{ "best_metric": 0.9057239057239057, "best_model_checkpoint": "videomae-base-finetuned-kinetics-finetuned-class-fc-35k/checkpoint-1288", "epoch": 1.0500777604976672, "eval_steps": 500, "global_step": 2576, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 9.999611197511664e-06, "loss": 0.6777, "step": 1 }, { "epoch": 0.0, "learning_rate": 9.99922239502333e-06, "loss": 0.6774, "step": 2 }, { "epoch": 0.0, "learning_rate": 9.998833592534993e-06, "loss": 0.7183, "step": 3 }, { "epoch": 0.0, "learning_rate": 9.998444790046657e-06, "loss": 0.7489, "step": 4 }, { "epoch": 0.0, "learning_rate": 9.998055987558322e-06, "loss": 0.7105, "step": 5 }, { "epoch": 0.0, "learning_rate": 9.997667185069986e-06, "loss": 0.6813, "step": 6 }, { "epoch": 0.0, "learning_rate": 9.99727838258165e-06, "loss": 0.6919, "step": 7 }, { "epoch": 0.0, "learning_rate": 9.996889580093313e-06, "loss": 0.7043, "step": 8 }, { "epoch": 0.0, "learning_rate": 9.996500777604977e-06, "loss": 0.689, "step": 9 }, { "epoch": 0.0, "learning_rate": 9.99611197511664e-06, "loss": 0.7073, "step": 10 }, { "epoch": 0.0, "learning_rate": 9.995723172628306e-06, "loss": 0.7, "step": 11 }, { "epoch": 0.0, "learning_rate": 9.99533437013997e-06, "loss": 0.6954, "step": 12 }, { "epoch": 0.0, "learning_rate": 9.994945567651633e-06, "loss": 0.6928, "step": 13 }, { "epoch": 0.0, "learning_rate": 9.994556765163298e-06, "loss": 0.6872, "step": 14 }, { "epoch": 0.0, "learning_rate": 9.994167962674962e-06, "loss": 0.6886, "step": 15 }, { "epoch": 0.0, "learning_rate": 9.993779160186626e-06, "loss": 0.6789, "step": 16 }, { "epoch": 0.0, "learning_rate": 9.993390357698291e-06, "loss": 0.6623, "step": 17 }, { "epoch": 0.0, "learning_rate": 9.993001555209955e-06, "loss": 0.6855, "step": 18 }, { "epoch": 0.0, "learning_rate": 9.992612752721618e-06, "loss": 0.7227, "step": 19 }, { "epoch": 0.0, "learning_rate": 9.992223950233282e-06, "loss": 0.649, "step": 20 }, { "epoch": 0.0, "learning_rate": 9.991835147744945e-06, "loss": 0.6797, "step": 21 }, { "epoch": 0.0, "learning_rate": 9.99144634525661e-06, "loss": 0.7441, "step": 22 }, { "epoch": 0.0, "learning_rate": 9.991057542768274e-06, "loss": 0.6967, "step": 23 }, { "epoch": 0.0, "learning_rate": 9.990668740279938e-06, "loss": 0.6495, "step": 24 }, { "epoch": 0.0, "learning_rate": 9.990279937791603e-06, "loss": 0.6557, "step": 25 }, { "epoch": 0.0, "learning_rate": 9.989891135303267e-06, "loss": 0.6386, "step": 26 }, { "epoch": 0.0, "learning_rate": 9.98950233281493e-06, "loss": 0.6615, "step": 27 }, { "epoch": 0.0, "learning_rate": 9.989113530326596e-06, "loss": 0.6628, "step": 28 }, { "epoch": 0.0, "learning_rate": 9.98872472783826e-06, "loss": 0.716, "step": 29 }, { "epoch": 0.0, "learning_rate": 9.988335925349923e-06, "loss": 0.6903, "step": 30 }, { "epoch": 0.0, "learning_rate": 9.987947122861587e-06, "loss": 0.6408, "step": 31 }, { "epoch": 0.0, "learning_rate": 9.98755832037325e-06, "loss": 0.711, "step": 32 }, { "epoch": 0.0, "learning_rate": 9.987169517884914e-06, "loss": 0.6702, "step": 33 }, { "epoch": 0.0, "learning_rate": 9.98678071539658e-06, "loss": 0.7442, "step": 34 }, { "epoch": 0.0, "learning_rate": 9.986391912908243e-06, "loss": 0.6251, "step": 35 }, { "epoch": 0.0, "learning_rate": 9.986003110419907e-06, "loss": 0.6892, "step": 36 }, { "epoch": 0.0, "learning_rate": 9.985614307931572e-06, "loss": 0.6633, "step": 37 }, { "epoch": 0.0, "learning_rate": 9.985225505443236e-06, "loss": 0.658, "step": 38 }, { "epoch": 0.0, "learning_rate": 9.9848367029549e-06, "loss": 0.6399, "step": 39 }, { "epoch": 0.0, "learning_rate": 9.984447900466565e-06, "loss": 0.6831, "step": 40 }, { "epoch": 0.0, "learning_rate": 9.984059097978228e-06, "loss": 0.6765, "step": 41 }, { "epoch": 0.0, "learning_rate": 9.983670295489892e-06, "loss": 0.7162, "step": 42 }, { "epoch": 0.0, "learning_rate": 9.983281493001557e-06, "loss": 0.6351, "step": 43 }, { "epoch": 0.0, "learning_rate": 9.98289269051322e-06, "loss": 0.6382, "step": 44 }, { "epoch": 0.0, "learning_rate": 9.982503888024884e-06, "loss": 0.6922, "step": 45 }, { "epoch": 0.0, "learning_rate": 9.982115085536548e-06, "loss": 0.6923, "step": 46 }, { "epoch": 0.0, "learning_rate": 9.981726283048212e-06, "loss": 0.6596, "step": 47 }, { "epoch": 0.0, "learning_rate": 9.981337480559877e-06, "loss": 0.6645, "step": 48 }, { "epoch": 0.0, "learning_rate": 9.98094867807154e-06, "loss": 0.6425, "step": 49 }, { "epoch": 0.0, "learning_rate": 9.980559875583204e-06, "loss": 0.6348, "step": 50 }, { "epoch": 0.0, "learning_rate": 9.980171073094868e-06, "loss": 0.6181, "step": 51 }, { "epoch": 0.0, "learning_rate": 9.979782270606533e-06, "loss": 0.6463, "step": 52 }, { "epoch": 0.0, "learning_rate": 9.979393468118197e-06, "loss": 0.6714, "step": 53 }, { "epoch": 0.0, "learning_rate": 9.97900466562986e-06, "loss": 0.6233, "step": 54 }, { "epoch": 0.0, "learning_rate": 9.978615863141526e-06, "loss": 0.7113, "step": 55 }, { "epoch": 0.0, "learning_rate": 9.97822706065319e-06, "loss": 0.6165, "step": 56 }, { "epoch": 0.0, "learning_rate": 9.977838258164853e-06, "loss": 0.602, "step": 57 }, { "epoch": 0.0, "learning_rate": 9.977449455676517e-06, "loss": 0.6596, "step": 58 }, { "epoch": 0.0, "learning_rate": 9.97706065318818e-06, "loss": 0.6407, "step": 59 }, { "epoch": 0.0, "learning_rate": 9.976671850699846e-06, "loss": 0.6076, "step": 60 }, { "epoch": 0.0, "learning_rate": 9.97628304821151e-06, "loss": 0.6342, "step": 61 }, { "epoch": 0.0, "learning_rate": 9.975894245723173e-06, "loss": 0.705, "step": 62 }, { "epoch": 0.0, "learning_rate": 9.975505443234838e-06, "loss": 0.6303, "step": 63 }, { "epoch": 0.0, "learning_rate": 9.975116640746502e-06, "loss": 0.6421, "step": 64 }, { "epoch": 0.0, "learning_rate": 9.974727838258165e-06, "loss": 0.5896, "step": 65 }, { "epoch": 0.0, "learning_rate": 9.97433903576983e-06, "loss": 0.6126, "step": 66 }, { "epoch": 0.0, "learning_rate": 9.973950233281494e-06, "loss": 0.6355, "step": 67 }, { "epoch": 0.0, "learning_rate": 9.973561430793158e-06, "loss": 0.5874, "step": 68 }, { "epoch": 0.0, "learning_rate": 9.973172628304822e-06, "loss": 0.5841, "step": 69 }, { "epoch": 0.0, "learning_rate": 9.972783825816485e-06, "loss": 0.6167, "step": 70 }, { "epoch": 0.0, "learning_rate": 9.972395023328149e-06, "loss": 0.6169, "step": 71 }, { "epoch": 0.0, "learning_rate": 9.972006220839814e-06, "loss": 0.5527, "step": 72 }, { "epoch": 0.0, "learning_rate": 9.971617418351478e-06, "loss": 0.6401, "step": 73 }, { "epoch": 0.0, "learning_rate": 9.971228615863141e-06, "loss": 0.6801, "step": 74 }, { "epoch": 0.0, "learning_rate": 9.970839813374807e-06, "loss": 0.6195, "step": 75 }, { "epoch": 0.0, "learning_rate": 9.97045101088647e-06, "loss": 0.5555, "step": 76 }, { "epoch": 0.0, "learning_rate": 9.970062208398134e-06, "loss": 0.4961, "step": 77 }, { "epoch": 0.0, "learning_rate": 9.9696734059098e-06, "loss": 0.6965, "step": 78 }, { "epoch": 0.0, "learning_rate": 9.969284603421463e-06, "loss": 0.5376, "step": 79 }, { "epoch": 0.0, "learning_rate": 9.968895800933127e-06, "loss": 0.6359, "step": 80 }, { "epoch": 0.0, "learning_rate": 9.968506998444792e-06, "loss": 0.7635, "step": 81 }, { "epoch": 0.0, "learning_rate": 9.968118195956456e-06, "loss": 0.7944, "step": 82 }, { "epoch": 0.0, "learning_rate": 9.96772939346812e-06, "loss": 0.5653, "step": 83 }, { "epoch": 0.0, "learning_rate": 9.967340590979783e-06, "loss": 0.5059, "step": 84 }, { "epoch": 0.0, "learning_rate": 9.966951788491446e-06, "loss": 0.6833, "step": 85 }, { "epoch": 0.0, "learning_rate": 9.966562986003112e-06, "loss": 0.6116, "step": 86 }, { "epoch": 0.0, "learning_rate": 9.966174183514775e-06, "loss": 0.5188, "step": 87 }, { "epoch": 0.0, "learning_rate": 9.965785381026439e-06, "loss": 0.6334, "step": 88 }, { "epoch": 0.0, "learning_rate": 9.965396578538103e-06, "loss": 0.4306, "step": 89 }, { "epoch": 0.0, "learning_rate": 9.965007776049768e-06, "loss": 0.5548, "step": 90 }, { "epoch": 0.0, "learning_rate": 9.964618973561432e-06, "loss": 0.5597, "step": 91 }, { "epoch": 0.0, "learning_rate": 9.964230171073095e-06, "loss": 0.5666, "step": 92 }, { "epoch": 0.0, "learning_rate": 9.96384136858476e-06, "loss": 0.4973, "step": 93 }, { "epoch": 0.0, "learning_rate": 9.963452566096424e-06, "loss": 0.5162, "step": 94 }, { "epoch": 0.0, "learning_rate": 9.963063763608088e-06, "loss": 0.5611, "step": 95 }, { "epoch": 0.0, "learning_rate": 9.962674961119751e-06, "loss": 0.4131, "step": 96 }, { "epoch": 0.0, "learning_rate": 9.962286158631415e-06, "loss": 0.4473, "step": 97 }, { "epoch": 0.0, "learning_rate": 9.96189735614308e-06, "loss": 0.6576, "step": 98 }, { "epoch": 0.0, "learning_rate": 9.961508553654744e-06, "loss": 0.4737, "step": 99 }, { "epoch": 0.0, "learning_rate": 9.961119751166408e-06, "loss": 0.5215, "step": 100 }, { "epoch": 0.0, "learning_rate": 9.960730948678073e-06, "loss": 0.5549, "step": 101 }, { "epoch": 0.0, "learning_rate": 9.960342146189737e-06, "loss": 0.5727, "step": 102 }, { "epoch": 0.0, "learning_rate": 9.9599533437014e-06, "loss": 0.319, "step": 103 }, { "epoch": 0.0, "learning_rate": 9.959564541213066e-06, "loss": 0.5689, "step": 104 }, { "epoch": 0.0, "learning_rate": 9.95917573872473e-06, "loss": 0.4203, "step": 105 }, { "epoch": 0.0, "learning_rate": 9.958786936236393e-06, "loss": 0.4436, "step": 106 }, { "epoch": 0.0, "learning_rate": 9.958398133748058e-06, "loss": 0.4524, "step": 107 }, { "epoch": 0.0, "learning_rate": 9.95800933125972e-06, "loss": 0.6149, "step": 108 }, { "epoch": 0.0, "learning_rate": 9.957620528771384e-06, "loss": 0.4411, "step": 109 }, { "epoch": 0.0, "learning_rate": 9.957231726283049e-06, "loss": 0.5067, "step": 110 }, { "epoch": 0.0, "learning_rate": 9.956842923794713e-06, "loss": 0.4483, "step": 111 }, { "epoch": 0.0, "learning_rate": 9.956454121306376e-06, "loss": 0.457, "step": 112 }, { "epoch": 0.0, "learning_rate": 9.956065318818042e-06, "loss": 0.3778, "step": 113 }, { "epoch": 0.0, "learning_rate": 9.955676516329705e-06, "loss": 0.448, "step": 114 }, { "epoch": 0.0, "learning_rate": 9.955287713841369e-06, "loss": 0.5136, "step": 115 }, { "epoch": 0.0, "learning_rate": 9.954898911353034e-06, "loss": 0.3659, "step": 116 }, { "epoch": 0.0, "learning_rate": 9.954510108864698e-06, "loss": 0.5226, "step": 117 }, { "epoch": 0.0, "learning_rate": 9.954121306376361e-06, "loss": 0.4151, "step": 118 }, { "epoch": 0.0, "learning_rate": 9.953732503888027e-06, "loss": 0.5927, "step": 119 }, { "epoch": 0.0, "learning_rate": 9.95334370139969e-06, "loss": 0.4243, "step": 120 }, { "epoch": 0.0, "learning_rate": 9.952954898911354e-06, "loss": 0.286, "step": 121 }, { "epoch": 0.0, "learning_rate": 9.952566096423018e-06, "loss": 0.6714, "step": 122 }, { "epoch": 0.0, "learning_rate": 9.952177293934681e-06, "loss": 0.3588, "step": 123 }, { "epoch": 0.0, "learning_rate": 9.951788491446347e-06, "loss": 0.4278, "step": 124 }, { "epoch": 0.0, "learning_rate": 9.95139968895801e-06, "loss": 0.496, "step": 125 }, { "epoch": 0.0, "learning_rate": 9.951010886469674e-06, "loss": 0.8661, "step": 126 }, { "epoch": 0.0, "learning_rate": 9.95062208398134e-06, "loss": 0.4017, "step": 127 }, { "epoch": 0.0, "learning_rate": 9.950233281493003e-06, "loss": 0.3287, "step": 128 }, { "epoch": 0.01, "learning_rate": 9.949844479004666e-06, "loss": 0.4936, "step": 129 }, { "epoch": 0.01, "learning_rate": 9.94945567651633e-06, "loss": 0.5004, "step": 130 }, { "epoch": 0.01, "learning_rate": 9.949066874027995e-06, "loss": 0.5015, "step": 131 }, { "epoch": 0.01, "learning_rate": 9.948678071539659e-06, "loss": 0.2217, "step": 132 }, { "epoch": 0.01, "learning_rate": 9.948289269051323e-06, "loss": 0.4372, "step": 133 }, { "epoch": 0.01, "learning_rate": 9.947900466562986e-06, "loss": 0.4201, "step": 134 }, { "epoch": 0.01, "learning_rate": 9.94751166407465e-06, "loss": 0.2846, "step": 135 }, { "epoch": 0.01, "learning_rate": 9.947122861586315e-06, "loss": 0.3995, "step": 136 }, { "epoch": 0.01, "learning_rate": 9.946734059097979e-06, "loss": 0.2506, "step": 137 }, { "epoch": 0.01, "learning_rate": 9.946345256609642e-06, "loss": 0.7523, "step": 138 }, { "epoch": 0.01, "learning_rate": 9.945956454121308e-06, "loss": 0.4691, "step": 139 }, { "epoch": 0.01, "learning_rate": 9.945567651632971e-06, "loss": 0.4785, "step": 140 }, { "epoch": 0.01, "learning_rate": 9.945178849144635e-06, "loss": 0.3503, "step": 141 }, { "epoch": 0.01, "learning_rate": 9.9447900466563e-06, "loss": 0.3068, "step": 142 }, { "epoch": 0.01, "learning_rate": 9.944401244167964e-06, "loss": 0.5316, "step": 143 }, { "epoch": 0.01, "learning_rate": 9.944012441679628e-06, "loss": 0.2478, "step": 144 }, { "epoch": 0.01, "learning_rate": 9.943623639191293e-06, "loss": 0.5122, "step": 145 }, { "epoch": 0.01, "learning_rate": 9.943234836702955e-06, "loss": 0.4335, "step": 146 }, { "epoch": 0.01, "learning_rate": 9.94284603421462e-06, "loss": 0.2705, "step": 147 }, { "epoch": 0.01, "learning_rate": 9.942457231726284e-06, "loss": 0.272, "step": 148 }, { "epoch": 0.01, "learning_rate": 9.942068429237947e-06, "loss": 0.2714, "step": 149 }, { "epoch": 0.01, "learning_rate": 9.941679626749611e-06, "loss": 0.6988, "step": 150 }, { "epoch": 0.01, "learning_rate": 9.941290824261276e-06, "loss": 0.6198, "step": 151 }, { "epoch": 0.01, "learning_rate": 9.94090202177294e-06, "loss": 0.2285, "step": 152 }, { "epoch": 0.01, "learning_rate": 9.940513219284604e-06, "loss": 0.5953, "step": 153 }, { "epoch": 0.01, "learning_rate": 9.940124416796269e-06, "loss": 0.294, "step": 154 }, { "epoch": 0.01, "learning_rate": 9.939735614307933e-06, "loss": 0.4563, "step": 155 }, { "epoch": 0.01, "learning_rate": 9.939346811819596e-06, "loss": 0.7298, "step": 156 }, { "epoch": 0.01, "learning_rate": 9.938958009331262e-06, "loss": 0.3239, "step": 157 }, { "epoch": 0.01, "learning_rate": 9.938569206842925e-06, "loss": 0.4191, "step": 158 }, { "epoch": 0.01, "learning_rate": 9.938180404354589e-06, "loss": 0.6343, "step": 159 }, { "epoch": 0.01, "learning_rate": 9.937791601866252e-06, "loss": 0.4537, "step": 160 }, { "epoch": 0.01, "learning_rate": 9.937402799377916e-06, "loss": 0.4265, "step": 161 }, { "epoch": 0.01, "learning_rate": 9.937013996889581e-06, "loss": 0.3927, "step": 162 }, { "epoch": 0.01, "learning_rate": 9.936625194401245e-06, "loss": 0.4147, "step": 163 }, { "epoch": 0.01, "learning_rate": 9.936236391912909e-06, "loss": 0.2799, "step": 164 }, { "epoch": 0.01, "learning_rate": 9.935847589424574e-06, "loss": 0.3504, "step": 165 }, { "epoch": 0.01, "learning_rate": 9.935458786936238e-06, "loss": 0.5513, "step": 166 }, { "epoch": 0.01, "learning_rate": 9.935069984447901e-06, "loss": 0.4448, "step": 167 }, { "epoch": 0.01, "learning_rate": 9.934681181959565e-06, "loss": 0.2058, "step": 168 }, { "epoch": 0.01, "learning_rate": 9.93429237947123e-06, "loss": 0.6277, "step": 169 }, { "epoch": 0.01, "learning_rate": 9.933903576982894e-06, "loss": 0.387, "step": 170 }, { "epoch": 0.01, "learning_rate": 9.933514774494557e-06, "loss": 0.4316, "step": 171 }, { "epoch": 0.01, "learning_rate": 9.933125972006221e-06, "loss": 0.2842, "step": 172 }, { "epoch": 0.01, "learning_rate": 9.932737169517885e-06, "loss": 0.6212, "step": 173 }, { "epoch": 0.01, "learning_rate": 9.93234836702955e-06, "loss": 0.5228, "step": 174 }, { "epoch": 0.01, "learning_rate": 9.931959564541214e-06, "loss": 0.3519, "step": 175 }, { "epoch": 0.01, "learning_rate": 9.931570762052877e-06, "loss": 0.3247, "step": 176 }, { "epoch": 0.01, "learning_rate": 9.931181959564543e-06, "loss": 0.5055, "step": 177 }, { "epoch": 0.01, "learning_rate": 9.930793157076206e-06, "loss": 0.3766, "step": 178 }, { "epoch": 0.01, "learning_rate": 9.93040435458787e-06, "loss": 0.3487, "step": 179 }, { "epoch": 0.01, "learning_rate": 9.930015552099535e-06, "loss": 0.3972, "step": 180 }, { "epoch": 0.01, "learning_rate": 9.929626749611199e-06, "loss": 0.227, "step": 181 }, { "epoch": 0.01, "learning_rate": 9.929237947122862e-06, "loss": 0.4875, "step": 182 }, { "epoch": 0.01, "learning_rate": 9.928849144634528e-06, "loss": 0.4304, "step": 183 }, { "epoch": 0.01, "learning_rate": 9.92846034214619e-06, "loss": 0.6761, "step": 184 }, { "epoch": 0.01, "learning_rate": 9.928071539657855e-06, "loss": 0.3669, "step": 185 }, { "epoch": 0.01, "learning_rate": 9.927682737169519e-06, "loss": 0.5097, "step": 186 }, { "epoch": 0.01, "learning_rate": 9.927293934681182e-06, "loss": 0.4293, "step": 187 }, { "epoch": 0.01, "learning_rate": 9.926905132192846e-06, "loss": 0.219, "step": 188 }, { "epoch": 0.01, "learning_rate": 9.926516329704511e-06, "loss": 0.3351, "step": 189 }, { "epoch": 0.01, "learning_rate": 9.926127527216175e-06, "loss": 0.2016, "step": 190 }, { "epoch": 0.01, "learning_rate": 9.925738724727838e-06, "loss": 0.4046, "step": 191 }, { "epoch": 0.01, "learning_rate": 9.925349922239504e-06, "loss": 0.4652, "step": 192 }, { "epoch": 0.01, "learning_rate": 9.924961119751167e-06, "loss": 0.6114, "step": 193 }, { "epoch": 0.01, "learning_rate": 9.924572317262831e-06, "loss": 0.4533, "step": 194 }, { "epoch": 0.01, "learning_rate": 9.924183514774496e-06, "loss": 0.4018, "step": 195 }, { "epoch": 0.01, "learning_rate": 9.92379471228616e-06, "loss": 0.7365, "step": 196 }, { "epoch": 0.01, "learning_rate": 9.923405909797824e-06, "loss": 0.3536, "step": 197 }, { "epoch": 0.01, "learning_rate": 9.923017107309487e-06, "loss": 0.3361, "step": 198 }, { "epoch": 0.01, "learning_rate": 9.922628304821151e-06, "loss": 0.4504, "step": 199 }, { "epoch": 0.01, "learning_rate": 9.922239502332816e-06, "loss": 0.2772, "step": 200 }, { "epoch": 0.01, "learning_rate": 9.92185069984448e-06, "loss": 0.3715, "step": 201 }, { "epoch": 0.01, "learning_rate": 9.921461897356143e-06, "loss": 0.3799, "step": 202 }, { "epoch": 0.01, "learning_rate": 9.921073094867809e-06, "loss": 0.5721, "step": 203 }, { "epoch": 0.01, "learning_rate": 9.920684292379472e-06, "loss": 0.4771, "step": 204 }, { "epoch": 0.01, "learning_rate": 9.920295489891136e-06, "loss": 0.2996, "step": 205 }, { "epoch": 0.01, "learning_rate": 9.919906687402801e-06, "loss": 0.1997, "step": 206 }, { "epoch": 0.01, "learning_rate": 9.919517884914465e-06, "loss": 0.4195, "step": 207 }, { "epoch": 0.01, "learning_rate": 9.919129082426129e-06, "loss": 0.3926, "step": 208 }, { "epoch": 0.01, "learning_rate": 9.918740279937792e-06, "loss": 0.4048, "step": 209 }, { "epoch": 0.01, "learning_rate": 9.918351477449456e-06, "loss": 0.3625, "step": 210 }, { "epoch": 0.01, "learning_rate": 9.91796267496112e-06, "loss": 0.5558, "step": 211 }, { "epoch": 0.01, "learning_rate": 9.917573872472785e-06, "loss": 0.2951, "step": 212 }, { "epoch": 0.01, "learning_rate": 9.917185069984448e-06, "loss": 0.3423, "step": 213 }, { "epoch": 0.01, "learning_rate": 9.916796267496112e-06, "loss": 0.5455, "step": 214 }, { "epoch": 0.01, "learning_rate": 9.916407465007777e-06, "loss": 0.3702, "step": 215 }, { "epoch": 0.01, "learning_rate": 9.916018662519441e-06, "loss": 0.2729, "step": 216 }, { "epoch": 0.01, "learning_rate": 9.915629860031105e-06, "loss": 0.3657, "step": 217 }, { "epoch": 0.01, "learning_rate": 9.91524105754277e-06, "loss": 0.4765, "step": 218 }, { "epoch": 0.01, "learning_rate": 9.914852255054434e-06, "loss": 0.2855, "step": 219 }, { "epoch": 0.01, "learning_rate": 9.914463452566097e-06, "loss": 0.4208, "step": 220 }, { "epoch": 0.01, "learning_rate": 9.914074650077763e-06, "loss": 0.4596, "step": 221 }, { "epoch": 0.01, "learning_rate": 9.913685847589424e-06, "loss": 0.4146, "step": 222 }, { "epoch": 0.01, "learning_rate": 9.91329704510109e-06, "loss": 0.4149, "step": 223 }, { "epoch": 0.01, "learning_rate": 9.912908242612753e-06, "loss": 0.312, "step": 224 }, { "epoch": 0.01, "learning_rate": 9.912519440124417e-06, "loss": 0.5113, "step": 225 }, { "epoch": 0.01, "learning_rate": 9.912130637636082e-06, "loss": 0.3813, "step": 226 }, { "epoch": 0.01, "learning_rate": 9.911741835147746e-06, "loss": 0.641, "step": 227 }, { "epoch": 0.01, "learning_rate": 9.91135303265941e-06, "loss": 0.4206, "step": 228 }, { "epoch": 0.01, "learning_rate": 9.910964230171073e-06, "loss": 0.3397, "step": 229 }, { "epoch": 0.01, "learning_rate": 9.910575427682739e-06, "loss": 0.2875, "step": 230 }, { "epoch": 0.01, "learning_rate": 9.910186625194402e-06, "loss": 0.4528, "step": 231 }, { "epoch": 0.01, "learning_rate": 9.909797822706066e-06, "loss": 0.3025, "step": 232 }, { "epoch": 0.01, "learning_rate": 9.909409020217731e-06, "loss": 0.3954, "step": 233 }, { "epoch": 0.01, "learning_rate": 9.909020217729395e-06, "loss": 0.2455, "step": 234 }, { "epoch": 0.01, "learning_rate": 9.908631415241058e-06, "loss": 0.3266, "step": 235 }, { "epoch": 0.01, "learning_rate": 9.908242612752722e-06, "loss": 0.3284, "step": 236 }, { "epoch": 0.01, "learning_rate": 9.907853810264386e-06, "loss": 0.3295, "step": 237 }, { "epoch": 0.01, "learning_rate": 9.907465007776051e-06, "loss": 0.5269, "step": 238 }, { "epoch": 0.01, "learning_rate": 9.907076205287715e-06, "loss": 0.2892, "step": 239 }, { "epoch": 0.01, "learning_rate": 9.906687402799378e-06, "loss": 0.2164, "step": 240 }, { "epoch": 0.01, "learning_rate": 9.906298600311044e-06, "loss": 0.1715, "step": 241 }, { "epoch": 0.01, "learning_rate": 9.905909797822707e-06, "loss": 0.4704, "step": 242 }, { "epoch": 0.01, "learning_rate": 9.90552099533437e-06, "loss": 0.5001, "step": 243 }, { "epoch": 0.01, "learning_rate": 9.905132192846036e-06, "loss": 0.2924, "step": 244 }, { "epoch": 0.01, "learning_rate": 9.9047433903577e-06, "loss": 0.3362, "step": 245 }, { "epoch": 0.01, "learning_rate": 9.904354587869363e-06, "loss": 0.6158, "step": 246 }, { "epoch": 0.01, "learning_rate": 9.903965785381027e-06, "loss": 0.4011, "step": 247 }, { "epoch": 0.01, "learning_rate": 9.90357698289269e-06, "loss": 0.3824, "step": 248 }, { "epoch": 0.01, "learning_rate": 9.903188180404354e-06, "loss": 0.3756, "step": 249 }, { "epoch": 0.01, "learning_rate": 9.90279937791602e-06, "loss": 0.1312, "step": 250 }, { "epoch": 0.01, "learning_rate": 9.902410575427683e-06, "loss": 0.2533, "step": 251 }, { "epoch": 0.01, "learning_rate": 9.902021772939347e-06, "loss": 0.3449, "step": 252 }, { "epoch": 0.01, "learning_rate": 9.901632970451012e-06, "loss": 0.3314, "step": 253 }, { "epoch": 0.01, "learning_rate": 9.901244167962676e-06, "loss": 0.5809, "step": 254 }, { "epoch": 0.01, "learning_rate": 9.90085536547434e-06, "loss": 0.1391, "step": 255 }, { "epoch": 0.01, "learning_rate": 9.900466562986005e-06, "loss": 0.318, "step": 256 }, { "epoch": 0.01, "learning_rate": 9.900077760497668e-06, "loss": 0.1318, "step": 257 }, { "epoch": 0.01, "learning_rate": 9.899688958009332e-06, "loss": 0.552, "step": 258 }, { "epoch": 0.01, "learning_rate": 9.899300155520997e-06, "loss": 0.7435, "step": 259 }, { "epoch": 0.01, "learning_rate": 9.89891135303266e-06, "loss": 0.2203, "step": 260 }, { "epoch": 0.01, "learning_rate": 9.898522550544325e-06, "loss": 0.5877, "step": 261 }, { "epoch": 0.01, "learning_rate": 9.898133748055988e-06, "loss": 0.4416, "step": 262 }, { "epoch": 0.01, "learning_rate": 9.897744945567652e-06, "loss": 0.2231, "step": 263 }, { "epoch": 0.01, "learning_rate": 9.897356143079317e-06, "loss": 0.313, "step": 264 }, { "epoch": 0.01, "learning_rate": 9.89696734059098e-06, "loss": 0.1442, "step": 265 }, { "epoch": 0.01, "learning_rate": 9.896578538102644e-06, "loss": 0.3188, "step": 266 }, { "epoch": 0.01, "learning_rate": 9.896189735614308e-06, "loss": 0.3906, "step": 267 }, { "epoch": 0.01, "learning_rate": 9.895800933125973e-06, "loss": 0.4107, "step": 268 }, { "epoch": 0.01, "learning_rate": 9.895412130637637e-06, "loss": 0.5213, "step": 269 }, { "epoch": 0.01, "learning_rate": 9.8950233281493e-06, "loss": 0.4627, "step": 270 }, { "epoch": 0.01, "learning_rate": 9.894634525660966e-06, "loss": 0.1631, "step": 271 }, { "epoch": 0.01, "learning_rate": 9.89424572317263e-06, "loss": 0.3353, "step": 272 }, { "epoch": 0.01, "learning_rate": 9.893856920684293e-06, "loss": 0.2015, "step": 273 }, { "epoch": 0.01, "learning_rate": 9.893468118195957e-06, "loss": 0.2398, "step": 274 }, { "epoch": 0.01, "learning_rate": 9.89307931570762e-06, "loss": 0.1922, "step": 275 }, { "epoch": 0.01, "learning_rate": 9.892690513219286e-06, "loss": 0.4644, "step": 276 }, { "epoch": 0.01, "learning_rate": 9.89230171073095e-06, "loss": 0.3709, "step": 277 }, { "epoch": 0.01, "learning_rate": 9.891912908242613e-06, "loss": 0.4182, "step": 278 }, { "epoch": 0.01, "learning_rate": 9.891524105754278e-06, "loss": 0.3994, "step": 279 }, { "epoch": 0.01, "learning_rate": 9.891135303265942e-06, "loss": 0.435, "step": 280 }, { "epoch": 0.01, "learning_rate": 9.890746500777606e-06, "loss": 0.3838, "step": 281 }, { "epoch": 0.01, "learning_rate": 9.890357698289271e-06, "loss": 0.333, "step": 282 }, { "epoch": 0.01, "learning_rate": 9.889968895800935e-06, "loss": 0.2271, "step": 283 }, { "epoch": 0.01, "learning_rate": 9.889580093312598e-06, "loss": 0.3545, "step": 284 }, { "epoch": 0.01, "learning_rate": 9.889191290824262e-06, "loss": 0.3729, "step": 285 }, { "epoch": 0.01, "learning_rate": 9.888802488335925e-06, "loss": 0.4665, "step": 286 }, { "epoch": 0.01, "learning_rate": 9.888413685847589e-06, "loss": 0.6178, "step": 287 }, { "epoch": 0.01, "learning_rate": 9.888024883359254e-06, "loss": 0.3168, "step": 288 }, { "epoch": 0.01, "learning_rate": 9.887636080870918e-06, "loss": 0.1806, "step": 289 }, { "epoch": 0.01, "learning_rate": 9.887247278382582e-06, "loss": 0.7655, "step": 290 }, { "epoch": 0.01, "learning_rate": 9.886858475894247e-06, "loss": 0.3271, "step": 291 }, { "epoch": 0.01, "learning_rate": 9.88646967340591e-06, "loss": 0.2243, "step": 292 }, { "epoch": 0.01, "learning_rate": 9.886080870917574e-06, "loss": 0.3793, "step": 293 }, { "epoch": 0.01, "learning_rate": 9.88569206842924e-06, "loss": 0.2042, "step": 294 }, { "epoch": 0.01, "learning_rate": 9.885303265940903e-06, "loss": 0.7342, "step": 295 }, { "epoch": 0.01, "learning_rate": 9.884914463452567e-06, "loss": 0.4231, "step": 296 }, { "epoch": 0.01, "learning_rate": 9.884525660964232e-06, "loss": 0.5069, "step": 297 }, { "epoch": 0.01, "learning_rate": 9.884136858475894e-06, "loss": 0.3039, "step": 298 }, { "epoch": 0.01, "learning_rate": 9.88374805598756e-06, "loss": 0.2563, "step": 299 }, { "epoch": 0.01, "learning_rate": 9.883359253499223e-06, "loss": 0.353, "step": 300 }, { "epoch": 0.01, "learning_rate": 9.882970451010887e-06, "loss": 0.6334, "step": 301 }, { "epoch": 0.01, "learning_rate": 9.882581648522552e-06, "loss": 0.2789, "step": 302 }, { "epoch": 0.01, "learning_rate": 9.882192846034216e-06, "loss": 0.3268, "step": 303 }, { "epoch": 0.01, "learning_rate": 9.88180404354588e-06, "loss": 0.215, "step": 304 }, { "epoch": 0.01, "learning_rate": 9.881415241057543e-06, "loss": 0.2369, "step": 305 }, { "epoch": 0.01, "learning_rate": 9.881026438569208e-06, "loss": 0.193, "step": 306 }, { "epoch": 0.01, "learning_rate": 9.880637636080872e-06, "loss": 0.2781, "step": 307 }, { "epoch": 0.01, "learning_rate": 9.880248833592535e-06, "loss": 0.4108, "step": 308 }, { "epoch": 0.01, "learning_rate": 9.8798600311042e-06, "loss": 0.4072, "step": 309 }, { "epoch": 0.01, "learning_rate": 9.879471228615864e-06, "loss": 0.2806, "step": 310 }, { "epoch": 0.01, "learning_rate": 9.879082426127528e-06, "loss": 0.3472, "step": 311 }, { "epoch": 0.01, "learning_rate": 9.878693623639192e-06, "loss": 0.2658, "step": 312 }, { "epoch": 0.01, "learning_rate": 9.878304821150855e-06, "loss": 0.6742, "step": 313 }, { "epoch": 0.01, "learning_rate": 9.87791601866252e-06, "loss": 0.5215, "step": 314 }, { "epoch": 0.01, "learning_rate": 9.877527216174184e-06, "loss": 0.2833, "step": 315 }, { "epoch": 0.01, "learning_rate": 9.877138413685848e-06, "loss": 0.3695, "step": 316 }, { "epoch": 0.01, "learning_rate": 9.876749611197513e-06, "loss": 0.5832, "step": 317 }, { "epoch": 0.01, "learning_rate": 9.876360808709177e-06, "loss": 0.3247, "step": 318 }, { "epoch": 0.01, "learning_rate": 9.87597200622084e-06, "loss": 0.4108, "step": 319 }, { "epoch": 0.01, "learning_rate": 9.875583203732506e-06, "loss": 0.3846, "step": 320 }, { "epoch": 0.01, "learning_rate": 9.87519440124417e-06, "loss": 0.3604, "step": 321 }, { "epoch": 0.01, "learning_rate": 9.874805598755833e-06, "loss": 0.4105, "step": 322 }, { "epoch": 0.01, "learning_rate": 9.874416796267497e-06, "loss": 0.6104, "step": 323 }, { "epoch": 0.01, "learning_rate": 9.87402799377916e-06, "loss": 0.4166, "step": 324 }, { "epoch": 0.01, "learning_rate": 9.873639191290824e-06, "loss": 0.6151, "step": 325 }, { "epoch": 0.01, "learning_rate": 9.87325038880249e-06, "loss": 0.3521, "step": 326 }, { "epoch": 0.01, "learning_rate": 9.872861586314153e-06, "loss": 0.3119, "step": 327 }, { "epoch": 0.01, "learning_rate": 9.872472783825816e-06, "loss": 0.4439, "step": 328 }, { "epoch": 0.01, "learning_rate": 9.872083981337482e-06, "loss": 0.3617, "step": 329 }, { "epoch": 0.01, "learning_rate": 9.871695178849145e-06, "loss": 0.2594, "step": 330 }, { "epoch": 0.01, "learning_rate": 9.871306376360809e-06, "loss": 0.3209, "step": 331 }, { "epoch": 0.01, "learning_rate": 9.870917573872474e-06, "loss": 0.4046, "step": 332 }, { "epoch": 0.01, "learning_rate": 9.870528771384138e-06, "loss": 0.3346, "step": 333 }, { "epoch": 0.01, "learning_rate": 9.870139968895802e-06, "loss": 0.3427, "step": 334 }, { "epoch": 0.01, "learning_rate": 9.869751166407467e-06, "loss": 0.4576, "step": 335 }, { "epoch": 0.01, "learning_rate": 9.869362363919129e-06, "loss": 0.3337, "step": 336 }, { "epoch": 0.01, "learning_rate": 9.868973561430794e-06, "loss": 0.4388, "step": 337 }, { "epoch": 0.01, "learning_rate": 9.868584758942458e-06, "loss": 0.2725, "step": 338 }, { "epoch": 0.01, "learning_rate": 9.868195956454121e-06, "loss": 0.4582, "step": 339 }, { "epoch": 0.01, "learning_rate": 9.867807153965787e-06, "loss": 0.3076, "step": 340 }, { "epoch": 0.01, "learning_rate": 9.86741835147745e-06, "loss": 0.6559, "step": 341 }, { "epoch": 0.01, "learning_rate": 9.867029548989114e-06, "loss": 0.1804, "step": 342 }, { "epoch": 0.01, "learning_rate": 9.86664074650078e-06, "loss": 0.3189, "step": 343 }, { "epoch": 0.01, "learning_rate": 9.866251944012443e-06, "loss": 0.3751, "step": 344 }, { "epoch": 0.01, "learning_rate": 9.865863141524107e-06, "loss": 0.3824, "step": 345 }, { "epoch": 0.01, "learning_rate": 9.86547433903577e-06, "loss": 0.445, "step": 346 }, { "epoch": 0.01, "learning_rate": 9.865085536547436e-06, "loss": 0.241, "step": 347 }, { "epoch": 0.01, "learning_rate": 9.8646967340591e-06, "loss": 0.6468, "step": 348 }, { "epoch": 0.01, "learning_rate": 9.864307931570763e-06, "loss": 0.243, "step": 349 }, { "epoch": 0.01, "learning_rate": 9.863919129082426e-06, "loss": 0.3332, "step": 350 }, { "epoch": 0.01, "learning_rate": 9.86353032659409e-06, "loss": 0.4973, "step": 351 }, { "epoch": 0.01, "learning_rate": 9.863141524105755e-06, "loss": 0.1949, "step": 352 }, { "epoch": 0.01, "learning_rate": 9.862752721617419e-06, "loss": 0.3253, "step": 353 }, { "epoch": 0.01, "learning_rate": 9.862363919129083e-06, "loss": 0.3263, "step": 354 }, { "epoch": 0.01, "learning_rate": 9.861975116640748e-06, "loss": 0.4494, "step": 355 }, { "epoch": 0.01, "learning_rate": 9.861586314152412e-06, "loss": 0.2436, "step": 356 }, { "epoch": 0.01, "learning_rate": 9.861197511664075e-06, "loss": 0.1651, "step": 357 }, { "epoch": 0.01, "learning_rate": 9.86080870917574e-06, "loss": 0.1387, "step": 358 }, { "epoch": 0.01, "learning_rate": 9.860419906687404e-06, "loss": 0.2604, "step": 359 }, { "epoch": 0.01, "learning_rate": 9.860031104199068e-06, "loss": 0.1694, "step": 360 }, { "epoch": 0.01, "learning_rate": 9.859642301710731e-06, "loss": 0.3507, "step": 361 }, { "epoch": 0.01, "learning_rate": 9.859253499222395e-06, "loss": 0.5129, "step": 362 }, { "epoch": 0.01, "learning_rate": 9.85886469673406e-06, "loss": 0.2048, "step": 363 }, { "epoch": 0.01, "learning_rate": 9.858475894245724e-06, "loss": 0.5124, "step": 364 }, { "epoch": 0.01, "learning_rate": 9.858087091757388e-06, "loss": 0.5399, "step": 365 }, { "epoch": 0.01, "learning_rate": 9.857698289269051e-06, "loss": 0.3691, "step": 366 }, { "epoch": 0.01, "learning_rate": 9.857309486780717e-06, "loss": 0.1534, "step": 367 }, { "epoch": 0.01, "learning_rate": 9.85692068429238e-06, "loss": 0.3788, "step": 368 }, { "epoch": 0.01, "learning_rate": 9.856531881804044e-06, "loss": 0.3987, "step": 369 }, { "epoch": 0.01, "learning_rate": 9.856143079315709e-06, "loss": 0.5371, "step": 370 }, { "epoch": 0.01, "learning_rate": 9.855754276827373e-06, "loss": 0.2463, "step": 371 }, { "epoch": 0.01, "learning_rate": 9.855365474339036e-06, "loss": 0.5395, "step": 372 }, { "epoch": 0.01, "learning_rate": 9.854976671850702e-06, "loss": 0.6109, "step": 373 }, { "epoch": 0.01, "learning_rate": 9.854587869362364e-06, "loss": 0.1901, "step": 374 }, { "epoch": 0.01, "learning_rate": 9.854199066874029e-06, "loss": 0.1791, "step": 375 }, { "epoch": 0.01, "learning_rate": 9.853810264385693e-06, "loss": 0.3077, "step": 376 }, { "epoch": 0.01, "learning_rate": 9.853421461897356e-06, "loss": 0.1874, "step": 377 }, { "epoch": 0.01, "learning_rate": 9.853032659409022e-06, "loss": 0.378, "step": 378 }, { "epoch": 0.01, "learning_rate": 9.852643856920685e-06, "loss": 0.148, "step": 379 }, { "epoch": 0.01, "learning_rate": 9.852255054432349e-06, "loss": 0.2914, "step": 380 }, { "epoch": 0.01, "learning_rate": 9.851866251944014e-06, "loss": 0.2887, "step": 381 }, { "epoch": 0.01, "learning_rate": 9.851477449455678e-06, "loss": 0.3898, "step": 382 }, { "epoch": 0.01, "learning_rate": 9.851088646967341e-06, "loss": 0.2087, "step": 383 }, { "epoch": 0.01, "learning_rate": 9.850699844479005e-06, "loss": 0.212, "step": 384 }, { "epoch": 0.01, "learning_rate": 9.85031104199067e-06, "loss": 0.4412, "step": 385 }, { "epoch": 0.02, "learning_rate": 9.849922239502334e-06, "loss": 0.3461, "step": 386 }, { "epoch": 0.02, "learning_rate": 9.849533437013998e-06, "loss": 0.2722, "step": 387 }, { "epoch": 0.02, "learning_rate": 9.849144634525661e-06, "loss": 0.3375, "step": 388 }, { "epoch": 0.02, "learning_rate": 9.848755832037325e-06, "loss": 0.3208, "step": 389 }, { "epoch": 0.02, "learning_rate": 9.84836702954899e-06, "loss": 0.2434, "step": 390 }, { "epoch": 0.02, "learning_rate": 9.847978227060654e-06, "loss": 0.3278, "step": 391 }, { "epoch": 0.02, "learning_rate": 9.847589424572317e-06, "loss": 0.1373, "step": 392 }, { "epoch": 0.02, "learning_rate": 9.847200622083983e-06, "loss": 0.3417, "step": 393 }, { "epoch": 0.02, "learning_rate": 9.846811819595646e-06, "loss": 0.304, "step": 394 }, { "epoch": 0.02, "learning_rate": 9.84642301710731e-06, "loss": 0.4361, "step": 395 }, { "epoch": 0.02, "learning_rate": 9.846034214618975e-06, "loss": 0.5287, "step": 396 }, { "epoch": 0.02, "learning_rate": 9.845645412130639e-06, "loss": 0.3049, "step": 397 }, { "epoch": 0.02, "learning_rate": 9.845256609642303e-06, "loss": 0.3699, "step": 398 }, { "epoch": 0.02, "learning_rate": 9.844867807153966e-06, "loss": 0.1589, "step": 399 }, { "epoch": 0.02, "learning_rate": 9.84447900466563e-06, "loss": 0.3877, "step": 400 }, { "epoch": 0.02, "learning_rate": 9.844090202177295e-06, "loss": 0.3018, "step": 401 }, { "epoch": 0.02, "learning_rate": 9.843701399688959e-06, "loss": 0.5712, "step": 402 }, { "epoch": 0.02, "learning_rate": 9.843312597200622e-06, "loss": 0.3674, "step": 403 }, { "epoch": 0.02, "learning_rate": 9.842923794712286e-06, "loss": 0.4342, "step": 404 }, { "epoch": 0.02, "learning_rate": 9.842534992223951e-06, "loss": 0.3791, "step": 405 }, { "epoch": 0.02, "learning_rate": 9.842146189735615e-06, "loss": 0.2897, "step": 406 }, { "epoch": 0.02, "learning_rate": 9.841757387247279e-06, "loss": 0.3437, "step": 407 }, { "epoch": 0.02, "learning_rate": 9.841368584758944e-06, "loss": 0.3404, "step": 408 }, { "epoch": 0.02, "learning_rate": 9.840979782270608e-06, "loss": 0.404, "step": 409 }, { "epoch": 0.02, "learning_rate": 9.840590979782271e-06, "loss": 0.1996, "step": 410 }, { "epoch": 0.02, "learning_rate": 9.840202177293937e-06, "loss": 0.204, "step": 411 }, { "epoch": 0.02, "learning_rate": 9.839813374805598e-06, "loss": 0.2184, "step": 412 }, { "epoch": 0.02, "learning_rate": 9.839424572317264e-06, "loss": 0.2333, "step": 413 }, { "epoch": 0.02, "learning_rate": 9.839035769828927e-06, "loss": 0.4191, "step": 414 }, { "epoch": 0.02, "learning_rate": 9.838646967340591e-06, "loss": 0.2754, "step": 415 }, { "epoch": 0.02, "learning_rate": 9.838258164852256e-06, "loss": 0.4373, "step": 416 }, { "epoch": 0.02, "learning_rate": 9.83786936236392e-06, "loss": 0.1077, "step": 417 }, { "epoch": 0.02, "learning_rate": 9.837480559875584e-06, "loss": 0.2873, "step": 418 }, { "epoch": 0.02, "learning_rate": 9.837091757387249e-06, "loss": 0.3369, "step": 419 }, { "epoch": 0.02, "learning_rate": 9.836702954898913e-06, "loss": 0.4217, "step": 420 }, { "epoch": 0.02, "learning_rate": 9.836314152410576e-06, "loss": 0.1227, "step": 421 }, { "epoch": 0.02, "learning_rate": 9.835925349922242e-06, "loss": 0.663, "step": 422 }, { "epoch": 0.02, "learning_rate": 9.835925349922242e-06, "loss": 0.5817, "step": 423 }, { "epoch": 0.02, "learning_rate": 9.835536547433905e-06, "loss": 0.4328, "step": 424 }, { "epoch": 0.02, "learning_rate": 9.835147744945569e-06, "loss": 0.6481, "step": 425 }, { "epoch": 0.02, "learning_rate": 9.834758942457232e-06, "loss": 0.1279, "step": 426 }, { "epoch": 0.02, "learning_rate": 9.834370139968896e-06, "loss": 0.9278, "step": 427 }, { "epoch": 0.02, "learning_rate": 9.83398133748056e-06, "loss": 0.5642, "step": 428 }, { "epoch": 0.02, "learning_rate": 9.833592534992225e-06, "loss": 0.499, "step": 429 }, { "epoch": 0.02, "learning_rate": 9.833203732503889e-06, "loss": 0.3602, "step": 430 }, { "epoch": 0.02, "learning_rate": 9.832814930015552e-06, "loss": 0.1525, "step": 431 }, { "epoch": 0.02, "learning_rate": 9.832426127527218e-06, "loss": 0.2212, "step": 432 }, { "epoch": 0.02, "learning_rate": 9.832037325038881e-06, "loss": 0.5552, "step": 433 }, { "epoch": 0.02, "learning_rate": 9.831648522550545e-06, "loss": 0.5691, "step": 434 }, { "epoch": 0.02, "learning_rate": 9.83125972006221e-06, "loss": 0.4612, "step": 435 }, { "epoch": 0.02, "learning_rate": 9.830870917573874e-06, "loss": 0.2771, "step": 436 }, { "epoch": 0.02, "learning_rate": 9.830482115085537e-06, "loss": 0.3139, "step": 437 }, { "epoch": 0.02, "learning_rate": 9.830093312597201e-06, "loss": 0.3454, "step": 438 }, { "epoch": 0.02, "learning_rate": 9.829704510108865e-06, "loss": 0.4506, "step": 439 }, { "epoch": 0.02, "learning_rate": 9.82931570762053e-06, "loss": 0.7831, "step": 440 }, { "epoch": 0.02, "learning_rate": 9.828926905132194e-06, "loss": 0.2347, "step": 441 }, { "epoch": 0.02, "learning_rate": 9.828538102643857e-06, "loss": 0.5763, "step": 442 }, { "epoch": 0.02, "learning_rate": 9.828149300155523e-06, "loss": 0.2286, "step": 443 }, { "epoch": 0.02, "learning_rate": 9.827760497667186e-06, "loss": 0.5154, "step": 444 }, { "epoch": 0.02, "learning_rate": 9.82737169517885e-06, "loss": 0.3131, "step": 445 }, { "epoch": 0.02, "learning_rate": 9.826982892690513e-06, "loss": 0.4719, "step": 446 }, { "epoch": 0.02, "learning_rate": 9.826594090202179e-06, "loss": 0.3151, "step": 447 }, { "epoch": 0.02, "learning_rate": 9.826205287713842e-06, "loss": 0.447, "step": 448 }, { "epoch": 0.02, "learning_rate": 9.825816485225506e-06, "loss": 0.2636, "step": 449 }, { "epoch": 0.02, "learning_rate": 9.825427682737171e-06, "loss": 0.3454, "step": 450 }, { "epoch": 0.02, "learning_rate": 9.825038880248833e-06, "loss": 0.3893, "step": 451 }, { "epoch": 0.02, "learning_rate": 9.824650077760499e-06, "loss": 0.251, "step": 452 }, { "epoch": 0.02, "learning_rate": 9.824261275272162e-06, "loss": 0.2372, "step": 453 }, { "epoch": 0.02, "learning_rate": 9.823872472783826e-06, "loss": 0.2133, "step": 454 }, { "epoch": 0.02, "learning_rate": 9.823483670295491e-06, "loss": 0.3505, "step": 455 }, { "epoch": 0.02, "learning_rate": 9.823094867807155e-06, "loss": 0.1206, "step": 456 }, { "epoch": 0.02, "learning_rate": 9.822706065318818e-06, "loss": 0.3378, "step": 457 }, { "epoch": 0.02, "learning_rate": 9.822317262830484e-06, "loss": 0.377, "step": 458 }, { "epoch": 0.02, "learning_rate": 9.821928460342147e-06, "loss": 0.3244, "step": 459 }, { "epoch": 0.02, "learning_rate": 9.821539657853811e-06, "loss": 0.4241, "step": 460 }, { "epoch": 0.02, "learning_rate": 9.821150855365476e-06, "loss": 0.544, "step": 461 }, { "epoch": 0.02, "learning_rate": 9.82076205287714e-06, "loss": 0.2368, "step": 462 }, { "epoch": 0.02, "learning_rate": 9.820373250388804e-06, "loss": 0.2996, "step": 463 }, { "epoch": 0.02, "learning_rate": 9.819984447900467e-06, "loss": 0.1721, "step": 464 }, { "epoch": 0.02, "learning_rate": 9.81959564541213e-06, "loss": 0.6556, "step": 465 }, { "epoch": 0.02, "learning_rate": 9.819206842923794e-06, "loss": 0.6221, "step": 466 }, { "epoch": 0.02, "learning_rate": 9.81881804043546e-06, "loss": 0.2403, "step": 467 }, { "epoch": 0.02, "learning_rate": 9.818429237947123e-06, "loss": 0.2641, "step": 468 }, { "epoch": 0.02, "learning_rate": 9.818040435458787e-06, "loss": 0.2079, "step": 469 }, { "epoch": 0.02, "learning_rate": 9.817651632970452e-06, "loss": 0.2692, "step": 470 }, { "epoch": 0.02, "learning_rate": 9.817262830482116e-06, "loss": 0.6408, "step": 471 }, { "epoch": 0.02, "learning_rate": 9.81687402799378e-06, "loss": 0.6249, "step": 472 }, { "epoch": 0.02, "learning_rate": 9.816485225505445e-06, "loss": 0.2792, "step": 473 }, { "epoch": 0.02, "learning_rate": 9.816096423017109e-06, "loss": 0.3716, "step": 474 }, { "epoch": 0.02, "learning_rate": 9.815707620528772e-06, "loss": 0.2558, "step": 475 }, { "epoch": 0.02, "learning_rate": 9.815318818040436e-06, "loss": 0.3263, "step": 476 }, { "epoch": 0.02, "learning_rate": 9.8149300155521e-06, "loss": 0.256, "step": 477 }, { "epoch": 0.02, "learning_rate": 9.814541213063765e-06, "loss": 0.2736, "step": 478 }, { "epoch": 0.02, "learning_rate": 9.814152410575428e-06, "loss": 0.8155, "step": 479 }, { "epoch": 0.02, "learning_rate": 9.813763608087092e-06, "loss": 0.365, "step": 480 }, { "epoch": 0.02, "learning_rate": 9.813374805598757e-06, "loss": 0.251, "step": 481 }, { "epoch": 0.02, "learning_rate": 9.812986003110421e-06, "loss": 0.4177, "step": 482 }, { "epoch": 0.02, "learning_rate": 9.812597200622085e-06, "loss": 0.4148, "step": 483 }, { "epoch": 0.02, "learning_rate": 9.812208398133748e-06, "loss": 0.3688, "step": 484 }, { "epoch": 0.02, "learning_rate": 9.811819595645414e-06, "loss": 0.2643, "step": 485 }, { "epoch": 0.02, "learning_rate": 9.811430793157077e-06, "loss": 0.3064, "step": 486 }, { "epoch": 0.02, "learning_rate": 9.81104199066874e-06, "loss": 0.2234, "step": 487 }, { "epoch": 0.02, "learning_rate": 9.810653188180406e-06, "loss": 0.5324, "step": 488 }, { "epoch": 0.02, "learning_rate": 9.810264385692068e-06, "loss": 0.1688, "step": 489 }, { "epoch": 0.02, "learning_rate": 9.809875583203733e-06, "loss": 0.161, "step": 490 }, { "epoch": 0.02, "learning_rate": 9.809486780715397e-06, "loss": 0.1732, "step": 491 }, { "epoch": 0.02, "learning_rate": 9.80909797822706e-06, "loss": 0.222, "step": 492 }, { "epoch": 0.02, "learning_rate": 9.808709175738726e-06, "loss": 0.0951, "step": 493 }, { "epoch": 0.02, "learning_rate": 9.80832037325039e-06, "loss": 0.1916, "step": 494 }, { "epoch": 0.02, "learning_rate": 9.807931570762053e-06, "loss": 0.3128, "step": 495 }, { "epoch": 0.02, "learning_rate": 9.807542768273719e-06, "loss": 0.273, "step": 496 }, { "epoch": 0.02, "learning_rate": 9.807153965785382e-06, "loss": 0.4427, "step": 497 }, { "epoch": 0.02, "learning_rate": 9.806765163297046e-06, "loss": 0.2708, "step": 498 }, { "epoch": 0.02, "learning_rate": 9.806376360808711e-06, "loss": 0.3988, "step": 499 }, { "epoch": 0.02, "learning_rate": 9.805987558320375e-06, "loss": 0.3634, "step": 500 }, { "epoch": 0.02, "learning_rate": 9.805598755832038e-06, "loss": 0.1661, "step": 501 }, { "epoch": 0.02, "learning_rate": 9.805209953343702e-06, "loss": 0.2886, "step": 502 }, { "epoch": 0.02, "learning_rate": 9.804821150855366e-06, "loss": 0.2934, "step": 503 }, { "epoch": 0.02, "learning_rate": 9.80443234836703e-06, "loss": 0.3531, "step": 504 }, { "epoch": 0.02, "learning_rate": 9.804043545878695e-06, "loss": 0.0836, "step": 505 }, { "epoch": 0.02, "learning_rate": 9.803654743390358e-06, "loss": 0.241, "step": 506 }, { "epoch": 0.02, "learning_rate": 9.803265940902022e-06, "loss": 0.5959, "step": 507 }, { "epoch": 0.02, "learning_rate": 9.802877138413687e-06, "loss": 0.2467, "step": 508 }, { "epoch": 0.02, "learning_rate": 9.80248833592535e-06, "loss": 0.5243, "step": 509 }, { "epoch": 0.02, "learning_rate": 9.802099533437014e-06, "loss": 0.2222, "step": 510 }, { "epoch": 0.02, "learning_rate": 9.80171073094868e-06, "loss": 0.4864, "step": 511 }, { "epoch": 0.02, "learning_rate": 9.801321928460343e-06, "loss": 0.4821, "step": 512 }, { "epoch": 0.02, "learning_rate": 9.800933125972007e-06, "loss": 0.3492, "step": 513 }, { "epoch": 0.02, "learning_rate": 9.80054432348367e-06, "loss": 0.3339, "step": 514 }, { "epoch": 0.02, "learning_rate": 9.800155520995334e-06, "loss": 0.3337, "step": 515 }, { "epoch": 0.02, "learning_rate": 9.799766718507e-06, "loss": 0.4931, "step": 516 }, { "epoch": 0.02, "learning_rate": 9.799377916018663e-06, "loss": 0.3333, "step": 517 }, { "epoch": 0.02, "learning_rate": 9.798989113530327e-06, "loss": 0.1827, "step": 518 }, { "epoch": 0.02, "learning_rate": 9.798600311041992e-06, "loss": 0.2077, "step": 519 }, { "epoch": 0.02, "learning_rate": 9.798211508553656e-06, "loss": 0.2294, "step": 520 }, { "epoch": 0.02, "learning_rate": 9.79782270606532e-06, "loss": 0.6992, "step": 521 }, { "epoch": 0.02, "learning_rate": 9.797433903576985e-06, "loss": 0.292, "step": 522 }, { "epoch": 0.02, "learning_rate": 9.797045101088648e-06, "loss": 0.2363, "step": 523 }, { "epoch": 0.02, "learning_rate": 9.796656298600312e-06, "loss": 0.1914, "step": 524 }, { "epoch": 0.02, "learning_rate": 9.796267496111976e-06, "loss": 0.3069, "step": 525 }, { "epoch": 0.02, "learning_rate": 9.795878693623641e-06, "loss": 0.3627, "step": 526 }, { "epoch": 0.02, "learning_rate": 9.795489891135303e-06, "loss": 0.4102, "step": 527 }, { "epoch": 0.02, "learning_rate": 9.795101088646968e-06, "loss": 0.2108, "step": 528 }, { "epoch": 0.02, "learning_rate": 9.794712286158632e-06, "loss": 0.2214, "step": 529 }, { "epoch": 0.02, "learning_rate": 9.794323483670295e-06, "loss": 0.3867, "step": 530 }, { "epoch": 0.02, "learning_rate": 9.79393468118196e-06, "loss": 0.2888, "step": 531 }, { "epoch": 0.02, "learning_rate": 9.793545878693624e-06, "loss": 0.2237, "step": 532 }, { "epoch": 0.02, "learning_rate": 9.793157076205288e-06, "loss": 0.1608, "step": 533 }, { "epoch": 0.02, "learning_rate": 9.792768273716953e-06, "loss": 0.2943, "step": 534 }, { "epoch": 0.02, "learning_rate": 9.792379471228617e-06, "loss": 0.2974, "step": 535 }, { "epoch": 0.02, "learning_rate": 9.79199066874028e-06, "loss": 0.5654, "step": 536 }, { "epoch": 0.02, "learning_rate": 9.791601866251946e-06, "loss": 0.606, "step": 537 }, { "epoch": 0.02, "learning_rate": 9.79121306376361e-06, "loss": 0.8816, "step": 538 }, { "epoch": 0.02, "learning_rate": 9.790824261275273e-06, "loss": 0.3602, "step": 539 }, { "epoch": 0.02, "learning_rate": 9.790435458786937e-06, "loss": 0.1224, "step": 540 }, { "epoch": 0.02, "learning_rate": 9.7900466562986e-06, "loss": 0.5199, "step": 541 }, { "epoch": 0.02, "learning_rate": 9.789657853810266e-06, "loss": 0.4114, "step": 542 }, { "epoch": 0.02, "learning_rate": 9.78926905132193e-06, "loss": 0.2521, "step": 543 }, { "epoch": 0.02, "learning_rate": 9.788880248833593e-06, "loss": 0.2931, "step": 544 }, { "epoch": 0.02, "learning_rate": 9.788491446345257e-06, "loss": 0.2026, "step": 545 }, { "epoch": 0.02, "learning_rate": 9.788102643856922e-06, "loss": 0.3164, "step": 546 }, { "epoch": 0.02, "learning_rate": 9.787713841368586e-06, "loss": 0.4304, "step": 547 }, { "epoch": 0.02, "learning_rate": 9.78732503888025e-06, "loss": 0.1011, "step": 548 }, { "epoch": 0.02, "learning_rate": 9.786936236391915e-06, "loss": 0.4743, "step": 549 }, { "epoch": 0.02, "learning_rate": 9.786547433903578e-06, "loss": 0.4145, "step": 550 }, { "epoch": 0.02, "learning_rate": 9.786158631415242e-06, "loss": 0.3914, "step": 551 }, { "epoch": 0.02, "learning_rate": 9.785769828926905e-06, "loss": 0.1508, "step": 552 }, { "epoch": 0.02, "learning_rate": 9.785381026438569e-06, "loss": 0.0859, "step": 553 }, { "epoch": 0.02, "learning_rate": 9.784992223950234e-06, "loss": 0.3515, "step": 554 }, { "epoch": 0.02, "learning_rate": 9.784603421461898e-06, "loss": 0.22, "step": 555 }, { "epoch": 0.02, "learning_rate": 9.784214618973562e-06, "loss": 0.4301, "step": 556 }, { "epoch": 0.02, "learning_rate": 9.783825816485227e-06, "loss": 0.3754, "step": 557 }, { "epoch": 0.02, "learning_rate": 9.78343701399689e-06, "loss": 0.2857, "step": 558 }, { "epoch": 0.02, "learning_rate": 9.783048211508554e-06, "loss": 0.2854, "step": 559 }, { "epoch": 0.02, "learning_rate": 9.78265940902022e-06, "loss": 0.1261, "step": 560 }, { "epoch": 0.02, "learning_rate": 9.782270606531883e-06, "loss": 0.2841, "step": 561 }, { "epoch": 0.02, "learning_rate": 9.781881804043547e-06, "loss": 0.5208, "step": 562 }, { "epoch": 0.02, "learning_rate": 9.78149300155521e-06, "loss": 0.1844, "step": 563 }, { "epoch": 0.02, "learning_rate": 9.781104199066876e-06, "loss": 0.2707, "step": 564 }, { "epoch": 0.02, "learning_rate": 9.780715396578538e-06, "loss": 0.384, "step": 565 }, { "epoch": 0.02, "learning_rate": 9.780326594090203e-06, "loss": 0.1589, "step": 566 }, { "epoch": 0.02, "learning_rate": 9.779937791601867e-06, "loss": 0.3899, "step": 567 }, { "epoch": 0.02, "learning_rate": 9.77954898911353e-06, "loss": 0.3033, "step": 568 }, { "epoch": 0.02, "learning_rate": 9.779160186625196e-06, "loss": 0.3751, "step": 569 }, { "epoch": 0.02, "learning_rate": 9.77877138413686e-06, "loss": 0.4019, "step": 570 }, { "epoch": 0.02, "learning_rate": 9.778382581648523e-06, "loss": 0.1689, "step": 571 }, { "epoch": 0.02, "learning_rate": 9.777993779160188e-06, "loss": 0.2733, "step": 572 }, { "epoch": 0.02, "learning_rate": 9.777604976671852e-06, "loss": 0.4909, "step": 573 }, { "epoch": 0.02, "learning_rate": 9.777216174183515e-06, "loss": 0.1028, "step": 574 }, { "epoch": 0.02, "learning_rate": 9.77682737169518e-06, "loss": 0.2735, "step": 575 }, { "epoch": 0.02, "learning_rate": 9.776438569206844e-06, "loss": 0.1836, "step": 576 }, { "epoch": 0.02, "learning_rate": 9.776049766718508e-06, "loss": 0.1237, "step": 577 }, { "epoch": 0.02, "learning_rate": 9.775660964230172e-06, "loss": 0.2771, "step": 578 }, { "epoch": 0.02, "learning_rate": 9.775272161741835e-06, "loss": 0.5872, "step": 579 }, { "epoch": 0.02, "learning_rate": 9.7748833592535e-06, "loss": 0.1442, "step": 580 }, { "epoch": 0.02, "learning_rate": 9.774494556765164e-06, "loss": 0.4006, "step": 581 }, { "epoch": 0.02, "learning_rate": 9.774105754276828e-06, "loss": 0.224, "step": 582 }, { "epoch": 0.02, "learning_rate": 9.773716951788491e-06, "loss": 0.1664, "step": 583 }, { "epoch": 0.02, "learning_rate": 9.773328149300157e-06, "loss": 0.0923, "step": 584 }, { "epoch": 0.02, "learning_rate": 9.77293934681182e-06, "loss": 0.2743, "step": 585 }, { "epoch": 0.02, "learning_rate": 9.772550544323484e-06, "loss": 0.3856, "step": 586 }, { "epoch": 0.02, "learning_rate": 9.77216174183515e-06, "loss": 0.2925, "step": 587 }, { "epoch": 0.02, "learning_rate": 9.771772939346813e-06, "loss": 0.1784, "step": 588 }, { "epoch": 0.02, "learning_rate": 9.771384136858477e-06, "loss": 0.0864, "step": 589 }, { "epoch": 0.02, "learning_rate": 9.77099533437014e-06, "loss": 0.8613, "step": 590 }, { "epoch": 0.02, "learning_rate": 9.770606531881804e-06, "loss": 0.2661, "step": 591 }, { "epoch": 0.02, "learning_rate": 9.770217729393469e-06, "loss": 0.602, "step": 592 }, { "epoch": 0.02, "learning_rate": 9.769828926905133e-06, "loss": 0.3537, "step": 593 }, { "epoch": 0.02, "learning_rate": 9.769440124416796e-06, "loss": 0.2646, "step": 594 }, { "epoch": 0.02, "learning_rate": 9.769051321928462e-06, "loss": 0.4871, "step": 595 }, { "epoch": 0.02, "learning_rate": 9.768662519440125e-06, "loss": 0.2425, "step": 596 }, { "epoch": 0.02, "learning_rate": 9.768273716951789e-06, "loss": 0.1259, "step": 597 }, { "epoch": 0.02, "learning_rate": 9.767884914463454e-06, "loss": 0.5255, "step": 598 }, { "epoch": 0.02, "learning_rate": 9.767496111975118e-06, "loss": 0.6901, "step": 599 }, { "epoch": 0.02, "learning_rate": 9.767107309486782e-06, "loss": 0.4286, "step": 600 }, { "epoch": 0.02, "learning_rate": 9.766718506998445e-06, "loss": 0.1908, "step": 601 }, { "epoch": 0.02, "learning_rate": 9.76632970451011e-06, "loss": 0.6688, "step": 602 }, { "epoch": 0.02, "learning_rate": 9.765940902021772e-06, "loss": 0.2619, "step": 603 }, { "epoch": 0.02, "learning_rate": 9.765552099533438e-06, "loss": 0.3867, "step": 604 }, { "epoch": 0.02, "learning_rate": 9.765163297045101e-06, "loss": 0.179, "step": 605 }, { "epoch": 0.02, "learning_rate": 9.764774494556765e-06, "loss": 0.3225, "step": 606 }, { "epoch": 0.02, "learning_rate": 9.76438569206843e-06, "loss": 0.1314, "step": 607 }, { "epoch": 0.02, "learning_rate": 9.763996889580094e-06, "loss": 0.3632, "step": 608 }, { "epoch": 0.02, "learning_rate": 9.763608087091758e-06, "loss": 0.4708, "step": 609 }, { "epoch": 0.02, "learning_rate": 9.763219284603423e-06, "loss": 0.4432, "step": 610 }, { "epoch": 0.02, "learning_rate": 9.762830482115087e-06, "loss": 0.3133, "step": 611 }, { "epoch": 0.02, "learning_rate": 9.76244167962675e-06, "loss": 0.3202, "step": 612 }, { "epoch": 0.02, "learning_rate": 9.762052877138416e-06, "loss": 0.3245, "step": 613 }, { "epoch": 0.02, "learning_rate": 9.761664074650079e-06, "loss": 0.3342, "step": 614 }, { "epoch": 0.02, "learning_rate": 9.761275272161743e-06, "loss": 0.1093, "step": 615 }, { "epoch": 0.02, "learning_rate": 9.760886469673406e-06, "loss": 0.6117, "step": 616 }, { "epoch": 0.02, "learning_rate": 9.76049766718507e-06, "loss": 0.4946, "step": 617 }, { "epoch": 0.02, "learning_rate": 9.760108864696735e-06, "loss": 0.215, "step": 618 }, { "epoch": 0.02, "learning_rate": 9.759720062208399e-06, "loss": 0.3151, "step": 619 }, { "epoch": 0.02, "learning_rate": 9.759331259720063e-06, "loss": 0.2215, "step": 620 }, { "epoch": 0.02, "learning_rate": 9.758942457231726e-06, "loss": 0.4238, "step": 621 }, { "epoch": 0.02, "learning_rate": 9.758553654743392e-06, "loss": 0.221, "step": 622 }, { "epoch": 0.02, "learning_rate": 9.758164852255055e-06, "loss": 0.4266, "step": 623 }, { "epoch": 0.02, "learning_rate": 9.757776049766719e-06, "loss": 0.4506, "step": 624 }, { "epoch": 0.02, "learning_rate": 9.757387247278384e-06, "loss": 0.3659, "step": 625 }, { "epoch": 0.02, "learning_rate": 9.756998444790048e-06, "loss": 0.1569, "step": 626 }, { "epoch": 0.02, "learning_rate": 9.756609642301711e-06, "loss": 0.5016, "step": 627 }, { "epoch": 0.02, "learning_rate": 9.756220839813375e-06, "loss": 0.1902, "step": 628 }, { "epoch": 0.02, "learning_rate": 9.755832037325039e-06, "loss": 0.3151, "step": 629 }, { "epoch": 0.02, "learning_rate": 9.755443234836704e-06, "loss": 0.2462, "step": 630 }, { "epoch": 0.02, "learning_rate": 9.755054432348368e-06, "loss": 0.1827, "step": 631 }, { "epoch": 0.02, "learning_rate": 9.754665629860031e-06, "loss": 0.3611, "step": 632 }, { "epoch": 0.02, "learning_rate": 9.754276827371697e-06, "loss": 0.3638, "step": 633 }, { "epoch": 0.02, "learning_rate": 9.75388802488336e-06, "loss": 0.2696, "step": 634 }, { "epoch": 0.02, "learning_rate": 9.753499222395024e-06, "loss": 0.3171, "step": 635 }, { "epoch": 0.02, "learning_rate": 9.753110419906689e-06, "loss": 0.2259, "step": 636 }, { "epoch": 0.02, "learning_rate": 9.752721617418353e-06, "loss": 0.1935, "step": 637 }, { "epoch": 0.02, "learning_rate": 9.752332814930016e-06, "loss": 0.2019, "step": 638 }, { "epoch": 0.02, "learning_rate": 9.751944012441682e-06, "loss": 0.4767, "step": 639 }, { "epoch": 0.02, "learning_rate": 9.751555209953345e-06, "loss": 0.2194, "step": 640 }, { "epoch": 0.02, "learning_rate": 9.751166407465007e-06, "loss": 0.1995, "step": 641 }, { "epoch": 0.02, "learning_rate": 9.750777604976673e-06, "loss": 0.191, "step": 642 }, { "epoch": 0.03, "learning_rate": 9.750388802488336e-06, "loss": 0.3502, "step": 643 }, { "epoch": 0.03, "learning_rate": 9.75e-06, "loss": 0.2512, "step": 644 }, { "epoch": 0.03, "learning_rate": 9.749611197511665e-06, "loss": 0.1942, "step": 645 }, { "epoch": 0.03, "learning_rate": 9.749222395023329e-06, "loss": 0.2508, "step": 646 }, { "epoch": 0.03, "learning_rate": 9.748833592534992e-06, "loss": 0.581, "step": 647 }, { "epoch": 0.03, "learning_rate": 9.748444790046658e-06, "loss": 0.1725, "step": 648 }, { "epoch": 0.03, "learning_rate": 9.748055987558321e-06, "loss": 0.2266, "step": 649 }, { "epoch": 0.03, "learning_rate": 9.747667185069985e-06, "loss": 0.3914, "step": 650 }, { "epoch": 0.03, "learning_rate": 9.74727838258165e-06, "loss": 0.2556, "step": 651 }, { "epoch": 0.03, "learning_rate": 9.746889580093314e-06, "loss": 0.3661, "step": 652 }, { "epoch": 0.03, "learning_rate": 9.746500777604978e-06, "loss": 0.359, "step": 653 }, { "epoch": 0.03, "learning_rate": 9.746111975116641e-06, "loss": 0.1122, "step": 654 }, { "epoch": 0.03, "learning_rate": 9.745723172628305e-06, "loss": 0.1793, "step": 655 }, { "epoch": 0.03, "learning_rate": 9.74533437013997e-06, "loss": 0.2673, "step": 656 }, { "epoch": 0.03, "learning_rate": 9.744945567651634e-06, "loss": 0.3434, "step": 657 }, { "epoch": 0.03, "learning_rate": 9.744556765163297e-06, "loss": 0.3515, "step": 658 }, { "epoch": 0.03, "learning_rate": 9.744167962674963e-06, "loss": 0.2324, "step": 659 }, { "epoch": 0.03, "learning_rate": 9.743779160186626e-06, "loss": 0.2545, "step": 660 }, { "epoch": 0.03, "learning_rate": 9.74339035769829e-06, "loss": 0.2109, "step": 661 }, { "epoch": 0.03, "learning_rate": 9.743001555209954e-06, "loss": 0.2935, "step": 662 }, { "epoch": 0.03, "learning_rate": 9.742612752721619e-06, "loss": 0.2756, "step": 663 }, { "epoch": 0.03, "learning_rate": 9.742223950233283e-06, "loss": 0.5134, "step": 664 }, { "epoch": 0.03, "learning_rate": 9.741835147744946e-06, "loss": 0.3644, "step": 665 }, { "epoch": 0.03, "learning_rate": 9.74144634525661e-06, "loss": 0.2963, "step": 666 }, { "epoch": 0.03, "learning_rate": 9.741057542768273e-06, "loss": 0.3968, "step": 667 }, { "epoch": 0.03, "learning_rate": 9.740668740279939e-06, "loss": 0.43, "step": 668 }, { "epoch": 0.03, "learning_rate": 9.740279937791602e-06, "loss": 0.1654, "step": 669 }, { "epoch": 0.03, "learning_rate": 9.739891135303266e-06, "loss": 0.3972, "step": 670 }, { "epoch": 0.03, "learning_rate": 9.739502332814931e-06, "loss": 0.4194, "step": 671 }, { "epoch": 0.03, "learning_rate": 9.739113530326595e-06, "loss": 0.4357, "step": 672 }, { "epoch": 0.03, "learning_rate": 9.738724727838259e-06, "loss": 0.481, "step": 673 }, { "epoch": 0.03, "learning_rate": 9.738335925349924e-06, "loss": 0.3384, "step": 674 }, { "epoch": 0.03, "learning_rate": 9.737947122861588e-06, "loss": 0.2639, "step": 675 }, { "epoch": 0.03, "learning_rate": 9.737558320373251e-06, "loss": 0.2649, "step": 676 }, { "epoch": 0.03, "learning_rate": 9.737169517884916e-06, "loss": 0.1043, "step": 677 }, { "epoch": 0.03, "learning_rate": 9.73678071539658e-06, "loss": 0.271, "step": 678 }, { "epoch": 0.03, "learning_rate": 9.736391912908244e-06, "loss": 0.289, "step": 679 }, { "epoch": 0.03, "learning_rate": 9.736003110419907e-06, "loss": 0.3727, "step": 680 }, { "epoch": 0.03, "learning_rate": 9.735614307931571e-06, "loss": 0.1608, "step": 681 }, { "epoch": 0.03, "learning_rate": 9.735225505443235e-06, "loss": 0.333, "step": 682 }, { "epoch": 0.03, "learning_rate": 9.7348367029549e-06, "loss": 0.4925, "step": 683 }, { "epoch": 0.03, "learning_rate": 9.734447900466564e-06, "loss": 0.3004, "step": 684 }, { "epoch": 0.03, "learning_rate": 9.734059097978227e-06, "loss": 0.3725, "step": 685 }, { "epoch": 0.03, "learning_rate": 9.733670295489893e-06, "loss": 0.4376, "step": 686 }, { "epoch": 0.03, "learning_rate": 9.733281493001556e-06, "loss": 0.1285, "step": 687 }, { "epoch": 0.03, "learning_rate": 9.73289269051322e-06, "loss": 0.3158, "step": 688 }, { "epoch": 0.03, "learning_rate": 9.732503888024885e-06, "loss": 0.1914, "step": 689 }, { "epoch": 0.03, "learning_rate": 9.732115085536549e-06, "loss": 0.176, "step": 690 }, { "epoch": 0.03, "learning_rate": 9.731726283048212e-06, "loss": 0.4963, "step": 691 }, { "epoch": 0.03, "learning_rate": 9.731337480559876e-06, "loss": 0.1348, "step": 692 }, { "epoch": 0.03, "learning_rate": 9.73094867807154e-06, "loss": 0.4394, "step": 693 }, { "epoch": 0.03, "learning_rate": 9.730559875583205e-06, "loss": 0.2421, "step": 694 }, { "epoch": 0.03, "learning_rate": 9.730171073094869e-06, "loss": 0.6994, "step": 695 }, { "epoch": 0.03, "learning_rate": 9.729782270606532e-06, "loss": 0.1768, "step": 696 }, { "epoch": 0.03, "learning_rate": 9.729393468118198e-06, "loss": 0.4496, "step": 697 }, { "epoch": 0.03, "learning_rate": 9.729004665629861e-06, "loss": 0.3153, "step": 698 }, { "epoch": 0.03, "learning_rate": 9.728615863141525e-06, "loss": 0.4059, "step": 699 }, { "epoch": 0.03, "learning_rate": 9.728227060653188e-06, "loss": 0.3255, "step": 700 }, { "epoch": 0.03, "learning_rate": 9.727838258164854e-06, "loss": 0.1741, "step": 701 }, { "epoch": 0.03, "learning_rate": 9.727449455676517e-06, "loss": 0.3983, "step": 702 }, { "epoch": 0.03, "learning_rate": 9.727060653188181e-06, "loss": 0.1888, "step": 703 }, { "epoch": 0.03, "learning_rate": 9.726671850699845e-06, "loss": 0.1855, "step": 704 }, { "epoch": 0.03, "learning_rate": 9.726283048211508e-06, "loss": 0.1492, "step": 705 }, { "epoch": 0.03, "learning_rate": 9.725894245723174e-06, "loss": 0.1796, "step": 706 }, { "epoch": 0.03, "learning_rate": 9.725505443234837e-06, "loss": 0.3521, "step": 707 }, { "epoch": 0.03, "learning_rate": 9.7251166407465e-06, "loss": 0.1364, "step": 708 }, { "epoch": 0.03, "learning_rate": 9.724727838258166e-06, "loss": 0.2321, "step": 709 }, { "epoch": 0.03, "learning_rate": 9.72433903576983e-06, "loss": 0.2843, "step": 710 }, { "epoch": 0.03, "learning_rate": 9.723950233281493e-06, "loss": 0.1088, "step": 711 }, { "epoch": 0.03, "learning_rate": 9.723561430793159e-06, "loss": 0.3013, "step": 712 }, { "epoch": 0.03, "learning_rate": 9.723172628304822e-06, "loss": 0.1026, "step": 713 }, { "epoch": 0.03, "learning_rate": 9.722783825816486e-06, "loss": 0.5426, "step": 714 }, { "epoch": 0.03, "learning_rate": 9.722395023328151e-06, "loss": 0.2009, "step": 715 }, { "epoch": 0.03, "learning_rate": 9.722006220839815e-06, "loss": 0.2278, "step": 716 }, { "epoch": 0.03, "learning_rate": 9.721617418351479e-06, "loss": 0.2442, "step": 717 }, { "epoch": 0.03, "learning_rate": 9.721228615863142e-06, "loss": 0.3392, "step": 718 }, { "epoch": 0.03, "learning_rate": 9.720839813374806e-06, "loss": 0.269, "step": 719 }, { "epoch": 0.03, "learning_rate": 9.72045101088647e-06, "loss": 0.3609, "step": 720 }, { "epoch": 0.03, "learning_rate": 9.720062208398135e-06, "loss": 0.3676, "step": 721 }, { "epoch": 0.03, "learning_rate": 9.719673405909798e-06, "loss": 0.3369, "step": 722 }, { "epoch": 0.03, "learning_rate": 9.719284603421462e-06, "loss": 0.2173, "step": 723 }, { "epoch": 0.03, "learning_rate": 9.718895800933127e-06, "loss": 0.3089, "step": 724 }, { "epoch": 0.03, "learning_rate": 9.718506998444791e-06, "loss": 0.1579, "step": 725 }, { "epoch": 0.03, "learning_rate": 9.718118195956455e-06, "loss": 0.4787, "step": 726 }, { "epoch": 0.03, "learning_rate": 9.71772939346812e-06, "loss": 0.3738, "step": 727 }, { "epoch": 0.03, "learning_rate": 9.717340590979784e-06, "loss": 0.227, "step": 728 }, { "epoch": 0.03, "learning_rate": 9.716951788491447e-06, "loss": 0.339, "step": 729 }, { "epoch": 0.03, "learning_rate": 9.71656298600311e-06, "loss": 0.1907, "step": 730 }, { "epoch": 0.03, "learning_rate": 9.716174183514774e-06, "loss": 0.2931, "step": 731 }, { "epoch": 0.03, "learning_rate": 9.71578538102644e-06, "loss": 0.2839, "step": 732 }, { "epoch": 0.03, "learning_rate": 9.715396578538103e-06, "loss": 0.2642, "step": 733 }, { "epoch": 0.03, "learning_rate": 9.715007776049767e-06, "loss": 0.286, "step": 734 }, { "epoch": 0.03, "learning_rate": 9.714618973561432e-06, "loss": 0.4578, "step": 735 }, { "epoch": 0.03, "learning_rate": 9.714230171073096e-06, "loss": 0.2426, "step": 736 }, { "epoch": 0.03, "learning_rate": 9.71384136858476e-06, "loss": 0.0828, "step": 737 }, { "epoch": 0.03, "learning_rate": 9.713452566096425e-06, "loss": 0.1925, "step": 738 }, { "epoch": 0.03, "learning_rate": 9.713063763608089e-06, "loss": 0.2377, "step": 739 }, { "epoch": 0.03, "learning_rate": 9.712674961119752e-06, "loss": 0.2843, "step": 740 }, { "epoch": 0.03, "learning_rate": 9.712286158631416e-06, "loss": 0.4177, "step": 741 }, { "epoch": 0.03, "learning_rate": 9.71189735614308e-06, "loss": 0.4184, "step": 742 }, { "epoch": 0.03, "learning_rate": 9.711508553654743e-06, "loss": 0.2539, "step": 743 }, { "epoch": 0.03, "learning_rate": 9.711119751166408e-06, "loss": 0.2982, "step": 744 }, { "epoch": 0.03, "learning_rate": 9.710730948678072e-06, "loss": 0.2884, "step": 745 }, { "epoch": 0.03, "learning_rate": 9.710342146189736e-06, "loss": 0.3609, "step": 746 }, { "epoch": 0.03, "learning_rate": 9.709953343701401e-06, "loss": 0.0973, "step": 747 }, { "epoch": 0.03, "learning_rate": 9.709564541213065e-06, "loss": 0.2413, "step": 748 }, { "epoch": 0.03, "learning_rate": 9.709175738724728e-06, "loss": 0.5373, "step": 749 }, { "epoch": 0.03, "learning_rate": 9.708786936236394e-06, "loss": 0.1344, "step": 750 }, { "epoch": 0.03, "learning_rate": 9.708398133748057e-06, "loss": 0.2647, "step": 751 }, { "epoch": 0.03, "learning_rate": 9.70800933125972e-06, "loss": 0.3368, "step": 752 }, { "epoch": 0.03, "learning_rate": 9.707620528771386e-06, "loss": 0.2527, "step": 753 }, { "epoch": 0.03, "learning_rate": 9.70723172628305e-06, "loss": 0.266, "step": 754 }, { "epoch": 0.03, "learning_rate": 9.706842923794713e-06, "loss": 0.5481, "step": 755 }, { "epoch": 0.03, "learning_rate": 9.706454121306377e-06, "loss": 0.1158, "step": 756 }, { "epoch": 0.03, "learning_rate": 9.70606531881804e-06, "loss": 0.1793, "step": 757 }, { "epoch": 0.03, "learning_rate": 9.705676516329706e-06, "loss": 0.1973, "step": 758 }, { "epoch": 0.03, "learning_rate": 9.70528771384137e-06, "loss": 0.316, "step": 759 }, { "epoch": 0.03, "learning_rate": 9.704898911353033e-06, "loss": 0.1433, "step": 760 }, { "epoch": 0.03, "learning_rate": 9.704510108864697e-06, "loss": 0.2546, "step": 761 }, { "epoch": 0.03, "learning_rate": 9.704121306376362e-06, "loss": 0.454, "step": 762 }, { "epoch": 0.03, "learning_rate": 9.703732503888026e-06, "loss": 0.5689, "step": 763 }, { "epoch": 0.03, "learning_rate": 9.70334370139969e-06, "loss": 0.3749, "step": 764 }, { "epoch": 0.03, "learning_rate": 9.702954898911355e-06, "loss": 0.4607, "step": 765 }, { "epoch": 0.03, "learning_rate": 9.702566096423018e-06, "loss": 0.3611, "step": 766 }, { "epoch": 0.03, "learning_rate": 9.702177293934682e-06, "loss": 0.4063, "step": 767 }, { "epoch": 0.03, "learning_rate": 9.701788491446346e-06, "loss": 0.283, "step": 768 }, { "epoch": 0.03, "learning_rate": 9.70139968895801e-06, "loss": 0.5277, "step": 769 }, { "epoch": 0.03, "learning_rate": 9.701010886469675e-06, "loss": 0.3291, "step": 770 }, { "epoch": 0.03, "learning_rate": 9.700622083981338e-06, "loss": 0.2658, "step": 771 }, { "epoch": 0.03, "learning_rate": 9.700233281493002e-06, "loss": 0.2162, "step": 772 }, { "epoch": 0.03, "learning_rate": 9.699844479004667e-06, "loss": 0.114, "step": 773 }, { "epoch": 0.03, "learning_rate": 9.69945567651633e-06, "loss": 0.286, "step": 774 }, { "epoch": 0.03, "learning_rate": 9.699066874027994e-06, "loss": 0.3039, "step": 775 }, { "epoch": 0.03, "learning_rate": 9.69867807153966e-06, "loss": 0.3146, "step": 776 }, { "epoch": 0.03, "learning_rate": 9.698289269051323e-06, "loss": 0.1943, "step": 777 }, { "epoch": 0.03, "learning_rate": 9.697900466562987e-06, "loss": 0.1569, "step": 778 }, { "epoch": 0.03, "learning_rate": 9.69751166407465e-06, "loss": 0.3697, "step": 779 }, { "epoch": 0.03, "learning_rate": 9.697122861586314e-06, "loss": 0.4015, "step": 780 }, { "epoch": 0.03, "learning_rate": 9.696734059097978e-06, "loss": 0.1424, "step": 781 }, { "epoch": 0.03, "learning_rate": 9.696345256609643e-06, "loss": 0.3123, "step": 782 }, { "epoch": 0.03, "learning_rate": 9.695956454121307e-06, "loss": 0.658, "step": 783 }, { "epoch": 0.03, "learning_rate": 9.69556765163297e-06, "loss": 0.3752, "step": 784 }, { "epoch": 0.03, "learning_rate": 9.695178849144636e-06, "loss": 0.1294, "step": 785 }, { "epoch": 0.03, "learning_rate": 9.6947900466563e-06, "loss": 0.3316, "step": 786 }, { "epoch": 0.03, "learning_rate": 9.694401244167963e-06, "loss": 0.3678, "step": 787 }, { "epoch": 0.03, "learning_rate": 9.694012441679628e-06, "loss": 0.3412, "step": 788 }, { "epoch": 0.03, "learning_rate": 9.693623639191292e-06, "loss": 0.2961, "step": 789 }, { "epoch": 0.03, "learning_rate": 9.693234836702956e-06, "loss": 0.2339, "step": 790 }, { "epoch": 0.03, "learning_rate": 9.692846034214621e-06, "loss": 0.4538, "step": 791 }, { "epoch": 0.03, "learning_rate": 9.692457231726285e-06, "loss": 0.3244, "step": 792 }, { "epoch": 0.03, "learning_rate": 9.692068429237948e-06, "loss": 0.4797, "step": 793 }, { "epoch": 0.03, "learning_rate": 9.691679626749612e-06, "loss": 0.1829, "step": 794 }, { "epoch": 0.03, "learning_rate": 9.691290824261275e-06, "loss": 0.3841, "step": 795 }, { "epoch": 0.03, "learning_rate": 9.69090202177294e-06, "loss": 0.1142, "step": 796 }, { "epoch": 0.03, "learning_rate": 9.690513219284604e-06, "loss": 0.5708, "step": 797 }, { "epoch": 0.03, "learning_rate": 9.690124416796268e-06, "loss": 0.4263, "step": 798 }, { "epoch": 0.03, "learning_rate": 9.689735614307932e-06, "loss": 0.3741, "step": 799 }, { "epoch": 0.03, "learning_rate": 9.689346811819597e-06, "loss": 0.2563, "step": 800 }, { "epoch": 0.03, "learning_rate": 9.68895800933126e-06, "loss": 0.4044, "step": 801 }, { "epoch": 0.03, "learning_rate": 9.688569206842924e-06, "loss": 0.2048, "step": 802 }, { "epoch": 0.03, "learning_rate": 9.68818040435459e-06, "loss": 0.5188, "step": 803 }, { "epoch": 0.03, "learning_rate": 9.687791601866253e-06, "loss": 0.1584, "step": 804 }, { "epoch": 0.03, "learning_rate": 9.687402799377917e-06, "loss": 0.5148, "step": 805 }, { "epoch": 0.03, "learning_rate": 9.68701399688958e-06, "loss": 0.0808, "step": 806 }, { "epoch": 0.03, "learning_rate": 9.686625194401244e-06, "loss": 0.2855, "step": 807 }, { "epoch": 0.03, "learning_rate": 9.68623639191291e-06, "loss": 0.257, "step": 808 }, { "epoch": 0.03, "learning_rate": 9.685847589424573e-06, "loss": 0.2373, "step": 809 }, { "epoch": 0.03, "learning_rate": 9.685458786936237e-06, "loss": 0.2443, "step": 810 }, { "epoch": 0.03, "learning_rate": 9.685069984447902e-06, "loss": 0.1325, "step": 811 }, { "epoch": 0.03, "learning_rate": 9.684681181959566e-06, "loss": 0.1397, "step": 812 }, { "epoch": 0.03, "learning_rate": 9.68429237947123e-06, "loss": 0.268, "step": 813 }, { "epoch": 0.03, "learning_rate": 9.683903576982894e-06, "loss": 0.2965, "step": 814 }, { "epoch": 0.03, "learning_rate": 9.683514774494558e-06, "loss": 0.2282, "step": 815 }, { "epoch": 0.03, "learning_rate": 9.683125972006222e-06, "loss": 0.4088, "step": 816 }, { "epoch": 0.03, "learning_rate": 9.682737169517887e-06, "loss": 0.2473, "step": 817 }, { "epoch": 0.03, "learning_rate": 9.682348367029549e-06, "loss": 0.2317, "step": 818 }, { "epoch": 0.03, "learning_rate": 9.681959564541213e-06, "loss": 0.331, "step": 819 }, { "epoch": 0.03, "learning_rate": 9.681570762052878e-06, "loss": 0.1415, "step": 820 }, { "epoch": 0.03, "learning_rate": 9.681181959564542e-06, "loss": 0.2774, "step": 821 }, { "epoch": 0.03, "learning_rate": 9.680793157076205e-06, "loss": 0.0921, "step": 822 }, { "epoch": 0.03, "learning_rate": 9.68040435458787e-06, "loss": 0.2875, "step": 823 }, { "epoch": 0.03, "learning_rate": 9.680015552099534e-06, "loss": 0.3964, "step": 824 }, { "epoch": 0.03, "learning_rate": 9.679626749611198e-06, "loss": 0.1238, "step": 825 }, { "epoch": 0.03, "learning_rate": 9.679237947122863e-06, "loss": 0.1146, "step": 826 }, { "epoch": 0.03, "learning_rate": 9.678849144634527e-06, "loss": 0.1646, "step": 827 }, { "epoch": 0.03, "learning_rate": 9.67846034214619e-06, "loss": 0.5589, "step": 828 }, { "epoch": 0.03, "learning_rate": 9.678071539657856e-06, "loss": 0.3212, "step": 829 }, { "epoch": 0.03, "learning_rate": 9.67768273716952e-06, "loss": 0.2594, "step": 830 }, { "epoch": 0.03, "learning_rate": 9.677293934681183e-06, "loss": 0.2898, "step": 831 }, { "epoch": 0.03, "learning_rate": 9.676905132192847e-06, "loss": 0.5834, "step": 832 }, { "epoch": 0.03, "learning_rate": 9.67651632970451e-06, "loss": 0.4943, "step": 833 }, { "epoch": 0.03, "learning_rate": 9.676127527216176e-06, "loss": 0.4246, "step": 834 }, { "epoch": 0.03, "learning_rate": 9.675738724727839e-06, "loss": 0.2321, "step": 835 }, { "epoch": 0.03, "learning_rate": 9.675349922239503e-06, "loss": 0.1967, "step": 836 }, { "epoch": 0.03, "learning_rate": 9.674961119751168e-06, "loss": 0.1043, "step": 837 }, { "epoch": 0.03, "learning_rate": 9.674572317262832e-06, "loss": 0.0751, "step": 838 }, { "epoch": 0.03, "learning_rate": 9.674183514774495e-06, "loss": 0.194, "step": 839 }, { "epoch": 0.03, "learning_rate": 9.673794712286159e-06, "loss": 0.193, "step": 840 }, { "epoch": 0.03, "learning_rate": 9.673405909797824e-06, "loss": 0.186, "step": 841 }, { "epoch": 0.03, "learning_rate": 9.673017107309488e-06, "loss": 0.0919, "step": 842 }, { "epoch": 0.03, "learning_rate": 9.672628304821152e-06, "loss": 0.09, "step": 843 }, { "epoch": 0.03, "learning_rate": 9.672239502332815e-06, "loss": 0.2045, "step": 844 }, { "epoch": 0.03, "learning_rate": 9.671850699844479e-06, "loss": 0.3879, "step": 845 }, { "epoch": 0.03, "learning_rate": 9.671461897356144e-06, "loss": 0.2673, "step": 846 }, { "epoch": 0.03, "learning_rate": 9.671073094867808e-06, "loss": 0.4591, "step": 847 }, { "epoch": 0.03, "learning_rate": 9.670684292379471e-06, "loss": 0.3149, "step": 848 }, { "epoch": 0.03, "learning_rate": 9.670295489891137e-06, "loss": 0.5389, "step": 849 }, { "epoch": 0.03, "learning_rate": 9.6699066874028e-06, "loss": 0.344, "step": 850 }, { "epoch": 0.03, "learning_rate": 9.669517884914464e-06, "loss": 0.3107, "step": 851 }, { "epoch": 0.03, "learning_rate": 9.66912908242613e-06, "loss": 0.5008, "step": 852 }, { "epoch": 0.03, "learning_rate": 9.668740279937793e-06, "loss": 0.3085, "step": 853 }, { "epoch": 0.03, "learning_rate": 9.668351477449457e-06, "loss": 0.2869, "step": 854 }, { "epoch": 0.03, "learning_rate": 9.667962674961122e-06, "loss": 0.2361, "step": 855 }, { "epoch": 0.03, "learning_rate": 9.667573872472784e-06, "loss": 0.2152, "step": 856 }, { "epoch": 0.03, "learning_rate": 9.667185069984449e-06, "loss": 0.5693, "step": 857 }, { "epoch": 0.03, "learning_rate": 9.666796267496113e-06, "loss": 0.3241, "step": 858 }, { "epoch": 0.03, "learning_rate": 9.666407465007776e-06, "loss": 0.3914, "step": 859 }, { "epoch": 0.03, "learning_rate": 9.66601866251944e-06, "loss": 0.094, "step": 860 }, { "epoch": 0.03, "learning_rate": 9.665629860031105e-06, "loss": 0.1983, "step": 861 }, { "epoch": 0.03, "learning_rate": 9.665241057542769e-06, "loss": 0.3519, "step": 862 }, { "epoch": 0.03, "learning_rate": 9.664852255054433e-06, "loss": 0.3966, "step": 863 }, { "epoch": 0.03, "learning_rate": 9.664463452566098e-06, "loss": 0.1577, "step": 864 }, { "epoch": 0.03, "learning_rate": 9.664074650077762e-06, "loss": 0.051, "step": 865 }, { "epoch": 0.03, "learning_rate": 9.663685847589425e-06, "loss": 0.0748, "step": 866 }, { "epoch": 0.03, "learning_rate": 9.66329704510109e-06, "loss": 0.274, "step": 867 }, { "epoch": 0.03, "learning_rate": 9.662908242612754e-06, "loss": 0.5193, "step": 868 }, { "epoch": 0.03, "learning_rate": 9.662519440124418e-06, "loss": 0.118, "step": 869 }, { "epoch": 0.03, "learning_rate": 9.662130637636081e-06, "loss": 0.3877, "step": 870 }, { "epoch": 0.03, "learning_rate": 9.661741835147745e-06, "loss": 0.2275, "step": 871 }, { "epoch": 0.03, "learning_rate": 9.66135303265941e-06, "loss": 0.0972, "step": 872 }, { "epoch": 0.03, "learning_rate": 9.660964230171074e-06, "loss": 0.3248, "step": 873 }, { "epoch": 0.03, "learning_rate": 9.660575427682738e-06, "loss": 0.0943, "step": 874 }, { "epoch": 0.03, "learning_rate": 9.660186625194403e-06, "loss": 0.238, "step": 875 }, { "epoch": 0.03, "learning_rate": 9.659797822706067e-06, "loss": 0.1239, "step": 876 }, { "epoch": 0.03, "learning_rate": 9.65940902021773e-06, "loss": 0.5799, "step": 877 }, { "epoch": 0.03, "learning_rate": 9.659020217729394e-06, "loss": 0.2254, "step": 878 }, { "epoch": 0.03, "learning_rate": 9.658631415241059e-06, "loss": 0.1088, "step": 879 }, { "epoch": 0.03, "learning_rate": 9.658242612752723e-06, "loss": 0.2804, "step": 880 }, { "epoch": 0.03, "learning_rate": 9.657853810264386e-06, "loss": 0.0971, "step": 881 }, { "epoch": 0.03, "learning_rate": 9.65746500777605e-06, "loss": 0.0934, "step": 882 }, { "epoch": 0.03, "learning_rate": 9.657076205287714e-06, "loss": 0.161, "step": 883 }, { "epoch": 0.03, "learning_rate": 9.656687402799379e-06, "loss": 0.1446, "step": 884 }, { "epoch": 0.03, "learning_rate": 9.656298600311043e-06, "loss": 0.407, "step": 885 }, { "epoch": 0.03, "learning_rate": 9.655909797822706e-06, "loss": 0.1894, "step": 886 }, { "epoch": 0.03, "learning_rate": 9.655520995334372e-06, "loss": 0.5367, "step": 887 }, { "epoch": 0.03, "learning_rate": 9.655132192846035e-06, "loss": 0.2051, "step": 888 }, { "epoch": 0.03, "learning_rate": 9.654743390357699e-06, "loss": 0.1271, "step": 889 }, { "epoch": 0.03, "learning_rate": 9.654354587869364e-06, "loss": 0.2521, "step": 890 }, { "epoch": 0.03, "learning_rate": 9.653965785381028e-06, "loss": 0.2675, "step": 891 }, { "epoch": 0.03, "learning_rate": 9.653576982892691e-06, "loss": 0.3395, "step": 892 }, { "epoch": 0.03, "learning_rate": 9.653188180404357e-06, "loss": 0.4135, "step": 893 }, { "epoch": 0.03, "learning_rate": 9.652799377916019e-06, "loss": 0.5682, "step": 894 }, { "epoch": 0.03, "learning_rate": 9.652410575427684e-06, "loss": 0.3266, "step": 895 }, { "epoch": 0.03, "learning_rate": 9.652021772939348e-06, "loss": 0.1009, "step": 896 }, { "epoch": 0.03, "learning_rate": 9.651632970451011e-06, "loss": 0.2875, "step": 897 }, { "epoch": 0.03, "learning_rate": 9.651244167962675e-06, "loss": 0.6444, "step": 898 }, { "epoch": 0.03, "learning_rate": 9.65085536547434e-06, "loss": 0.0967, "step": 899 }, { "epoch": 0.03, "learning_rate": 9.650466562986004e-06, "loss": 0.2109, "step": 900 }, { "epoch": 0.04, "learning_rate": 9.650077760497667e-06, "loss": 0.3348, "step": 901 }, { "epoch": 0.04, "learning_rate": 9.649688958009333e-06, "loss": 0.2001, "step": 902 }, { "epoch": 0.04, "learning_rate": 9.649300155520996e-06, "loss": 0.3632, "step": 903 }, { "epoch": 0.04, "learning_rate": 9.64891135303266e-06, "loss": 0.222, "step": 904 }, { "epoch": 0.04, "learning_rate": 9.648522550544325e-06, "loss": 0.8221, "step": 905 }, { "epoch": 0.04, "learning_rate": 9.648133748055989e-06, "loss": 0.1915, "step": 906 }, { "epoch": 0.04, "learning_rate": 9.647744945567653e-06, "loss": 0.0764, "step": 907 }, { "epoch": 0.04, "learning_rate": 9.647356143079316e-06, "loss": 0.21, "step": 908 }, { "epoch": 0.04, "learning_rate": 9.64696734059098e-06, "loss": 0.3826, "step": 909 }, { "epoch": 0.04, "learning_rate": 9.646578538102645e-06, "loss": 0.1969, "step": 910 }, { "epoch": 0.04, "learning_rate": 9.646189735614309e-06, "loss": 0.347, "step": 911 }, { "epoch": 0.04, "learning_rate": 9.645800933125972e-06, "loss": 0.47, "step": 912 }, { "epoch": 0.04, "learning_rate": 9.645412130637638e-06, "loss": 0.1558, "step": 913 }, { "epoch": 0.04, "learning_rate": 9.645023328149301e-06, "loss": 0.147, "step": 914 }, { "epoch": 0.04, "learning_rate": 9.644634525660965e-06, "loss": 0.3294, "step": 915 }, { "epoch": 0.04, "learning_rate": 9.644245723172629e-06, "loss": 0.2702, "step": 916 }, { "epoch": 0.04, "learning_rate": 9.643856920684294e-06, "loss": 0.202, "step": 917 }, { "epoch": 0.04, "learning_rate": 9.643468118195958e-06, "loss": 0.3488, "step": 918 }, { "epoch": 0.04, "learning_rate": 9.643079315707621e-06, "loss": 0.2424, "step": 919 }, { "epoch": 0.04, "learning_rate": 9.642690513219285e-06, "loss": 0.1368, "step": 920 }, { "epoch": 0.04, "learning_rate": 9.642301710730948e-06, "loss": 0.2748, "step": 921 }, { "epoch": 0.04, "learning_rate": 9.641912908242614e-06, "loss": 0.155, "step": 922 }, { "epoch": 0.04, "learning_rate": 9.641524105754277e-06, "loss": 0.2842, "step": 923 }, { "epoch": 0.04, "learning_rate": 9.641135303265941e-06, "loss": 0.1456, "step": 924 }, { "epoch": 0.04, "learning_rate": 9.640746500777606e-06, "loss": 0.1101, "step": 925 }, { "epoch": 0.04, "learning_rate": 9.64035769828927e-06, "loss": 0.2293, "step": 926 }, { "epoch": 0.04, "learning_rate": 9.639968895800934e-06, "loss": 0.3571, "step": 927 }, { "epoch": 0.04, "learning_rate": 9.639580093312599e-06, "loss": 0.147, "step": 928 }, { "epoch": 0.04, "learning_rate": 9.639191290824263e-06, "loss": 0.1624, "step": 929 }, { "epoch": 0.04, "learning_rate": 9.638802488335926e-06, "loss": 0.1512, "step": 930 }, { "epoch": 0.04, "learning_rate": 9.638413685847591e-06, "loss": 0.2992, "step": 931 }, { "epoch": 0.04, "learning_rate": 9.638024883359253e-06, "loss": 0.0411, "step": 932 }, { "epoch": 0.04, "learning_rate": 9.637636080870919e-06, "loss": 0.3962, "step": 933 }, { "epoch": 0.04, "learning_rate": 9.637247278382582e-06, "loss": 0.2922, "step": 934 }, { "epoch": 0.04, "learning_rate": 9.636858475894246e-06, "loss": 0.6288, "step": 935 }, { "epoch": 0.04, "learning_rate": 9.63646967340591e-06, "loss": 0.3375, "step": 936 }, { "epoch": 0.04, "learning_rate": 9.636080870917575e-06, "loss": 0.2004, "step": 937 }, { "epoch": 0.04, "learning_rate": 9.635692068429239e-06, "loss": 0.2668, "step": 938 }, { "epoch": 0.04, "learning_rate": 9.635303265940902e-06, "loss": 0.2955, "step": 939 }, { "epoch": 0.04, "learning_rate": 9.634914463452567e-06, "loss": 0.4016, "step": 940 }, { "epoch": 0.04, "learning_rate": 9.634525660964231e-06, "loss": 0.1973, "step": 941 }, { "epoch": 0.04, "learning_rate": 9.634136858475895e-06, "loss": 0.1107, "step": 942 }, { "epoch": 0.04, "learning_rate": 9.63374805598756e-06, "loss": 0.2318, "step": 943 }, { "epoch": 0.04, "learning_rate": 9.633359253499222e-06, "loss": 0.1997, "step": 944 }, { "epoch": 0.04, "learning_rate": 9.632970451010887e-06, "loss": 0.2637, "step": 945 }, { "epoch": 0.04, "learning_rate": 9.632581648522551e-06, "loss": 0.2032, "step": 946 }, { "epoch": 0.04, "learning_rate": 9.632192846034215e-06, "loss": 0.2442, "step": 947 }, { "epoch": 0.04, "learning_rate": 9.63180404354588e-06, "loss": 0.3417, "step": 948 }, { "epoch": 0.04, "learning_rate": 9.631415241057544e-06, "loss": 0.1683, "step": 949 }, { "epoch": 0.04, "learning_rate": 9.631026438569207e-06, "loss": 0.3454, "step": 950 }, { "epoch": 0.04, "learning_rate": 9.630637636080872e-06, "loss": 0.2864, "step": 951 }, { "epoch": 0.04, "learning_rate": 9.630248833592536e-06, "loss": 0.1863, "step": 952 }, { "epoch": 0.04, "learning_rate": 9.6298600311042e-06, "loss": 0.6117, "step": 953 }, { "epoch": 0.04, "learning_rate": 9.629471228615865e-06, "loss": 0.3763, "step": 954 }, { "epoch": 0.04, "learning_rate": 9.629082426127529e-06, "loss": 0.1631, "step": 955 }, { "epoch": 0.04, "learning_rate": 9.628693623639192e-06, "loss": 0.3133, "step": 956 }, { "epoch": 0.04, "learning_rate": 9.628304821150856e-06, "loss": 0.0387, "step": 957 }, { "epoch": 0.04, "learning_rate": 9.62791601866252e-06, "loss": 0.0833, "step": 958 }, { "epoch": 0.04, "learning_rate": 9.627527216174183e-06, "loss": 0.3068, "step": 959 }, { "epoch": 0.04, "learning_rate": 9.627138413685849e-06, "loss": 0.1721, "step": 960 }, { "epoch": 0.04, "learning_rate": 9.626749611197512e-06, "loss": 0.0872, "step": 961 }, { "epoch": 0.04, "learning_rate": 9.626360808709176e-06, "loss": 0.1253, "step": 962 }, { "epoch": 0.04, "learning_rate": 9.625972006220841e-06, "loss": 0.0956, "step": 963 }, { "epoch": 0.04, "learning_rate": 9.625583203732505e-06, "loss": 0.3492, "step": 964 }, { "epoch": 0.04, "learning_rate": 9.625194401244168e-06, "loss": 0.4105, "step": 965 }, { "epoch": 0.04, "learning_rate": 9.624805598755834e-06, "loss": 0.8928, "step": 966 }, { "epoch": 0.04, "learning_rate": 9.624416796267497e-06, "loss": 0.0943, "step": 967 }, { "epoch": 0.04, "learning_rate": 9.624027993779161e-06, "loss": 0.4384, "step": 968 }, { "epoch": 0.04, "learning_rate": 9.623639191290826e-06, "loss": 0.5032, "step": 969 }, { "epoch": 0.04, "learning_rate": 9.623250388802488e-06, "loss": 0.3937, "step": 970 }, { "epoch": 0.04, "learning_rate": 9.622861586314154e-06, "loss": 0.5512, "step": 971 }, { "epoch": 0.04, "learning_rate": 9.622472783825817e-06, "loss": 0.2168, "step": 972 }, { "epoch": 0.04, "learning_rate": 9.62208398133748e-06, "loss": 0.3217, "step": 973 }, { "epoch": 0.04, "learning_rate": 9.621695178849146e-06, "loss": 0.4605, "step": 974 }, { "epoch": 0.04, "learning_rate": 9.62130637636081e-06, "loss": 0.5344, "step": 975 }, { "epoch": 0.04, "learning_rate": 9.620917573872473e-06, "loss": 0.573, "step": 976 }, { "epoch": 0.04, "learning_rate": 9.620528771384137e-06, "loss": 0.3458, "step": 977 }, { "epoch": 0.04, "learning_rate": 9.620139968895802e-06, "loss": 0.0701, "step": 978 }, { "epoch": 0.04, "learning_rate": 9.619751166407466e-06, "loss": 0.3819, "step": 979 }, { "epoch": 0.04, "learning_rate": 9.61936236391913e-06, "loss": 0.1538, "step": 980 }, { "epoch": 0.04, "learning_rate": 9.618973561430795e-06, "loss": 0.0886, "step": 981 }, { "epoch": 0.04, "learning_rate": 9.618584758942457e-06, "loss": 0.1767, "step": 982 }, { "epoch": 0.04, "learning_rate": 9.618195956454122e-06, "loss": 0.1681, "step": 983 }, { "epoch": 0.04, "learning_rate": 9.617807153965786e-06, "loss": 0.0678, "step": 984 }, { "epoch": 0.04, "learning_rate": 9.61741835147745e-06, "loss": 0.4475, "step": 985 }, { "epoch": 0.04, "learning_rate": 9.617029548989115e-06, "loss": 0.2384, "step": 986 }, { "epoch": 0.04, "learning_rate": 9.616640746500778e-06, "loss": 0.2286, "step": 987 }, { "epoch": 0.04, "learning_rate": 9.616251944012442e-06, "loss": 0.2321, "step": 988 }, { "epoch": 0.04, "learning_rate": 9.615863141524107e-06, "loss": 0.5778, "step": 989 }, { "epoch": 0.04, "learning_rate": 9.615474339035771e-06, "loss": 0.126, "step": 990 }, { "epoch": 0.04, "learning_rate": 9.615085536547435e-06, "loss": 0.1656, "step": 991 }, { "epoch": 0.04, "learning_rate": 9.6146967340591e-06, "loss": 0.1866, "step": 992 }, { "epoch": 0.04, "learning_rate": 9.614307931570763e-06, "loss": 0.117, "step": 993 }, { "epoch": 0.04, "learning_rate": 9.613919129082427e-06, "loss": 0.2247, "step": 994 }, { "epoch": 0.04, "learning_rate": 9.61353032659409e-06, "loss": 0.0907, "step": 995 }, { "epoch": 0.04, "learning_rate": 9.613141524105754e-06, "loss": 0.3421, "step": 996 }, { "epoch": 0.04, "learning_rate": 9.612752721617418e-06, "loss": 0.0703, "step": 997 }, { "epoch": 0.04, "learning_rate": 9.612363919129083e-06, "loss": 0.3701, "step": 998 }, { "epoch": 0.04, "learning_rate": 9.611975116640747e-06, "loss": 0.2224, "step": 999 }, { "epoch": 0.04, "learning_rate": 9.61158631415241e-06, "loss": 0.2193, "step": 1000 }, { "epoch": 0.04, "learning_rate": 9.611197511664076e-06, "loss": 0.2086, "step": 1001 }, { "epoch": 0.04, "learning_rate": 9.61080870917574e-06, "loss": 0.1062, "step": 1002 }, { "epoch": 0.04, "learning_rate": 9.610419906687403e-06, "loss": 0.3624, "step": 1003 }, { "epoch": 0.04, "learning_rate": 9.610031104199068e-06, "loss": 0.4092, "step": 1004 }, { "epoch": 0.04, "learning_rate": 9.609642301710732e-06, "loss": 0.3711, "step": 1005 }, { "epoch": 0.04, "learning_rate": 9.609253499222396e-06, "loss": 0.3915, "step": 1006 }, { "epoch": 0.04, "learning_rate": 9.608864696734061e-06, "loss": 0.0392, "step": 1007 }, { "epoch": 0.04, "learning_rate": 9.608475894245723e-06, "loss": 0.3999, "step": 1008 }, { "epoch": 0.04, "learning_rate": 9.608087091757388e-06, "loss": 0.0874, "step": 1009 }, { "epoch": 0.04, "learning_rate": 9.607698289269052e-06, "loss": 0.3629, "step": 1010 }, { "epoch": 0.04, "learning_rate": 9.607309486780716e-06, "loss": 0.3872, "step": 1011 }, { "epoch": 0.04, "learning_rate": 9.606920684292381e-06, "loss": 0.3183, "step": 1012 }, { "epoch": 0.04, "learning_rate": 9.606531881804045e-06, "loss": 0.1177, "step": 1013 }, { "epoch": 0.04, "learning_rate": 9.606143079315708e-06, "loss": 0.1676, "step": 1014 }, { "epoch": 0.04, "learning_rate": 9.605754276827372e-06, "loss": 0.305, "step": 1015 }, { "epoch": 0.04, "learning_rate": 9.605365474339037e-06, "loss": 0.4595, "step": 1016 }, { "epoch": 0.04, "learning_rate": 9.6049766718507e-06, "loss": 0.1438, "step": 1017 }, { "epoch": 0.04, "learning_rate": 9.604587869362364e-06, "loss": 0.2665, "step": 1018 }, { "epoch": 0.04, "learning_rate": 9.60419906687403e-06, "loss": 0.2334, "step": 1019 }, { "epoch": 0.04, "learning_rate": 9.603810264385692e-06, "loss": 0.4926, "step": 1020 }, { "epoch": 0.04, "learning_rate": 9.603421461897357e-06, "loss": 0.1835, "step": 1021 }, { "epoch": 0.04, "learning_rate": 9.60303265940902e-06, "loss": 0.3326, "step": 1022 }, { "epoch": 0.04, "learning_rate": 9.602643856920684e-06, "loss": 0.1247, "step": 1023 }, { "epoch": 0.04, "learning_rate": 9.60225505443235e-06, "loss": 0.4599, "step": 1024 }, { "epoch": 0.04, "learning_rate": 9.601866251944013e-06, "loss": 0.4059, "step": 1025 }, { "epoch": 0.04, "learning_rate": 9.601477449455677e-06, "loss": 0.3197, "step": 1026 }, { "epoch": 0.04, "learning_rate": 9.601088646967342e-06, "loss": 0.2844, "step": 1027 }, { "epoch": 0.04, "learning_rate": 9.600699844479006e-06, "loss": 0.3649, "step": 1028 }, { "epoch": 0.04, "learning_rate": 9.60031104199067e-06, "loss": 0.0622, "step": 1029 }, { "epoch": 0.04, "learning_rate": 9.599922239502335e-06, "loss": 0.1762, "step": 1030 }, { "epoch": 0.04, "learning_rate": 9.599533437013998e-06, "loss": 0.5385, "step": 1031 }, { "epoch": 0.04, "learning_rate": 9.599144634525662e-06, "loss": 0.0649, "step": 1032 }, { "epoch": 0.04, "learning_rate": 9.598755832037326e-06, "loss": 0.1415, "step": 1033 }, { "epoch": 0.04, "learning_rate": 9.59836702954899e-06, "loss": 0.3054, "step": 1034 }, { "epoch": 0.04, "learning_rate": 9.597978227060653e-06, "loss": 0.2198, "step": 1035 }, { "epoch": 0.04, "learning_rate": 9.597589424572318e-06, "loss": 0.0901, "step": 1036 }, { "epoch": 0.04, "learning_rate": 9.597200622083982e-06, "loss": 0.0566, "step": 1037 }, { "epoch": 0.04, "learning_rate": 9.596811819595645e-06, "loss": 0.3068, "step": 1038 }, { "epoch": 0.04, "learning_rate": 9.59642301710731e-06, "loss": 0.0889, "step": 1039 }, { "epoch": 0.04, "learning_rate": 9.596034214618974e-06, "loss": 0.3096, "step": 1040 }, { "epoch": 0.04, "learning_rate": 9.595645412130638e-06, "loss": 0.1358, "step": 1041 }, { "epoch": 0.04, "learning_rate": 9.595256609642303e-06, "loss": 0.2586, "step": 1042 }, { "epoch": 0.04, "learning_rate": 9.594867807153967e-06, "loss": 0.1209, "step": 1043 }, { "epoch": 0.04, "learning_rate": 9.59447900466563e-06, "loss": 0.4153, "step": 1044 }, { "epoch": 0.04, "learning_rate": 9.594090202177296e-06, "loss": 0.3723, "step": 1045 }, { "epoch": 0.04, "learning_rate": 9.593701399688958e-06, "loss": 0.1584, "step": 1046 }, { "epoch": 0.04, "learning_rate": 9.593312597200623e-06, "loss": 0.3855, "step": 1047 }, { "epoch": 0.04, "learning_rate": 9.592923794712287e-06, "loss": 0.1464, "step": 1048 }, { "epoch": 0.04, "learning_rate": 9.59253499222395e-06, "loss": 0.582, "step": 1049 }, { "epoch": 0.04, "learning_rate": 9.592146189735616e-06, "loss": 0.2644, "step": 1050 }, { "epoch": 0.04, "learning_rate": 9.59175738724728e-06, "loss": 0.3607, "step": 1051 }, { "epoch": 0.04, "learning_rate": 9.591368584758943e-06, "loss": 0.364, "step": 1052 }, { "epoch": 0.04, "learning_rate": 9.590979782270608e-06, "loss": 0.4309, "step": 1053 }, { "epoch": 0.04, "learning_rate": 9.590590979782272e-06, "loss": 0.5067, "step": 1054 }, { "epoch": 0.04, "learning_rate": 9.590202177293936e-06, "loss": 0.0924, "step": 1055 }, { "epoch": 0.04, "learning_rate": 9.589813374805599e-06, "loss": 0.5266, "step": 1056 }, { "epoch": 0.04, "learning_rate": 9.589424572317264e-06, "loss": 0.2002, "step": 1057 }, { "epoch": 0.04, "learning_rate": 9.589035769828926e-06, "loss": 0.2927, "step": 1058 }, { "epoch": 0.04, "learning_rate": 9.588646967340592e-06, "loss": 0.4308, "step": 1059 }, { "epoch": 0.04, "learning_rate": 9.588258164852255e-06, "loss": 0.9281, "step": 1060 }, { "epoch": 0.04, "learning_rate": 9.587869362363919e-06, "loss": 0.3286, "step": 1061 }, { "epoch": 0.04, "learning_rate": 9.587480559875584e-06, "loss": 0.2827, "step": 1062 }, { "epoch": 0.04, "learning_rate": 9.587091757387248e-06, "loss": 0.283, "step": 1063 }, { "epoch": 0.04, "learning_rate": 9.586702954898912e-06, "loss": 0.2303, "step": 1064 }, { "epoch": 0.04, "learning_rate": 9.586314152410577e-06, "loss": 0.1425, "step": 1065 }, { "epoch": 0.04, "learning_rate": 9.58592534992224e-06, "loss": 0.1383, "step": 1066 }, { "epoch": 0.04, "learning_rate": 9.585536547433904e-06, "loss": 0.2675, "step": 1067 }, { "epoch": 0.04, "learning_rate": 9.58514774494557e-06, "loss": 0.1712, "step": 1068 }, { "epoch": 0.04, "learning_rate": 9.584758942457233e-06, "loss": 0.6265, "step": 1069 }, { "epoch": 0.04, "learning_rate": 9.584370139968897e-06, "loss": 0.1452, "step": 1070 }, { "epoch": 0.04, "learning_rate": 9.58398133748056e-06, "loss": 0.1902, "step": 1071 }, { "epoch": 0.04, "learning_rate": 9.583592534992224e-06, "loss": 0.3767, "step": 1072 }, { "epoch": 0.04, "learning_rate": 9.58320373250389e-06, "loss": 0.5267, "step": 1073 }, { "epoch": 0.04, "learning_rate": 9.582814930015553e-06, "loss": 0.0655, "step": 1074 }, { "epoch": 0.04, "learning_rate": 9.582426127527217e-06, "loss": 0.1212, "step": 1075 }, { "epoch": 0.04, "learning_rate": 9.58203732503888e-06, "loss": 0.2565, "step": 1076 }, { "epoch": 0.04, "learning_rate": 9.581648522550546e-06, "loss": 0.1773, "step": 1077 }, { "epoch": 0.04, "learning_rate": 9.581259720062209e-06, "loss": 0.447, "step": 1078 }, { "epoch": 0.04, "learning_rate": 9.580870917573873e-06, "loss": 0.3435, "step": 1079 }, { "epoch": 0.04, "learning_rate": 9.580482115085538e-06, "loss": 0.3929, "step": 1080 }, { "epoch": 0.04, "learning_rate": 9.580093312597202e-06, "loss": 0.2924, "step": 1081 }, { "epoch": 0.04, "learning_rate": 9.579704510108865e-06, "loss": 0.3289, "step": 1082 }, { "epoch": 0.04, "learning_rate": 9.57931570762053e-06, "loss": 0.0825, "step": 1083 }, { "epoch": 0.04, "learning_rate": 9.578926905132193e-06, "loss": 0.3187, "step": 1084 }, { "epoch": 0.04, "learning_rate": 9.578538102643858e-06, "loss": 0.6953, "step": 1085 }, { "epoch": 0.04, "learning_rate": 9.578149300155522e-06, "loss": 0.4257, "step": 1086 }, { "epoch": 0.04, "learning_rate": 9.577760497667185e-06, "loss": 0.2937, "step": 1087 }, { "epoch": 0.04, "learning_rate": 9.57737169517885e-06, "loss": 0.1927, "step": 1088 }, { "epoch": 0.04, "learning_rate": 9.576982892690514e-06, "loss": 0.1278, "step": 1089 }, { "epoch": 0.04, "learning_rate": 9.576594090202178e-06, "loss": 0.2732, "step": 1090 }, { "epoch": 0.04, "learning_rate": 9.576205287713843e-06, "loss": 0.1419, "step": 1091 }, { "epoch": 0.04, "learning_rate": 9.575816485225507e-06, "loss": 0.2082, "step": 1092 }, { "epoch": 0.04, "learning_rate": 9.57542768273717e-06, "loss": 0.3084, "step": 1093 }, { "epoch": 0.04, "learning_rate": 9.575038880248834e-06, "loss": 0.4917, "step": 1094 }, { "epoch": 0.04, "learning_rate": 9.5746500777605e-06, "loss": 0.379, "step": 1095 }, { "epoch": 0.04, "learning_rate": 9.574261275272161e-06, "loss": 0.1492, "step": 1096 }, { "epoch": 0.04, "learning_rate": 9.573872472783827e-06, "loss": 0.1071, "step": 1097 }, { "epoch": 0.04, "learning_rate": 9.57348367029549e-06, "loss": 0.2809, "step": 1098 }, { "epoch": 0.04, "learning_rate": 9.573094867807154e-06, "loss": 0.147, "step": 1099 }, { "epoch": 0.04, "learning_rate": 9.572706065318819e-06, "loss": 0.2112, "step": 1100 }, { "epoch": 0.04, "learning_rate": 9.572317262830483e-06, "loss": 0.1178, "step": 1101 }, { "epoch": 0.04, "learning_rate": 9.571928460342146e-06, "loss": 0.1656, "step": 1102 }, { "epoch": 0.04, "learning_rate": 9.571539657853812e-06, "loss": 0.0708, "step": 1103 }, { "epoch": 0.04, "learning_rate": 9.571150855365475e-06, "loss": 0.3682, "step": 1104 }, { "epoch": 0.04, "learning_rate": 9.570762052877139e-06, "loss": 0.1873, "step": 1105 }, { "epoch": 0.04, "learning_rate": 9.570373250388804e-06, "loss": 0.3121, "step": 1106 }, { "epoch": 0.04, "learning_rate": 9.569984447900468e-06, "loss": 0.2131, "step": 1107 }, { "epoch": 0.04, "learning_rate": 9.569595645412132e-06, "loss": 0.1777, "step": 1108 }, { "epoch": 0.04, "learning_rate": 9.569206842923795e-06, "loss": 0.2771, "step": 1109 }, { "epoch": 0.04, "learning_rate": 9.568818040435459e-06, "loss": 0.1467, "step": 1110 }, { "epoch": 0.04, "learning_rate": 9.568429237947124e-06, "loss": 0.1265, "step": 1111 }, { "epoch": 0.04, "learning_rate": 9.568040435458788e-06, "loss": 0.1361, "step": 1112 }, { "epoch": 0.04, "learning_rate": 9.567651632970451e-06, "loss": 0.1137, "step": 1113 }, { "epoch": 0.04, "learning_rate": 9.567262830482115e-06, "loss": 0.2308, "step": 1114 }, { "epoch": 0.04, "learning_rate": 9.56687402799378e-06, "loss": 0.2651, "step": 1115 }, { "epoch": 0.04, "learning_rate": 9.566485225505444e-06, "loss": 0.1002, "step": 1116 }, { "epoch": 0.04, "learning_rate": 9.566096423017108e-06, "loss": 0.2063, "step": 1117 }, { "epoch": 0.04, "learning_rate": 9.565707620528773e-06, "loss": 0.2906, "step": 1118 }, { "epoch": 0.04, "learning_rate": 9.565318818040437e-06, "loss": 0.1807, "step": 1119 }, { "epoch": 0.04, "learning_rate": 9.5649300155521e-06, "loss": 0.2838, "step": 1120 }, { "epoch": 0.04, "learning_rate": 9.564541213063765e-06, "loss": 0.3407, "step": 1121 }, { "epoch": 0.04, "learning_rate": 9.564152410575427e-06, "loss": 0.0893, "step": 1122 }, { "epoch": 0.04, "learning_rate": 9.563763608087093e-06, "loss": 0.1093, "step": 1123 }, { "epoch": 0.04, "learning_rate": 9.563374805598756e-06, "loss": 0.2561, "step": 1124 }, { "epoch": 0.04, "learning_rate": 9.56298600311042e-06, "loss": 0.3185, "step": 1125 }, { "epoch": 0.04, "learning_rate": 9.562597200622085e-06, "loss": 0.203, "step": 1126 }, { "epoch": 0.04, "learning_rate": 9.562208398133749e-06, "loss": 0.2906, "step": 1127 }, { "epoch": 0.04, "learning_rate": 9.561819595645413e-06, "loss": 0.509, "step": 1128 }, { "epoch": 0.04, "learning_rate": 9.561430793157078e-06, "loss": 0.1434, "step": 1129 }, { "epoch": 0.04, "learning_rate": 9.561041990668741e-06, "loss": 0.279, "step": 1130 }, { "epoch": 0.04, "learning_rate": 9.560653188180405e-06, "loss": 0.2164, "step": 1131 }, { "epoch": 0.04, "learning_rate": 9.56026438569207e-06, "loss": 0.4589, "step": 1132 }, { "epoch": 0.04, "learning_rate": 9.559875583203734e-06, "loss": 0.1028, "step": 1133 }, { "epoch": 0.04, "learning_rate": 9.559486780715396e-06, "loss": 0.2088, "step": 1134 }, { "epoch": 0.04, "learning_rate": 9.559097978227061e-06, "loss": 0.0532, "step": 1135 }, { "epoch": 0.04, "learning_rate": 9.558709175738725e-06, "loss": 0.3605, "step": 1136 }, { "epoch": 0.04, "learning_rate": 9.558320373250389e-06, "loss": 0.7769, "step": 1137 }, { "epoch": 0.04, "learning_rate": 9.557931570762054e-06, "loss": 0.2525, "step": 1138 }, { "epoch": 0.04, "learning_rate": 9.557542768273718e-06, "loss": 0.149, "step": 1139 }, { "epoch": 0.04, "learning_rate": 9.557153965785381e-06, "loss": 0.2859, "step": 1140 }, { "epoch": 0.04, "learning_rate": 9.556765163297046e-06, "loss": 0.2493, "step": 1141 }, { "epoch": 0.04, "learning_rate": 9.55637636080871e-06, "loss": 0.143, "step": 1142 }, { "epoch": 0.04, "learning_rate": 9.555987558320374e-06, "loss": 0.1191, "step": 1143 }, { "epoch": 0.04, "learning_rate": 9.555598755832039e-06, "loss": 0.5667, "step": 1144 }, { "epoch": 0.04, "learning_rate": 9.555209953343703e-06, "loss": 0.1953, "step": 1145 }, { "epoch": 0.04, "learning_rate": 9.554821150855366e-06, "loss": 0.2038, "step": 1146 }, { "epoch": 0.04, "learning_rate": 9.55443234836703e-06, "loss": 0.1807, "step": 1147 }, { "epoch": 0.04, "learning_rate": 9.554043545878694e-06, "loss": 0.4901, "step": 1148 }, { "epoch": 0.04, "learning_rate": 9.553654743390359e-06, "loss": 0.0615, "step": 1149 }, { "epoch": 0.04, "learning_rate": 9.553265940902023e-06, "loss": 0.5038, "step": 1150 }, { "epoch": 0.04, "learning_rate": 9.552877138413686e-06, "loss": 0.4333, "step": 1151 }, { "epoch": 0.04, "learning_rate": 9.552488335925351e-06, "loss": 0.1783, "step": 1152 }, { "epoch": 0.04, "learning_rate": 9.552099533437015e-06, "loss": 0.4721, "step": 1153 }, { "epoch": 0.04, "learning_rate": 9.551710730948679e-06, "loss": 0.2149, "step": 1154 }, { "epoch": 0.04, "learning_rate": 9.551321928460342e-06, "loss": 0.1631, "step": 1155 }, { "epoch": 0.04, "learning_rate": 9.550933125972008e-06, "loss": 0.4971, "step": 1156 }, { "epoch": 0.04, "learning_rate": 9.550544323483671e-06, "loss": 0.3352, "step": 1157 }, { "epoch": 0.05, "learning_rate": 9.550155520995335e-06, "loss": 0.1878, "step": 1158 }, { "epoch": 0.05, "learning_rate": 9.549766718507e-06, "loss": 0.3925, "step": 1159 }, { "epoch": 0.05, "learning_rate": 9.549377916018662e-06, "loss": 0.3108, "step": 1160 }, { "epoch": 0.05, "learning_rate": 9.548989113530328e-06, "loss": 0.3683, "step": 1161 }, { "epoch": 0.05, "learning_rate": 9.548600311041991e-06, "loss": 0.1305, "step": 1162 }, { "epoch": 0.05, "learning_rate": 9.548211508553655e-06, "loss": 0.2767, "step": 1163 }, { "epoch": 0.05, "learning_rate": 9.54782270606532e-06, "loss": 0.4385, "step": 1164 }, { "epoch": 0.05, "learning_rate": 9.547433903576984e-06, "loss": 0.3987, "step": 1165 }, { "epoch": 0.05, "learning_rate": 9.547045101088647e-06, "loss": 0.3361, "step": 1166 }, { "epoch": 0.05, "learning_rate": 9.546656298600313e-06, "loss": 0.8082, "step": 1167 }, { "epoch": 0.05, "learning_rate": 9.546267496111976e-06, "loss": 0.3187, "step": 1168 }, { "epoch": 0.05, "learning_rate": 9.54587869362364e-06, "loss": 0.3012, "step": 1169 }, { "epoch": 0.05, "learning_rate": 9.545489891135305e-06, "loss": 0.3004, "step": 1170 }, { "epoch": 0.05, "learning_rate": 9.545101088646969e-06, "loss": 0.3351, "step": 1171 }, { "epoch": 0.05, "learning_rate": 9.544712286158632e-06, "loss": 0.4383, "step": 1172 }, { "epoch": 0.05, "learning_rate": 9.544323483670296e-06, "loss": 0.4744, "step": 1173 }, { "epoch": 0.05, "learning_rate": 9.54393468118196e-06, "loss": 0.4835, "step": 1174 }, { "epoch": 0.05, "learning_rate": 9.543545878693623e-06, "loss": 0.2741, "step": 1175 }, { "epoch": 0.05, "learning_rate": 9.543157076205289e-06, "loss": 0.0874, "step": 1176 }, { "epoch": 0.05, "learning_rate": 9.542768273716952e-06, "loss": 0.3466, "step": 1177 }, { "epoch": 0.05, "learning_rate": 9.542379471228616e-06, "loss": 0.4201, "step": 1178 }, { "epoch": 0.05, "learning_rate": 9.541990668740281e-06, "loss": 0.3088, "step": 1179 }, { "epoch": 0.05, "learning_rate": 9.541601866251945e-06, "loss": 0.3396, "step": 1180 }, { "epoch": 0.05, "learning_rate": 9.541213063763609e-06, "loss": 0.2388, "step": 1181 }, { "epoch": 0.05, "learning_rate": 9.540824261275274e-06, "loss": 0.2306, "step": 1182 }, { "epoch": 0.05, "learning_rate": 9.540435458786937e-06, "loss": 0.177, "step": 1183 }, { "epoch": 0.05, "learning_rate": 9.540046656298601e-06, "loss": 0.1942, "step": 1184 }, { "epoch": 0.05, "learning_rate": 9.539657853810265e-06, "loss": 0.1981, "step": 1185 }, { "epoch": 0.05, "learning_rate": 9.539269051321928e-06, "loss": 0.2048, "step": 1186 }, { "epoch": 0.05, "learning_rate": 9.538880248833594e-06, "loss": 0.1647, "step": 1187 }, { "epoch": 0.05, "learning_rate": 9.538491446345257e-06, "loss": 0.1369, "step": 1188 }, { "epoch": 0.05, "learning_rate": 9.538102643856921e-06, "loss": 0.0711, "step": 1189 }, { "epoch": 0.05, "learning_rate": 9.537713841368586e-06, "loss": 0.1554, "step": 1190 }, { "epoch": 0.05, "learning_rate": 9.53732503888025e-06, "loss": 0.4194, "step": 1191 }, { "epoch": 0.05, "learning_rate": 9.536936236391914e-06, "loss": 0.1282, "step": 1192 }, { "epoch": 0.05, "learning_rate": 9.536547433903577e-06, "loss": 0.1597, "step": 1193 }, { "epoch": 0.05, "learning_rate": 9.536158631415242e-06, "loss": 0.2463, "step": 1194 }, { "epoch": 0.05, "learning_rate": 9.535769828926906e-06, "loss": 0.385, "step": 1195 }, { "epoch": 0.05, "learning_rate": 9.53538102643857e-06, "loss": 0.1665, "step": 1196 }, { "epoch": 0.05, "learning_rate": 9.534992223950235e-06, "loss": 0.0934, "step": 1197 }, { "epoch": 0.05, "learning_rate": 9.534603421461897e-06, "loss": 0.1417, "step": 1198 }, { "epoch": 0.05, "learning_rate": 9.534214618973562e-06, "loss": 0.1926, "step": 1199 }, { "epoch": 0.05, "learning_rate": 9.533825816485226e-06, "loss": 0.1861, "step": 1200 }, { "epoch": 0.05, "learning_rate": 9.53343701399689e-06, "loss": 0.0473, "step": 1201 }, { "epoch": 0.05, "learning_rate": 9.533048211508555e-06, "loss": 0.1355, "step": 1202 }, { "epoch": 0.05, "learning_rate": 9.532659409020219e-06, "loss": 0.417, "step": 1203 }, { "epoch": 0.05, "learning_rate": 9.532270606531882e-06, "loss": 0.4478, "step": 1204 }, { "epoch": 0.05, "learning_rate": 9.531881804043547e-06, "loss": 0.3083, "step": 1205 }, { "epoch": 0.05, "learning_rate": 9.531493001555211e-06, "loss": 0.366, "step": 1206 }, { "epoch": 0.05, "learning_rate": 9.531104199066875e-06, "loss": 0.277, "step": 1207 }, { "epoch": 0.05, "learning_rate": 9.53071539657854e-06, "loss": 0.2609, "step": 1208 }, { "epoch": 0.05, "learning_rate": 9.530326594090204e-06, "loss": 0.0725, "step": 1209 }, { "epoch": 0.05, "learning_rate": 9.529937791601867e-06, "loss": 0.0796, "step": 1210 }, { "epoch": 0.05, "learning_rate": 9.529548989113531e-06, "loss": 0.2443, "step": 1211 }, { "epoch": 0.05, "learning_rate": 9.529160186625195e-06, "loss": 0.15, "step": 1212 }, { "epoch": 0.05, "learning_rate": 9.528771384136858e-06, "loss": 0.5069, "step": 1213 }, { "epoch": 0.05, "learning_rate": 9.528382581648524e-06, "loss": 0.1067, "step": 1214 }, { "epoch": 0.05, "learning_rate": 9.527993779160187e-06, "loss": 0.4105, "step": 1215 }, { "epoch": 0.05, "learning_rate": 9.52760497667185e-06, "loss": 0.6277, "step": 1216 }, { "epoch": 0.05, "learning_rate": 9.527216174183516e-06, "loss": 0.2554, "step": 1217 }, { "epoch": 0.05, "learning_rate": 9.52682737169518e-06, "loss": 0.0421, "step": 1218 }, { "epoch": 0.05, "learning_rate": 9.526438569206843e-06, "loss": 0.1085, "step": 1219 }, { "epoch": 0.05, "learning_rate": 9.526049766718509e-06, "loss": 0.078, "step": 1220 }, { "epoch": 0.05, "learning_rate": 9.525660964230172e-06, "loss": 0.1253, "step": 1221 }, { "epoch": 0.05, "learning_rate": 9.525272161741836e-06, "loss": 0.4346, "step": 1222 }, { "epoch": 0.05, "learning_rate": 9.5248833592535e-06, "loss": 0.1803, "step": 1223 }, { "epoch": 0.05, "learning_rate": 9.524494556765163e-06, "loss": 0.3959, "step": 1224 }, { "epoch": 0.05, "learning_rate": 9.524105754276828e-06, "loss": 0.2432, "step": 1225 }, { "epoch": 0.05, "learning_rate": 9.523716951788492e-06, "loss": 0.2392, "step": 1226 }, { "epoch": 0.05, "learning_rate": 9.523328149300156e-06, "loss": 0.337, "step": 1227 }, { "epoch": 0.05, "learning_rate": 9.522939346811821e-06, "loss": 0.1713, "step": 1228 }, { "epoch": 0.05, "learning_rate": 9.522550544323485e-06, "loss": 0.2212, "step": 1229 }, { "epoch": 0.05, "learning_rate": 9.522161741835148e-06, "loss": 0.0479, "step": 1230 }, { "epoch": 0.05, "learning_rate": 9.521772939346812e-06, "loss": 0.3133, "step": 1231 }, { "epoch": 0.05, "learning_rate": 9.521384136858477e-06, "loss": 0.1916, "step": 1232 }, { "epoch": 0.05, "learning_rate": 9.520995334370141e-06, "loss": 0.3304, "step": 1233 }, { "epoch": 0.05, "learning_rate": 9.520606531881805e-06, "loss": 0.247, "step": 1234 }, { "epoch": 0.05, "learning_rate": 9.52021772939347e-06, "loss": 0.0719, "step": 1235 }, { "epoch": 0.05, "learning_rate": 9.519828926905132e-06, "loss": 0.101, "step": 1236 }, { "epoch": 0.05, "learning_rate": 9.519440124416797e-06, "loss": 0.1432, "step": 1237 }, { "epoch": 0.05, "learning_rate": 9.51905132192846e-06, "loss": 0.2337, "step": 1238 }, { "epoch": 0.05, "learning_rate": 9.518662519440124e-06, "loss": 0.082, "step": 1239 }, { "epoch": 0.05, "learning_rate": 9.51827371695179e-06, "loss": 0.1544, "step": 1240 }, { "epoch": 0.05, "learning_rate": 9.517884914463453e-06, "loss": 0.1412, "step": 1241 }, { "epoch": 0.05, "learning_rate": 9.517496111975117e-06, "loss": 0.3831, "step": 1242 }, { "epoch": 0.05, "learning_rate": 9.517107309486782e-06, "loss": 0.0391, "step": 1243 }, { "epoch": 0.05, "learning_rate": 9.516718506998446e-06, "loss": 0.2215, "step": 1244 }, { "epoch": 0.05, "learning_rate": 9.51632970451011e-06, "loss": 0.323, "step": 1245 }, { "epoch": 0.05, "learning_rate": 9.515940902021775e-06, "loss": 0.1856, "step": 1246 }, { "epoch": 0.05, "learning_rate": 9.515552099533438e-06, "loss": 0.1835, "step": 1247 }, { "epoch": 0.05, "learning_rate": 9.515163297045102e-06, "loss": 0.2884, "step": 1248 }, { "epoch": 0.05, "learning_rate": 9.514774494556766e-06, "loss": 0.2574, "step": 1249 }, { "epoch": 0.05, "learning_rate": 9.51438569206843e-06, "loss": 0.3203, "step": 1250 }, { "epoch": 0.05, "learning_rate": 9.513996889580093e-06, "loss": 0.1982, "step": 1251 }, { "epoch": 0.05, "learning_rate": 9.513608087091758e-06, "loss": 0.3264, "step": 1252 }, { "epoch": 0.05, "learning_rate": 9.513219284603422e-06, "loss": 0.0661, "step": 1253 }, { "epoch": 0.05, "learning_rate": 9.512830482115086e-06, "loss": 0.2388, "step": 1254 }, { "epoch": 0.05, "learning_rate": 9.512441679626751e-06, "loss": 0.3092, "step": 1255 }, { "epoch": 0.05, "learning_rate": 9.512052877138415e-06, "loss": 0.3457, "step": 1256 }, { "epoch": 0.05, "learning_rate": 9.511664074650078e-06, "loss": 0.1369, "step": 1257 }, { "epoch": 0.05, "learning_rate": 9.511275272161743e-06, "loss": 0.3852, "step": 1258 }, { "epoch": 0.05, "learning_rate": 9.510886469673407e-06, "loss": 0.2789, "step": 1259 }, { "epoch": 0.05, "learning_rate": 9.51049766718507e-06, "loss": 0.2737, "step": 1260 }, { "epoch": 0.05, "learning_rate": 9.510108864696734e-06, "loss": 0.103, "step": 1261 }, { "epoch": 0.05, "learning_rate": 9.509720062208398e-06, "loss": 0.5511, "step": 1262 }, { "epoch": 0.05, "learning_rate": 9.509331259720063e-06, "loss": 0.1608, "step": 1263 }, { "epoch": 0.05, "learning_rate": 9.508942457231727e-06, "loss": 0.2027, "step": 1264 }, { "epoch": 0.05, "learning_rate": 9.50855365474339e-06, "loss": 0.2975, "step": 1265 }, { "epoch": 0.05, "learning_rate": 9.508164852255056e-06, "loss": 0.1845, "step": 1266 }, { "epoch": 0.05, "learning_rate": 9.50777604976672e-06, "loss": 0.131, "step": 1267 }, { "epoch": 0.05, "learning_rate": 9.507387247278383e-06, "loss": 0.1445, "step": 1268 }, { "epoch": 0.05, "learning_rate": 9.506998444790048e-06, "loss": 0.3579, "step": 1269 }, { "epoch": 0.05, "learning_rate": 9.506609642301712e-06, "loss": 0.1513, "step": 1270 }, { "epoch": 0.05, "learning_rate": 9.506220839813376e-06, "loss": 0.2406, "step": 1271 }, { "epoch": 0.05, "learning_rate": 9.50583203732504e-06, "loss": 0.1414, "step": 1272 }, { "epoch": 0.05, "learning_rate": 9.505443234836705e-06, "loss": 0.2524, "step": 1273 }, { "epoch": 0.05, "learning_rate": 9.505054432348367e-06, "loss": 0.4142, "step": 1274 }, { "epoch": 0.05, "learning_rate": 9.504665629860032e-06, "loss": 0.1059, "step": 1275 }, { "epoch": 0.05, "learning_rate": 9.504276827371696e-06, "loss": 0.1223, "step": 1276 }, { "epoch": 0.05, "learning_rate": 9.50388802488336e-06, "loss": 0.347, "step": 1277 }, { "epoch": 0.05, "learning_rate": 9.503499222395024e-06, "loss": 0.2948, "step": 1278 }, { "epoch": 0.05, "learning_rate": 9.503110419906688e-06, "loss": 0.0839, "step": 1279 }, { "epoch": 0.05, "learning_rate": 9.502721617418352e-06, "loss": 0.1206, "step": 1280 }, { "epoch": 0.05, "learning_rate": 9.502332814930017e-06, "loss": 0.4611, "step": 1281 }, { "epoch": 0.05, "learning_rate": 9.50194401244168e-06, "loss": 0.2354, "step": 1282 }, { "epoch": 0.05, "learning_rate": 9.501555209953344e-06, "loss": 0.3542, "step": 1283 }, { "epoch": 0.05, "learning_rate": 9.50116640746501e-06, "loss": 0.1319, "step": 1284 }, { "epoch": 0.05, "learning_rate": 9.500777604976673e-06, "loss": 0.3193, "step": 1285 }, { "epoch": 0.05, "learning_rate": 9.500388802488337e-06, "loss": 0.3044, "step": 1286 }, { "epoch": 0.05, "learning_rate": 9.5e-06, "loss": 0.1491, "step": 1287 }, { "epoch": 0.05, "learning_rate": 9.499611197511664e-06, "loss": 0.4631, "step": 1288 }, { "epoch": 0.05, "eval_accuracy": 0.9057239057239057, "eval_loss": 0.257296085357666, "eval_runtime": 105.0935, "eval_samples_per_second": 2.826, "eval_steps_per_second": 0.143, "step": 1288 }, { "epoch": 1.0, "learning_rate": 9.49922239502333e-06, "loss": 0.4902, "step": 1289 }, { "epoch": 1.0, "learning_rate": 9.498833592534993e-06, "loss": 0.1533, "step": 1290 }, { "epoch": 1.0, "learning_rate": 9.498444790046657e-06, "loss": 0.0345, "step": 1291 }, { "epoch": 1.0, "learning_rate": 9.49805598755832e-06, "loss": 0.1821, "step": 1292 }, { "epoch": 1.0, "learning_rate": 9.497667185069986e-06, "loss": 0.2972, "step": 1293 }, { "epoch": 1.0, "learning_rate": 9.49727838258165e-06, "loss": 0.1218, "step": 1294 }, { "epoch": 1.0, "learning_rate": 9.496889580093313e-06, "loss": 0.1239, "step": 1295 }, { "epoch": 1.0, "learning_rate": 9.496500777604978e-06, "loss": 0.1852, "step": 1296 }, { "epoch": 1.0, "learning_rate": 9.496111975116642e-06, "loss": 0.1171, "step": 1297 }, { "epoch": 1.0, "learning_rate": 9.495723172628306e-06, "loss": 0.0385, "step": 1298 }, { "epoch": 1.0, "learning_rate": 9.495334370139969e-06, "loss": 0.1769, "step": 1299 }, { "epoch": 1.0, "learning_rate": 9.494945567651633e-06, "loss": 0.3138, "step": 1300 }, { "epoch": 1.0, "learning_rate": 9.494556765163298e-06, "loss": 0.038, "step": 1301 }, { "epoch": 1.0, "learning_rate": 9.494167962674962e-06, "loss": 0.0699, "step": 1302 }, { "epoch": 1.0, "learning_rate": 9.493779160186625e-06, "loss": 0.1214, "step": 1303 }, { "epoch": 1.0, "learning_rate": 9.49339035769829e-06, "loss": 0.3679, "step": 1304 }, { "epoch": 1.0, "learning_rate": 9.493001555209954e-06, "loss": 0.3804, "step": 1305 }, { "epoch": 1.0, "learning_rate": 9.492612752721618e-06, "loss": 0.0508, "step": 1306 }, { "epoch": 1.0, "learning_rate": 9.492223950233283e-06, "loss": 0.1821, "step": 1307 }, { "epoch": 1.0, "learning_rate": 9.491835147744947e-06, "loss": 0.3268, "step": 1308 }, { "epoch": 1.0, "learning_rate": 9.49144634525661e-06, "loss": 0.0351, "step": 1309 }, { "epoch": 1.0, "learning_rate": 9.491057542768274e-06, "loss": 0.1384, "step": 1310 }, { "epoch": 1.0, "learning_rate": 9.49066874027994e-06, "loss": 0.2318, "step": 1311 }, { "epoch": 1.0, "learning_rate": 9.490279937791601e-06, "loss": 0.2263, "step": 1312 }, { "epoch": 1.0, "learning_rate": 9.489891135303267e-06, "loss": 0.254, "step": 1313 }, { "epoch": 1.0, "learning_rate": 9.48950233281493e-06, "loss": 0.2162, "step": 1314 }, { "epoch": 1.0, "learning_rate": 9.489113530326594e-06, "loss": 0.1666, "step": 1315 }, { "epoch": 1.0, "learning_rate": 9.48872472783826e-06, "loss": 0.0572, "step": 1316 }, { "epoch": 1.0, "learning_rate": 9.488335925349923e-06, "loss": 0.0378, "step": 1317 }, { "epoch": 1.0, "learning_rate": 9.487947122861587e-06, "loss": 0.0107, "step": 1318 }, { "epoch": 1.0, "learning_rate": 9.487558320373252e-06, "loss": 0.2524, "step": 1319 }, { "epoch": 1.0, "learning_rate": 9.487169517884915e-06, "loss": 0.3704, "step": 1320 }, { "epoch": 1.0, "learning_rate": 9.486780715396579e-06, "loss": 0.3886, "step": 1321 }, { "epoch": 1.0, "learning_rate": 9.486391912908244e-06, "loss": 0.227, "step": 1322 }, { "epoch": 1.0, "learning_rate": 9.486003110419908e-06, "loss": 0.1881, "step": 1323 }, { "epoch": 1.0, "learning_rate": 9.485614307931572e-06, "loss": 0.5063, "step": 1324 }, { "epoch": 1.0, "learning_rate": 9.485225505443235e-06, "loss": 0.1228, "step": 1325 }, { "epoch": 1.0, "learning_rate": 9.484836702954899e-06, "loss": 0.4521, "step": 1326 }, { "epoch": 1.0, "learning_rate": 9.484447900466564e-06, "loss": 0.0652, "step": 1327 }, { "epoch": 1.0, "learning_rate": 9.484059097978228e-06, "loss": 0.1983, "step": 1328 }, { "epoch": 1.0, "learning_rate": 9.483670295489892e-06, "loss": 0.2743, "step": 1329 }, { "epoch": 1.0, "learning_rate": 9.483281493001555e-06, "loss": 0.1696, "step": 1330 }, { "epoch": 1.0, "learning_rate": 9.48289269051322e-06, "loss": 0.0819, "step": 1331 }, { "epoch": 1.0, "learning_rate": 9.482503888024884e-06, "loss": 0.0758, "step": 1332 }, { "epoch": 1.0, "learning_rate": 9.482115085536548e-06, "loss": 0.1698, "step": 1333 }, { "epoch": 1.0, "learning_rate": 9.481726283048213e-06, "loss": 0.0392, "step": 1334 }, { "epoch": 1.0, "learning_rate": 9.481337480559877e-06, "loss": 0.1879, "step": 1335 }, { "epoch": 1.0, "learning_rate": 9.48094867807154e-06, "loss": 0.3429, "step": 1336 }, { "epoch": 1.0, "learning_rate": 9.480559875583204e-06, "loss": 0.0849, "step": 1337 }, { "epoch": 1.0, "learning_rate": 9.480171073094868e-06, "loss": 0.3589, "step": 1338 }, { "epoch": 1.0, "learning_rate": 9.479782270606533e-06, "loss": 0.7134, "step": 1339 }, { "epoch": 1.0, "learning_rate": 9.479393468118197e-06, "loss": 0.7085, "step": 1340 }, { "epoch": 1.0, "learning_rate": 9.47900466562986e-06, "loss": 0.4835, "step": 1341 }, { "epoch": 1.0, "learning_rate": 9.478615863141525e-06, "loss": 0.2878, "step": 1342 }, { "epoch": 1.0, "learning_rate": 9.478227060653189e-06, "loss": 0.0828, "step": 1343 }, { "epoch": 1.0, "learning_rate": 9.477838258164853e-06, "loss": 0.1302, "step": 1344 }, { "epoch": 1.0, "learning_rate": 9.477449455676518e-06, "loss": 0.3051, "step": 1345 }, { "epoch": 1.0, "learning_rate": 9.477060653188182e-06, "loss": 0.0677, "step": 1346 }, { "epoch": 1.0, "learning_rate": 9.476671850699845e-06, "loss": 0.1132, "step": 1347 }, { "epoch": 1.0, "learning_rate": 9.47628304821151e-06, "loss": 0.1261, "step": 1348 }, { "epoch": 1.0, "learning_rate": 9.475894245723174e-06, "loss": 0.1946, "step": 1349 }, { "epoch": 1.0, "learning_rate": 9.475505443234836e-06, "loss": 0.0732, "step": 1350 }, { "epoch": 1.0, "learning_rate": 9.475116640746502e-06, "loss": 0.4509, "step": 1351 }, { "epoch": 1.0, "learning_rate": 9.474727838258165e-06, "loss": 0.0282, "step": 1352 }, { "epoch": 1.0, "learning_rate": 9.474339035769829e-06, "loss": 0.0286, "step": 1353 }, { "epoch": 1.0, "learning_rate": 9.473950233281494e-06, "loss": 0.3231, "step": 1354 }, { "epoch": 1.0, "learning_rate": 9.473561430793158e-06, "loss": 0.0394, "step": 1355 }, { "epoch": 1.0, "learning_rate": 9.473172628304821e-06, "loss": 0.1106, "step": 1356 }, { "epoch": 1.0, "learning_rate": 9.472783825816487e-06, "loss": 0.3595, "step": 1357 }, { "epoch": 1.0, "learning_rate": 9.47239502332815e-06, "loss": 0.3928, "step": 1358 }, { "epoch": 1.0, "learning_rate": 9.472006220839814e-06, "loss": 0.1939, "step": 1359 }, { "epoch": 1.0, "learning_rate": 9.47161741835148e-06, "loss": 0.0491, "step": 1360 }, { "epoch": 1.0, "learning_rate": 9.471228615863143e-06, "loss": 0.237, "step": 1361 }, { "epoch": 1.0, "learning_rate": 9.470839813374806e-06, "loss": 0.0638, "step": 1362 }, { "epoch": 1.0, "learning_rate": 9.47045101088647e-06, "loss": 0.1608, "step": 1363 }, { "epoch": 1.0, "learning_rate": 9.470062208398134e-06, "loss": 0.2538, "step": 1364 }, { "epoch": 1.0, "learning_rate": 9.469673405909799e-06, "loss": 0.0312, "step": 1365 }, { "epoch": 1.0, "learning_rate": 9.469284603421463e-06, "loss": 0.0623, "step": 1366 }, { "epoch": 1.0, "learning_rate": 9.468895800933126e-06, "loss": 0.0418, "step": 1367 }, { "epoch": 1.0, "learning_rate": 9.468506998444792e-06, "loss": 0.0119, "step": 1368 }, { "epoch": 1.0, "learning_rate": 9.468118195956455e-06, "loss": 0.0492, "step": 1369 }, { "epoch": 1.0, "learning_rate": 9.467729393468119e-06, "loss": 0.1452, "step": 1370 }, { "epoch": 1.0, "learning_rate": 9.467340590979783e-06, "loss": 0.1062, "step": 1371 }, { "epoch": 1.0, "learning_rate": 9.466951788491448e-06, "loss": 0.3965, "step": 1372 }, { "epoch": 1.0, "learning_rate": 9.466562986003111e-06, "loss": 0.1641, "step": 1373 }, { "epoch": 1.0, "learning_rate": 9.466174183514775e-06, "loss": 0.1109, "step": 1374 }, { "epoch": 1.0, "learning_rate": 9.465785381026439e-06, "loss": 0.1648, "step": 1375 }, { "epoch": 1.0, "learning_rate": 9.465396578538102e-06, "loss": 0.021, "step": 1376 }, { "epoch": 1.0, "learning_rate": 9.465007776049768e-06, "loss": 0.0466, "step": 1377 }, { "epoch": 1.0, "learning_rate": 9.464618973561431e-06, "loss": 0.0371, "step": 1378 }, { "epoch": 1.0, "learning_rate": 9.464230171073095e-06, "loss": 0.1536, "step": 1379 }, { "epoch": 1.0, "learning_rate": 9.46384136858476e-06, "loss": 0.0146, "step": 1380 }, { "epoch": 1.0, "learning_rate": 9.463452566096424e-06, "loss": 0.1456, "step": 1381 }, { "epoch": 1.0, "learning_rate": 9.463063763608088e-06, "loss": 0.2029, "step": 1382 }, { "epoch": 1.0, "learning_rate": 9.462674961119753e-06, "loss": 0.1092, "step": 1383 }, { "epoch": 1.0, "learning_rate": 9.462286158631416e-06, "loss": 0.3602, "step": 1384 }, { "epoch": 1.0, "learning_rate": 9.46189735614308e-06, "loss": 0.4914, "step": 1385 }, { "epoch": 1.0, "learning_rate": 9.461508553654745e-06, "loss": 0.0421, "step": 1386 }, { "epoch": 1.0, "learning_rate": 9.461119751166407e-06, "loss": 0.3205, "step": 1387 }, { "epoch": 1.0, "learning_rate": 9.460730948678073e-06, "loss": 0.4944, "step": 1388 }, { "epoch": 1.0, "learning_rate": 9.460342146189736e-06, "loss": 0.2796, "step": 1389 }, { "epoch": 1.0, "learning_rate": 9.4599533437014e-06, "loss": 0.157, "step": 1390 }, { "epoch": 1.0, "learning_rate": 9.459564541213064e-06, "loss": 0.1229, "step": 1391 }, { "epoch": 1.0, "learning_rate": 9.459175738724729e-06, "loss": 0.3618, "step": 1392 }, { "epoch": 1.0, "learning_rate": 9.458786936236393e-06, "loss": 0.1857, "step": 1393 }, { "epoch": 1.0, "learning_rate": 9.458398133748056e-06, "loss": 0.2176, "step": 1394 }, { "epoch": 1.0, "learning_rate": 9.458009331259721e-06, "loss": 0.0598, "step": 1395 }, { "epoch": 1.0, "learning_rate": 9.457620528771385e-06, "loss": 0.7116, "step": 1396 }, { "epoch": 1.0, "learning_rate": 9.457231726283049e-06, "loss": 0.0597, "step": 1397 }, { "epoch": 1.0, "learning_rate": 9.456842923794714e-06, "loss": 0.3643, "step": 1398 }, { "epoch": 1.0, "learning_rate": 9.456454121306378e-06, "loss": 0.0155, "step": 1399 }, { "epoch": 1.0, "learning_rate": 9.456065318818041e-06, "loss": 0.0302, "step": 1400 }, { "epoch": 1.0, "learning_rate": 9.455676516329705e-06, "loss": 0.127, "step": 1401 }, { "epoch": 1.0, "learning_rate": 9.455287713841369e-06, "loss": 0.1886, "step": 1402 }, { "epoch": 1.0, "learning_rate": 9.454898911353034e-06, "loss": 0.2174, "step": 1403 }, { "epoch": 1.0, "learning_rate": 9.454510108864697e-06, "loss": 0.1672, "step": 1404 }, { "epoch": 1.0, "learning_rate": 9.454121306376361e-06, "loss": 0.0635, "step": 1405 }, { "epoch": 1.0, "learning_rate": 9.453732503888026e-06, "loss": 0.2117, "step": 1406 }, { "epoch": 1.0, "learning_rate": 9.45334370139969e-06, "loss": 0.4005, "step": 1407 }, { "epoch": 1.0, "learning_rate": 9.452954898911354e-06, "loss": 0.1736, "step": 1408 }, { "epoch": 1.0, "learning_rate": 9.452566096423017e-06, "loss": 0.3427, "step": 1409 }, { "epoch": 1.0, "learning_rate": 9.452177293934683e-06, "loss": 0.0295, "step": 1410 }, { "epoch": 1.0, "learning_rate": 9.451788491446346e-06, "loss": 0.4953, "step": 1411 }, { "epoch": 1.0, "learning_rate": 9.45139968895801e-06, "loss": 0.0606, "step": 1412 }, { "epoch": 1.0, "learning_rate": 9.451010886469674e-06, "loss": 0.2012, "step": 1413 }, { "epoch": 1.0, "learning_rate": 9.450622083981337e-06, "loss": 0.16, "step": 1414 }, { "epoch": 1.0, "learning_rate": 9.450233281493002e-06, "loss": 0.0927, "step": 1415 }, { "epoch": 1.0, "learning_rate": 9.449844479004666e-06, "loss": 0.0215, "step": 1416 }, { "epoch": 1.01, "learning_rate": 9.44945567651633e-06, "loss": 0.2488, "step": 1417 }, { "epoch": 1.01, "learning_rate": 9.449066874027995e-06, "loss": 0.4219, "step": 1418 }, { "epoch": 1.01, "learning_rate": 9.448678071539659e-06, "loss": 0.0958, "step": 1419 }, { "epoch": 1.01, "learning_rate": 9.448289269051322e-06, "loss": 0.1126, "step": 1420 }, { "epoch": 1.01, "learning_rate": 9.447900466562988e-06, "loss": 0.0866, "step": 1421 }, { "epoch": 1.01, "learning_rate": 9.447511664074651e-06, "loss": 0.0643, "step": 1422 }, { "epoch": 1.01, "learning_rate": 9.447122861586315e-06, "loss": 0.0252, "step": 1423 }, { "epoch": 1.01, "learning_rate": 9.44673405909798e-06, "loss": 0.2337, "step": 1424 }, { "epoch": 1.01, "learning_rate": 9.446345256609642e-06, "loss": 0.6457, "step": 1425 }, { "epoch": 1.01, "learning_rate": 9.445956454121307e-06, "loss": 0.0566, "step": 1426 }, { "epoch": 1.01, "learning_rate": 9.445567651632971e-06, "loss": 0.351, "step": 1427 }, { "epoch": 1.01, "learning_rate": 9.445178849144635e-06, "loss": 0.4883, "step": 1428 }, { "epoch": 1.01, "learning_rate": 9.444790046656298e-06, "loss": 0.2493, "step": 1429 }, { "epoch": 1.01, "learning_rate": 9.444401244167964e-06, "loss": 0.1887, "step": 1430 }, { "epoch": 1.01, "learning_rate": 9.444012441679627e-06, "loss": 0.0695, "step": 1431 }, { "epoch": 1.01, "learning_rate": 9.443623639191291e-06, "loss": 0.2781, "step": 1432 }, { "epoch": 1.01, "learning_rate": 9.443234836702956e-06, "loss": 0.0297, "step": 1433 }, { "epoch": 1.01, "learning_rate": 9.44284603421462e-06, "loss": 0.4265, "step": 1434 }, { "epoch": 1.01, "learning_rate": 9.442457231726284e-06, "loss": 0.1619, "step": 1435 }, { "epoch": 1.01, "learning_rate": 9.442068429237949e-06, "loss": 0.1236, "step": 1436 }, { "epoch": 1.01, "learning_rate": 9.441679626749612e-06, "loss": 0.025, "step": 1437 }, { "epoch": 1.01, "learning_rate": 9.441290824261276e-06, "loss": 0.1226, "step": 1438 }, { "epoch": 1.01, "learning_rate": 9.44090202177294e-06, "loss": 0.1589, "step": 1439 }, { "epoch": 1.01, "learning_rate": 9.440513219284603e-06, "loss": 0.293, "step": 1440 }, { "epoch": 1.01, "learning_rate": 9.440124416796269e-06, "loss": 0.1914, "step": 1441 }, { "epoch": 1.01, "learning_rate": 9.439735614307932e-06, "loss": 0.1911, "step": 1442 }, { "epoch": 1.01, "learning_rate": 9.439346811819596e-06, "loss": 0.0499, "step": 1443 }, { "epoch": 1.01, "learning_rate": 9.438958009331261e-06, "loss": 0.1356, "step": 1444 }, { "epoch": 1.01, "learning_rate": 9.438569206842925e-06, "loss": 0.2486, "step": 1445 }, { "epoch": 1.01, "learning_rate": 9.438180404354589e-06, "loss": 0.0797, "step": 1446 }, { "epoch": 1.01, "learning_rate": 9.437791601866254e-06, "loss": 0.0349, "step": 1447 }, { "epoch": 1.01, "learning_rate": 9.437402799377917e-06, "loss": 0.0336, "step": 1448 }, { "epoch": 1.01, "learning_rate": 9.437013996889581e-06, "loss": 0.1931, "step": 1449 }, { "epoch": 1.01, "learning_rate": 9.436625194401245e-06, "loss": 0.0179, "step": 1450 }, { "epoch": 1.01, "learning_rate": 9.436236391912908e-06, "loss": 0.143, "step": 1451 }, { "epoch": 1.01, "learning_rate": 9.435847589424572e-06, "loss": 0.2398, "step": 1452 }, { "epoch": 1.01, "learning_rate": 9.435458786936237e-06, "loss": 0.3084, "step": 1453 }, { "epoch": 1.01, "learning_rate": 9.435069984447901e-06, "loss": 0.04, "step": 1454 }, { "epoch": 1.01, "learning_rate": 9.434681181959565e-06, "loss": 0.0188, "step": 1455 }, { "epoch": 1.01, "learning_rate": 9.43429237947123e-06, "loss": 0.1592, "step": 1456 }, { "epoch": 1.01, "learning_rate": 9.433903576982893e-06, "loss": 0.4175, "step": 1457 }, { "epoch": 1.01, "learning_rate": 9.433514774494557e-06, "loss": 0.1157, "step": 1458 }, { "epoch": 1.01, "learning_rate": 9.433125972006222e-06, "loss": 0.376, "step": 1459 }, { "epoch": 1.01, "learning_rate": 9.432737169517886e-06, "loss": 0.0513, "step": 1460 }, { "epoch": 1.01, "learning_rate": 9.43234836702955e-06, "loss": 0.3912, "step": 1461 }, { "epoch": 1.01, "learning_rate": 9.431959564541215e-06, "loss": 0.0139, "step": 1462 }, { "epoch": 1.01, "learning_rate": 9.431570762052877e-06, "loss": 0.0494, "step": 1463 }, { "epoch": 1.01, "learning_rate": 9.431181959564542e-06, "loss": 0.0096, "step": 1464 }, { "epoch": 1.01, "learning_rate": 9.430793157076206e-06, "loss": 0.4317, "step": 1465 }, { "epoch": 1.01, "learning_rate": 9.43040435458787e-06, "loss": 0.0517, "step": 1466 }, { "epoch": 1.01, "learning_rate": 9.430015552099535e-06, "loss": 0.1694, "step": 1467 }, { "epoch": 1.01, "learning_rate": 9.429626749611198e-06, "loss": 0.1497, "step": 1468 }, { "epoch": 1.01, "learning_rate": 9.429237947122862e-06, "loss": 0.4988, "step": 1469 }, { "epoch": 1.01, "learning_rate": 9.428849144634526e-06, "loss": 0.1226, "step": 1470 }, { "epoch": 1.01, "learning_rate": 9.428460342146191e-06, "loss": 0.0471, "step": 1471 }, { "epoch": 1.01, "learning_rate": 9.428071539657855e-06, "loss": 0.0843, "step": 1472 }, { "epoch": 1.01, "learning_rate": 9.427682737169518e-06, "loss": 0.3228, "step": 1473 }, { "epoch": 1.01, "learning_rate": 9.427293934681184e-06, "loss": 0.0273, "step": 1474 }, { "epoch": 1.01, "learning_rate": 9.426905132192847e-06, "loss": 0.4052, "step": 1475 }, { "epoch": 1.01, "learning_rate": 9.426516329704511e-06, "loss": 0.2288, "step": 1476 }, { "epoch": 1.01, "learning_rate": 9.426127527216175e-06, "loss": 0.6146, "step": 1477 }, { "epoch": 1.01, "learning_rate": 9.425738724727838e-06, "loss": 0.108, "step": 1478 }, { "epoch": 1.01, "learning_rate": 9.425349922239503e-06, "loss": 0.2364, "step": 1479 }, { "epoch": 1.01, "learning_rate": 9.424961119751167e-06, "loss": 0.0896, "step": 1480 }, { "epoch": 1.01, "learning_rate": 9.42457231726283e-06, "loss": 0.2409, "step": 1481 }, { "epoch": 1.01, "learning_rate": 9.424183514774496e-06, "loss": 0.5078, "step": 1482 }, { "epoch": 1.01, "learning_rate": 9.42379471228616e-06, "loss": 0.1722, "step": 1483 }, { "epoch": 1.01, "learning_rate": 9.423405909797823e-06, "loss": 0.1389, "step": 1484 }, { "epoch": 1.01, "learning_rate": 9.423017107309489e-06, "loss": 0.1323, "step": 1485 }, { "epoch": 1.01, "learning_rate": 9.422628304821152e-06, "loss": 0.3259, "step": 1486 }, { "epoch": 1.01, "learning_rate": 9.422239502332816e-06, "loss": 0.1138, "step": 1487 }, { "epoch": 1.01, "learning_rate": 9.42185069984448e-06, "loss": 0.1349, "step": 1488 }, { "epoch": 1.01, "learning_rate": 9.421461897356143e-06, "loss": 0.3092, "step": 1489 }, { "epoch": 1.01, "learning_rate": 9.421073094867807e-06, "loss": 0.3889, "step": 1490 }, { "epoch": 1.01, "learning_rate": 9.420684292379472e-06, "loss": 0.1352, "step": 1491 }, { "epoch": 1.01, "learning_rate": 9.420295489891136e-06, "loss": 0.1062, "step": 1492 }, { "epoch": 1.01, "learning_rate": 9.4199066874028e-06, "loss": 0.0362, "step": 1493 }, { "epoch": 1.01, "learning_rate": 9.419517884914465e-06, "loss": 0.0523, "step": 1494 }, { "epoch": 1.01, "learning_rate": 9.419129082426128e-06, "loss": 0.2051, "step": 1495 }, { "epoch": 1.01, "learning_rate": 9.418740279937792e-06, "loss": 0.0482, "step": 1496 }, { "epoch": 1.01, "learning_rate": 9.418351477449457e-06, "loss": 0.0924, "step": 1497 }, { "epoch": 1.01, "learning_rate": 9.417962674961121e-06, "loss": 0.0211, "step": 1498 }, { "epoch": 1.01, "learning_rate": 9.417573872472784e-06, "loss": 0.4273, "step": 1499 }, { "epoch": 1.01, "learning_rate": 9.41718506998445e-06, "loss": 0.0266, "step": 1500 }, { "epoch": 1.01, "learning_rate": 9.416796267496112e-06, "loss": 0.1192, "step": 1501 }, { "epoch": 1.01, "learning_rate": 9.416407465007777e-06, "loss": 0.0288, "step": 1502 }, { "epoch": 1.01, "learning_rate": 9.41601866251944e-06, "loss": 0.1388, "step": 1503 }, { "epoch": 1.01, "learning_rate": 9.415629860031104e-06, "loss": 0.333, "step": 1504 }, { "epoch": 1.01, "learning_rate": 9.41524105754277e-06, "loss": 0.1633, "step": 1505 }, { "epoch": 1.01, "learning_rate": 9.414852255054433e-06, "loss": 0.215, "step": 1506 }, { "epoch": 1.01, "learning_rate": 9.414463452566097e-06, "loss": 0.8097, "step": 1507 }, { "epoch": 1.01, "learning_rate": 9.41407465007776e-06, "loss": 0.071, "step": 1508 }, { "epoch": 1.01, "learning_rate": 9.413685847589426e-06, "loss": 0.0615, "step": 1509 }, { "epoch": 1.01, "learning_rate": 9.41329704510109e-06, "loss": 0.0384, "step": 1510 }, { "epoch": 1.01, "learning_rate": 9.412908242612753e-06, "loss": 0.0381, "step": 1511 }, { "epoch": 1.01, "learning_rate": 9.412519440124418e-06, "loss": 0.2003, "step": 1512 }, { "epoch": 1.01, "learning_rate": 9.412130637636082e-06, "loss": 0.1979, "step": 1513 }, { "epoch": 1.01, "learning_rate": 9.411741835147746e-06, "loss": 0.2551, "step": 1514 }, { "epoch": 1.01, "learning_rate": 9.41135303265941e-06, "loss": 0.1195, "step": 1515 }, { "epoch": 1.01, "learning_rate": 9.410964230171073e-06, "loss": 0.0376, "step": 1516 }, { "epoch": 1.01, "learning_rate": 9.410575427682738e-06, "loss": 0.0066, "step": 1517 }, { "epoch": 1.01, "learning_rate": 9.410186625194402e-06, "loss": 0.0614, "step": 1518 }, { "epoch": 1.01, "learning_rate": 9.409797822706066e-06, "loss": 0.0196, "step": 1519 }, { "epoch": 1.01, "learning_rate": 9.40940902021773e-06, "loss": 0.3958, "step": 1520 }, { "epoch": 1.01, "learning_rate": 9.409020217729394e-06, "loss": 0.0281, "step": 1521 }, { "epoch": 1.01, "learning_rate": 9.408631415241058e-06, "loss": 0.0313, "step": 1522 }, { "epoch": 1.01, "learning_rate": 9.408242612752723e-06, "loss": 0.315, "step": 1523 }, { "epoch": 1.01, "learning_rate": 9.407853810264387e-06, "loss": 0.0464, "step": 1524 }, { "epoch": 1.01, "learning_rate": 9.40746500777605e-06, "loss": 0.3318, "step": 1525 }, { "epoch": 1.01, "learning_rate": 9.407076205287714e-06, "loss": 0.4158, "step": 1526 }, { "epoch": 1.01, "learning_rate": 9.406687402799378e-06, "loss": 0.3579, "step": 1527 }, { "epoch": 1.01, "learning_rate": 9.406298600311042e-06, "loss": 0.414, "step": 1528 }, { "epoch": 1.01, "learning_rate": 9.405909797822707e-06, "loss": 0.3017, "step": 1529 }, { "epoch": 1.01, "learning_rate": 9.40552099533437e-06, "loss": 0.1845, "step": 1530 }, { "epoch": 1.01, "learning_rate": 9.405132192846034e-06, "loss": 0.1648, "step": 1531 }, { "epoch": 1.01, "learning_rate": 9.4047433903577e-06, "loss": 0.0415, "step": 1532 }, { "epoch": 1.01, "learning_rate": 9.404354587869363e-06, "loss": 0.0508, "step": 1533 }, { "epoch": 1.01, "learning_rate": 9.403965785381027e-06, "loss": 0.1402, "step": 1534 }, { "epoch": 1.01, "learning_rate": 9.403576982892692e-06, "loss": 0.0419, "step": 1535 }, { "epoch": 1.01, "learning_rate": 9.403188180404356e-06, "loss": 0.132, "step": 1536 }, { "epoch": 1.01, "learning_rate": 9.40279937791602e-06, "loss": 0.074, "step": 1537 }, { "epoch": 1.01, "learning_rate": 9.402410575427685e-06, "loss": 0.0753, "step": 1538 }, { "epoch": 1.01, "learning_rate": 9.402021772939347e-06, "loss": 0.109, "step": 1539 }, { "epoch": 1.01, "learning_rate": 9.401632970451012e-06, "loss": 0.5405, "step": 1540 }, { "epoch": 1.01, "learning_rate": 9.401244167962675e-06, "loss": 0.5226, "step": 1541 }, { "epoch": 1.01, "learning_rate": 9.400855365474339e-06, "loss": 0.2252, "step": 1542 }, { "epoch": 1.01, "learning_rate": 9.400466562986004e-06, "loss": 0.2047, "step": 1543 }, { "epoch": 1.01, "learning_rate": 9.400077760497668e-06, "loss": 0.1423, "step": 1544 }, { "epoch": 1.01, "learning_rate": 9.399688958009332e-06, "loss": 0.1723, "step": 1545 }, { "epoch": 1.01, "learning_rate": 9.399300155520995e-06, "loss": 0.12, "step": 1546 }, { "epoch": 1.01, "learning_rate": 9.39891135303266e-06, "loss": 0.0659, "step": 1547 }, { "epoch": 1.01, "learning_rate": 9.398522550544324e-06, "loss": 0.364, "step": 1548 }, { "epoch": 1.01, "learning_rate": 9.398133748055988e-06, "loss": 0.3239, "step": 1549 }, { "epoch": 1.01, "learning_rate": 9.397744945567653e-06, "loss": 0.1835, "step": 1550 }, { "epoch": 1.01, "learning_rate": 9.397356143079317e-06, "loss": 0.0241, "step": 1551 }, { "epoch": 1.01, "learning_rate": 9.39696734059098e-06, "loss": 0.2064, "step": 1552 }, { "epoch": 1.01, "learning_rate": 9.396578538102644e-06, "loss": 0.3437, "step": 1553 }, { "epoch": 1.01, "learning_rate": 9.396189735614308e-06, "loss": 0.1416, "step": 1554 }, { "epoch": 1.01, "learning_rate": 9.395800933125973e-06, "loss": 0.0844, "step": 1555 }, { "epoch": 1.01, "learning_rate": 9.395412130637637e-06, "loss": 0.0631, "step": 1556 }, { "epoch": 1.01, "learning_rate": 9.3950233281493e-06, "loss": 0.0734, "step": 1557 }, { "epoch": 1.01, "learning_rate": 9.394634525660966e-06, "loss": 0.2028, "step": 1558 }, { "epoch": 1.01, "learning_rate": 9.39424572317263e-06, "loss": 0.281, "step": 1559 }, { "epoch": 1.01, "learning_rate": 9.393856920684293e-06, "loss": 0.0462, "step": 1560 }, { "epoch": 1.01, "learning_rate": 9.393468118195958e-06, "loss": 0.0189, "step": 1561 }, { "epoch": 1.01, "learning_rate": 9.393079315707622e-06, "loss": 0.1664, "step": 1562 }, { "epoch": 1.01, "learning_rate": 9.392690513219285e-06, "loss": 0.1428, "step": 1563 }, { "epoch": 1.01, "learning_rate": 9.39230171073095e-06, "loss": 0.233, "step": 1564 }, { "epoch": 1.01, "learning_rate": 9.391912908242613e-06, "loss": 0.2665, "step": 1565 }, { "epoch": 1.01, "learning_rate": 9.391524105754276e-06, "loss": 0.5003, "step": 1566 }, { "epoch": 1.01, "learning_rate": 9.391135303265942e-06, "loss": 0.1092, "step": 1567 }, { "epoch": 1.01, "learning_rate": 9.390746500777605e-06, "loss": 0.6797, "step": 1568 }, { "epoch": 1.01, "learning_rate": 9.390357698289269e-06, "loss": 0.2455, "step": 1569 }, { "epoch": 1.01, "learning_rate": 9.389968895800934e-06, "loss": 0.2659, "step": 1570 }, { "epoch": 1.01, "learning_rate": 9.389580093312598e-06, "loss": 0.0306, "step": 1571 }, { "epoch": 1.01, "learning_rate": 9.389191290824262e-06, "loss": 0.0724, "step": 1572 }, { "epoch": 1.01, "learning_rate": 9.388802488335927e-06, "loss": 0.1052, "step": 1573 }, { "epoch": 1.01, "learning_rate": 9.38841368584759e-06, "loss": 0.5527, "step": 1574 }, { "epoch": 1.01, "learning_rate": 9.388024883359254e-06, "loss": 0.0537, "step": 1575 }, { "epoch": 1.01, "learning_rate": 9.38763608087092e-06, "loss": 0.2395, "step": 1576 }, { "epoch": 1.01, "learning_rate": 9.387247278382581e-06, "loss": 0.0695, "step": 1577 }, { "epoch": 1.01, "learning_rate": 9.386858475894247e-06, "loss": 0.3492, "step": 1578 }, { "epoch": 1.01, "learning_rate": 9.38646967340591e-06, "loss": 0.2284, "step": 1579 }, { "epoch": 1.01, "learning_rate": 9.386080870917574e-06, "loss": 0.0283, "step": 1580 }, { "epoch": 1.01, "learning_rate": 9.38569206842924e-06, "loss": 0.0236, "step": 1581 }, { "epoch": 1.01, "learning_rate": 9.385303265940903e-06, "loss": 0.3261, "step": 1582 }, { "epoch": 1.01, "learning_rate": 9.384914463452567e-06, "loss": 0.3218, "step": 1583 }, { "epoch": 1.01, "learning_rate": 9.384525660964232e-06, "loss": 0.4604, "step": 1584 }, { "epoch": 1.01, "learning_rate": 9.384136858475895e-06, "loss": 0.0628, "step": 1585 }, { "epoch": 1.01, "learning_rate": 9.383748055987559e-06, "loss": 0.2295, "step": 1586 }, { "epoch": 1.01, "learning_rate": 9.383359253499223e-06, "loss": 0.1658, "step": 1587 }, { "epoch": 1.01, "learning_rate": 9.382970451010888e-06, "loss": 0.1807, "step": 1588 }, { "epoch": 1.01, "learning_rate": 9.382581648522552e-06, "loss": 0.2142, "step": 1589 }, { "epoch": 1.01, "learning_rate": 9.382192846034215e-06, "loss": 0.0767, "step": 1590 }, { "epoch": 1.01, "learning_rate": 9.381804043545879e-06, "loss": 0.0576, "step": 1591 }, { "epoch": 1.01, "learning_rate": 9.381415241057543e-06, "loss": 0.0204, "step": 1592 }, { "epoch": 1.01, "learning_rate": 9.381026438569208e-06, "loss": 0.2992, "step": 1593 }, { "epoch": 1.01, "learning_rate": 9.380637636080871e-06, "loss": 0.0425, "step": 1594 }, { "epoch": 1.01, "learning_rate": 9.380248833592535e-06, "loss": 0.1695, "step": 1595 }, { "epoch": 1.01, "learning_rate": 9.3798600311042e-06, "loss": 0.3405, "step": 1596 }, { "epoch": 1.01, "learning_rate": 9.379471228615864e-06, "loss": 0.2162, "step": 1597 }, { "epoch": 1.01, "learning_rate": 9.379082426127528e-06, "loss": 0.2889, "step": 1598 }, { "epoch": 1.01, "learning_rate": 9.378693623639193e-06, "loss": 0.4753, "step": 1599 }, { "epoch": 1.01, "learning_rate": 9.378304821150857e-06, "loss": 0.0432, "step": 1600 }, { "epoch": 1.01, "learning_rate": 9.37791601866252e-06, "loss": 0.2025, "step": 1601 }, { "epoch": 1.01, "learning_rate": 9.377527216174186e-06, "loss": 0.2727, "step": 1602 }, { "epoch": 1.01, "learning_rate": 9.377138413685848e-06, "loss": 0.4965, "step": 1603 }, { "epoch": 1.01, "learning_rate": 9.376749611197513e-06, "loss": 0.0449, "step": 1604 }, { "epoch": 1.01, "learning_rate": 9.376360808709176e-06, "loss": 0.496, "step": 1605 }, { "epoch": 1.01, "learning_rate": 9.37597200622084e-06, "loss": 0.0283, "step": 1606 }, { "epoch": 1.01, "learning_rate": 9.375583203732504e-06, "loss": 0.0692, "step": 1607 }, { "epoch": 1.01, "learning_rate": 9.375194401244169e-06, "loss": 0.3561, "step": 1608 }, { "epoch": 1.01, "learning_rate": 9.374805598755833e-06, "loss": 0.234, "step": 1609 }, { "epoch": 1.01, "learning_rate": 9.374416796267496e-06, "loss": 0.0319, "step": 1610 }, { "epoch": 1.01, "learning_rate": 9.374027993779162e-06, "loss": 0.2625, "step": 1611 }, { "epoch": 1.01, "learning_rate": 9.373639191290825e-06, "loss": 0.2874, "step": 1612 }, { "epoch": 1.01, "learning_rate": 9.373250388802489e-06, "loss": 0.1666, "step": 1613 }, { "epoch": 1.01, "learning_rate": 9.372861586314154e-06, "loss": 0.2988, "step": 1614 }, { "epoch": 1.01, "learning_rate": 9.372472783825816e-06, "loss": 0.1319, "step": 1615 }, { "epoch": 1.01, "learning_rate": 9.372083981337481e-06, "loss": 0.2732, "step": 1616 }, { "epoch": 1.01, "learning_rate": 9.371695178849145e-06, "loss": 0.152, "step": 1617 }, { "epoch": 1.01, "learning_rate": 9.371306376360809e-06, "loss": 0.2733, "step": 1618 }, { "epoch": 1.01, "learning_rate": 9.370917573872474e-06, "loss": 0.2458, "step": 1619 }, { "epoch": 1.01, "learning_rate": 9.370528771384138e-06, "loss": 0.075, "step": 1620 }, { "epoch": 1.01, "learning_rate": 9.370139968895801e-06, "loss": 0.1346, "step": 1621 }, { "epoch": 1.01, "learning_rate": 9.369751166407467e-06, "loss": 0.3822, "step": 1622 }, { "epoch": 1.01, "learning_rate": 9.36936236391913e-06, "loss": 0.2462, "step": 1623 }, { "epoch": 1.01, "learning_rate": 9.368973561430794e-06, "loss": 0.112, "step": 1624 }, { "epoch": 1.01, "learning_rate": 9.368584758942458e-06, "loss": 0.0788, "step": 1625 }, { "epoch": 1.01, "learning_rate": 9.368195956454123e-06, "loss": 0.4486, "step": 1626 }, { "epoch": 1.01, "learning_rate": 9.367807153965786e-06, "loss": 0.4051, "step": 1627 }, { "epoch": 1.01, "learning_rate": 9.36741835147745e-06, "loss": 0.0637, "step": 1628 }, { "epoch": 1.01, "learning_rate": 9.367029548989114e-06, "loss": 0.0502, "step": 1629 }, { "epoch": 1.01, "learning_rate": 9.366640746500777e-06, "loss": 0.4564, "step": 1630 }, { "epoch": 1.01, "learning_rate": 9.366251944012443e-06, "loss": 0.0546, "step": 1631 }, { "epoch": 1.01, "learning_rate": 9.365863141524106e-06, "loss": 0.1365, "step": 1632 }, { "epoch": 1.01, "learning_rate": 9.36547433903577e-06, "loss": 0.0925, "step": 1633 }, { "epoch": 1.01, "learning_rate": 9.365085536547435e-06, "loss": 0.2342, "step": 1634 }, { "epoch": 1.01, "learning_rate": 9.364696734059099e-06, "loss": 0.1431, "step": 1635 }, { "epoch": 1.01, "learning_rate": 9.364307931570762e-06, "loss": 0.2959, "step": 1636 }, { "epoch": 1.01, "learning_rate": 9.363919129082428e-06, "loss": 0.135, "step": 1637 }, { "epoch": 1.01, "learning_rate": 9.363530326594091e-06, "loss": 0.1158, "step": 1638 }, { "epoch": 1.01, "learning_rate": 9.363141524105755e-06, "loss": 0.1799, "step": 1639 }, { "epoch": 1.01, "learning_rate": 9.36275272161742e-06, "loss": 0.0616, "step": 1640 }, { "epoch": 1.01, "learning_rate": 9.362363919129082e-06, "loss": 0.2731, "step": 1641 }, { "epoch": 1.01, "learning_rate": 9.361975116640748e-06, "loss": 0.2046, "step": 1642 }, { "epoch": 1.01, "learning_rate": 9.361586314152411e-06, "loss": 0.1113, "step": 1643 }, { "epoch": 1.01, "learning_rate": 9.361197511664075e-06, "loss": 0.238, "step": 1644 }, { "epoch": 1.01, "learning_rate": 9.360808709175739e-06, "loss": 0.1511, "step": 1645 }, { "epoch": 1.01, "learning_rate": 9.360419906687404e-06, "loss": 0.2855, "step": 1646 }, { "epoch": 1.01, "learning_rate": 9.360031104199067e-06, "loss": 0.034, "step": 1647 }, { "epoch": 1.01, "learning_rate": 9.359642301710731e-06, "loss": 0.2088, "step": 1648 }, { "epoch": 1.01, "learning_rate": 9.359253499222396e-06, "loss": 0.508, "step": 1649 }, { "epoch": 1.01, "learning_rate": 9.35886469673406e-06, "loss": 0.0721, "step": 1650 }, { "epoch": 1.01, "learning_rate": 9.358475894245724e-06, "loss": 0.2424, "step": 1651 }, { "epoch": 1.01, "learning_rate": 9.358087091757389e-06, "loss": 0.0332, "step": 1652 }, { "epoch": 1.01, "learning_rate": 9.357698289269051e-06, "loss": 0.1447, "step": 1653 }, { "epoch": 1.01, "learning_rate": 9.357309486780716e-06, "loss": 0.1305, "step": 1654 }, { "epoch": 1.01, "learning_rate": 9.35692068429238e-06, "loss": 0.1609, "step": 1655 }, { "epoch": 1.01, "learning_rate": 9.356531881804044e-06, "loss": 0.2838, "step": 1656 }, { "epoch": 1.01, "learning_rate": 9.356143079315709e-06, "loss": 0.3214, "step": 1657 }, { "epoch": 1.01, "learning_rate": 9.355754276827372e-06, "loss": 0.1251, "step": 1658 }, { "epoch": 1.01, "learning_rate": 9.355365474339036e-06, "loss": 0.1979, "step": 1659 }, { "epoch": 1.01, "learning_rate": 9.354976671850701e-06, "loss": 0.2186, "step": 1660 }, { "epoch": 1.01, "learning_rate": 9.354587869362365e-06, "loss": 0.2569, "step": 1661 }, { "epoch": 1.01, "learning_rate": 9.354199066874029e-06, "loss": 0.0857, "step": 1662 }, { "epoch": 1.01, "learning_rate": 9.353810264385694e-06, "loss": 0.3073, "step": 1663 }, { "epoch": 1.01, "learning_rate": 9.353421461897358e-06, "loss": 0.1655, "step": 1664 }, { "epoch": 1.01, "learning_rate": 9.353032659409021e-06, "loss": 0.1851, "step": 1665 }, { "epoch": 1.01, "learning_rate": 9.352643856920685e-06, "loss": 0.0477, "step": 1666 }, { "epoch": 1.01, "learning_rate": 9.352255054432349e-06, "loss": 0.0523, "step": 1667 }, { "epoch": 1.01, "learning_rate": 9.351866251944012e-06, "loss": 0.3695, "step": 1668 }, { "epoch": 1.01, "learning_rate": 9.351477449455677e-06, "loss": 0.482, "step": 1669 }, { "epoch": 1.01, "learning_rate": 9.351088646967341e-06, "loss": 0.3429, "step": 1670 }, { "epoch": 1.01, "learning_rate": 9.350699844479005e-06, "loss": 0.0359, "step": 1671 }, { "epoch": 1.01, "learning_rate": 9.35031104199067e-06, "loss": 0.3392, "step": 1672 }, { "epoch": 1.01, "learning_rate": 9.349922239502334e-06, "loss": 0.1464, "step": 1673 }, { "epoch": 1.02, "learning_rate": 9.349533437013997e-06, "loss": 0.2492, "step": 1674 }, { "epoch": 1.02, "learning_rate": 9.349144634525663e-06, "loss": 0.0684, "step": 1675 }, { "epoch": 1.02, "learning_rate": 9.348755832037326e-06, "loss": 0.0652, "step": 1676 }, { "epoch": 1.02, "learning_rate": 9.34836702954899e-06, "loss": 0.0408, "step": 1677 }, { "epoch": 1.02, "learning_rate": 9.347978227060655e-06, "loss": 0.0652, "step": 1678 }, { "epoch": 1.02, "learning_rate": 9.347589424572317e-06, "loss": 0.2294, "step": 1679 }, { "epoch": 1.02, "learning_rate": 9.347200622083982e-06, "loss": 0.0991, "step": 1680 }, { "epoch": 1.02, "learning_rate": 9.346811819595646e-06, "loss": 0.1708, "step": 1681 }, { "epoch": 1.02, "learning_rate": 9.34642301710731e-06, "loss": 0.4075, "step": 1682 }, { "epoch": 1.02, "learning_rate": 9.346034214618975e-06, "loss": 0.0275, "step": 1683 }, { "epoch": 1.02, "learning_rate": 9.345645412130639e-06, "loss": 0.1534, "step": 1684 }, { "epoch": 1.02, "learning_rate": 9.345256609642302e-06, "loss": 0.2642, "step": 1685 }, { "epoch": 1.02, "learning_rate": 9.344867807153966e-06, "loss": 0.0696, "step": 1686 }, { "epoch": 1.02, "learning_rate": 9.344479004665631e-06, "loss": 0.0376, "step": 1687 }, { "epoch": 1.02, "learning_rate": 9.344090202177295e-06, "loss": 0.0184, "step": 1688 }, { "epoch": 1.02, "learning_rate": 9.343701399688958e-06, "loss": 0.0149, "step": 1689 }, { "epoch": 1.02, "learning_rate": 9.343312597200624e-06, "loss": 0.0547, "step": 1690 }, { "epoch": 1.02, "learning_rate": 9.342923794712286e-06, "loss": 0.3754, "step": 1691 }, { "epoch": 1.02, "learning_rate": 9.342534992223951e-06, "loss": 0.1052, "step": 1692 }, { "epoch": 1.02, "learning_rate": 9.342146189735615e-06, "loss": 0.5078, "step": 1693 }, { "epoch": 1.02, "learning_rate": 9.341757387247278e-06, "loss": 0.2297, "step": 1694 }, { "epoch": 1.02, "learning_rate": 9.341368584758944e-06, "loss": 0.1717, "step": 1695 }, { "epoch": 1.02, "learning_rate": 9.340979782270607e-06, "loss": 0.2892, "step": 1696 }, { "epoch": 1.02, "learning_rate": 9.340590979782271e-06, "loss": 0.208, "step": 1697 }, { "epoch": 1.02, "learning_rate": 9.340202177293936e-06, "loss": 0.2257, "step": 1698 }, { "epoch": 1.02, "learning_rate": 9.3398133748056e-06, "loss": 0.0205, "step": 1699 }, { "epoch": 1.02, "learning_rate": 9.339424572317263e-06, "loss": 0.4235, "step": 1700 }, { "epoch": 1.02, "learning_rate": 9.339035769828929e-06, "loss": 0.0562, "step": 1701 }, { "epoch": 1.02, "learning_rate": 9.338646967340592e-06, "loss": 0.2608, "step": 1702 }, { "epoch": 1.02, "learning_rate": 9.338258164852256e-06, "loss": 0.2003, "step": 1703 }, { "epoch": 1.02, "learning_rate": 9.33786936236392e-06, "loss": 0.1279, "step": 1704 }, { "epoch": 1.02, "learning_rate": 9.337480559875583e-06, "loss": 0.1773, "step": 1705 }, { "epoch": 1.02, "learning_rate": 9.337091757387247e-06, "loss": 0.0917, "step": 1706 }, { "epoch": 1.02, "learning_rate": 9.336702954898912e-06, "loss": 0.0367, "step": 1707 }, { "epoch": 1.02, "learning_rate": 9.336314152410576e-06, "loss": 0.181, "step": 1708 }, { "epoch": 1.02, "learning_rate": 9.33592534992224e-06, "loss": 0.0881, "step": 1709 }, { "epoch": 1.02, "learning_rate": 9.335536547433905e-06, "loss": 0.246, "step": 1710 }, { "epoch": 1.02, "learning_rate": 9.335147744945568e-06, "loss": 0.0467, "step": 1711 }, { "epoch": 1.02, "learning_rate": 9.334758942457232e-06, "loss": 0.0352, "step": 1712 }, { "epoch": 1.02, "learning_rate": 9.334370139968897e-06, "loss": 0.0347, "step": 1713 }, { "epoch": 1.02, "learning_rate": 9.333981337480561e-06, "loss": 0.4142, "step": 1714 }, { "epoch": 1.02, "learning_rate": 9.333592534992225e-06, "loss": 0.1047, "step": 1715 }, { "epoch": 1.02, "learning_rate": 9.33320373250389e-06, "loss": 0.4396, "step": 1716 }, { "epoch": 1.02, "learning_rate": 9.332814930015552e-06, "loss": 0.0469, "step": 1717 }, { "epoch": 1.02, "learning_rate": 9.332426127527217e-06, "loss": 0.0579, "step": 1718 }, { "epoch": 1.02, "learning_rate": 9.332037325038881e-06, "loss": 0.2603, "step": 1719 }, { "epoch": 1.02, "learning_rate": 9.331648522550545e-06, "loss": 0.0306, "step": 1720 }, { "epoch": 1.02, "learning_rate": 9.33125972006221e-06, "loss": 0.2927, "step": 1721 }, { "epoch": 1.02, "learning_rate": 9.330870917573873e-06, "loss": 0.0125, "step": 1722 }, { "epoch": 1.02, "learning_rate": 9.330482115085537e-06, "loss": 0.114, "step": 1723 }, { "epoch": 1.02, "learning_rate": 9.3300933125972e-06, "loss": 0.3716, "step": 1724 }, { "epoch": 1.02, "learning_rate": 9.329704510108866e-06, "loss": 0.0893, "step": 1725 }, { "epoch": 1.02, "learning_rate": 9.32931570762053e-06, "loss": 0.0263, "step": 1726 }, { "epoch": 1.02, "learning_rate": 9.328926905132193e-06, "loss": 0.4877, "step": 1727 }, { "epoch": 1.02, "learning_rate": 9.328538102643859e-06, "loss": 0.1352, "step": 1728 }, { "epoch": 1.02, "learning_rate": 9.32814930015552e-06, "loss": 0.038, "step": 1729 }, { "epoch": 1.02, "learning_rate": 9.327760497667186e-06, "loss": 0.1589, "step": 1730 }, { "epoch": 1.02, "learning_rate": 9.32737169517885e-06, "loss": 0.3767, "step": 1731 }, { "epoch": 1.02, "learning_rate": 9.326982892690513e-06, "loss": 0.4261, "step": 1732 }, { "epoch": 1.02, "learning_rate": 9.326594090202178e-06, "loss": 0.0114, "step": 1733 }, { "epoch": 1.02, "learning_rate": 9.326205287713842e-06, "loss": 0.1184, "step": 1734 }, { "epoch": 1.02, "learning_rate": 9.325816485225506e-06, "loss": 0.3191, "step": 1735 }, { "epoch": 1.02, "learning_rate": 9.325427682737171e-06, "loss": 0.4599, "step": 1736 }, { "epoch": 1.02, "learning_rate": 9.325038880248835e-06, "loss": 0.4027, "step": 1737 }, { "epoch": 1.02, "learning_rate": 9.324650077760498e-06, "loss": 0.519, "step": 1738 }, { "epoch": 1.02, "learning_rate": 9.324261275272164e-06, "loss": 0.0996, "step": 1739 }, { "epoch": 1.02, "learning_rate": 9.323872472783827e-06, "loss": 0.7154, "step": 1740 }, { "epoch": 1.02, "learning_rate": 9.323483670295491e-06, "loss": 0.3166, "step": 1741 }, { "epoch": 1.02, "learning_rate": 9.323094867807154e-06, "loss": 0.4782, "step": 1742 }, { "epoch": 1.02, "learning_rate": 9.322706065318818e-06, "loss": 0.3677, "step": 1743 }, { "epoch": 1.02, "learning_rate": 9.322317262830482e-06, "loss": 0.0091, "step": 1744 }, { "epoch": 1.02, "learning_rate": 9.321928460342147e-06, "loss": 0.1036, "step": 1745 }, { "epoch": 1.02, "learning_rate": 9.32153965785381e-06, "loss": 0.1165, "step": 1746 }, { "epoch": 1.02, "learning_rate": 9.321150855365474e-06, "loss": 0.3944, "step": 1747 }, { "epoch": 1.02, "learning_rate": 9.32076205287714e-06, "loss": 0.2841, "step": 1748 }, { "epoch": 1.02, "learning_rate": 9.320373250388803e-06, "loss": 0.1014, "step": 1749 }, { "epoch": 1.02, "learning_rate": 9.319984447900467e-06, "loss": 0.2997, "step": 1750 }, { "epoch": 1.02, "learning_rate": 9.319595645412132e-06, "loss": 0.1841, "step": 1751 }, { "epoch": 1.02, "learning_rate": 9.319206842923796e-06, "loss": 0.0999, "step": 1752 }, { "epoch": 1.02, "learning_rate": 9.31881804043546e-06, "loss": 0.237, "step": 1753 }, { "epoch": 1.02, "learning_rate": 9.318429237947125e-06, "loss": 0.3468, "step": 1754 }, { "epoch": 1.02, "learning_rate": 9.318040435458787e-06, "loss": 0.2008, "step": 1755 }, { "epoch": 1.02, "learning_rate": 9.317651632970452e-06, "loss": 0.0731, "step": 1756 }, { "epoch": 1.02, "learning_rate": 9.317262830482116e-06, "loss": 0.2346, "step": 1757 }, { "epoch": 1.02, "learning_rate": 9.31687402799378e-06, "loss": 0.2998, "step": 1758 }, { "epoch": 1.02, "learning_rate": 9.316485225505445e-06, "loss": 0.1179, "step": 1759 }, { "epoch": 1.02, "learning_rate": 9.316096423017108e-06, "loss": 0.0699, "step": 1760 }, { "epoch": 1.02, "learning_rate": 9.315707620528772e-06, "loss": 0.0258, "step": 1761 }, { "epoch": 1.02, "learning_rate": 9.315318818040437e-06, "loss": 0.1239, "step": 1762 }, { "epoch": 1.02, "learning_rate": 9.3149300155521e-06, "loss": 0.0473, "step": 1763 }, { "epoch": 1.02, "learning_rate": 9.314541213063764e-06, "loss": 0.1858, "step": 1764 }, { "epoch": 1.02, "learning_rate": 9.314152410575428e-06, "loss": 0.2126, "step": 1765 }, { "epoch": 1.02, "learning_rate": 9.313763608087093e-06, "loss": 0.038, "step": 1766 }, { "epoch": 1.02, "learning_rate": 9.313374805598755e-06, "loss": 0.0887, "step": 1767 }, { "epoch": 1.02, "learning_rate": 9.31298600311042e-06, "loss": 0.0711, "step": 1768 }, { "epoch": 1.02, "learning_rate": 9.312597200622084e-06, "loss": 0.0849, "step": 1769 }, { "epoch": 1.02, "learning_rate": 9.312208398133748e-06, "loss": 0.2492, "step": 1770 }, { "epoch": 1.02, "learning_rate": 9.311819595645413e-06, "loss": 0.2369, "step": 1771 }, { "epoch": 1.02, "learning_rate": 9.311430793157077e-06, "loss": 0.1984, "step": 1772 }, { "epoch": 1.02, "learning_rate": 9.31104199066874e-06, "loss": 0.09, "step": 1773 }, { "epoch": 1.02, "learning_rate": 9.310653188180406e-06, "loss": 0.1006, "step": 1774 }, { "epoch": 1.02, "learning_rate": 9.31026438569207e-06, "loss": 0.2268, "step": 1775 }, { "epoch": 1.02, "learning_rate": 9.309875583203733e-06, "loss": 0.0488, "step": 1776 }, { "epoch": 1.02, "learning_rate": 9.309486780715398e-06, "loss": 0.1512, "step": 1777 }, { "epoch": 1.02, "learning_rate": 9.309097978227062e-06, "loss": 0.15, "step": 1778 }, { "epoch": 1.02, "learning_rate": 9.308709175738726e-06, "loss": 0.0414, "step": 1779 }, { "epoch": 1.02, "learning_rate": 9.30832037325039e-06, "loss": 0.2899, "step": 1780 }, { "epoch": 1.02, "learning_rate": 9.307931570762053e-06, "loss": 0.1295, "step": 1781 }, { "epoch": 1.02, "learning_rate": 9.307542768273718e-06, "loss": 0.0115, "step": 1782 }, { "epoch": 1.02, "learning_rate": 9.307153965785382e-06, "loss": 0.0522, "step": 1783 }, { "epoch": 1.02, "learning_rate": 9.306765163297045e-06, "loss": 0.3618, "step": 1784 }, { "epoch": 1.02, "learning_rate": 9.306376360808709e-06, "loss": 0.1552, "step": 1785 }, { "epoch": 1.02, "learning_rate": 9.305987558320374e-06, "loss": 0.081, "step": 1786 }, { "epoch": 1.02, "learning_rate": 9.305598755832038e-06, "loss": 0.0814, "step": 1787 }, { "epoch": 1.02, "learning_rate": 9.305209953343702e-06, "loss": 0.2485, "step": 1788 }, { "epoch": 1.02, "learning_rate": 9.304821150855367e-06, "loss": 0.0618, "step": 1789 }, { "epoch": 1.02, "learning_rate": 9.30443234836703e-06, "loss": 0.3847, "step": 1790 }, { "epoch": 1.02, "learning_rate": 9.304043545878694e-06, "loss": 0.0891, "step": 1791 }, { "epoch": 1.02, "learning_rate": 9.30365474339036e-06, "loss": 0.1733, "step": 1792 }, { "epoch": 1.02, "learning_rate": 9.303265940902022e-06, "loss": 0.0623, "step": 1793 }, { "epoch": 1.02, "learning_rate": 9.302877138413687e-06, "loss": 0.4657, "step": 1794 }, { "epoch": 1.02, "learning_rate": 9.30248833592535e-06, "loss": 0.2565, "step": 1795 }, { "epoch": 1.02, "learning_rate": 9.302099533437014e-06, "loss": 0.1637, "step": 1796 }, { "epoch": 1.02, "learning_rate": 9.30171073094868e-06, "loss": 0.1733, "step": 1797 }, { "epoch": 1.02, "learning_rate": 9.301321928460343e-06, "loss": 0.1951, "step": 1798 }, { "epoch": 1.02, "learning_rate": 9.300933125972007e-06, "loss": 0.2244, "step": 1799 }, { "epoch": 1.02, "learning_rate": 9.300544323483672e-06, "loss": 0.21, "step": 1800 }, { "epoch": 1.02, "learning_rate": 9.300155520995336e-06, "loss": 0.0983, "step": 1801 }, { "epoch": 1.02, "learning_rate": 9.299766718507e-06, "loss": 0.0688, "step": 1802 }, { "epoch": 1.02, "learning_rate": 9.299377916018663e-06, "loss": 0.2419, "step": 1803 }, { "epoch": 1.02, "learning_rate": 9.298989113530328e-06, "loss": 0.1777, "step": 1804 }, { "epoch": 1.02, "learning_rate": 9.29860031104199e-06, "loss": 0.2685, "step": 1805 }, { "epoch": 1.02, "learning_rate": 9.298211508553655e-06, "loss": 0.0173, "step": 1806 }, { "epoch": 1.02, "learning_rate": 9.297822706065319e-06, "loss": 0.3807, "step": 1807 }, { "epoch": 1.02, "learning_rate": 9.297433903576983e-06, "loss": 0.0412, "step": 1808 }, { "epoch": 1.02, "learning_rate": 9.297045101088648e-06, "loss": 0.1516, "step": 1809 }, { "epoch": 1.02, "learning_rate": 9.296656298600312e-06, "loss": 0.2473, "step": 1810 }, { "epoch": 1.02, "learning_rate": 9.296267496111975e-06, "loss": 0.0899, "step": 1811 }, { "epoch": 1.02, "learning_rate": 9.29587869362364e-06, "loss": 0.1035, "step": 1812 }, { "epoch": 1.02, "learning_rate": 9.295489891135304e-06, "loss": 0.0091, "step": 1813 }, { "epoch": 1.02, "learning_rate": 9.295101088646968e-06, "loss": 0.4296, "step": 1814 }, { "epoch": 1.02, "learning_rate": 9.294712286158633e-06, "loss": 0.0354, "step": 1815 }, { "epoch": 1.02, "learning_rate": 9.294323483670297e-06, "loss": 0.1235, "step": 1816 }, { "epoch": 1.02, "learning_rate": 9.29393468118196e-06, "loss": 0.0463, "step": 1817 }, { "epoch": 1.02, "learning_rate": 9.293545878693624e-06, "loss": 0.0314, "step": 1818 }, { "epoch": 1.02, "learning_rate": 9.293157076205288e-06, "loss": 1.0577, "step": 1819 }, { "epoch": 1.02, "learning_rate": 9.292768273716953e-06, "loss": 0.0752, "step": 1820 }, { "epoch": 1.02, "learning_rate": 9.292379471228617e-06, "loss": 0.1536, "step": 1821 }, { "epoch": 1.02, "learning_rate": 9.29199066874028e-06, "loss": 0.493, "step": 1822 }, { "epoch": 1.02, "learning_rate": 9.291601866251944e-06, "loss": 0.0763, "step": 1823 }, { "epoch": 1.02, "learning_rate": 9.29121306376361e-06, "loss": 0.0154, "step": 1824 }, { "epoch": 1.02, "learning_rate": 9.290824261275273e-06, "loss": 0.3542, "step": 1825 }, { "epoch": 1.02, "learning_rate": 9.290435458786936e-06, "loss": 0.1882, "step": 1826 }, { "epoch": 1.02, "learning_rate": 9.290046656298602e-06, "loss": 0.3221, "step": 1827 }, { "epoch": 1.02, "learning_rate": 9.289657853810265e-06, "loss": 0.1174, "step": 1828 }, { "epoch": 1.02, "learning_rate": 9.289269051321929e-06, "loss": 0.0925, "step": 1829 }, { "epoch": 1.02, "learning_rate": 9.288880248833594e-06, "loss": 0.1875, "step": 1830 }, { "epoch": 1.02, "learning_rate": 9.288491446345256e-06, "loss": 0.594, "step": 1831 }, { "epoch": 1.02, "learning_rate": 9.288102643856922e-06, "loss": 0.2103, "step": 1832 }, { "epoch": 1.02, "learning_rate": 9.287713841368585e-06, "loss": 0.0082, "step": 1833 }, { "epoch": 1.02, "learning_rate": 9.287325038880249e-06, "loss": 0.0363, "step": 1834 }, { "epoch": 1.02, "learning_rate": 9.286936236391914e-06, "loss": 0.2852, "step": 1835 }, { "epoch": 1.02, "learning_rate": 9.286547433903578e-06, "loss": 0.0755, "step": 1836 }, { "epoch": 1.02, "learning_rate": 9.286158631415241e-06, "loss": 0.0583, "step": 1837 }, { "epoch": 1.02, "learning_rate": 9.285769828926907e-06, "loss": 0.069, "step": 1838 }, { "epoch": 1.02, "learning_rate": 9.28538102643857e-06, "loss": 0.0166, "step": 1839 }, { "epoch": 1.02, "learning_rate": 9.284992223950234e-06, "loss": 0.1092, "step": 1840 }, { "epoch": 1.02, "learning_rate": 9.284603421461898e-06, "loss": 0.4382, "step": 1841 }, { "epoch": 1.02, "learning_rate": 9.284214618973563e-06, "loss": 0.0366, "step": 1842 }, { "epoch": 1.02, "learning_rate": 9.283825816485225e-06, "loss": 0.0122, "step": 1843 }, { "epoch": 1.02, "learning_rate": 9.28343701399689e-06, "loss": 0.4246, "step": 1844 }, { "epoch": 1.02, "learning_rate": 9.283048211508554e-06, "loss": 0.3089, "step": 1845 }, { "epoch": 1.02, "learning_rate": 9.282659409020218e-06, "loss": 0.1896, "step": 1846 }, { "epoch": 1.02, "learning_rate": 9.282270606531883e-06, "loss": 0.0588, "step": 1847 }, { "epoch": 1.02, "learning_rate": 9.281881804043546e-06, "loss": 0.0223, "step": 1848 }, { "epoch": 1.02, "learning_rate": 9.28149300155521e-06, "loss": 0.102, "step": 1849 }, { "epoch": 1.02, "learning_rate": 9.281104199066875e-06, "loss": 0.0696, "step": 1850 }, { "epoch": 1.02, "learning_rate": 9.280715396578539e-06, "loss": 0.0223, "step": 1851 }, { "epoch": 1.02, "learning_rate": 9.280326594090203e-06, "loss": 0.5061, "step": 1852 }, { "epoch": 1.02, "learning_rate": 9.279937791601868e-06, "loss": 0.0495, "step": 1853 }, { "epoch": 1.02, "learning_rate": 9.279548989113532e-06, "loss": 0.0628, "step": 1854 }, { "epoch": 1.02, "learning_rate": 9.279160186625195e-06, "loss": 0.158, "step": 1855 }, { "epoch": 1.02, "learning_rate": 9.278771384136859e-06, "loss": 0.0364, "step": 1856 }, { "epoch": 1.02, "learning_rate": 9.278382581648523e-06, "loss": 0.0303, "step": 1857 }, { "epoch": 1.02, "learning_rate": 9.277993779160188e-06, "loss": 0.3238, "step": 1858 }, { "epoch": 1.02, "learning_rate": 9.277604976671851e-06, "loss": 0.0563, "step": 1859 }, { "epoch": 1.02, "learning_rate": 9.277216174183515e-06, "loss": 0.4929, "step": 1860 }, { "epoch": 1.02, "learning_rate": 9.276827371695179e-06, "loss": 0.5223, "step": 1861 }, { "epoch": 1.02, "learning_rate": 9.276438569206844e-06, "loss": 0.0953, "step": 1862 }, { "epoch": 1.02, "learning_rate": 9.276049766718508e-06, "loss": 0.2854, "step": 1863 }, { "epoch": 1.02, "learning_rate": 9.275660964230171e-06, "loss": 0.0588, "step": 1864 }, { "epoch": 1.02, "learning_rate": 9.275272161741837e-06, "loss": 0.0248, "step": 1865 }, { "epoch": 1.02, "learning_rate": 9.2748833592535e-06, "loss": 0.1818, "step": 1866 }, { "epoch": 1.02, "learning_rate": 9.274494556765164e-06, "loss": 0.2607, "step": 1867 }, { "epoch": 1.02, "learning_rate": 9.274105754276827e-06, "loss": 0.0374, "step": 1868 }, { "epoch": 1.02, "learning_rate": 9.273716951788491e-06, "loss": 0.0858, "step": 1869 }, { "epoch": 1.02, "learning_rate": 9.273328149300156e-06, "loss": 0.1695, "step": 1870 }, { "epoch": 1.02, "learning_rate": 9.27293934681182e-06, "loss": 0.331, "step": 1871 }, { "epoch": 1.02, "learning_rate": 9.272550544323484e-06, "loss": 0.0869, "step": 1872 }, { "epoch": 1.02, "learning_rate": 9.272161741835149e-06, "loss": 0.0869, "step": 1873 }, { "epoch": 1.02, "learning_rate": 9.271772939346813e-06, "loss": 0.0306, "step": 1874 }, { "epoch": 1.02, "learning_rate": 9.271384136858476e-06, "loss": 0.3155, "step": 1875 }, { "epoch": 1.02, "learning_rate": 9.270995334370142e-06, "loss": 0.012, "step": 1876 }, { "epoch": 1.02, "learning_rate": 9.270606531881805e-06, "loss": 0.1138, "step": 1877 }, { "epoch": 1.02, "learning_rate": 9.270217729393469e-06, "loss": 0.2024, "step": 1878 }, { "epoch": 1.02, "learning_rate": 9.269828926905134e-06, "loss": 0.041, "step": 1879 }, { "epoch": 1.02, "learning_rate": 9.269440124416798e-06, "loss": 0.1643, "step": 1880 }, { "epoch": 1.02, "learning_rate": 9.26905132192846e-06, "loss": 0.2612, "step": 1881 }, { "epoch": 1.02, "learning_rate": 9.268662519440125e-06, "loss": 0.0491, "step": 1882 }, { "epoch": 1.02, "learning_rate": 9.268273716951789e-06, "loss": 0.0852, "step": 1883 }, { "epoch": 1.02, "learning_rate": 9.267884914463452e-06, "loss": 0.1133, "step": 1884 }, { "epoch": 1.02, "learning_rate": 9.267496111975118e-06, "loss": 0.136, "step": 1885 }, { "epoch": 1.02, "learning_rate": 9.267107309486781e-06, "loss": 0.0918, "step": 1886 }, { "epoch": 1.02, "learning_rate": 9.266718506998445e-06, "loss": 0.3722, "step": 1887 }, { "epoch": 1.02, "learning_rate": 9.26632970451011e-06, "loss": 0.1999, "step": 1888 }, { "epoch": 1.02, "learning_rate": 9.265940902021774e-06, "loss": 0.1822, "step": 1889 }, { "epoch": 1.02, "learning_rate": 9.265552099533437e-06, "loss": 0.1402, "step": 1890 }, { "epoch": 1.02, "learning_rate": 9.265163297045103e-06, "loss": 0.1795, "step": 1891 }, { "epoch": 1.02, "learning_rate": 9.264774494556766e-06, "loss": 0.203, "step": 1892 }, { "epoch": 1.02, "learning_rate": 9.26438569206843e-06, "loss": 0.0735, "step": 1893 }, { "epoch": 1.02, "learning_rate": 9.263996889580094e-06, "loss": 0.2483, "step": 1894 }, { "epoch": 1.02, "learning_rate": 9.263608087091757e-06, "loss": 0.4336, "step": 1895 }, { "epoch": 1.02, "learning_rate": 9.263219284603423e-06, "loss": 0.022, "step": 1896 }, { "epoch": 1.02, "learning_rate": 9.262830482115086e-06, "loss": 0.0861, "step": 1897 }, { "epoch": 1.02, "learning_rate": 9.26244167962675e-06, "loss": 0.3558, "step": 1898 }, { "epoch": 1.02, "learning_rate": 9.262052877138415e-06, "loss": 0.1461, "step": 1899 }, { "epoch": 1.02, "learning_rate": 9.261664074650079e-06, "loss": 0.1346, "step": 1900 }, { "epoch": 1.02, "learning_rate": 9.261275272161742e-06, "loss": 0.1671, "step": 1901 }, { "epoch": 1.02, "learning_rate": 9.260886469673406e-06, "loss": 0.0688, "step": 1902 }, { "epoch": 1.02, "learning_rate": 9.260497667185071e-06, "loss": 0.0976, "step": 1903 }, { "epoch": 1.02, "learning_rate": 9.260108864696735e-06, "loss": 0.2788, "step": 1904 }, { "epoch": 1.02, "learning_rate": 9.259720062208399e-06, "loss": 0.0376, "step": 1905 }, { "epoch": 1.02, "learning_rate": 9.259331259720062e-06, "loss": 0.035, "step": 1906 }, { "epoch": 1.02, "learning_rate": 9.258942457231726e-06, "loss": 0.331, "step": 1907 }, { "epoch": 1.02, "learning_rate": 9.258553654743391e-06, "loss": 0.1505, "step": 1908 }, { "epoch": 1.02, "learning_rate": 9.258164852255055e-06, "loss": 0.1914, "step": 1909 }, { "epoch": 1.02, "learning_rate": 9.257776049766719e-06, "loss": 0.1604, "step": 1910 }, { "epoch": 1.02, "learning_rate": 9.257387247278384e-06, "loss": 0.0847, "step": 1911 }, { "epoch": 1.02, "learning_rate": 9.256998444790047e-06, "loss": 0.129, "step": 1912 }, { "epoch": 1.02, "learning_rate": 9.256609642301711e-06, "loss": 0.0206, "step": 1913 }, { "epoch": 1.02, "learning_rate": 9.256220839813376e-06, "loss": 0.0685, "step": 1914 }, { "epoch": 1.02, "learning_rate": 9.25583203732504e-06, "loss": 0.1589, "step": 1915 }, { "epoch": 1.02, "learning_rate": 9.255443234836704e-06, "loss": 0.0787, "step": 1916 }, { "epoch": 1.02, "learning_rate": 9.255054432348369e-06, "loss": 0.2236, "step": 1917 }, { "epoch": 1.02, "learning_rate": 9.254665629860033e-06, "loss": 0.0781, "step": 1918 }, { "epoch": 1.02, "learning_rate": 9.254276827371696e-06, "loss": 0.0858, "step": 1919 }, { "epoch": 1.02, "learning_rate": 9.25388802488336e-06, "loss": 0.1072, "step": 1920 }, { "epoch": 1.02, "learning_rate": 9.253499222395023e-06, "loss": 0.4599, "step": 1921 }, { "epoch": 1.02, "learning_rate": 9.253110419906687e-06, "loss": 0.7524, "step": 1922 }, { "epoch": 1.02, "learning_rate": 9.252721617418352e-06, "loss": 0.3459, "step": 1923 }, { "epoch": 1.02, "learning_rate": 9.252332814930016e-06, "loss": 0.2434, "step": 1924 }, { "epoch": 1.02, "learning_rate": 9.25194401244168e-06, "loss": 0.0659, "step": 1925 }, { "epoch": 1.02, "learning_rate": 9.251555209953345e-06, "loss": 0.221, "step": 1926 }, { "epoch": 1.02, "learning_rate": 9.251166407465009e-06, "loss": 0.0449, "step": 1927 }, { "epoch": 1.02, "learning_rate": 9.250777604976672e-06, "loss": 0.2406, "step": 1928 }, { "epoch": 1.02, "learning_rate": 9.250388802488338e-06, "loss": 0.2071, "step": 1929 }, { "epoch": 1.02, "learning_rate": 9.250000000000001e-06, "loss": 0.1128, "step": 1930 }, { "epoch": 1.02, "learning_rate": 9.249611197511665e-06, "loss": 0.1389, "step": 1931 }, { "epoch": 1.03, "learning_rate": 9.249222395023328e-06, "loss": 0.1408, "step": 1932 }, { "epoch": 1.03, "learning_rate": 9.248833592534992e-06, "loss": 0.3758, "step": 1933 }, { "epoch": 1.03, "learning_rate": 9.248444790046657e-06, "loss": 0.0983, "step": 1934 }, { "epoch": 1.03, "learning_rate": 9.248055987558321e-06, "loss": 0.2595, "step": 1935 }, { "epoch": 1.03, "learning_rate": 9.247667185069985e-06, "loss": 0.0713, "step": 1936 }, { "epoch": 1.03, "learning_rate": 9.24727838258165e-06, "loss": 0.1421, "step": 1937 }, { "epoch": 1.03, "learning_rate": 9.246889580093314e-06, "loss": 0.2434, "step": 1938 }, { "epoch": 1.03, "learning_rate": 9.246500777604977e-06, "loss": 0.1376, "step": 1939 }, { "epoch": 1.03, "learning_rate": 9.246111975116641e-06, "loss": 0.0645, "step": 1940 }, { "epoch": 1.03, "learning_rate": 9.245723172628306e-06, "loss": 0.018, "step": 1941 }, { "epoch": 1.03, "learning_rate": 9.24533437013997e-06, "loss": 0.125, "step": 1942 }, { "epoch": 1.03, "learning_rate": 9.244945567651633e-06, "loss": 0.1867, "step": 1943 }, { "epoch": 1.03, "learning_rate": 9.244556765163297e-06, "loss": 0.0265, "step": 1944 }, { "epoch": 1.03, "learning_rate": 9.24416796267496e-06, "loss": 0.3988, "step": 1945 }, { "epoch": 1.03, "learning_rate": 9.243779160186626e-06, "loss": 0.0181, "step": 1946 }, { "epoch": 1.03, "learning_rate": 9.24339035769829e-06, "loss": 0.3952, "step": 1947 }, { "epoch": 1.03, "learning_rate": 9.243001555209953e-06, "loss": 0.122, "step": 1948 }, { "epoch": 1.03, "learning_rate": 9.242612752721619e-06, "loss": 0.0358, "step": 1949 }, { "epoch": 1.03, "learning_rate": 9.242223950233282e-06, "loss": 0.4276, "step": 1950 }, { "epoch": 1.03, "learning_rate": 9.241835147744946e-06, "loss": 0.2127, "step": 1951 }, { "epoch": 1.03, "learning_rate": 9.241446345256611e-06, "loss": 0.0164, "step": 1952 }, { "epoch": 1.03, "learning_rate": 9.241057542768275e-06, "loss": 0.3519, "step": 1953 }, { "epoch": 1.03, "learning_rate": 9.240668740279938e-06, "loss": 0.0118, "step": 1954 }, { "epoch": 1.03, "learning_rate": 9.240279937791604e-06, "loss": 0.0298, "step": 1955 }, { "epoch": 1.03, "learning_rate": 9.239891135303267e-06, "loss": 0.1483, "step": 1956 }, { "epoch": 1.03, "learning_rate": 9.239502332814931e-06, "loss": 0.0644, "step": 1957 }, { "epoch": 1.03, "learning_rate": 9.239113530326595e-06, "loss": 0.1005, "step": 1958 }, { "epoch": 1.03, "learning_rate": 9.238724727838258e-06, "loss": 0.2816, "step": 1959 }, { "epoch": 1.03, "learning_rate": 9.238335925349922e-06, "loss": 0.1219, "step": 1960 }, { "epoch": 1.03, "learning_rate": 9.237947122861587e-06, "loss": 0.0553, "step": 1961 }, { "epoch": 1.03, "learning_rate": 9.237558320373251e-06, "loss": 0.2403, "step": 1962 }, { "epoch": 1.03, "learning_rate": 9.237169517884914e-06, "loss": 0.3062, "step": 1963 }, { "epoch": 1.03, "learning_rate": 9.23678071539658e-06, "loss": 0.3507, "step": 1964 }, { "epoch": 1.03, "learning_rate": 9.236391912908243e-06, "loss": 0.0337, "step": 1965 }, { "epoch": 1.03, "learning_rate": 9.236003110419907e-06, "loss": 0.2399, "step": 1966 }, { "epoch": 1.03, "learning_rate": 9.235614307931572e-06, "loss": 0.0822, "step": 1967 }, { "epoch": 1.03, "learning_rate": 9.235225505443236e-06, "loss": 0.1122, "step": 1968 }, { "epoch": 1.03, "learning_rate": 9.2348367029549e-06, "loss": 0.0744, "step": 1969 }, { "epoch": 1.03, "learning_rate": 9.234447900466563e-06, "loss": 0.1215, "step": 1970 }, { "epoch": 1.03, "learning_rate": 9.234059097978227e-06, "loss": 0.1941, "step": 1971 }, { "epoch": 1.03, "learning_rate": 9.233670295489892e-06, "loss": 0.0234, "step": 1972 }, { "epoch": 1.03, "learning_rate": 9.233281493001556e-06, "loss": 0.2682, "step": 1973 }, { "epoch": 1.03, "learning_rate": 9.23289269051322e-06, "loss": 0.2154, "step": 1974 }, { "epoch": 1.03, "learning_rate": 9.232503888024885e-06, "loss": 0.1301, "step": 1975 }, { "epoch": 1.03, "learning_rate": 9.232115085536548e-06, "loss": 0.3585, "step": 1976 }, { "epoch": 1.03, "learning_rate": 9.231726283048212e-06, "loss": 0.1588, "step": 1977 }, { "epoch": 1.03, "learning_rate": 9.231337480559877e-06, "loss": 0.4043, "step": 1978 }, { "epoch": 1.03, "learning_rate": 9.230948678071541e-06, "loss": 0.0914, "step": 1979 }, { "epoch": 1.03, "learning_rate": 9.230559875583205e-06, "loss": 0.3926, "step": 1980 }, { "epoch": 1.03, "learning_rate": 9.230171073094868e-06, "loss": 0.2245, "step": 1981 }, { "epoch": 1.03, "learning_rate": 9.229782270606532e-06, "loss": 0.1161, "step": 1982 }, { "epoch": 1.03, "learning_rate": 9.229393468118196e-06, "loss": 0.0992, "step": 1983 }, { "epoch": 1.03, "learning_rate": 9.22900466562986e-06, "loss": 0.0465, "step": 1984 }, { "epoch": 1.03, "learning_rate": 9.228615863141524e-06, "loss": 0.1598, "step": 1985 }, { "epoch": 1.03, "learning_rate": 9.228227060653188e-06, "loss": 0.0851, "step": 1986 }, { "epoch": 1.03, "learning_rate": 9.227838258164853e-06, "loss": 0.0194, "step": 1987 }, { "epoch": 1.03, "learning_rate": 9.227449455676517e-06, "loss": 0.2965, "step": 1988 }, { "epoch": 1.03, "learning_rate": 9.22706065318818e-06, "loss": 0.0487, "step": 1989 }, { "epoch": 1.03, "learning_rate": 9.226671850699846e-06, "loss": 0.1006, "step": 1990 }, { "epoch": 1.03, "learning_rate": 9.22628304821151e-06, "loss": 0.0382, "step": 1991 }, { "epoch": 1.03, "learning_rate": 9.225894245723173e-06, "loss": 0.0629, "step": 1992 }, { "epoch": 1.03, "learning_rate": 9.225505443234839e-06, "loss": 0.1536, "step": 1993 }, { "epoch": 1.03, "learning_rate": 9.225116640746502e-06, "loss": 0.137, "step": 1994 }, { "epoch": 1.03, "learning_rate": 9.224727838258166e-06, "loss": 0.0149, "step": 1995 }, { "epoch": 1.03, "learning_rate": 9.22433903576983e-06, "loss": 0.1402, "step": 1996 }, { "epoch": 1.03, "learning_rate": 9.223950233281493e-06, "loss": 0.0507, "step": 1997 }, { "epoch": 1.03, "learning_rate": 9.223561430793158e-06, "loss": 0.2039, "step": 1998 }, { "epoch": 1.03, "learning_rate": 9.223172628304822e-06, "loss": 0.2127, "step": 1999 }, { "epoch": 1.03, "learning_rate": 9.222783825816486e-06, "loss": 0.3252, "step": 2000 }, { "epoch": 1.03, "learning_rate": 9.22239502332815e-06, "loss": 0.2706, "step": 2001 }, { "epoch": 1.03, "learning_rate": 9.222006220839815e-06, "loss": 0.223, "step": 2002 }, { "epoch": 1.03, "learning_rate": 9.221617418351478e-06, "loss": 0.0846, "step": 2003 }, { "epoch": 1.03, "learning_rate": 9.221228615863142e-06, "loss": 0.2458, "step": 2004 }, { "epoch": 1.03, "learning_rate": 9.220839813374807e-06, "loss": 0.4346, "step": 2005 }, { "epoch": 1.03, "learning_rate": 9.22045101088647e-06, "loss": 0.03, "step": 2006 }, { "epoch": 1.03, "learning_rate": 9.220062208398134e-06, "loss": 0.3246, "step": 2007 }, { "epoch": 1.03, "learning_rate": 9.219673405909798e-06, "loss": 0.4976, "step": 2008 }, { "epoch": 1.03, "learning_rate": 9.219284603421462e-06, "loss": 0.1872, "step": 2009 }, { "epoch": 1.03, "learning_rate": 9.218895800933127e-06, "loss": 0.3931, "step": 2010 }, { "epoch": 1.03, "learning_rate": 9.21850699844479e-06, "loss": 0.0957, "step": 2011 }, { "epoch": 1.03, "learning_rate": 9.218118195956454e-06, "loss": 0.1713, "step": 2012 }, { "epoch": 1.03, "learning_rate": 9.21772939346812e-06, "loss": 0.0634, "step": 2013 }, { "epoch": 1.03, "learning_rate": 9.217340590979783e-06, "loss": 0.2035, "step": 2014 }, { "epoch": 1.03, "learning_rate": 9.216951788491447e-06, "loss": 0.2668, "step": 2015 }, { "epoch": 1.03, "learning_rate": 9.216562986003112e-06, "loss": 0.7613, "step": 2016 }, { "epoch": 1.03, "learning_rate": 9.216174183514776e-06, "loss": 0.4616, "step": 2017 }, { "epoch": 1.03, "learning_rate": 9.21578538102644e-06, "loss": 0.078, "step": 2018 }, { "epoch": 1.03, "learning_rate": 9.215396578538103e-06, "loss": 0.0678, "step": 2019 }, { "epoch": 1.03, "learning_rate": 9.215007776049767e-06, "loss": 0.6502, "step": 2020 }, { "epoch": 1.03, "learning_rate": 9.21461897356143e-06, "loss": 0.0135, "step": 2021 }, { "epoch": 1.03, "learning_rate": 9.214230171073096e-06, "loss": 0.4649, "step": 2022 }, { "epoch": 1.03, "learning_rate": 9.21384136858476e-06, "loss": 0.0711, "step": 2023 }, { "epoch": 1.03, "learning_rate": 9.213452566096423e-06, "loss": 0.2811, "step": 2024 }, { "epoch": 1.03, "learning_rate": 9.213063763608088e-06, "loss": 0.1241, "step": 2025 }, { "epoch": 1.03, "learning_rate": 9.212674961119752e-06, "loss": 0.2415, "step": 2026 }, { "epoch": 1.03, "learning_rate": 9.212286158631415e-06, "loss": 0.2204, "step": 2027 }, { "epoch": 1.03, "learning_rate": 9.21189735614308e-06, "loss": 0.093, "step": 2028 }, { "epoch": 1.03, "learning_rate": 9.211508553654744e-06, "loss": 0.0447, "step": 2029 }, { "epoch": 1.03, "learning_rate": 9.211119751166408e-06, "loss": 0.3865, "step": 2030 }, { "epoch": 1.03, "learning_rate": 9.210730948678073e-06, "loss": 0.3614, "step": 2031 }, { "epoch": 1.03, "learning_rate": 9.210342146189737e-06, "loss": 0.2758, "step": 2032 }, { "epoch": 1.03, "learning_rate": 9.2099533437014e-06, "loss": 0.0944, "step": 2033 }, { "epoch": 1.03, "learning_rate": 9.209564541213064e-06, "loss": 0.0527, "step": 2034 }, { "epoch": 1.03, "learning_rate": 9.209175738724728e-06, "loss": 0.2293, "step": 2035 }, { "epoch": 1.03, "learning_rate": 9.208786936236393e-06, "loss": 0.3074, "step": 2036 }, { "epoch": 1.03, "learning_rate": 9.208398133748057e-06, "loss": 0.1193, "step": 2037 }, { "epoch": 1.03, "learning_rate": 9.20800933125972e-06, "loss": 0.1131, "step": 2038 }, { "epoch": 1.03, "learning_rate": 9.207620528771384e-06, "loss": 0.0419, "step": 2039 }, { "epoch": 1.03, "learning_rate": 9.20723172628305e-06, "loss": 0.291, "step": 2040 }, { "epoch": 1.03, "learning_rate": 9.206842923794713e-06, "loss": 0.2059, "step": 2041 }, { "epoch": 1.03, "learning_rate": 9.206454121306377e-06, "loss": 0.012, "step": 2042 }, { "epoch": 1.03, "learning_rate": 9.206065318818042e-06, "loss": 0.3031, "step": 2043 }, { "epoch": 1.03, "learning_rate": 9.205676516329706e-06, "loss": 0.4171, "step": 2044 }, { "epoch": 1.03, "learning_rate": 9.20528771384137e-06, "loss": 0.0468, "step": 2045 }, { "epoch": 1.03, "learning_rate": 9.204898911353033e-06, "loss": 0.2005, "step": 2046 }, { "epoch": 1.03, "learning_rate": 9.204510108864697e-06, "loss": 0.1082, "step": 2047 }, { "epoch": 1.03, "learning_rate": 9.204121306376362e-06, "loss": 0.1517, "step": 2048 }, { "epoch": 1.03, "learning_rate": 9.203732503888025e-06, "loss": 0.033, "step": 2049 }, { "epoch": 1.03, "learning_rate": 9.203343701399689e-06, "loss": 0.4207, "step": 2050 }, { "epoch": 1.03, "learning_rate": 9.202954898911354e-06, "loss": 0.0208, "step": 2051 }, { "epoch": 1.03, "learning_rate": 9.202566096423018e-06, "loss": 0.0817, "step": 2052 }, { "epoch": 1.03, "learning_rate": 9.202177293934682e-06, "loss": 0.0306, "step": 2053 }, { "epoch": 1.03, "learning_rate": 9.201788491446347e-06, "loss": 0.1394, "step": 2054 }, { "epoch": 1.03, "learning_rate": 9.20139968895801e-06, "loss": 0.3997, "step": 2055 }, { "epoch": 1.03, "learning_rate": 9.201010886469674e-06, "loss": 0.0466, "step": 2056 }, { "epoch": 1.03, "learning_rate": 9.20062208398134e-06, "loss": 0.0584, "step": 2057 }, { "epoch": 1.03, "learning_rate": 9.200233281493001e-06, "loss": 0.0295, "step": 2058 }, { "epoch": 1.03, "learning_rate": 9.199844479004665e-06, "loss": 0.0714, "step": 2059 }, { "epoch": 1.03, "learning_rate": 9.19945567651633e-06, "loss": 0.2511, "step": 2060 }, { "epoch": 1.03, "learning_rate": 9.199066874027994e-06, "loss": 0.015, "step": 2061 }, { "epoch": 1.03, "learning_rate": 9.198678071539658e-06, "loss": 0.1993, "step": 2062 }, { "epoch": 1.03, "learning_rate": 9.198289269051323e-06, "loss": 0.1831, "step": 2063 }, { "epoch": 1.03, "learning_rate": 9.197900466562987e-06, "loss": 0.3363, "step": 2064 }, { "epoch": 1.03, "learning_rate": 9.19751166407465e-06, "loss": 0.0964, "step": 2065 }, { "epoch": 1.03, "learning_rate": 9.197122861586316e-06, "loss": 0.2925, "step": 2066 }, { "epoch": 1.03, "learning_rate": 9.19673405909798e-06, "loss": 0.1733, "step": 2067 }, { "epoch": 1.03, "learning_rate": 9.196345256609643e-06, "loss": 0.0919, "step": 2068 }, { "epoch": 1.03, "learning_rate": 9.195956454121308e-06, "loss": 0.4342, "step": 2069 }, { "epoch": 1.03, "learning_rate": 9.195567651632972e-06, "loss": 0.1799, "step": 2070 }, { "epoch": 1.03, "learning_rate": 9.195178849144635e-06, "loss": 0.1297, "step": 2071 }, { "epoch": 1.03, "learning_rate": 9.194790046656299e-06, "loss": 0.3064, "step": 2072 }, { "epoch": 1.03, "learning_rate": 9.194401244167963e-06, "loss": 0.0086, "step": 2073 }, { "epoch": 1.03, "learning_rate": 9.194012441679628e-06, "loss": 0.1347, "step": 2074 }, { "epoch": 1.03, "learning_rate": 9.193623639191292e-06, "loss": 0.2339, "step": 2075 }, { "epoch": 1.03, "learning_rate": 9.193234836702955e-06, "loss": 0.1733, "step": 2076 }, { "epoch": 1.03, "learning_rate": 9.19284603421462e-06, "loss": 0.2689, "step": 2077 }, { "epoch": 1.03, "learning_rate": 9.192457231726284e-06, "loss": 0.0287, "step": 2078 }, { "epoch": 1.03, "learning_rate": 9.192068429237948e-06, "loss": 0.0402, "step": 2079 }, { "epoch": 1.03, "learning_rate": 9.191679626749611e-06, "loss": 0.034, "step": 2080 }, { "epoch": 1.03, "learning_rate": 9.191290824261277e-06, "loss": 0.1295, "step": 2081 }, { "epoch": 1.03, "learning_rate": 9.19090202177294e-06, "loss": 0.0547, "step": 2082 }, { "epoch": 1.03, "learning_rate": 9.190513219284604e-06, "loss": 0.056, "step": 2083 }, { "epoch": 1.03, "learning_rate": 9.190124416796268e-06, "loss": 0.0188, "step": 2084 }, { "epoch": 1.03, "learning_rate": 9.189735614307931e-06, "loss": 0.2779, "step": 2085 }, { "epoch": 1.03, "learning_rate": 9.189346811819597e-06, "loss": 0.0263, "step": 2086 }, { "epoch": 1.03, "learning_rate": 9.18895800933126e-06, "loss": 0.0424, "step": 2087 }, { "epoch": 1.03, "learning_rate": 9.188569206842924e-06, "loss": 0.1946, "step": 2088 }, { "epoch": 1.03, "learning_rate": 9.18818040435459e-06, "loss": 0.1897, "step": 2089 }, { "epoch": 1.03, "learning_rate": 9.187791601866253e-06, "loss": 0.2076, "step": 2090 }, { "epoch": 1.03, "learning_rate": 9.187402799377916e-06, "loss": 0.5188, "step": 2091 }, { "epoch": 1.03, "learning_rate": 9.187013996889582e-06, "loss": 0.141, "step": 2092 }, { "epoch": 1.03, "learning_rate": 9.186625194401245e-06, "loss": 0.4416, "step": 2093 }, { "epoch": 1.03, "learning_rate": 9.186236391912909e-06, "loss": 0.1021, "step": 2094 }, { "epoch": 1.03, "learning_rate": 9.185847589424574e-06, "loss": 0.1199, "step": 2095 }, { "epoch": 1.03, "learning_rate": 9.185458786936236e-06, "loss": 0.1616, "step": 2096 }, { "epoch": 1.03, "learning_rate": 9.185069984447902e-06, "loss": 0.0374, "step": 2097 }, { "epoch": 1.03, "learning_rate": 9.184681181959565e-06, "loss": 0.0243, "step": 2098 }, { "epoch": 1.03, "learning_rate": 9.184292379471229e-06, "loss": 0.4664, "step": 2099 }, { "epoch": 1.03, "learning_rate": 9.183903576982892e-06, "loss": 0.0573, "step": 2100 }, { "epoch": 1.03, "learning_rate": 9.183514774494558e-06, "loss": 0.2916, "step": 2101 }, { "epoch": 1.03, "learning_rate": 9.183125972006221e-06, "loss": 0.1788, "step": 2102 }, { "epoch": 1.03, "learning_rate": 9.182737169517885e-06, "loss": 0.1074, "step": 2103 }, { "epoch": 1.03, "learning_rate": 9.18234836702955e-06, "loss": 0.0356, "step": 2104 }, { "epoch": 1.03, "learning_rate": 9.181959564541214e-06, "loss": 0.0285, "step": 2105 }, { "epoch": 1.03, "learning_rate": 9.181570762052878e-06, "loss": 0.2546, "step": 2106 }, { "epoch": 1.03, "learning_rate": 9.181181959564543e-06, "loss": 0.1834, "step": 2107 }, { "epoch": 1.03, "learning_rate": 9.180793157076207e-06, "loss": 0.0549, "step": 2108 }, { "epoch": 1.03, "learning_rate": 9.18040435458787e-06, "loss": 0.2625, "step": 2109 }, { "epoch": 1.03, "learning_rate": 9.180015552099534e-06, "loss": 0.0814, "step": 2110 }, { "epoch": 1.03, "learning_rate": 9.179626749611197e-06, "loss": 0.2796, "step": 2111 }, { "epoch": 1.03, "learning_rate": 9.179237947122863e-06, "loss": 0.6206, "step": 2112 }, { "epoch": 1.03, "learning_rate": 9.178849144634526e-06, "loss": 0.3182, "step": 2113 }, { "epoch": 1.03, "learning_rate": 9.17846034214619e-06, "loss": 0.1425, "step": 2114 }, { "epoch": 1.03, "learning_rate": 9.178071539657855e-06, "loss": 0.3073, "step": 2115 }, { "epoch": 1.03, "learning_rate": 9.177682737169519e-06, "loss": 0.0747, "step": 2116 }, { "epoch": 1.03, "learning_rate": 9.177293934681183e-06, "loss": 0.0703, "step": 2117 }, { "epoch": 1.03, "learning_rate": 9.176905132192846e-06, "loss": 0.1281, "step": 2118 }, { "epoch": 1.03, "learning_rate": 9.176516329704512e-06, "loss": 0.144, "step": 2119 }, { "epoch": 1.03, "learning_rate": 9.176127527216175e-06, "loss": 0.3825, "step": 2120 }, { "epoch": 1.03, "learning_rate": 9.175738724727839e-06, "loss": 0.1561, "step": 2121 }, { "epoch": 1.03, "learning_rate": 9.175349922239502e-06, "loss": 0.0837, "step": 2122 }, { "epoch": 1.03, "learning_rate": 9.174961119751166e-06, "loss": 0.0533, "step": 2123 }, { "epoch": 1.03, "learning_rate": 9.174572317262831e-06, "loss": 0.1615, "step": 2124 }, { "epoch": 1.03, "learning_rate": 9.174183514774495e-06, "loss": 0.0518, "step": 2125 }, { "epoch": 1.03, "learning_rate": 9.173794712286159e-06, "loss": 0.0515, "step": 2126 }, { "epoch": 1.03, "learning_rate": 9.173405909797824e-06, "loss": 0.0791, "step": 2127 }, { "epoch": 1.03, "learning_rate": 9.173017107309488e-06, "loss": 0.1705, "step": 2128 }, { "epoch": 1.03, "learning_rate": 9.172628304821151e-06, "loss": 0.1648, "step": 2129 }, { "epoch": 1.03, "learning_rate": 9.172239502332817e-06, "loss": 0.0761, "step": 2130 }, { "epoch": 1.03, "learning_rate": 9.17185069984448e-06, "loss": 0.1281, "step": 2131 }, { "epoch": 1.03, "learning_rate": 9.171461897356144e-06, "loss": 0.1734, "step": 2132 }, { "epoch": 1.03, "learning_rate": 9.17107309486781e-06, "loss": 0.3448, "step": 2133 }, { "epoch": 1.03, "learning_rate": 9.170684292379471e-06, "loss": 0.4711, "step": 2134 }, { "epoch": 1.03, "learning_rate": 9.170295489891136e-06, "loss": 0.0575, "step": 2135 }, { "epoch": 1.03, "learning_rate": 9.1699066874028e-06, "loss": 0.2863, "step": 2136 }, { "epoch": 1.03, "learning_rate": 9.169517884914464e-06, "loss": 0.1427, "step": 2137 }, { "epoch": 1.03, "learning_rate": 9.169129082426127e-06, "loss": 0.2099, "step": 2138 }, { "epoch": 1.03, "learning_rate": 9.168740279937793e-06, "loss": 0.4044, "step": 2139 }, { "epoch": 1.03, "learning_rate": 9.168351477449456e-06, "loss": 0.4015, "step": 2140 }, { "epoch": 1.03, "learning_rate": 9.16796267496112e-06, "loss": 0.3841, "step": 2141 }, { "epoch": 1.03, "learning_rate": 9.167573872472785e-06, "loss": 0.2389, "step": 2142 }, { "epoch": 1.03, "learning_rate": 9.167185069984449e-06, "loss": 0.0808, "step": 2143 }, { "epoch": 1.03, "learning_rate": 9.166796267496112e-06, "loss": 0.336, "step": 2144 }, { "epoch": 1.03, "learning_rate": 9.166407465007778e-06, "loss": 0.199, "step": 2145 }, { "epoch": 1.03, "learning_rate": 9.166018662519441e-06, "loss": 0.1317, "step": 2146 }, { "epoch": 1.03, "learning_rate": 9.165629860031105e-06, "loss": 0.173, "step": 2147 }, { "epoch": 1.03, "learning_rate": 9.165241057542769e-06, "loss": 0.0805, "step": 2148 }, { "epoch": 1.03, "learning_rate": 9.164852255054432e-06, "loss": 0.184, "step": 2149 }, { "epoch": 1.03, "learning_rate": 9.164463452566098e-06, "loss": 0.1191, "step": 2150 }, { "epoch": 1.03, "learning_rate": 9.164074650077761e-06, "loss": 0.0763, "step": 2151 }, { "epoch": 1.03, "learning_rate": 9.163685847589425e-06, "loss": 0.2125, "step": 2152 }, { "epoch": 1.03, "learning_rate": 9.16329704510109e-06, "loss": 0.2239, "step": 2153 }, { "epoch": 1.03, "learning_rate": 9.162908242612754e-06, "loss": 0.1523, "step": 2154 }, { "epoch": 1.03, "learning_rate": 9.162519440124417e-06, "loss": 0.0902, "step": 2155 }, { "epoch": 1.03, "learning_rate": 9.162130637636081e-06, "loss": 0.1702, "step": 2156 }, { "epoch": 1.03, "learning_rate": 9.161741835147746e-06, "loss": 0.1614, "step": 2157 }, { "epoch": 1.03, "learning_rate": 9.16135303265941e-06, "loss": 0.2787, "step": 2158 }, { "epoch": 1.03, "learning_rate": 9.160964230171074e-06, "loss": 0.397, "step": 2159 }, { "epoch": 1.03, "learning_rate": 9.160575427682737e-06, "loss": 0.0956, "step": 2160 }, { "epoch": 1.03, "learning_rate": 9.160186625194401e-06, "loss": 0.2929, "step": 2161 }, { "epoch": 1.03, "learning_rate": 9.159797822706066e-06, "loss": 0.1356, "step": 2162 }, { "epoch": 1.03, "learning_rate": 9.15940902021773e-06, "loss": 0.276, "step": 2163 }, { "epoch": 1.03, "learning_rate": 9.159020217729393e-06, "loss": 0.3742, "step": 2164 }, { "epoch": 1.03, "learning_rate": 9.158631415241059e-06, "loss": 0.2922, "step": 2165 }, { "epoch": 1.03, "learning_rate": 9.158242612752722e-06, "loss": 0.5301, "step": 2166 }, { "epoch": 1.03, "learning_rate": 9.157853810264386e-06, "loss": 0.1368, "step": 2167 }, { "epoch": 1.03, "learning_rate": 9.157465007776051e-06, "loss": 0.0988, "step": 2168 }, { "epoch": 1.03, "learning_rate": 9.157076205287715e-06, "loss": 0.2317, "step": 2169 }, { "epoch": 1.03, "learning_rate": 9.156687402799379e-06, "loss": 0.3727, "step": 2170 }, { "epoch": 1.03, "learning_rate": 9.156298600311044e-06, "loss": 0.0761, "step": 2171 }, { "epoch": 1.03, "learning_rate": 9.155909797822706e-06, "loss": 0.1929, "step": 2172 }, { "epoch": 1.03, "learning_rate": 9.155520995334371e-06, "loss": 0.128, "step": 2173 }, { "epoch": 1.03, "learning_rate": 9.155132192846035e-06, "loss": 0.0448, "step": 2174 }, { "epoch": 1.03, "learning_rate": 9.154743390357698e-06, "loss": 0.1198, "step": 2175 }, { "epoch": 1.03, "learning_rate": 9.154354587869362e-06, "loss": 0.076, "step": 2176 }, { "epoch": 1.03, "learning_rate": 9.153965785381027e-06, "loss": 0.201, "step": 2177 }, { "epoch": 1.03, "learning_rate": 9.153576982892691e-06, "loss": 0.0479, "step": 2178 }, { "epoch": 1.03, "learning_rate": 9.153188180404355e-06, "loss": 0.1251, "step": 2179 }, { "epoch": 1.03, "learning_rate": 9.15279937791602e-06, "loss": 0.0249, "step": 2180 }, { "epoch": 1.03, "learning_rate": 9.152410575427684e-06, "loss": 0.1347, "step": 2181 }, { "epoch": 1.03, "learning_rate": 9.152021772939347e-06, "loss": 0.6095, "step": 2182 }, { "epoch": 1.03, "learning_rate": 9.151632970451013e-06, "loss": 0.2463, "step": 2183 }, { "epoch": 1.03, "learning_rate": 9.151244167962676e-06, "loss": 0.1312, "step": 2184 }, { "epoch": 1.03, "learning_rate": 9.15085536547434e-06, "loss": 0.2229, "step": 2185 }, { "epoch": 1.03, "learning_rate": 9.150466562986003e-06, "loss": 0.1114, "step": 2186 }, { "epoch": 1.03, "learning_rate": 9.150077760497667e-06, "loss": 0.1937, "step": 2187 }, { "epoch": 1.03, "learning_rate": 9.149688958009332e-06, "loss": 0.2663, "step": 2188 }, { "epoch": 1.04, "learning_rate": 9.149300155520996e-06, "loss": 0.1072, "step": 2189 }, { "epoch": 1.04, "learning_rate": 9.14891135303266e-06, "loss": 0.1416, "step": 2190 }, { "epoch": 1.04, "learning_rate": 9.148522550544325e-06, "loss": 0.1853, "step": 2191 }, { "epoch": 1.04, "learning_rate": 9.148133748055989e-06, "loss": 0.1775, "step": 2192 }, { "epoch": 1.04, "learning_rate": 9.147744945567652e-06, "loss": 0.0216, "step": 2193 }, { "epoch": 1.04, "learning_rate": 9.147356143079318e-06, "loss": 0.0433, "step": 2194 }, { "epoch": 1.04, "learning_rate": 9.146967340590981e-06, "loss": 0.0594, "step": 2195 }, { "epoch": 1.04, "learning_rate": 9.146578538102645e-06, "loss": 0.0486, "step": 2196 }, { "epoch": 1.04, "learning_rate": 9.146189735614308e-06, "loss": 0.0693, "step": 2197 }, { "epoch": 1.04, "learning_rate": 9.145800933125972e-06, "loss": 0.0545, "step": 2198 }, { "epoch": 1.04, "learning_rate": 9.145412130637636e-06, "loss": 0.1858, "step": 2199 }, { "epoch": 1.04, "learning_rate": 9.145023328149301e-06, "loss": 0.4332, "step": 2200 }, { "epoch": 1.04, "learning_rate": 9.144634525660965e-06, "loss": 0.0675, "step": 2201 }, { "epoch": 1.04, "learning_rate": 9.144245723172628e-06, "loss": 0.2609, "step": 2202 }, { "epoch": 1.04, "learning_rate": 9.143856920684294e-06, "loss": 0.2915, "step": 2203 }, { "epoch": 1.04, "learning_rate": 9.143468118195957e-06, "loss": 0.0917, "step": 2204 }, { "epoch": 1.04, "learning_rate": 9.143079315707621e-06, "loss": 0.2044, "step": 2205 }, { "epoch": 1.04, "learning_rate": 9.142690513219286e-06, "loss": 0.0465, "step": 2206 }, { "epoch": 1.04, "learning_rate": 9.14230171073095e-06, "loss": 0.2987, "step": 2207 }, { "epoch": 1.04, "learning_rate": 9.141912908242613e-06, "loss": 0.1705, "step": 2208 }, { "epoch": 1.04, "learning_rate": 9.141524105754279e-06, "loss": 0.0558, "step": 2209 }, { "epoch": 1.04, "learning_rate": 9.14113530326594e-06, "loss": 0.1444, "step": 2210 }, { "epoch": 1.04, "learning_rate": 9.140746500777606e-06, "loss": 0.3228, "step": 2211 }, { "epoch": 1.04, "learning_rate": 9.14035769828927e-06, "loss": 0.1607, "step": 2212 }, { "epoch": 1.04, "learning_rate": 9.139968895800933e-06, "loss": 0.4304, "step": 2213 }, { "epoch": 1.04, "learning_rate": 9.139580093312599e-06, "loss": 0.3535, "step": 2214 }, { "epoch": 1.04, "learning_rate": 9.139191290824262e-06, "loss": 0.1693, "step": 2215 }, { "epoch": 1.04, "learning_rate": 9.138802488335926e-06, "loss": 0.28, "step": 2216 }, { "epoch": 1.04, "learning_rate": 9.13841368584759e-06, "loss": 0.0576, "step": 2217 }, { "epoch": 1.04, "learning_rate": 9.138024883359255e-06, "loss": 0.1936, "step": 2218 }, { "epoch": 1.04, "learning_rate": 9.137636080870918e-06, "loss": 0.0722, "step": 2219 }, { "epoch": 1.04, "learning_rate": 9.137247278382582e-06, "loss": 0.0532, "step": 2220 }, { "epoch": 1.04, "learning_rate": 9.136858475894247e-06, "loss": 0.1405, "step": 2221 }, { "epoch": 1.04, "learning_rate": 9.136469673405911e-06, "loss": 0.1317, "step": 2222 }, { "epoch": 1.04, "learning_rate": 9.136080870917575e-06, "loss": 0.0239, "step": 2223 }, { "epoch": 1.04, "learning_rate": 9.135692068429238e-06, "loss": 0.0209, "step": 2224 }, { "epoch": 1.04, "learning_rate": 9.135303265940902e-06, "loss": 0.1581, "step": 2225 }, { "epoch": 1.04, "learning_rate": 9.134914463452567e-06, "loss": 0.2041, "step": 2226 }, { "epoch": 1.04, "learning_rate": 9.13452566096423e-06, "loss": 0.0653, "step": 2227 }, { "epoch": 1.04, "learning_rate": 9.134136858475894e-06, "loss": 0.1227, "step": 2228 }, { "epoch": 1.04, "learning_rate": 9.13374805598756e-06, "loss": 0.1687, "step": 2229 }, { "epoch": 1.04, "learning_rate": 9.133359253499223e-06, "loss": 0.05, "step": 2230 }, { "epoch": 1.04, "learning_rate": 9.132970451010887e-06, "loss": 0.4971, "step": 2231 }, { "epoch": 1.04, "learning_rate": 9.132581648522552e-06, "loss": 0.093, "step": 2232 }, { "epoch": 1.04, "learning_rate": 9.132192846034216e-06, "loss": 0.2021, "step": 2233 }, { "epoch": 1.04, "learning_rate": 9.13180404354588e-06, "loss": 0.1419, "step": 2234 }, { "epoch": 1.04, "learning_rate": 9.131415241057543e-06, "loss": 0.1551, "step": 2235 }, { "epoch": 1.04, "learning_rate": 9.131026438569207e-06, "loss": 0.1635, "step": 2236 }, { "epoch": 1.04, "learning_rate": 9.13063763608087e-06, "loss": 0.338, "step": 2237 }, { "epoch": 1.04, "learning_rate": 9.130248833592536e-06, "loss": 0.3465, "step": 2238 }, { "epoch": 1.04, "learning_rate": 9.1298600311042e-06, "loss": 0.0868, "step": 2239 }, { "epoch": 1.04, "learning_rate": 9.129471228615863e-06, "loss": 0.4092, "step": 2240 }, { "epoch": 1.04, "learning_rate": 9.129082426127528e-06, "loss": 0.0342, "step": 2241 }, { "epoch": 1.04, "learning_rate": 9.128693623639192e-06, "loss": 0.2807, "step": 2242 }, { "epoch": 1.04, "learning_rate": 9.128304821150856e-06, "loss": 0.0612, "step": 2243 }, { "epoch": 1.04, "learning_rate": 9.127916018662521e-06, "loss": 0.0903, "step": 2244 }, { "epoch": 1.04, "learning_rate": 9.127527216174185e-06, "loss": 0.0579, "step": 2245 }, { "epoch": 1.04, "learning_rate": 9.127138413685848e-06, "loss": 0.0527, "step": 2246 }, { "epoch": 1.04, "learning_rate": 9.126749611197514e-06, "loss": 0.5689, "step": 2247 }, { "epoch": 1.04, "learning_rate": 9.126360808709175e-06, "loss": 0.1618, "step": 2248 }, { "epoch": 1.04, "learning_rate": 9.12597200622084e-06, "loss": 0.2708, "step": 2249 }, { "epoch": 1.04, "learning_rate": 9.125583203732504e-06, "loss": 0.1514, "step": 2250 }, { "epoch": 1.04, "learning_rate": 9.125194401244168e-06, "loss": 0.2461, "step": 2251 }, { "epoch": 1.04, "learning_rate": 9.124805598755833e-06, "loss": 0.1348, "step": 2252 }, { "epoch": 1.04, "learning_rate": 9.124416796267497e-06, "loss": 0.414, "step": 2253 }, { "epoch": 1.04, "learning_rate": 9.12402799377916e-06, "loss": 0.0615, "step": 2254 }, { "epoch": 1.04, "learning_rate": 9.123639191290824e-06, "loss": 0.1735, "step": 2255 }, { "epoch": 1.04, "learning_rate": 9.12325038880249e-06, "loss": 0.1409, "step": 2256 }, { "epoch": 1.04, "learning_rate": 9.122861586314153e-06, "loss": 0.0551, "step": 2257 }, { "epoch": 1.04, "learning_rate": 9.122472783825817e-06, "loss": 0.369, "step": 2258 }, { "epoch": 1.04, "learning_rate": 9.122083981337482e-06, "loss": 0.1605, "step": 2259 }, { "epoch": 1.04, "learning_rate": 9.121695178849146e-06, "loss": 0.2866, "step": 2260 }, { "epoch": 1.04, "learning_rate": 9.12130637636081e-06, "loss": 0.1278, "step": 2261 }, { "epoch": 1.04, "learning_rate": 9.120917573872473e-06, "loss": 0.4768, "step": 2262 }, { "epoch": 1.04, "learning_rate": 9.120528771384137e-06, "loss": 0.1194, "step": 2263 }, { "epoch": 1.04, "learning_rate": 9.120139968895802e-06, "loss": 0.4112, "step": 2264 }, { "epoch": 1.04, "learning_rate": 9.119751166407466e-06, "loss": 0.0688, "step": 2265 }, { "epoch": 1.04, "learning_rate": 9.11936236391913e-06, "loss": 0.0639, "step": 2266 }, { "epoch": 1.04, "learning_rate": 9.118973561430795e-06, "loss": 0.153, "step": 2267 }, { "epoch": 1.04, "learning_rate": 9.118584758942458e-06, "loss": 0.15, "step": 2268 }, { "epoch": 1.04, "learning_rate": 9.118195956454122e-06, "loss": 0.1737, "step": 2269 }, { "epoch": 1.04, "learning_rate": 9.117807153965787e-06, "loss": 0.147, "step": 2270 }, { "epoch": 1.04, "learning_rate": 9.11741835147745e-06, "loss": 0.1724, "step": 2271 }, { "epoch": 1.04, "learning_rate": 9.117029548989114e-06, "loss": 0.1077, "step": 2272 }, { "epoch": 1.04, "learning_rate": 9.11664074650078e-06, "loss": 0.2009, "step": 2273 }, { "epoch": 1.04, "learning_rate": 9.116251944012442e-06, "loss": 0.0616, "step": 2274 }, { "epoch": 1.04, "learning_rate": 9.115863141524105e-06, "loss": 0.1085, "step": 2275 }, { "epoch": 1.04, "learning_rate": 9.11547433903577e-06, "loss": 0.0478, "step": 2276 }, { "epoch": 1.04, "learning_rate": 9.115085536547434e-06, "loss": 0.0574, "step": 2277 }, { "epoch": 1.04, "learning_rate": 9.114696734059098e-06, "loss": 0.0991, "step": 2278 }, { "epoch": 1.04, "learning_rate": 9.114307931570763e-06, "loss": 0.11, "step": 2279 }, { "epoch": 1.04, "learning_rate": 9.113919129082427e-06, "loss": 0.2109, "step": 2280 }, { "epoch": 1.04, "learning_rate": 9.11353032659409e-06, "loss": 0.302, "step": 2281 }, { "epoch": 1.04, "learning_rate": 9.113141524105756e-06, "loss": 0.1429, "step": 2282 }, { "epoch": 1.04, "learning_rate": 9.11275272161742e-06, "loss": 0.2027, "step": 2283 }, { "epoch": 1.04, "learning_rate": 9.112363919129083e-06, "loss": 0.2817, "step": 2284 }, { "epoch": 1.04, "learning_rate": 9.111975116640748e-06, "loss": 0.4526, "step": 2285 }, { "epoch": 1.04, "learning_rate": 9.11158631415241e-06, "loss": 0.1254, "step": 2286 }, { "epoch": 1.04, "learning_rate": 9.111197511664076e-06, "loss": 0.1589, "step": 2287 }, { "epoch": 1.04, "learning_rate": 9.11080870917574e-06, "loss": 0.1132, "step": 2288 }, { "epoch": 1.04, "learning_rate": 9.110419906687403e-06, "loss": 0.1083, "step": 2289 }, { "epoch": 1.04, "learning_rate": 9.110031104199068e-06, "loss": 0.0271, "step": 2290 }, { "epoch": 1.04, "learning_rate": 9.109642301710732e-06, "loss": 0.0753, "step": 2291 }, { "epoch": 1.04, "learning_rate": 9.109253499222395e-06, "loss": 0.3856, "step": 2292 }, { "epoch": 1.04, "learning_rate": 9.10886469673406e-06, "loss": 0.0703, "step": 2293 }, { "epoch": 1.04, "learning_rate": 9.108475894245724e-06, "loss": 0.1714, "step": 2294 }, { "epoch": 1.04, "learning_rate": 9.108087091757388e-06, "loss": 0.0999, "step": 2295 }, { "epoch": 1.04, "learning_rate": 9.107698289269052e-06, "loss": 0.3001, "step": 2296 }, { "epoch": 1.04, "learning_rate": 9.107309486780717e-06, "loss": 0.1755, "step": 2297 }, { "epoch": 1.04, "learning_rate": 9.10692068429238e-06, "loss": 0.0855, "step": 2298 }, { "epoch": 1.04, "learning_rate": 9.106531881804044e-06, "loss": 0.1889, "step": 2299 }, { "epoch": 1.04, "learning_rate": 9.106143079315708e-06, "loss": 0.5226, "step": 2300 }, { "epoch": 1.04, "learning_rate": 9.105754276827371e-06, "loss": 0.1536, "step": 2301 }, { "epoch": 1.04, "learning_rate": 9.105365474339037e-06, "loss": 0.181, "step": 2302 }, { "epoch": 1.04, "learning_rate": 9.1049766718507e-06, "loss": 0.2, "step": 2303 }, { "epoch": 1.04, "learning_rate": 9.104587869362364e-06, "loss": 0.2412, "step": 2304 }, { "epoch": 1.04, "learning_rate": 9.10419906687403e-06, "loss": 0.041, "step": 2305 }, { "epoch": 1.04, "learning_rate": 9.103810264385693e-06, "loss": 0.1163, "step": 2306 }, { "epoch": 1.04, "learning_rate": 9.103421461897357e-06, "loss": 0.2905, "step": 2307 }, { "epoch": 1.04, "learning_rate": 9.103032659409022e-06, "loss": 0.5034, "step": 2308 }, { "epoch": 1.04, "learning_rate": 9.102643856920686e-06, "loss": 0.2256, "step": 2309 }, { "epoch": 1.04, "learning_rate": 9.10225505443235e-06, "loss": 0.0694, "step": 2310 }, { "epoch": 1.04, "learning_rate": 9.101866251944013e-06, "loss": 0.1907, "step": 2311 }, { "epoch": 1.04, "learning_rate": 9.101477449455676e-06, "loss": 0.0202, "step": 2312 }, { "epoch": 1.04, "learning_rate": 9.101088646967342e-06, "loss": 0.0358, "step": 2313 }, { "epoch": 1.04, "learning_rate": 9.100699844479005e-06, "loss": 0.1365, "step": 2314 }, { "epoch": 1.04, "learning_rate": 9.100311041990669e-06, "loss": 0.4403, "step": 2315 }, { "epoch": 1.04, "learning_rate": 9.099922239502333e-06, "loss": 0.1718, "step": 2316 }, { "epoch": 1.04, "learning_rate": 9.099533437013998e-06, "loss": 0.0393, "step": 2317 }, { "epoch": 1.04, "learning_rate": 9.099144634525662e-06, "loss": 0.5932, "step": 2318 }, { "epoch": 1.04, "learning_rate": 9.098755832037325e-06, "loss": 0.0243, "step": 2319 }, { "epoch": 1.04, "learning_rate": 9.09836702954899e-06, "loss": 0.1379, "step": 2320 }, { "epoch": 1.04, "learning_rate": 9.097978227060654e-06, "loss": 0.3299, "step": 2321 }, { "epoch": 1.04, "learning_rate": 9.097589424572318e-06, "loss": 0.4305, "step": 2322 }, { "epoch": 1.04, "learning_rate": 9.097200622083983e-06, "loss": 0.2403, "step": 2323 }, { "epoch": 1.04, "learning_rate": 9.096811819595645e-06, "loss": 0.2285, "step": 2324 }, { "epoch": 1.04, "learning_rate": 9.09642301710731e-06, "loss": 0.1079, "step": 2325 }, { "epoch": 1.04, "learning_rate": 9.096034214618974e-06, "loss": 0.0825, "step": 2326 }, { "epoch": 1.04, "learning_rate": 9.095645412130638e-06, "loss": 0.0591, "step": 2327 }, { "epoch": 1.04, "learning_rate": 9.095256609642303e-06, "loss": 0.0474, "step": 2328 }, { "epoch": 1.04, "learning_rate": 9.094867807153967e-06, "loss": 0.1951, "step": 2329 }, { "epoch": 1.04, "learning_rate": 9.09447900466563e-06, "loss": 0.4083, "step": 2330 }, { "epoch": 1.04, "learning_rate": 9.094090202177296e-06, "loss": 0.0204, "step": 2331 }, { "epoch": 1.04, "learning_rate": 9.09370139968896e-06, "loss": 0.0101, "step": 2332 }, { "epoch": 1.04, "learning_rate": 9.093312597200623e-06, "loss": 0.0942, "step": 2333 }, { "epoch": 1.04, "learning_rate": 9.092923794712286e-06, "loss": 0.0701, "step": 2334 }, { "epoch": 1.04, "learning_rate": 9.092534992223952e-06, "loss": 0.1543, "step": 2335 }, { "epoch": 1.04, "learning_rate": 9.092146189735615e-06, "loss": 0.4366, "step": 2336 }, { "epoch": 1.04, "learning_rate": 9.091757387247279e-06, "loss": 0.1969, "step": 2337 }, { "epoch": 1.04, "learning_rate": 9.091368584758943e-06, "loss": 0.4397, "step": 2338 }, { "epoch": 1.04, "learning_rate": 9.090979782270606e-06, "loss": 0.0204, "step": 2339 }, { "epoch": 1.04, "learning_rate": 9.090590979782272e-06, "loss": 0.5049, "step": 2340 }, { "epoch": 1.04, "learning_rate": 9.090202177293935e-06, "loss": 0.4141, "step": 2341 }, { "epoch": 1.04, "learning_rate": 9.089813374805599e-06, "loss": 0.5404, "step": 2342 }, { "epoch": 1.04, "learning_rate": 9.089424572317264e-06, "loss": 0.0757, "step": 2343 }, { "epoch": 1.04, "learning_rate": 9.089035769828928e-06, "loss": 0.0392, "step": 2344 }, { "epoch": 1.04, "learning_rate": 9.088646967340591e-06, "loss": 0.1201, "step": 2345 }, { "epoch": 1.04, "learning_rate": 9.088258164852257e-06, "loss": 0.2555, "step": 2346 }, { "epoch": 1.04, "learning_rate": 9.08786936236392e-06, "loss": 0.5941, "step": 2347 }, { "epoch": 1.04, "learning_rate": 9.087480559875584e-06, "loss": 0.0794, "step": 2348 }, { "epoch": 1.04, "learning_rate": 9.087091757387248e-06, "loss": 0.2705, "step": 2349 }, { "epoch": 1.04, "learning_rate": 9.086702954898911e-06, "loss": 0.2934, "step": 2350 }, { "epoch": 1.04, "learning_rate": 9.086314152410577e-06, "loss": 0.033, "step": 2351 }, { "epoch": 1.04, "learning_rate": 9.08592534992224e-06, "loss": 0.4122, "step": 2352 }, { "epoch": 1.04, "learning_rate": 9.085536547433904e-06, "loss": 0.6044, "step": 2353 }, { "epoch": 1.04, "learning_rate": 9.085147744945567e-06, "loss": 0.0725, "step": 2354 }, { "epoch": 1.04, "learning_rate": 9.084758942457233e-06, "loss": 0.1177, "step": 2355 }, { "epoch": 1.04, "learning_rate": 9.084370139968896e-06, "loss": 0.3759, "step": 2356 }, { "epoch": 1.04, "learning_rate": 9.08398133748056e-06, "loss": 0.3695, "step": 2357 }, { "epoch": 1.04, "learning_rate": 9.083592534992225e-06, "loss": 0.4094, "step": 2358 }, { "epoch": 1.04, "learning_rate": 9.083203732503889e-06, "loss": 0.0639, "step": 2359 }, { "epoch": 1.04, "learning_rate": 9.082814930015553e-06, "loss": 0.0705, "step": 2360 }, { "epoch": 1.04, "learning_rate": 9.082426127527218e-06, "loss": 0.3501, "step": 2361 }, { "epoch": 1.04, "learning_rate": 9.08203732503888e-06, "loss": 0.187, "step": 2362 }, { "epoch": 1.04, "learning_rate": 9.081648522550545e-06, "loss": 0.2368, "step": 2363 }, { "epoch": 1.04, "learning_rate": 9.081259720062209e-06, "loss": 0.2375, "step": 2364 }, { "epoch": 1.04, "learning_rate": 9.080870917573872e-06, "loss": 0.0433, "step": 2365 }, { "epoch": 1.04, "learning_rate": 9.080482115085538e-06, "loss": 0.3242, "step": 2366 }, { "epoch": 1.04, "learning_rate": 9.080093312597201e-06, "loss": 0.0542, "step": 2367 }, { "epoch": 1.04, "learning_rate": 9.079704510108865e-06, "loss": 0.0863, "step": 2368 }, { "epoch": 1.04, "learning_rate": 9.07931570762053e-06, "loss": 0.2503, "step": 2369 }, { "epoch": 1.04, "learning_rate": 9.078926905132194e-06, "loss": 0.1313, "step": 2370 }, { "epoch": 1.04, "learning_rate": 9.078538102643858e-06, "loss": 0.2434, "step": 2371 }, { "epoch": 1.04, "learning_rate": 9.078149300155523e-06, "loss": 0.0963, "step": 2372 }, { "epoch": 1.04, "learning_rate": 9.077760497667187e-06, "loss": 0.0953, "step": 2373 }, { "epoch": 1.04, "learning_rate": 9.07737169517885e-06, "loss": 0.0888, "step": 2374 }, { "epoch": 1.04, "learning_rate": 9.076982892690514e-06, "loss": 0.0094, "step": 2375 }, { "epoch": 1.04, "learning_rate": 9.076594090202177e-06, "loss": 0.1068, "step": 2376 }, { "epoch": 1.04, "learning_rate": 9.076205287713841e-06, "loss": 0.112, "step": 2377 }, { "epoch": 1.04, "learning_rate": 9.075816485225506e-06, "loss": 0.0513, "step": 2378 }, { "epoch": 1.04, "learning_rate": 9.07542768273717e-06, "loss": 0.431, "step": 2379 }, { "epoch": 1.04, "learning_rate": 9.075038880248834e-06, "loss": 0.1089, "step": 2380 }, { "epoch": 1.04, "learning_rate": 9.074650077760499e-06, "loss": 0.2635, "step": 2381 }, { "epoch": 1.04, "learning_rate": 9.074261275272163e-06, "loss": 0.0546, "step": 2382 }, { "epoch": 1.04, "learning_rate": 9.073872472783826e-06, "loss": 0.1943, "step": 2383 }, { "epoch": 1.04, "learning_rate": 9.073483670295492e-06, "loss": 0.008, "step": 2384 }, { "epoch": 1.04, "learning_rate": 9.073094867807155e-06, "loss": 0.1826, "step": 2385 }, { "epoch": 1.04, "learning_rate": 9.072706065318819e-06, "loss": 0.2387, "step": 2386 }, { "epoch": 1.04, "learning_rate": 9.072317262830482e-06, "loss": 0.5912, "step": 2387 }, { "epoch": 1.04, "learning_rate": 9.071928460342146e-06, "loss": 0.1032, "step": 2388 }, { "epoch": 1.04, "learning_rate": 9.071539657853811e-06, "loss": 0.1341, "step": 2389 }, { "epoch": 1.04, "learning_rate": 9.071150855365475e-06, "loss": 0.2837, "step": 2390 }, { "epoch": 1.04, "learning_rate": 9.070762052877139e-06, "loss": 0.2613, "step": 2391 }, { "epoch": 1.04, "learning_rate": 9.070373250388804e-06, "loss": 0.1762, "step": 2392 }, { "epoch": 1.04, "learning_rate": 9.069984447900468e-06, "loss": 0.2447, "step": 2393 }, { "epoch": 1.04, "learning_rate": 9.069595645412131e-06, "loss": 0.0488, "step": 2394 }, { "epoch": 1.04, "learning_rate": 9.069206842923795e-06, "loss": 0.0713, "step": 2395 }, { "epoch": 1.04, "learning_rate": 9.06881804043546e-06, "loss": 0.4743, "step": 2396 }, { "epoch": 1.04, "learning_rate": 9.068429237947124e-06, "loss": 0.2294, "step": 2397 }, { "epoch": 1.04, "learning_rate": 9.068040435458787e-06, "loss": 0.1397, "step": 2398 }, { "epoch": 1.04, "learning_rate": 9.067651632970453e-06, "loss": 0.2697, "step": 2399 }, { "epoch": 1.04, "learning_rate": 9.067262830482115e-06, "loss": 0.0328, "step": 2400 }, { "epoch": 1.04, "learning_rate": 9.06687402799378e-06, "loss": 0.182, "step": 2401 }, { "epoch": 1.04, "learning_rate": 9.066485225505444e-06, "loss": 0.4465, "step": 2402 }, { "epoch": 1.04, "learning_rate": 9.066096423017107e-06, "loss": 0.1129, "step": 2403 }, { "epoch": 1.04, "learning_rate": 9.065707620528773e-06, "loss": 0.2309, "step": 2404 }, { "epoch": 1.04, "learning_rate": 9.065318818040436e-06, "loss": 0.0614, "step": 2405 }, { "epoch": 1.04, "learning_rate": 9.0649300155521e-06, "loss": 0.5575, "step": 2406 }, { "epoch": 1.04, "learning_rate": 9.064541213063765e-06, "loss": 0.0253, "step": 2407 }, { "epoch": 1.04, "learning_rate": 9.064152410575429e-06, "loss": 0.1558, "step": 2408 }, { "epoch": 1.04, "learning_rate": 9.063763608087092e-06, "loss": 0.0898, "step": 2409 }, { "epoch": 1.04, "learning_rate": 9.063374805598758e-06, "loss": 0.1959, "step": 2410 }, { "epoch": 1.04, "learning_rate": 9.062986003110421e-06, "loss": 0.2282, "step": 2411 }, { "epoch": 1.04, "learning_rate": 9.062597200622085e-06, "loss": 0.1775, "step": 2412 }, { "epoch": 1.04, "learning_rate": 9.062208398133749e-06, "loss": 0.109, "step": 2413 }, { "epoch": 1.04, "learning_rate": 9.061819595645412e-06, "loss": 0.0907, "step": 2414 }, { "epoch": 1.04, "learning_rate": 9.061430793157076e-06, "loss": 0.0212, "step": 2415 }, { "epoch": 1.04, "learning_rate": 9.061041990668741e-06, "loss": 0.047, "step": 2416 }, { "epoch": 1.04, "learning_rate": 9.060653188180405e-06, "loss": 0.183, "step": 2417 }, { "epoch": 1.04, "learning_rate": 9.060264385692068e-06, "loss": 0.0871, "step": 2418 }, { "epoch": 1.04, "learning_rate": 9.059875583203734e-06, "loss": 0.2984, "step": 2419 }, { "epoch": 1.04, "learning_rate": 9.059486780715397e-06, "loss": 0.2404, "step": 2420 }, { "epoch": 1.04, "learning_rate": 9.059097978227061e-06, "loss": 0.0538, "step": 2421 }, { "epoch": 1.04, "learning_rate": 9.058709175738726e-06, "loss": 0.1195, "step": 2422 }, { "epoch": 1.04, "learning_rate": 9.05832037325039e-06, "loss": 0.1141, "step": 2423 }, { "epoch": 1.04, "learning_rate": 9.057931570762054e-06, "loss": 0.1045, "step": 2424 }, { "epoch": 1.04, "learning_rate": 9.057542768273717e-06, "loss": 0.0368, "step": 2425 }, { "epoch": 1.04, "learning_rate": 9.057153965785381e-06, "loss": 0.154, "step": 2426 }, { "epoch": 1.04, "learning_rate": 9.056765163297046e-06, "loss": 0.0729, "step": 2427 }, { "epoch": 1.04, "learning_rate": 9.05637636080871e-06, "loss": 0.3106, "step": 2428 }, { "epoch": 1.04, "learning_rate": 9.055987558320373e-06, "loss": 0.0652, "step": 2429 }, { "epoch": 1.04, "learning_rate": 9.055598755832039e-06, "loss": 0.0274, "step": 2430 }, { "epoch": 1.04, "learning_rate": 9.055209953343702e-06, "loss": 0.281, "step": 2431 }, { "epoch": 1.04, "learning_rate": 9.054821150855366e-06, "loss": 0.087, "step": 2432 }, { "epoch": 1.04, "learning_rate": 9.05443234836703e-06, "loss": 0.1117, "step": 2433 }, { "epoch": 1.04, "learning_rate": 9.054043545878695e-06, "loss": 0.0221, "step": 2434 }, { "epoch": 1.04, "learning_rate": 9.053654743390359e-06, "loss": 0.1381, "step": 2435 }, { "epoch": 1.04, "learning_rate": 9.053265940902022e-06, "loss": 0.0819, "step": 2436 }, { "epoch": 1.04, "learning_rate": 9.052877138413688e-06, "loss": 0.1796, "step": 2437 }, { "epoch": 1.04, "learning_rate": 9.05248833592535e-06, "loss": 0.1353, "step": 2438 }, { "epoch": 1.04, "learning_rate": 9.052099533437015e-06, "loss": 0.1694, "step": 2439 }, { "epoch": 1.04, "learning_rate": 9.051710730948678e-06, "loss": 0.4143, "step": 2440 }, { "epoch": 1.04, "learning_rate": 9.051321928460342e-06, "loss": 0.074, "step": 2441 }, { "epoch": 1.04, "learning_rate": 9.050933125972007e-06, "loss": 0.0706, "step": 2442 }, { "epoch": 1.04, "learning_rate": 9.050544323483671e-06, "loss": 0.1417, "step": 2443 }, { "epoch": 1.04, "learning_rate": 9.050155520995335e-06, "loss": 0.176, "step": 2444 }, { "epoch": 1.04, "learning_rate": 9.049766718507e-06, "loss": 0.285, "step": 2445 }, { "epoch": 1.05, "learning_rate": 9.049377916018664e-06, "loss": 0.0995, "step": 2446 }, { "epoch": 1.05, "learning_rate": 9.048989113530327e-06, "loss": 0.2537, "step": 2447 }, { "epoch": 1.05, "learning_rate": 9.048600311041993e-06, "loss": 0.1283, "step": 2448 }, { "epoch": 1.05, "learning_rate": 9.048211508553656e-06, "loss": 0.2188, "step": 2449 }, { "epoch": 1.05, "learning_rate": 9.04782270606532e-06, "loss": 0.1546, "step": 2450 }, { "epoch": 1.05, "learning_rate": 9.047433903576983e-06, "loss": 0.3002, "step": 2451 }, { "epoch": 1.05, "learning_rate": 9.047045101088647e-06, "loss": 0.0491, "step": 2452 }, { "epoch": 1.05, "learning_rate": 9.04665629860031e-06, "loss": 0.0766, "step": 2453 }, { "epoch": 1.05, "learning_rate": 9.046267496111976e-06, "loss": 0.209, "step": 2454 }, { "epoch": 1.05, "learning_rate": 9.04587869362364e-06, "loss": 0.1019, "step": 2455 }, { "epoch": 1.05, "learning_rate": 9.045489891135303e-06, "loss": 0.0345, "step": 2456 }, { "epoch": 1.05, "learning_rate": 9.045101088646969e-06, "loss": 0.1428, "step": 2457 }, { "epoch": 1.05, "learning_rate": 9.044712286158632e-06, "loss": 0.1958, "step": 2458 }, { "epoch": 1.05, "learning_rate": 9.044323483670296e-06, "loss": 0.2167, "step": 2459 }, { "epoch": 1.05, "learning_rate": 9.043934681181961e-06, "loss": 0.1877, "step": 2460 }, { "epoch": 1.05, "learning_rate": 9.043545878693625e-06, "loss": 0.1432, "step": 2461 }, { "epoch": 1.05, "learning_rate": 9.043157076205288e-06, "loss": 0.0112, "step": 2462 }, { "epoch": 1.05, "learning_rate": 9.042768273716952e-06, "loss": 0.1324, "step": 2463 }, { "epoch": 1.05, "learning_rate": 9.042379471228616e-06, "loss": 0.2944, "step": 2464 }, { "epoch": 1.05, "learning_rate": 9.041990668740281e-06, "loss": 0.5872, "step": 2465 }, { "epoch": 1.05, "learning_rate": 9.041601866251945e-06, "loss": 0.1328, "step": 2466 }, { "epoch": 1.05, "learning_rate": 9.041213063763608e-06, "loss": 0.4866, "step": 2467 }, { "epoch": 1.05, "learning_rate": 9.040824261275274e-06, "loss": 0.0536, "step": 2468 }, { "epoch": 1.05, "learning_rate": 9.040435458786937e-06, "loss": 0.1268, "step": 2469 }, { "epoch": 1.05, "learning_rate": 9.0400466562986e-06, "loss": 0.1876, "step": 2470 }, { "epoch": 1.05, "learning_rate": 9.039657853810264e-06, "loss": 0.1027, "step": 2471 }, { "epoch": 1.05, "learning_rate": 9.03926905132193e-06, "loss": 0.1363, "step": 2472 }, { "epoch": 1.05, "learning_rate": 9.038880248833593e-06, "loss": 0.1948, "step": 2473 }, { "epoch": 1.05, "learning_rate": 9.038491446345257e-06, "loss": 0.1996, "step": 2474 }, { "epoch": 1.05, "learning_rate": 9.038102643856922e-06, "loss": 0.0642, "step": 2475 }, { "epoch": 1.05, "learning_rate": 9.037713841368584e-06, "loss": 0.2957, "step": 2476 }, { "epoch": 1.05, "learning_rate": 9.03732503888025e-06, "loss": 0.0532, "step": 2477 }, { "epoch": 1.05, "learning_rate": 9.036936236391913e-06, "loss": 0.3708, "step": 2478 }, { "epoch": 1.05, "learning_rate": 9.036547433903577e-06, "loss": 0.5191, "step": 2479 }, { "epoch": 1.05, "learning_rate": 9.036158631415242e-06, "loss": 0.1678, "step": 2480 }, { "epoch": 1.05, "learning_rate": 9.035769828926906e-06, "loss": 0.0448, "step": 2481 }, { "epoch": 1.05, "learning_rate": 9.03538102643857e-06, "loss": 0.1911, "step": 2482 }, { "epoch": 1.05, "learning_rate": 9.034992223950235e-06, "loss": 0.3601, "step": 2483 }, { "epoch": 1.05, "learning_rate": 9.034603421461898e-06, "loss": 0.2061, "step": 2484 }, { "epoch": 1.05, "learning_rate": 9.034214618973562e-06, "loss": 0.0367, "step": 2485 }, { "epoch": 1.05, "learning_rate": 9.033825816485227e-06, "loss": 0.2684, "step": 2486 }, { "epoch": 1.05, "learning_rate": 9.033437013996891e-06, "loss": 0.0946, "step": 2487 }, { "epoch": 1.05, "learning_rate": 9.033048211508555e-06, "loss": 0.087, "step": 2488 }, { "epoch": 1.05, "learning_rate": 9.032659409020218e-06, "loss": 0.1925, "step": 2489 }, { "epoch": 1.05, "learning_rate": 9.032270606531882e-06, "loss": 0.3394, "step": 2490 }, { "epoch": 1.05, "learning_rate": 9.031881804043545e-06, "loss": 0.1327, "step": 2491 }, { "epoch": 1.05, "learning_rate": 9.03149300155521e-06, "loss": 0.1116, "step": 2492 }, { "epoch": 1.05, "learning_rate": 9.031104199066874e-06, "loss": 0.2798, "step": 2493 }, { "epoch": 1.05, "learning_rate": 9.030715396578538e-06, "loss": 0.1146, "step": 2494 }, { "epoch": 1.05, "learning_rate": 9.030326594090203e-06, "loss": 0.6748, "step": 2495 }, { "epoch": 1.05, "learning_rate": 9.029937791601867e-06, "loss": 0.1318, "step": 2496 }, { "epoch": 1.05, "learning_rate": 9.02954898911353e-06, "loss": 0.2022, "step": 2497 }, { "epoch": 1.05, "learning_rate": 9.029160186625196e-06, "loss": 0.1034, "step": 2498 }, { "epoch": 1.05, "learning_rate": 9.02877138413686e-06, "loss": 0.3419, "step": 2499 }, { "epoch": 1.05, "learning_rate": 9.028382581648523e-06, "loss": 0.0605, "step": 2500 }, { "epoch": 1.05, "learning_rate": 9.027993779160187e-06, "loss": 0.1356, "step": 2501 }, { "epoch": 1.05, "learning_rate": 9.02760497667185e-06, "loss": 0.0373, "step": 2502 }, { "epoch": 1.05, "learning_rate": 9.027216174183516e-06, "loss": 0.1668, "step": 2503 }, { "epoch": 1.05, "learning_rate": 9.02682737169518e-06, "loss": 0.4398, "step": 2504 }, { "epoch": 1.05, "learning_rate": 9.026438569206843e-06, "loss": 0.0639, "step": 2505 }, { "epoch": 1.05, "learning_rate": 9.026049766718508e-06, "loss": 0.1162, "step": 2506 }, { "epoch": 1.05, "learning_rate": 9.025660964230172e-06, "loss": 0.0555, "step": 2507 }, { "epoch": 1.05, "learning_rate": 9.025272161741836e-06, "loss": 0.3419, "step": 2508 }, { "epoch": 1.05, "learning_rate": 9.024883359253501e-06, "loss": 0.2192, "step": 2509 }, { "epoch": 1.05, "learning_rate": 9.024494556765165e-06, "loss": 0.112, "step": 2510 }, { "epoch": 1.05, "learning_rate": 9.024105754276828e-06, "loss": 0.0759, "step": 2511 }, { "epoch": 1.05, "learning_rate": 9.023716951788492e-06, "loss": 0.0565, "step": 2512 }, { "epoch": 1.05, "learning_rate": 9.023328149300157e-06, "loss": 0.0802, "step": 2513 }, { "epoch": 1.05, "learning_rate": 9.022939346811819e-06, "loss": 0.1112, "step": 2514 }, { "epoch": 1.05, "learning_rate": 9.022550544323484e-06, "loss": 0.3185, "step": 2515 }, { "epoch": 1.05, "learning_rate": 9.022161741835148e-06, "loss": 0.1148, "step": 2516 }, { "epoch": 1.05, "learning_rate": 9.021772939346812e-06, "loss": 0.1815, "step": 2517 }, { "epoch": 1.05, "learning_rate": 9.021384136858477e-06, "loss": 0.2411, "step": 2518 }, { "epoch": 1.05, "learning_rate": 9.02099533437014e-06, "loss": 0.1816, "step": 2519 }, { "epoch": 1.05, "learning_rate": 9.020606531881804e-06, "loss": 0.183, "step": 2520 }, { "epoch": 1.05, "learning_rate": 9.02021772939347e-06, "loss": 0.1262, "step": 2521 }, { "epoch": 1.05, "learning_rate": 9.019828926905133e-06, "loss": 0.103, "step": 2522 }, { "epoch": 1.05, "learning_rate": 9.019440124416797e-06, "loss": 0.32, "step": 2523 }, { "epoch": 1.05, "learning_rate": 9.019051321928462e-06, "loss": 0.3453, "step": 2524 }, { "epoch": 1.05, "learning_rate": 9.018662519440126e-06, "loss": 0.0431, "step": 2525 }, { "epoch": 1.05, "learning_rate": 9.01827371695179e-06, "loss": 0.0287, "step": 2526 }, { "epoch": 1.05, "learning_rate": 9.017884914463453e-06, "loss": 0.1438, "step": 2527 }, { "epoch": 1.05, "learning_rate": 9.017496111975117e-06, "loss": 0.3484, "step": 2528 }, { "epoch": 1.05, "learning_rate": 9.017107309486782e-06, "loss": 0.0494, "step": 2529 }, { "epoch": 1.05, "learning_rate": 9.016718506998446e-06, "loss": 0.159, "step": 2530 }, { "epoch": 1.05, "learning_rate": 9.01632970451011e-06, "loss": 0.1874, "step": 2531 }, { "epoch": 1.05, "learning_rate": 9.015940902021773e-06, "loss": 0.049, "step": 2532 }, { "epoch": 1.05, "learning_rate": 9.015552099533438e-06, "loss": 0.276, "step": 2533 }, { "epoch": 1.05, "learning_rate": 9.015163297045102e-06, "loss": 0.2519, "step": 2534 }, { "epoch": 1.05, "learning_rate": 9.014774494556765e-06, "loss": 0.0761, "step": 2535 }, { "epoch": 1.05, "learning_rate": 9.01438569206843e-06, "loss": 0.1249, "step": 2536 }, { "epoch": 1.05, "learning_rate": 9.013996889580094e-06, "loss": 0.0548, "step": 2537 }, { "epoch": 1.05, "learning_rate": 9.013608087091758e-06, "loss": 0.12, "step": 2538 }, { "epoch": 1.05, "learning_rate": 9.013219284603422e-06, "loss": 0.1253, "step": 2539 }, { "epoch": 1.05, "learning_rate": 9.012830482115085e-06, "loss": 0.0405, "step": 2540 }, { "epoch": 1.05, "learning_rate": 9.01244167962675e-06, "loss": 0.0932, "step": 2541 }, { "epoch": 1.05, "learning_rate": 9.012052877138414e-06, "loss": 0.3505, "step": 2542 }, { "epoch": 1.05, "learning_rate": 9.011664074650078e-06, "loss": 0.111, "step": 2543 }, { "epoch": 1.05, "learning_rate": 9.011275272161743e-06, "loss": 0.1072, "step": 2544 }, { "epoch": 1.05, "learning_rate": 9.010886469673407e-06, "loss": 0.1627, "step": 2545 }, { "epoch": 1.05, "learning_rate": 9.01049766718507e-06, "loss": 0.073, "step": 2546 }, { "epoch": 1.05, "learning_rate": 9.010108864696736e-06, "loss": 0.3296, "step": 2547 }, { "epoch": 1.05, "learning_rate": 9.0097200622084e-06, "loss": 0.0367, "step": 2548 }, { "epoch": 1.05, "learning_rate": 9.009331259720063e-06, "loss": 0.1589, "step": 2549 }, { "epoch": 1.05, "learning_rate": 9.008942457231727e-06, "loss": 0.0595, "step": 2550 }, { "epoch": 1.05, "learning_rate": 9.008553654743392e-06, "loss": 0.1303, "step": 2551 }, { "epoch": 1.05, "learning_rate": 9.008164852255054e-06, "loss": 0.1647, "step": 2552 }, { "epoch": 1.05, "learning_rate": 9.00777604976672e-06, "loss": 0.4254, "step": 2553 }, { "epoch": 1.05, "learning_rate": 9.007387247278383e-06, "loss": 0.4797, "step": 2554 }, { "epoch": 1.05, "learning_rate": 9.006998444790046e-06, "loss": 0.2126, "step": 2555 }, { "epoch": 1.05, "learning_rate": 9.006609642301712e-06, "loss": 0.3049, "step": 2556 }, { "epoch": 1.05, "learning_rate": 9.006220839813375e-06, "loss": 0.245, "step": 2557 }, { "epoch": 1.05, "learning_rate": 9.005832037325039e-06, "loss": 0.0728, "step": 2558 }, { "epoch": 1.05, "learning_rate": 9.005443234836704e-06, "loss": 0.3131, "step": 2559 }, { "epoch": 1.05, "learning_rate": 9.005054432348368e-06, "loss": 0.3286, "step": 2560 }, { "epoch": 1.05, "learning_rate": 9.004665629860032e-06, "loss": 0.2511, "step": 2561 }, { "epoch": 1.05, "learning_rate": 9.004276827371697e-06, "loss": 0.0491, "step": 2562 }, { "epoch": 1.05, "learning_rate": 9.00388802488336e-06, "loss": 0.0436, "step": 2563 }, { "epoch": 1.05, "learning_rate": 9.003499222395024e-06, "loss": 0.0908, "step": 2564 }, { "epoch": 1.05, "learning_rate": 9.003110419906688e-06, "loss": 0.1036, "step": 2565 }, { "epoch": 1.05, "learning_rate": 9.003110419906688e-06, "loss": 0.051, "step": 2566 }, { "epoch": 1.05, "learning_rate": 9.002721617418351e-06, "loss": 0.2821, "step": 2567 }, { "epoch": 1.05, "learning_rate": 9.002332814930017e-06, "loss": 0.1973, "step": 2568 }, { "epoch": 1.05, "learning_rate": 9.00194401244168e-06, "loss": 0.2327, "step": 2569 }, { "epoch": 1.05, "learning_rate": 9.001555209953344e-06, "loss": 0.2189, "step": 2570 }, { "epoch": 1.05, "learning_rate": 9.001166407465008e-06, "loss": 0.0147, "step": 2571 }, { "epoch": 1.05, "learning_rate": 9.000777604976673e-06, "loss": 0.0274, "step": 2572 }, { "epoch": 1.05, "learning_rate": 9.000388802488337e-06, "loss": 0.1116, "step": 2573 }, { "epoch": 1.05, "learning_rate": 9e-06, "loss": 0.1127, "step": 2574 }, { "epoch": 1.05, "learning_rate": 8.999611197511666e-06, "loss": 0.0925, "step": 2575 }, { "epoch": 1.05, "learning_rate": 8.99922239502333e-06, "loss": 0.4764, "step": 2576 }, { "epoch": 1.05, "eval_accuracy": 0.9023569023569024, "eval_loss": 0.25402989983558655, "eval_runtime": 106.8411, "eval_samples_per_second": 2.78, "eval_steps_per_second": 0.14, "step": 2576 } ], "logging_steps": 1, "max_steps": 25720, "num_train_epochs": 9223372036854775807, "save_steps": 500, "total_flos": 6.413494442011066e+19, "trial_name": null, "trial_params": null }