": 2055 }, { "epoch": 1.29, "learning_rate": 2.8723662003238508e-06, "loss": 0.146, "step": 2056 }, { "epoch": 1.29, "learning_rate": 2.867857468520321e-06, "loss": 0.142, "step": 2057 }, { "epoch": 1.29, "learning_rate": 2.863350854802478e-06, "loss": 0.131, "step": 2058 }, { "epoch": 1.29, "learning_rate": 2.8588463636472266e-06, "loss": 0.1422, "step": 2059 }, { "epoch": 1.29, "learning_rate": 2.8543439995293532e-06, "loss": 0.1364, "step": 2060 }, { "epoch": 1.29, "learning_rate": 2.849843766921543e-06, "loss": 0.1443, "step": 2061 }, { "epoch": 1.3, "learning_rate": 2.8453456702943604e-06, "loss": 0.1247, "step": 2062 }, { "epoch": 1.3, "learning_rate": 2.840849714116244e-06, "loss": 0.1333, "step": 2063 }, { "epoch": 1.3, "learning_rate": 2.8363559028535127e-06, "loss": 0.1464, "step": 2064 }, { "epoch": 1.3, "learning_rate": 2.8318642409703468e-06, "loss": 0.1413, "step": 2065 }, { "epoch": 1.3, "learning_rate": 2.827374732928797e-06, "loss": 0.1536, "step": 2066 }, { "epoch": 1.3, "learning_rate": 2.822887383188775e-06, "loss": 0.1323, "step": 2067 }, { "epoch": 1.3, "learning_rate": 2.818402196208042e-06, "loss": 0.1428, "step": 2068 }, { "epoch": 1.3, "learning_rate": 2.8139191764422177e-06, "loss": 0.174, "step": 2069 }, { "epoch": 1.3, "learning_rate": 2.809438328344767e-06, "loss": 0.1409, "step": 2070 }, { "epoch": 1.3, "learning_rate": 2.8049596563669936e-06, "loss": 0.1417, "step": 2071 }, { "epoch": 1.3, "learning_rate": 2.8004831649580455e-06, "loss": 0.1365, "step": 2072 }, { "epoch": 1.3, "learning_rate": 2.7960088585649004e-06, "loss": 0.1165, "step": 2073 }, { "epoch": 1.3, "learning_rate": 2.7915367416323635e-06, "loss": 0.1183, "step": 2074 }, { "epoch": 1.3, "learning_rate": 2.787066818603072e-06, "loss": 0.1284, "step": 2075 }, { "epoch": 1.3, "learning_rate": 2.782599093917476e-06, "loss": 0.14, "step": 2076 }, { "epoch": 1.3, "learning_rate": 2.778133572013846e-06, "loss": 0.1079, "step": 2077 }, { "epoch": 1.31, "learning_rate": 2.773670257328267e-06, "loss": 0.1316, "step": 2078 }, { "epoch": 1.31, "learning_rate": 2.769209154294623e-06, "loss": 0.1631, "step": 2079 }, { "epoch": 1.31, "learning_rate": 2.764750267344608e-06, "loss": 0.1195, "step": 2080 }, { "epoch": 1.31, "learning_rate": 2.760293600907714e-06, "loss": 0.1432, "step": 2081 }, { "epoch": 1.31, "learning_rate": 2.7558391594112216e-06, "loss": 0.1474, "step": 2082 }, { "epoch": 1.31, "learning_rate": 2.751386947280209e-06, "loss": 0.178, "step": 2083 }, { "epoch": 1.31, "learning_rate": 2.7469369689375314e-06, "loss": 0.1613, "step": 2084 }, { "epoch": 1.31, "learning_rate": 2.742489228803832e-06, "loss": 0.1196, "step": 2085 }, { "epoch": 1.31, "learning_rate": 2.738043731297529e-06, "loss": 0.1326, "step": 2086 }, { "epoch": 1.31, "learning_rate": 2.7336004808348094e-06, "loss": 0.1181, "step": 2087 }, { "epoch": 1.31, "learning_rate": 2.7291594818296285e-06, "loss": 0.1532, "step": 2088 }, { "epoch": 1.31, "learning_rate": 2.7247207386937102e-06, "loss": 0.1543, "step": 2089 }, { "epoch": 1.31, "learning_rate": 2.7202842558365304e-06, "loss": 0.1256, "step": 2090 }, { "epoch": 1.31, "learning_rate": 2.715850037665327e-06, "loss": 0.1401, "step": 2091 }, { "epoch": 1.31, "learning_rate": 2.7114180885850795e-06, "loss": 0.1188, "step": 2092 }, { "epoch": 1.31, "learning_rate": 2.7069884129985193e-06, "loss": 0.132, "step": 2093 }, { "epoch": 1.32, "learning_rate": 2.7025610153061214e-06, "loss": 0.1434, "step": 2094 }, { "epoch": 1.32, "learning_rate": 2.69813589990609e-06, "loss": 0.1812, "step": 2095 }, { "epoch": 1.32, "learning_rate": 2.693713071194368e-06, "loss": 0.1329, "step": 2096 }, { "epoch": 1.32, "learning_rate": 2.689292533564628e-06, "loss": 0.1361, "step": 2097 }, { "epoch": 1.32, "learning_rate": 2.6848742914082584e-06, "loss": 0.1327, "step": 2098 }, { "epoch": 1.32, "learning_rate": 2.680458349114378e-06, "loss": 0.1315, "step": 2099 }, { "epoch": 1.32, "learning_rate": 2.6760447110698107e-06, "loss": 0.1171, "step": 2100 }, { "epoch": 1.32, "learning_rate": 2.671633381659101e-06, "loss": 0.1663, "step": 2101 }, { "epoch": 1.32, "learning_rate": 2.6672243652644926e-06, "loss": 0.1395, "step": 2102 }, { "epoch": 1.32, "learning_rate": 2.662817666265932e-06, "loss": 0.1538, "step": 2103 }, { "epoch": 1.32, "learning_rate": 2.658413289041068e-06, "loss": 0.1434, "step": 2104 }, { "epoch": 1.32, "learning_rate": 2.654011237965242e-06, "loss": 0.1467, "step": 2105 }, { "epoch": 1.32, "learning_rate": 2.6496115174114795e-06, "loss": 0.1472, "step": 2106 }, { "epoch": 1.32, "learning_rate": 2.645214131750498e-06, "loss": 0.1488, "step": 2107 }, { "epoch": 1.32, "learning_rate": 2.640819085350689e-06, "loss": 0.1409, "step": 2108 }, { "epoch": 1.32, "learning_rate": 2.6364263825781238e-06, "loss": 0.1473, "step": 2109 }, { "epoch": 1.33, "learning_rate": 2.632036027796547e-06, "loss": 0.1239, "step": 2110 }, { "epoch": 1.33, "learning_rate": 2.6276480253673663e-06, "loss": 0.1398, "step": 2111 }, { "epoch": 1.33, "learning_rate": 2.6232623796496547e-06, "loss": 0.111, "step": 2112 }, { "epoch": 1.33, "learning_rate": 2.6188790950001475e-06, "loss": 0.1558, "step": 2113 }, { "epoch": 1.33, "learning_rate": 2.6144981757732263e-06, "loss": 0.1262, "step": 2114 }, { "epoch": 1.33, "learning_rate": 2.6101196263209315e-06, "loss": 0.142, "step": 2115 }, { "epoch": 1.33, "learning_rate": 2.6057434509929447e-06, "loss": 0.1208, "step": 2116 }, { "epoch": 1.33, "learning_rate": 2.601369654136586e-06, "loss": 0.1274, "step": 2117 }, { "epoch": 1.33, "learning_rate": 2.596998240096823e-06, "loss": 0.1498, "step": 2118 }, { "epoch": 1.33, "learning_rate": 2.5926292132162432e-06, "loss": 0.0963, "step": 2119 }, { "epoch": 1.33, "learning_rate": 2.588262577835073e-06, "loss": 0.1471, "step": 2120 }, { "epoch": 1.33, "learning_rate": 2.5838983382911613e-06, "loss": 0.1383, "step": 2121 }, { "epoch": 1.33, "learning_rate": 2.57953649891997e-06, "loss": 0.1559, "step": 2122 }, { "epoch": 1.33, "learning_rate": 2.5751770640545863e-06, "loss": 0.1282, "step": 2123 }, { "epoch": 1.33, "learning_rate": 2.5708200380257e-06, "loss": 0.1546, "step": 2124 }, { "epoch": 1.33, "learning_rate": 2.566465425161614e-06, "loss": 0.1316, "step": 2125 }, { "epoch": 1.34, "learning_rate": 2.5621132297882332e-06, "loss": 0.151, "step": 2126 }, { "epoch": 1.34, "learning_rate": 2.5577634562290567e-06, "loss": 0.1255, "step": 2127 }, { "epoch": 1.34, "learning_rate": 2.5534161088051823e-06, "loss": 0.116, "step": 2128 }, { "epoch": 1.34, "learning_rate": 2.5490711918353e-06, "loss": 0.1616, "step": 2129 }, { "epoch": 1.34, "learning_rate": 2.5447287096356743e-06, "loss": 0.1645, "step": 2130 }, { "epoch": 1.34, "learning_rate": 2.5403886665201617e-06, "loss": 0.1292, "step": 2131 }, { "epoch": 1.34, "learning_rate": 2.536051066800193e-06, "loss": 0.159, "step": 2132 }, { "epoch": 1.34, "learning_rate": 2.5317159147847686e-06, "loss": 0.124, "step": 2133 }, { "epoch": 1.34, "learning_rate": 2.5273832147804634e-06, "loss": 0.1562, "step": 2134 }, { "epoch": 1.34, "learning_rate": 2.5230529710914074e-06, "loss": 0.1243, "step": 2135 }, { "epoch": 1.34, "learning_rate": 2.518725188019299e-06, "loss": 0.1228, "step": 2136 }, { "epoch": 1.34, "learning_rate": 2.51439986986339e-06, "loss": 0.1212, "step": 2137 }, { "epoch": 1.34, "learning_rate": 2.5100770209204796e-06, "loss": 0.1292, "step": 2138 }, { "epoch": 1.34, "learning_rate": 2.5057566454849196e-06, "loss": 0.113, "step": 2139 }, { "epoch": 1.34, "learning_rate": 2.501438747848599e-06, "loss": 0.1464, "step": 2140 }, { "epoch": 1.34, "learning_rate": 2.49712333230095e-06, "loss": 0.1385, "step": 2141 }, { "epoch": 1.35, "learning_rate": 2.49281040312894e-06, "loss": 0.1802, "step": 2142 }, { "epoch": 1.35, "learning_rate": 2.48849996461706e-06, "loss": 0.1502, "step": 2143 }, { "epoch": 1.35, "learning_rate": 2.48419202104733e-06, "loss": 0.1357, "step": 2144 }, { "epoch": 1.35, "learning_rate": 2.4798865766992958e-06, "loss": 0.1425, "step": 2145 }, { "epoch": 1.35, "learning_rate": 2.475583635850013e-06, "loss": 0.1355, "step": 2146 }, { "epoch": 1.35, "learning_rate": 2.4712832027740545e-06, "loss": 0.136, "step": 2147 }, { "epoch": 1.35, "learning_rate": 2.4669852817435053e-06, "loss": 0.119, "step": 2148 }, { "epoch": 1.35, "learning_rate": 2.4626898770279455e-06, "loss": 0.1391, "step": 2149 }, { "epoch": 1.35, "learning_rate": 2.4583969928944657e-06, "loss": 0.1429, "step": 2150 }, { "epoch": 1.35, "learning_rate": 2.4541066336076434e-06, "loss": 0.1383, "step": 2151 }, { "epoch": 1.35, "learning_rate": 2.449818803429555e-06, "loss": 0.1324, "step": 2152 }, { "epoch": 1.35, "learning_rate": 2.445533506619763e-06, "loss": 0.122, "step": 2153 }, { "epoch": 1.35, "learning_rate": 2.441250747435308e-06, "loss": 0.1489, "step": 2154 }, { "epoch": 1.35, "learning_rate": 2.4369705301307185e-06, "loss": 0.1424, "step": 2155 }, { "epoch": 1.35, "learning_rate": 2.432692858957988e-06, "loss": 0.1202, "step": 2156 }, { "epoch": 1.35, "learning_rate": 2.4284177381665903e-06, "loss": 0.128, "step": 2157 }, { "epoch": 1.36, "learning_rate": 2.4241451720034586e-06, "loss": 0.118, "step": 2158 }, { "epoch": 1.36, "learning_rate": 2.4198751647129896e-06, "loss": 0.1249, "step": 2159 }, { "epoch": 1.36, "learning_rate": 2.4156077205370408e-06, "loss": 0.1521, "step": 2160 }, { "epoch": 1.36, "learning_rate": 2.411342843714924e-06, "loss": 0.1438, "step": 2161 }, { "epoch": 1.36, "learning_rate": 2.4070805384833938e-06, "loss": 0.1482, "step": 2162 }, { "epoch": 1.36, "learning_rate": 2.402820809076658e-06, "loss": 0.147, "step": 2163 }, { "epoch": 1.36, "learning_rate": 2.3985636597263627e-06, "loss": 0.1268, "step": 2164 }, { "epoch": 1.36, "learning_rate": 2.394309094661588e-06, "loss": 0.1155, "step": 2165 }, { "epoch": 1.36, "learning_rate": 2.390057118108853e-06, "loss": 0.1179, "step": 2166 }, { "epoch": 1.36, "learning_rate": 2.385807734292097e-06, "loss": 0.1093, "step": 2167 }, { "epoch": 1.36, "learning_rate": 2.381560947432691e-06, "loss": 0.1694, "step": 2168 }, { "epoch": 1.36, "learning_rate": 2.3773167617494258e-06, "loss": 0.1446, "step": 2169 }, { "epoch": 1.36, "learning_rate": 2.3730751814585013e-06, "loss": 0.1563, "step": 2170 }, { "epoch": 1.36, "learning_rate": 2.368836210773538e-06, "loss": 0.1413, "step": 2171 }, { "epoch": 1.36, "learning_rate": 2.3645998539055582e-06, "loss": 0.1384, "step": 2172 }, { "epoch": 1.36, "learning_rate": 2.3603661150629865e-06, "loss": 0.1419, "step": 2173 }, { "epoch": 1.37, "learning_rate": 2.3561349984516535e-06, "loss": 0.1259, "step": 2174 }, { "epoch": 1.37, "learning_rate": 2.3519065082747777e-06, "loss": 0.1358, "step": 2175 }, { "epoch": 1.37, "learning_rate": 2.347680648732973e-06, "loss": 0.1073, "step": 2176 }, { "epoch": 1.37, "learning_rate": 2.3434574240242398e-06, "loss": 0.142, "step": 2177 }, { "epoch": 1.37, "learning_rate": 2.3392368383439568e-06, "loss": 0.1663, "step": 2178 }, { "epoch": 1.37, "learning_rate": 2.335018895884886e-06, "loss": 0.1601, "step": 2179 }, { "epoch": 1.37, "learning_rate": 2.330803600837164e-06, "loss": 0.1418, "step": 2180 }, { "epoch": 1.37, "learning_rate": 2.32659095738829e-06, "loss": 0.1201, "step": 2181 }, { "epoch": 1.37, "learning_rate": 2.3223809697231386e-06, "loss": 0.1519, "step": 2182 }, { "epoch": 1.37, "learning_rate": 2.318173642023939e-06, "loss": 0.1456, "step": 2183 }, { "epoch": 1.37, "learning_rate": 2.3139689784702813e-06, "loss": 0.1445, "step": 2184 }, { "epoch": 1.37, "learning_rate": 2.3097669832391124e-06, "loss": 0.1274, "step": 2185 }, { "epoch": 1.37, "learning_rate": 2.3055676605047173e-06, "loss": 0.1207, "step": 2186 }, { "epoch": 1.37, "learning_rate": 2.3013710144387374e-06, "loss": 0.137, "step": 2187 }, { "epoch": 1.37, "learning_rate": 2.2971770492101525e-06, "loss": 0.1229, "step": 2188 }, { "epoch": 1.37, "learning_rate": 2.2929857689852737e-06, "loss": 0.1205, "step": 2189 }, { "epoch": 1.38, "learning_rate": 2.288797177927753e-06, "loss": 0.1632, "step": 2190 }, { "epoch": 1.38, "learning_rate": 2.284611280198563e-06, "loss": 0.1342, "step": 2191 }, { "epoch": 1.38, "learning_rate": 2.2804280799560064e-06, "loss": 0.1227, "step": 2192 }, { "epoch": 1.38, "learning_rate": 2.276247581355706e-06, "loss": 0.1325, "step": 2193 }, { "epoch": 1.38, "learning_rate": 2.2720697885505955e-06, "loss": 0.1203, "step": 2194 }, { "epoch": 1.38, "learning_rate": 2.267894705690926e-06, "loss": 0.1404, "step": 2195 }, { "epoch": 1.38, "learning_rate": 2.263722336924257e-06, "loss": 0.1306, "step": 2196 }, { "epoch": 1.38, "learning_rate": 2.259552686395446e-06, "loss": 0.1159, "step": 2197 }, { "epoch": 1.38, "learning_rate": 2.2553857582466567e-06, "loss": 0.1511, "step": 2198 }, { "epoch": 1.38, "learning_rate": 2.251221556617344e-06, "loss": 0.1239, "step": 2199 }, { "epoch": 1.38, "learning_rate": 2.2470600856442533e-06, "loss": 0.1257, "step": 2200 }, { "epoch": 1.38, "learning_rate": 2.2429013494614245e-06, "loss": 0.1097, "step": 2201 }, { "epoch": 1.38, "learning_rate": 2.2387453522001713e-06, "loss": 0.1142, "step": 2202 }, { "epoch": 1.38, "learning_rate": 2.2345920979890933e-06, "loss": 0.1572, "step": 2203 }, { "epoch": 1.38, "learning_rate": 2.230441590954065e-06, "loss": 0.1282, "step": 2204 }, { "epoch": 1.38, "learning_rate": 2.2262938352182257e-06, "loss": 0.1454, "step": 2205 }, { "epoch": 1.39, "learning_rate": 2.2221488349019903e-06, "loss": 0.132, "step": 2206 }, { "epoch": 1.39, "learning_rate": 2.218006594123028e-06, "loss": 0.1174, "step": 2207 }, { "epoch": 1.39, "learning_rate": 2.213867116996271e-06, "loss": 0.1306, "step": 2208 }, { "epoch": 1.39, "learning_rate": 2.209730407633909e-06, "loss": 0.1276, "step": 2209 }, { "epoch": 1.39, "learning_rate": 2.205596470145375e-06, "loss": 0.1337, "step": 2210 }, { "epoch": 1.39, "learning_rate": 2.201465308637352e-06, "loss": 0.1101, "step": 2211 }, { "epoch": 1.39, "learning_rate": 2.1973369272137703e-06, "loss": 0.1377, "step": 2212 }, { "epoch": 1.39, "learning_rate": 2.1932113299757874e-06, "loss": 0.1329, "step": 2213 }, { "epoch": 1.39, "learning_rate": 2.1890885210218066e-06, "loss": 0.1418, "step": 2214 }, { "epoch": 1.39, "learning_rate": 2.184968504447453e-06, "loss": 0.1289, "step": 2215 }, { "epoch": 1.39, "learning_rate": 2.1808512843455793e-06, "loss": 0.1408, "step": 2216 }, { "epoch": 1.39, "learning_rate": 2.1767368648062654e-06, "loss": 0.129, "step": 2217 }, { "epoch": 1.39, "learning_rate": 2.1726252499168003e-06, "loss": 0.157, "step": 2218 }, { "epoch": 1.39, "learning_rate": 2.168516443761695e-06, "loss": 0.1295, "step": 2219 }, { "epoch": 1.39, "learning_rate": 2.164410450422668e-06, "loss": 0.1274, "step": 2220 }, { "epoch": 1.39, "learning_rate": 2.160307273978639e-06, "loss": 0.116, "step": 2221 }, { "epoch": 1.4, "learning_rate": 2.156206918505735e-06, "loss": 0.155, "step": 2222 }, { "epoch": 1.4, "learning_rate": 2.15210938807728e-06, "loss": 0.1446, "step": 2223 }, { "epoch": 1.4, "learning_rate": 2.1480146867637867e-06, "loss": 0.1233, "step": 2224 }, { "epoch": 1.4, "learning_rate": 2.1439228186329647e-06, "loss": 0.1287, "step": 2225 }, { "epoch": 1.4, "learning_rate": 2.139833787749701e-06, "loss": 0.1216, "step": 2226 }, { "epoch": 1.4, "learning_rate": 2.1357475981760704e-06, "loss": 0.162, "step": 2227 }, { "epoch": 1.4, "learning_rate": 2.1316642539713267e-06, "loss": 0.1344, "step": 2228 }, { "epoch": 1.4, "learning_rate": 2.1275837591918854e-06, "loss": 0.1474, "step": 2229 }, { "epoch": 1.4, "learning_rate": 2.123506117891343e-06, "loss": 0.1244, "step": 2230 }, { "epoch": 1.4, "learning_rate": 2.11943133412046e-06, "loss": 0.1177, "step": 2231 }, { "epoch": 1.4, "learning_rate": 2.115359411927152e-06, "loss": 0.1359, "step": 2232 }, { "epoch": 1.4, "learning_rate": 2.111290355356499e-06, "loss": 0.1511, "step": 2233 }, { "epoch": 1.4, "learning_rate": 2.1072241684507277e-06, "loss": 0.1476, "step": 2234 }, { "epoch": 1.4, "learning_rate": 2.1031608552492187e-06, "loss": 0.1286, "step": 2235 }, { "epoch": 1.4, "learning_rate": 2.0991004197884995e-06, "loss": 0.1311, "step": 2236 }, { "epoch": 1.4, "learning_rate": 2.095042866102231e-06, "loss": 0.106, "step": 2237 }, { "epoch": 1.41, "learning_rate": 2.090988198221218e-06, "loss": 0.1211, "step": 2238 }, { "epoch": 1.41, "learning_rate": 2.086936420173399e-06, "loss": 0.1244, "step": 2239 }, { "epoch": 1.41, "learning_rate": 2.0828875359838357e-06, "loss": 0.1099, "step": 2240 }, { "epoch": 1.41, "learning_rate": 2.078841549674722e-06, "loss": 0.1379, "step": 2241 }, { "epoch": 1.41, "learning_rate": 2.074798465265368e-06, "loss": 0.1205, "step": 2242 }, { "epoch": 1.41, "learning_rate": 2.0707582867722008e-06, "loss": 0.1269, "step": 2243 }, { "epoch": 1.41, "learning_rate": 2.066721018208767e-06, "loss": 0.1326, "step": 2244 }, { "epoch": 1.41, "learning_rate": 2.062686663585714e-06, "loss": 0.122, "step": 2245 }, { "epoch": 1.41, "learning_rate": 2.058655226910801e-06, "loss": 0.1525, "step": 2246 }, { "epoch": 1.41, "learning_rate": 2.0546267121888863e-06, "loss": 0.1385, "step": 2247 }, { "epoch": 1.41, "learning_rate": 2.050601123421923e-06, "loss": 0.1289, "step": 2248 }, { "epoch": 1.41, "learning_rate": 2.0465784646089644e-06, "loss": 0.119, "step": 2249 }, { "epoch": 1.41, "learning_rate": 2.0425587397461425e-06, "loss": 0.1186, "step": 2250 }, { "epoch": 1.41, "learning_rate": 2.038541952826685e-06, "loss": 0.1395, "step": 2251 }, { "epoch": 1.41, "learning_rate": 2.034528107840897e-06, "loss": 0.1208, "step": 2252 }, { "epoch": 1.41, "learning_rate": 2.0305172087761588e-06, "loss": 0.1472, "step": 2253 }, { "epoch": 1.42, "learning_rate": 2.0265092596169265e-06, "loss": 0.1224, "step": 2254 }, { "epoch": 1.42, "learning_rate": 2.0225042643447283e-06, "loss": 0.1364, "step": 2255 }, { "epoch": 1.42, "learning_rate": 2.018502226938154e-06, "loss": 0.136, "step": 2256 }, { "epoch": 1.42, "learning_rate": 2.014503151372853e-06, "loss": 0.1241, "step": 2257 }, { "epoch": 1.42, "learning_rate": 2.01050704162154e-06, "loss": 0.1903, "step": 2258 }, { "epoch": 1.42, "learning_rate": 2.006513901653975e-06, "loss": 0.1105, "step": 2259 }, { "epoch": 1.42, "learning_rate": 2.002523735436977e-06, "loss": 0.1319, "step": 2260 }, { "epoch": 1.42, "learning_rate": 1.998536546934401e-06, "loss": 0.1181, "step": 2261 }, { "epoch": 1.42, "learning_rate": 1.99455234010715e-06, "loss": 0.1402, "step": 2262 }, { "epoch": 1.42, "learning_rate": 1.990571118913166e-06, "loss": 0.1071, "step": 2263 }, { "epoch": 1.42, "learning_rate": 1.98659288730742e-06, "loss": 0.1389, "step": 2264 }, { "epoch": 1.42, "learning_rate": 1.9826176492419175e-06, "loss": 0.11, "step": 2265 }, { "epoch": 1.42, "learning_rate": 1.9786454086656865e-06, "loss": 0.1343, "step": 2266 }, { "epoch": 1.42, "learning_rate": 1.9746761695247803e-06, "loss": 0.142, "step": 2267 }, { "epoch": 1.42, "learning_rate": 1.970709935762271e-06, "loss": 0.1344, "step": 2268 }, { "epoch": 1.43, "learning_rate": 1.9667467113182416e-06, "loss": 0.1318, "step": 2269 }, { "epoch": 1.43, "learning_rate": 1.96278650012979e-06, "loss": 0.1274, "step": 2270 }, { "epoch": 1.43, "learning_rate": 1.9588293061310165e-06, "loss": 0.1227, "step": 2271 }, { "epoch": 1.43, "learning_rate": 1.9548751332530253e-06, "loss": 0.1505, "step": 2272 }, { "epoch": 1.43, "learning_rate": 1.950923985423921e-06, "loss": 0.1366, "step": 2273 }, { "epoch": 1.43, "learning_rate": 1.946975866568804e-06, "loss": 0.1193, "step": 2274 }, { "epoch": 1.43, "learning_rate": 1.9430307806097605e-06, "loss": 0.121, "step": 2275 }, { "epoch": 1.43, "learning_rate": 1.9390887314658708e-06, "loss": 0.1221, "step": 2276 }, { "epoch": 1.43, "learning_rate": 1.935149723053191e-06, "loss": 0.1186, "step": 2277 }, { "epoch": 1.43, "learning_rate": 1.931213759284763e-06, "loss": 0.1401, "step": 2278 }, { "epoch": 1.43, "learning_rate": 1.9272808440706024e-06, "loss": 0.1283, "step": 2279 }, { "epoch": 1.43, "learning_rate": 1.9233509813176916e-06, "loss": 0.1093, "step": 2280 }, { "epoch": 1.43, "learning_rate": 1.9194241749299884e-06, "loss": 0.133, "step": 2281 }, { "epoch": 1.43, "learning_rate": 1.9155004288084074e-06, "loss": 0.1232, "step": 2282 }, { "epoch": 1.43, "learning_rate": 1.911579746850827e-06, "loss": 0.1522, "step": 2283 }, { "epoch": 1.43, "learning_rate": 1.9076621329520843e-06, "loss": 0.1311, "step": 2284 }, { "epoch": 1.44, "learning_rate": 1.9037475910039593e-06, "loss": 0.1373, "step": 2285 }, { "epoch": 1.44, "learning_rate": 1.899836124895188e-06, "loss": 0.1451, "step": 2286 }, { "epoch": 1.44, "learning_rate": 1.8959277385114516e-06, "loss": 0.133, "step": 2287 }, { "epoch": 1.44, "learning_rate": 1.892022435735365e-06, "loss": 0.1435, "step": 2288 }, { "epoch": 1.44, "learning_rate": 1.8881202204464866e-06, "loss": 0.1323, "step": 2289 }, { "epoch": 1.44, "learning_rate": 1.884221096521306e-06, "loss": 0.1398, "step": 2290 }, { "epoch": 1.44, "learning_rate": 1.880325067833238e-06, "loss": 0.1324, "step": 2291 }, { "epoch": 1.44, "learning_rate": 1.8764321382526285e-06, "loss": 0.1363, "step": 2292 }, { "epoch": 1.44, "learning_rate": 1.8725423116467384e-06, "loss": 0.1483, "step": 2293 }, { "epoch": 1.44, "learning_rate": 1.868655591879751e-06, "loss": 0.1236, "step": 2294 }, { "epoch": 1.44, "learning_rate": 1.864771982812763e-06, "loss": 0.1214, "step": 2295 }, { "epoch": 1.44, "learning_rate": 1.860891488303777e-06, "loss": 0.1382, "step": 2296 }, { "epoch": 1.44, "learning_rate": 1.8570141122077068e-06, "loss": 0.1205, "step": 2297 }, { "epoch": 1.44, "learning_rate": 1.8531398583763637e-06, "loss": 0.1257, "step": 2298 }, { "epoch": 1.44, "learning_rate": 1.8492687306584583e-06, "loss": 0.1289, "step": 2299 }, { "epoch": 1.44, "learning_rate": 1.8454007328995992e-06, "loss": 0.1313, "step": 2300 }, { "epoch": 1.45, "learning_rate": 1.8415358689422807e-06, "loss": 0.1457, "step": 2301 }, { "epoch": 1.45, "learning_rate": 1.837674142625887e-06, "loss": 0.114, "step": 2302 }, { "epoch": 1.45, "learning_rate": 1.8338155577866873e-06, "loss": 0.1025, "step": 2303 }, { "epoch": 1.45, "learning_rate": 1.8299601182578241e-06, "loss": 0.1201, "step": 2304 }, { "epoch": 1.45, "learning_rate": 1.8261078278693195e-06, "loss": 0.1568, "step": 2305 }, { "epoch": 1.45, "learning_rate": 1.8222586904480688e-06, "loss": 0.1345, "step": 2306 }, { "epoch": 1.45, "learning_rate": 1.8184127098178288e-06, "loss": 0.1167, "step": 2307 }, { "epoch": 1.45, "learning_rate": 1.8145698897992282e-06, "loss": 0.115, "step": 2308 }, { "epoch": 1.45, "learning_rate": 1.8107302342097487e-06, "loss": 0.1037, "step": 2309 }, { "epoch": 1.45, "learning_rate": 1.8068937468637326e-06, "loss": 0.1129, "step": 2310 }, { "epoch": 1.45, "learning_rate": 1.8030604315723765e-06, "loss": 0.1234, "step": 2311 }, { "epoch": 1.45, "learning_rate": 1.799230292143721e-06, "loss": 0.1136, "step": 2312 }, { "epoch": 1.45, "learning_rate": 1.7954033323826524e-06, "loss": 0.1483, "step": 2313 }, { "epoch": 1.45, "learning_rate": 1.7915795560909043e-06, "loss": 0.112, "step": 2314 }, { "epoch": 1.45, "learning_rate": 1.7877589670670386e-06, "loss": 0.1234, "step": 2315 }, { "epoch": 1.45, "learning_rate": 1.783941569106461e-06, "loss": 0.1408, "step": 2316 }, { "epoch": 1.46, "learning_rate": 1.7801273660013968e-06, "loss": 0.1054, "step": 2317 }, { "epoch": 1.46, "learning_rate": 1.7763163615409057e-06, "loss": 0.1274, "step": 2318 }, { "epoch": 1.46, "learning_rate": 1.7725085595108682e-06, "loss": 0.1115, "step": 2319 }, { "epoch": 1.46, "learning_rate": 1.7687039636939785e-06, "loss": 0.1604, "step": 2320 }, { "epoch": 1.46, "learning_rate": 1.764902577869751e-06, "loss": 0.1236, "step": 2321 }, { "epoch": 1.46, "learning_rate": 1.7611044058145121e-06, "loss": 0.1506, "step": 2322 }, { "epoch": 1.46, "learning_rate": 1.7573094513013893e-06, "loss": 0.1163, "step": 2323 }, { "epoch": 1.46, "learning_rate": 1.753517718100321e-06, "loss": 0.1288, "step": 2324 }, { "epoch": 1.46, "learning_rate": 1.7497292099780384e-06, "loss": 0.1508, "step": 2325 }, { "epoch": 1.46, "learning_rate": 1.745943930698077e-06, "loss": 0.1099, "step": 2326 }, { "epoch": 1.46, "learning_rate": 1.7421618840207576e-06, "loss": 0.1374, "step": 2327 }, { "epoch": 1.46, "learning_rate": 1.738383073703192e-06, "loss": 0.1162, "step": 2328 }, { "epoch": 1.46, "learning_rate": 1.7346075034992782e-06, "loss": 0.1338, "step": 2329 }, { "epoch": 1.46, "learning_rate": 1.7308351771596964e-06, "loss": 0.1474, "step": 2330 }, { "epoch": 1.46, "learning_rate": 1.7270660984318998e-06, "loss": 0.1302, "step": 2331 }, { "epoch": 1.46, "learning_rate": 1.7233002710601215e-06, "loss": 0.1325, "step": 2332 }, { "epoch": 1.47, "learning_rate": 1.719537698785358e-06, "loss": 0.1273, "step": 2333 }, { "epoch": 1.47, "learning_rate": 1.7157783853453775e-06, "loss": 0.166, "step": 2334 }, { "epoch": 1.47, "learning_rate": 1.71202233447471e-06, "loss": 0.1408, "step": 2335 }, { "epoch": 1.47, "learning_rate": 1.7082695499046403e-06, "loss": 0.1169, "step": 2336 }, { "epoch": 1.47, "learning_rate": 1.7045200353632135e-06, "loss": 0.1026, "step": 2337 }, { "epoch": 1.47, "learning_rate": 1.7007737945752257e-06, "loss": 0.1491, "step": 2338 }, { "epoch": 1.47, "learning_rate": 1.6970308312622157e-06, "loss": 0.1122, "step": 2339 }, { "epoch": 1.47, "learning_rate": 1.6932911491424743e-06, "loss": 0.1289, "step": 2340 }, { "epoch": 1.47, "learning_rate": 1.6895547519310258e-06, "loss": 0.117, "step": 2341 }, { "epoch": 1.47, "learning_rate": 1.685821643339633e-06, "loss": 0.1254, "step": 2342 }, { "epoch": 1.47, "learning_rate": 1.682091827076796e-06, "loss": 0.1094, "step": 2343 }, { "epoch": 1.47, "learning_rate": 1.678365306847738e-06, "loss": 0.123, "step": 2344 }, { "epoch": 1.47, "learning_rate": 1.6746420863544115e-06, "loss": 0.1006, "step": 2345 }, { "epoch": 1.47, "learning_rate": 1.6709221692954936e-06, "loss": 0.1158, "step": 2346 }, { "epoch": 1.47, "learning_rate": 1.667205559366372e-06, "loss": 0.153, "step": 2347 }, { "epoch": 1.47, "learning_rate": 1.6634922602591564e-06, "loss": 0.1206, "step": 2348 }, { "epoch": 1.48, "learning_rate": 1.6597822756626658e-06, "loss": 0.1498, "step": 2349 }, { "epoch": 1.48, "learning_rate": 1.6560756092624219e-06, "loss": 0.1387, "step": 2350 }, { "epoch": 1.48, "learning_rate": 1.6523722647406575e-06, "loss": 0.1214, "step": 2351 }, { "epoch": 1.48, "learning_rate": 1.6486722457762983e-06, "loss": 0.1249, "step": 2352 }, { "epoch": 1.48, "learning_rate": 1.6449755560449715e-06, "loss": 0.1291, "step": 2353 }, { "epoch": 1.48, "learning_rate": 1.6412821992189976e-06, "loss": 0.1442, "step": 2354 }, { "epoch": 1.48, "learning_rate": 1.6375921789673781e-06, "loss": 0.1435, "step": 2355 }, { "epoch": 1.48, "learning_rate": 1.6339054989558085e-06, "loss": 0.113, "step": 2356 }, { "epoch": 1.48, "learning_rate": 1.6302221628466653e-06, "loss": 0.1275, "step": 2357 }, { "epoch": 1.48, "learning_rate": 1.6265421742989968e-06, "loss": 0.1277, "step": 2358 }, { "epoch": 1.48, "learning_rate": 1.6228655369685342e-06, "loss": 0.1223, "step": 2359 }, { "epoch": 1.48, "learning_rate": 1.6191922545076717e-06, "loss": 0.1301, "step": 2360 }, { "epoch": 1.48, "learning_rate": 1.6155223305654755e-06, "loss": 0.1508, "step": 2361 }, { "epoch": 1.48, "learning_rate": 1.6118557687876774e-06, "loss": 0.1479, "step": 2362 }, { "epoch": 1.48, "learning_rate": 1.608192572816662e-06, "loss": 0.1221, "step": 2363 }, { "epoch": 1.48, "learning_rate": 1.6045327462914755e-06, "loss": 0.1064, "step": 2364 }, { "epoch": 1.49, "learning_rate": 1.600876292847819e-06, "loss": 0.1458, "step": 2365 }, { "epoch": 1.49, "learning_rate": 1.5972232161180352e-06, "loss": 0.1365, "step": 2366 }, { "epoch": 1.49, "learning_rate": 1.5935735197311204e-06, "loss": 0.1173, "step": 2367 }, { "epoch": 1.49, "learning_rate": 1.5899272073127075e-06, "loss": 0.1127, "step": 2368 }, { "epoch": 1.49, "learning_rate": 1.5862842824850677e-06, "loss": 0.1344, "step": 2369 }, { "epoch": 1.49, "learning_rate": 1.5826447488671114e-06, "loss": 0.1167, "step": 2370 }, { "epoch": 1.49, "learning_rate": 1.5790086100743756e-06, "loss": 0.129, "step": 2371 }, { "epoch": 1.49, "learning_rate": 1.5753758697190268e-06, "loss": 0.1042, "step": 2372 }, { "epoch": 1.49, "learning_rate": 1.5717465314098573e-06, "loss": 0.1243, "step": 2373 }, { "epoch": 1.49, "learning_rate": 1.5681205987522752e-06, "loss": 0.1084, "step": 2374 }, { "epoch": 1.49, "learning_rate": 1.5644980753483109e-06, "loss": 0.1428, "step": 2375 }, { "epoch": 1.49, "learning_rate": 1.5608789647966015e-06, "loss": 0.1204, "step": 2376 }, { "epoch": 1.49, "learning_rate": 1.5572632706923995e-06, "loss": 0.1094, "step": 2377 }, { "epoch": 1.49, "learning_rate": 1.553650996627563e-06, "loss": 0.1235, "step": 2378 }, { "epoch": 1.49, "learning_rate": 1.5500421461905475e-06, "loss": 0.1447, "step": 2379 }, { "epoch": 1.49, "learning_rate": 1.5464367229664124e-06, "loss": 0.1111, "step": 2380 }, { "epoch": 1.5, "learning_rate": 1.542834730536813e-06, "loss": 0.1291, "step": 2381 }, { "epoch": 1.5, "learning_rate": 1.5392361724799925e-06, "loss": 0.1131, "step": 2382 }, { "epoch": 1.5, "learning_rate": 1.5356410523707828e-06, "loss": 0.1143, "step": 2383 }, { "epoch": 1.5, "learning_rate": 1.5320493737806052e-06, "loss": 0.1229, "step": 2384 }, { "epoch": 1.5, "learning_rate": 1.5284611402774557e-06, "loss": 0.1438, "step": 2385 }, { "epoch": 1.5, "learning_rate": 1.5248763554259143e-06, "loss": 0.1158, "step": 2386 }, { "epoch": 1.5, "learning_rate": 1.5212950227871292e-06, "loss": 0.1193, "step": 2387 }, { "epoch": 1.5, "learning_rate": 1.5177171459188234e-06, "loss": 0.1213, "step": 2388 }, { "epoch": 1.5, "learning_rate": 1.5141427283752869e-06, "loss": 0.1323, "step": 2389 }, { "epoch": 1.5, "learning_rate": 1.5105717737073682e-06, "loss": 0.1176, "step": 2390 }, { "epoch": 1.5, "learning_rate": 1.5070042854624833e-06, "loss": 0.1366, "step": 2391 }, { "epoch": 1.5, "learning_rate": 1.5034402671845972e-06, "loss": 0.1452, "step": 2392 }, { "epoch": 1.5, "learning_rate": 1.4998797224142319e-06, "loss": 0.1049, "step": 2393 }, { "epoch": 1.5, "learning_rate": 1.4963226546884607e-06, "loss": 0.1319, "step": 2394 }, { "epoch": 1.5, "learning_rate": 1.4927690675408969e-06, "loss": 0.1173, "step": 2395 }, { "epoch": 1.5, "learning_rate": 1.4892189645017019e-06, "loss": 0.1222, "step": 2396 }, { "epoch": 1.51, "learning_rate": 1.4856723490975733e-06, "loss": 0.1386, "step": 2397 }, { "epoch": 1.51, "learning_rate": 1.482129224851741e-06, "loss": 0.1182, "step": 2398 }, { "epoch": 1.51, "learning_rate": 1.4785895952839735e-06, "loss": 0.1197, "step": 2399 }, { "epoch": 1.51, "learning_rate": 1.475053463910564e-06, "loss": 0.1231, "step": 2400 }, { "epoch": 1.51, "learning_rate": 1.4715208342443288e-06, "loss": 0.1049, "step": 2401 }, { "epoch": 1.51, "learning_rate": 1.46799170979461e-06, "loss": 0.112, "step": 2402 }, { "epoch": 1.51, "learning_rate": 1.4644660940672628e-06, "loss": 0.135, "step": 2403 }, { "epoch": 1.51, "learning_rate": 1.4609439905646599e-06, "loss": 0.1197, "step": 2404 }, { "epoch": 1.51, "learning_rate": 1.4574254027856866e-06, "loss": 0.1091, "step": 2405 }, { "epoch": 1.51, "learning_rate": 1.4539103342257287e-06, "loss": 0.0912, "step": 2406 }, { "epoch": 1.51, "learning_rate": 1.4503987883766857e-06, "loss": 0.114, "step": 2407 }, { "epoch": 1.51, "learning_rate": 1.4468907687269474e-06, "loss": 0.1182, "step": 2408 }, { "epoch": 1.51, "learning_rate": 1.4433862787614083e-06, "loss": 0.1188, "step": 2409 }, { "epoch": 1.51, "learning_rate": 1.439885321961455e-06, "loss": 0.1093, "step": 2410 }, { "epoch": 1.51, "learning_rate": 1.436387901804962e-06, "loss": 0.1299, "step": 2411 }, { "epoch": 1.51, "learning_rate": 1.4328940217662896e-06, "loss": 0.1185, "step": 2412 }, { "epoch": 1.52, "learning_rate": 1.4294036853162863e-06, "loss": 0.0956, "step": 2413 }, { "epoch": 1.52, "learning_rate": 1.4259168959222742e-06, "loss": 0.114, "step": 2414 }, { "epoch": 1.52, "learning_rate": 1.4224336570480574e-06, "loss": 0.1305, "step": 2415 }, { "epoch": 1.52, "learning_rate": 1.4189539721539107e-06, "loss": 0.1364, "step": 2416 }, { "epoch": 1.52, "learning_rate": 1.4154778446965761e-06, "loss": 0.1291, "step": 2417 }, { "epoch": 1.52, "learning_rate": 1.4120052781292654e-06, "loss": 0.1266, "step": 2418 }, { "epoch": 1.52, "learning_rate": 1.4085362759016491e-06, "loss": 0.1467, "step": 2419 }, { "epoch": 1.52, "learning_rate": 1.4050708414598602e-06, "loss": 0.0983, "step": 2420 }, { "epoch": 1.52, "learning_rate": 1.401608978246487e-06, "loss": 0.1283, "step": 2421 }, { "epoch": 1.52, "learning_rate": 1.398150689700567e-06, "loss": 0.1133, "step": 2422 }, { "epoch": 1.52, "learning_rate": 1.3946959792575915e-06, "loss": 0.1466, "step": 2423 }, { "epoch": 1.52, "learning_rate": 1.391244850349492e-06, "loss": 0.1364, "step": 2424 }, { "epoch": 1.52, "learning_rate": 1.3877973064046467e-06, "loss": 0.1323, "step": 2425 }, { "epoch": 1.52, "learning_rate": 1.3843533508478707e-06, "loss": 0.1088, "step": 2426 }, { "epoch": 1.52, "learning_rate": 1.3809129871004107e-06, "loss": 0.1076, "step": 2427 }, { "epoch": 1.52, "learning_rate": 1.3774762185799517e-06, "loss": 0.1288, "step": 2428 }, { "epoch": 1.53, "learning_rate": 1.374043048700605e-06, "loss": 0.1213, "step": 2429 }, { "epoch": 1.53, "learning_rate": 1.3706134808729037e-06, "loss": 0.1274, "step": 2430 }, { "epoch": 1.53, "learning_rate": 1.3671875185038064e-06, "loss": 0.1348, "step": 2431 }, { "epoch": 1.53, "learning_rate": 1.3637651649966893e-06, "loss": 0.1051, "step": 2432 }, { "epoch": 1.53, "learning_rate": 1.3603464237513414e-06, "loss": 0.1287, "step": 2433 }, { "epoch": 1.53, "learning_rate": 1.3569312981639664e-06, "loss": 0.146, "step": 2434 }, { "epoch": 1.53, "learning_rate": 1.353519791627173e-06, "loss": 0.1181, "step": 2435 }, { "epoch": 1.53, "learning_rate": 1.3501119075299763e-06, "loss": 0.1193, "step": 2436 }, { "epoch": 1.53, "learning_rate": 1.3467076492577952e-06, "loss": 0.1576, "step": 2437 }, { "epoch": 1.53, "learning_rate": 1.3433070201924414e-06, "loss": 0.1246, "step": 2438 }, { "epoch": 1.53, "learning_rate": 1.3399100237121266e-06, "loss": 0.1119, "step": 2439 }, { "epoch": 1.53, "learning_rate": 1.33651666319145e-06, "loss": 0.1155, "step": 2440 }, { "epoch": 1.53, "learning_rate": 1.3331269420013992e-06, "loss": 0.1104, "step": 2441 }, { "epoch": 1.53, "learning_rate": 1.3297408635093506e-06, "loss": 0.0974, "step": 2442 }, { "epoch": 1.53, "learning_rate": 1.3263584310790557e-06, "loss": 0.1154, "step": 2443 }, { "epoch": 1.53, "learning_rate": 1.3229796480706486e-06, "loss": 0.127, "step": 2444 }, { "epoch": 1.54, "learning_rate": 1.319604517840638e-06, "loss": 0.1377, "step": 2445 }, { "epoch": 1.54, "learning_rate": 1.3162330437418991e-06, "loss": 0.1254, "step": 2446 }, { "epoch": 1.54, "learning_rate": 1.312865229123681e-06, "loss": 0.129, "step": 2447 }, { "epoch": 1.54, "learning_rate": 1.3095010773315957e-06, "loss": 0.1301, "step": 2448 }, { "epoch": 1.54, "learning_rate": 1.3061405917076126e-06, "loss": 0.1055, "step": 2449 }, { "epoch": 1.54, "learning_rate": 1.3027837755900658e-06, "loss": 0.1135, "step": 2450 }, { "epoch": 1.54, "learning_rate": 1.2994306323136368e-06, "loss": 0.0978, "step": 2451 }, { "epoch": 1.54, "learning_rate": 1.296081165209364e-06, "loss": 0.0862, "step": 2452 }, { "epoch": 1.54, "learning_rate": 1.2927353776046352e-06, "loss": 0.1436, "step": 2453 }, { "epoch": 1.54, "learning_rate": 1.289393272823174e-06, "loss": 0.1348, "step": 2454 }, { "epoch": 1.54, "learning_rate": 1.2860548541850542e-06, "loss": 0.1342, "step": 2455 }, { "epoch": 1.54, "learning_rate": 1.2827201250066857e-06, "loss": 0.1181, "step": 2456 }, { "epoch": 1.54, "learning_rate": 1.2793890886008098e-06, "loss": 0.1325, "step": 2457 }, { "epoch": 1.54, "learning_rate": 1.2760617482765047e-06, "loss": 0.1343, "step": 2458 }, { "epoch": 1.54, "learning_rate": 1.272738107339172e-06, "loss": 0.1174, "step": 2459 }, { "epoch": 1.54, "learning_rate": 1.269418169090541e-06, "loss": 0.138, "step": 2460 }, { "epoch": 1.55, "learning_rate": 1.2661019368286636e-06, "loss": 0.1617, "step": 2461 }, { "epoch": 1.55, "learning_rate": 1.262789413847907e-06, "loss": 0.1433, "step": 2462 }, { "epoch": 1.55, "learning_rate": 1.2594806034389556e-06, "loss": 0.1039, "step": 2463 }, { "epoch": 1.55, "learning_rate": 1.2561755088888072e-06, "loss": 0.156, "step": 2464 }, { "epoch": 1.55, "learning_rate": 1.2528741334807632e-06, "loss": 0.1328, "step": 2465 }, { "epoch": 1.55, "learning_rate": 1.2495764804944372e-06, "loss": 0.1286, "step": 2466 }, { "epoch": 1.55, "learning_rate": 1.2462825532057394e-06, "loss": 0.1189, "step": 2467 }, { "epoch": 1.55, "learning_rate": 1.2429923548868795e-06, "loss": 0.1116, "step": 2468 }, { "epoch": 1.55, "learning_rate": 1.2397058888063668e-06, "loss": 0.1474, "step": 2469 }, { "epoch": 1.55, "learning_rate": 1.2364231582289976e-06, "loss": 0.1155, "step": 2470 }, { "epoch": 1.55, "learning_rate": 1.233144166415861e-06, "loss": 0.1355, "step": 2471 }, { "epoch": 1.55, "learning_rate": 1.2298689166243338e-06, "loss": 0.1287, "step": 2472 }, { "epoch": 1.55, "learning_rate": 1.2265974121080682e-06, "loss": 0.1036, "step": 2473 }, { "epoch": 1.55, "learning_rate": 1.2233296561170021e-06, "loss": 0.1227, "step": 2474 }, { "epoch": 1.55, "learning_rate": 1.2200656518973497e-06, "loss": 0.1447, "step": 2475 }, { "epoch": 1.56, "learning_rate": 1.2168054026915931e-06, "loss": 0.1418, "step": 2476 }, { "epoch": 1.56, "learning_rate": 1.2135489117384897e-06, "loss": 0.115, "step": 2477 }, { "epoch": 1.56, "learning_rate": 1.2102961822730586e-06, "loss": 0.1113, "step": 2478 }, { "epoch": 1.56, "learning_rate": 1.2070472175265857e-06, "loss": 0.1033, "step": 2479 }, { "epoch": 1.56, "learning_rate": 1.2038020207266176e-06, "loss": 0.1115, "step": 2480 }, { "epoch": 1.56, "learning_rate": 1.2005605950969544e-06, "loss": 0.1076, "step": 2481 }, { "epoch": 1.56, "learning_rate": 1.1973229438576506e-06, "loss": 0.1316, "step": 2482 }, { "epoch": 1.56, "learning_rate": 1.1940890702250142e-06, "loss": 0.1197, "step": 2483 }, { "epoch": 1.56, "learning_rate": 1.1908589774115964e-06, "loss": 0.1046, "step": 2484 }, { "epoch": 1.56, "learning_rate": 1.1876326686261968e-06, "loss": 0.1289, "step": 2485 }, { "epoch": 1.56, "learning_rate": 1.1844101470738516e-06, "loss": 0.1231, "step": 2486 }, { "epoch": 1.56, "learning_rate": 1.1811914159558374e-06, "loss": 0.1317, "step": 2487 }, { "epoch": 1.56, "learning_rate": 1.1779764784696669e-06, "loss": 0.1123, "step": 2488 }, { "epoch": 1.56, "learning_rate": 1.174765337809079e-06, "loss": 0.117, "step": 2489 }, { "epoch": 1.56, "learning_rate": 1.171557997164045e-06, "loss": 0.1059, "step": 2490 }, { "epoch": 1.56, "learning_rate": 1.1683544597207613e-06, "loss": 0.1327, "step": 2491 }, { "epoch": 1.57, "learning_rate": 1.1651547286616432e-06, "loss": 0.1085, "step": 2492 }, { "epoch": 1.57, "learning_rate": 1.1619588071653288e-06, "loss": 0.1145, "step": 2493 }, { "epoch": 1.57, "learning_rate": 1.158766698406666e-06, "loss": 0.1295, "step": 2494 }, { "epoch": 1.57, "learning_rate": 1.155578405556722e-06, "loss": 0.1278, "step": 2495 }, { "epoch": 1.57, "learning_rate": 1.1523939317827686e-06, "loss": 0.109, "step": 2496 }, { "epoch": 1.57, "learning_rate": 1.149213280248283e-06, "loss": 0.1282, "step": 2497 }, { "epoch": 1.57, "learning_rate": 1.1460364541129487e-06, "loss": 0.1162, "step": 2498 }, { "epoch": 1.57, "learning_rate": 1.1428634565326491e-06, "loss": 0.1257, "step": 2499 }, { "epoch": 1.57, "learning_rate": 1.13969429065946e-06, "loss": 0.1331, "step": 2500 }, { "epoch": 1.57, "learning_rate": 1.1365289596416562e-06, "loss": 0.1402, "step": 2501 }, { "epoch": 1.57, "learning_rate": 1.1333674666236972e-06, "loss": 0.1392, "step": 2502 }, { "epoch": 1.57, "learning_rate": 1.1302098147462348e-06, "loss": 0.1127, "step": 2503 }, { "epoch": 1.57, "learning_rate": 1.1270560071461045e-06, "loss": 0.1061, "step": 2504 }, { "epoch": 1.57, "learning_rate": 1.1239060469563178e-06, "loss": 0.1239, "step": 2505 }, { "epoch": 1.57, "learning_rate": 1.1207599373060696e-06, "loss": 0.1618, "step": 2506 }, { "epoch": 1.57, "learning_rate": 1.117617681320729e-06, "loss": 0.1195, "step": 2507 }, { "epoch": 1.58, "learning_rate": 1.1144792821218337e-06, "loss": 0.133, "step": 2508 }, { "epoch": 1.58, "learning_rate": 1.1113447428270935e-06, "loss": 0.118, "step": 2509 }, { "epoch": 1.58, "learning_rate": 1.1082140665503815e-06, "loss": 0.1188, "step": 2510 }, { "epoch": 1.58, "learning_rate": 1.1050872564017329e-06, "loss": 0.1248, "step": 2511 }, { "epoch": 1.58, "learning_rate": 1.1019643154873455e-06, "loss": 0.138, "step": 2512 }, { "epoch": 1.58, "learning_rate": 1.0988452469095683e-06, "loss": 0.119, "step": 2513 }, { "epoch": 1.58, "learning_rate": 1.0957300537669085e-06, "loss": 0.1094, "step": 2514 }, { "epoch": 1.58, "learning_rate": 1.0926187391540222e-06, "loss": 0.1626, "step": 2515 }, { "epoch": 1.58, "learning_rate": 1.089511306161709e-06, "loss": 0.1302, "step": 2516 }, { "epoch": 1.58, "learning_rate": 1.0864077578769183e-06, "loss": 0.1021, "step": 2517 }, { "epoch": 1.58, "learning_rate": 1.0833080973827336e-06, "loss": 0.0974, "step": 2518 }, { "epoch": 1.58, "learning_rate": 1.080212327758382e-06, "loss": 0.1095, "step": 2519 }, { "epoch": 1.58, "learning_rate": 1.0771204520792244e-06, "loss": 0.1004, "step": 2520 }, { "epoch": 1.58, "learning_rate": 1.074032473416749e-06, "loss": 0.106, "step": 2521 }, { "epoch": 1.58, "learning_rate": 1.0709483948385785e-06, "loss": 0.1172, "step": 2522 }, { "epoch": 1.58, "learning_rate": 1.0678682194084605e-06, "loss": 0.0993, "step": 2523 }, { "epoch": 1.59, "learning_rate": 1.064791950186258e-06, "loss": 0.1311, "step": 2524 }, { "epoch": 1.59, "learning_rate": 1.061719590227962e-06, "loss": 0.1127, "step": 2525 }, { "epoch": 1.59, "learning_rate": 1.0586511425856777e-06, "loss": 0.1351, "step": 2526 }, { "epoch": 1.59, "learning_rate": 1.0555866103076212e-06, "loss": 0.0901, "step": 2527 }, { "epoch": 1.59, "learning_rate": 1.0525259964381224e-06, "loss": 0.1075, "step": 2528 }, { "epoch": 1.59, "learning_rate": 1.0494693040176152e-06, "loss": 0.1214, "step": 2529 }, { "epoch": 1.59, "learning_rate": 1.046416536082641e-06, "loss": 0.1244, "step": 2530 }, { "epoch": 1.59, "learning_rate": 1.043367695665843e-06, "loss": 0.133, "step": 2531 }, { "epoch": 1.59, "learning_rate": 1.0403227857959575e-06, "loss": 0.1386, "step": 2532 }, { "epoch": 1.59, "learning_rate": 1.0372818094978233e-06, "loss": 0.1072, "step": 2533 }, { "epoch": 1.59, "learning_rate": 1.0342447697923653e-06, "loss": 0.1071, "step": 2534 }, { "epoch": 1.59, "learning_rate": 1.0312116696966012e-06, "loss": 0.1178, "step": 2535 }, { "epoch": 1.59, "learning_rate": 1.0281825122236366e-06, "loss": 0.1211, "step": 2536 }, { "epoch": 1.59, "learning_rate": 1.0251573003826565e-06, "loss": 0.1092, "step": 2537 }, { "epoch": 1.59, "learning_rate": 1.022136037178927e-06, "loss": 0.1481, "step": 2538 }, { "epoch": 1.59, "learning_rate": 1.0191187256137946e-06, "loss": 0.1096, "step": 2539 }, { "epoch": 1.6, "learning_rate": 1.0161053686846767e-06, "loss": 0.0883, "step": 2540 }, { "epoch": 1.6, "learning_rate": 1.013095969385064e-06, "loss": 0.1066, "step": 2541 }, { "epoch": 1.6, "learning_rate": 1.0100905307045171e-06, "loss": 0.0979, "step": 2542 }, { "epoch": 1.6, "learning_rate": 1.0070890556286578e-06, "loss": 0.1447, "step": 2543 }, { "epoch": 1.6, "learning_rate": 1.004091547139176e-06, "loss": 0.1328, "step": 2544 }, { "epoch": 1.6, "learning_rate": 1.0010980082138144e-06, "loss": 0.134, "step": 2545 }, { "epoch": 1.6, "learning_rate": 9.98108441826378e-07, "loss": 0.1105, "step": 2546 }, { "epoch": 1.6, "learning_rate": 9.951228509467248e-07, "loss": 0.1175, "step": 2547 }, { "epoch": 1.6, "learning_rate": 9.921412385407597e-07, "loss": 0.1346, "step": 2548 }, { "epoch": 1.6, "learning_rate": 9.891636075704403e-07, "loss": 0.1201, "step": 2549 }, { "epoch": 1.6, "learning_rate": 9.861899609937636e-07, "loss": 0.1378, "step": 2550 }, { "epoch": 1.6, "learning_rate": 9.832203017647746e-07, "loss": 0.1228, "step": 2551 }, { "epoch": 1.6, "learning_rate": 9.802546328335528e-07, "loss": 0.1255, "step": 2552 }, { "epoch": 1.6, "learning_rate": 9.77292957146213e-07, "loss": 0.1336, "step": 2553 }, { "epoch": 1.6, "learning_rate": 9.743352776449073e-07, "loss": 0.1196, "step": 2554 }, { "epoch": 1.6, "learning_rate": 9.713815972678165e-07, "loss": 0.1204, "step": 2555 }, { "epoch": 1.61, "learning_rate": 9.684319189491454e-07, "loss": 0.1126, "step": 2556 }, { "epoch": 1.61, "learning_rate": 9.654862456191265e-07, "loss": 0.107, "step": 2557 }, { "epoch": 1.61, "learning_rate": 9.625445802040151e-07, "loss": 0.1322, "step": 2558 }, { "epoch": 1.61, "learning_rate": 9.596069256260792e-07, "loss": 0.1148, "step": 2559 }, { "epoch": 1.61, "learning_rate": 9.566732848036087e-07, "loss": 0.113, "step": 2560 }, { "epoch": 1.61, "learning_rate": 9.53743660650901e-07, "loss": 0.12, "step": 2561 }, { "epoch": 1.61, "learning_rate": 9.50818056078267e-07, "loss": 0.1164, "step": 2562 }, { "epoch": 1.61, "learning_rate": 9.478964739920248e-07, "loss": 0.1361, "step": 2563 }, { "epoch": 1.61, "learning_rate": 9.449789172944923e-07, "loss": 0.1417, "step": 2564 }, { "epoch": 1.61, "learning_rate": 9.420653888839937e-07, "loss": 0.1108, "step": 2565 }, { "epoch": 1.61, "learning_rate": 9.391558916548477e-07, "loss": 0.1375, "step": 2566 }, { "epoch": 1.61, "learning_rate": 9.362504284973683e-07, "loss": 0.1223, "step": 2567 }, { "epoch": 1.61, "learning_rate": 9.333490022978665e-07, "loss": 0.1141, "step": 2568 }, { "epoch": 1.61, "learning_rate": 9.304516159386368e-07, "loss": 0.1294, "step": 2569 }, { "epoch": 1.61, "learning_rate": 9.275582722979654e-07, "loss": 0.131, "step": 2570 }, { "epoch": 1.61, "learning_rate": 9.246689742501213e-07, "loss": 0.1332, "step": 2571 }, { "epoch": 1.62, "learning_rate": 9.217837246653522e-07, "loss": 0.1178, "step": 2572 }, { "epoch": 1.62, "learning_rate": 9.189025264098861e-07, "loss": 0.1062, "step": 2573 }, { "epoch": 1.62, "learning_rate": 9.160253823459276e-07, "loss": 0.1261, "step": 2574 }, { "epoch": 1.62, "learning_rate": 9.131522953316502e-07, "loss": 0.144, "step": 2575 }, { "epoch": 1.62, "learning_rate": 9.102832682212009e-07, "loss": 0.1253, "step": 2576 }, { "epoch": 1.62, "learning_rate": 9.0741830386469e-07, "loss": 0.0929, "step": 2577 }, { "epoch": 1.62, "learning_rate": 9.04557405108194e-07, "loss": 0.1103, "step": 2578 }, { "epoch": 1.62, "learning_rate": 9.017005747937541e-07, "loss": 0.1255, "step": 2579 }, { "epoch": 1.62, "learning_rate": 8.9884781575936e-07, "loss": 0.103, "step": 2580 }, { "epoch": 1.62, "learning_rate": 8.959991308389654e-07, "loss": 0.1362, "step": 2581 }, { "epoch": 1.62, "learning_rate": 8.93154522862476e-07, "loss": 0.1264, "step": 2582 }, { "epoch": 1.62, "learning_rate": 8.903139946557437e-07, "loss": 0.1125, "step": 2583 }, { "epoch": 1.62, "learning_rate": 8.874775490405712e-07, "loss": 0.0946, "step": 2584 }, { "epoch": 1.62, "learning_rate": 8.846451888347018e-07, "loss": 0.1455, "step": 2585 }, { "epoch": 1.62, "learning_rate": 8.818169168518237e-07, "loss": 0.1205, "step": 2586 }, { "epoch": 1.62, "learning_rate": 8.789927359015643e-07, "loss": 0.0996, "step": 2587 }, { "epoch": 1.63, "learning_rate": 8.761726487894817e-07, "loss": 0.1368, "step": 2588 }, { "epoch": 1.63, "learning_rate": 8.733566583170727e-07, "loss": 0.1171, "step": 2589 }, { "epoch": 1.63, "learning_rate": 8.705447672817635e-07, "loss": 0.1186, "step": 2590 }, { "epoch": 1.63, "learning_rate": 8.677369784769041e-07, "loss": 0.1258, "step": 2591 }, { "epoch": 1.63, "learning_rate": 8.649332946917749e-07, "loss": 0.1108, "step": 2592 }, { "epoch": 1.63, "learning_rate": 8.621337187115742e-07, "loss": 0.1202, "step": 2593 }, { "epoch": 1.63, "learning_rate": 8.593382533174194e-07, "loss": 0.1161, "step": 2594 }, { "epoch": 1.63, "learning_rate": 8.565469012863481e-07, "loss": 0.139, "step": 2595 }, { "epoch": 1.63, "learning_rate": 8.537596653913072e-07, "loss": 0.1205, "step": 2596 }, { "epoch": 1.63, "learning_rate": 8.50976548401159e-07, "loss": 0.0995, "step": 2597 }, { "epoch": 1.63, "learning_rate": 8.481975530806713e-07, "loss": 0.1266, "step": 2598 }, { "epoch": 1.63, "learning_rate": 8.454226821905171e-07, "loss": 0.1323, "step": 2599 }, { "epoch": 1.63, "learning_rate": 8.426519384872733e-07, "loss": 0.1013, "step": 2600 } ], "logging_steps": 1.0, "max_steps": 3184, "num_train_epochs": 2, "save_steps": 50, "total_flos": 0.0, "trial_name": null, "trial_params": null }