{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9998663637578511, "global_step": 3741, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 9.997326917936381e-08, "loss": 0.0471, "step": 1 }, { "epoch": 0.0, "learning_rate": 9.99465383587276e-08, "loss": 0.0646, "step": 2 }, { "epoch": 0.0, "learning_rate": 9.991980753809142e-08, "loss": 0.0485, "step": 3 }, { "epoch": 0.0, "learning_rate": 9.989307671745521e-08, "loss": 0.0568, "step": 4 }, { "epoch": 0.0, "learning_rate": 9.986634589681903e-08, "loss": 0.0643, "step": 5 }, { "epoch": 0.0, "learning_rate": 9.983961507618284e-08, "loss": 0.0609, "step": 6 }, { "epoch": 0.0, "learning_rate": 9.981288425554664e-08, "loss": 0.0433, "step": 7 }, { "epoch": 0.0, "learning_rate": 9.978615343491045e-08, "loss": 0.0803, "step": 8 }, { "epoch": 0.0, "learning_rate": 9.975942261427426e-08, "loss": 0.0717, "step": 9 }, { "epoch": 0.0, "learning_rate": 9.973269179363806e-08, "loss": 0.0614, "step": 10 }, { "epoch": 0.0, "learning_rate": 9.970596097300187e-08, "loss": 0.0442, "step": 11 }, { "epoch": 0.0, "learning_rate": 9.967923015236567e-08, "loss": 0.077, "step": 12 }, { "epoch": 0.0, "learning_rate": 9.965249933172948e-08, "loss": 0.065, "step": 13 }, { "epoch": 0.0, "learning_rate": 9.962576851109329e-08, "loss": 0.0608, "step": 14 }, { "epoch": 0.0, "learning_rate": 9.959903769045709e-08, "loss": 0.0594, "step": 15 }, { "epoch": 0.0, "learning_rate": 9.95723068698209e-08, "loss": 0.0759, "step": 16 }, { "epoch": 0.0, "learning_rate": 9.954557604918471e-08, "loss": 0.0681, "step": 17 }, { "epoch": 0.0, "learning_rate": 9.951884522854851e-08, "loss": 0.0709, "step": 18 }, { "epoch": 0.01, "learning_rate": 9.949211440791232e-08, "loss": 0.054, "step": 19 }, { "epoch": 0.01, "learning_rate": 9.946538358727612e-08, "loss": 0.0793, "step": 20 }, { "epoch": 0.01, "learning_rate": 9.943865276663993e-08, "loss": 0.0574, "step": 21 }, { "epoch": 0.01, "learning_rate": 9.941192194600374e-08, "loss": 0.0908, "step": 22 }, { "epoch": 0.01, "learning_rate": 9.938519112536754e-08, "loss": 0.0604, "step": 23 }, { "epoch": 0.01, "learning_rate": 9.935846030473135e-08, "loss": 0.0483, "step": 24 }, { "epoch": 0.01, "learning_rate": 9.933172948409516e-08, "loss": 0.0516, "step": 25 }, { "epoch": 0.01, "learning_rate": 9.930499866345896e-08, "loss": 0.0667, "step": 26 }, { "epoch": 0.01, "learning_rate": 9.927826784282277e-08, "loss": 0.0549, "step": 27 }, { "epoch": 0.01, "learning_rate": 9.925153702218657e-08, "loss": 0.0683, "step": 28 }, { "epoch": 0.01, "learning_rate": 9.922480620155038e-08, "loss": 0.0673, "step": 29 }, { "epoch": 0.01, "learning_rate": 9.919807538091419e-08, "loss": 0.0609, "step": 30 }, { "epoch": 0.01, "learning_rate": 9.917134456027799e-08, "loss": 0.0606, "step": 31 }, { "epoch": 0.01, "learning_rate": 9.91446137396418e-08, "loss": 0.0645, "step": 32 }, { "epoch": 0.01, "learning_rate": 9.911788291900561e-08, "loss": 0.0554, "step": 33 }, { "epoch": 0.01, "learning_rate": 9.909115209836941e-08, "loss": 0.0758, "step": 34 }, { "epoch": 0.01, "learning_rate": 9.906442127773322e-08, "loss": 0.0524, "step": 35 }, { "epoch": 0.01, "learning_rate": 9.903769045709702e-08, "loss": 0.0772, "step": 36 }, { "epoch": 0.01, "learning_rate": 9.901095963646083e-08, "loss": 0.0508, "step": 37 }, { "epoch": 0.01, "learning_rate": 9.898422881582464e-08, "loss": 0.0464, "step": 38 }, { "epoch": 0.01, "learning_rate": 9.895749799518844e-08, "loss": 0.0554, "step": 39 }, { "epoch": 0.01, "learning_rate": 9.893076717455225e-08, "loss": 0.0662, "step": 40 }, { "epoch": 0.01, "learning_rate": 9.890403635391606e-08, "loss": 0.07, "step": 41 }, { "epoch": 0.01, "learning_rate": 9.887730553327986e-08, "loss": 0.0616, "step": 42 }, { "epoch": 0.01, "learning_rate": 9.885057471264367e-08, "loss": 0.0406, "step": 43 }, { "epoch": 0.01, "learning_rate": 9.882384389200747e-08, "loss": 0.0452, "step": 44 }, { "epoch": 0.01, "learning_rate": 9.879711307137128e-08, "loss": 0.0582, "step": 45 }, { "epoch": 0.01, "learning_rate": 9.877038225073509e-08, "loss": 0.0712, "step": 46 }, { "epoch": 0.01, "learning_rate": 9.874365143009889e-08, "loss": 0.0499, "step": 47 }, { "epoch": 0.01, "learning_rate": 9.87169206094627e-08, "loss": 0.0545, "step": 48 }, { "epoch": 0.01, "learning_rate": 9.869018978882651e-08, "loss": 0.0811, "step": 49 }, { "epoch": 0.01, "learning_rate": 9.866345896819031e-08, "loss": 0.0604, "step": 50 }, { "epoch": 0.01, "learning_rate": 9.863672814755412e-08, "loss": 0.1121, "step": 51 }, { "epoch": 0.01, "learning_rate": 9.860999732691792e-08, "loss": 0.0809, "step": 52 }, { "epoch": 0.01, "learning_rate": 9.858326650628173e-08, "loss": 0.0485, "step": 53 }, { "epoch": 0.01, "learning_rate": 9.855653568564554e-08, "loss": 0.0559, "step": 54 }, { "epoch": 0.01, "learning_rate": 9.852980486500934e-08, "loss": 0.0611, "step": 55 }, { "epoch": 0.01, "learning_rate": 9.850307404437315e-08, "loss": 0.0583, "step": 56 }, { "epoch": 0.02, "learning_rate": 9.847634322373696e-08, "loss": 0.0869, "step": 57 }, { "epoch": 0.02, "learning_rate": 9.844961240310076e-08, "loss": 0.0737, "step": 58 }, { "epoch": 0.02, "learning_rate": 9.842288158246457e-08, "loss": 0.0612, "step": 59 }, { "epoch": 0.02, "learning_rate": 9.839615076182837e-08, "loss": 0.0463, "step": 60 }, { "epoch": 0.02, "learning_rate": 9.836941994119218e-08, "loss": 0.0462, "step": 61 }, { "epoch": 0.02, "learning_rate": 9.8342689120556e-08, "loss": 0.08, "step": 62 }, { "epoch": 0.02, "learning_rate": 9.831595829991979e-08, "loss": 0.0513, "step": 63 }, { "epoch": 0.02, "learning_rate": 9.82892274792836e-08, "loss": 0.0702, "step": 64 }, { "epoch": 0.02, "learning_rate": 9.826249665864742e-08, "loss": 0.0642, "step": 65 }, { "epoch": 0.02, "learning_rate": 9.823576583801121e-08, "loss": 0.0604, "step": 66 }, { "epoch": 0.02, "learning_rate": 9.820903501737503e-08, "loss": 0.0616, "step": 67 }, { "epoch": 0.02, "learning_rate": 9.818230419673882e-08, "loss": 0.0928, "step": 68 }, { "epoch": 0.02, "learning_rate": 9.815557337610263e-08, "loss": 0.0818, "step": 69 }, { "epoch": 0.02, "learning_rate": 9.812884255546645e-08, "loss": 0.0716, "step": 70 }, { "epoch": 0.02, "learning_rate": 9.810211173483024e-08, "loss": 0.0577, "step": 71 }, { "epoch": 0.02, "learning_rate": 9.807538091419406e-08, "loss": 0.0625, "step": 72 }, { "epoch": 0.02, "learning_rate": 9.804865009355787e-08, "loss": 0.062, "step": 73 }, { "epoch": 0.02, "learning_rate": 9.802191927292167e-08, "loss": 0.0445, "step": 74 }, { "epoch": 0.02, "learning_rate": 9.799518845228548e-08, "loss": 0.0543, "step": 75 }, { "epoch": 0.02, "learning_rate": 9.796845763164929e-08, "loss": 0.0729, "step": 76 }, { "epoch": 0.02, "learning_rate": 9.794172681101309e-08, "loss": 0.0539, "step": 77 }, { "epoch": 0.02, "learning_rate": 9.79149959903769e-08, "loss": 0.0906, "step": 78 }, { "epoch": 0.02, "learning_rate": 9.788826516974071e-08, "loss": 0.0766, "step": 79 }, { "epoch": 0.02, "learning_rate": 9.786153434910451e-08, "loss": 0.0569, "step": 80 }, { "epoch": 0.02, "learning_rate": 9.783480352846832e-08, "loss": 0.0574, "step": 81 }, { "epoch": 0.02, "learning_rate": 9.780807270783213e-08, "loss": 0.0827, "step": 82 }, { "epoch": 0.02, "learning_rate": 9.778134188719593e-08, "loss": 0.0517, "step": 83 }, { "epoch": 0.02, "learning_rate": 9.775461106655974e-08, "loss": 0.0921, "step": 84 }, { "epoch": 0.02, "learning_rate": 9.772788024592355e-08, "loss": 0.0633, "step": 85 }, { "epoch": 0.02, "learning_rate": 9.770114942528735e-08, "loss": 0.0567, "step": 86 }, { "epoch": 0.02, "learning_rate": 9.767441860465116e-08, "loss": 0.098, "step": 87 }, { "epoch": 0.02, "learning_rate": 9.764768778401497e-08, "loss": 0.0567, "step": 88 }, { "epoch": 0.02, "learning_rate": 9.762095696337877e-08, "loss": 0.0498, "step": 89 }, { "epoch": 0.02, "learning_rate": 9.759422614274258e-08, "loss": 0.0674, "step": 90 }, { "epoch": 0.02, "learning_rate": 9.756749532210639e-08, "loss": 0.0695, "step": 91 }, { "epoch": 0.02, "learning_rate": 9.754076450147019e-08, "loss": 0.0618, "step": 92 }, { "epoch": 0.02, "learning_rate": 9.7514033680834e-08, "loss": 0.0637, "step": 93 }, { "epoch": 0.03, "learning_rate": 9.748730286019781e-08, "loss": 0.0389, "step": 94 }, { "epoch": 0.03, "learning_rate": 9.746057203956161e-08, "loss": 0.0681, "step": 95 }, { "epoch": 0.03, "learning_rate": 9.743384121892542e-08, "loss": 0.0508, "step": 96 }, { "epoch": 0.03, "learning_rate": 9.740711039828923e-08, "loss": 0.0496, "step": 97 }, { "epoch": 0.03, "learning_rate": 9.738037957765303e-08, "loss": 0.0544, "step": 98 }, { "epoch": 0.03, "learning_rate": 9.735364875701684e-08, "loss": 0.0687, "step": 99 }, { "epoch": 0.03, "learning_rate": 9.732691793638064e-08, "loss": 0.0718, "step": 100 }, { "epoch": 0.03, "learning_rate": 9.730018711574445e-08, "loss": 0.0595, "step": 101 }, { "epoch": 0.03, "learning_rate": 9.727345629510826e-08, "loss": 0.06, "step": 102 }, { "epoch": 0.03, "learning_rate": 9.724672547447206e-08, "loss": 0.0537, "step": 103 }, { "epoch": 0.03, "learning_rate": 9.721999465383587e-08, "loss": 0.0704, "step": 104 }, { "epoch": 0.03, "learning_rate": 9.719326383319968e-08, "loss": 0.0492, "step": 105 }, { "epoch": 0.03, "learning_rate": 9.716653301256348e-08, "loss": 0.0748, "step": 106 }, { "epoch": 0.03, "learning_rate": 9.71398021919273e-08, "loss": 0.0529, "step": 107 }, { "epoch": 0.03, "learning_rate": 9.711307137129109e-08, "loss": 0.0534, "step": 108 }, { "epoch": 0.03, "learning_rate": 9.70863405506549e-08, "loss": 0.0781, "step": 109 }, { "epoch": 0.03, "learning_rate": 9.705960973001872e-08, "loss": 0.0735, "step": 110 }, { "epoch": 0.03, "learning_rate": 9.703287890938251e-08, "loss": 0.0721, "step": 111 }, { "epoch": 0.03, "learning_rate": 9.700614808874632e-08, "loss": 0.0679, "step": 112 }, { "epoch": 0.03, "learning_rate": 9.697941726811014e-08, "loss": 0.097, "step": 113 }, { "epoch": 0.03, "learning_rate": 9.695268644747393e-08, "loss": 0.0686, "step": 114 }, { "epoch": 0.03, "learning_rate": 9.692595562683775e-08, "loss": 0.0652, "step": 115 }, { "epoch": 0.03, "learning_rate": 9.689922480620154e-08, "loss": 0.07, "step": 116 }, { "epoch": 0.03, "learning_rate": 9.687249398556535e-08, "loss": 0.0518, "step": 117 }, { "epoch": 0.03, "learning_rate": 9.684576316492917e-08, "loss": 0.06, "step": 118 }, { "epoch": 0.03, "learning_rate": 9.681903234429296e-08, "loss": 0.0704, "step": 119 }, { "epoch": 0.03, "learning_rate": 9.679230152365678e-08, "loss": 0.042, "step": 120 }, { "epoch": 0.03, "learning_rate": 9.676557070302059e-08, "loss": 0.0755, "step": 121 }, { "epoch": 0.03, "learning_rate": 9.673883988238439e-08, "loss": 0.0556, "step": 122 }, { "epoch": 0.03, "learning_rate": 9.67121090617482e-08, "loss": 0.0534, "step": 123 }, { "epoch": 0.03, "learning_rate": 9.6685378241112e-08, "loss": 0.0644, "step": 124 }, { "epoch": 0.03, "learning_rate": 9.66586474204758e-08, "loss": 0.0484, "step": 125 }, { "epoch": 0.03, "learning_rate": 9.663191659983962e-08, "loss": 0.0798, "step": 126 }, { "epoch": 0.03, "learning_rate": 9.660518577920342e-08, "loss": 0.053, "step": 127 }, { "epoch": 0.03, "learning_rate": 9.657845495856723e-08, "loss": 0.0521, "step": 128 }, { "epoch": 0.03, "learning_rate": 9.655172413793104e-08, "loss": 0.0494, "step": 129 }, { "epoch": 0.03, "learning_rate": 9.652499331729484e-08, "loss": 0.0691, "step": 130 }, { "epoch": 0.04, "learning_rate": 9.649826249665865e-08, "loss": 0.0489, "step": 131 }, { "epoch": 0.04, "learning_rate": 9.647153167602245e-08, "loss": 0.0885, "step": 132 }, { "epoch": 0.04, "learning_rate": 9.644480085538626e-08, "loss": 0.0679, "step": 133 }, { "epoch": 0.04, "learning_rate": 9.641807003475007e-08, "loss": 0.0791, "step": 134 }, { "epoch": 0.04, "learning_rate": 9.639133921411387e-08, "loss": 0.0589, "step": 135 }, { "epoch": 0.04, "learning_rate": 9.636460839347768e-08, "loss": 0.0515, "step": 136 }, { "epoch": 0.04, "learning_rate": 9.633787757284149e-08, "loss": 0.074, "step": 137 }, { "epoch": 0.04, "learning_rate": 9.631114675220529e-08, "loss": 0.0604, "step": 138 }, { "epoch": 0.04, "learning_rate": 9.62844159315691e-08, "loss": 0.0586, "step": 139 }, { "epoch": 0.04, "learning_rate": 9.62576851109329e-08, "loss": 0.0615, "step": 140 }, { "epoch": 0.04, "learning_rate": 9.623095429029671e-08, "loss": 0.0739, "step": 141 }, { "epoch": 0.04, "learning_rate": 9.620422346966052e-08, "loss": 0.0581, "step": 142 }, { "epoch": 0.04, "learning_rate": 9.617749264902432e-08, "loss": 0.0401, "step": 143 }, { "epoch": 0.04, "learning_rate": 9.615076182838813e-08, "loss": 0.052, "step": 144 }, { "epoch": 0.04, "learning_rate": 9.612403100775194e-08, "loss": 0.0584, "step": 145 }, { "epoch": 0.04, "learning_rate": 9.609730018711574e-08, "loss": 0.0658, "step": 146 }, { "epoch": 0.04, "learning_rate": 9.607056936647955e-08, "loss": 0.0594, "step": 147 }, { "epoch": 0.04, "learning_rate": 9.604383854584335e-08, "loss": 0.0668, "step": 148 }, { "epoch": 0.04, "learning_rate": 9.601710772520716e-08, "loss": 0.0585, "step": 149 }, { "epoch": 0.04, "learning_rate": 9.599037690457097e-08, "loss": 0.0487, "step": 150 }, { "epoch": 0.04, "learning_rate": 9.596364608393477e-08, "loss": 0.0703, "step": 151 }, { "epoch": 0.04, "learning_rate": 9.593691526329858e-08, "loss": 0.0552, "step": 152 }, { "epoch": 0.04, "learning_rate": 9.591018444266239e-08, "loss": 0.0699, "step": 153 }, { "epoch": 0.04, "learning_rate": 9.588345362202619e-08, "loss": 0.0712, "step": 154 }, { "epoch": 0.04, "learning_rate": 9.585672280139e-08, "loss": 0.0524, "step": 155 }, { "epoch": 0.04, "learning_rate": 9.58299919807538e-08, "loss": 0.06, "step": 156 }, { "epoch": 0.04, "learning_rate": 9.580326116011761e-08, "loss": 0.0766, "step": 157 }, { "epoch": 0.04, "learning_rate": 9.577653033948142e-08, "loss": 0.0825, "step": 158 }, { "epoch": 0.04, "learning_rate": 9.574979951884522e-08, "loss": 0.0552, "step": 159 }, { "epoch": 0.04, "learning_rate": 9.572306869820903e-08, "loss": 0.053, "step": 160 }, { "epoch": 0.04, "learning_rate": 9.569633787757284e-08, "loss": 0.0583, "step": 161 }, { "epoch": 0.04, "learning_rate": 9.566960705693664e-08, "loss": 0.0546, "step": 162 }, { "epoch": 0.04, "learning_rate": 9.564287623630045e-08, "loss": 0.0797, "step": 163 }, { "epoch": 0.04, "learning_rate": 9.561614541566425e-08, "loss": 0.0633, "step": 164 }, { "epoch": 0.04, "learning_rate": 9.558941459502806e-08, "loss": 0.0618, "step": 165 }, { "epoch": 0.04, "learning_rate": 9.556268377439187e-08, "loss": 0.0556, "step": 166 }, { "epoch": 0.04, "learning_rate": 9.553595295375567e-08, "loss": 0.0797, "step": 167 }, { "epoch": 0.04, "learning_rate": 9.550922213311948e-08, "loss": 0.0559, "step": 168 }, { "epoch": 0.05, "learning_rate": 9.54824913124833e-08, "loss": 0.0693, "step": 169 }, { "epoch": 0.05, "learning_rate": 9.545576049184709e-08, "loss": 0.05, "step": 170 }, { "epoch": 0.05, "learning_rate": 9.54290296712109e-08, "loss": 0.0548, "step": 171 }, { "epoch": 0.05, "learning_rate": 9.54022988505747e-08, "loss": 0.0785, "step": 172 }, { "epoch": 0.05, "learning_rate": 9.537556802993851e-08, "loss": 0.0833, "step": 173 }, { "epoch": 0.05, "learning_rate": 9.534883720930232e-08, "loss": 0.0501, "step": 174 }, { "epoch": 0.05, "learning_rate": 9.532210638866612e-08, "loss": 0.0677, "step": 175 }, { "epoch": 0.05, "learning_rate": 9.529537556802993e-08, "loss": 0.057, "step": 176 }, { "epoch": 0.05, "learning_rate": 9.526864474739375e-08, "loss": 0.0661, "step": 177 }, { "epoch": 0.05, "learning_rate": 9.524191392675754e-08, "loss": 0.0839, "step": 178 }, { "epoch": 0.05, "learning_rate": 9.521518310612135e-08, "loss": 0.0536, "step": 179 }, { "epoch": 0.05, "learning_rate": 9.518845228548515e-08, "loss": 0.0536, "step": 180 }, { "epoch": 0.05, "learning_rate": 9.516172146484896e-08, "loss": 0.055, "step": 181 }, { "epoch": 0.05, "learning_rate": 9.513499064421278e-08, "loss": 0.0912, "step": 182 }, { "epoch": 0.05, "learning_rate": 9.510825982357657e-08, "loss": 0.0423, "step": 183 }, { "epoch": 0.05, "learning_rate": 9.508152900294038e-08, "loss": 0.0509, "step": 184 }, { "epoch": 0.05, "learning_rate": 9.50547981823042e-08, "loss": 0.0719, "step": 185 }, { "epoch": 0.05, "learning_rate": 9.5028067361668e-08, "loss": 0.0631, "step": 186 }, { "epoch": 0.05, "learning_rate": 9.50013365410318e-08, "loss": 0.0522, "step": 187 }, { "epoch": 0.05, "learning_rate": 9.49746057203956e-08, "loss": 0.0647, "step": 188 }, { "epoch": 0.05, "learning_rate": 9.494787489975942e-08, "loss": 0.061, "step": 189 }, { "epoch": 0.05, "learning_rate": 9.492114407912323e-08, "loss": 0.0662, "step": 190 }, { "epoch": 0.05, "learning_rate": 9.489441325848702e-08, "loss": 0.0712, "step": 191 }, { "epoch": 0.05, "learning_rate": 9.486768243785084e-08, "loss": 0.0564, "step": 192 }, { "epoch": 0.05, "learning_rate": 9.484095161721465e-08, "loss": 0.0823, "step": 193 }, { "epoch": 0.05, "learning_rate": 9.481422079657845e-08, "loss": 0.079, "step": 194 }, { "epoch": 0.05, "learning_rate": 9.478748997594226e-08, "loss": 0.0544, "step": 195 }, { "epoch": 0.05, "learning_rate": 9.476075915530606e-08, "loss": 0.0769, "step": 196 }, { "epoch": 0.05, "learning_rate": 9.473402833466987e-08, "loss": 0.0731, "step": 197 }, { "epoch": 0.05, "learning_rate": 9.470729751403368e-08, "loss": 0.0701, "step": 198 }, { "epoch": 0.05, "learning_rate": 9.468056669339748e-08, "loss": 0.0726, "step": 199 }, { "epoch": 0.05, "learning_rate": 9.465383587276129e-08, "loss": 0.0672, "step": 200 }, { "epoch": 0.05, "learning_rate": 9.46271050521251e-08, "loss": 0.0743, "step": 201 }, { "epoch": 0.05, "learning_rate": 9.46003742314889e-08, "loss": 0.0661, "step": 202 }, { "epoch": 0.05, "learning_rate": 9.457364341085271e-08, "loss": 0.0777, "step": 203 }, { "epoch": 0.05, "learning_rate": 9.45469125902165e-08, "loss": 0.0694, "step": 204 }, { "epoch": 0.05, "learning_rate": 9.452018176958032e-08, "loss": 0.0923, "step": 205 }, { "epoch": 0.06, "learning_rate": 9.449345094894413e-08, "loss": 0.0505, "step": 206 }, { "epoch": 0.06, "learning_rate": 9.446672012830793e-08, "loss": 0.0682, "step": 207 }, { "epoch": 0.06, "learning_rate": 9.443998930767174e-08, "loss": 0.0605, "step": 208 }, { "epoch": 0.06, "learning_rate": 9.441325848703555e-08, "loss": 0.0698, "step": 209 }, { "epoch": 0.06, "learning_rate": 9.438652766639935e-08, "loss": 0.0449, "step": 210 }, { "epoch": 0.06, "learning_rate": 9.435979684576316e-08, "loss": 0.0509, "step": 211 }, { "epoch": 0.06, "learning_rate": 9.433306602512696e-08, "loss": 0.0664, "step": 212 }, { "epoch": 0.06, "learning_rate": 9.430633520449077e-08, "loss": 0.053, "step": 213 }, { "epoch": 0.06, "learning_rate": 9.427960438385458e-08, "loss": 0.0474, "step": 214 }, { "epoch": 0.06, "learning_rate": 9.425287356321838e-08, "loss": 0.0806, "step": 215 }, { "epoch": 0.06, "learning_rate": 9.422614274258219e-08, "loss": 0.0787, "step": 216 }, { "epoch": 0.06, "learning_rate": 9.4199411921946e-08, "loss": 0.0557, "step": 217 }, { "epoch": 0.06, "learning_rate": 9.41726811013098e-08, "loss": 0.0571, "step": 218 }, { "epoch": 0.06, "learning_rate": 9.414595028067361e-08, "loss": 0.0387, "step": 219 }, { "epoch": 0.06, "learning_rate": 9.411921946003741e-08, "loss": 0.0506, "step": 220 }, { "epoch": 0.06, "learning_rate": 9.409248863940122e-08, "loss": 0.0607, "step": 221 }, { "epoch": 0.06, "learning_rate": 9.406575781876503e-08, "loss": 0.0797, "step": 222 }, { "epoch": 0.06, "learning_rate": 9.403902699812883e-08, "loss": 0.0887, "step": 223 }, { "epoch": 0.06, "learning_rate": 9.401229617749264e-08, "loss": 0.0933, "step": 224 }, { "epoch": 0.06, "learning_rate": 9.398556535685645e-08, "loss": 0.0661, "step": 225 }, { "epoch": 0.06, "learning_rate": 9.395883453622025e-08, "loss": 0.0473, "step": 226 }, { "epoch": 0.06, "learning_rate": 9.393210371558406e-08, "loss": 0.0473, "step": 227 }, { "epoch": 0.06, "learning_rate": 9.390537289494786e-08, "loss": 0.0573, "step": 228 }, { "epoch": 0.06, "learning_rate": 9.387864207431167e-08, "loss": 0.0897, "step": 229 }, { "epoch": 0.06, "learning_rate": 9.385191125367548e-08, "loss": 0.0723, "step": 230 }, { "epoch": 0.06, "learning_rate": 9.382518043303928e-08, "loss": 0.0546, "step": 231 }, { "epoch": 0.06, "learning_rate": 9.379844961240309e-08, "loss": 0.0522, "step": 232 }, { "epoch": 0.06, "learning_rate": 9.37717187917669e-08, "loss": 0.0593, "step": 233 }, { "epoch": 0.06, "learning_rate": 9.37449879711307e-08, "loss": 0.0645, "step": 234 }, { "epoch": 0.06, "learning_rate": 9.371825715049451e-08, "loss": 0.0646, "step": 235 }, { "epoch": 0.06, "learning_rate": 9.369152632985831e-08, "loss": 0.0673, "step": 236 }, { "epoch": 0.06, "learning_rate": 9.366479550922212e-08, "loss": 0.0604, "step": 237 }, { "epoch": 0.06, "learning_rate": 9.363806468858593e-08, "loss": 0.0474, "step": 238 }, { "epoch": 0.06, "learning_rate": 9.361133386794973e-08, "loss": 0.0633, "step": 239 }, { "epoch": 0.06, "learning_rate": 9.358460304731354e-08, "loss": 0.0569, "step": 240 }, { "epoch": 0.06, "learning_rate": 9.355787222667735e-08, "loss": 0.0631, "step": 241 }, { "epoch": 0.06, "learning_rate": 9.353114140604115e-08, "loss": 0.044, "step": 242 }, { "epoch": 0.06, "learning_rate": 9.350441058540496e-08, "loss": 0.068, "step": 243 }, { "epoch": 0.07, "learning_rate": 9.347767976476878e-08, "loss": 0.0631, "step": 244 }, { "epoch": 0.07, "learning_rate": 9.345094894413257e-08, "loss": 0.0571, "step": 245 }, { "epoch": 0.07, "learning_rate": 9.342421812349638e-08, "loss": 0.0691, "step": 246 }, { "epoch": 0.07, "learning_rate": 9.33974873028602e-08, "loss": 0.0614, "step": 247 }, { "epoch": 0.07, "learning_rate": 9.3370756482224e-08, "loss": 0.05, "step": 248 }, { "epoch": 0.07, "learning_rate": 9.33440256615878e-08, "loss": 0.0571, "step": 249 }, { "epoch": 0.07, "learning_rate": 9.331729484095162e-08, "loss": 0.0535, "step": 250 }, { "epoch": 0.07, "learning_rate": 9.329056402031541e-08, "loss": 0.069, "step": 251 }, { "epoch": 0.07, "learning_rate": 9.326383319967923e-08, "loss": 0.0602, "step": 252 }, { "epoch": 0.07, "learning_rate": 9.323710237904304e-08, "loss": 0.0595, "step": 253 }, { "epoch": 0.07, "learning_rate": 9.321037155840684e-08, "loss": 0.0608, "step": 254 }, { "epoch": 0.07, "learning_rate": 9.318364073777065e-08, "loss": 0.0583, "step": 255 }, { "epoch": 0.07, "learning_rate": 9.315690991713446e-08, "loss": 0.0762, "step": 256 }, { "epoch": 0.07, "learning_rate": 9.313017909649826e-08, "loss": 0.0624, "step": 257 }, { "epoch": 0.07, "learning_rate": 9.310344827586207e-08, "loss": 0.0693, "step": 258 }, { "epoch": 0.07, "learning_rate": 9.307671745522588e-08, "loss": 0.0526, "step": 259 }, { "epoch": 0.07, "learning_rate": 9.304998663458968e-08, "loss": 0.0576, "step": 260 }, { "epoch": 0.07, "learning_rate": 9.302325581395349e-08, "loss": 0.0619, "step": 261 }, { "epoch": 0.07, "learning_rate": 9.299652499331729e-08, "loss": 0.0679, "step": 262 }, { "epoch": 0.07, "learning_rate": 9.29697941726811e-08, "loss": 0.0515, "step": 263 }, { "epoch": 0.07, "learning_rate": 9.294306335204491e-08, "loss": 0.0648, "step": 264 }, { "epoch": 0.07, "learning_rate": 9.291633253140871e-08, "loss": 0.0615, "step": 265 }, { "epoch": 0.07, "learning_rate": 9.288960171077252e-08, "loss": 0.0581, "step": 266 }, { "epoch": 0.07, "learning_rate": 9.286287089013633e-08, "loss": 0.0659, "step": 267 }, { "epoch": 0.07, "learning_rate": 9.283614006950013e-08, "loss": 0.0611, "step": 268 }, { "epoch": 0.07, "learning_rate": 9.280940924886394e-08, "loss": 0.0789, "step": 269 }, { "epoch": 0.07, "learning_rate": 9.278267842822775e-08, "loss": 0.0491, "step": 270 }, { "epoch": 0.07, "learning_rate": 9.275594760759155e-08, "loss": 0.0809, "step": 271 }, { "epoch": 0.07, "learning_rate": 9.272921678695536e-08, "loss": 0.0619, "step": 272 }, { "epoch": 0.07, "learning_rate": 9.270248596631917e-08, "loss": 0.056, "step": 273 }, { "epoch": 0.07, "learning_rate": 9.267575514568297e-08, "loss": 0.0699, "step": 274 }, { "epoch": 0.07, "learning_rate": 9.264902432504678e-08, "loss": 0.0512, "step": 275 }, { "epoch": 0.07, "learning_rate": 9.262229350441058e-08, "loss": 0.0722, "step": 276 }, { "epoch": 0.07, "learning_rate": 9.259556268377439e-08, "loss": 0.0643, "step": 277 }, { "epoch": 0.07, "learning_rate": 9.25688318631382e-08, "loss": 0.0499, "step": 278 }, { "epoch": 0.07, "learning_rate": 9.2542101042502e-08, "loss": 0.0736, "step": 279 }, { "epoch": 0.07, "learning_rate": 9.251537022186581e-08, "loss": 0.0676, "step": 280 }, { "epoch": 0.08, "learning_rate": 9.248863940122962e-08, "loss": 0.0596, "step": 281 }, { "epoch": 0.08, "learning_rate": 9.246190858059342e-08, "loss": 0.0772, "step": 282 }, { "epoch": 0.08, "learning_rate": 9.243517775995723e-08, "loss": 0.0485, "step": 283 }, { "epoch": 0.08, "learning_rate": 9.240844693932103e-08, "loss": 0.0695, "step": 284 }, { "epoch": 0.08, "learning_rate": 9.238171611868484e-08, "loss": 0.0853, "step": 285 }, { "epoch": 0.08, "learning_rate": 9.235498529804865e-08, "loss": 0.0721, "step": 286 }, { "epoch": 0.08, "learning_rate": 9.232825447741245e-08, "loss": 0.0621, "step": 287 }, { "epoch": 0.08, "learning_rate": 9.230152365677626e-08, "loss": 0.0332, "step": 288 }, { "epoch": 0.08, "learning_rate": 9.227479283614007e-08, "loss": 0.0576, "step": 289 }, { "epoch": 0.08, "learning_rate": 9.224806201550387e-08, "loss": 0.0684, "step": 290 }, { "epoch": 0.08, "learning_rate": 9.222133119486768e-08, "loss": 0.0714, "step": 291 }, { "epoch": 0.08, "learning_rate": 9.219460037423148e-08, "loss": 0.0536, "step": 292 }, { "epoch": 0.08, "learning_rate": 9.216786955359529e-08, "loss": 0.0468, "step": 293 }, { "epoch": 0.08, "learning_rate": 9.21411387329591e-08, "loss": 0.0628, "step": 294 }, { "epoch": 0.08, "learning_rate": 9.21144079123229e-08, "loss": 0.0658, "step": 295 }, { "epoch": 0.08, "learning_rate": 9.208767709168671e-08, "loss": 0.0436, "step": 296 }, { "epoch": 0.08, "learning_rate": 9.206094627105053e-08, "loss": 0.07, "step": 297 }, { "epoch": 0.08, "learning_rate": 9.203421545041432e-08, "loss": 0.0367, "step": 298 }, { "epoch": 0.08, "learning_rate": 9.200748462977814e-08, "loss": 0.0711, "step": 299 }, { "epoch": 0.08, "learning_rate": 9.198075380914193e-08, "loss": 0.0618, "step": 300 }, { "epoch": 0.08, "learning_rate": 9.195402298850574e-08, "loss": 0.0666, "step": 301 }, { "epoch": 0.08, "learning_rate": 9.192729216786956e-08, "loss": 0.0551, "step": 302 }, { "epoch": 0.08, "learning_rate": 9.190056134723335e-08, "loss": 0.0711, "step": 303 }, { "epoch": 0.08, "learning_rate": 9.187383052659717e-08, "loss": 0.0853, "step": 304 }, { "epoch": 0.08, "learning_rate": 9.184709970596098e-08, "loss": 0.0642, "step": 305 }, { "epoch": 0.08, "learning_rate": 9.182036888532477e-08, "loss": 0.0574, "step": 306 }, { "epoch": 0.08, "learning_rate": 9.179363806468859e-08, "loss": 0.0877, "step": 307 }, { "epoch": 0.08, "learning_rate": 9.176690724405238e-08, "loss": 0.0522, "step": 308 }, { "epoch": 0.08, "learning_rate": 9.17401764234162e-08, "loss": 0.0669, "step": 309 }, { "epoch": 0.08, "learning_rate": 9.171344560278001e-08, "loss": 0.0565, "step": 310 }, { "epoch": 0.08, "learning_rate": 9.16867147821438e-08, "loss": 0.0656, "step": 311 }, { "epoch": 0.08, "learning_rate": 9.165998396150762e-08, "loss": 0.0587, "step": 312 }, { "epoch": 0.08, "learning_rate": 9.163325314087143e-08, "loss": 0.0571, "step": 313 }, { "epoch": 0.08, "learning_rate": 9.160652232023523e-08, "loss": 0.0594, "step": 314 }, { "epoch": 0.08, "learning_rate": 9.157979149959904e-08, "loss": 0.0533, "step": 315 }, { "epoch": 0.08, "learning_rate": 9.155306067896284e-08, "loss": 0.0762, "step": 316 }, { "epoch": 0.08, "learning_rate": 9.152632985832665e-08, "loss": 0.0608, "step": 317 }, { "epoch": 0.08, "learning_rate": 9.149959903769046e-08, "loss": 0.0796, "step": 318 }, { "epoch": 0.09, "learning_rate": 9.147286821705426e-08, "loss": 0.0811, "step": 319 }, { "epoch": 0.09, "learning_rate": 9.144613739641807e-08, "loss": 0.0725, "step": 320 }, { "epoch": 0.09, "learning_rate": 9.141940657578188e-08, "loss": 0.0701, "step": 321 }, { "epoch": 0.09, "learning_rate": 9.139267575514568e-08, "loss": 0.0639, "step": 322 }, { "epoch": 0.09, "learning_rate": 9.136594493450949e-08, "loss": 0.0654, "step": 323 }, { "epoch": 0.09, "learning_rate": 9.133921411387329e-08, "loss": 0.0583, "step": 324 }, { "epoch": 0.09, "learning_rate": 9.13124832932371e-08, "loss": 0.0918, "step": 325 }, { "epoch": 0.09, "learning_rate": 9.128575247260091e-08, "loss": 0.0503, "step": 326 }, { "epoch": 0.09, "learning_rate": 9.125902165196471e-08, "loss": 0.0732, "step": 327 }, { "epoch": 0.09, "learning_rate": 9.123229083132852e-08, "loss": 0.0717, "step": 328 }, { "epoch": 0.09, "learning_rate": 9.120556001069233e-08, "loss": 0.0607, "step": 329 }, { "epoch": 0.09, "learning_rate": 9.117882919005613e-08, "loss": 0.0634, "step": 330 }, { "epoch": 0.09, "learning_rate": 9.115209836941994e-08, "loss": 0.0559, "step": 331 }, { "epoch": 0.09, "learning_rate": 9.112536754878374e-08, "loss": 0.0527, "step": 332 }, { "epoch": 0.09, "learning_rate": 9.109863672814755e-08, "loss": 0.0666, "step": 333 }, { "epoch": 0.09, "learning_rate": 9.107190590751136e-08, "loss": 0.0499, "step": 334 }, { "epoch": 0.09, "learning_rate": 9.104517508687516e-08, "loss": 0.0603, "step": 335 }, { "epoch": 0.09, "learning_rate": 9.101844426623897e-08, "loss": 0.0497, "step": 336 }, { "epoch": 0.09, "learning_rate": 9.099171344560278e-08, "loss": 0.0938, "step": 337 }, { "epoch": 0.09, "learning_rate": 9.096498262496658e-08, "loss": 0.0678, "step": 338 }, { "epoch": 0.09, "learning_rate": 9.093825180433039e-08, "loss": 0.0417, "step": 339 }, { "epoch": 0.09, "learning_rate": 9.091152098369419e-08, "loss": 0.053, "step": 340 }, { "epoch": 0.09, "learning_rate": 9.0884790163058e-08, "loss": 0.0494, "step": 341 }, { "epoch": 0.09, "learning_rate": 9.085805934242181e-08, "loss": 0.0695, "step": 342 }, { "epoch": 0.09, "learning_rate": 9.083132852178561e-08, "loss": 0.0877, "step": 343 }, { "epoch": 0.09, "learning_rate": 9.080459770114942e-08, "loss": 0.0708, "step": 344 }, { "epoch": 0.09, "learning_rate": 9.077786688051323e-08, "loss": 0.0736, "step": 345 }, { "epoch": 0.09, "learning_rate": 9.075113605987703e-08, "loss": 0.0597, "step": 346 }, { "epoch": 0.09, "learning_rate": 9.072440523924084e-08, "loss": 0.069, "step": 347 }, { "epoch": 0.09, "learning_rate": 9.069767441860464e-08, "loss": 0.0534, "step": 348 }, { "epoch": 0.09, "learning_rate": 9.067094359796845e-08, "loss": 0.0701, "step": 349 }, { "epoch": 0.09, "learning_rate": 9.064421277733226e-08, "loss": 0.0427, "step": 350 }, { "epoch": 0.09, "learning_rate": 9.061748195669606e-08, "loss": 0.074, "step": 351 }, { "epoch": 0.09, "learning_rate": 9.059075113605987e-08, "loss": 0.0826, "step": 352 }, { "epoch": 0.09, "learning_rate": 9.056402031542368e-08, "loss": 0.0649, "step": 353 }, { "epoch": 0.09, "learning_rate": 9.053728949478748e-08, "loss": 0.0556, "step": 354 }, { "epoch": 0.09, "learning_rate": 9.051055867415129e-08, "loss": 0.0576, "step": 355 }, { "epoch": 0.1, "learning_rate": 9.048382785351509e-08, "loss": 0.0539, "step": 356 }, { "epoch": 0.1, "learning_rate": 9.04570970328789e-08, "loss": 0.0468, "step": 357 }, { "epoch": 0.1, "learning_rate": 9.043036621224271e-08, "loss": 0.0639, "step": 358 }, { "epoch": 0.1, "learning_rate": 9.040363539160651e-08, "loss": 0.0714, "step": 359 }, { "epoch": 0.1, "learning_rate": 9.037690457097032e-08, "loss": 0.0638, "step": 360 }, { "epoch": 0.1, "learning_rate": 9.035017375033413e-08, "loss": 0.0591, "step": 361 }, { "epoch": 0.1, "learning_rate": 9.032344292969793e-08, "loss": 0.0605, "step": 362 }, { "epoch": 0.1, "learning_rate": 9.029671210906174e-08, "loss": 0.0606, "step": 363 }, { "epoch": 0.1, "learning_rate": 9.026998128842554e-08, "loss": 0.0682, "step": 364 }, { "epoch": 0.1, "learning_rate": 9.024325046778935e-08, "loss": 0.0803, "step": 365 }, { "epoch": 0.1, "learning_rate": 9.021651964715316e-08, "loss": 0.0705, "step": 366 }, { "epoch": 0.1, "learning_rate": 9.018978882651696e-08, "loss": 0.0617, "step": 367 }, { "epoch": 0.1, "learning_rate": 9.016305800588077e-08, "loss": 0.088, "step": 368 }, { "epoch": 0.1, "learning_rate": 9.013632718524459e-08, "loss": 0.0739, "step": 369 }, { "epoch": 0.1, "learning_rate": 9.010959636460838e-08, "loss": 0.0786, "step": 370 }, { "epoch": 0.1, "learning_rate": 9.00828655439722e-08, "loss": 0.0622, "step": 371 }, { "epoch": 0.1, "learning_rate": 9.0056134723336e-08, "loss": 0.0506, "step": 372 }, { "epoch": 0.1, "learning_rate": 9.00294039026998e-08, "loss": 0.0545, "step": 373 }, { "epoch": 0.1, "learning_rate": 9.000267308206362e-08, "loss": 0.0581, "step": 374 }, { "epoch": 0.1, "learning_rate": 8.997594226142741e-08, "loss": 0.0709, "step": 375 }, { "epoch": 0.1, "learning_rate": 8.994921144079123e-08, "loss": 0.0809, "step": 376 }, { "epoch": 0.1, "learning_rate": 8.992248062015504e-08, "loss": 0.0531, "step": 377 }, { "epoch": 0.1, "learning_rate": 8.989574979951884e-08, "loss": 0.0569, "step": 378 }, { "epoch": 0.1, "learning_rate": 8.986901897888265e-08, "loss": 0.0773, "step": 379 }, { "epoch": 0.1, "learning_rate": 8.984228815824644e-08, "loss": 0.0465, "step": 380 }, { "epoch": 0.1, "learning_rate": 8.981555733761026e-08, "loss": 0.0756, "step": 381 }, { "epoch": 0.1, "learning_rate": 8.978882651697407e-08, "loss": 0.0596, "step": 382 }, { "epoch": 0.1, "learning_rate": 8.976209569633787e-08, "loss": 0.0667, "step": 383 }, { "epoch": 0.1, "learning_rate": 8.973536487570168e-08, "loss": 0.0657, "step": 384 }, { "epoch": 0.1, "learning_rate": 8.970863405506549e-08, "loss": 0.0531, "step": 385 }, { "epoch": 0.1, "learning_rate": 8.968190323442929e-08, "loss": 0.048, "step": 386 }, { "epoch": 0.1, "learning_rate": 8.96551724137931e-08, "loss": 0.0572, "step": 387 }, { "epoch": 0.1, "learning_rate": 8.96284415931569e-08, "loss": 0.0731, "step": 388 }, { "epoch": 0.1, "learning_rate": 8.960171077252071e-08, "loss": 0.0519, "step": 389 }, { "epoch": 0.1, "learning_rate": 8.957497995188452e-08, "loss": 0.0729, "step": 390 }, { "epoch": 0.1, "learning_rate": 8.954824913124832e-08, "loss": 0.0529, "step": 391 }, { "epoch": 0.1, "learning_rate": 8.952151831061213e-08, "loss": 0.0796, "step": 392 }, { "epoch": 0.11, "learning_rate": 8.949478748997594e-08, "loss": 0.0845, "step": 393 }, { "epoch": 0.11, "learning_rate": 8.946805666933974e-08, "loss": 0.076, "step": 394 }, { "epoch": 0.11, "learning_rate": 8.944132584870355e-08, "loss": 0.0742, "step": 395 }, { "epoch": 0.11, "learning_rate": 8.941459502806735e-08, "loss": 0.0593, "step": 396 }, { "epoch": 0.11, "learning_rate": 8.938786420743116e-08, "loss": 0.077, "step": 397 }, { "epoch": 0.11, "learning_rate": 8.936113338679497e-08, "loss": 0.0675, "step": 398 }, { "epoch": 0.11, "learning_rate": 8.933440256615877e-08, "loss": 0.0643, "step": 399 }, { "epoch": 0.11, "learning_rate": 8.930767174552258e-08, "loss": 0.0719, "step": 400 }, { "epoch": 0.11, "learning_rate": 8.928094092488639e-08, "loss": 0.0675, "step": 401 }, { "epoch": 0.11, "learning_rate": 8.925421010425019e-08, "loss": 0.0485, "step": 402 }, { "epoch": 0.11, "learning_rate": 8.9227479283614e-08, "loss": 0.0613, "step": 403 }, { "epoch": 0.11, "learning_rate": 8.92007484629778e-08, "loss": 0.0661, "step": 404 }, { "epoch": 0.11, "learning_rate": 8.917401764234161e-08, "loss": 0.0902, "step": 405 }, { "epoch": 0.11, "learning_rate": 8.914728682170542e-08, "loss": 0.0712, "step": 406 }, { "epoch": 0.11, "learning_rate": 8.912055600106922e-08, "loss": 0.0572, "step": 407 }, { "epoch": 0.11, "learning_rate": 8.909382518043303e-08, "loss": 0.0628, "step": 408 }, { "epoch": 0.11, "learning_rate": 8.906709435979684e-08, "loss": 0.0609, "step": 409 }, { "epoch": 0.11, "learning_rate": 8.904036353916064e-08, "loss": 0.0694, "step": 410 }, { "epoch": 0.11, "learning_rate": 8.901363271852445e-08, "loss": 0.0637, "step": 411 }, { "epoch": 0.11, "learning_rate": 8.898690189788826e-08, "loss": 0.0582, "step": 412 }, { "epoch": 0.11, "learning_rate": 8.896017107725206e-08, "loss": 0.0544, "step": 413 }, { "epoch": 0.11, "learning_rate": 8.893344025661587e-08, "loss": 0.059, "step": 414 }, { "epoch": 0.11, "learning_rate": 8.890670943597967e-08, "loss": 0.0679, "step": 415 }, { "epoch": 0.11, "learning_rate": 8.887997861534348e-08, "loss": 0.0393, "step": 416 }, { "epoch": 0.11, "learning_rate": 8.885324779470729e-08, "loss": 0.0552, "step": 417 }, { "epoch": 0.11, "learning_rate": 8.882651697407109e-08, "loss": 0.071, "step": 418 }, { "epoch": 0.11, "learning_rate": 8.87997861534349e-08, "loss": 0.0812, "step": 419 }, { "epoch": 0.11, "learning_rate": 8.877305533279871e-08, "loss": 0.0874, "step": 420 }, { "epoch": 0.11, "learning_rate": 8.874632451216251e-08, "loss": 0.0524, "step": 421 }, { "epoch": 0.11, "learning_rate": 8.871959369152632e-08, "loss": 0.057, "step": 422 }, { "epoch": 0.11, "learning_rate": 8.869286287089013e-08, "loss": 0.0631, "step": 423 }, { "epoch": 0.11, "learning_rate": 8.866613205025393e-08, "loss": 0.0614, "step": 424 }, { "epoch": 0.11, "learning_rate": 8.863940122961774e-08, "loss": 0.0816, "step": 425 }, { "epoch": 0.11, "learning_rate": 8.861267040898156e-08, "loss": 0.0787, "step": 426 }, { "epoch": 0.11, "learning_rate": 8.858593958834535e-08, "loss": 0.0685, "step": 427 }, { "epoch": 0.11, "learning_rate": 8.855920876770916e-08, "loss": 0.0747, "step": 428 }, { "epoch": 0.11, "learning_rate": 8.853247794707298e-08, "loss": 0.0657, "step": 429 }, { "epoch": 0.11, "learning_rate": 8.850574712643677e-08, "loss": 0.0571, "step": 430 }, { "epoch": 0.12, "learning_rate": 8.847901630580059e-08, "loss": 0.0469, "step": 431 }, { "epoch": 0.12, "learning_rate": 8.84522854851644e-08, "loss": 0.0576, "step": 432 }, { "epoch": 0.12, "learning_rate": 8.84255546645282e-08, "loss": 0.0538, "step": 433 }, { "epoch": 0.12, "learning_rate": 8.8398823843892e-08, "loss": 0.0549, "step": 434 }, { "epoch": 0.12, "learning_rate": 8.837209302325582e-08, "loss": 0.0642, "step": 435 }, { "epoch": 0.12, "learning_rate": 8.834536220261962e-08, "loss": 0.0864, "step": 436 }, { "epoch": 0.12, "learning_rate": 8.831863138198343e-08, "loss": 0.0601, "step": 437 }, { "epoch": 0.12, "learning_rate": 8.829190056134724e-08, "loss": 0.0666, "step": 438 }, { "epoch": 0.12, "learning_rate": 8.826516974071104e-08, "loss": 0.0612, "step": 439 }, { "epoch": 0.12, "learning_rate": 8.823843892007485e-08, "loss": 0.0461, "step": 440 }, { "epoch": 0.12, "learning_rate": 8.821170809943866e-08, "loss": 0.0634, "step": 441 }, { "epoch": 0.12, "learning_rate": 8.818497727880246e-08, "loss": 0.0717, "step": 442 }, { "epoch": 0.12, "learning_rate": 8.815824645816627e-08, "loss": 0.0692, "step": 443 }, { "epoch": 0.12, "learning_rate": 8.813151563753007e-08, "loss": 0.0698, "step": 444 }, { "epoch": 0.12, "learning_rate": 8.810478481689388e-08, "loss": 0.0443, "step": 445 }, { "epoch": 0.12, "learning_rate": 8.807805399625769e-08, "loss": 0.0648, "step": 446 }, { "epoch": 0.12, "learning_rate": 8.805132317562149e-08, "loss": 0.0673, "step": 447 }, { "epoch": 0.12, "learning_rate": 8.80245923549853e-08, "loss": 0.0618, "step": 448 }, { "epoch": 0.12, "learning_rate": 8.799786153434911e-08, "loss": 0.0675, "step": 449 }, { "epoch": 0.12, "learning_rate": 8.797113071371291e-08, "loss": 0.0547, "step": 450 }, { "epoch": 0.12, "learning_rate": 8.794439989307672e-08, "loss": 0.0726, "step": 451 }, { "epoch": 0.12, "learning_rate": 8.791766907244052e-08, "loss": 0.0457, "step": 452 }, { "epoch": 0.12, "learning_rate": 8.789093825180433e-08, "loss": 0.0674, "step": 453 }, { "epoch": 0.12, "learning_rate": 8.786420743116814e-08, "loss": 0.0715, "step": 454 }, { "epoch": 0.12, "learning_rate": 8.783747661053194e-08, "loss": 0.051, "step": 455 }, { "epoch": 0.12, "learning_rate": 8.781074578989575e-08, "loss": 0.0709, "step": 456 }, { "epoch": 0.12, "learning_rate": 8.778401496925956e-08, "loss": 0.0765, "step": 457 }, { "epoch": 0.12, "learning_rate": 8.775728414862336e-08, "loss": 0.0693, "step": 458 }, { "epoch": 0.12, "learning_rate": 8.773055332798717e-08, "loss": 0.0543, "step": 459 }, { "epoch": 0.12, "learning_rate": 8.770382250735097e-08, "loss": 0.0581, "step": 460 }, { "epoch": 0.12, "learning_rate": 8.767709168671478e-08, "loss": 0.0537, "step": 461 }, { "epoch": 0.12, "learning_rate": 8.765036086607859e-08, "loss": 0.0801, "step": 462 }, { "epoch": 0.12, "learning_rate": 8.762363004544239e-08, "loss": 0.0884, "step": 463 }, { "epoch": 0.12, "learning_rate": 8.75968992248062e-08, "loss": 0.0531, "step": 464 }, { "epoch": 0.12, "learning_rate": 8.757016840417001e-08, "loss": 0.1051, "step": 465 }, { "epoch": 0.12, "learning_rate": 8.754343758353381e-08, "loss": 0.0495, "step": 466 }, { "epoch": 0.12, "learning_rate": 8.751670676289762e-08, "loss": 0.0533, "step": 467 }, { "epoch": 0.13, "learning_rate": 8.748997594226142e-08, "loss": 0.0719, "step": 468 }, { "epoch": 0.13, "learning_rate": 8.746324512162523e-08, "loss": 0.0665, "step": 469 }, { "epoch": 0.13, "learning_rate": 8.743651430098904e-08, "loss": 0.0951, "step": 470 }, { "epoch": 0.13, "learning_rate": 8.740978348035284e-08, "loss": 0.0603, "step": 471 }, { "epoch": 0.13, "learning_rate": 8.738305265971665e-08, "loss": 0.0628, "step": 472 }, { "epoch": 0.13, "learning_rate": 8.735632183908046e-08, "loss": 0.0769, "step": 473 }, { "epoch": 0.13, "learning_rate": 8.732959101844426e-08, "loss": 0.0483, "step": 474 }, { "epoch": 0.13, "learning_rate": 8.730286019780807e-08, "loss": 0.0745, "step": 475 }, { "epoch": 0.13, "learning_rate": 8.727612937717187e-08, "loss": 0.0608, "step": 476 }, { "epoch": 0.13, "learning_rate": 8.724939855653568e-08, "loss": 0.0684, "step": 477 }, { "epoch": 0.13, "learning_rate": 8.72226677358995e-08, "loss": 0.0732, "step": 478 }, { "epoch": 0.13, "learning_rate": 8.719593691526329e-08, "loss": 0.0761, "step": 479 }, { "epoch": 0.13, "learning_rate": 8.71692060946271e-08, "loss": 0.0544, "step": 480 }, { "epoch": 0.13, "learning_rate": 8.714247527399092e-08, "loss": 0.0674, "step": 481 }, { "epoch": 0.13, "learning_rate": 8.711574445335471e-08, "loss": 0.0476, "step": 482 }, { "epoch": 0.13, "learning_rate": 8.708901363271852e-08, "loss": 0.0764, "step": 483 }, { "epoch": 0.13, "learning_rate": 8.706228281208232e-08, "loss": 0.0658, "step": 484 }, { "epoch": 0.13, "learning_rate": 8.703555199144613e-08, "loss": 0.0765, "step": 485 }, { "epoch": 0.13, "learning_rate": 8.700882117080995e-08, "loss": 0.069, "step": 486 }, { "epoch": 0.13, "learning_rate": 8.698209035017374e-08, "loss": 0.0577, "step": 487 }, { "epoch": 0.13, "learning_rate": 8.695535952953755e-08, "loss": 0.0569, "step": 488 }, { "epoch": 0.13, "learning_rate": 8.692862870890137e-08, "loss": 0.0675, "step": 489 }, { "epoch": 0.13, "learning_rate": 8.690189788826516e-08, "loss": 0.069, "step": 490 }, { "epoch": 0.13, "learning_rate": 8.687516706762898e-08, "loss": 0.0733, "step": 491 }, { "epoch": 0.13, "learning_rate": 8.684843624699277e-08, "loss": 0.0548, "step": 492 }, { "epoch": 0.13, "learning_rate": 8.682170542635659e-08, "loss": 0.0609, "step": 493 }, { "epoch": 0.13, "learning_rate": 8.67949746057204e-08, "loss": 0.0434, "step": 494 }, { "epoch": 0.13, "learning_rate": 8.67682437850842e-08, "loss": 0.0598, "step": 495 }, { "epoch": 0.13, "learning_rate": 8.6741512964448e-08, "loss": 0.0786, "step": 496 }, { "epoch": 0.13, "learning_rate": 8.671478214381182e-08, "loss": 0.0619, "step": 497 }, { "epoch": 0.13, "learning_rate": 8.668805132317562e-08, "loss": 0.068, "step": 498 }, { "epoch": 0.13, "learning_rate": 8.666132050253943e-08, "loss": 0.0673, "step": 499 }, { "epoch": 0.13, "learning_rate": 8.663458968190322e-08, "loss": 0.0883, "step": 500 }, { "epoch": 0.13, "learning_rate": 8.660785886126704e-08, "loss": 0.0756, "step": 501 }, { "epoch": 0.13, "learning_rate": 8.658112804063085e-08, "loss": 0.0714, "step": 502 }, { "epoch": 0.13, "learning_rate": 8.655439721999465e-08, "loss": 0.0575, "step": 503 }, { "epoch": 0.13, "learning_rate": 8.652766639935846e-08, "loss": 0.0735, "step": 504 }, { "epoch": 0.13, "learning_rate": 8.650093557872227e-08, "loss": 0.0559, "step": 505 }, { "epoch": 0.14, "learning_rate": 8.647420475808607e-08, "loss": 0.0777, "step": 506 }, { "epoch": 0.14, "learning_rate": 8.644747393744988e-08, "loss": 0.0569, "step": 507 }, { "epoch": 0.14, "learning_rate": 8.642074311681368e-08, "loss": 0.0839, "step": 508 }, { "epoch": 0.14, "learning_rate": 8.639401229617749e-08, "loss": 0.0561, "step": 509 }, { "epoch": 0.14, "learning_rate": 8.63672814755413e-08, "loss": 0.0635, "step": 510 }, { "epoch": 0.14, "learning_rate": 8.63405506549051e-08, "loss": 0.0679, "step": 511 }, { "epoch": 0.14, "learning_rate": 8.631381983426891e-08, "loss": 0.0579, "step": 512 }, { "epoch": 0.14, "learning_rate": 8.628708901363272e-08, "loss": 0.0627, "step": 513 }, { "epoch": 0.14, "learning_rate": 8.626035819299652e-08, "loss": 0.0574, "step": 514 }, { "epoch": 0.14, "learning_rate": 8.623362737236033e-08, "loss": 0.0448, "step": 515 }, { "epoch": 0.14, "learning_rate": 8.620689655172413e-08, "loss": 0.0601, "step": 516 }, { "epoch": 0.14, "learning_rate": 8.618016573108794e-08, "loss": 0.0782, "step": 517 }, { "epoch": 0.14, "learning_rate": 8.615343491045175e-08, "loss": 0.0544, "step": 518 }, { "epoch": 0.14, "learning_rate": 8.612670408981555e-08, "loss": 0.0718, "step": 519 }, { "epoch": 0.14, "learning_rate": 8.609997326917936e-08, "loss": 0.064, "step": 520 }, { "epoch": 0.14, "learning_rate": 8.607324244854317e-08, "loss": 0.0625, "step": 521 }, { "epoch": 0.14, "learning_rate": 8.604651162790697e-08, "loss": 0.0806, "step": 522 }, { "epoch": 0.14, "learning_rate": 8.601978080727078e-08, "loss": 0.0529, "step": 523 }, { "epoch": 0.14, "learning_rate": 8.599304998663458e-08, "loss": 0.0689, "step": 524 }, { "epoch": 0.14, "learning_rate": 8.596631916599839e-08, "loss": 0.0782, "step": 525 }, { "epoch": 0.14, "learning_rate": 8.59395883453622e-08, "loss": 0.0605, "step": 526 }, { "epoch": 0.14, "learning_rate": 8.5912857524726e-08, "loss": 0.048, "step": 527 }, { "epoch": 0.14, "learning_rate": 8.588612670408981e-08, "loss": 0.0491, "step": 528 }, { "epoch": 0.14, "learning_rate": 8.585939588345362e-08, "loss": 0.0754, "step": 529 }, { "epoch": 0.14, "learning_rate": 8.583266506281742e-08, "loss": 0.0866, "step": 530 }, { "epoch": 0.14, "learning_rate": 8.580593424218123e-08, "loss": 0.0711, "step": 531 }, { "epoch": 0.14, "learning_rate": 8.577920342154503e-08, "loss": 0.0736, "step": 532 }, { "epoch": 0.14, "learning_rate": 8.575247260090884e-08, "loss": 0.0774, "step": 533 }, { "epoch": 0.14, "learning_rate": 8.572574178027265e-08, "loss": 0.0689, "step": 534 }, { "epoch": 0.14, "learning_rate": 8.569901095963645e-08, "loss": 0.0496, "step": 535 }, { "epoch": 0.14, "learning_rate": 8.567228013900026e-08, "loss": 0.0598, "step": 536 }, { "epoch": 0.14, "learning_rate": 8.564554931836407e-08, "loss": 0.0522, "step": 537 }, { "epoch": 0.14, "learning_rate": 8.561881849772787e-08, "loss": 0.0593, "step": 538 }, { "epoch": 0.14, "learning_rate": 8.559208767709168e-08, "loss": 0.0703, "step": 539 }, { "epoch": 0.14, "learning_rate": 8.556535685645548e-08, "loss": 0.0662, "step": 540 }, { "epoch": 0.14, "learning_rate": 8.553862603581929e-08, "loss": 0.0677, "step": 541 }, { "epoch": 0.14, "learning_rate": 8.55118952151831e-08, "loss": 0.0806, "step": 542 }, { "epoch": 0.15, "learning_rate": 8.54851643945469e-08, "loss": 0.0625, "step": 543 }, { "epoch": 0.15, "learning_rate": 8.545843357391071e-08, "loss": 0.0642, "step": 544 }, { "epoch": 0.15, "learning_rate": 8.543170275327452e-08, "loss": 0.0566, "step": 545 }, { "epoch": 0.15, "learning_rate": 8.540497193263832e-08, "loss": 0.063, "step": 546 }, { "epoch": 0.15, "learning_rate": 8.537824111200213e-08, "loss": 0.0814, "step": 547 }, { "epoch": 0.15, "learning_rate": 8.535151029136593e-08, "loss": 0.087, "step": 548 }, { "epoch": 0.15, "learning_rate": 8.532477947072974e-08, "loss": 0.0729, "step": 549 }, { "epoch": 0.15, "learning_rate": 8.529804865009355e-08, "loss": 0.0939, "step": 550 }, { "epoch": 0.15, "learning_rate": 8.527131782945735e-08, "loss": 0.0723, "step": 551 }, { "epoch": 0.15, "learning_rate": 8.524458700882116e-08, "loss": 0.0595, "step": 552 }, { "epoch": 0.15, "learning_rate": 8.521785618818498e-08, "loss": 0.0536, "step": 553 }, { "epoch": 0.15, "learning_rate": 8.519112536754877e-08, "loss": 0.0562, "step": 554 }, { "epoch": 0.15, "learning_rate": 8.516439454691258e-08, "loss": 0.0473, "step": 555 }, { "epoch": 0.15, "learning_rate": 8.513766372627638e-08, "loss": 0.0715, "step": 556 }, { "epoch": 0.15, "learning_rate": 8.51109329056402e-08, "loss": 0.0458, "step": 557 }, { "epoch": 0.15, "learning_rate": 8.5084202085004e-08, "loss": 0.0545, "step": 558 }, { "epoch": 0.15, "learning_rate": 8.50574712643678e-08, "loss": 0.0849, "step": 559 }, { "epoch": 0.15, "learning_rate": 8.503074044373162e-08, "loss": 0.0839, "step": 560 }, { "epoch": 0.15, "learning_rate": 8.500400962309543e-08, "loss": 0.0476, "step": 561 }, { "epoch": 0.15, "learning_rate": 8.497727880245922e-08, "loss": 0.0697, "step": 562 }, { "epoch": 0.15, "learning_rate": 8.495054798182304e-08, "loss": 0.0792, "step": 563 }, { "epoch": 0.15, "learning_rate": 8.492381716118683e-08, "loss": 0.0635, "step": 564 }, { "epoch": 0.15, "learning_rate": 8.489708634055065e-08, "loss": 0.0599, "step": 565 }, { "epoch": 0.15, "learning_rate": 8.487035551991446e-08, "loss": 0.0485, "step": 566 }, { "epoch": 0.15, "learning_rate": 8.484362469927825e-08, "loss": 0.0616, "step": 567 }, { "epoch": 0.15, "learning_rate": 8.481689387864207e-08, "loss": 0.0749, "step": 568 }, { "epoch": 0.15, "learning_rate": 8.479016305800588e-08, "loss": 0.0568, "step": 569 }, { "epoch": 0.15, "learning_rate": 8.476343223736968e-08, "loss": 0.1255, "step": 570 }, { "epoch": 0.15, "learning_rate": 8.473670141673349e-08, "loss": 0.0717, "step": 571 }, { "epoch": 0.15, "learning_rate": 8.470997059609729e-08, "loss": 0.0734, "step": 572 }, { "epoch": 0.15, "learning_rate": 8.46832397754611e-08, "loss": 0.0496, "step": 573 }, { "epoch": 0.15, "learning_rate": 8.465650895482491e-08, "loss": 0.0703, "step": 574 }, { "epoch": 0.15, "learning_rate": 8.46297781341887e-08, "loss": 0.0558, "step": 575 }, { "epoch": 0.15, "learning_rate": 8.460304731355252e-08, "loss": 0.08, "step": 576 }, { "epoch": 0.15, "learning_rate": 8.457631649291633e-08, "loss": 0.0837, "step": 577 }, { "epoch": 0.15, "learning_rate": 8.454958567228013e-08, "loss": 0.0509, "step": 578 }, { "epoch": 0.15, "learning_rate": 8.452285485164394e-08, "loss": 0.0889, "step": 579 }, { "epoch": 0.16, "learning_rate": 8.449612403100774e-08, "loss": 0.0777, "step": 580 }, { "epoch": 0.16, "learning_rate": 8.446939321037155e-08, "loss": 0.0642, "step": 581 }, { "epoch": 0.16, "learning_rate": 8.444266238973536e-08, "loss": 0.0594, "step": 582 }, { "epoch": 0.16, "learning_rate": 8.441593156909916e-08, "loss": 0.0554, "step": 583 }, { "epoch": 0.16, "learning_rate": 8.438920074846297e-08, "loss": 0.0681, "step": 584 }, { "epoch": 0.16, "learning_rate": 8.436246992782678e-08, "loss": 0.0791, "step": 585 }, { "epoch": 0.16, "learning_rate": 8.433573910719058e-08, "loss": 0.0725, "step": 586 }, { "epoch": 0.16, "learning_rate": 8.430900828655439e-08, "loss": 0.0736, "step": 587 }, { "epoch": 0.16, "learning_rate": 8.42822774659182e-08, "loss": 0.1051, "step": 588 }, { "epoch": 0.16, "learning_rate": 8.4255546645282e-08, "loss": 0.0628, "step": 589 }, { "epoch": 0.16, "learning_rate": 8.422881582464581e-08, "loss": 0.064, "step": 590 }, { "epoch": 0.16, "learning_rate": 8.420208500400962e-08, "loss": 0.0544, "step": 591 }, { "epoch": 0.16, "learning_rate": 8.417535418337342e-08, "loss": 0.0684, "step": 592 }, { "epoch": 0.16, "learning_rate": 8.414862336273723e-08, "loss": 0.0522, "step": 593 }, { "epoch": 0.16, "learning_rate": 8.412189254210104e-08, "loss": 0.0589, "step": 594 }, { "epoch": 0.16, "learning_rate": 8.409516172146484e-08, "loss": 0.0564, "step": 595 }, { "epoch": 0.16, "learning_rate": 8.406843090082865e-08, "loss": 0.0679, "step": 596 }, { "epoch": 0.16, "learning_rate": 8.404170008019246e-08, "loss": 0.0615, "step": 597 }, { "epoch": 0.16, "learning_rate": 8.401496925955626e-08, "loss": 0.0597, "step": 598 }, { "epoch": 0.16, "learning_rate": 8.398823843892007e-08, "loss": 0.0488, "step": 599 }, { "epoch": 0.16, "learning_rate": 8.396150761828388e-08, "loss": 0.0656, "step": 600 }, { "epoch": 0.16, "learning_rate": 8.393477679764768e-08, "loss": 0.0794, "step": 601 }, { "epoch": 0.16, "learning_rate": 8.39080459770115e-08, "loss": 0.059, "step": 602 }, { "epoch": 0.16, "learning_rate": 8.38813151563753e-08, "loss": 0.0572, "step": 603 }, { "epoch": 0.16, "learning_rate": 8.38545843357391e-08, "loss": 0.0732, "step": 604 }, { "epoch": 0.16, "learning_rate": 8.382785351510291e-08, "loss": 0.0698, "step": 605 }, { "epoch": 0.16, "learning_rate": 8.380112269446673e-08, "loss": 0.0839, "step": 606 }, { "epoch": 0.16, "learning_rate": 8.377439187383052e-08, "loss": 0.0686, "step": 607 }, { "epoch": 0.16, "learning_rate": 8.374766105319434e-08, "loss": 0.0793, "step": 608 }, { "epoch": 0.16, "learning_rate": 8.372093023255815e-08, "loss": 0.0576, "step": 609 }, { "epoch": 0.16, "learning_rate": 8.369419941192194e-08, "loss": 0.076, "step": 610 }, { "epoch": 0.16, "learning_rate": 8.366746859128576e-08, "loss": 0.0837, "step": 611 }, { "epoch": 0.16, "learning_rate": 8.364073777064955e-08, "loss": 0.0643, "step": 612 }, { "epoch": 0.16, "learning_rate": 8.361400695001337e-08, "loss": 0.0626, "step": 613 }, { "epoch": 0.16, "learning_rate": 8.358727612937718e-08, "loss": 0.0807, "step": 614 }, { "epoch": 0.16, "learning_rate": 8.356054530874097e-08, "loss": 0.0588, "step": 615 }, { "epoch": 0.16, "learning_rate": 8.353381448810479e-08, "loss": 0.0747, "step": 616 }, { "epoch": 0.16, "learning_rate": 8.35070836674686e-08, "loss": 0.0689, "step": 617 }, { "epoch": 0.17, "learning_rate": 8.34803528468324e-08, "loss": 0.0682, "step": 618 }, { "epoch": 0.17, "learning_rate": 8.345362202619621e-08, "loss": 0.0581, "step": 619 }, { "epoch": 0.17, "learning_rate": 8.342689120556e-08, "loss": 0.0547, "step": 620 }, { "epoch": 0.17, "learning_rate": 8.340016038492382e-08, "loss": 0.0772, "step": 621 }, { "epoch": 0.17, "learning_rate": 8.337342956428763e-08, "loss": 0.048, "step": 622 }, { "epoch": 0.17, "learning_rate": 8.334669874365143e-08, "loss": 0.0608, "step": 623 }, { "epoch": 0.17, "learning_rate": 8.331996792301524e-08, "loss": 0.0547, "step": 624 }, { "epoch": 0.17, "learning_rate": 8.329323710237905e-08, "loss": 0.0822, "step": 625 }, { "epoch": 0.17, "learning_rate": 8.326650628174285e-08, "loss": 0.058, "step": 626 }, { "epoch": 0.17, "learning_rate": 8.323977546110666e-08, "loss": 0.0519, "step": 627 }, { "epoch": 0.17, "learning_rate": 8.321304464047046e-08, "loss": 0.0678, "step": 628 }, { "epoch": 0.17, "learning_rate": 8.318631381983427e-08, "loss": 0.0717, "step": 629 }, { "epoch": 0.17, "learning_rate": 8.315958299919808e-08, "loss": 0.0779, "step": 630 }, { "epoch": 0.17, "learning_rate": 8.313285217856188e-08, "loss": 0.0606, "step": 631 }, { "epoch": 0.17, "learning_rate": 8.310612135792569e-08, "loss": 0.0749, "step": 632 }, { "epoch": 0.17, "learning_rate": 8.30793905372895e-08, "loss": 0.0637, "step": 633 }, { "epoch": 0.17, "learning_rate": 8.30526597166533e-08, "loss": 0.0554, "step": 634 }, { "epoch": 0.17, "learning_rate": 8.302592889601711e-08, "loss": 0.0765, "step": 635 }, { "epoch": 0.17, "learning_rate": 8.299919807538091e-08, "loss": 0.0534, "step": 636 }, { "epoch": 0.17, "learning_rate": 8.297246725474472e-08, "loss": 0.0538, "step": 637 }, { "epoch": 0.17, "learning_rate": 8.294573643410853e-08, "loss": 0.0587, "step": 638 }, { "epoch": 0.17, "learning_rate": 8.291900561347233e-08, "loss": 0.0441, "step": 639 }, { "epoch": 0.17, "learning_rate": 8.289227479283614e-08, "loss": 0.0961, "step": 640 }, { "epoch": 0.17, "learning_rate": 8.286554397219995e-08, "loss": 0.0862, "step": 641 }, { "epoch": 0.17, "learning_rate": 8.283881315156375e-08, "loss": 0.0575, "step": 642 }, { "epoch": 0.17, "learning_rate": 8.281208233092756e-08, "loss": 0.083, "step": 643 }, { "epoch": 0.17, "learning_rate": 8.278535151029136e-08, "loss": 0.059, "step": 644 }, { "epoch": 0.17, "learning_rate": 8.275862068965517e-08, "loss": 0.0746, "step": 645 }, { "epoch": 0.17, "learning_rate": 8.273188986901898e-08, "loss": 0.0495, "step": 646 }, { "epoch": 0.17, "learning_rate": 8.270515904838278e-08, "loss": 0.0657, "step": 647 }, { "epoch": 0.17, "learning_rate": 8.267842822774659e-08, "loss": 0.061, "step": 648 }, { "epoch": 0.17, "learning_rate": 8.26516974071104e-08, "loss": 0.0615, "step": 649 }, { "epoch": 0.17, "learning_rate": 8.26249665864742e-08, "loss": 0.0586, "step": 650 }, { "epoch": 0.17, "learning_rate": 8.259823576583801e-08, "loss": 0.0587, "step": 651 }, { "epoch": 0.17, "learning_rate": 8.257150494520181e-08, "loss": 0.0502, "step": 652 }, { "epoch": 0.17, "learning_rate": 8.254477412456562e-08, "loss": 0.0868, "step": 653 }, { "epoch": 0.17, "learning_rate": 8.251804330392943e-08, "loss": 0.0776, "step": 654 }, { "epoch": 0.18, "learning_rate": 8.249131248329323e-08, "loss": 0.0639, "step": 655 }, { "epoch": 0.18, "learning_rate": 8.246458166265704e-08, "loss": 0.0832, "step": 656 }, { "epoch": 0.18, "learning_rate": 8.243785084202085e-08, "loss": 0.0576, "step": 657 }, { "epoch": 0.18, "learning_rate": 8.241112002138465e-08, "loss": 0.074, "step": 658 }, { "epoch": 0.18, "learning_rate": 8.238438920074846e-08, "loss": 0.0445, "step": 659 }, { "epoch": 0.18, "learning_rate": 8.235765838011226e-08, "loss": 0.0411, "step": 660 }, { "epoch": 0.18, "learning_rate": 8.233092755947607e-08, "loss": 0.0631, "step": 661 }, { "epoch": 0.18, "learning_rate": 8.230419673883988e-08, "loss": 0.0793, "step": 662 }, { "epoch": 0.18, "learning_rate": 8.227746591820368e-08, "loss": 0.0737, "step": 663 }, { "epoch": 0.18, "learning_rate": 8.225073509756749e-08, "loss": 0.0534, "step": 664 }, { "epoch": 0.18, "learning_rate": 8.22240042769313e-08, "loss": 0.0599, "step": 665 }, { "epoch": 0.18, "learning_rate": 8.21972734562951e-08, "loss": 0.0766, "step": 666 }, { "epoch": 0.18, "learning_rate": 8.217054263565891e-08, "loss": 0.0557, "step": 667 }, { "epoch": 0.18, "learning_rate": 8.214381181502271e-08, "loss": 0.0877, "step": 668 }, { "epoch": 0.18, "learning_rate": 8.211708099438652e-08, "loss": 0.0657, "step": 669 }, { "epoch": 0.18, "learning_rate": 8.209035017375033e-08, "loss": 0.0802, "step": 670 }, { "epoch": 0.18, "learning_rate": 8.206361935311413e-08, "loss": 0.0616, "step": 671 }, { "epoch": 0.18, "learning_rate": 8.203688853247794e-08, "loss": 0.0641, "step": 672 }, { "epoch": 0.18, "learning_rate": 8.201015771184176e-08, "loss": 0.0607, "step": 673 }, { "epoch": 0.18, "learning_rate": 8.198342689120555e-08, "loss": 0.0577, "step": 674 }, { "epoch": 0.18, "learning_rate": 8.195669607056937e-08, "loss": 0.0607, "step": 675 }, { "epoch": 0.18, "learning_rate": 8.192996524993316e-08, "loss": 0.0681, "step": 676 }, { "epoch": 0.18, "learning_rate": 8.190323442929697e-08, "loss": 0.0714, "step": 677 }, { "epoch": 0.18, "learning_rate": 8.187650360866079e-08, "loss": 0.0636, "step": 678 }, { "epoch": 0.18, "learning_rate": 8.184977278802458e-08, "loss": 0.0637, "step": 679 }, { "epoch": 0.18, "learning_rate": 8.18230419673884e-08, "loss": 0.0573, "step": 680 }, { "epoch": 0.18, "learning_rate": 8.179631114675221e-08, "loss": 0.0583, "step": 681 }, { "epoch": 0.18, "learning_rate": 8.1769580326116e-08, "loss": 0.0656, "step": 682 }, { "epoch": 0.18, "learning_rate": 8.174284950547982e-08, "loss": 0.0861, "step": 683 }, { "epoch": 0.18, "learning_rate": 8.171611868484361e-08, "loss": 0.0722, "step": 684 }, { "epoch": 0.18, "learning_rate": 8.168938786420743e-08, "loss": 0.0731, "step": 685 }, { "epoch": 0.18, "learning_rate": 8.166265704357124e-08, "loss": 0.0825, "step": 686 }, { "epoch": 0.18, "learning_rate": 8.163592622293504e-08, "loss": 0.078, "step": 687 }, { "epoch": 0.18, "learning_rate": 8.160919540229885e-08, "loss": 0.065, "step": 688 }, { "epoch": 0.18, "learning_rate": 8.158246458166266e-08, "loss": 0.0413, "step": 689 }, { "epoch": 0.18, "learning_rate": 8.155573376102646e-08, "loss": 0.0563, "step": 690 }, { "epoch": 0.18, "learning_rate": 8.152900294039027e-08, "loss": 0.062, "step": 691 }, { "epoch": 0.18, "learning_rate": 8.150227211975407e-08, "loss": 0.0599, "step": 692 }, { "epoch": 0.19, "learning_rate": 8.147554129911788e-08, "loss": 0.0483, "step": 693 }, { "epoch": 0.19, "learning_rate": 8.144881047848169e-08, "loss": 0.0582, "step": 694 }, { "epoch": 0.19, "learning_rate": 8.142207965784549e-08, "loss": 0.0422, "step": 695 }, { "epoch": 0.19, "learning_rate": 8.13953488372093e-08, "loss": 0.0722, "step": 696 }, { "epoch": 0.19, "learning_rate": 8.136861801657311e-08, "loss": 0.0713, "step": 697 }, { "epoch": 0.19, "learning_rate": 8.134188719593691e-08, "loss": 0.0719, "step": 698 }, { "epoch": 0.19, "learning_rate": 8.131515637530072e-08, "loss": 0.0547, "step": 699 }, { "epoch": 0.19, "learning_rate": 8.128842555466452e-08, "loss": 0.0628, "step": 700 }, { "epoch": 0.19, "learning_rate": 8.126169473402833e-08, "loss": 0.0591, "step": 701 }, { "epoch": 0.19, "learning_rate": 8.123496391339214e-08, "loss": 0.078, "step": 702 }, { "epoch": 0.19, "learning_rate": 8.120823309275594e-08, "loss": 0.0727, "step": 703 }, { "epoch": 0.19, "learning_rate": 8.118150227211975e-08, "loss": 0.0555, "step": 704 }, { "epoch": 0.19, "learning_rate": 8.115477145148356e-08, "loss": 0.066, "step": 705 }, { "epoch": 0.19, "learning_rate": 8.112804063084736e-08, "loss": 0.0849, "step": 706 }, { "epoch": 0.19, "learning_rate": 8.110130981021117e-08, "loss": 0.0564, "step": 707 }, { "epoch": 0.19, "learning_rate": 8.107457898957497e-08, "loss": 0.0596, "step": 708 }, { "epoch": 0.19, "learning_rate": 8.104784816893878e-08, "loss": 0.0556, "step": 709 }, { "epoch": 0.19, "learning_rate": 8.102111734830259e-08, "loss": 0.0571, "step": 710 }, { "epoch": 0.19, "learning_rate": 8.099438652766639e-08, "loss": 0.0828, "step": 711 }, { "epoch": 0.19, "learning_rate": 8.09676557070302e-08, "loss": 0.0748, "step": 712 }, { "epoch": 0.19, "learning_rate": 8.094092488639401e-08, "loss": 0.0741, "step": 713 }, { "epoch": 0.19, "learning_rate": 8.091419406575781e-08, "loss": 0.0759, "step": 714 }, { "epoch": 0.19, "learning_rate": 8.088746324512162e-08, "loss": 0.0797, "step": 715 }, { "epoch": 0.19, "learning_rate": 8.086073242448542e-08, "loss": 0.093, "step": 716 }, { "epoch": 0.19, "learning_rate": 8.083400160384923e-08, "loss": 0.0743, "step": 717 }, { "epoch": 0.19, "learning_rate": 8.080727078321304e-08, "loss": 0.0626, "step": 718 }, { "epoch": 0.19, "learning_rate": 8.078053996257684e-08, "loss": 0.0538, "step": 719 }, { "epoch": 0.19, "learning_rate": 8.075380914194065e-08, "loss": 0.0515, "step": 720 }, { "epoch": 0.19, "learning_rate": 8.072707832130446e-08, "loss": 0.0633, "step": 721 }, { "epoch": 0.19, "learning_rate": 8.070034750066826e-08, "loss": 0.0709, "step": 722 }, { "epoch": 0.19, "learning_rate": 8.067361668003207e-08, "loss": 0.0544, "step": 723 }, { "epoch": 0.19, "learning_rate": 8.064688585939587e-08, "loss": 0.0621, "step": 724 }, { "epoch": 0.19, "learning_rate": 8.062015503875968e-08, "loss": 0.0572, "step": 725 }, { "epoch": 0.19, "learning_rate": 8.059342421812349e-08, "loss": 0.0737, "step": 726 }, { "epoch": 0.19, "learning_rate": 8.056669339748729e-08, "loss": 0.0556, "step": 727 }, { "epoch": 0.19, "learning_rate": 8.05399625768511e-08, "loss": 0.0466, "step": 728 }, { "epoch": 0.19, "learning_rate": 8.051323175621491e-08, "loss": 0.0912, "step": 729 }, { "epoch": 0.2, "learning_rate": 8.048650093557871e-08, "loss": 0.0848, "step": 730 }, { "epoch": 0.2, "learning_rate": 8.045977011494252e-08, "loss": 0.0723, "step": 731 }, { "epoch": 0.2, "learning_rate": 8.043303929430632e-08, "loss": 0.0637, "step": 732 }, { "epoch": 0.2, "learning_rate": 8.040630847367013e-08, "loss": 0.0487, "step": 733 }, { "epoch": 0.2, "learning_rate": 8.037957765303394e-08, "loss": 0.0557, "step": 734 }, { "epoch": 0.2, "learning_rate": 8.035284683239774e-08, "loss": 0.0569, "step": 735 }, { "epoch": 0.2, "learning_rate": 8.032611601176155e-08, "loss": 0.0683, "step": 736 }, { "epoch": 0.2, "learning_rate": 8.029938519112536e-08, "loss": 0.0756, "step": 737 }, { "epoch": 0.2, "learning_rate": 8.027265437048916e-08, "loss": 0.0901, "step": 738 }, { "epoch": 0.2, "learning_rate": 8.024592354985297e-08, "loss": 0.0479, "step": 739 }, { "epoch": 0.2, "learning_rate": 8.021919272921677e-08, "loss": 0.0665, "step": 740 }, { "epoch": 0.2, "learning_rate": 8.019246190858058e-08, "loss": 0.065, "step": 741 }, { "epoch": 0.2, "learning_rate": 8.01657310879444e-08, "loss": 0.0657, "step": 742 }, { "epoch": 0.2, "learning_rate": 8.013900026730819e-08, "loss": 0.0468, "step": 743 }, { "epoch": 0.2, "learning_rate": 8.0112269446672e-08, "loss": 0.0771, "step": 744 }, { "epoch": 0.2, "learning_rate": 8.008553862603582e-08, "loss": 0.0679, "step": 745 }, { "epoch": 0.2, "learning_rate": 8.005880780539961e-08, "loss": 0.0543, "step": 746 }, { "epoch": 0.2, "learning_rate": 8.003207698476343e-08, "loss": 0.1008, "step": 747 }, { "epoch": 0.2, "learning_rate": 8.000534616412722e-08, "loss": 0.0574, "step": 748 }, { "epoch": 0.2, "learning_rate": 7.997861534349103e-08, "loss": 0.0576, "step": 749 }, { "epoch": 0.2, "learning_rate": 7.995188452285485e-08, "loss": 0.0597, "step": 750 }, { "epoch": 0.2, "learning_rate": 7.992515370221864e-08, "loss": 0.0629, "step": 751 }, { "epoch": 0.2, "learning_rate": 7.989842288158246e-08, "loss": 0.0745, "step": 752 }, { "epoch": 0.2, "learning_rate": 7.987169206094627e-08, "loss": 0.0492, "step": 753 }, { "epoch": 0.2, "learning_rate": 7.984496124031007e-08, "loss": 0.0648, "step": 754 }, { "epoch": 0.2, "learning_rate": 7.981823041967388e-08, "loss": 0.0493, "step": 755 }, { "epoch": 0.2, "learning_rate": 7.979149959903769e-08, "loss": 0.0568, "step": 756 }, { "epoch": 0.2, "learning_rate": 7.976476877840149e-08, "loss": 0.0635, "step": 757 }, { "epoch": 0.2, "learning_rate": 7.97380379577653e-08, "loss": 0.0619, "step": 758 }, { "epoch": 0.2, "learning_rate": 7.971130713712911e-08, "loss": 0.0712, "step": 759 }, { "epoch": 0.2, "learning_rate": 7.968457631649291e-08, "loss": 0.0508, "step": 760 }, { "epoch": 0.2, "learning_rate": 7.965784549585672e-08, "loss": 0.0604, "step": 761 }, { "epoch": 0.2, "learning_rate": 7.963111467522053e-08, "loss": 0.065, "step": 762 }, { "epoch": 0.2, "learning_rate": 7.960438385458433e-08, "loss": 0.0682, "step": 763 }, { "epoch": 0.2, "learning_rate": 7.957765303394814e-08, "loss": 0.0895, "step": 764 }, { "epoch": 0.2, "learning_rate": 7.955092221331195e-08, "loss": 0.0528, "step": 765 }, { "epoch": 0.2, "learning_rate": 7.952419139267575e-08, "loss": 0.0583, "step": 766 }, { "epoch": 0.2, "learning_rate": 7.949746057203956e-08, "loss": 0.0627, "step": 767 }, { "epoch": 0.21, "learning_rate": 7.947072975140337e-08, "loss": 0.0839, "step": 768 }, { "epoch": 0.21, "learning_rate": 7.944399893076717e-08, "loss": 0.0477, "step": 769 }, { "epoch": 0.21, "learning_rate": 7.941726811013098e-08, "loss": 0.0578, "step": 770 }, { "epoch": 0.21, "learning_rate": 7.939053728949479e-08, "loss": 0.0691, "step": 771 }, { "epoch": 0.21, "learning_rate": 7.936380646885859e-08, "loss": 0.0619, "step": 772 }, { "epoch": 0.21, "learning_rate": 7.93370756482224e-08, "loss": 0.0625, "step": 773 }, { "epoch": 0.21, "learning_rate": 7.931034482758621e-08, "loss": 0.0511, "step": 774 }, { "epoch": 0.21, "learning_rate": 7.928361400695001e-08, "loss": 0.0784, "step": 775 }, { "epoch": 0.21, "learning_rate": 7.925688318631382e-08, "loss": 0.0534, "step": 776 }, { "epoch": 0.21, "learning_rate": 7.923015236567763e-08, "loss": 0.0558, "step": 777 }, { "epoch": 0.21, "learning_rate": 7.920342154504143e-08, "loss": 0.0697, "step": 778 }, { "epoch": 0.21, "learning_rate": 7.917669072440524e-08, "loss": 0.0544, "step": 779 }, { "epoch": 0.21, "learning_rate": 7.914995990376904e-08, "loss": 0.0399, "step": 780 }, { "epoch": 0.21, "learning_rate": 7.912322908313285e-08, "loss": 0.0777, "step": 781 }, { "epoch": 0.21, "learning_rate": 7.909649826249666e-08, "loss": 0.054, "step": 782 }, { "epoch": 0.21, "learning_rate": 7.906976744186046e-08, "loss": 0.0667, "step": 783 }, { "epoch": 0.21, "learning_rate": 7.904303662122427e-08, "loss": 0.0773, "step": 784 }, { "epoch": 0.21, "learning_rate": 7.901630580058808e-08, "loss": 0.0627, "step": 785 }, { "epoch": 0.21, "learning_rate": 7.898957497995188e-08, "loss": 0.0897, "step": 786 }, { "epoch": 0.21, "learning_rate": 7.89628441593157e-08, "loss": 0.0928, "step": 787 }, { "epoch": 0.21, "learning_rate": 7.893611333867949e-08, "loss": 0.0814, "step": 788 }, { "epoch": 0.21, "learning_rate": 7.89093825180433e-08, "loss": 0.0559, "step": 789 }, { "epoch": 0.21, "learning_rate": 7.888265169740712e-08, "loss": 0.0573, "step": 790 }, { "epoch": 0.21, "learning_rate": 7.885592087677091e-08, "loss": 0.0697, "step": 791 }, { "epoch": 0.21, "learning_rate": 7.882919005613472e-08, "loss": 0.0561, "step": 792 }, { "epoch": 0.21, "learning_rate": 7.880245923549854e-08, "loss": 0.0797, "step": 793 }, { "epoch": 0.21, "learning_rate": 7.877572841486233e-08, "loss": 0.058, "step": 794 }, { "epoch": 0.21, "learning_rate": 7.874899759422615e-08, "loss": 0.0649, "step": 795 }, { "epoch": 0.21, "learning_rate": 7.872226677358994e-08, "loss": 0.0748, "step": 796 }, { "epoch": 0.21, "learning_rate": 7.869553595295375e-08, "loss": 0.0625, "step": 797 }, { "epoch": 0.21, "learning_rate": 7.866880513231757e-08, "loss": 0.0768, "step": 798 }, { "epoch": 0.21, "learning_rate": 7.864207431168136e-08, "loss": 0.0615, "step": 799 }, { "epoch": 0.21, "learning_rate": 7.861534349104518e-08, "loss": 0.0737, "step": 800 }, { "epoch": 0.21, "learning_rate": 7.858861267040899e-08, "loss": 0.0632, "step": 801 }, { "epoch": 0.21, "learning_rate": 7.856188184977279e-08, "loss": 0.1221, "step": 802 }, { "epoch": 0.21, "learning_rate": 7.85351510291366e-08, "loss": 0.0608, "step": 803 }, { "epoch": 0.21, "learning_rate": 7.85084202085004e-08, "loss": 0.0708, "step": 804 }, { "epoch": 0.22, "learning_rate": 7.84816893878642e-08, "loss": 0.0508, "step": 805 }, { "epoch": 0.22, "learning_rate": 7.845495856722802e-08, "loss": 0.0807, "step": 806 }, { "epoch": 0.22, "learning_rate": 7.842822774659182e-08, "loss": 0.0415, "step": 807 }, { "epoch": 0.22, "learning_rate": 7.840149692595563e-08, "loss": 0.0702, "step": 808 }, { "epoch": 0.22, "learning_rate": 7.837476610531944e-08, "loss": 0.0634, "step": 809 }, { "epoch": 0.22, "learning_rate": 7.834803528468324e-08, "loss": 0.0711, "step": 810 }, { "epoch": 0.22, "learning_rate": 7.832130446404705e-08, "loss": 0.0526, "step": 811 }, { "epoch": 0.22, "learning_rate": 7.829457364341085e-08, "loss": 0.0601, "step": 812 }, { "epoch": 0.22, "learning_rate": 7.826784282277466e-08, "loss": 0.0508, "step": 813 }, { "epoch": 0.22, "learning_rate": 7.824111200213847e-08, "loss": 0.0513, "step": 814 }, { "epoch": 0.22, "learning_rate": 7.821438118150227e-08, "loss": 0.0606, "step": 815 }, { "epoch": 0.22, "learning_rate": 7.818765036086608e-08, "loss": 0.0636, "step": 816 }, { "epoch": 0.22, "learning_rate": 7.816091954022989e-08, "loss": 0.0877, "step": 817 }, { "epoch": 0.22, "learning_rate": 7.813418871959369e-08, "loss": 0.0687, "step": 818 }, { "epoch": 0.22, "learning_rate": 7.81074578989575e-08, "loss": 0.0603, "step": 819 }, { "epoch": 0.22, "learning_rate": 7.80807270783213e-08, "loss": 0.0551, "step": 820 }, { "epoch": 0.22, "learning_rate": 7.805399625768511e-08, "loss": 0.0663, "step": 821 }, { "epoch": 0.22, "learning_rate": 7.802726543704892e-08, "loss": 0.0729, "step": 822 }, { "epoch": 0.22, "learning_rate": 7.800053461641272e-08, "loss": 0.0624, "step": 823 }, { "epoch": 0.22, "learning_rate": 7.797380379577653e-08, "loss": 0.0514, "step": 824 }, { "epoch": 0.22, "learning_rate": 7.794707297514034e-08, "loss": 0.0872, "step": 825 }, { "epoch": 0.22, "learning_rate": 7.792034215450414e-08, "loss": 0.059, "step": 826 }, { "epoch": 0.22, "learning_rate": 7.789361133386795e-08, "loss": 0.078, "step": 827 }, { "epoch": 0.22, "learning_rate": 7.786688051323175e-08, "loss": 0.0559, "step": 828 }, { "epoch": 0.22, "learning_rate": 7.784014969259556e-08, "loss": 0.0518, "step": 829 }, { "epoch": 0.22, "learning_rate": 7.781341887195937e-08, "loss": 0.0615, "step": 830 }, { "epoch": 0.22, "learning_rate": 7.778668805132317e-08, "loss": 0.0489, "step": 831 }, { "epoch": 0.22, "learning_rate": 7.775995723068698e-08, "loss": 0.066, "step": 832 }, { "epoch": 0.22, "learning_rate": 7.773322641005079e-08, "loss": 0.059, "step": 833 }, { "epoch": 0.22, "learning_rate": 7.770649558941459e-08, "loss": 0.0476, "step": 834 }, { "epoch": 0.22, "learning_rate": 7.76797647687784e-08, "loss": 0.068, "step": 835 }, { "epoch": 0.22, "learning_rate": 7.76530339481422e-08, "loss": 0.0681, "step": 836 }, { "epoch": 0.22, "learning_rate": 7.762630312750601e-08, "loss": 0.0785, "step": 837 }, { "epoch": 0.22, "learning_rate": 7.759957230686982e-08, "loss": 0.0638, "step": 838 }, { "epoch": 0.22, "learning_rate": 7.757284148623362e-08, "loss": 0.0589, "step": 839 }, { "epoch": 0.22, "learning_rate": 7.754611066559743e-08, "loss": 0.0501, "step": 840 }, { "epoch": 0.22, "learning_rate": 7.751937984496124e-08, "loss": 0.0627, "step": 841 }, { "epoch": 0.23, "learning_rate": 7.749264902432504e-08, "loss": 0.0673, "step": 842 }, { "epoch": 0.23, "learning_rate": 7.746591820368885e-08, "loss": 0.0672, "step": 843 }, { "epoch": 0.23, "learning_rate": 7.743918738305265e-08, "loss": 0.077, "step": 844 }, { "epoch": 0.23, "learning_rate": 7.741245656241646e-08, "loss": 0.0573, "step": 845 }, { "epoch": 0.23, "learning_rate": 7.738572574178027e-08, "loss": 0.0544, "step": 846 }, { "epoch": 0.23, "learning_rate": 7.735899492114407e-08, "loss": 0.049, "step": 847 }, { "epoch": 0.23, "learning_rate": 7.733226410050788e-08, "loss": 0.0572, "step": 848 }, { "epoch": 0.23, "learning_rate": 7.73055332798717e-08, "loss": 0.0789, "step": 849 }, { "epoch": 0.23, "learning_rate": 7.727880245923549e-08, "loss": 0.0617, "step": 850 }, { "epoch": 0.23, "learning_rate": 7.72520716385993e-08, "loss": 0.0564, "step": 851 }, { "epoch": 0.23, "learning_rate": 7.72253408179631e-08, "loss": 0.0783, "step": 852 }, { "epoch": 0.23, "learning_rate": 7.719860999732691e-08, "loss": 0.0727, "step": 853 }, { "epoch": 0.23, "learning_rate": 7.717187917669072e-08, "loss": 0.0573, "step": 854 }, { "epoch": 0.23, "learning_rate": 7.714514835605452e-08, "loss": 0.0662, "step": 855 }, { "epoch": 0.23, "learning_rate": 7.711841753541833e-08, "loss": 0.06, "step": 856 }, { "epoch": 0.23, "learning_rate": 7.709168671478215e-08, "loss": 0.0669, "step": 857 }, { "epoch": 0.23, "learning_rate": 7.706495589414594e-08, "loss": 0.0713, "step": 858 }, { "epoch": 0.23, "learning_rate": 7.703822507350975e-08, "loss": 0.0558, "step": 859 }, { "epoch": 0.23, "learning_rate": 7.701149425287355e-08, "loss": 0.0551, "step": 860 }, { "epoch": 0.23, "learning_rate": 7.698476343223736e-08, "loss": 0.0583, "step": 861 }, { "epoch": 0.23, "learning_rate": 7.695803261160118e-08, "loss": 0.0622, "step": 862 }, { "epoch": 0.23, "learning_rate": 7.693130179096497e-08, "loss": 0.096, "step": 863 }, { "epoch": 0.23, "learning_rate": 7.690457097032878e-08, "loss": 0.0647, "step": 864 }, { "epoch": 0.23, "learning_rate": 7.68778401496926e-08, "loss": 0.0811, "step": 865 }, { "epoch": 0.23, "learning_rate": 7.68511093290564e-08, "loss": 0.0634, "step": 866 }, { "epoch": 0.23, "learning_rate": 7.68243785084202e-08, "loss": 0.0652, "step": 867 }, { "epoch": 0.23, "learning_rate": 7.6797647687784e-08, "loss": 0.055, "step": 868 }, { "epoch": 0.23, "learning_rate": 7.677091686714782e-08, "loss": 0.0537, "step": 869 }, { "epoch": 0.23, "learning_rate": 7.674418604651163e-08, "loss": 0.0712, "step": 870 }, { "epoch": 0.23, "learning_rate": 7.671745522587542e-08, "loss": 0.0794, "step": 871 }, { "epoch": 0.23, "learning_rate": 7.669072440523924e-08, "loss": 0.0584, "step": 872 }, { "epoch": 0.23, "learning_rate": 7.666399358460305e-08, "loss": 0.0614, "step": 873 }, { "epoch": 0.23, "learning_rate": 7.663726276396685e-08, "loss": 0.0525, "step": 874 }, { "epoch": 0.23, "learning_rate": 7.661053194333066e-08, "loss": 0.0488, "step": 875 }, { "epoch": 0.23, "learning_rate": 7.658380112269446e-08, "loss": 0.0768, "step": 876 }, { "epoch": 0.23, "learning_rate": 7.655707030205827e-08, "loss": 0.0525, "step": 877 }, { "epoch": 0.23, "learning_rate": 7.653033948142208e-08, "loss": 0.0547, "step": 878 }, { "epoch": 0.23, "learning_rate": 7.650360866078588e-08, "loss": 0.049, "step": 879 }, { "epoch": 0.24, "learning_rate": 7.647687784014969e-08, "loss": 0.0644, "step": 880 }, { "epoch": 0.24, "learning_rate": 7.64501470195135e-08, "loss": 0.0788, "step": 881 }, { "epoch": 0.24, "learning_rate": 7.64234161988773e-08, "loss": 0.0631, "step": 882 }, { "epoch": 0.24, "learning_rate": 7.639668537824111e-08, "loss": 0.0727, "step": 883 }, { "epoch": 0.24, "learning_rate": 7.63699545576049e-08, "loss": 0.0383, "step": 884 }, { "epoch": 0.24, "learning_rate": 7.634322373696872e-08, "loss": 0.0683, "step": 885 }, { "epoch": 0.24, "learning_rate": 7.631649291633253e-08, "loss": 0.0572, "step": 886 }, { "epoch": 0.24, "learning_rate": 7.628976209569633e-08, "loss": 0.0693, "step": 887 }, { "epoch": 0.24, "learning_rate": 7.626303127506014e-08, "loss": 0.058, "step": 888 }, { "epoch": 0.24, "learning_rate": 7.623630045442395e-08, "loss": 0.0682, "step": 889 }, { "epoch": 0.24, "learning_rate": 7.620956963378775e-08, "loss": 0.0667, "step": 890 }, { "epoch": 0.24, "learning_rate": 7.618283881315156e-08, "loss": 0.0817, "step": 891 }, { "epoch": 0.24, "learning_rate": 7.615610799251536e-08, "loss": 0.0501, "step": 892 }, { "epoch": 0.24, "learning_rate": 7.612937717187917e-08, "loss": 0.072, "step": 893 }, { "epoch": 0.24, "learning_rate": 7.610264635124298e-08, "loss": 0.0675, "step": 894 }, { "epoch": 0.24, "learning_rate": 7.607591553060678e-08, "loss": 0.0583, "step": 895 }, { "epoch": 0.24, "learning_rate": 7.604918470997059e-08, "loss": 0.071, "step": 896 }, { "epoch": 0.24, "learning_rate": 7.60224538893344e-08, "loss": 0.0693, "step": 897 }, { "epoch": 0.24, "learning_rate": 7.59957230686982e-08, "loss": 0.0518, "step": 898 }, { "epoch": 0.24, "learning_rate": 7.596899224806201e-08, "loss": 0.0663, "step": 899 }, { "epoch": 0.24, "learning_rate": 7.594226142742581e-08, "loss": 0.0744, "step": 900 }, { "epoch": 0.24, "learning_rate": 7.591553060678962e-08, "loss": 0.0631, "step": 901 }, { "epoch": 0.24, "learning_rate": 7.588879978615343e-08, "loss": 0.072, "step": 902 }, { "epoch": 0.24, "learning_rate": 7.586206896551723e-08, "loss": 0.0689, "step": 903 }, { "epoch": 0.24, "learning_rate": 7.583533814488104e-08, "loss": 0.0438, "step": 904 }, { "epoch": 0.24, "learning_rate": 7.580860732424485e-08, "loss": 0.0425, "step": 905 }, { "epoch": 0.24, "learning_rate": 7.578187650360865e-08, "loss": 0.0714, "step": 906 }, { "epoch": 0.24, "learning_rate": 7.575514568297246e-08, "loss": 0.0579, "step": 907 }, { "epoch": 0.24, "learning_rate": 7.572841486233626e-08, "loss": 0.0564, "step": 908 }, { "epoch": 0.24, "learning_rate": 7.570168404170007e-08, "loss": 0.0666, "step": 909 }, { "epoch": 0.24, "learning_rate": 7.567495322106388e-08, "loss": 0.0588, "step": 910 }, { "epoch": 0.24, "learning_rate": 7.564822240042768e-08, "loss": 0.0694, "step": 911 }, { "epoch": 0.24, "learning_rate": 7.562149157979149e-08, "loss": 0.0404, "step": 912 }, { "epoch": 0.24, "learning_rate": 7.55947607591553e-08, "loss": 0.0758, "step": 913 }, { "epoch": 0.24, "learning_rate": 7.55680299385191e-08, "loss": 0.0463, "step": 914 }, { "epoch": 0.24, "learning_rate": 7.554129911788291e-08, "loss": 0.0502, "step": 915 }, { "epoch": 0.24, "learning_rate": 7.551456829724671e-08, "loss": 0.0948, "step": 916 }, { "epoch": 0.25, "learning_rate": 7.548783747661052e-08, "loss": 0.0652, "step": 917 }, { "epoch": 0.25, "learning_rate": 7.546110665597433e-08, "loss": 0.0956, "step": 918 }, { "epoch": 0.25, "learning_rate": 7.543437583533813e-08, "loss": 0.0503, "step": 919 }, { "epoch": 0.25, "learning_rate": 7.540764501470194e-08, "loss": 0.1055, "step": 920 }, { "epoch": 0.25, "learning_rate": 7.538091419406575e-08, "loss": 0.0703, "step": 921 }, { "epoch": 0.25, "learning_rate": 7.535418337342955e-08, "loss": 0.057, "step": 922 }, { "epoch": 0.25, "learning_rate": 7.532745255279336e-08, "loss": 0.08, "step": 923 }, { "epoch": 0.25, "learning_rate": 7.530072173215718e-08, "loss": 0.0594, "step": 924 }, { "epoch": 0.25, "learning_rate": 7.527399091152097e-08, "loss": 0.0736, "step": 925 }, { "epoch": 0.25, "learning_rate": 7.524726009088478e-08, "loss": 0.0603, "step": 926 }, { "epoch": 0.25, "learning_rate": 7.52205292702486e-08, "loss": 0.0765, "step": 927 }, { "epoch": 0.25, "learning_rate": 7.51937984496124e-08, "loss": 0.072, "step": 928 }, { "epoch": 0.25, "learning_rate": 7.51670676289762e-08, "loss": 0.0729, "step": 929 }, { "epoch": 0.25, "learning_rate": 7.514033680834002e-08, "loss": 0.0626, "step": 930 }, { "epoch": 0.25, "learning_rate": 7.511360598770381e-08, "loss": 0.0422, "step": 931 }, { "epoch": 0.25, "learning_rate": 7.508687516706763e-08, "loss": 0.0655, "step": 932 }, { "epoch": 0.25, "learning_rate": 7.506014434643144e-08, "loss": 0.0587, "step": 933 }, { "epoch": 0.25, "learning_rate": 7.503341352579524e-08, "loss": 0.0611, "step": 934 }, { "epoch": 0.25, "learning_rate": 7.500668270515905e-08, "loss": 0.0597, "step": 935 }, { "epoch": 0.25, "learning_rate": 7.497995188452285e-08, "loss": 0.0606, "step": 936 }, { "epoch": 0.25, "learning_rate": 7.495322106388666e-08, "loss": 0.076, "step": 937 }, { "epoch": 0.25, "learning_rate": 7.492649024325047e-08, "loss": 0.058, "step": 938 }, { "epoch": 0.25, "learning_rate": 7.489975942261427e-08, "loss": 0.0631, "step": 939 }, { "epoch": 0.25, "learning_rate": 7.487302860197808e-08, "loss": 0.0427, "step": 940 }, { "epoch": 0.25, "learning_rate": 7.484629778134189e-08, "loss": 0.0743, "step": 941 }, { "epoch": 0.25, "learning_rate": 7.481956696070569e-08, "loss": 0.0638, "step": 942 }, { "epoch": 0.25, "learning_rate": 7.47928361400695e-08, "loss": 0.0757, "step": 943 }, { "epoch": 0.25, "learning_rate": 7.476610531943331e-08, "loss": 0.064, "step": 944 }, { "epoch": 0.25, "learning_rate": 7.473937449879711e-08, "loss": 0.0644, "step": 945 }, { "epoch": 0.25, "learning_rate": 7.471264367816092e-08, "loss": 0.0465, "step": 946 }, { "epoch": 0.25, "learning_rate": 7.468591285752473e-08, "loss": 0.0697, "step": 947 }, { "epoch": 0.25, "learning_rate": 7.465918203688853e-08, "loss": 0.0827, "step": 948 }, { "epoch": 0.25, "learning_rate": 7.463245121625234e-08, "loss": 0.0756, "step": 949 }, { "epoch": 0.25, "learning_rate": 7.460572039561615e-08, "loss": 0.0535, "step": 950 }, { "epoch": 0.25, "learning_rate": 7.457898957497995e-08, "loss": 0.0711, "step": 951 }, { "epoch": 0.25, "learning_rate": 7.455225875434376e-08, "loss": 0.0539, "step": 952 }, { "epoch": 0.25, "learning_rate": 7.452552793370757e-08, "loss": 0.0395, "step": 953 }, { "epoch": 0.25, "learning_rate": 7.449879711307137e-08, "loss": 0.0488, "step": 954 }, { "epoch": 0.26, "learning_rate": 7.447206629243518e-08, "loss": 0.0361, "step": 955 }, { "epoch": 0.26, "learning_rate": 7.444533547179898e-08, "loss": 0.0821, "step": 956 }, { "epoch": 0.26, "learning_rate": 7.441860465116279e-08, "loss": 0.0714, "step": 957 }, { "epoch": 0.26, "learning_rate": 7.43918738305266e-08, "loss": 0.0568, "step": 958 }, { "epoch": 0.26, "learning_rate": 7.43651430098904e-08, "loss": 0.0477, "step": 959 }, { "epoch": 0.26, "learning_rate": 7.433841218925421e-08, "loss": 0.036, "step": 960 }, { "epoch": 0.26, "learning_rate": 7.431168136861802e-08, "loss": 0.0416, "step": 961 }, { "epoch": 0.26, "learning_rate": 7.428495054798182e-08, "loss": 0.0864, "step": 962 }, { "epoch": 0.26, "learning_rate": 7.425821972734563e-08, "loss": 0.0581, "step": 963 }, { "epoch": 0.26, "learning_rate": 7.423148890670943e-08, "loss": 0.065, "step": 964 }, { "epoch": 0.26, "learning_rate": 7.420475808607324e-08, "loss": 0.0589, "step": 965 }, { "epoch": 0.26, "learning_rate": 7.417802726543705e-08, "loss": 0.0394, "step": 966 }, { "epoch": 0.26, "learning_rate": 7.415129644480085e-08, "loss": 0.0675, "step": 967 }, { "epoch": 0.26, "learning_rate": 7.412456562416466e-08, "loss": 0.0626, "step": 968 }, { "epoch": 0.26, "learning_rate": 7.409783480352847e-08, "loss": 0.0593, "step": 969 }, { "epoch": 0.26, "learning_rate": 7.407110398289227e-08, "loss": 0.0695, "step": 970 }, { "epoch": 0.26, "learning_rate": 7.404437316225608e-08, "loss": 0.0563, "step": 971 }, { "epoch": 0.26, "learning_rate": 7.401764234161988e-08, "loss": 0.0463, "step": 972 }, { "epoch": 0.26, "learning_rate": 7.399091152098369e-08, "loss": 0.0789, "step": 973 }, { "epoch": 0.26, "learning_rate": 7.39641807003475e-08, "loss": 0.066, "step": 974 }, { "epoch": 0.26, "learning_rate": 7.39374498797113e-08, "loss": 0.054, "step": 975 }, { "epoch": 0.26, "learning_rate": 7.391071905907511e-08, "loss": 0.0713, "step": 976 }, { "epoch": 0.26, "learning_rate": 7.388398823843893e-08, "loss": 0.0629, "step": 977 }, { "epoch": 0.26, "learning_rate": 7.385725741780272e-08, "loss": 0.0824, "step": 978 }, { "epoch": 0.26, "learning_rate": 7.383052659716653e-08, "loss": 0.0715, "step": 979 }, { "epoch": 0.26, "learning_rate": 7.380379577653033e-08, "loss": 0.0802, "step": 980 }, { "epoch": 0.26, "learning_rate": 7.377706495589414e-08, "loss": 0.0576, "step": 981 }, { "epoch": 0.26, "learning_rate": 7.375033413525796e-08, "loss": 0.0566, "step": 982 }, { "epoch": 0.26, "learning_rate": 7.372360331462175e-08, "loss": 0.0567, "step": 983 }, { "epoch": 0.26, "learning_rate": 7.369687249398557e-08, "loss": 0.0536, "step": 984 }, { "epoch": 0.26, "learning_rate": 7.367014167334938e-08, "loss": 0.0651, "step": 985 }, { "epoch": 0.26, "learning_rate": 7.364341085271317e-08, "loss": 0.0834, "step": 986 }, { "epoch": 0.26, "learning_rate": 7.361668003207699e-08, "loss": 0.0624, "step": 987 }, { "epoch": 0.26, "learning_rate": 7.358994921144078e-08, "loss": 0.0558, "step": 988 }, { "epoch": 0.26, "learning_rate": 7.35632183908046e-08, "loss": 0.0449, "step": 989 }, { "epoch": 0.26, "learning_rate": 7.353648757016841e-08, "loss": 0.072, "step": 990 }, { "epoch": 0.26, "learning_rate": 7.35097567495322e-08, "loss": 0.0868, "step": 991 }, { "epoch": 0.27, "learning_rate": 7.348302592889602e-08, "loss": 0.0562, "step": 992 }, { "epoch": 0.27, "learning_rate": 7.345629510825983e-08, "loss": 0.0644, "step": 993 }, { "epoch": 0.27, "learning_rate": 7.342956428762363e-08, "loss": 0.0737, "step": 994 }, { "epoch": 0.27, "learning_rate": 7.340283346698744e-08, "loss": 0.0669, "step": 995 }, { "epoch": 0.27, "learning_rate": 7.337610264635124e-08, "loss": 0.0437, "step": 996 }, { "epoch": 0.27, "learning_rate": 7.334937182571505e-08, "loss": 0.0491, "step": 997 }, { "epoch": 0.27, "learning_rate": 7.332264100507886e-08, "loss": 0.0615, "step": 998 }, { "epoch": 0.27, "learning_rate": 7.329591018444266e-08, "loss": 0.0612, "step": 999 }, { "epoch": 0.27, "learning_rate": 7.326917936380647e-08, "loss": 0.0667, "step": 1000 }, { "epoch": 0.27, "learning_rate": 7.324244854317028e-08, "loss": 0.048, "step": 1001 }, { "epoch": 0.27, "learning_rate": 7.321571772253408e-08, "loss": 0.057, "step": 1002 }, { "epoch": 0.27, "learning_rate": 7.318898690189789e-08, "loss": 0.0607, "step": 1003 }, { "epoch": 0.27, "learning_rate": 7.316225608126169e-08, "loss": 0.0512, "step": 1004 }, { "epoch": 0.27, "learning_rate": 7.31355252606255e-08, "loss": 0.0744, "step": 1005 }, { "epoch": 0.27, "learning_rate": 7.310879443998931e-08, "loss": 0.09, "step": 1006 }, { "epoch": 0.27, "learning_rate": 7.308206361935311e-08, "loss": 0.0724, "step": 1007 }, { "epoch": 0.27, "learning_rate": 7.305533279871692e-08, "loss": 0.0758, "step": 1008 }, { "epoch": 0.27, "learning_rate": 7.302860197808073e-08, "loss": 0.0607, "step": 1009 }, { "epoch": 0.27, "learning_rate": 7.300187115744453e-08, "loss": 0.1092, "step": 1010 }, { "epoch": 0.27, "learning_rate": 7.297514033680834e-08, "loss": 0.0684, "step": 1011 }, { "epoch": 0.27, "learning_rate": 7.294840951617214e-08, "loss": 0.0733, "step": 1012 }, { "epoch": 0.27, "learning_rate": 7.292167869553595e-08, "loss": 0.051, "step": 1013 }, { "epoch": 0.27, "learning_rate": 7.289494787489976e-08, "loss": 0.0571, "step": 1014 }, { "epoch": 0.27, "learning_rate": 7.286821705426356e-08, "loss": 0.0691, "step": 1015 }, { "epoch": 0.27, "learning_rate": 7.284148623362737e-08, "loss": 0.0676, "step": 1016 }, { "epoch": 0.27, "learning_rate": 7.281475541299118e-08, "loss": 0.0661, "step": 1017 }, { "epoch": 0.27, "learning_rate": 7.278802459235498e-08, "loss": 0.0687, "step": 1018 }, { "epoch": 0.27, "learning_rate": 7.276129377171879e-08, "loss": 0.0745, "step": 1019 }, { "epoch": 0.27, "learning_rate": 7.273456295108259e-08, "loss": 0.0633, "step": 1020 }, { "epoch": 0.27, "learning_rate": 7.27078321304464e-08, "loss": 0.0489, "step": 1021 }, { "epoch": 0.27, "learning_rate": 7.268110130981021e-08, "loss": 0.064, "step": 1022 }, { "epoch": 0.27, "learning_rate": 7.265437048917401e-08, "loss": 0.0517, "step": 1023 }, { "epoch": 0.27, "learning_rate": 7.262763966853782e-08, "loss": 0.0619, "step": 1024 }, { "epoch": 0.27, "learning_rate": 7.260090884790163e-08, "loss": 0.0571, "step": 1025 }, { "epoch": 0.27, "learning_rate": 7.257417802726543e-08, "loss": 0.0629, "step": 1026 }, { "epoch": 0.27, "learning_rate": 7.254744720662924e-08, "loss": 0.0615, "step": 1027 }, { "epoch": 0.27, "learning_rate": 7.252071638599304e-08, "loss": 0.0707, "step": 1028 }, { "epoch": 0.28, "learning_rate": 7.249398556535685e-08, "loss": 0.0629, "step": 1029 }, { "epoch": 0.28, "learning_rate": 7.246725474472066e-08, "loss": 0.0659, "step": 1030 }, { "epoch": 0.28, "learning_rate": 7.244052392408446e-08, "loss": 0.0572, "step": 1031 }, { "epoch": 0.28, "learning_rate": 7.241379310344827e-08, "loss": 0.0658, "step": 1032 }, { "epoch": 0.28, "learning_rate": 7.238706228281208e-08, "loss": 0.0796, "step": 1033 }, { "epoch": 0.28, "learning_rate": 7.236033146217588e-08, "loss": 0.0605, "step": 1034 }, { "epoch": 0.28, "learning_rate": 7.233360064153969e-08, "loss": 0.0585, "step": 1035 }, { "epoch": 0.28, "learning_rate": 7.230686982090349e-08, "loss": 0.0726, "step": 1036 }, { "epoch": 0.28, "learning_rate": 7.22801390002673e-08, "loss": 0.0615, "step": 1037 }, { "epoch": 0.28, "learning_rate": 7.225340817963111e-08, "loss": 0.0616, "step": 1038 }, { "epoch": 0.28, "learning_rate": 7.222667735899491e-08, "loss": 0.0971, "step": 1039 }, { "epoch": 0.28, "learning_rate": 7.219994653835872e-08, "loss": 0.0853, "step": 1040 }, { "epoch": 0.28, "learning_rate": 7.217321571772253e-08, "loss": 0.0736, "step": 1041 }, { "epoch": 0.28, "learning_rate": 7.214648489708633e-08, "loss": 0.0736, "step": 1042 }, { "epoch": 0.28, "learning_rate": 7.211975407645014e-08, "loss": 0.0736, "step": 1043 }, { "epoch": 0.28, "learning_rate": 7.209302325581394e-08, "loss": 0.0686, "step": 1044 }, { "epoch": 0.28, "learning_rate": 7.206629243517775e-08, "loss": 0.0561, "step": 1045 }, { "epoch": 0.28, "learning_rate": 7.203956161454156e-08, "loss": 0.0436, "step": 1046 }, { "epoch": 0.28, "learning_rate": 7.201283079390536e-08, "loss": 0.0482, "step": 1047 }, { "epoch": 0.28, "learning_rate": 7.198609997326917e-08, "loss": 0.0505, "step": 1048 }, { "epoch": 0.28, "learning_rate": 7.195936915263299e-08, "loss": 0.0603, "step": 1049 }, { "epoch": 0.28, "learning_rate": 7.193263833199678e-08, "loss": 0.059, "step": 1050 }, { "epoch": 0.28, "learning_rate": 7.19059075113606e-08, "loss": 0.0739, "step": 1051 }, { "epoch": 0.28, "learning_rate": 7.187917669072439e-08, "loss": 0.0723, "step": 1052 }, { "epoch": 0.28, "learning_rate": 7.18524458700882e-08, "loss": 0.0499, "step": 1053 }, { "epoch": 0.28, "learning_rate": 7.182571504945202e-08, "loss": 0.0612, "step": 1054 }, { "epoch": 0.28, "learning_rate": 7.179898422881581e-08, "loss": 0.0612, "step": 1055 }, { "epoch": 0.28, "learning_rate": 7.177225340817963e-08, "loss": 0.0692, "step": 1056 }, { "epoch": 0.28, "learning_rate": 7.174552258754344e-08, "loss": 0.1065, "step": 1057 }, { "epoch": 0.28, "learning_rate": 7.171879176690724e-08, "loss": 0.0578, "step": 1058 }, { "epoch": 0.28, "learning_rate": 7.169206094627105e-08, "loss": 0.0534, "step": 1059 }, { "epoch": 0.28, "learning_rate": 7.166533012563484e-08, "loss": 0.0571, "step": 1060 }, { "epoch": 0.28, "learning_rate": 7.163859930499866e-08, "loss": 0.051, "step": 1061 }, { "epoch": 0.28, "learning_rate": 7.161186848436247e-08, "loss": 0.0496, "step": 1062 }, { "epoch": 0.28, "learning_rate": 7.158513766372627e-08, "loss": 0.0587, "step": 1063 }, { "epoch": 0.28, "learning_rate": 7.155840684309008e-08, "loss": 0.0638, "step": 1064 }, { "epoch": 0.28, "learning_rate": 7.153167602245389e-08, "loss": 0.0471, "step": 1065 }, { "epoch": 0.28, "learning_rate": 7.150494520181769e-08, "loss": 0.0523, "step": 1066 }, { "epoch": 0.29, "learning_rate": 7.14782143811815e-08, "loss": 0.0636, "step": 1067 }, { "epoch": 0.29, "learning_rate": 7.14514835605453e-08, "loss": 0.0662, "step": 1068 }, { "epoch": 0.29, "learning_rate": 7.142475273990911e-08, "loss": 0.0737, "step": 1069 }, { "epoch": 0.29, "learning_rate": 7.139802191927292e-08, "loss": 0.0829, "step": 1070 }, { "epoch": 0.29, "learning_rate": 7.137129109863672e-08, "loss": 0.0609, "step": 1071 }, { "epoch": 0.29, "learning_rate": 7.134456027800053e-08, "loss": 0.0736, "step": 1072 }, { "epoch": 0.29, "learning_rate": 7.131782945736434e-08, "loss": 0.0782, "step": 1073 }, { "epoch": 0.29, "learning_rate": 7.129109863672814e-08, "loss": 0.054, "step": 1074 }, { "epoch": 0.29, "learning_rate": 7.126436781609195e-08, "loss": 0.0718, "step": 1075 }, { "epoch": 0.29, "learning_rate": 7.123763699545575e-08, "loss": 0.0502, "step": 1076 }, { "epoch": 0.29, "learning_rate": 7.121090617481956e-08, "loss": 0.0692, "step": 1077 }, { "epoch": 0.29, "learning_rate": 7.118417535418337e-08, "loss": 0.0551, "step": 1078 }, { "epoch": 0.29, "learning_rate": 7.115744453354717e-08, "loss": 0.0494, "step": 1079 }, { "epoch": 0.29, "learning_rate": 7.113071371291098e-08, "loss": 0.0735, "step": 1080 }, { "epoch": 0.29, "learning_rate": 7.110398289227479e-08, "loss": 0.0539, "step": 1081 }, { "epoch": 0.29, "learning_rate": 7.107725207163859e-08, "loss": 0.0575, "step": 1082 }, { "epoch": 0.29, "learning_rate": 7.10505212510024e-08, "loss": 0.0668, "step": 1083 }, { "epoch": 0.29, "learning_rate": 7.10237904303662e-08, "loss": 0.0613, "step": 1084 }, { "epoch": 0.29, "learning_rate": 7.099705960973001e-08, "loss": 0.0708, "step": 1085 }, { "epoch": 0.29, "learning_rate": 7.097032878909382e-08, "loss": 0.0715, "step": 1086 }, { "epoch": 0.29, "learning_rate": 7.094359796845762e-08, "loss": 0.0642, "step": 1087 }, { "epoch": 0.29, "learning_rate": 7.091686714782143e-08, "loss": 0.0783, "step": 1088 }, { "epoch": 0.29, "learning_rate": 7.089013632718524e-08, "loss": 0.076, "step": 1089 }, { "epoch": 0.29, "learning_rate": 7.086340550654904e-08, "loss": 0.0817, "step": 1090 }, { "epoch": 0.29, "learning_rate": 7.083667468591285e-08, "loss": 0.0515, "step": 1091 }, { "epoch": 0.29, "learning_rate": 7.080994386527665e-08, "loss": 0.0577, "step": 1092 }, { "epoch": 0.29, "learning_rate": 7.078321304464046e-08, "loss": 0.0722, "step": 1093 }, { "epoch": 0.29, "learning_rate": 7.075648222400427e-08, "loss": 0.0566, "step": 1094 }, { "epoch": 0.29, "learning_rate": 7.072975140336807e-08, "loss": 0.0695, "step": 1095 }, { "epoch": 0.29, "learning_rate": 7.070302058273188e-08, "loss": 0.0558, "step": 1096 }, { "epoch": 0.29, "learning_rate": 7.067628976209569e-08, "loss": 0.0563, "step": 1097 }, { "epoch": 0.29, "learning_rate": 7.064955894145949e-08, "loss": 0.0609, "step": 1098 }, { "epoch": 0.29, "learning_rate": 7.06228281208233e-08, "loss": 0.0496, "step": 1099 }, { "epoch": 0.29, "learning_rate": 7.059609730018711e-08, "loss": 0.0711, "step": 1100 }, { "epoch": 0.29, "learning_rate": 7.056936647955091e-08, "loss": 0.0576, "step": 1101 }, { "epoch": 0.29, "learning_rate": 7.054263565891472e-08, "loss": 0.0487, "step": 1102 }, { "epoch": 0.29, "learning_rate": 7.051590483827853e-08, "loss": 0.0585, "step": 1103 }, { "epoch": 0.3, "learning_rate": 7.048917401764233e-08, "loss": 0.0591, "step": 1104 }, { "epoch": 0.3, "learning_rate": 7.046244319700614e-08, "loss": 0.0539, "step": 1105 }, { "epoch": 0.3, "learning_rate": 7.043571237636996e-08, "loss": 0.0584, "step": 1106 }, { "epoch": 0.3, "learning_rate": 7.040898155573375e-08, "loss": 0.0746, "step": 1107 }, { "epoch": 0.3, "learning_rate": 7.038225073509756e-08, "loss": 0.073, "step": 1108 }, { "epoch": 0.3, "learning_rate": 7.035551991446138e-08, "loss": 0.0739, "step": 1109 }, { "epoch": 0.3, "learning_rate": 7.032878909382517e-08, "loss": 0.0533, "step": 1110 }, { "epoch": 0.3, "learning_rate": 7.030205827318899e-08, "loss": 0.0713, "step": 1111 }, { "epoch": 0.3, "learning_rate": 7.02753274525528e-08, "loss": 0.0463, "step": 1112 }, { "epoch": 0.3, "learning_rate": 7.02485966319166e-08, "loss": 0.0511, "step": 1113 }, { "epoch": 0.3, "learning_rate": 7.02218658112804e-08, "loss": 0.0595, "step": 1114 }, { "epoch": 0.3, "learning_rate": 7.019513499064422e-08, "loss": 0.0648, "step": 1115 }, { "epoch": 0.3, "learning_rate": 7.016840417000802e-08, "loss": 0.0764, "step": 1116 }, { "epoch": 0.3, "learning_rate": 7.014167334937183e-08, "loss": 0.0705, "step": 1117 }, { "epoch": 0.3, "learning_rate": 7.011494252873564e-08, "loss": 0.0659, "step": 1118 }, { "epoch": 0.3, "learning_rate": 7.008821170809944e-08, "loss": 0.0687, "step": 1119 }, { "epoch": 0.3, "learning_rate": 7.006148088746325e-08, "loss": 0.0598, "step": 1120 }, { "epoch": 0.3, "learning_rate": 7.003475006682706e-08, "loss": 0.0497, "step": 1121 }, { "epoch": 0.3, "learning_rate": 7.000801924619086e-08, "loss": 0.0796, "step": 1122 }, { "epoch": 0.3, "learning_rate": 6.998128842555467e-08, "loss": 0.064, "step": 1123 }, { "epoch": 0.3, "learning_rate": 6.995455760491847e-08, "loss": 0.0621, "step": 1124 }, { "epoch": 0.3, "learning_rate": 6.992782678428228e-08, "loss": 0.0715, "step": 1125 }, { "epoch": 0.3, "learning_rate": 6.990109596364609e-08, "loss": 0.0633, "step": 1126 }, { "epoch": 0.3, "learning_rate": 6.987436514300989e-08, "loss": 0.0673, "step": 1127 }, { "epoch": 0.3, "learning_rate": 6.98476343223737e-08, "loss": 0.0497, "step": 1128 }, { "epoch": 0.3, "learning_rate": 6.982090350173751e-08, "loss": 0.0525, "step": 1129 }, { "epoch": 0.3, "learning_rate": 6.979417268110131e-08, "loss": 0.0557, "step": 1130 }, { "epoch": 0.3, "learning_rate": 6.976744186046512e-08, "loss": 0.0435, "step": 1131 }, { "epoch": 0.3, "learning_rate": 6.974071103982892e-08, "loss": 0.0702, "step": 1132 }, { "epoch": 0.3, "learning_rate": 6.971398021919273e-08, "loss": 0.0913, "step": 1133 }, { "epoch": 0.3, "learning_rate": 6.968724939855654e-08, "loss": 0.0744, "step": 1134 }, { "epoch": 0.3, "learning_rate": 6.966051857792034e-08, "loss": 0.0676, "step": 1135 }, { "epoch": 0.3, "learning_rate": 6.963378775728415e-08, "loss": 0.0551, "step": 1136 }, { "epoch": 0.3, "learning_rate": 6.960705693664796e-08, "loss": 0.0851, "step": 1137 }, { "epoch": 0.3, "learning_rate": 6.958032611601176e-08, "loss": 0.0528, "step": 1138 }, { "epoch": 0.3, "learning_rate": 6.955359529537557e-08, "loss": 0.0758, "step": 1139 }, { "epoch": 0.3, "learning_rate": 6.952686447473937e-08, "loss": 0.0738, "step": 1140 }, { "epoch": 0.3, "learning_rate": 6.950013365410318e-08, "loss": 0.0544, "step": 1141 }, { "epoch": 0.31, "learning_rate": 6.947340283346699e-08, "loss": 0.069, "step": 1142 }, { "epoch": 0.31, "learning_rate": 6.944667201283079e-08, "loss": 0.0555, "step": 1143 }, { "epoch": 0.31, "learning_rate": 6.94199411921946e-08, "loss": 0.0651, "step": 1144 }, { "epoch": 0.31, "learning_rate": 6.939321037155841e-08, "loss": 0.0537, "step": 1145 }, { "epoch": 0.31, "learning_rate": 6.936647955092221e-08, "loss": 0.079, "step": 1146 }, { "epoch": 0.31, "learning_rate": 6.933974873028602e-08, "loss": 0.0651, "step": 1147 }, { "epoch": 0.31, "learning_rate": 6.931301790964982e-08, "loss": 0.0967, "step": 1148 }, { "epoch": 0.31, "learning_rate": 6.928628708901363e-08, "loss": 0.0468, "step": 1149 }, { "epoch": 0.31, "learning_rate": 6.925955626837744e-08, "loss": 0.056, "step": 1150 }, { "epoch": 0.31, "learning_rate": 6.923282544774124e-08, "loss": 0.0553, "step": 1151 }, { "epoch": 0.31, "learning_rate": 6.920609462710505e-08, "loss": 0.0657, "step": 1152 }, { "epoch": 0.31, "learning_rate": 6.917936380646886e-08, "loss": 0.0665, "step": 1153 }, { "epoch": 0.31, "learning_rate": 6.915263298583266e-08, "loss": 0.0644, "step": 1154 }, { "epoch": 0.31, "learning_rate": 6.912590216519647e-08, "loss": 0.0574, "step": 1155 }, { "epoch": 0.31, "learning_rate": 6.909917134456027e-08, "loss": 0.066, "step": 1156 }, { "epoch": 0.31, "learning_rate": 6.907244052392408e-08, "loss": 0.0589, "step": 1157 }, { "epoch": 0.31, "learning_rate": 6.90457097032879e-08, "loss": 0.0633, "step": 1158 }, { "epoch": 0.31, "learning_rate": 6.901897888265169e-08, "loss": 0.0603, "step": 1159 }, { "epoch": 0.31, "learning_rate": 6.89922480620155e-08, "loss": 0.0422, "step": 1160 }, { "epoch": 0.31, "learning_rate": 6.896551724137931e-08, "loss": 0.0737, "step": 1161 }, { "epoch": 0.31, "learning_rate": 6.893878642074311e-08, "loss": 0.0613, "step": 1162 }, { "epoch": 0.31, "learning_rate": 6.891205560010692e-08, "loss": 0.0613, "step": 1163 }, { "epoch": 0.31, "learning_rate": 6.888532477947072e-08, "loss": 0.0445, "step": 1164 }, { "epoch": 0.31, "learning_rate": 6.885859395883453e-08, "loss": 0.0654, "step": 1165 }, { "epoch": 0.31, "learning_rate": 6.883186313819835e-08, "loss": 0.0648, "step": 1166 }, { "epoch": 0.31, "learning_rate": 6.880513231756214e-08, "loss": 0.0738, "step": 1167 }, { "epoch": 0.31, "learning_rate": 6.877840149692595e-08, "loss": 0.0991, "step": 1168 }, { "epoch": 0.31, "learning_rate": 6.875167067628977e-08, "loss": 0.0826, "step": 1169 }, { "epoch": 0.31, "learning_rate": 6.872493985565356e-08, "loss": 0.0746, "step": 1170 }, { "epoch": 0.31, "learning_rate": 6.869820903501738e-08, "loss": 0.0468, "step": 1171 }, { "epoch": 0.31, "learning_rate": 6.867147821438117e-08, "loss": 0.0571, "step": 1172 }, { "epoch": 0.31, "learning_rate": 6.864474739374499e-08, "loss": 0.049, "step": 1173 }, { "epoch": 0.31, "learning_rate": 6.86180165731088e-08, "loss": 0.048, "step": 1174 }, { "epoch": 0.31, "learning_rate": 6.85912857524726e-08, "loss": 0.0489, "step": 1175 }, { "epoch": 0.31, "learning_rate": 6.85645549318364e-08, "loss": 0.0422, "step": 1176 }, { "epoch": 0.31, "learning_rate": 6.853782411120022e-08, "loss": 0.0801, "step": 1177 }, { "epoch": 0.31, "learning_rate": 6.851109329056402e-08, "loss": 0.0737, "step": 1178 }, { "epoch": 0.32, "learning_rate": 6.848436246992783e-08, "loss": 0.0647, "step": 1179 }, { "epoch": 0.32, "learning_rate": 6.845763164929162e-08, "loss": 0.0621, "step": 1180 }, { "epoch": 0.32, "learning_rate": 6.843090082865544e-08, "loss": 0.0611, "step": 1181 }, { "epoch": 0.32, "learning_rate": 6.840417000801925e-08, "loss": 0.0645, "step": 1182 }, { "epoch": 0.32, "learning_rate": 6.837743918738305e-08, "loss": 0.0613, "step": 1183 }, { "epoch": 0.32, "learning_rate": 6.835070836674686e-08, "loss": 0.0686, "step": 1184 }, { "epoch": 0.32, "learning_rate": 6.832397754611067e-08, "loss": 0.076, "step": 1185 }, { "epoch": 0.32, "learning_rate": 6.829724672547447e-08, "loss": 0.0558, "step": 1186 }, { "epoch": 0.32, "learning_rate": 6.827051590483828e-08, "loss": 0.0537, "step": 1187 }, { "epoch": 0.32, "learning_rate": 6.824378508420208e-08, "loss": 0.0513, "step": 1188 }, { "epoch": 0.32, "learning_rate": 6.821705426356589e-08, "loss": 0.0753, "step": 1189 }, { "epoch": 0.32, "learning_rate": 6.81903234429297e-08, "loss": 0.0667, "step": 1190 }, { "epoch": 0.32, "learning_rate": 6.81635926222935e-08, "loss": 0.0654, "step": 1191 }, { "epoch": 0.32, "learning_rate": 6.813686180165731e-08, "loss": 0.0615, "step": 1192 }, { "epoch": 0.32, "learning_rate": 6.811013098102112e-08, "loss": 0.0711, "step": 1193 }, { "epoch": 0.32, "learning_rate": 6.808340016038492e-08, "loss": 0.0786, "step": 1194 }, { "epoch": 0.32, "learning_rate": 6.805666933974873e-08, "loss": 0.0431, "step": 1195 }, { "epoch": 0.32, "learning_rate": 6.802993851911253e-08, "loss": 0.0566, "step": 1196 }, { "epoch": 0.32, "learning_rate": 6.800320769847634e-08, "loss": 0.0759, "step": 1197 }, { "epoch": 0.32, "learning_rate": 6.797647687784015e-08, "loss": 0.058, "step": 1198 }, { "epoch": 0.32, "learning_rate": 6.794974605720395e-08, "loss": 0.0622, "step": 1199 }, { "epoch": 0.32, "learning_rate": 6.792301523656776e-08, "loss": 0.0433, "step": 1200 }, { "epoch": 0.32, "learning_rate": 6.789628441593157e-08, "loss": 0.0669, "step": 1201 }, { "epoch": 0.32, "learning_rate": 6.786955359529537e-08, "loss": 0.0659, "step": 1202 }, { "epoch": 0.32, "learning_rate": 6.784282277465918e-08, "loss": 0.0687, "step": 1203 }, { "epoch": 0.32, "learning_rate": 6.781609195402298e-08, "loss": 0.0562, "step": 1204 }, { "epoch": 0.32, "learning_rate": 6.778936113338679e-08, "loss": 0.0698, "step": 1205 }, { "epoch": 0.32, "learning_rate": 6.77626303127506e-08, "loss": 0.0572, "step": 1206 }, { "epoch": 0.32, "learning_rate": 6.77358994921144e-08, "loss": 0.0601, "step": 1207 }, { "epoch": 0.32, "learning_rate": 6.770916867147821e-08, "loss": 0.0625, "step": 1208 }, { "epoch": 0.32, "learning_rate": 6.768243785084202e-08, "loss": 0.0653, "step": 1209 }, { "epoch": 0.32, "learning_rate": 6.765570703020582e-08, "loss": 0.0637, "step": 1210 }, { "epoch": 0.32, "learning_rate": 6.762897620956963e-08, "loss": 0.0695, "step": 1211 }, { "epoch": 0.32, "learning_rate": 6.760224538893343e-08, "loss": 0.0556, "step": 1212 }, { "epoch": 0.32, "learning_rate": 6.757551456829724e-08, "loss": 0.0532, "step": 1213 }, { "epoch": 0.32, "learning_rate": 6.754878374766105e-08, "loss": 0.0538, "step": 1214 }, { "epoch": 0.32, "learning_rate": 6.752205292702485e-08, "loss": 0.0551, "step": 1215 }, { "epoch": 0.33, "learning_rate": 6.749532210638866e-08, "loss": 0.0676, "step": 1216 }, { "epoch": 0.33, "learning_rate": 6.746859128575247e-08, "loss": 0.0599, "step": 1217 }, { "epoch": 0.33, "learning_rate": 6.744186046511627e-08, "loss": 0.0566, "step": 1218 }, { "epoch": 0.33, "learning_rate": 6.741512964448008e-08, "loss": 0.0628, "step": 1219 }, { "epoch": 0.33, "learning_rate": 6.738839882384388e-08, "loss": 0.049, "step": 1220 }, { "epoch": 0.33, "learning_rate": 6.736166800320769e-08, "loss": 0.0425, "step": 1221 }, { "epoch": 0.33, "learning_rate": 6.73349371825715e-08, "loss": 0.0717, "step": 1222 }, { "epoch": 0.33, "learning_rate": 6.73082063619353e-08, "loss": 0.0461, "step": 1223 }, { "epoch": 0.33, "learning_rate": 6.728147554129911e-08, "loss": 0.0709, "step": 1224 }, { "epoch": 0.33, "learning_rate": 6.725474472066292e-08, "loss": 0.0542, "step": 1225 }, { "epoch": 0.33, "learning_rate": 6.722801390002672e-08, "loss": 0.0537, "step": 1226 }, { "epoch": 0.33, "learning_rate": 6.720128307939053e-08, "loss": 0.0605, "step": 1227 }, { "epoch": 0.33, "learning_rate": 6.717455225875433e-08, "loss": 0.0583, "step": 1228 }, { "epoch": 0.33, "learning_rate": 6.714782143811814e-08, "loss": 0.046, "step": 1229 }, { "epoch": 0.33, "learning_rate": 6.712109061748195e-08, "loss": 0.0559, "step": 1230 }, { "epoch": 0.33, "learning_rate": 6.709435979684575e-08, "loss": 0.0525, "step": 1231 }, { "epoch": 0.33, "learning_rate": 6.706762897620956e-08, "loss": 0.0906, "step": 1232 }, { "epoch": 0.33, "learning_rate": 6.704089815557338e-08, "loss": 0.0891, "step": 1233 }, { "epoch": 0.33, "learning_rate": 6.701416733493717e-08, "loss": 0.0748, "step": 1234 }, { "epoch": 0.33, "learning_rate": 6.698743651430098e-08, "loss": 0.0866, "step": 1235 }, { "epoch": 0.33, "learning_rate": 6.696070569366478e-08, "loss": 0.0751, "step": 1236 }, { "epoch": 0.33, "learning_rate": 6.69339748730286e-08, "loss": 0.0531, "step": 1237 }, { "epoch": 0.33, "learning_rate": 6.69072440523924e-08, "loss": 0.0645, "step": 1238 }, { "epoch": 0.33, "learning_rate": 6.68805132317562e-08, "loss": 0.0609, "step": 1239 }, { "epoch": 0.33, "learning_rate": 6.685378241112002e-08, "loss": 0.0592, "step": 1240 }, { "epoch": 0.33, "learning_rate": 6.682705159048383e-08, "loss": 0.0852, "step": 1241 }, { "epoch": 0.33, "learning_rate": 6.680032076984762e-08, "loss": 0.0501, "step": 1242 }, { "epoch": 0.33, "learning_rate": 6.677358994921144e-08, "loss": 0.0503, "step": 1243 }, { "epoch": 0.33, "learning_rate": 6.674685912857523e-08, "loss": 0.0472, "step": 1244 }, { "epoch": 0.33, "learning_rate": 6.672012830793905e-08, "loss": 0.0906, "step": 1245 }, { "epoch": 0.33, "learning_rate": 6.669339748730286e-08, "loss": 0.0421, "step": 1246 }, { "epoch": 0.33, "learning_rate": 6.666666666666665e-08, "loss": 0.0622, "step": 1247 }, { "epoch": 0.33, "learning_rate": 6.663993584603047e-08, "loss": 0.0913, "step": 1248 }, { "epoch": 0.33, "learning_rate": 6.661320502539428e-08, "loss": 0.0535, "step": 1249 }, { "epoch": 0.33, "learning_rate": 6.658647420475808e-08, "loss": 0.06, "step": 1250 }, { "epoch": 0.33, "learning_rate": 6.655974338412189e-08, "loss": 0.0717, "step": 1251 }, { "epoch": 0.33, "learning_rate": 6.653301256348569e-08, "loss": 0.0877, "step": 1252 }, { "epoch": 0.33, "learning_rate": 6.65062817428495e-08, "loss": 0.07, "step": 1253 }, { "epoch": 0.34, "learning_rate": 6.647955092221331e-08, "loss": 0.0535, "step": 1254 }, { "epoch": 0.34, "learning_rate": 6.64528201015771e-08, "loss": 0.0702, "step": 1255 }, { "epoch": 0.34, "learning_rate": 6.642608928094092e-08, "loss": 0.059, "step": 1256 }, { "epoch": 0.34, "learning_rate": 6.639935846030473e-08, "loss": 0.0638, "step": 1257 }, { "epoch": 0.34, "learning_rate": 6.637262763966853e-08, "loss": 0.083, "step": 1258 }, { "epoch": 0.34, "learning_rate": 6.634589681903234e-08, "loss": 0.0753, "step": 1259 }, { "epoch": 0.34, "learning_rate": 6.631916599839614e-08, "loss": 0.0587, "step": 1260 }, { "epoch": 0.34, "learning_rate": 6.629243517775995e-08, "loss": 0.0588, "step": 1261 }, { "epoch": 0.34, "learning_rate": 6.626570435712376e-08, "loss": 0.0715, "step": 1262 }, { "epoch": 0.34, "learning_rate": 6.623897353648756e-08, "loss": 0.0551, "step": 1263 }, { "epoch": 0.34, "learning_rate": 6.621224271585137e-08, "loss": 0.0792, "step": 1264 }, { "epoch": 0.34, "learning_rate": 6.618551189521518e-08, "loss": 0.046, "step": 1265 }, { "epoch": 0.34, "learning_rate": 6.615878107457898e-08, "loss": 0.0693, "step": 1266 }, { "epoch": 0.34, "learning_rate": 6.613205025394279e-08, "loss": 0.0663, "step": 1267 }, { "epoch": 0.34, "learning_rate": 6.61053194333066e-08, "loss": 0.0585, "step": 1268 }, { "epoch": 0.34, "learning_rate": 6.60785886126704e-08, "loss": 0.0507, "step": 1269 }, { "epoch": 0.34, "learning_rate": 6.605185779203421e-08, "loss": 0.0594, "step": 1270 }, { "epoch": 0.34, "learning_rate": 6.602512697139802e-08, "loss": 0.0518, "step": 1271 }, { "epoch": 0.34, "learning_rate": 6.599839615076182e-08, "loss": 0.0551, "step": 1272 }, { "epoch": 0.34, "learning_rate": 6.597166533012563e-08, "loss": 0.0613, "step": 1273 }, { "epoch": 0.34, "learning_rate": 6.594493450948944e-08, "loss": 0.0557, "step": 1274 }, { "epoch": 0.34, "learning_rate": 6.591820368885324e-08, "loss": 0.0649, "step": 1275 }, { "epoch": 0.34, "learning_rate": 6.589147286821705e-08, "loss": 0.0587, "step": 1276 }, { "epoch": 0.34, "learning_rate": 6.586474204758086e-08, "loss": 0.082, "step": 1277 }, { "epoch": 0.34, "learning_rate": 6.583801122694466e-08, "loss": 0.0723, "step": 1278 }, { "epoch": 0.34, "learning_rate": 6.581128040630847e-08, "loss": 0.0523, "step": 1279 }, { "epoch": 0.34, "learning_rate": 6.578454958567228e-08, "loss": 0.0753, "step": 1280 }, { "epoch": 0.34, "learning_rate": 6.575781876503608e-08, "loss": 0.063, "step": 1281 }, { "epoch": 0.34, "learning_rate": 6.57310879443999e-08, "loss": 0.0732, "step": 1282 }, { "epoch": 0.34, "learning_rate": 6.57043571237637e-08, "loss": 0.0678, "step": 1283 }, { "epoch": 0.34, "learning_rate": 6.56776263031275e-08, "loss": 0.1202, "step": 1284 }, { "epoch": 0.34, "learning_rate": 6.565089548249131e-08, "loss": 0.0538, "step": 1285 }, { "epoch": 0.34, "learning_rate": 6.562416466185513e-08, "loss": 0.0829, "step": 1286 }, { "epoch": 0.34, "learning_rate": 6.559743384121892e-08, "loss": 0.0546, "step": 1287 }, { "epoch": 0.34, "learning_rate": 6.557070302058274e-08, "loss": 0.0527, "step": 1288 }, { "epoch": 0.34, "learning_rate": 6.554397219994655e-08, "loss": 0.0711, "step": 1289 }, { "epoch": 0.34, "learning_rate": 6.551724137931034e-08, "loss": 0.0614, "step": 1290 }, { "epoch": 0.35, "learning_rate": 6.549051055867416e-08, "loss": 0.0699, "step": 1291 }, { "epoch": 0.35, "learning_rate": 6.546377973803795e-08, "loss": 0.0492, "step": 1292 }, { "epoch": 0.35, "learning_rate": 6.543704891740177e-08, "loss": 0.0675, "step": 1293 }, { "epoch": 0.35, "learning_rate": 6.541031809676558e-08, "loss": 0.0536, "step": 1294 }, { "epoch": 0.35, "learning_rate": 6.538358727612937e-08, "loss": 0.0626, "step": 1295 }, { "epoch": 0.35, "learning_rate": 6.535685645549319e-08, "loss": 0.0694, "step": 1296 }, { "epoch": 0.35, "learning_rate": 6.5330125634857e-08, "loss": 0.0743, "step": 1297 }, { "epoch": 0.35, "learning_rate": 6.53033948142208e-08, "loss": 0.0557, "step": 1298 }, { "epoch": 0.35, "learning_rate": 6.527666399358461e-08, "loss": 0.0592, "step": 1299 }, { "epoch": 0.35, "learning_rate": 6.52499331729484e-08, "loss": 0.0476, "step": 1300 }, { "epoch": 0.35, "learning_rate": 6.522320235231222e-08, "loss": 0.0547, "step": 1301 }, { "epoch": 0.35, "learning_rate": 6.519647153167603e-08, "loss": 0.0543, "step": 1302 }, { "epoch": 0.35, "learning_rate": 6.516974071103983e-08, "loss": 0.075, "step": 1303 }, { "epoch": 0.35, "learning_rate": 6.514300989040364e-08, "loss": 0.0617, "step": 1304 }, { "epoch": 0.35, "learning_rate": 6.511627906976745e-08, "loss": 0.0584, "step": 1305 }, { "epoch": 0.35, "learning_rate": 6.508954824913125e-08, "loss": 0.061, "step": 1306 }, { "epoch": 0.35, "learning_rate": 6.506281742849506e-08, "loss": 0.0687, "step": 1307 }, { "epoch": 0.35, "learning_rate": 6.503608660785886e-08, "loss": 0.0473, "step": 1308 }, { "epoch": 0.35, "learning_rate": 6.500935578722267e-08, "loss": 0.0892, "step": 1309 }, { "epoch": 0.35, "learning_rate": 6.498262496658648e-08, "loss": 0.0552, "step": 1310 }, { "epoch": 0.35, "learning_rate": 6.495589414595028e-08, "loss": 0.0523, "step": 1311 }, { "epoch": 0.35, "learning_rate": 6.492916332531409e-08, "loss": 0.0839, "step": 1312 }, { "epoch": 0.35, "learning_rate": 6.49024325046779e-08, "loss": 0.0593, "step": 1313 }, { "epoch": 0.35, "learning_rate": 6.48757016840417e-08, "loss": 0.0596, "step": 1314 }, { "epoch": 0.35, "learning_rate": 6.484897086340551e-08, "loss": 0.0626, "step": 1315 }, { "epoch": 0.35, "learning_rate": 6.482224004276931e-08, "loss": 0.0772, "step": 1316 }, { "epoch": 0.35, "learning_rate": 6.479550922213312e-08, "loss": 0.0448, "step": 1317 }, { "epoch": 0.35, "learning_rate": 6.476877840149693e-08, "loss": 0.0901, "step": 1318 }, { "epoch": 0.35, "learning_rate": 6.474204758086073e-08, "loss": 0.0655, "step": 1319 }, { "epoch": 0.35, "learning_rate": 6.471531676022454e-08, "loss": 0.0878, "step": 1320 }, { "epoch": 0.35, "learning_rate": 6.468858593958835e-08, "loss": 0.0583, "step": 1321 }, { "epoch": 0.35, "learning_rate": 6.466185511895215e-08, "loss": 0.0787, "step": 1322 }, { "epoch": 0.35, "learning_rate": 6.463512429831596e-08, "loss": 0.0501, "step": 1323 }, { "epoch": 0.35, "learning_rate": 6.460839347767976e-08, "loss": 0.0499, "step": 1324 }, { "epoch": 0.35, "learning_rate": 6.458166265704357e-08, "loss": 0.066, "step": 1325 }, { "epoch": 0.35, "learning_rate": 6.455493183640738e-08, "loss": 0.0763, "step": 1326 }, { "epoch": 0.35, "learning_rate": 6.452820101577118e-08, "loss": 0.0495, "step": 1327 }, { "epoch": 0.35, "learning_rate": 6.450147019513499e-08, "loss": 0.0799, "step": 1328 }, { "epoch": 0.36, "learning_rate": 6.44747393744988e-08, "loss": 0.0401, "step": 1329 }, { "epoch": 0.36, "learning_rate": 6.44480085538626e-08, "loss": 0.0877, "step": 1330 }, { "epoch": 0.36, "learning_rate": 6.442127773322641e-08, "loss": 0.0711, "step": 1331 }, { "epoch": 0.36, "learning_rate": 6.439454691259021e-08, "loss": 0.0515, "step": 1332 }, { "epoch": 0.36, "learning_rate": 6.436781609195402e-08, "loss": 0.0479, "step": 1333 }, { "epoch": 0.36, "learning_rate": 6.434108527131783e-08, "loss": 0.0851, "step": 1334 }, { "epoch": 0.36, "learning_rate": 6.431435445068163e-08, "loss": 0.0484, "step": 1335 }, { "epoch": 0.36, "learning_rate": 6.428762363004544e-08, "loss": 0.0569, "step": 1336 }, { "epoch": 0.36, "learning_rate": 6.426089280940925e-08, "loss": 0.055, "step": 1337 }, { "epoch": 0.36, "learning_rate": 6.423416198877305e-08, "loss": 0.0737, "step": 1338 }, { "epoch": 0.36, "learning_rate": 6.420743116813686e-08, "loss": 0.0622, "step": 1339 }, { "epoch": 0.36, "learning_rate": 6.418070034750066e-08, "loss": 0.0698, "step": 1340 }, { "epoch": 0.36, "learning_rate": 6.415396952686447e-08, "loss": 0.0526, "step": 1341 }, { "epoch": 0.36, "learning_rate": 6.412723870622828e-08, "loss": 0.0698, "step": 1342 }, { "epoch": 0.36, "learning_rate": 6.410050788559208e-08, "loss": 0.0636, "step": 1343 }, { "epoch": 0.36, "learning_rate": 6.407377706495589e-08, "loss": 0.0854, "step": 1344 }, { "epoch": 0.36, "learning_rate": 6.40470462443197e-08, "loss": 0.0636, "step": 1345 }, { "epoch": 0.36, "learning_rate": 6.40203154236835e-08, "loss": 0.0466, "step": 1346 }, { "epoch": 0.36, "learning_rate": 6.399358460304731e-08, "loss": 0.066, "step": 1347 }, { "epoch": 0.36, "learning_rate": 6.396685378241111e-08, "loss": 0.0512, "step": 1348 }, { "epoch": 0.36, "learning_rate": 6.394012296177492e-08, "loss": 0.0673, "step": 1349 }, { "epoch": 0.36, "learning_rate": 6.391339214113873e-08, "loss": 0.0946, "step": 1350 }, { "epoch": 0.36, "learning_rate": 6.388666132050253e-08, "loss": 0.0576, "step": 1351 }, { "epoch": 0.36, "learning_rate": 6.385993049986634e-08, "loss": 0.0475, "step": 1352 }, { "epoch": 0.36, "learning_rate": 6.383319967923016e-08, "loss": 0.0642, "step": 1353 }, { "epoch": 0.36, "learning_rate": 6.380646885859395e-08, "loss": 0.059, "step": 1354 }, { "epoch": 0.36, "learning_rate": 6.377973803795777e-08, "loss": 0.0666, "step": 1355 }, { "epoch": 0.36, "learning_rate": 6.375300721732156e-08, "loss": 0.0732, "step": 1356 }, { "epoch": 0.36, "learning_rate": 6.372627639668537e-08, "loss": 0.0713, "step": 1357 }, { "epoch": 0.36, "learning_rate": 6.369954557604919e-08, "loss": 0.051, "step": 1358 }, { "epoch": 0.36, "learning_rate": 6.367281475541298e-08, "loss": 0.0672, "step": 1359 }, { "epoch": 0.36, "learning_rate": 6.36460839347768e-08, "loss": 0.0651, "step": 1360 }, { "epoch": 0.36, "learning_rate": 6.361935311414061e-08, "loss": 0.074, "step": 1361 }, { "epoch": 0.36, "learning_rate": 6.35926222935044e-08, "loss": 0.0667, "step": 1362 }, { "epoch": 0.36, "learning_rate": 6.356589147286822e-08, "loss": 0.068, "step": 1363 }, { "epoch": 0.36, "learning_rate": 6.353916065223201e-08, "loss": 0.0822, "step": 1364 }, { "epoch": 0.36, "learning_rate": 6.351242983159583e-08, "loss": 0.0725, "step": 1365 }, { "epoch": 0.37, "learning_rate": 6.348569901095964e-08, "loss": 0.0805, "step": 1366 }, { "epoch": 0.37, "learning_rate": 6.345896819032344e-08, "loss": 0.0685, "step": 1367 }, { "epoch": 0.37, "learning_rate": 6.343223736968725e-08, "loss": 0.0889, "step": 1368 }, { "epoch": 0.37, "learning_rate": 6.340550654905106e-08, "loss": 0.055, "step": 1369 }, { "epoch": 0.37, "learning_rate": 6.337877572841486e-08, "loss": 0.061, "step": 1370 }, { "epoch": 0.37, "learning_rate": 6.335204490777867e-08, "loss": 0.0587, "step": 1371 }, { "epoch": 0.37, "learning_rate": 6.332531408714247e-08, "loss": 0.0576, "step": 1372 }, { "epoch": 0.37, "learning_rate": 6.329858326650628e-08, "loss": 0.0556, "step": 1373 }, { "epoch": 0.37, "learning_rate": 6.327185244587009e-08, "loss": 0.0513, "step": 1374 }, { "epoch": 0.37, "learning_rate": 6.324512162523389e-08, "loss": 0.068, "step": 1375 }, { "epoch": 0.37, "learning_rate": 6.32183908045977e-08, "loss": 0.065, "step": 1376 }, { "epoch": 0.37, "learning_rate": 6.319165998396151e-08, "loss": 0.0588, "step": 1377 }, { "epoch": 0.37, "learning_rate": 6.316492916332531e-08, "loss": 0.0623, "step": 1378 }, { "epoch": 0.37, "learning_rate": 6.313819834268912e-08, "loss": 0.0523, "step": 1379 }, { "epoch": 0.37, "learning_rate": 6.311146752205292e-08, "loss": 0.0509, "step": 1380 }, { "epoch": 0.37, "learning_rate": 6.308473670141673e-08, "loss": 0.0776, "step": 1381 }, { "epoch": 0.37, "learning_rate": 6.305800588078054e-08, "loss": 0.0888, "step": 1382 }, { "epoch": 0.37, "learning_rate": 6.303127506014434e-08, "loss": 0.0656, "step": 1383 }, { "epoch": 0.37, "learning_rate": 6.300454423950815e-08, "loss": 0.0518, "step": 1384 }, { "epoch": 0.37, "learning_rate": 6.297781341887196e-08, "loss": 0.0452, "step": 1385 }, { "epoch": 0.37, "learning_rate": 6.295108259823576e-08, "loss": 0.087, "step": 1386 }, { "epoch": 0.37, "learning_rate": 6.292435177759957e-08, "loss": 0.0485, "step": 1387 }, { "epoch": 0.37, "learning_rate": 6.289762095696337e-08, "loss": 0.0622, "step": 1388 }, { "epoch": 0.37, "learning_rate": 6.287089013632718e-08, "loss": 0.05, "step": 1389 }, { "epoch": 0.37, "learning_rate": 6.284415931569099e-08, "loss": 0.0543, "step": 1390 }, { "epoch": 0.37, "learning_rate": 6.281742849505479e-08, "loss": 0.0696, "step": 1391 }, { "epoch": 0.37, "learning_rate": 6.27906976744186e-08, "loss": 0.0573, "step": 1392 }, { "epoch": 0.37, "learning_rate": 6.276396685378241e-08, "loss": 0.0765, "step": 1393 }, { "epoch": 0.37, "learning_rate": 6.273723603314621e-08, "loss": 0.061, "step": 1394 }, { "epoch": 0.37, "learning_rate": 6.271050521251002e-08, "loss": 0.067, "step": 1395 }, { "epoch": 0.37, "learning_rate": 6.268377439187382e-08, "loss": 0.0494, "step": 1396 }, { "epoch": 0.37, "learning_rate": 6.265704357123763e-08, "loss": 0.055, "step": 1397 }, { "epoch": 0.37, "learning_rate": 6.263031275060144e-08, "loss": 0.0628, "step": 1398 }, { "epoch": 0.37, "learning_rate": 6.260358192996524e-08, "loss": 0.0529, "step": 1399 }, { "epoch": 0.37, "learning_rate": 6.257685110932905e-08, "loss": 0.0798, "step": 1400 }, { "epoch": 0.37, "learning_rate": 6.255012028869286e-08, "loss": 0.0709, "step": 1401 }, { "epoch": 0.37, "learning_rate": 6.252338946805666e-08, "loss": 0.0557, "step": 1402 }, { "epoch": 0.37, "learning_rate": 6.249665864742047e-08, "loss": 0.0533, "step": 1403 }, { "epoch": 0.38, "learning_rate": 6.246992782678427e-08, "loss": 0.0576, "step": 1404 }, { "epoch": 0.38, "learning_rate": 6.244319700614808e-08, "loss": 0.0646, "step": 1405 }, { "epoch": 0.38, "learning_rate": 6.241646618551189e-08, "loss": 0.0682, "step": 1406 }, { "epoch": 0.38, "learning_rate": 6.238973536487569e-08, "loss": 0.0539, "step": 1407 }, { "epoch": 0.38, "learning_rate": 6.23630045442395e-08, "loss": 0.0741, "step": 1408 }, { "epoch": 0.38, "learning_rate": 6.233627372360331e-08, "loss": 0.0915, "step": 1409 }, { "epoch": 0.38, "learning_rate": 6.230954290296711e-08, "loss": 0.0551, "step": 1410 }, { "epoch": 0.38, "learning_rate": 6.228281208233092e-08, "loss": 0.0715, "step": 1411 }, { "epoch": 0.38, "learning_rate": 6.225608126169472e-08, "loss": 0.0589, "step": 1412 }, { "epoch": 0.38, "learning_rate": 6.222935044105853e-08, "loss": 0.0624, "step": 1413 }, { "epoch": 0.38, "learning_rate": 6.220261962042234e-08, "loss": 0.0563, "step": 1414 }, { "epoch": 0.38, "learning_rate": 6.217588879978614e-08, "loss": 0.0782, "step": 1415 }, { "epoch": 0.38, "learning_rate": 6.214915797914995e-08, "loss": 0.0637, "step": 1416 }, { "epoch": 0.38, "learning_rate": 6.212242715851376e-08, "loss": 0.0722, "step": 1417 }, { "epoch": 0.38, "learning_rate": 6.209569633787756e-08, "loss": 0.0535, "step": 1418 }, { "epoch": 0.38, "learning_rate": 6.206896551724137e-08, "loss": 0.057, "step": 1419 }, { "epoch": 0.38, "learning_rate": 6.204223469660517e-08, "loss": 0.0647, "step": 1420 }, { "epoch": 0.38, "learning_rate": 6.201550387596898e-08, "loss": 0.0623, "step": 1421 }, { "epoch": 0.38, "learning_rate": 6.19887730553328e-08, "loss": 0.0513, "step": 1422 }, { "epoch": 0.38, "learning_rate": 6.196204223469659e-08, "loss": 0.0896, "step": 1423 }, { "epoch": 0.38, "learning_rate": 6.19353114140604e-08, "loss": 0.0509, "step": 1424 }, { "epoch": 0.38, "learning_rate": 6.190858059342422e-08, "loss": 0.0467, "step": 1425 }, { "epoch": 0.38, "learning_rate": 6.188184977278801e-08, "loss": 0.064, "step": 1426 }, { "epoch": 0.38, "learning_rate": 6.185511895215183e-08, "loss": 0.0728, "step": 1427 }, { "epoch": 0.38, "learning_rate": 6.182838813151562e-08, "loss": 0.0751, "step": 1428 }, { "epoch": 0.38, "learning_rate": 6.180165731087943e-08, "loss": 0.0739, "step": 1429 }, { "epoch": 0.38, "learning_rate": 6.177492649024325e-08, "loss": 0.0546, "step": 1430 }, { "epoch": 0.38, "learning_rate": 6.174819566960704e-08, "loss": 0.0688, "step": 1431 }, { "epoch": 0.38, "learning_rate": 6.172146484897086e-08, "loss": 0.0563, "step": 1432 }, { "epoch": 0.38, "learning_rate": 6.169473402833467e-08, "loss": 0.0451, "step": 1433 }, { "epoch": 0.38, "learning_rate": 6.166800320769847e-08, "loss": 0.0523, "step": 1434 }, { "epoch": 0.38, "learning_rate": 6.164127238706228e-08, "loss": 0.0727, "step": 1435 }, { "epoch": 0.38, "learning_rate": 6.161454156642609e-08, "loss": 0.0638, "step": 1436 }, { "epoch": 0.38, "learning_rate": 6.158781074578989e-08, "loss": 0.0541, "step": 1437 }, { "epoch": 0.38, "learning_rate": 6.15610799251537e-08, "loss": 0.0684, "step": 1438 }, { "epoch": 0.38, "learning_rate": 6.153434910451751e-08, "loss": 0.0711, "step": 1439 }, { "epoch": 0.38, "learning_rate": 6.150761828388131e-08, "loss": 0.0455, "step": 1440 }, { "epoch": 0.39, "learning_rate": 6.148088746324512e-08, "loss": 0.065, "step": 1441 }, { "epoch": 0.39, "learning_rate": 6.145415664260893e-08, "loss": 0.0588, "step": 1442 }, { "epoch": 0.39, "learning_rate": 6.142742582197273e-08, "loss": 0.0748, "step": 1443 }, { "epoch": 0.39, "learning_rate": 6.140069500133654e-08, "loss": 0.0614, "step": 1444 }, { "epoch": 0.39, "learning_rate": 6.137396418070035e-08, "loss": 0.0721, "step": 1445 }, { "epoch": 0.39, "learning_rate": 6.134723336006415e-08, "loss": 0.0616, "step": 1446 }, { "epoch": 0.39, "learning_rate": 6.132050253942796e-08, "loss": 0.0694, "step": 1447 }, { "epoch": 0.39, "learning_rate": 6.129377171879177e-08, "loss": 0.0601, "step": 1448 }, { "epoch": 0.39, "learning_rate": 6.126704089815557e-08, "loss": 0.0615, "step": 1449 }, { "epoch": 0.39, "learning_rate": 6.124031007751938e-08, "loss": 0.0729, "step": 1450 }, { "epoch": 0.39, "learning_rate": 6.121357925688319e-08, "loss": 0.0608, "step": 1451 }, { "epoch": 0.39, "learning_rate": 6.118684843624699e-08, "loss": 0.0643, "step": 1452 }, { "epoch": 0.39, "learning_rate": 6.11601176156108e-08, "loss": 0.0649, "step": 1453 }, { "epoch": 0.39, "learning_rate": 6.113338679497461e-08, "loss": 0.0456, "step": 1454 }, { "epoch": 0.39, "learning_rate": 6.110665597433841e-08, "loss": 0.0579, "step": 1455 }, { "epoch": 0.39, "learning_rate": 6.107992515370222e-08, "loss": 0.0668, "step": 1456 }, { "epoch": 0.39, "learning_rate": 6.105319433306603e-08, "loss": 0.0556, "step": 1457 }, { "epoch": 0.39, "learning_rate": 6.102646351242983e-08, "loss": 0.0738, "step": 1458 }, { "epoch": 0.39, "learning_rate": 6.099973269179364e-08, "loss": 0.094, "step": 1459 }, { "epoch": 0.39, "learning_rate": 6.097300187115744e-08, "loss": 0.0638, "step": 1460 }, { "epoch": 0.39, "learning_rate": 6.094627105052125e-08, "loss": 0.0766, "step": 1461 }, { "epoch": 0.39, "learning_rate": 6.091954022988506e-08, "loss": 0.0708, "step": 1462 }, { "epoch": 0.39, "learning_rate": 6.089280940924886e-08, "loss": 0.0578, "step": 1463 }, { "epoch": 0.39, "learning_rate": 6.086607858861267e-08, "loss": 0.0562, "step": 1464 }, { "epoch": 0.39, "learning_rate": 6.083934776797648e-08, "loss": 0.0822, "step": 1465 }, { "epoch": 0.39, "learning_rate": 6.081261694734028e-08, "loss": 0.086, "step": 1466 }, { "epoch": 0.39, "learning_rate": 6.07858861267041e-08, "loss": 0.06, "step": 1467 }, { "epoch": 0.39, "learning_rate": 6.075915530606789e-08, "loss": 0.0704, "step": 1468 }, { "epoch": 0.39, "learning_rate": 6.07324244854317e-08, "loss": 0.0744, "step": 1469 }, { "epoch": 0.39, "learning_rate": 6.070569366479552e-08, "loss": 0.0581, "step": 1470 }, { "epoch": 0.39, "learning_rate": 6.067896284415931e-08, "loss": 0.0775, "step": 1471 }, { "epoch": 0.39, "learning_rate": 6.065223202352312e-08, "loss": 0.0793, "step": 1472 }, { "epoch": 0.39, "learning_rate": 6.062550120288694e-08, "loss": 0.0578, "step": 1473 }, { "epoch": 0.39, "learning_rate": 6.059877038225073e-08, "loss": 0.068, "step": 1474 }, { "epoch": 0.39, "learning_rate": 6.057203956161455e-08, "loss": 0.0737, "step": 1475 }, { "epoch": 0.39, "learning_rate": 6.054530874097834e-08, "loss": 0.0456, "step": 1476 }, { "epoch": 0.39, "learning_rate": 6.051857792034215e-08, "loss": 0.0671, "step": 1477 }, { "epoch": 0.4, "learning_rate": 6.049184709970597e-08, "loss": 0.0479, "step": 1478 }, { "epoch": 0.4, "learning_rate": 6.046511627906976e-08, "loss": 0.0844, "step": 1479 }, { "epoch": 0.4, "learning_rate": 6.043838545843358e-08, "loss": 0.0514, "step": 1480 }, { "epoch": 0.4, "learning_rate": 6.041165463779739e-08, "loss": 0.0582, "step": 1481 }, { "epoch": 0.4, "learning_rate": 6.038492381716119e-08, "loss": 0.0595, "step": 1482 }, { "epoch": 0.4, "learning_rate": 6.0358192996525e-08, "loss": 0.0561, "step": 1483 }, { "epoch": 0.4, "learning_rate": 6.03314621758888e-08, "loss": 0.0633, "step": 1484 }, { "epoch": 0.4, "learning_rate": 6.03047313552526e-08, "loss": 0.0575, "step": 1485 }, { "epoch": 0.4, "learning_rate": 6.027800053461642e-08, "loss": 0.1027, "step": 1486 }, { "epoch": 0.4, "learning_rate": 6.025126971398022e-08, "loss": 0.0453, "step": 1487 }, { "epoch": 0.4, "learning_rate": 6.022453889334403e-08, "loss": 0.0664, "step": 1488 }, { "epoch": 0.4, "learning_rate": 6.019780807270784e-08, "loss": 0.063, "step": 1489 }, { "epoch": 0.4, "learning_rate": 6.017107725207164e-08, "loss": 0.1012, "step": 1490 }, { "epoch": 0.4, "learning_rate": 6.014434643143545e-08, "loss": 0.0742, "step": 1491 }, { "epoch": 0.4, "learning_rate": 6.011761561079925e-08, "loss": 0.0626, "step": 1492 }, { "epoch": 0.4, "learning_rate": 6.009088479016306e-08, "loss": 0.0936, "step": 1493 }, { "epoch": 0.4, "learning_rate": 6.006415396952687e-08, "loss": 0.0599, "step": 1494 }, { "epoch": 0.4, "learning_rate": 6.003742314889067e-08, "loss": 0.0632, "step": 1495 }, { "epoch": 0.4, "learning_rate": 6.001069232825448e-08, "loss": 0.0622, "step": 1496 }, { "epoch": 0.4, "learning_rate": 5.998396150761829e-08, "loss": 0.0518, "step": 1497 }, { "epoch": 0.4, "learning_rate": 5.995723068698209e-08, "loss": 0.0587, "step": 1498 }, { "epoch": 0.4, "learning_rate": 5.99304998663459e-08, "loss": 0.076, "step": 1499 }, { "epoch": 0.4, "learning_rate": 5.99037690457097e-08, "loss": 0.0574, "step": 1500 }, { "epoch": 0.4, "learning_rate": 5.987703822507351e-08, "loss": 0.0807, "step": 1501 }, { "epoch": 0.4, "learning_rate": 5.985030740443732e-08, "loss": 0.0518, "step": 1502 }, { "epoch": 0.4, "learning_rate": 5.982357658380112e-08, "loss": 0.0679, "step": 1503 }, { "epoch": 0.4, "learning_rate": 5.979684576316493e-08, "loss": 0.0514, "step": 1504 }, { "epoch": 0.4, "learning_rate": 5.977011494252874e-08, "loss": 0.0718, "step": 1505 }, { "epoch": 0.4, "learning_rate": 5.974338412189254e-08, "loss": 0.0834, "step": 1506 }, { "epoch": 0.4, "learning_rate": 5.971665330125635e-08, "loss": 0.0681, "step": 1507 }, { "epoch": 0.4, "learning_rate": 5.968992248062015e-08, "loss": 0.0559, "step": 1508 }, { "epoch": 0.4, "learning_rate": 5.966319165998396e-08, "loss": 0.0562, "step": 1509 }, { "epoch": 0.4, "learning_rate": 5.963646083934777e-08, "loss": 0.0618, "step": 1510 }, { "epoch": 0.4, "learning_rate": 5.960973001871157e-08, "loss": 0.0515, "step": 1511 }, { "epoch": 0.4, "learning_rate": 5.958299919807538e-08, "loss": 0.0826, "step": 1512 }, { "epoch": 0.4, "learning_rate": 5.955626837743919e-08, "loss": 0.0619, "step": 1513 }, { "epoch": 0.4, "learning_rate": 5.952953755680299e-08, "loss": 0.0724, "step": 1514 }, { "epoch": 0.4, "learning_rate": 5.95028067361668e-08, "loss": 0.0724, "step": 1515 }, { "epoch": 0.41, "learning_rate": 5.94760759155306e-08, "loss": 0.0704, "step": 1516 }, { "epoch": 0.41, "learning_rate": 5.944934509489441e-08, "loss": 0.0434, "step": 1517 }, { "epoch": 0.41, "learning_rate": 5.942261427425822e-08, "loss": 0.074, "step": 1518 }, { "epoch": 0.41, "learning_rate": 5.939588345362202e-08, "loss": 0.0883, "step": 1519 }, { "epoch": 0.41, "learning_rate": 5.936915263298583e-08, "loss": 0.074, "step": 1520 }, { "epoch": 0.41, "learning_rate": 5.934242181234964e-08, "loss": 0.0488, "step": 1521 }, { "epoch": 0.41, "learning_rate": 5.931569099171344e-08, "loss": 0.0439, "step": 1522 }, { "epoch": 0.41, "learning_rate": 5.928896017107725e-08, "loss": 0.0649, "step": 1523 }, { "epoch": 0.41, "learning_rate": 5.926222935044105e-08, "loss": 0.07, "step": 1524 }, { "epoch": 0.41, "learning_rate": 5.923549852980486e-08, "loss": 0.054, "step": 1525 }, { "epoch": 0.41, "learning_rate": 5.920876770916867e-08, "loss": 0.0546, "step": 1526 }, { "epoch": 0.41, "learning_rate": 5.918203688853247e-08, "loss": 0.0664, "step": 1527 }, { "epoch": 0.41, "learning_rate": 5.915530606789628e-08, "loss": 0.066, "step": 1528 }, { "epoch": 0.41, "learning_rate": 5.9128575247260094e-08, "loss": 0.054, "step": 1529 }, { "epoch": 0.41, "learning_rate": 5.910184442662389e-08, "loss": 0.0556, "step": 1530 }, { "epoch": 0.41, "learning_rate": 5.9075113605987703e-08, "loss": 0.0768, "step": 1531 }, { "epoch": 0.41, "learning_rate": 5.90483827853515e-08, "loss": 0.0598, "step": 1532 }, { "epoch": 0.41, "learning_rate": 5.902165196471531e-08, "loss": 0.0574, "step": 1533 }, { "epoch": 0.41, "learning_rate": 5.8994921144079124e-08, "loss": 0.0645, "step": 1534 }, { "epoch": 0.41, "learning_rate": 5.896819032344292e-08, "loss": 0.0793, "step": 1535 }, { "epoch": 0.41, "learning_rate": 5.8941459502806734e-08, "loss": 0.0584, "step": 1536 }, { "epoch": 0.41, "learning_rate": 5.8914728682170545e-08, "loss": 0.0799, "step": 1537 }, { "epoch": 0.41, "learning_rate": 5.888799786153434e-08, "loss": 0.0762, "step": 1538 }, { "epoch": 0.41, "learning_rate": 5.8861267040898155e-08, "loss": 0.0624, "step": 1539 }, { "epoch": 0.41, "learning_rate": 5.883453622026195e-08, "loss": 0.0463, "step": 1540 }, { "epoch": 0.41, "learning_rate": 5.8807805399625764e-08, "loss": 0.0678, "step": 1541 }, { "epoch": 0.41, "learning_rate": 5.8781074578989575e-08, "loss": 0.0472, "step": 1542 }, { "epoch": 0.41, "learning_rate": 5.8754343758353374e-08, "loss": 0.0591, "step": 1543 }, { "epoch": 0.41, "learning_rate": 5.8727612937717185e-08, "loss": 0.0635, "step": 1544 }, { "epoch": 0.41, "learning_rate": 5.8700882117080996e-08, "loss": 0.0648, "step": 1545 }, { "epoch": 0.41, "learning_rate": 5.8674151296444794e-08, "loss": 0.0687, "step": 1546 }, { "epoch": 0.41, "learning_rate": 5.8647420475808606e-08, "loss": 0.0636, "step": 1547 }, { "epoch": 0.41, "learning_rate": 5.8620689655172404e-08, "loss": 0.0917, "step": 1548 }, { "epoch": 0.41, "learning_rate": 5.8593958834536215e-08, "loss": 0.087, "step": 1549 }, { "epoch": 0.41, "learning_rate": 5.8567228013900027e-08, "loss": 0.0782, "step": 1550 }, { "epoch": 0.41, "learning_rate": 5.8540497193263825e-08, "loss": 0.0937, "step": 1551 }, { "epoch": 0.41, "learning_rate": 5.8513766372627636e-08, "loss": 0.066, "step": 1552 }, { "epoch": 0.42, "learning_rate": 5.848703555199145e-08, "loss": 0.0406, "step": 1553 }, { "epoch": 0.42, "learning_rate": 5.8460304731355246e-08, "loss": 0.0435, "step": 1554 }, { "epoch": 0.42, "learning_rate": 5.843357391071906e-08, "loss": 0.0746, "step": 1555 }, { "epoch": 0.42, "learning_rate": 5.8406843090082855e-08, "loss": 0.0607, "step": 1556 }, { "epoch": 0.42, "learning_rate": 5.8380112269446666e-08, "loss": 0.0576, "step": 1557 }, { "epoch": 0.42, "learning_rate": 5.835338144881048e-08, "loss": 0.0681, "step": 1558 }, { "epoch": 0.42, "learning_rate": 5.8326650628174276e-08, "loss": 0.0461, "step": 1559 }, { "epoch": 0.42, "learning_rate": 5.829991980753809e-08, "loss": 0.0679, "step": 1560 }, { "epoch": 0.42, "learning_rate": 5.82731889869019e-08, "loss": 0.0671, "step": 1561 }, { "epoch": 0.42, "learning_rate": 5.8246458166265697e-08, "loss": 0.0656, "step": 1562 }, { "epoch": 0.42, "learning_rate": 5.821972734562951e-08, "loss": 0.0582, "step": 1563 }, { "epoch": 0.42, "learning_rate": 5.819299652499331e-08, "loss": 0.0412, "step": 1564 }, { "epoch": 0.42, "learning_rate": 5.816626570435712e-08, "loss": 0.0882, "step": 1565 }, { "epoch": 0.42, "learning_rate": 5.813953488372093e-08, "loss": 0.0627, "step": 1566 }, { "epoch": 0.42, "learning_rate": 5.8112804063084734e-08, "loss": 0.0619, "step": 1567 }, { "epoch": 0.42, "learning_rate": 5.808607324244854e-08, "loss": 0.0724, "step": 1568 }, { "epoch": 0.42, "learning_rate": 5.805934242181235e-08, "loss": 0.0688, "step": 1569 }, { "epoch": 0.42, "learning_rate": 5.8032611601176154e-08, "loss": 0.058, "step": 1570 }, { "epoch": 0.42, "learning_rate": 5.800588078053996e-08, "loss": 0.0483, "step": 1571 }, { "epoch": 0.42, "learning_rate": 5.7979149959903764e-08, "loss": 0.0506, "step": 1572 }, { "epoch": 0.42, "learning_rate": 5.7952419139267575e-08, "loss": 0.0716, "step": 1573 }, { "epoch": 0.42, "learning_rate": 5.792568831863138e-08, "loss": 0.0691, "step": 1574 }, { "epoch": 0.42, "learning_rate": 5.7898957497995185e-08, "loss": 0.0562, "step": 1575 }, { "epoch": 0.42, "learning_rate": 5.7872226677358996e-08, "loss": 0.0796, "step": 1576 }, { "epoch": 0.42, "learning_rate": 5.78454958567228e-08, "loss": 0.0545, "step": 1577 }, { "epoch": 0.42, "learning_rate": 5.7818765036086606e-08, "loss": 0.0483, "step": 1578 }, { "epoch": 0.42, "learning_rate": 5.779203421545042e-08, "loss": 0.0566, "step": 1579 }, { "epoch": 0.42, "learning_rate": 5.7765303394814215e-08, "loss": 0.0512, "step": 1580 }, { "epoch": 0.42, "learning_rate": 5.7738572574178026e-08, "loss": 0.0613, "step": 1581 }, { "epoch": 0.42, "learning_rate": 5.771184175354184e-08, "loss": 0.0895, "step": 1582 }, { "epoch": 0.42, "learning_rate": 5.7685110932905636e-08, "loss": 0.0518, "step": 1583 }, { "epoch": 0.42, "learning_rate": 5.765838011226945e-08, "loss": 0.0493, "step": 1584 }, { "epoch": 0.42, "learning_rate": 5.763164929163326e-08, "loss": 0.0504, "step": 1585 }, { "epoch": 0.42, "learning_rate": 5.7604918470997057e-08, "loss": 0.0605, "step": 1586 }, { "epoch": 0.42, "learning_rate": 5.757818765036087e-08, "loss": 0.0545, "step": 1587 }, { "epoch": 0.42, "learning_rate": 5.7551456829724666e-08, "loss": 0.0448, "step": 1588 }, { "epoch": 0.42, "learning_rate": 5.752472600908848e-08, "loss": 0.053, "step": 1589 }, { "epoch": 0.42, "learning_rate": 5.749799518845229e-08, "loss": 0.0758, "step": 1590 }, { "epoch": 0.43, "learning_rate": 5.747126436781609e-08, "loss": 0.078, "step": 1591 }, { "epoch": 0.43, "learning_rate": 5.74445335471799e-08, "loss": 0.0942, "step": 1592 }, { "epoch": 0.43, "learning_rate": 5.741780272654371e-08, "loss": 0.0557, "step": 1593 }, { "epoch": 0.43, "learning_rate": 5.739107190590751e-08, "loss": 0.0674, "step": 1594 }, { "epoch": 0.43, "learning_rate": 5.736434108527132e-08, "loss": 0.0859, "step": 1595 }, { "epoch": 0.43, "learning_rate": 5.733761026463512e-08, "loss": 0.0588, "step": 1596 }, { "epoch": 0.43, "learning_rate": 5.731087944399893e-08, "loss": 0.0669, "step": 1597 }, { "epoch": 0.43, "learning_rate": 5.728414862336274e-08, "loss": 0.0561, "step": 1598 }, { "epoch": 0.43, "learning_rate": 5.725741780272654e-08, "loss": 0.0704, "step": 1599 }, { "epoch": 0.43, "learning_rate": 5.723068698209035e-08, "loss": 0.0573, "step": 1600 }, { "epoch": 0.43, "learning_rate": 5.720395616145416e-08, "loss": 0.0677, "step": 1601 }, { "epoch": 0.43, "learning_rate": 5.717722534081796e-08, "loss": 0.0638, "step": 1602 }, { "epoch": 0.43, "learning_rate": 5.715049452018177e-08, "loss": 0.0571, "step": 1603 }, { "epoch": 0.43, "learning_rate": 5.712376369954557e-08, "loss": 0.0541, "step": 1604 }, { "epoch": 0.43, "learning_rate": 5.709703287890938e-08, "loss": 0.0663, "step": 1605 }, { "epoch": 0.43, "learning_rate": 5.707030205827319e-08, "loss": 0.0704, "step": 1606 }, { "epoch": 0.43, "learning_rate": 5.704357123763699e-08, "loss": 0.0669, "step": 1607 }, { "epoch": 0.43, "learning_rate": 5.70168404170008e-08, "loss": 0.0715, "step": 1608 }, { "epoch": 0.43, "learning_rate": 5.699010959636461e-08, "loss": 0.0576, "step": 1609 }, { "epoch": 0.43, "learning_rate": 5.696337877572841e-08, "loss": 0.0621, "step": 1610 }, { "epoch": 0.43, "learning_rate": 5.693664795509222e-08, "loss": 0.0638, "step": 1611 }, { "epoch": 0.43, "learning_rate": 5.690991713445602e-08, "loss": 0.0563, "step": 1612 }, { "epoch": 0.43, "learning_rate": 5.688318631381983e-08, "loss": 0.0694, "step": 1613 }, { "epoch": 0.43, "learning_rate": 5.685645549318364e-08, "loss": 0.0777, "step": 1614 }, { "epoch": 0.43, "learning_rate": 5.682972467254744e-08, "loss": 0.0654, "step": 1615 }, { "epoch": 0.43, "learning_rate": 5.680299385191125e-08, "loss": 0.0714, "step": 1616 }, { "epoch": 0.43, "learning_rate": 5.677626303127506e-08, "loss": 0.0565, "step": 1617 }, { "epoch": 0.43, "learning_rate": 5.674953221063886e-08, "loss": 0.0756, "step": 1618 }, { "epoch": 0.43, "learning_rate": 5.672280139000267e-08, "loss": 0.0574, "step": 1619 }, { "epoch": 0.43, "learning_rate": 5.669607056936647e-08, "loss": 0.0566, "step": 1620 }, { "epoch": 0.43, "learning_rate": 5.666933974873028e-08, "loss": 0.0556, "step": 1621 }, { "epoch": 0.43, "learning_rate": 5.6642608928094093e-08, "loss": 0.0566, "step": 1622 }, { "epoch": 0.43, "learning_rate": 5.661587810745789e-08, "loss": 0.0637, "step": 1623 }, { "epoch": 0.43, "learning_rate": 5.65891472868217e-08, "loss": 0.0845, "step": 1624 }, { "epoch": 0.43, "learning_rate": 5.6562416466185514e-08, "loss": 0.0675, "step": 1625 }, { "epoch": 0.43, "learning_rate": 5.653568564554931e-08, "loss": 0.0636, "step": 1626 }, { "epoch": 0.43, "learning_rate": 5.6508954824913124e-08, "loss": 0.0634, "step": 1627 }, { "epoch": 0.44, "learning_rate": 5.648222400427692e-08, "loss": 0.0579, "step": 1628 }, { "epoch": 0.44, "learning_rate": 5.645549318364073e-08, "loss": 0.0592, "step": 1629 }, { "epoch": 0.44, "learning_rate": 5.6428762363004545e-08, "loss": 0.0646, "step": 1630 }, { "epoch": 0.44, "learning_rate": 5.640203154236834e-08, "loss": 0.0791, "step": 1631 }, { "epoch": 0.44, "learning_rate": 5.6375300721732154e-08, "loss": 0.0759, "step": 1632 }, { "epoch": 0.44, "learning_rate": 5.6348569901095965e-08, "loss": 0.0531, "step": 1633 }, { "epoch": 0.44, "learning_rate": 5.6321839080459764e-08, "loss": 0.0624, "step": 1634 }, { "epoch": 0.44, "learning_rate": 5.6295108259823575e-08, "loss": 0.0714, "step": 1635 }, { "epoch": 0.44, "learning_rate": 5.626837743918737e-08, "loss": 0.071, "step": 1636 }, { "epoch": 0.44, "learning_rate": 5.6241646618551184e-08, "loss": 0.0722, "step": 1637 }, { "epoch": 0.44, "learning_rate": 5.6214915797914996e-08, "loss": 0.0542, "step": 1638 }, { "epoch": 0.44, "learning_rate": 5.6188184977278794e-08, "loss": 0.0682, "step": 1639 }, { "epoch": 0.44, "learning_rate": 5.6161454156642605e-08, "loss": 0.0554, "step": 1640 }, { "epoch": 0.44, "learning_rate": 5.6134723336006417e-08, "loss": 0.0597, "step": 1641 }, { "epoch": 0.44, "learning_rate": 5.6107992515370215e-08, "loss": 0.0552, "step": 1642 }, { "epoch": 0.44, "learning_rate": 5.6081261694734026e-08, "loss": 0.0909, "step": 1643 }, { "epoch": 0.44, "learning_rate": 5.6054530874097824e-08, "loss": 0.0769, "step": 1644 }, { "epoch": 0.44, "learning_rate": 5.6027800053461636e-08, "loss": 0.0475, "step": 1645 }, { "epoch": 0.44, "learning_rate": 5.600106923282545e-08, "loss": 0.0861, "step": 1646 }, { "epoch": 0.44, "learning_rate": 5.5974338412189245e-08, "loss": 0.0717, "step": 1647 }, { "epoch": 0.44, "learning_rate": 5.5947607591553056e-08, "loss": 0.0636, "step": 1648 }, { "epoch": 0.44, "learning_rate": 5.592087677091687e-08, "loss": 0.0607, "step": 1649 }, { "epoch": 0.44, "learning_rate": 5.5894145950280666e-08, "loss": 0.0508, "step": 1650 }, { "epoch": 0.44, "learning_rate": 5.586741512964448e-08, "loss": 0.0477, "step": 1651 }, { "epoch": 0.44, "learning_rate": 5.584068430900828e-08, "loss": 0.0656, "step": 1652 }, { "epoch": 0.44, "learning_rate": 5.5813953488372087e-08, "loss": 0.0443, "step": 1653 }, { "epoch": 0.44, "learning_rate": 5.57872226677359e-08, "loss": 0.0527, "step": 1654 }, { "epoch": 0.44, "learning_rate": 5.57604918470997e-08, "loss": 0.066, "step": 1655 }, { "epoch": 0.44, "learning_rate": 5.573376102646351e-08, "loss": 0.0733, "step": 1656 }, { "epoch": 0.44, "learning_rate": 5.570703020582732e-08, "loss": 0.0579, "step": 1657 }, { "epoch": 0.44, "learning_rate": 5.5680299385191124e-08, "loss": 0.0734, "step": 1658 }, { "epoch": 0.44, "learning_rate": 5.565356856455493e-08, "loss": 0.0618, "step": 1659 }, { "epoch": 0.44, "learning_rate": 5.562683774391873e-08, "loss": 0.065, "step": 1660 }, { "epoch": 0.44, "learning_rate": 5.5600106923282544e-08, "loss": 0.0655, "step": 1661 }, { "epoch": 0.44, "learning_rate": 5.557337610264635e-08, "loss": 0.0498, "step": 1662 }, { "epoch": 0.44, "learning_rate": 5.5546645282010154e-08, "loss": 0.0594, "step": 1663 }, { "epoch": 0.44, "learning_rate": 5.5519914461373965e-08, "loss": 0.0636, "step": 1664 }, { "epoch": 0.45, "learning_rate": 5.549318364073777e-08, "loss": 0.0529, "step": 1665 }, { "epoch": 0.45, "learning_rate": 5.5466452820101575e-08, "loss": 0.0597, "step": 1666 }, { "epoch": 0.45, "learning_rate": 5.5439721999465386e-08, "loss": 0.0928, "step": 1667 }, { "epoch": 0.45, "learning_rate": 5.5412991178829184e-08, "loss": 0.0627, "step": 1668 }, { "epoch": 0.45, "learning_rate": 5.5386260358192996e-08, "loss": 0.0784, "step": 1669 }, { "epoch": 0.45, "learning_rate": 5.535952953755681e-08, "loss": 0.0563, "step": 1670 }, { "epoch": 0.45, "learning_rate": 5.5332798716920605e-08, "loss": 0.0755, "step": 1671 }, { "epoch": 0.45, "learning_rate": 5.5306067896284416e-08, "loss": 0.0768, "step": 1672 }, { "epoch": 0.45, "learning_rate": 5.527933707564823e-08, "loss": 0.0495, "step": 1673 }, { "epoch": 0.45, "learning_rate": 5.5252606255012026e-08, "loss": 0.0607, "step": 1674 }, { "epoch": 0.45, "learning_rate": 5.522587543437584e-08, "loss": 0.0636, "step": 1675 }, { "epoch": 0.45, "learning_rate": 5.5199144613739635e-08, "loss": 0.0593, "step": 1676 }, { "epoch": 0.45, "learning_rate": 5.517241379310345e-08, "loss": 0.0606, "step": 1677 }, { "epoch": 0.45, "learning_rate": 5.514568297246726e-08, "loss": 0.0693, "step": 1678 }, { "epoch": 0.45, "learning_rate": 5.5118952151831056e-08, "loss": 0.0656, "step": 1679 }, { "epoch": 0.45, "learning_rate": 5.509222133119487e-08, "loss": 0.0566, "step": 1680 }, { "epoch": 0.45, "learning_rate": 5.506549051055868e-08, "loss": 0.0704, "step": 1681 }, { "epoch": 0.45, "learning_rate": 5.503875968992248e-08, "loss": 0.0741, "step": 1682 }, { "epoch": 0.45, "learning_rate": 5.501202886928629e-08, "loss": 0.0583, "step": 1683 }, { "epoch": 0.45, "learning_rate": 5.4985298048650086e-08, "loss": 0.0596, "step": 1684 }, { "epoch": 0.45, "learning_rate": 5.49585672280139e-08, "loss": 0.0914, "step": 1685 }, { "epoch": 0.45, "learning_rate": 5.493183640737771e-08, "loss": 0.0829, "step": 1686 }, { "epoch": 0.45, "learning_rate": 5.490510558674151e-08, "loss": 0.0672, "step": 1687 }, { "epoch": 0.45, "learning_rate": 5.487837476610532e-08, "loss": 0.078, "step": 1688 }, { "epoch": 0.45, "learning_rate": 5.485164394546913e-08, "loss": 0.0581, "step": 1689 }, { "epoch": 0.45, "learning_rate": 5.482491312483293e-08, "loss": 0.0724, "step": 1690 }, { "epoch": 0.45, "learning_rate": 5.479818230419674e-08, "loss": 0.0742, "step": 1691 }, { "epoch": 0.45, "learning_rate": 5.477145148356054e-08, "loss": 0.064, "step": 1692 }, { "epoch": 0.45, "learning_rate": 5.474472066292435e-08, "loss": 0.043, "step": 1693 }, { "epoch": 0.45, "learning_rate": 5.471798984228816e-08, "loss": 0.0419, "step": 1694 }, { "epoch": 0.45, "learning_rate": 5.469125902165196e-08, "loss": 0.0509, "step": 1695 }, { "epoch": 0.45, "learning_rate": 5.466452820101577e-08, "loss": 0.0529, "step": 1696 }, { "epoch": 0.45, "learning_rate": 5.463779738037958e-08, "loss": 0.0718, "step": 1697 }, { "epoch": 0.45, "learning_rate": 5.461106655974338e-08, "loss": 0.0485, "step": 1698 }, { "epoch": 0.45, "learning_rate": 5.458433573910719e-08, "loss": 0.0947, "step": 1699 }, { "epoch": 0.45, "learning_rate": 5.455760491847099e-08, "loss": 0.0655, "step": 1700 }, { "epoch": 0.45, "learning_rate": 5.45308740978348e-08, "loss": 0.0528, "step": 1701 }, { "epoch": 0.45, "learning_rate": 5.450414327719861e-08, "loss": 0.0545, "step": 1702 }, { "epoch": 0.46, "learning_rate": 5.447741245656241e-08, "loss": 0.07, "step": 1703 }, { "epoch": 0.46, "learning_rate": 5.445068163592622e-08, "loss": 0.0372, "step": 1704 }, { "epoch": 0.46, "learning_rate": 5.442395081529003e-08, "loss": 0.0594, "step": 1705 }, { "epoch": 0.46, "learning_rate": 5.439721999465383e-08, "loss": 0.058, "step": 1706 }, { "epoch": 0.46, "learning_rate": 5.437048917401764e-08, "loss": 0.0638, "step": 1707 }, { "epoch": 0.46, "learning_rate": 5.434375835338144e-08, "loss": 0.0767, "step": 1708 }, { "epoch": 0.46, "learning_rate": 5.431702753274525e-08, "loss": 0.0723, "step": 1709 }, { "epoch": 0.46, "learning_rate": 5.429029671210906e-08, "loss": 0.0725, "step": 1710 }, { "epoch": 0.46, "learning_rate": 5.426356589147286e-08, "loss": 0.0641, "step": 1711 }, { "epoch": 0.46, "learning_rate": 5.423683507083667e-08, "loss": 0.0839, "step": 1712 }, { "epoch": 0.46, "learning_rate": 5.4210104250200483e-08, "loss": 0.0576, "step": 1713 }, { "epoch": 0.46, "learning_rate": 5.418337342956428e-08, "loss": 0.0502, "step": 1714 }, { "epoch": 0.46, "learning_rate": 5.415664260892809e-08, "loss": 0.0499, "step": 1715 }, { "epoch": 0.46, "learning_rate": 5.412991178829189e-08, "loss": 0.0593, "step": 1716 }, { "epoch": 0.46, "learning_rate": 5.41031809676557e-08, "loss": 0.0693, "step": 1717 }, { "epoch": 0.46, "learning_rate": 5.4076450147019514e-08, "loss": 0.0862, "step": 1718 }, { "epoch": 0.46, "learning_rate": 5.404971932638331e-08, "loss": 0.0491, "step": 1719 }, { "epoch": 0.46, "learning_rate": 5.402298850574712e-08, "loss": 0.0451, "step": 1720 }, { "epoch": 0.46, "learning_rate": 5.3996257685110935e-08, "loss": 0.0717, "step": 1721 }, { "epoch": 0.46, "learning_rate": 5.396952686447473e-08, "loss": 0.0441, "step": 1722 }, { "epoch": 0.46, "learning_rate": 5.3942796043838544e-08, "loss": 0.0574, "step": 1723 }, { "epoch": 0.46, "learning_rate": 5.391606522320234e-08, "loss": 0.0793, "step": 1724 }, { "epoch": 0.46, "learning_rate": 5.3889334402566154e-08, "loss": 0.0728, "step": 1725 }, { "epoch": 0.46, "learning_rate": 5.3862603581929965e-08, "loss": 0.063, "step": 1726 }, { "epoch": 0.46, "learning_rate": 5.383587276129376e-08, "loss": 0.0575, "step": 1727 }, { "epoch": 0.46, "learning_rate": 5.3809141940657574e-08, "loss": 0.0773, "step": 1728 }, { "epoch": 0.46, "learning_rate": 5.3782411120021386e-08, "loss": 0.0752, "step": 1729 }, { "epoch": 0.46, "learning_rate": 5.3755680299385184e-08, "loss": 0.0764, "step": 1730 }, { "epoch": 0.46, "learning_rate": 5.3728949478748995e-08, "loss": 0.0721, "step": 1731 }, { "epoch": 0.46, "learning_rate": 5.37022186581128e-08, "loss": 0.0723, "step": 1732 }, { "epoch": 0.46, "learning_rate": 5.3675487837476605e-08, "loss": 0.0565, "step": 1733 }, { "epoch": 0.46, "learning_rate": 5.3648757016840416e-08, "loss": 0.0664, "step": 1734 }, { "epoch": 0.46, "learning_rate": 5.362202619620422e-08, "loss": 0.0596, "step": 1735 }, { "epoch": 0.46, "learning_rate": 5.3595295375568026e-08, "loss": 0.0786, "step": 1736 }, { "epoch": 0.46, "learning_rate": 5.356856455493184e-08, "loss": 0.0798, "step": 1737 }, { "epoch": 0.46, "learning_rate": 5.354183373429564e-08, "loss": 0.0984, "step": 1738 }, { "epoch": 0.46, "learning_rate": 5.3515102913659446e-08, "loss": 0.0493, "step": 1739 }, { "epoch": 0.47, "learning_rate": 5.348837209302325e-08, "loss": 0.0606, "step": 1740 }, { "epoch": 0.47, "learning_rate": 5.3461641272387056e-08, "loss": 0.0687, "step": 1741 }, { "epoch": 0.47, "learning_rate": 5.343491045175087e-08, "loss": 0.076, "step": 1742 }, { "epoch": 0.47, "learning_rate": 5.340817963111467e-08, "loss": 0.0566, "step": 1743 }, { "epoch": 0.47, "learning_rate": 5.3381448810478477e-08, "loss": 0.0419, "step": 1744 }, { "epoch": 0.47, "learning_rate": 5.335471798984229e-08, "loss": 0.0587, "step": 1745 }, { "epoch": 0.47, "learning_rate": 5.332798716920609e-08, "loss": 0.0557, "step": 1746 }, { "epoch": 0.47, "learning_rate": 5.33012563485699e-08, "loss": 0.0702, "step": 1747 }, { "epoch": 0.47, "learning_rate": 5.32745255279337e-08, "loss": 0.06, "step": 1748 }, { "epoch": 0.47, "learning_rate": 5.3247794707297514e-08, "loss": 0.0665, "step": 1749 }, { "epoch": 0.47, "learning_rate": 5.322106388666132e-08, "loss": 0.0686, "step": 1750 }, { "epoch": 0.47, "learning_rate": 5.319433306602512e-08, "loss": 0.0568, "step": 1751 }, { "epoch": 0.47, "learning_rate": 5.3167602245388934e-08, "loss": 0.0567, "step": 1752 }, { "epoch": 0.47, "learning_rate": 5.314087142475274e-08, "loss": 0.0655, "step": 1753 }, { "epoch": 0.47, "learning_rate": 5.3114140604116544e-08, "loss": 0.0615, "step": 1754 }, { "epoch": 0.47, "learning_rate": 5.3087409783480355e-08, "loss": 0.0616, "step": 1755 }, { "epoch": 0.47, "learning_rate": 5.3060678962844153e-08, "loss": 0.0734, "step": 1756 }, { "epoch": 0.47, "learning_rate": 5.3033948142207965e-08, "loss": 0.0693, "step": 1757 }, { "epoch": 0.47, "learning_rate": 5.3007217321571776e-08, "loss": 0.0458, "step": 1758 }, { "epoch": 0.47, "learning_rate": 5.2980486500935574e-08, "loss": 0.0749, "step": 1759 }, { "epoch": 0.47, "learning_rate": 5.2953755680299386e-08, "loss": 0.0825, "step": 1760 }, { "epoch": 0.47, "learning_rate": 5.29270248596632e-08, "loss": 0.06, "step": 1761 }, { "epoch": 0.47, "learning_rate": 5.2900294039026995e-08, "loss": 0.0778, "step": 1762 }, { "epoch": 0.47, "learning_rate": 5.2873563218390806e-08, "loss": 0.0543, "step": 1763 }, { "epoch": 0.47, "learning_rate": 5.2846832397754605e-08, "loss": 0.0425, "step": 1764 }, { "epoch": 0.47, "learning_rate": 5.2820101577118416e-08, "loss": 0.0534, "step": 1765 }, { "epoch": 0.47, "learning_rate": 5.279337075648223e-08, "loss": 0.0606, "step": 1766 }, { "epoch": 0.47, "learning_rate": 5.2766639935846025e-08, "loss": 0.078, "step": 1767 }, { "epoch": 0.47, "learning_rate": 5.273990911520984e-08, "loss": 0.0581, "step": 1768 }, { "epoch": 0.47, "learning_rate": 5.271317829457365e-08, "loss": 0.0599, "step": 1769 }, { "epoch": 0.47, "learning_rate": 5.2686447473937446e-08, "loss": 0.0465, "step": 1770 }, { "epoch": 0.47, "learning_rate": 5.265971665330126e-08, "loss": 0.0496, "step": 1771 }, { "epoch": 0.47, "learning_rate": 5.2632985832665056e-08, "loss": 0.064, "step": 1772 }, { "epoch": 0.47, "learning_rate": 5.260625501202887e-08, "loss": 0.0534, "step": 1773 }, { "epoch": 0.47, "learning_rate": 5.257952419139268e-08, "loss": 0.0778, "step": 1774 }, { "epoch": 0.47, "learning_rate": 5.2552793370756476e-08, "loss": 0.0718, "step": 1775 }, { "epoch": 0.47, "learning_rate": 5.252606255012029e-08, "loss": 0.0538, "step": 1776 }, { "epoch": 0.47, "learning_rate": 5.24993317294841e-08, "loss": 0.0551, "step": 1777 }, { "epoch": 0.48, "learning_rate": 5.24726009088479e-08, "loss": 0.0536, "step": 1778 }, { "epoch": 0.48, "learning_rate": 5.244587008821171e-08, "loss": 0.059, "step": 1779 }, { "epoch": 0.48, "learning_rate": 5.241913926757551e-08, "loss": 0.0651, "step": 1780 }, { "epoch": 0.48, "learning_rate": 5.239240844693932e-08, "loss": 0.0761, "step": 1781 }, { "epoch": 0.48, "learning_rate": 5.236567762630313e-08, "loss": 0.0734, "step": 1782 }, { "epoch": 0.48, "learning_rate": 5.233894680566693e-08, "loss": 0.0745, "step": 1783 }, { "epoch": 0.48, "learning_rate": 5.231221598503074e-08, "loss": 0.079, "step": 1784 }, { "epoch": 0.48, "learning_rate": 5.228548516439455e-08, "loss": 0.0736, "step": 1785 }, { "epoch": 0.48, "learning_rate": 5.225875434375835e-08, "loss": 0.0503, "step": 1786 }, { "epoch": 0.48, "learning_rate": 5.223202352312216e-08, "loss": 0.049, "step": 1787 }, { "epoch": 0.48, "learning_rate": 5.220529270248596e-08, "loss": 0.0528, "step": 1788 }, { "epoch": 0.48, "learning_rate": 5.217856188184977e-08, "loss": 0.0704, "step": 1789 }, { "epoch": 0.48, "learning_rate": 5.215183106121358e-08, "loss": 0.0647, "step": 1790 }, { "epoch": 0.48, "learning_rate": 5.212510024057738e-08, "loss": 0.0481, "step": 1791 }, { "epoch": 0.48, "learning_rate": 5.209836941994119e-08, "loss": 0.0762, "step": 1792 }, { "epoch": 0.48, "learning_rate": 5.2071638599305e-08, "loss": 0.0685, "step": 1793 }, { "epoch": 0.48, "learning_rate": 5.20449077786688e-08, "loss": 0.0589, "step": 1794 }, { "epoch": 0.48, "learning_rate": 5.201817695803261e-08, "loss": 0.0757, "step": 1795 }, { "epoch": 0.48, "learning_rate": 5.199144613739641e-08, "loss": 0.065, "step": 1796 }, { "epoch": 0.48, "learning_rate": 5.196471531676022e-08, "loss": 0.0651, "step": 1797 }, { "epoch": 0.48, "learning_rate": 5.193798449612403e-08, "loss": 0.065, "step": 1798 }, { "epoch": 0.48, "learning_rate": 5.191125367548783e-08, "loss": 0.056, "step": 1799 }, { "epoch": 0.48, "learning_rate": 5.188452285485164e-08, "loss": 0.0552, "step": 1800 }, { "epoch": 0.48, "learning_rate": 5.185779203421545e-08, "loss": 0.0508, "step": 1801 }, { "epoch": 0.48, "learning_rate": 5.183106121357925e-08, "loss": 0.0768, "step": 1802 }, { "epoch": 0.48, "learning_rate": 5.180433039294306e-08, "loss": 0.0641, "step": 1803 }, { "epoch": 0.48, "learning_rate": 5.177759957230686e-08, "loss": 0.056, "step": 1804 }, { "epoch": 0.48, "learning_rate": 5.175086875167067e-08, "loss": 0.0689, "step": 1805 }, { "epoch": 0.48, "learning_rate": 5.172413793103448e-08, "loss": 0.0441, "step": 1806 }, { "epoch": 0.48, "learning_rate": 5.169740711039828e-08, "loss": 0.0557, "step": 1807 }, { "epoch": 0.48, "learning_rate": 5.167067628976209e-08, "loss": 0.0632, "step": 1808 }, { "epoch": 0.48, "learning_rate": 5.1643945469125904e-08, "loss": 0.0628, "step": 1809 }, { "epoch": 0.48, "learning_rate": 5.16172146484897e-08, "loss": 0.0709, "step": 1810 }, { "epoch": 0.48, "learning_rate": 5.159048382785351e-08, "loss": 0.0468, "step": 1811 }, { "epoch": 0.48, "learning_rate": 5.156375300721731e-08, "loss": 0.0812, "step": 1812 }, { "epoch": 0.48, "learning_rate": 5.153702218658112e-08, "loss": 0.0724, "step": 1813 }, { "epoch": 0.48, "learning_rate": 5.1510291365944934e-08, "loss": 0.057, "step": 1814 }, { "epoch": 0.49, "learning_rate": 5.148356054530873e-08, "loss": 0.071, "step": 1815 }, { "epoch": 0.49, "learning_rate": 5.1456829724672544e-08, "loss": 0.0712, "step": 1816 }, { "epoch": 0.49, "learning_rate": 5.1430098904036355e-08, "loss": 0.0694, "step": 1817 }, { "epoch": 0.49, "learning_rate": 5.140336808340015e-08, "loss": 0.1107, "step": 1818 }, { "epoch": 0.49, "learning_rate": 5.1376637262763964e-08, "loss": 0.0558, "step": 1819 }, { "epoch": 0.49, "learning_rate": 5.134990644212777e-08, "loss": 0.0564, "step": 1820 }, { "epoch": 0.49, "learning_rate": 5.1323175621491574e-08, "loss": 0.0764, "step": 1821 }, { "epoch": 0.49, "learning_rate": 5.1296444800855385e-08, "loss": 0.0481, "step": 1822 }, { "epoch": 0.49, "learning_rate": 5.126971398021919e-08, "loss": 0.0585, "step": 1823 }, { "epoch": 0.49, "learning_rate": 5.1242983159582995e-08, "loss": 0.0631, "step": 1824 }, { "epoch": 0.49, "learning_rate": 5.1216252338946806e-08, "loss": 0.0779, "step": 1825 }, { "epoch": 0.49, "learning_rate": 5.118952151831061e-08, "loss": 0.0459, "step": 1826 }, { "epoch": 0.49, "learning_rate": 5.1162790697674416e-08, "loss": 0.0784, "step": 1827 }, { "epoch": 0.49, "learning_rate": 5.113605987703822e-08, "loss": 0.0557, "step": 1828 }, { "epoch": 0.49, "learning_rate": 5.110932905640203e-08, "loss": 0.0822, "step": 1829 }, { "epoch": 0.49, "learning_rate": 5.1082598235765836e-08, "loss": 0.0669, "step": 1830 }, { "epoch": 0.49, "learning_rate": 5.105586741512964e-08, "loss": 0.0611, "step": 1831 }, { "epoch": 0.49, "learning_rate": 5.102913659449345e-08, "loss": 0.0488, "step": 1832 }, { "epoch": 0.49, "learning_rate": 5.100240577385726e-08, "loss": 0.058, "step": 1833 }, { "epoch": 0.49, "learning_rate": 5.097567495322106e-08, "loss": 0.0665, "step": 1834 }, { "epoch": 0.49, "learning_rate": 5.094894413258487e-08, "loss": 0.0626, "step": 1835 }, { "epoch": 0.49, "learning_rate": 5.092221331194867e-08, "loss": 0.0531, "step": 1836 }, { "epoch": 0.49, "learning_rate": 5.089548249131248e-08, "loss": 0.0618, "step": 1837 }, { "epoch": 0.49, "learning_rate": 5.0868751670676294e-08, "loss": 0.0598, "step": 1838 }, { "epoch": 0.49, "learning_rate": 5.084202085004009e-08, "loss": 0.083, "step": 1839 }, { "epoch": 0.49, "learning_rate": 5.0815290029403904e-08, "loss": 0.0643, "step": 1840 }, { "epoch": 0.49, "learning_rate": 5.0788559208767715e-08, "loss": 0.0717, "step": 1841 }, { "epoch": 0.49, "learning_rate": 5.076182838813151e-08, "loss": 0.06, "step": 1842 }, { "epoch": 0.49, "learning_rate": 5.0735097567495324e-08, "loss": 0.0568, "step": 1843 }, { "epoch": 0.49, "learning_rate": 5.070836674685912e-08, "loss": 0.047, "step": 1844 }, { "epoch": 0.49, "learning_rate": 5.0681635926222934e-08, "loss": 0.0484, "step": 1845 }, { "epoch": 0.49, "learning_rate": 5.0654905105586745e-08, "loss": 0.0593, "step": 1846 }, { "epoch": 0.49, "learning_rate": 5.0628174284950543e-08, "loss": 0.0681, "step": 1847 }, { "epoch": 0.49, "learning_rate": 5.0601443464314355e-08, "loss": 0.0543, "step": 1848 }, { "epoch": 0.49, "learning_rate": 5.0574712643678166e-08, "loss": 0.0476, "step": 1849 }, { "epoch": 0.49, "learning_rate": 5.0547981823041964e-08, "loss": 0.0579, "step": 1850 }, { "epoch": 0.49, "learning_rate": 5.0521251002405776e-08, "loss": 0.0549, "step": 1851 }, { "epoch": 0.49, "learning_rate": 5.0494520181769574e-08, "loss": 0.0646, "step": 1852 }, { "epoch": 0.5, "learning_rate": 5.0467789361133385e-08, "loss": 0.0504, "step": 1853 }, { "epoch": 0.5, "learning_rate": 5.0441058540497196e-08, "loss": 0.0617, "step": 1854 }, { "epoch": 0.5, "learning_rate": 5.0414327719860995e-08, "loss": 0.0608, "step": 1855 }, { "epoch": 0.5, "learning_rate": 5.0387596899224806e-08, "loss": 0.0612, "step": 1856 }, { "epoch": 0.5, "learning_rate": 5.036086607858862e-08, "loss": 0.0817, "step": 1857 }, { "epoch": 0.5, "learning_rate": 5.0334135257952415e-08, "loss": 0.0793, "step": 1858 }, { "epoch": 0.5, "learning_rate": 5.030740443731623e-08, "loss": 0.0594, "step": 1859 }, { "epoch": 0.5, "learning_rate": 5.0280673616680025e-08, "loss": 0.0521, "step": 1860 }, { "epoch": 0.5, "learning_rate": 5.0253942796043836e-08, "loss": 0.0735, "step": 1861 }, { "epoch": 0.5, "learning_rate": 5.022721197540765e-08, "loss": 0.0602, "step": 1862 }, { "epoch": 0.5, "learning_rate": 5.0200481154771446e-08, "loss": 0.0717, "step": 1863 }, { "epoch": 0.5, "learning_rate": 5.017375033413526e-08, "loss": 0.0639, "step": 1864 }, { "epoch": 0.5, "learning_rate": 5.014701951349907e-08, "loss": 0.0711, "step": 1865 }, { "epoch": 0.5, "learning_rate": 5.0120288692862866e-08, "loss": 0.0693, "step": 1866 }, { "epoch": 0.5, "learning_rate": 5.009355787222668e-08, "loss": 0.0596, "step": 1867 }, { "epoch": 0.5, "learning_rate": 5.0066827051590476e-08, "loss": 0.053, "step": 1868 }, { "epoch": 0.5, "learning_rate": 5.004009623095429e-08, "loss": 0.0513, "step": 1869 }, { "epoch": 0.5, "learning_rate": 5.00133654103181e-08, "loss": 0.0525, "step": 1870 }, { "epoch": 0.5, "learning_rate": 4.9986634589681903e-08, "loss": 0.0826, "step": 1871 }, { "epoch": 0.5, "learning_rate": 4.995990376904571e-08, "loss": 0.0656, "step": 1872 }, { "epoch": 0.5, "learning_rate": 4.993317294840951e-08, "loss": 0.0615, "step": 1873 }, { "epoch": 0.5, "learning_rate": 4.990644212777332e-08, "loss": 0.0755, "step": 1874 }, { "epoch": 0.5, "learning_rate": 4.987971130713713e-08, "loss": 0.0766, "step": 1875 }, { "epoch": 0.5, "learning_rate": 4.9852980486500934e-08, "loss": 0.0701, "step": 1876 }, { "epoch": 0.5, "learning_rate": 4.982624966586474e-08, "loss": 0.0679, "step": 1877 }, { "epoch": 0.5, "learning_rate": 4.979951884522854e-08, "loss": 0.0397, "step": 1878 }, { "epoch": 0.5, "learning_rate": 4.9772788024592355e-08, "loss": 0.0808, "step": 1879 }, { "epoch": 0.5, "learning_rate": 4.974605720395616e-08, "loss": 0.0741, "step": 1880 }, { "epoch": 0.5, "learning_rate": 4.9719326383319964e-08, "loss": 0.0563, "step": 1881 }, { "epoch": 0.5, "learning_rate": 4.969259556268377e-08, "loss": 0.0645, "step": 1882 }, { "epoch": 0.5, "learning_rate": 4.966586474204758e-08, "loss": 0.0886, "step": 1883 }, { "epoch": 0.5, "learning_rate": 4.9639133921411385e-08, "loss": 0.0573, "step": 1884 }, { "epoch": 0.5, "learning_rate": 4.961240310077519e-08, "loss": 0.0715, "step": 1885 }, { "epoch": 0.5, "learning_rate": 4.9585672280138994e-08, "loss": 0.072, "step": 1886 }, { "epoch": 0.5, "learning_rate": 4.9558941459502806e-08, "loss": 0.0662, "step": 1887 }, { "epoch": 0.5, "learning_rate": 4.953221063886661e-08, "loss": 0.0777, "step": 1888 }, { "epoch": 0.5, "learning_rate": 4.9505479818230415e-08, "loss": 0.081, "step": 1889 }, { "epoch": 0.51, "learning_rate": 4.947874899759422e-08, "loss": 0.0588, "step": 1890 }, { "epoch": 0.51, "learning_rate": 4.945201817695803e-08, "loss": 0.0392, "step": 1891 }, { "epoch": 0.51, "learning_rate": 4.9425287356321836e-08, "loss": 0.0541, "step": 1892 }, { "epoch": 0.51, "learning_rate": 4.939855653568564e-08, "loss": 0.0761, "step": 1893 }, { "epoch": 0.51, "learning_rate": 4.9371825715049445e-08, "loss": 0.0488, "step": 1894 }, { "epoch": 0.51, "learning_rate": 4.934509489441326e-08, "loss": 0.052, "step": 1895 }, { "epoch": 0.51, "learning_rate": 4.931836407377706e-08, "loss": 0.057, "step": 1896 }, { "epoch": 0.51, "learning_rate": 4.9291633253140866e-08, "loss": 0.06, "step": 1897 }, { "epoch": 0.51, "learning_rate": 4.926490243250467e-08, "loss": 0.0412, "step": 1898 }, { "epoch": 0.51, "learning_rate": 4.923817161186848e-08, "loss": 0.0648, "step": 1899 }, { "epoch": 0.51, "learning_rate": 4.921144079123229e-08, "loss": 0.067, "step": 1900 }, { "epoch": 0.51, "learning_rate": 4.918470997059609e-08, "loss": 0.0577, "step": 1901 }, { "epoch": 0.51, "learning_rate": 4.9157979149959897e-08, "loss": 0.0499, "step": 1902 }, { "epoch": 0.51, "learning_rate": 4.913124832932371e-08, "loss": 0.096, "step": 1903 }, { "epoch": 0.51, "learning_rate": 4.910451750868751e-08, "loss": 0.0488, "step": 1904 }, { "epoch": 0.51, "learning_rate": 4.907778668805132e-08, "loss": 0.0561, "step": 1905 }, { "epoch": 0.51, "learning_rate": 4.905105586741512e-08, "loss": 0.0503, "step": 1906 }, { "epoch": 0.51, "learning_rate": 4.9024325046778934e-08, "loss": 0.0735, "step": 1907 }, { "epoch": 0.51, "learning_rate": 4.899759422614274e-08, "loss": 0.0675, "step": 1908 }, { "epoch": 0.51, "learning_rate": 4.897086340550654e-08, "loss": 0.0611, "step": 1909 }, { "epoch": 0.51, "learning_rate": 4.8944132584870354e-08, "loss": 0.0757, "step": 1910 }, { "epoch": 0.51, "learning_rate": 4.891740176423416e-08, "loss": 0.0644, "step": 1911 }, { "epoch": 0.51, "learning_rate": 4.8890670943597964e-08, "loss": 0.0588, "step": 1912 }, { "epoch": 0.51, "learning_rate": 4.8863940122961775e-08, "loss": 0.0696, "step": 1913 }, { "epoch": 0.51, "learning_rate": 4.883720930232558e-08, "loss": 0.0633, "step": 1914 }, { "epoch": 0.51, "learning_rate": 4.8810478481689385e-08, "loss": 0.0576, "step": 1915 }, { "epoch": 0.51, "learning_rate": 4.8783747661053196e-08, "loss": 0.05, "step": 1916 }, { "epoch": 0.51, "learning_rate": 4.8757016840417e-08, "loss": 0.0701, "step": 1917 }, { "epoch": 0.51, "learning_rate": 4.8730286019780806e-08, "loss": 0.0705, "step": 1918 }, { "epoch": 0.51, "learning_rate": 4.870355519914462e-08, "loss": 0.0721, "step": 1919 }, { "epoch": 0.51, "learning_rate": 4.867682437850842e-08, "loss": 0.089, "step": 1920 }, { "epoch": 0.51, "learning_rate": 4.8650093557872226e-08, "loss": 0.0524, "step": 1921 }, { "epoch": 0.51, "learning_rate": 4.862336273723603e-08, "loss": 0.0695, "step": 1922 }, { "epoch": 0.51, "learning_rate": 4.859663191659984e-08, "loss": 0.0748, "step": 1923 }, { "epoch": 0.51, "learning_rate": 4.856990109596365e-08, "loss": 0.0552, "step": 1924 }, { "epoch": 0.51, "learning_rate": 4.854317027532745e-08, "loss": 0.0453, "step": 1925 }, { "epoch": 0.51, "learning_rate": 4.8516439454691257e-08, "loss": 0.0621, "step": 1926 }, { "epoch": 0.52, "learning_rate": 4.848970863405507e-08, "loss": 0.0534, "step": 1927 }, { "epoch": 0.52, "learning_rate": 4.846297781341887e-08, "loss": 0.0586, "step": 1928 }, { "epoch": 0.52, "learning_rate": 4.843624699278268e-08, "loss": 0.0957, "step": 1929 }, { "epoch": 0.52, "learning_rate": 4.840951617214648e-08, "loss": 0.0511, "step": 1930 }, { "epoch": 0.52, "learning_rate": 4.8382785351510294e-08, "loss": 0.044, "step": 1931 }, { "epoch": 0.52, "learning_rate": 4.83560545308741e-08, "loss": 0.0673, "step": 1932 }, { "epoch": 0.52, "learning_rate": 4.83293237102379e-08, "loss": 0.081, "step": 1933 }, { "epoch": 0.52, "learning_rate": 4.830259288960171e-08, "loss": 0.1062, "step": 1934 }, { "epoch": 0.52, "learning_rate": 4.827586206896552e-08, "loss": 0.0537, "step": 1935 }, { "epoch": 0.52, "learning_rate": 4.8249131248329324e-08, "loss": 0.0557, "step": 1936 }, { "epoch": 0.52, "learning_rate": 4.822240042769313e-08, "loss": 0.061, "step": 1937 }, { "epoch": 0.52, "learning_rate": 4.8195669607056933e-08, "loss": 0.0581, "step": 1938 }, { "epoch": 0.52, "learning_rate": 4.8168938786420745e-08, "loss": 0.0534, "step": 1939 }, { "epoch": 0.52, "learning_rate": 4.814220796578455e-08, "loss": 0.0837, "step": 1940 }, { "epoch": 0.52, "learning_rate": 4.8115477145148354e-08, "loss": 0.0453, "step": 1941 }, { "epoch": 0.52, "learning_rate": 4.808874632451216e-08, "loss": 0.0813, "step": 1942 }, { "epoch": 0.52, "learning_rate": 4.806201550387597e-08, "loss": 0.0511, "step": 1943 }, { "epoch": 0.52, "learning_rate": 4.8035284683239775e-08, "loss": 0.0623, "step": 1944 }, { "epoch": 0.52, "learning_rate": 4.800855386260358e-08, "loss": 0.0782, "step": 1945 }, { "epoch": 0.52, "learning_rate": 4.7981823041967385e-08, "loss": 0.0548, "step": 1946 }, { "epoch": 0.52, "learning_rate": 4.7955092221331196e-08, "loss": 0.0628, "step": 1947 }, { "epoch": 0.52, "learning_rate": 4.7928361400695e-08, "loss": 0.0556, "step": 1948 }, { "epoch": 0.52, "learning_rate": 4.7901630580058805e-08, "loss": 0.0492, "step": 1949 }, { "epoch": 0.52, "learning_rate": 4.787489975942261e-08, "loss": 0.068, "step": 1950 }, { "epoch": 0.52, "learning_rate": 4.784816893878642e-08, "loss": 0.0572, "step": 1951 }, { "epoch": 0.52, "learning_rate": 4.7821438118150226e-08, "loss": 0.0643, "step": 1952 }, { "epoch": 0.52, "learning_rate": 4.779470729751403e-08, "loss": 0.0381, "step": 1953 }, { "epoch": 0.52, "learning_rate": 4.7767976476877836e-08, "loss": 0.0636, "step": 1954 }, { "epoch": 0.52, "learning_rate": 4.774124565624165e-08, "loss": 0.0558, "step": 1955 }, { "epoch": 0.52, "learning_rate": 4.771451483560545e-08, "loss": 0.0642, "step": 1956 }, { "epoch": 0.52, "learning_rate": 4.7687784014969256e-08, "loss": 0.0543, "step": 1957 }, { "epoch": 0.52, "learning_rate": 4.766105319433306e-08, "loss": 0.0494, "step": 1958 }, { "epoch": 0.52, "learning_rate": 4.763432237369687e-08, "loss": 0.0594, "step": 1959 }, { "epoch": 0.52, "learning_rate": 4.760759155306068e-08, "loss": 0.0764, "step": 1960 }, { "epoch": 0.52, "learning_rate": 4.758086073242448e-08, "loss": 0.0745, "step": 1961 }, { "epoch": 0.52, "learning_rate": 4.755412991178829e-08, "loss": 0.0746, "step": 1962 }, { "epoch": 0.52, "learning_rate": 4.75273990911521e-08, "loss": 0.0659, "step": 1963 }, { "epoch": 0.52, "learning_rate": 4.75006682705159e-08, "loss": 0.091, "step": 1964 }, { "epoch": 0.53, "learning_rate": 4.747393744987971e-08, "loss": 0.0407, "step": 1965 }, { "epoch": 0.53, "learning_rate": 4.744720662924351e-08, "loss": 0.0619, "step": 1966 }, { "epoch": 0.53, "learning_rate": 4.7420475808607324e-08, "loss": 0.0483, "step": 1967 }, { "epoch": 0.53, "learning_rate": 4.739374498797113e-08, "loss": 0.0532, "step": 1968 }, { "epoch": 0.53, "learning_rate": 4.736701416733493e-08, "loss": 0.0569, "step": 1969 }, { "epoch": 0.53, "learning_rate": 4.734028334669874e-08, "loss": 0.0584, "step": 1970 }, { "epoch": 0.53, "learning_rate": 4.731355252606255e-08, "loss": 0.0564, "step": 1971 }, { "epoch": 0.53, "learning_rate": 4.7286821705426354e-08, "loss": 0.0419, "step": 1972 }, { "epoch": 0.53, "learning_rate": 4.726009088479016e-08, "loss": 0.0706, "step": 1973 }, { "epoch": 0.53, "learning_rate": 4.7233360064153964e-08, "loss": 0.0694, "step": 1974 }, { "epoch": 0.53, "learning_rate": 4.7206629243517775e-08, "loss": 0.0699, "step": 1975 }, { "epoch": 0.53, "learning_rate": 4.717989842288158e-08, "loss": 0.0731, "step": 1976 }, { "epoch": 0.53, "learning_rate": 4.7153167602245384e-08, "loss": 0.0387, "step": 1977 }, { "epoch": 0.53, "learning_rate": 4.712643678160919e-08, "loss": 0.0779, "step": 1978 }, { "epoch": 0.53, "learning_rate": 4.7099705960973e-08, "loss": 0.0628, "step": 1979 }, { "epoch": 0.53, "learning_rate": 4.7072975140336805e-08, "loss": 0.0619, "step": 1980 }, { "epoch": 0.53, "learning_rate": 4.704624431970061e-08, "loss": 0.0524, "step": 1981 }, { "epoch": 0.53, "learning_rate": 4.7019513499064415e-08, "loss": 0.0694, "step": 1982 }, { "epoch": 0.53, "learning_rate": 4.6992782678428226e-08, "loss": 0.0773, "step": 1983 }, { "epoch": 0.53, "learning_rate": 4.696605185779203e-08, "loss": 0.065, "step": 1984 }, { "epoch": 0.53, "learning_rate": 4.6939321037155835e-08, "loss": 0.0517, "step": 1985 }, { "epoch": 0.53, "learning_rate": 4.691259021651964e-08, "loss": 0.0774, "step": 1986 }, { "epoch": 0.53, "learning_rate": 4.688585939588345e-08, "loss": 0.0713, "step": 1987 }, { "epoch": 0.53, "learning_rate": 4.6859128575247256e-08, "loss": 0.0453, "step": 1988 }, { "epoch": 0.53, "learning_rate": 4.683239775461106e-08, "loss": 0.0751, "step": 1989 }, { "epoch": 0.53, "learning_rate": 4.6805666933974866e-08, "loss": 0.057, "step": 1990 }, { "epoch": 0.53, "learning_rate": 4.677893611333868e-08, "loss": 0.0782, "step": 1991 }, { "epoch": 0.53, "learning_rate": 4.675220529270248e-08, "loss": 0.07, "step": 1992 }, { "epoch": 0.53, "learning_rate": 4.6725474472066287e-08, "loss": 0.0703, "step": 1993 }, { "epoch": 0.53, "learning_rate": 4.66987436514301e-08, "loss": 0.0653, "step": 1994 }, { "epoch": 0.53, "learning_rate": 4.66720128307939e-08, "loss": 0.0655, "step": 1995 }, { "epoch": 0.53, "learning_rate": 4.664528201015771e-08, "loss": 0.0654, "step": 1996 }, { "epoch": 0.53, "learning_rate": 4.661855118952152e-08, "loss": 0.0542, "step": 1997 }, { "epoch": 0.53, "learning_rate": 4.6591820368885324e-08, "loss": 0.0546, "step": 1998 }, { "epoch": 0.53, "learning_rate": 4.656508954824913e-08, "loss": 0.0774, "step": 1999 }, { "epoch": 0.53, "learning_rate": 4.653835872761294e-08, "loss": 0.0489, "step": 2000 }, { "epoch": 0.53, "learning_rate": 4.6511627906976744e-08, "loss": 0.072, "step": 2001 }, { "epoch": 0.54, "learning_rate": 4.648489708634055e-08, "loss": 0.0617, "step": 2002 }, { "epoch": 0.54, "learning_rate": 4.6458166265704354e-08, "loss": 0.0458, "step": 2003 }, { "epoch": 0.54, "learning_rate": 4.6431435445068165e-08, "loss": 0.0548, "step": 2004 }, { "epoch": 0.54, "learning_rate": 4.640470462443197e-08, "loss": 0.0473, "step": 2005 }, { "epoch": 0.54, "learning_rate": 4.6377973803795775e-08, "loss": 0.0717, "step": 2006 }, { "epoch": 0.54, "learning_rate": 4.6351242983159586e-08, "loss": 0.0632, "step": 2007 }, { "epoch": 0.54, "learning_rate": 4.632451216252339e-08, "loss": 0.0678, "step": 2008 }, { "epoch": 0.54, "learning_rate": 4.6297781341887196e-08, "loss": 0.0819, "step": 2009 }, { "epoch": 0.54, "learning_rate": 4.6271050521251e-08, "loss": 0.0516, "step": 2010 }, { "epoch": 0.54, "learning_rate": 4.624431970061481e-08, "loss": 0.0708, "step": 2011 }, { "epoch": 0.54, "learning_rate": 4.6217588879978616e-08, "loss": 0.0704, "step": 2012 }, { "epoch": 0.54, "learning_rate": 4.619085805934242e-08, "loss": 0.0618, "step": 2013 }, { "epoch": 0.54, "learning_rate": 4.6164127238706226e-08, "loss": 0.0663, "step": 2014 }, { "epoch": 0.54, "learning_rate": 4.613739641807004e-08, "loss": 0.0597, "step": 2015 }, { "epoch": 0.54, "learning_rate": 4.611066559743384e-08, "loss": 0.0327, "step": 2016 }, { "epoch": 0.54, "learning_rate": 4.6083934776797647e-08, "loss": 0.0493, "step": 2017 }, { "epoch": 0.54, "learning_rate": 4.605720395616145e-08, "loss": 0.0613, "step": 2018 }, { "epoch": 0.54, "learning_rate": 4.603047313552526e-08, "loss": 0.0736, "step": 2019 }, { "epoch": 0.54, "learning_rate": 4.600374231488907e-08, "loss": 0.0689, "step": 2020 }, { "epoch": 0.54, "learning_rate": 4.597701149425287e-08, "loss": 0.0467, "step": 2021 }, { "epoch": 0.54, "learning_rate": 4.595028067361668e-08, "loss": 0.0563, "step": 2022 }, { "epoch": 0.54, "learning_rate": 4.592354985298049e-08, "loss": 0.0607, "step": 2023 }, { "epoch": 0.54, "learning_rate": 4.589681903234429e-08, "loss": 0.0753, "step": 2024 }, { "epoch": 0.54, "learning_rate": 4.58700882117081e-08, "loss": 0.0755, "step": 2025 }, { "epoch": 0.54, "learning_rate": 4.58433573910719e-08, "loss": 0.0605, "step": 2026 }, { "epoch": 0.54, "learning_rate": 4.5816626570435714e-08, "loss": 0.065, "step": 2027 }, { "epoch": 0.54, "learning_rate": 4.578989574979952e-08, "loss": 0.0577, "step": 2028 }, { "epoch": 0.54, "learning_rate": 4.5763164929163323e-08, "loss": 0.0494, "step": 2029 }, { "epoch": 0.54, "learning_rate": 4.573643410852713e-08, "loss": 0.0547, "step": 2030 }, { "epoch": 0.54, "learning_rate": 4.570970328789094e-08, "loss": 0.049, "step": 2031 }, { "epoch": 0.54, "learning_rate": 4.5682972467254744e-08, "loss": 0.054, "step": 2032 }, { "epoch": 0.54, "learning_rate": 4.565624164661855e-08, "loss": 0.0533, "step": 2033 }, { "epoch": 0.54, "learning_rate": 4.5629510825982354e-08, "loss": 0.0886, "step": 2034 }, { "epoch": 0.54, "learning_rate": 4.5602780005346165e-08, "loss": 0.0761, "step": 2035 }, { "epoch": 0.54, "learning_rate": 4.557604918470997e-08, "loss": 0.062, "step": 2036 }, { "epoch": 0.54, "learning_rate": 4.5549318364073775e-08, "loss": 0.0602, "step": 2037 }, { "epoch": 0.54, "learning_rate": 4.552258754343758e-08, "loss": 0.0528, "step": 2038 }, { "epoch": 0.54, "learning_rate": 4.549585672280139e-08, "loss": 0.062, "step": 2039 }, { "epoch": 0.55, "learning_rate": 4.5469125902165195e-08, "loss": 0.0471, "step": 2040 }, { "epoch": 0.55, "learning_rate": 4.5442395081529e-08, "loss": 0.0818, "step": 2041 }, { "epoch": 0.55, "learning_rate": 4.5415664260892805e-08, "loss": 0.0682, "step": 2042 }, { "epoch": 0.55, "learning_rate": 4.5388933440256616e-08, "loss": 0.0662, "step": 2043 }, { "epoch": 0.55, "learning_rate": 4.536220261962042e-08, "loss": 0.0612, "step": 2044 }, { "epoch": 0.55, "learning_rate": 4.5335471798984226e-08, "loss": 0.0502, "step": 2045 }, { "epoch": 0.55, "learning_rate": 4.530874097834803e-08, "loss": 0.0558, "step": 2046 }, { "epoch": 0.55, "learning_rate": 4.528201015771184e-08, "loss": 0.0582, "step": 2047 }, { "epoch": 0.55, "learning_rate": 4.5255279337075646e-08, "loss": 0.0508, "step": 2048 }, { "epoch": 0.55, "learning_rate": 4.522854851643945e-08, "loss": 0.0555, "step": 2049 }, { "epoch": 0.55, "learning_rate": 4.5201817695803256e-08, "loss": 0.0489, "step": 2050 }, { "epoch": 0.55, "learning_rate": 4.517508687516707e-08, "loss": 0.0798, "step": 2051 }, { "epoch": 0.55, "learning_rate": 4.514835605453087e-08, "loss": 0.0439, "step": 2052 }, { "epoch": 0.55, "learning_rate": 4.512162523389468e-08, "loss": 0.0541, "step": 2053 }, { "epoch": 0.55, "learning_rate": 4.509489441325848e-08, "loss": 0.0659, "step": 2054 }, { "epoch": 0.55, "learning_rate": 4.506816359262229e-08, "loss": 0.0585, "step": 2055 }, { "epoch": 0.55, "learning_rate": 4.50414327719861e-08, "loss": 0.0864, "step": 2056 }, { "epoch": 0.55, "learning_rate": 4.50147019513499e-08, "loss": 0.051, "step": 2057 }, { "epoch": 0.55, "learning_rate": 4.498797113071371e-08, "loss": 0.0515, "step": 2058 }, { "epoch": 0.55, "learning_rate": 4.496124031007752e-08, "loss": 0.0548, "step": 2059 }, { "epoch": 0.55, "learning_rate": 4.493450948944132e-08, "loss": 0.0622, "step": 2060 }, { "epoch": 0.55, "learning_rate": 4.490777866880513e-08, "loss": 0.0665, "step": 2061 }, { "epoch": 0.55, "learning_rate": 4.488104784816893e-08, "loss": 0.0581, "step": 2062 }, { "epoch": 0.55, "learning_rate": 4.4854317027532744e-08, "loss": 0.0634, "step": 2063 }, { "epoch": 0.55, "learning_rate": 4.482758620689655e-08, "loss": 0.0458, "step": 2064 }, { "epoch": 0.55, "learning_rate": 4.4800855386260354e-08, "loss": 0.074, "step": 2065 }, { "epoch": 0.55, "learning_rate": 4.477412456562416e-08, "loss": 0.0684, "step": 2066 }, { "epoch": 0.55, "learning_rate": 4.474739374498797e-08, "loss": 0.06, "step": 2067 }, { "epoch": 0.55, "learning_rate": 4.4720662924351774e-08, "loss": 0.0498, "step": 2068 }, { "epoch": 0.55, "learning_rate": 4.469393210371558e-08, "loss": 0.0753, "step": 2069 }, { "epoch": 0.55, "learning_rate": 4.4667201283079384e-08, "loss": 0.0722, "step": 2070 }, { "epoch": 0.55, "learning_rate": 4.4640470462443195e-08, "loss": 0.079, "step": 2071 }, { "epoch": 0.55, "learning_rate": 4.4613739641807e-08, "loss": 0.0539, "step": 2072 }, { "epoch": 0.55, "learning_rate": 4.4587008821170805e-08, "loss": 0.0828, "step": 2073 }, { "epoch": 0.55, "learning_rate": 4.456027800053461e-08, "loss": 0.0428, "step": 2074 }, { "epoch": 0.55, "learning_rate": 4.453354717989842e-08, "loss": 0.0735, "step": 2075 }, { "epoch": 0.55, "learning_rate": 4.4506816359262225e-08, "loss": 0.0687, "step": 2076 }, { "epoch": 0.56, "learning_rate": 4.448008553862603e-08, "loss": 0.0816, "step": 2077 }, { "epoch": 0.56, "learning_rate": 4.4453354717989835e-08, "loss": 0.0479, "step": 2078 }, { "epoch": 0.56, "learning_rate": 4.4426623897353646e-08, "loss": 0.0886, "step": 2079 }, { "epoch": 0.56, "learning_rate": 4.439989307671745e-08, "loss": 0.0489, "step": 2080 }, { "epoch": 0.56, "learning_rate": 4.4373162256081256e-08, "loss": 0.077, "step": 2081 }, { "epoch": 0.56, "learning_rate": 4.434643143544507e-08, "loss": 0.0642, "step": 2082 }, { "epoch": 0.56, "learning_rate": 4.431970061480887e-08, "loss": 0.0991, "step": 2083 }, { "epoch": 0.56, "learning_rate": 4.4292969794172677e-08, "loss": 0.0662, "step": 2084 }, { "epoch": 0.56, "learning_rate": 4.426623897353649e-08, "loss": 0.0594, "step": 2085 }, { "epoch": 0.56, "learning_rate": 4.423950815290029e-08, "loss": 0.0506, "step": 2086 }, { "epoch": 0.56, "learning_rate": 4.42127773322641e-08, "loss": 0.057, "step": 2087 }, { "epoch": 0.56, "learning_rate": 4.418604651162791e-08, "loss": 0.055, "step": 2088 }, { "epoch": 0.56, "learning_rate": 4.4159315690991714e-08, "loss": 0.0738, "step": 2089 }, { "epoch": 0.56, "learning_rate": 4.413258487035552e-08, "loss": 0.0682, "step": 2090 }, { "epoch": 0.56, "learning_rate": 4.410585404971933e-08, "loss": 0.0547, "step": 2091 }, { "epoch": 0.56, "learning_rate": 4.4079123229083134e-08, "loss": 0.0702, "step": 2092 }, { "epoch": 0.56, "learning_rate": 4.405239240844694e-08, "loss": 0.0799, "step": 2093 }, { "epoch": 0.56, "learning_rate": 4.4025661587810744e-08, "loss": 0.0625, "step": 2094 }, { "epoch": 0.56, "learning_rate": 4.3998930767174555e-08, "loss": 0.0465, "step": 2095 }, { "epoch": 0.56, "learning_rate": 4.397219994653836e-08, "loss": 0.0653, "step": 2096 }, { "epoch": 0.56, "learning_rate": 4.3945469125902165e-08, "loss": 0.0753, "step": 2097 }, { "epoch": 0.56, "learning_rate": 4.391873830526597e-08, "loss": 0.0573, "step": 2098 }, { "epoch": 0.56, "learning_rate": 4.389200748462978e-08, "loss": 0.0601, "step": 2099 }, { "epoch": 0.56, "learning_rate": 4.3865276663993586e-08, "loss": 0.0619, "step": 2100 }, { "epoch": 0.56, "learning_rate": 4.383854584335739e-08, "loss": 0.0745, "step": 2101 }, { "epoch": 0.56, "learning_rate": 4.3811815022721195e-08, "loss": 0.0566, "step": 2102 }, { "epoch": 0.56, "learning_rate": 4.3785084202085006e-08, "loss": 0.0628, "step": 2103 }, { "epoch": 0.56, "learning_rate": 4.375835338144881e-08, "loss": 0.0672, "step": 2104 }, { "epoch": 0.56, "learning_rate": 4.3731622560812616e-08, "loss": 0.0683, "step": 2105 }, { "epoch": 0.56, "learning_rate": 4.370489174017642e-08, "loss": 0.0709, "step": 2106 }, { "epoch": 0.56, "learning_rate": 4.367816091954023e-08, "loss": 0.0816, "step": 2107 }, { "epoch": 0.56, "learning_rate": 4.3651430098904037e-08, "loss": 0.0633, "step": 2108 }, { "epoch": 0.56, "learning_rate": 4.362469927826784e-08, "loss": 0.0455, "step": 2109 }, { "epoch": 0.56, "learning_rate": 4.3597968457631646e-08, "loss": 0.0586, "step": 2110 }, { "epoch": 0.56, "learning_rate": 4.357123763699546e-08, "loss": 0.0798, "step": 2111 }, { "epoch": 0.56, "learning_rate": 4.354450681635926e-08, "loss": 0.0497, "step": 2112 }, { "epoch": 0.56, "learning_rate": 4.351777599572307e-08, "loss": 0.0501, "step": 2113 }, { "epoch": 0.57, "learning_rate": 4.349104517508687e-08, "loss": 0.0752, "step": 2114 }, { "epoch": 0.57, "learning_rate": 4.346431435445068e-08, "loss": 0.078, "step": 2115 }, { "epoch": 0.57, "learning_rate": 4.343758353381449e-08, "loss": 0.0552, "step": 2116 }, { "epoch": 0.57, "learning_rate": 4.341085271317829e-08, "loss": 0.06, "step": 2117 }, { "epoch": 0.57, "learning_rate": 4.33841218925421e-08, "loss": 0.0519, "step": 2118 }, { "epoch": 0.57, "learning_rate": 4.335739107190591e-08, "loss": 0.0531, "step": 2119 }, { "epoch": 0.57, "learning_rate": 4.3330660251269713e-08, "loss": 0.0748, "step": 2120 }, { "epoch": 0.57, "learning_rate": 4.330392943063352e-08, "loss": 0.0512, "step": 2121 }, { "epoch": 0.57, "learning_rate": 4.327719860999732e-08, "loss": 0.061, "step": 2122 }, { "epoch": 0.57, "learning_rate": 4.3250467789361134e-08, "loss": 0.0586, "step": 2123 }, { "epoch": 0.57, "learning_rate": 4.322373696872494e-08, "loss": 0.0442, "step": 2124 }, { "epoch": 0.57, "learning_rate": 4.3197006148088744e-08, "loss": 0.0607, "step": 2125 }, { "epoch": 0.57, "learning_rate": 4.317027532745255e-08, "loss": 0.0613, "step": 2126 }, { "epoch": 0.57, "learning_rate": 4.314354450681636e-08, "loss": 0.0615, "step": 2127 }, { "epoch": 0.57, "learning_rate": 4.3116813686180165e-08, "loss": 0.0499, "step": 2128 }, { "epoch": 0.57, "learning_rate": 4.309008286554397e-08, "loss": 0.0653, "step": 2129 }, { "epoch": 0.57, "learning_rate": 4.3063352044907774e-08, "loss": 0.0642, "step": 2130 }, { "epoch": 0.57, "learning_rate": 4.3036621224271585e-08, "loss": 0.0704, "step": 2131 }, { "epoch": 0.57, "learning_rate": 4.300989040363539e-08, "loss": 0.0661, "step": 2132 }, { "epoch": 0.57, "learning_rate": 4.2983159582999195e-08, "loss": 0.0662, "step": 2133 }, { "epoch": 0.57, "learning_rate": 4.2956428762363e-08, "loss": 0.0799, "step": 2134 }, { "epoch": 0.57, "learning_rate": 4.292969794172681e-08, "loss": 0.0547, "step": 2135 }, { "epoch": 0.57, "learning_rate": 4.2902967121090616e-08, "loss": 0.0825, "step": 2136 }, { "epoch": 0.57, "learning_rate": 4.287623630045442e-08, "loss": 0.0606, "step": 2137 }, { "epoch": 0.57, "learning_rate": 4.2849505479818225e-08, "loss": 0.0853, "step": 2138 }, { "epoch": 0.57, "learning_rate": 4.2822774659182037e-08, "loss": 0.0678, "step": 2139 }, { "epoch": 0.57, "learning_rate": 4.279604383854584e-08, "loss": 0.0619, "step": 2140 }, { "epoch": 0.57, "learning_rate": 4.2769313017909646e-08, "loss": 0.0558, "step": 2141 }, { "epoch": 0.57, "learning_rate": 4.274258219727345e-08, "loss": 0.072, "step": 2142 }, { "epoch": 0.57, "learning_rate": 4.271585137663726e-08, "loss": 0.0575, "step": 2143 }, { "epoch": 0.57, "learning_rate": 4.268912055600107e-08, "loss": 0.0627, "step": 2144 }, { "epoch": 0.57, "learning_rate": 4.266238973536487e-08, "loss": 0.0568, "step": 2145 }, { "epoch": 0.57, "learning_rate": 4.2635658914728676e-08, "loss": 0.0734, "step": 2146 }, { "epoch": 0.57, "learning_rate": 4.260892809409249e-08, "loss": 0.0641, "step": 2147 }, { "epoch": 0.57, "learning_rate": 4.258219727345629e-08, "loss": 0.0659, "step": 2148 }, { "epoch": 0.57, "learning_rate": 4.25554664528201e-08, "loss": 0.0528, "step": 2149 }, { "epoch": 0.57, "learning_rate": 4.25287356321839e-08, "loss": 0.071, "step": 2150 }, { "epoch": 0.57, "learning_rate": 4.250200481154771e-08, "loss": 0.0489, "step": 2151 }, { "epoch": 0.58, "learning_rate": 4.247527399091152e-08, "loss": 0.0796, "step": 2152 }, { "epoch": 0.58, "learning_rate": 4.244854317027532e-08, "loss": 0.0713, "step": 2153 }, { "epoch": 0.58, "learning_rate": 4.242181234963913e-08, "loss": 0.0472, "step": 2154 }, { "epoch": 0.58, "learning_rate": 4.239508152900294e-08, "loss": 0.0502, "step": 2155 }, { "epoch": 0.58, "learning_rate": 4.2368350708366744e-08, "loss": 0.0834, "step": 2156 }, { "epoch": 0.58, "learning_rate": 4.234161988773055e-08, "loss": 0.0684, "step": 2157 }, { "epoch": 0.58, "learning_rate": 4.231488906709435e-08, "loss": 0.0549, "step": 2158 }, { "epoch": 0.58, "learning_rate": 4.2288158246458164e-08, "loss": 0.0645, "step": 2159 }, { "epoch": 0.58, "learning_rate": 4.226142742582197e-08, "loss": 0.0769, "step": 2160 }, { "epoch": 0.58, "learning_rate": 4.2234696605185774e-08, "loss": 0.0674, "step": 2161 }, { "epoch": 0.58, "learning_rate": 4.220796578454958e-08, "loss": 0.084, "step": 2162 }, { "epoch": 0.58, "learning_rate": 4.218123496391339e-08, "loss": 0.0603, "step": 2163 }, { "epoch": 0.58, "learning_rate": 4.2154504143277195e-08, "loss": 0.0632, "step": 2164 }, { "epoch": 0.58, "learning_rate": 4.2127773322641e-08, "loss": 0.0826, "step": 2165 }, { "epoch": 0.58, "learning_rate": 4.210104250200481e-08, "loss": 0.0557, "step": 2166 }, { "epoch": 0.58, "learning_rate": 4.2074311681368616e-08, "loss": 0.0556, "step": 2167 }, { "epoch": 0.58, "learning_rate": 4.204758086073242e-08, "loss": 0.0473, "step": 2168 }, { "epoch": 0.58, "learning_rate": 4.202085004009623e-08, "loss": 0.066, "step": 2169 }, { "epoch": 0.58, "learning_rate": 4.1994119219460036e-08, "loss": 0.05, "step": 2170 }, { "epoch": 0.58, "learning_rate": 4.196738839882384e-08, "loss": 0.0794, "step": 2171 }, { "epoch": 0.58, "learning_rate": 4.194065757818765e-08, "loss": 0.059, "step": 2172 }, { "epoch": 0.58, "learning_rate": 4.191392675755146e-08, "loss": 0.056, "step": 2173 }, { "epoch": 0.58, "learning_rate": 4.188719593691526e-08, "loss": 0.0797, "step": 2174 }, { "epoch": 0.58, "learning_rate": 4.186046511627907e-08, "loss": 0.0895, "step": 2175 }, { "epoch": 0.58, "learning_rate": 4.183373429564288e-08, "loss": 0.0744, "step": 2176 }, { "epoch": 0.58, "learning_rate": 4.180700347500668e-08, "loss": 0.0678, "step": 2177 }, { "epoch": 0.58, "learning_rate": 4.178027265437049e-08, "loss": 0.0599, "step": 2178 }, { "epoch": 0.58, "learning_rate": 4.17535418337343e-08, "loss": 0.0595, "step": 2179 }, { "epoch": 0.58, "learning_rate": 4.1726811013098104e-08, "loss": 0.0632, "step": 2180 }, { "epoch": 0.58, "learning_rate": 4.170008019246191e-08, "loss": 0.0603, "step": 2181 }, { "epoch": 0.58, "learning_rate": 4.167334937182571e-08, "loss": 0.0585, "step": 2182 }, { "epoch": 0.58, "learning_rate": 4.1646618551189524e-08, "loss": 0.0605, "step": 2183 }, { "epoch": 0.58, "learning_rate": 4.161988773055333e-08, "loss": 0.0734, "step": 2184 }, { "epoch": 0.58, "learning_rate": 4.1593156909917134e-08, "loss": 0.0634, "step": 2185 }, { "epoch": 0.58, "learning_rate": 4.156642608928094e-08, "loss": 0.0708, "step": 2186 }, { "epoch": 0.58, "learning_rate": 4.153969526864475e-08, "loss": 0.0648, "step": 2187 }, { "epoch": 0.58, "learning_rate": 4.1512964448008555e-08, "loss": 0.0808, "step": 2188 }, { "epoch": 0.59, "learning_rate": 4.148623362737236e-08, "loss": 0.0674, "step": 2189 }, { "epoch": 0.59, "learning_rate": 4.1459502806736164e-08, "loss": 0.0593, "step": 2190 }, { "epoch": 0.59, "learning_rate": 4.1432771986099976e-08, "loss": 0.0693, "step": 2191 }, { "epoch": 0.59, "learning_rate": 4.140604116546378e-08, "loss": 0.0544, "step": 2192 }, { "epoch": 0.59, "learning_rate": 4.1379310344827585e-08, "loss": 0.0652, "step": 2193 }, { "epoch": 0.59, "learning_rate": 4.135257952419139e-08, "loss": 0.0422, "step": 2194 }, { "epoch": 0.59, "learning_rate": 4.13258487035552e-08, "loss": 0.0815, "step": 2195 }, { "epoch": 0.59, "learning_rate": 4.1299117882919006e-08, "loss": 0.0646, "step": 2196 }, { "epoch": 0.59, "learning_rate": 4.127238706228281e-08, "loss": 0.0714, "step": 2197 }, { "epoch": 0.59, "learning_rate": 4.1245656241646615e-08, "loss": 0.0546, "step": 2198 }, { "epoch": 0.59, "learning_rate": 4.1218925421010427e-08, "loss": 0.0935, "step": 2199 }, { "epoch": 0.59, "learning_rate": 4.119219460037423e-08, "loss": 0.0686, "step": 2200 }, { "epoch": 0.59, "learning_rate": 4.1165463779738036e-08, "loss": 0.0729, "step": 2201 }, { "epoch": 0.59, "learning_rate": 4.113873295910184e-08, "loss": 0.0544, "step": 2202 }, { "epoch": 0.59, "learning_rate": 4.111200213846565e-08, "loss": 0.0609, "step": 2203 }, { "epoch": 0.59, "learning_rate": 4.108527131782946e-08, "loss": 0.0856, "step": 2204 }, { "epoch": 0.59, "learning_rate": 4.105854049719326e-08, "loss": 0.0627, "step": 2205 }, { "epoch": 0.59, "learning_rate": 4.1031809676557066e-08, "loss": 0.0451, "step": 2206 }, { "epoch": 0.59, "learning_rate": 4.100507885592088e-08, "loss": 0.0549, "step": 2207 }, { "epoch": 0.59, "learning_rate": 4.097834803528468e-08, "loss": 0.0647, "step": 2208 }, { "epoch": 0.59, "learning_rate": 4.095161721464849e-08, "loss": 0.0512, "step": 2209 }, { "epoch": 0.59, "learning_rate": 4.092488639401229e-08, "loss": 0.0425, "step": 2210 }, { "epoch": 0.59, "learning_rate": 4.0898155573376103e-08, "loss": 0.0525, "step": 2211 }, { "epoch": 0.59, "learning_rate": 4.087142475273991e-08, "loss": 0.0839, "step": 2212 }, { "epoch": 0.59, "learning_rate": 4.084469393210371e-08, "loss": 0.0781, "step": 2213 }, { "epoch": 0.59, "learning_rate": 4.081796311146752e-08, "loss": 0.067, "step": 2214 }, { "epoch": 0.59, "learning_rate": 4.079123229083133e-08, "loss": 0.0685, "step": 2215 }, { "epoch": 0.59, "learning_rate": 4.0764501470195134e-08, "loss": 0.0619, "step": 2216 }, { "epoch": 0.59, "learning_rate": 4.073777064955894e-08, "loss": 0.0469, "step": 2217 }, { "epoch": 0.59, "learning_rate": 4.071103982892274e-08, "loss": 0.0509, "step": 2218 }, { "epoch": 0.59, "learning_rate": 4.0684309008286555e-08, "loss": 0.0619, "step": 2219 }, { "epoch": 0.59, "learning_rate": 4.065757818765036e-08, "loss": 0.0822, "step": 2220 }, { "epoch": 0.59, "learning_rate": 4.0630847367014164e-08, "loss": 0.0588, "step": 2221 }, { "epoch": 0.59, "learning_rate": 4.060411654637797e-08, "loss": 0.0706, "step": 2222 }, { "epoch": 0.59, "learning_rate": 4.057738572574178e-08, "loss": 0.0723, "step": 2223 }, { "epoch": 0.59, "learning_rate": 4.0550654905105585e-08, "loss": 0.0596, "step": 2224 }, { "epoch": 0.59, "learning_rate": 4.052392408446939e-08, "loss": 0.0855, "step": 2225 }, { "epoch": 0.59, "learning_rate": 4.0497193263833194e-08, "loss": 0.0689, "step": 2226 }, { "epoch": 0.6, "learning_rate": 4.0470462443197006e-08, "loss": 0.0698, "step": 2227 }, { "epoch": 0.6, "learning_rate": 4.044373162256081e-08, "loss": 0.0458, "step": 2228 }, { "epoch": 0.6, "learning_rate": 4.0417000801924615e-08, "loss": 0.0613, "step": 2229 }, { "epoch": 0.6, "learning_rate": 4.039026998128842e-08, "loss": 0.0748, "step": 2230 }, { "epoch": 0.6, "learning_rate": 4.036353916065223e-08, "loss": 0.068, "step": 2231 }, { "epoch": 0.6, "learning_rate": 4.0336808340016036e-08, "loss": 0.0436, "step": 2232 }, { "epoch": 0.6, "learning_rate": 4.031007751937984e-08, "loss": 0.0586, "step": 2233 }, { "epoch": 0.6, "learning_rate": 4.0283346698743645e-08, "loss": 0.0636, "step": 2234 }, { "epoch": 0.6, "learning_rate": 4.025661587810746e-08, "loss": 0.0556, "step": 2235 }, { "epoch": 0.6, "learning_rate": 4.022988505747126e-08, "loss": 0.0585, "step": 2236 }, { "epoch": 0.6, "learning_rate": 4.0203154236835066e-08, "loss": 0.0506, "step": 2237 }, { "epoch": 0.6, "learning_rate": 4.017642341619887e-08, "loss": 0.0616, "step": 2238 }, { "epoch": 0.6, "learning_rate": 4.014969259556268e-08, "loss": 0.0599, "step": 2239 }, { "epoch": 0.6, "learning_rate": 4.012296177492649e-08, "loss": 0.0942, "step": 2240 }, { "epoch": 0.6, "learning_rate": 4.009623095429029e-08, "loss": 0.0612, "step": 2241 }, { "epoch": 0.6, "learning_rate": 4.0069500133654097e-08, "loss": 0.0552, "step": 2242 }, { "epoch": 0.6, "learning_rate": 4.004276931301791e-08, "loss": 0.0564, "step": 2243 }, { "epoch": 0.6, "learning_rate": 4.001603849238171e-08, "loss": 0.061, "step": 2244 }, { "epoch": 0.6, "learning_rate": 3.998930767174552e-08, "loss": 0.0836, "step": 2245 }, { "epoch": 0.6, "learning_rate": 3.996257685110932e-08, "loss": 0.0583, "step": 2246 }, { "epoch": 0.6, "learning_rate": 3.9935846030473134e-08, "loss": 0.0688, "step": 2247 }, { "epoch": 0.6, "learning_rate": 3.990911520983694e-08, "loss": 0.0532, "step": 2248 }, { "epoch": 0.6, "learning_rate": 3.988238438920074e-08, "loss": 0.1414, "step": 2249 }, { "epoch": 0.6, "learning_rate": 3.9855653568564554e-08, "loss": 0.0641, "step": 2250 }, { "epoch": 0.6, "learning_rate": 3.982892274792836e-08, "loss": 0.0693, "step": 2251 }, { "epoch": 0.6, "learning_rate": 3.9802191927292164e-08, "loss": 0.0479, "step": 2252 }, { "epoch": 0.6, "learning_rate": 3.9775461106655975e-08, "loss": 0.0567, "step": 2253 }, { "epoch": 0.6, "learning_rate": 3.974873028601978e-08, "loss": 0.0786, "step": 2254 }, { "epoch": 0.6, "learning_rate": 3.9721999465383585e-08, "loss": 0.0709, "step": 2255 }, { "epoch": 0.6, "learning_rate": 3.9695268644747396e-08, "loss": 0.0621, "step": 2256 }, { "epoch": 0.6, "learning_rate": 3.96685378241112e-08, "loss": 0.0699, "step": 2257 }, { "epoch": 0.6, "learning_rate": 3.9641807003475006e-08, "loss": 0.0603, "step": 2258 }, { "epoch": 0.6, "learning_rate": 3.961507618283882e-08, "loss": 0.0605, "step": 2259 }, { "epoch": 0.6, "learning_rate": 3.958834536220262e-08, "loss": 0.0728, "step": 2260 }, { "epoch": 0.6, "learning_rate": 3.9561614541566426e-08, "loss": 0.0573, "step": 2261 }, { "epoch": 0.6, "learning_rate": 3.953488372093023e-08, "loss": 0.0822, "step": 2262 }, { "epoch": 0.6, "learning_rate": 3.950815290029404e-08, "loss": 0.0582, "step": 2263 }, { "epoch": 0.61, "learning_rate": 3.948142207965785e-08, "loss": 0.0584, "step": 2264 }, { "epoch": 0.61, "learning_rate": 3.945469125902165e-08, "loss": 0.0583, "step": 2265 }, { "epoch": 0.61, "learning_rate": 3.9427960438385457e-08, "loss": 0.074, "step": 2266 }, { "epoch": 0.61, "learning_rate": 3.940122961774927e-08, "loss": 0.0584, "step": 2267 }, { "epoch": 0.61, "learning_rate": 3.937449879711307e-08, "loss": 0.0659, "step": 2268 }, { "epoch": 0.61, "learning_rate": 3.934776797647688e-08, "loss": 0.0504, "step": 2269 }, { "epoch": 0.61, "learning_rate": 3.932103715584068e-08, "loss": 0.0699, "step": 2270 }, { "epoch": 0.61, "learning_rate": 3.9294306335204494e-08, "loss": 0.0508, "step": 2271 }, { "epoch": 0.61, "learning_rate": 3.92675755145683e-08, "loss": 0.0667, "step": 2272 }, { "epoch": 0.61, "learning_rate": 3.92408446939321e-08, "loss": 0.0469, "step": 2273 }, { "epoch": 0.61, "learning_rate": 3.921411387329591e-08, "loss": 0.0671, "step": 2274 }, { "epoch": 0.61, "learning_rate": 3.918738305265972e-08, "loss": 0.0495, "step": 2275 }, { "epoch": 0.61, "learning_rate": 3.9160652232023524e-08, "loss": 0.0868, "step": 2276 }, { "epoch": 0.61, "learning_rate": 3.913392141138733e-08, "loss": 0.0747, "step": 2277 }, { "epoch": 0.61, "learning_rate": 3.9107190590751133e-08, "loss": 0.0746, "step": 2278 }, { "epoch": 0.61, "learning_rate": 3.9080459770114945e-08, "loss": 0.0495, "step": 2279 }, { "epoch": 0.61, "learning_rate": 3.905372894947875e-08, "loss": 0.0566, "step": 2280 }, { "epoch": 0.61, "learning_rate": 3.9026998128842554e-08, "loss": 0.0636, "step": 2281 }, { "epoch": 0.61, "learning_rate": 3.900026730820636e-08, "loss": 0.0619, "step": 2282 }, { "epoch": 0.61, "learning_rate": 3.897353648757017e-08, "loss": 0.0715, "step": 2283 }, { "epoch": 0.61, "learning_rate": 3.8946805666933975e-08, "loss": 0.0589, "step": 2284 }, { "epoch": 0.61, "learning_rate": 3.892007484629778e-08, "loss": 0.0663, "step": 2285 }, { "epoch": 0.61, "learning_rate": 3.8893344025661585e-08, "loss": 0.0621, "step": 2286 }, { "epoch": 0.61, "learning_rate": 3.8866613205025396e-08, "loss": 0.0582, "step": 2287 }, { "epoch": 0.61, "learning_rate": 3.88398823843892e-08, "loss": 0.0526, "step": 2288 }, { "epoch": 0.61, "learning_rate": 3.8813151563753005e-08, "loss": 0.0609, "step": 2289 }, { "epoch": 0.61, "learning_rate": 3.878642074311681e-08, "loss": 0.0614, "step": 2290 }, { "epoch": 0.61, "learning_rate": 3.875968992248062e-08, "loss": 0.0845, "step": 2291 }, { "epoch": 0.61, "learning_rate": 3.8732959101844426e-08, "loss": 0.0587, "step": 2292 }, { "epoch": 0.61, "learning_rate": 3.870622828120823e-08, "loss": 0.06, "step": 2293 }, { "epoch": 0.61, "learning_rate": 3.8679497460572036e-08, "loss": 0.0624, "step": 2294 }, { "epoch": 0.61, "learning_rate": 3.865276663993585e-08, "loss": 0.0613, "step": 2295 }, { "epoch": 0.61, "learning_rate": 3.862603581929965e-08, "loss": 0.0453, "step": 2296 }, { "epoch": 0.61, "learning_rate": 3.8599304998663456e-08, "loss": 0.0503, "step": 2297 }, { "epoch": 0.61, "learning_rate": 3.857257417802726e-08, "loss": 0.0765, "step": 2298 }, { "epoch": 0.61, "learning_rate": 3.854584335739107e-08, "loss": 0.0646, "step": 2299 }, { "epoch": 0.61, "learning_rate": 3.851911253675488e-08, "loss": 0.0503, "step": 2300 }, { "epoch": 0.61, "learning_rate": 3.849238171611868e-08, "loss": 0.061, "step": 2301 }, { "epoch": 0.62, "learning_rate": 3.846565089548249e-08, "loss": 0.0431, "step": 2302 }, { "epoch": 0.62, "learning_rate": 3.84389200748463e-08, "loss": 0.0597, "step": 2303 }, { "epoch": 0.62, "learning_rate": 3.84121892542101e-08, "loss": 0.0543, "step": 2304 }, { "epoch": 0.62, "learning_rate": 3.838545843357391e-08, "loss": 0.0552, "step": 2305 }, { "epoch": 0.62, "learning_rate": 3.835872761293771e-08, "loss": 0.0526, "step": 2306 }, { "epoch": 0.62, "learning_rate": 3.8331996792301524e-08, "loss": 0.0588, "step": 2307 }, { "epoch": 0.62, "learning_rate": 3.830526597166533e-08, "loss": 0.0524, "step": 2308 }, { "epoch": 0.62, "learning_rate": 3.827853515102913e-08, "loss": 0.0646, "step": 2309 }, { "epoch": 0.62, "learning_rate": 3.825180433039294e-08, "loss": 0.0586, "step": 2310 }, { "epoch": 0.62, "learning_rate": 3.822507350975675e-08, "loss": 0.0574, "step": 2311 }, { "epoch": 0.62, "learning_rate": 3.8198342689120554e-08, "loss": 0.0515, "step": 2312 }, { "epoch": 0.62, "learning_rate": 3.817161186848436e-08, "loss": 0.0648, "step": 2313 }, { "epoch": 0.62, "learning_rate": 3.8144881047848163e-08, "loss": 0.0689, "step": 2314 }, { "epoch": 0.62, "learning_rate": 3.8118150227211975e-08, "loss": 0.0597, "step": 2315 }, { "epoch": 0.62, "learning_rate": 3.809141940657578e-08, "loss": 0.0499, "step": 2316 }, { "epoch": 0.62, "learning_rate": 3.8064688585939584e-08, "loss": 0.066, "step": 2317 }, { "epoch": 0.62, "learning_rate": 3.803795776530339e-08, "loss": 0.0672, "step": 2318 }, { "epoch": 0.62, "learning_rate": 3.80112269446672e-08, "loss": 0.047, "step": 2319 }, { "epoch": 0.62, "learning_rate": 3.7984496124031005e-08, "loss": 0.0748, "step": 2320 }, { "epoch": 0.62, "learning_rate": 3.795776530339481e-08, "loss": 0.0521, "step": 2321 }, { "epoch": 0.62, "learning_rate": 3.7931034482758615e-08, "loss": 0.0515, "step": 2322 }, { "epoch": 0.62, "learning_rate": 3.7904303662122426e-08, "loss": 0.0886, "step": 2323 }, { "epoch": 0.62, "learning_rate": 3.787757284148623e-08, "loss": 0.0657, "step": 2324 }, { "epoch": 0.62, "learning_rate": 3.7850842020850035e-08, "loss": 0.0467, "step": 2325 }, { "epoch": 0.62, "learning_rate": 3.782411120021384e-08, "loss": 0.0521, "step": 2326 }, { "epoch": 0.62, "learning_rate": 3.779738037957765e-08, "loss": 0.0505, "step": 2327 }, { "epoch": 0.62, "learning_rate": 3.7770649558941456e-08, "loss": 0.0583, "step": 2328 }, { "epoch": 0.62, "learning_rate": 3.774391873830526e-08, "loss": 0.0676, "step": 2329 }, { "epoch": 0.62, "learning_rate": 3.7717187917669066e-08, "loss": 0.0688, "step": 2330 }, { "epoch": 0.62, "learning_rate": 3.769045709703288e-08, "loss": 0.067, "step": 2331 }, { "epoch": 0.62, "learning_rate": 3.766372627639668e-08, "loss": 0.0521, "step": 2332 }, { "epoch": 0.62, "learning_rate": 3.7636995455760487e-08, "loss": 0.0493, "step": 2333 }, { "epoch": 0.62, "learning_rate": 3.76102646351243e-08, "loss": 0.0845, "step": 2334 }, { "epoch": 0.62, "learning_rate": 3.75835338144881e-08, "loss": 0.0472, "step": 2335 }, { "epoch": 0.62, "learning_rate": 3.755680299385191e-08, "loss": 0.0628, "step": 2336 }, { "epoch": 0.62, "learning_rate": 3.753007217321572e-08, "loss": 0.0632, "step": 2337 }, { "epoch": 0.62, "learning_rate": 3.7503341352579524e-08, "loss": 0.0817, "step": 2338 }, { "epoch": 0.63, "learning_rate": 3.747661053194333e-08, "loss": 0.0474, "step": 2339 }, { "epoch": 0.63, "learning_rate": 3.744987971130713e-08, "loss": 0.052, "step": 2340 }, { "epoch": 0.63, "learning_rate": 3.7423148890670944e-08, "loss": 0.0614, "step": 2341 }, { "epoch": 0.63, "learning_rate": 3.739641807003475e-08, "loss": 0.0646, "step": 2342 }, { "epoch": 0.63, "learning_rate": 3.7369687249398554e-08, "loss": 0.0719, "step": 2343 }, { "epoch": 0.63, "learning_rate": 3.7342956428762365e-08, "loss": 0.0598, "step": 2344 }, { "epoch": 0.63, "learning_rate": 3.731622560812617e-08, "loss": 0.0609, "step": 2345 }, { "epoch": 0.63, "learning_rate": 3.7289494787489975e-08, "loss": 0.0796, "step": 2346 }, { "epoch": 0.63, "learning_rate": 3.7262763966853786e-08, "loss": 0.0686, "step": 2347 }, { "epoch": 0.63, "learning_rate": 3.723603314621759e-08, "loss": 0.0576, "step": 2348 }, { "epoch": 0.63, "learning_rate": 3.7209302325581396e-08, "loss": 0.07, "step": 2349 }, { "epoch": 0.63, "learning_rate": 3.71825715049452e-08, "loss": 0.0854, "step": 2350 }, { "epoch": 0.63, "learning_rate": 3.715584068430901e-08, "loss": 0.0856, "step": 2351 }, { "epoch": 0.63, "learning_rate": 3.7129109863672816e-08, "loss": 0.0326, "step": 2352 }, { "epoch": 0.63, "learning_rate": 3.710237904303662e-08, "loss": 0.055, "step": 2353 }, { "epoch": 0.63, "learning_rate": 3.7075648222400426e-08, "loss": 0.0769, "step": 2354 }, { "epoch": 0.63, "learning_rate": 3.704891740176424e-08, "loss": 0.0658, "step": 2355 }, { "epoch": 0.63, "learning_rate": 3.702218658112804e-08, "loss": 0.0789, "step": 2356 }, { "epoch": 0.63, "learning_rate": 3.6995455760491847e-08, "loss": 0.0593, "step": 2357 }, { "epoch": 0.63, "learning_rate": 3.696872493985565e-08, "loss": 0.0654, "step": 2358 }, { "epoch": 0.63, "learning_rate": 3.694199411921946e-08, "loss": 0.0608, "step": 2359 }, { "epoch": 0.63, "learning_rate": 3.691526329858327e-08, "loss": 0.0707, "step": 2360 }, { "epoch": 0.63, "learning_rate": 3.688853247794707e-08, "loss": 0.0649, "step": 2361 }, { "epoch": 0.63, "learning_rate": 3.686180165731088e-08, "loss": 0.0661, "step": 2362 }, { "epoch": 0.63, "learning_rate": 3.683507083667469e-08, "loss": 0.0697, "step": 2363 }, { "epoch": 0.63, "learning_rate": 3.680834001603849e-08, "loss": 0.0673, "step": 2364 }, { "epoch": 0.63, "learning_rate": 3.67816091954023e-08, "loss": 0.0619, "step": 2365 }, { "epoch": 0.63, "learning_rate": 3.67548783747661e-08, "loss": 0.0647, "step": 2366 }, { "epoch": 0.63, "learning_rate": 3.6728147554129914e-08, "loss": 0.0622, "step": 2367 }, { "epoch": 0.63, "learning_rate": 3.670141673349372e-08, "loss": 0.0612, "step": 2368 }, { "epoch": 0.63, "learning_rate": 3.6674685912857523e-08, "loss": 0.0955, "step": 2369 }, { "epoch": 0.63, "learning_rate": 3.664795509222133e-08, "loss": 0.0534, "step": 2370 }, { "epoch": 0.63, "learning_rate": 3.662122427158514e-08, "loss": 0.0658, "step": 2371 }, { "epoch": 0.63, "learning_rate": 3.6594493450948944e-08, "loss": 0.059, "step": 2372 }, { "epoch": 0.63, "learning_rate": 3.656776263031275e-08, "loss": 0.046, "step": 2373 }, { "epoch": 0.63, "learning_rate": 3.6541031809676554e-08, "loss": 0.0611, "step": 2374 }, { "epoch": 0.63, "learning_rate": 3.6514300989040365e-08, "loss": 0.0551, "step": 2375 }, { "epoch": 0.64, "learning_rate": 3.648757016840417e-08, "loss": 0.0562, "step": 2376 }, { "epoch": 0.64, "learning_rate": 3.6460839347767975e-08, "loss": 0.0529, "step": 2377 }, { "epoch": 0.64, "learning_rate": 3.643410852713178e-08, "loss": 0.0724, "step": 2378 }, { "epoch": 0.64, "learning_rate": 3.640737770649559e-08, "loss": 0.0634, "step": 2379 }, { "epoch": 0.64, "learning_rate": 3.6380646885859395e-08, "loss": 0.0509, "step": 2380 }, { "epoch": 0.64, "learning_rate": 3.63539160652232e-08, "loss": 0.054, "step": 2381 }, { "epoch": 0.64, "learning_rate": 3.6327185244587005e-08, "loss": 0.0672, "step": 2382 }, { "epoch": 0.64, "learning_rate": 3.6300454423950816e-08, "loss": 0.0579, "step": 2383 }, { "epoch": 0.64, "learning_rate": 3.627372360331462e-08, "loss": 0.0544, "step": 2384 }, { "epoch": 0.64, "learning_rate": 3.6246992782678426e-08, "loss": 0.0617, "step": 2385 }, { "epoch": 0.64, "learning_rate": 3.622026196204223e-08, "loss": 0.0679, "step": 2386 }, { "epoch": 0.64, "learning_rate": 3.619353114140604e-08, "loss": 0.0582, "step": 2387 }, { "epoch": 0.64, "learning_rate": 3.6166800320769846e-08, "loss": 0.0716, "step": 2388 }, { "epoch": 0.64, "learning_rate": 3.614006950013365e-08, "loss": 0.0652, "step": 2389 }, { "epoch": 0.64, "learning_rate": 3.6113338679497456e-08, "loss": 0.0578, "step": 2390 }, { "epoch": 0.64, "learning_rate": 3.608660785886127e-08, "loss": 0.0576, "step": 2391 }, { "epoch": 0.64, "learning_rate": 3.605987703822507e-08, "loss": 0.0682, "step": 2392 }, { "epoch": 0.64, "learning_rate": 3.603314621758888e-08, "loss": 0.0647, "step": 2393 }, { "epoch": 0.64, "learning_rate": 3.600641539695268e-08, "loss": 0.0712, "step": 2394 }, { "epoch": 0.64, "learning_rate": 3.597968457631649e-08, "loss": 0.0632, "step": 2395 }, { "epoch": 0.64, "learning_rate": 3.59529537556803e-08, "loss": 0.0853, "step": 2396 }, { "epoch": 0.64, "learning_rate": 3.59262229350441e-08, "loss": 0.0615, "step": 2397 }, { "epoch": 0.64, "learning_rate": 3.589949211440791e-08, "loss": 0.0894, "step": 2398 }, { "epoch": 0.64, "learning_rate": 3.587276129377172e-08, "loss": 0.0662, "step": 2399 }, { "epoch": 0.64, "learning_rate": 3.584603047313552e-08, "loss": 0.0686, "step": 2400 }, { "epoch": 0.64, "learning_rate": 3.581929965249933e-08, "loss": 0.0608, "step": 2401 }, { "epoch": 0.64, "learning_rate": 3.579256883186313e-08, "loss": 0.0551, "step": 2402 }, { "epoch": 0.64, "learning_rate": 3.5765838011226944e-08, "loss": 0.0612, "step": 2403 }, { "epoch": 0.64, "learning_rate": 3.573910719059075e-08, "loss": 0.0552, "step": 2404 }, { "epoch": 0.64, "learning_rate": 3.5712376369954554e-08, "loss": 0.0899, "step": 2405 }, { "epoch": 0.64, "learning_rate": 3.568564554931836e-08, "loss": 0.0709, "step": 2406 }, { "epoch": 0.64, "learning_rate": 3.565891472868217e-08, "loss": 0.0646, "step": 2407 }, { "epoch": 0.64, "learning_rate": 3.5632183908045974e-08, "loss": 0.044, "step": 2408 }, { "epoch": 0.64, "learning_rate": 3.560545308740978e-08, "loss": 0.0559, "step": 2409 }, { "epoch": 0.64, "learning_rate": 3.5578722266773584e-08, "loss": 0.0753, "step": 2410 }, { "epoch": 0.64, "learning_rate": 3.5551991446137395e-08, "loss": 0.0585, "step": 2411 }, { "epoch": 0.64, "learning_rate": 3.55252606255012e-08, "loss": 0.0752, "step": 2412 }, { "epoch": 0.64, "learning_rate": 3.5498529804865005e-08, "loss": 0.0767, "step": 2413 }, { "epoch": 0.65, "learning_rate": 3.547179898422881e-08, "loss": 0.0612, "step": 2414 }, { "epoch": 0.65, "learning_rate": 3.544506816359262e-08, "loss": 0.049, "step": 2415 }, { "epoch": 0.65, "learning_rate": 3.5418337342956425e-08, "loss": 0.0586, "step": 2416 }, { "epoch": 0.65, "learning_rate": 3.539160652232023e-08, "loss": 0.0702, "step": 2417 }, { "epoch": 0.65, "learning_rate": 3.5364875701684035e-08, "loss": 0.0506, "step": 2418 }, { "epoch": 0.65, "learning_rate": 3.5338144881047846e-08, "loss": 0.0802, "step": 2419 }, { "epoch": 0.65, "learning_rate": 3.531141406041165e-08, "loss": 0.0677, "step": 2420 }, { "epoch": 0.65, "learning_rate": 3.5284683239775456e-08, "loss": 0.0613, "step": 2421 }, { "epoch": 0.65, "learning_rate": 3.525795241913927e-08, "loss": 0.0757, "step": 2422 }, { "epoch": 0.65, "learning_rate": 3.523122159850307e-08, "loss": 0.0781, "step": 2423 }, { "epoch": 0.65, "learning_rate": 3.5204490777866877e-08, "loss": 0.0592, "step": 2424 }, { "epoch": 0.65, "learning_rate": 3.517775995723069e-08, "loss": 0.0417, "step": 2425 }, { "epoch": 0.65, "learning_rate": 3.515102913659449e-08, "loss": 0.0686, "step": 2426 }, { "epoch": 0.65, "learning_rate": 3.51242983159583e-08, "loss": 0.0634, "step": 2427 }, { "epoch": 0.65, "learning_rate": 3.509756749532211e-08, "loss": 0.0622, "step": 2428 }, { "epoch": 0.65, "learning_rate": 3.5070836674685914e-08, "loss": 0.0526, "step": 2429 }, { "epoch": 0.65, "learning_rate": 3.504410585404972e-08, "loss": 0.0626, "step": 2430 }, { "epoch": 0.65, "learning_rate": 3.501737503341353e-08, "loss": 0.0824, "step": 2431 }, { "epoch": 0.65, "learning_rate": 3.4990644212777334e-08, "loss": 0.0593, "step": 2432 }, { "epoch": 0.65, "learning_rate": 3.496391339214114e-08, "loss": 0.05, "step": 2433 }, { "epoch": 0.65, "learning_rate": 3.4937182571504944e-08, "loss": 0.0628, "step": 2434 }, { "epoch": 0.65, "learning_rate": 3.4910451750868755e-08, "loss": 0.068, "step": 2435 }, { "epoch": 0.65, "learning_rate": 3.488372093023256e-08, "loss": 0.0551, "step": 2436 }, { "epoch": 0.65, "learning_rate": 3.4856990109596365e-08, "loss": 0.0643, "step": 2437 }, { "epoch": 0.65, "learning_rate": 3.483025928896017e-08, "loss": 0.0518, "step": 2438 }, { "epoch": 0.65, "learning_rate": 3.480352846832398e-08, "loss": 0.0502, "step": 2439 }, { "epoch": 0.65, "learning_rate": 3.4776797647687786e-08, "loss": 0.0722, "step": 2440 }, { "epoch": 0.65, "learning_rate": 3.475006682705159e-08, "loss": 0.0542, "step": 2441 }, { "epoch": 0.65, "learning_rate": 3.4723336006415395e-08, "loss": 0.0655, "step": 2442 }, { "epoch": 0.65, "learning_rate": 3.4696605185779206e-08, "loss": 0.0638, "step": 2443 }, { "epoch": 0.65, "learning_rate": 3.466987436514301e-08, "loss": 0.0815, "step": 2444 }, { "epoch": 0.65, "learning_rate": 3.4643143544506816e-08, "loss": 0.0593, "step": 2445 }, { "epoch": 0.65, "learning_rate": 3.461641272387062e-08, "loss": 0.0513, "step": 2446 }, { "epoch": 0.65, "learning_rate": 3.458968190323443e-08, "loss": 0.0813, "step": 2447 }, { "epoch": 0.65, "learning_rate": 3.4562951082598237e-08, "loss": 0.048, "step": 2448 }, { "epoch": 0.65, "learning_rate": 3.453622026196204e-08, "loss": 0.054, "step": 2449 }, { "epoch": 0.65, "learning_rate": 3.4509489441325846e-08, "loss": 0.0607, "step": 2450 }, { "epoch": 0.66, "learning_rate": 3.448275862068966e-08, "loss": 0.0665, "step": 2451 }, { "epoch": 0.66, "learning_rate": 3.445602780005346e-08, "loss": 0.052, "step": 2452 }, { "epoch": 0.66, "learning_rate": 3.442929697941727e-08, "loss": 0.0538, "step": 2453 }, { "epoch": 0.66, "learning_rate": 3.440256615878107e-08, "loss": 0.0775, "step": 2454 }, { "epoch": 0.66, "learning_rate": 3.437583533814488e-08, "loss": 0.0541, "step": 2455 }, { "epoch": 0.66, "learning_rate": 3.434910451750869e-08, "loss": 0.0645, "step": 2456 }, { "epoch": 0.66, "learning_rate": 3.432237369687249e-08, "loss": 0.0548, "step": 2457 }, { "epoch": 0.66, "learning_rate": 3.42956428762363e-08, "loss": 0.0601, "step": 2458 }, { "epoch": 0.66, "learning_rate": 3.426891205560011e-08, "loss": 0.0562, "step": 2459 }, { "epoch": 0.66, "learning_rate": 3.4242181234963913e-08, "loss": 0.0611, "step": 2460 }, { "epoch": 0.66, "learning_rate": 3.421545041432772e-08, "loss": 0.0665, "step": 2461 }, { "epoch": 0.66, "learning_rate": 3.418871959369152e-08, "loss": 0.0566, "step": 2462 }, { "epoch": 0.66, "learning_rate": 3.4161988773055334e-08, "loss": 0.0822, "step": 2463 }, { "epoch": 0.66, "learning_rate": 3.413525795241914e-08, "loss": 0.0495, "step": 2464 }, { "epoch": 0.66, "learning_rate": 3.4108527131782944e-08, "loss": 0.0425, "step": 2465 }, { "epoch": 0.66, "learning_rate": 3.408179631114675e-08, "loss": 0.0651, "step": 2466 }, { "epoch": 0.66, "learning_rate": 3.405506549051056e-08, "loss": 0.0493, "step": 2467 }, { "epoch": 0.66, "learning_rate": 3.4028334669874365e-08, "loss": 0.0704, "step": 2468 }, { "epoch": 0.66, "learning_rate": 3.400160384923817e-08, "loss": 0.1064, "step": 2469 }, { "epoch": 0.66, "learning_rate": 3.3974873028601974e-08, "loss": 0.0524, "step": 2470 }, { "epoch": 0.66, "learning_rate": 3.3948142207965785e-08, "loss": 0.0521, "step": 2471 }, { "epoch": 0.66, "learning_rate": 3.392141138732959e-08, "loss": 0.065, "step": 2472 }, { "epoch": 0.66, "learning_rate": 3.3894680566693395e-08, "loss": 0.0517, "step": 2473 }, { "epoch": 0.66, "learning_rate": 3.38679497460572e-08, "loss": 0.0615, "step": 2474 }, { "epoch": 0.66, "learning_rate": 3.384121892542101e-08, "loss": 0.0592, "step": 2475 }, { "epoch": 0.66, "learning_rate": 3.3814488104784816e-08, "loss": 0.0532, "step": 2476 }, { "epoch": 0.66, "learning_rate": 3.378775728414862e-08, "loss": 0.0557, "step": 2477 }, { "epoch": 0.66, "learning_rate": 3.3761026463512425e-08, "loss": 0.0691, "step": 2478 }, { "epoch": 0.66, "learning_rate": 3.3734295642876236e-08, "loss": 0.0736, "step": 2479 }, { "epoch": 0.66, "learning_rate": 3.370756482224004e-08, "loss": 0.0594, "step": 2480 }, { "epoch": 0.66, "learning_rate": 3.3680834001603846e-08, "loss": 0.0789, "step": 2481 }, { "epoch": 0.66, "learning_rate": 3.365410318096765e-08, "loss": 0.0755, "step": 2482 }, { "epoch": 0.66, "learning_rate": 3.362737236033146e-08, "loss": 0.0506, "step": 2483 }, { "epoch": 0.66, "learning_rate": 3.360064153969527e-08, "loss": 0.0581, "step": 2484 }, { "epoch": 0.66, "learning_rate": 3.357391071905907e-08, "loss": 0.0505, "step": 2485 }, { "epoch": 0.66, "learning_rate": 3.3547179898422876e-08, "loss": 0.0537, "step": 2486 }, { "epoch": 0.66, "learning_rate": 3.352044907778669e-08, "loss": 0.0656, "step": 2487 }, { "epoch": 0.66, "learning_rate": 3.349371825715049e-08, "loss": 0.0708, "step": 2488 }, { "epoch": 0.67, "learning_rate": 3.34669874365143e-08, "loss": 0.0572, "step": 2489 }, { "epoch": 0.67, "learning_rate": 3.34402566158781e-08, "loss": 0.0403, "step": 2490 }, { "epoch": 0.67, "learning_rate": 3.341352579524191e-08, "loss": 0.0605, "step": 2491 }, { "epoch": 0.67, "learning_rate": 3.338679497460572e-08, "loss": 0.0621, "step": 2492 }, { "epoch": 0.67, "learning_rate": 3.336006415396952e-08, "loss": 0.0439, "step": 2493 }, { "epoch": 0.67, "learning_rate": 3.333333333333333e-08, "loss": 0.08, "step": 2494 }, { "epoch": 0.67, "learning_rate": 3.330660251269714e-08, "loss": 0.0712, "step": 2495 }, { "epoch": 0.67, "learning_rate": 3.3279871692060944e-08, "loss": 0.0803, "step": 2496 }, { "epoch": 0.67, "learning_rate": 3.325314087142475e-08, "loss": 0.0625, "step": 2497 }, { "epoch": 0.67, "learning_rate": 3.322641005078855e-08, "loss": 0.0611, "step": 2498 }, { "epoch": 0.67, "learning_rate": 3.3199679230152364e-08, "loss": 0.076, "step": 2499 }, { "epoch": 0.67, "learning_rate": 3.317294840951617e-08, "loss": 0.0628, "step": 2500 }, { "epoch": 0.67, "learning_rate": 3.3146217588879974e-08, "loss": 0.0593, "step": 2501 }, { "epoch": 0.67, "learning_rate": 3.311948676824378e-08, "loss": 0.1012, "step": 2502 }, { "epoch": 0.67, "learning_rate": 3.309275594760759e-08, "loss": 0.0624, "step": 2503 }, { "epoch": 0.67, "learning_rate": 3.3066025126971395e-08, "loss": 0.0639, "step": 2504 }, { "epoch": 0.67, "learning_rate": 3.30392943063352e-08, "loss": 0.0753, "step": 2505 }, { "epoch": 0.67, "learning_rate": 3.301256348569901e-08, "loss": 0.0638, "step": 2506 }, { "epoch": 0.67, "learning_rate": 3.2985832665062815e-08, "loss": 0.08, "step": 2507 }, { "epoch": 0.67, "learning_rate": 3.295910184442662e-08, "loss": 0.0789, "step": 2508 }, { "epoch": 0.67, "learning_rate": 3.293237102379043e-08, "loss": 0.0783, "step": 2509 }, { "epoch": 0.67, "learning_rate": 3.2905640203154236e-08, "loss": 0.0676, "step": 2510 }, { "epoch": 0.67, "learning_rate": 3.287890938251804e-08, "loss": 0.1095, "step": 2511 }, { "epoch": 0.67, "learning_rate": 3.285217856188185e-08, "loss": 0.0714, "step": 2512 }, { "epoch": 0.67, "learning_rate": 3.282544774124566e-08, "loss": 0.0663, "step": 2513 }, { "epoch": 0.67, "learning_rate": 3.279871692060946e-08, "loss": 0.0659, "step": 2514 }, { "epoch": 0.67, "learning_rate": 3.277198609997327e-08, "loss": 0.0774, "step": 2515 }, { "epoch": 0.67, "learning_rate": 3.274525527933708e-08, "loss": 0.0887, "step": 2516 }, { "epoch": 0.67, "learning_rate": 3.271852445870088e-08, "loss": 0.0622, "step": 2517 }, { "epoch": 0.67, "learning_rate": 3.269179363806469e-08, "loss": 0.0665, "step": 2518 }, { "epoch": 0.67, "learning_rate": 3.26650628174285e-08, "loss": 0.0873, "step": 2519 }, { "epoch": 0.67, "learning_rate": 3.2638331996792304e-08, "loss": 0.1119, "step": 2520 }, { "epoch": 0.67, "learning_rate": 3.261160117615611e-08, "loss": 0.0729, "step": 2521 }, { "epoch": 0.67, "learning_rate": 3.258487035551991e-08, "loss": 0.0758, "step": 2522 }, { "epoch": 0.67, "learning_rate": 3.2558139534883724e-08, "loss": 0.0862, "step": 2523 }, { "epoch": 0.67, "learning_rate": 3.253140871424753e-08, "loss": 0.055, "step": 2524 }, { "epoch": 0.67, "learning_rate": 3.2504677893611334e-08, "loss": 0.0964, "step": 2525 }, { "epoch": 0.68, "learning_rate": 3.247794707297514e-08, "loss": 0.1061, "step": 2526 }, { "epoch": 0.68, "learning_rate": 3.245121625233895e-08, "loss": 0.0808, "step": 2527 }, { "epoch": 0.68, "learning_rate": 3.2424485431702755e-08, "loss": 0.0713, "step": 2528 }, { "epoch": 0.68, "learning_rate": 3.239775461106656e-08, "loss": 0.0513, "step": 2529 }, { "epoch": 0.68, "learning_rate": 3.2371023790430364e-08, "loss": 0.0856, "step": 2530 }, { "epoch": 0.68, "learning_rate": 3.2344292969794176e-08, "loss": 0.0846, "step": 2531 }, { "epoch": 0.68, "learning_rate": 3.231756214915798e-08, "loss": 0.0789, "step": 2532 }, { "epoch": 0.68, "learning_rate": 3.2290831328521785e-08, "loss": 0.0536, "step": 2533 }, { "epoch": 0.68, "learning_rate": 3.226410050788559e-08, "loss": 0.0853, "step": 2534 }, { "epoch": 0.68, "learning_rate": 3.22373696872494e-08, "loss": 0.0712, "step": 2535 }, { "epoch": 0.68, "learning_rate": 3.2210638866613206e-08, "loss": 0.0545, "step": 2536 }, { "epoch": 0.68, "learning_rate": 3.218390804597701e-08, "loss": 0.0546, "step": 2537 }, { "epoch": 0.68, "learning_rate": 3.2157177225340815e-08, "loss": 0.0764, "step": 2538 }, { "epoch": 0.68, "learning_rate": 3.2130446404704627e-08, "loss": 0.0749, "step": 2539 }, { "epoch": 0.68, "learning_rate": 3.210371558406843e-08, "loss": 0.0777, "step": 2540 }, { "epoch": 0.68, "learning_rate": 3.2076984763432236e-08, "loss": 0.0726, "step": 2541 }, { "epoch": 0.68, "learning_rate": 3.205025394279604e-08, "loss": 0.0841, "step": 2542 }, { "epoch": 0.68, "learning_rate": 3.202352312215985e-08, "loss": 0.078, "step": 2543 }, { "epoch": 0.68, "learning_rate": 3.199679230152366e-08, "loss": 0.0804, "step": 2544 }, { "epoch": 0.68, "learning_rate": 3.197006148088746e-08, "loss": 0.068, "step": 2545 }, { "epoch": 0.68, "learning_rate": 3.1943330660251266e-08, "loss": 0.0648, "step": 2546 }, { "epoch": 0.68, "learning_rate": 3.191659983961508e-08, "loss": 0.058, "step": 2547 }, { "epoch": 0.68, "learning_rate": 3.188986901897888e-08, "loss": 0.0619, "step": 2548 }, { "epoch": 0.68, "learning_rate": 3.186313819834269e-08, "loss": 0.0903, "step": 2549 }, { "epoch": 0.68, "learning_rate": 3.183640737770649e-08, "loss": 0.0743, "step": 2550 }, { "epoch": 0.68, "learning_rate": 3.1809676557070303e-08, "loss": 0.0654, "step": 2551 }, { "epoch": 0.68, "learning_rate": 3.178294573643411e-08, "loss": 0.0591, "step": 2552 }, { "epoch": 0.68, "learning_rate": 3.175621491579791e-08, "loss": 0.0592, "step": 2553 }, { "epoch": 0.68, "learning_rate": 3.172948409516172e-08, "loss": 0.0702, "step": 2554 }, { "epoch": 0.68, "learning_rate": 3.170275327452553e-08, "loss": 0.0877, "step": 2555 }, { "epoch": 0.68, "learning_rate": 3.1676022453889334e-08, "loss": 0.0909, "step": 2556 }, { "epoch": 0.68, "learning_rate": 3.164929163325314e-08, "loss": 0.0623, "step": 2557 }, { "epoch": 0.68, "learning_rate": 3.162256081261694e-08, "loss": 0.0477, "step": 2558 }, { "epoch": 0.68, "learning_rate": 3.1595829991980755e-08, "loss": 0.077, "step": 2559 }, { "epoch": 0.68, "learning_rate": 3.156909917134456e-08, "loss": 0.1278, "step": 2560 }, { "epoch": 0.68, "learning_rate": 3.1542368350708364e-08, "loss": 0.0923, "step": 2561 }, { "epoch": 0.68, "learning_rate": 3.151563753007217e-08, "loss": 0.1066, "step": 2562 }, { "epoch": 0.69, "learning_rate": 3.148890670943598e-08, "loss": 0.0734, "step": 2563 }, { "epoch": 0.69, "learning_rate": 3.1462175888799785e-08, "loss": 0.0724, "step": 2564 }, { "epoch": 0.69, "learning_rate": 3.143544506816359e-08, "loss": 0.0815, "step": 2565 }, { "epoch": 0.69, "learning_rate": 3.1408714247527394e-08, "loss": 0.0689, "step": 2566 }, { "epoch": 0.69, "learning_rate": 3.1381983426891206e-08, "loss": 0.073, "step": 2567 }, { "epoch": 0.69, "learning_rate": 3.135525260625501e-08, "loss": 0.0854, "step": 2568 }, { "epoch": 0.69, "learning_rate": 3.1328521785618815e-08, "loss": 0.0721, "step": 2569 }, { "epoch": 0.69, "learning_rate": 3.130179096498262e-08, "loss": 0.0635, "step": 2570 }, { "epoch": 0.69, "learning_rate": 3.127506014434643e-08, "loss": 0.0653, "step": 2571 }, { "epoch": 0.69, "learning_rate": 3.1248329323710236e-08, "loss": 0.0809, "step": 2572 }, { "epoch": 0.69, "learning_rate": 3.122159850307404e-08, "loss": 0.0594, "step": 2573 }, { "epoch": 0.69, "learning_rate": 3.1194867682437845e-08, "loss": 0.0732, "step": 2574 }, { "epoch": 0.69, "learning_rate": 3.116813686180166e-08, "loss": 0.0926, "step": 2575 }, { "epoch": 0.69, "learning_rate": 3.114140604116546e-08, "loss": 0.0932, "step": 2576 }, { "epoch": 0.69, "learning_rate": 3.1114675220529266e-08, "loss": 0.0922, "step": 2577 }, { "epoch": 0.69, "learning_rate": 3.108794439989307e-08, "loss": 0.0698, "step": 2578 }, { "epoch": 0.69, "learning_rate": 3.106121357925688e-08, "loss": 0.0701, "step": 2579 }, { "epoch": 0.69, "learning_rate": 3.103448275862069e-08, "loss": 0.088, "step": 2580 }, { "epoch": 0.69, "learning_rate": 3.100775193798449e-08, "loss": 0.0731, "step": 2581 }, { "epoch": 0.69, "learning_rate": 3.0981021117348297e-08, "loss": 0.0783, "step": 2582 }, { "epoch": 0.69, "learning_rate": 3.095429029671211e-08, "loss": 0.0776, "step": 2583 }, { "epoch": 0.69, "learning_rate": 3.092755947607591e-08, "loss": 0.0907, "step": 2584 }, { "epoch": 0.69, "learning_rate": 3.090082865543972e-08, "loss": 0.0836, "step": 2585 }, { "epoch": 0.69, "learning_rate": 3.087409783480352e-08, "loss": 0.073, "step": 2586 }, { "epoch": 0.69, "learning_rate": 3.0847367014167334e-08, "loss": 0.0975, "step": 2587 }, { "epoch": 0.69, "learning_rate": 3.082063619353114e-08, "loss": 0.088, "step": 2588 }, { "epoch": 0.69, "learning_rate": 3.079390537289494e-08, "loss": 0.0725, "step": 2589 }, { "epoch": 0.69, "learning_rate": 3.0767174552258754e-08, "loss": 0.0793, "step": 2590 }, { "epoch": 0.69, "learning_rate": 3.074044373162256e-08, "loss": 0.0863, "step": 2591 }, { "epoch": 0.69, "learning_rate": 3.0713712910986364e-08, "loss": 0.1005, "step": 2592 }, { "epoch": 0.69, "learning_rate": 3.0686982090350175e-08, "loss": 0.0879, "step": 2593 }, { "epoch": 0.69, "learning_rate": 3.066025126971398e-08, "loss": 0.0617, "step": 2594 }, { "epoch": 0.69, "learning_rate": 3.0633520449077785e-08, "loss": 0.0487, "step": 2595 }, { "epoch": 0.69, "learning_rate": 3.0606789628441596e-08, "loss": 0.1063, "step": 2596 }, { "epoch": 0.69, "learning_rate": 3.05800588078054e-08, "loss": 0.0962, "step": 2597 }, { "epoch": 0.69, "learning_rate": 3.0553327987169205e-08, "loss": 0.0868, "step": 2598 }, { "epoch": 0.69, "learning_rate": 3.052659716653302e-08, "loss": 0.072, "step": 2599 }, { "epoch": 0.69, "learning_rate": 3.049986634589682e-08, "loss": 0.0657, "step": 2600 }, { "epoch": 0.7, "learning_rate": 3.0473135525260626e-08, "loss": 0.1249, "step": 2601 }, { "epoch": 0.7, "learning_rate": 3.044640470462443e-08, "loss": 0.0874, "step": 2602 }, { "epoch": 0.7, "learning_rate": 3.041967388398824e-08, "loss": 0.0842, "step": 2603 }, { "epoch": 0.7, "learning_rate": 3.039294306335205e-08, "loss": 0.058, "step": 2604 }, { "epoch": 0.7, "learning_rate": 3.036621224271585e-08, "loss": 0.0938, "step": 2605 }, { "epoch": 0.7, "learning_rate": 3.0339481422079657e-08, "loss": 0.0609, "step": 2606 }, { "epoch": 0.7, "learning_rate": 3.031275060144347e-08, "loss": 0.0797, "step": 2607 }, { "epoch": 0.7, "learning_rate": 3.028601978080727e-08, "loss": 0.0651, "step": 2608 }, { "epoch": 0.7, "learning_rate": 3.025928896017108e-08, "loss": 0.0684, "step": 2609 }, { "epoch": 0.7, "learning_rate": 3.023255813953488e-08, "loss": 0.0739, "step": 2610 }, { "epoch": 0.7, "learning_rate": 3.0205827318898694e-08, "loss": 0.0987, "step": 2611 }, { "epoch": 0.7, "learning_rate": 3.01790964982625e-08, "loss": 0.0875, "step": 2612 }, { "epoch": 0.7, "learning_rate": 3.01523656776263e-08, "loss": 0.0672, "step": 2613 }, { "epoch": 0.7, "learning_rate": 3.012563485699011e-08, "loss": 0.0781, "step": 2614 }, { "epoch": 0.7, "learning_rate": 3.009890403635392e-08, "loss": 0.0804, "step": 2615 }, { "epoch": 0.7, "learning_rate": 3.0072173215717724e-08, "loss": 0.0727, "step": 2616 }, { "epoch": 0.7, "learning_rate": 3.004544239508153e-08, "loss": 0.093, "step": 2617 }, { "epoch": 0.7, "learning_rate": 3.0018711574445333e-08, "loss": 0.0614, "step": 2618 }, { "epoch": 0.7, "learning_rate": 2.9991980753809145e-08, "loss": 0.0923, "step": 2619 }, { "epoch": 0.7, "learning_rate": 2.996524993317295e-08, "loss": 0.0641, "step": 2620 }, { "epoch": 0.7, "learning_rate": 2.9938519112536754e-08, "loss": 0.0878, "step": 2621 }, { "epoch": 0.7, "learning_rate": 2.991178829190056e-08, "loss": 0.0799, "step": 2622 }, { "epoch": 0.7, "learning_rate": 2.988505747126437e-08, "loss": 0.074, "step": 2623 }, { "epoch": 0.7, "learning_rate": 2.9858326650628175e-08, "loss": 0.0466, "step": 2624 }, { "epoch": 0.7, "learning_rate": 2.983159582999198e-08, "loss": 0.0717, "step": 2625 }, { "epoch": 0.7, "learning_rate": 2.9804865009355784e-08, "loss": 0.0959, "step": 2626 }, { "epoch": 0.7, "learning_rate": 2.9778134188719596e-08, "loss": 0.0729, "step": 2627 }, { "epoch": 0.7, "learning_rate": 2.97514033680834e-08, "loss": 0.0548, "step": 2628 }, { "epoch": 0.7, "learning_rate": 2.9724672547447205e-08, "loss": 0.0546, "step": 2629 }, { "epoch": 0.7, "learning_rate": 2.969794172681101e-08, "loss": 0.0892, "step": 2630 }, { "epoch": 0.7, "learning_rate": 2.967121090617482e-08, "loss": 0.0878, "step": 2631 }, { "epoch": 0.7, "learning_rate": 2.9644480085538626e-08, "loss": 0.1249, "step": 2632 }, { "epoch": 0.7, "learning_rate": 2.961774926490243e-08, "loss": 0.0737, "step": 2633 }, { "epoch": 0.7, "learning_rate": 2.9591018444266236e-08, "loss": 0.1088, "step": 2634 }, { "epoch": 0.7, "learning_rate": 2.9564287623630047e-08, "loss": 0.0688, "step": 2635 }, { "epoch": 0.7, "learning_rate": 2.9537556802993852e-08, "loss": 0.0883, "step": 2636 }, { "epoch": 0.7, "learning_rate": 2.9510825982357656e-08, "loss": 0.0529, "step": 2637 }, { "epoch": 0.71, "learning_rate": 2.948409516172146e-08, "loss": 0.0676, "step": 2638 }, { "epoch": 0.71, "learning_rate": 2.9457364341085273e-08, "loss": 0.0838, "step": 2639 }, { "epoch": 0.71, "learning_rate": 2.9430633520449077e-08, "loss": 0.0567, "step": 2640 }, { "epoch": 0.71, "learning_rate": 2.9403902699812882e-08, "loss": 0.0724, "step": 2641 }, { "epoch": 0.71, "learning_rate": 2.9377171879176687e-08, "loss": 0.063, "step": 2642 }, { "epoch": 0.71, "learning_rate": 2.9350441058540498e-08, "loss": 0.0733, "step": 2643 }, { "epoch": 0.71, "learning_rate": 2.9323710237904303e-08, "loss": 0.0588, "step": 2644 }, { "epoch": 0.71, "learning_rate": 2.9296979417268108e-08, "loss": 0.0616, "step": 2645 }, { "epoch": 0.71, "learning_rate": 2.9270248596631912e-08, "loss": 0.0669, "step": 2646 }, { "epoch": 0.71, "learning_rate": 2.9243517775995724e-08, "loss": 0.0856, "step": 2647 }, { "epoch": 0.71, "learning_rate": 2.921678695535953e-08, "loss": 0.0753, "step": 2648 }, { "epoch": 0.71, "learning_rate": 2.9190056134723333e-08, "loss": 0.0772, "step": 2649 }, { "epoch": 0.71, "learning_rate": 2.9163325314087138e-08, "loss": 0.0749, "step": 2650 }, { "epoch": 0.71, "learning_rate": 2.913659449345095e-08, "loss": 0.071, "step": 2651 }, { "epoch": 0.71, "learning_rate": 2.9109863672814754e-08, "loss": 0.0827, "step": 2652 }, { "epoch": 0.71, "learning_rate": 2.908313285217856e-08, "loss": 0.0694, "step": 2653 }, { "epoch": 0.71, "learning_rate": 2.9056402031542367e-08, "loss": 0.081, "step": 2654 }, { "epoch": 0.71, "learning_rate": 2.9029671210906175e-08, "loss": 0.068, "step": 2655 }, { "epoch": 0.71, "learning_rate": 2.900294039026998e-08, "loss": 0.0699, "step": 2656 }, { "epoch": 0.71, "learning_rate": 2.8976209569633788e-08, "loss": 0.0856, "step": 2657 }, { "epoch": 0.71, "learning_rate": 2.8976209569633788e-08, "loss": 0.0668, "step": 2658 }, { "epoch": 0.71, "learning_rate": 2.8949478748997592e-08, "loss": 0.0605, "step": 2659 }, { "epoch": 0.71, "learning_rate": 2.89227479283614e-08, "loss": 0.0636, "step": 2660 }, { "epoch": 0.71, "learning_rate": 2.889601710772521e-08, "loss": 0.0693, "step": 2661 }, { "epoch": 0.71, "learning_rate": 2.8869286287089013e-08, "loss": 0.0774, "step": 2662 }, { "epoch": 0.71, "learning_rate": 2.8842555466452818e-08, "loss": 0.0693, "step": 2663 }, { "epoch": 0.71, "learning_rate": 2.881582464581663e-08, "loss": 0.059, "step": 2664 }, { "epoch": 0.71, "learning_rate": 2.8789093825180434e-08, "loss": 0.0702, "step": 2665 }, { "epoch": 0.71, "learning_rate": 2.876236300454424e-08, "loss": 0.0707, "step": 2666 }, { "epoch": 0.71, "learning_rate": 2.8735632183908043e-08, "loss": 0.0934, "step": 2667 }, { "epoch": 0.71, "learning_rate": 2.8708901363271855e-08, "loss": 0.0862, "step": 2668 }, { "epoch": 0.71, "learning_rate": 2.868217054263566e-08, "loss": 0.0631, "step": 2669 }, { "epoch": 0.71, "learning_rate": 2.8655439721999464e-08, "loss": 0.0803, "step": 2670 }, { "epoch": 0.71, "learning_rate": 2.862870890136327e-08, "loss": 0.057, "step": 2671 }, { "epoch": 0.71, "learning_rate": 2.860197808072708e-08, "loss": 0.0655, "step": 2672 }, { "epoch": 0.71, "learning_rate": 2.8575247260090885e-08, "loss": 0.0729, "step": 2673 }, { "epoch": 0.71, "learning_rate": 2.854851643945469e-08, "loss": 0.0893, "step": 2674 }, { "epoch": 0.71, "learning_rate": 2.8521785618818495e-08, "loss": 0.0847, "step": 2675 }, { "epoch": 0.72, "learning_rate": 2.8495054798182306e-08, "loss": 0.0859, "step": 2676 }, { "epoch": 0.72, "learning_rate": 2.846832397754611e-08, "loss": 0.0644, "step": 2677 }, { "epoch": 0.72, "learning_rate": 2.8441593156909915e-08, "loss": 0.0758, "step": 2678 }, { "epoch": 0.72, "learning_rate": 2.841486233627372e-08, "loss": 0.0761, "step": 2679 }, { "epoch": 0.72, "learning_rate": 2.838813151563753e-08, "loss": 0.0674, "step": 2680 }, { "epoch": 0.72, "learning_rate": 2.8361400695001336e-08, "loss": 0.0622, "step": 2681 }, { "epoch": 0.72, "learning_rate": 2.833466987436514e-08, "loss": 0.0839, "step": 2682 }, { "epoch": 0.72, "learning_rate": 2.8307939053728946e-08, "loss": 0.073, "step": 2683 }, { "epoch": 0.72, "learning_rate": 2.8281208233092757e-08, "loss": 0.0603, "step": 2684 }, { "epoch": 0.72, "learning_rate": 2.8254477412456562e-08, "loss": 0.0852, "step": 2685 }, { "epoch": 0.72, "learning_rate": 2.8227746591820367e-08, "loss": 0.0975, "step": 2686 }, { "epoch": 0.72, "learning_rate": 2.820101577118417e-08, "loss": 0.0837, "step": 2687 }, { "epoch": 0.72, "learning_rate": 2.8174284950547983e-08, "loss": 0.0671, "step": 2688 }, { "epoch": 0.72, "learning_rate": 2.8147554129911787e-08, "loss": 0.0678, "step": 2689 }, { "epoch": 0.72, "learning_rate": 2.8120823309275592e-08, "loss": 0.0705, "step": 2690 }, { "epoch": 0.72, "learning_rate": 2.8094092488639397e-08, "loss": 0.0755, "step": 2691 }, { "epoch": 0.72, "learning_rate": 2.8067361668003208e-08, "loss": 0.0563, "step": 2692 }, { "epoch": 0.72, "learning_rate": 2.8040630847367013e-08, "loss": 0.0731, "step": 2693 }, { "epoch": 0.72, "learning_rate": 2.8013900026730818e-08, "loss": 0.057, "step": 2694 }, { "epoch": 0.72, "learning_rate": 2.7987169206094622e-08, "loss": 0.0757, "step": 2695 }, { "epoch": 0.72, "learning_rate": 2.7960438385458434e-08, "loss": 0.0724, "step": 2696 }, { "epoch": 0.72, "learning_rate": 2.793370756482224e-08, "loss": 0.0595, "step": 2697 }, { "epoch": 0.72, "learning_rate": 2.7906976744186043e-08, "loss": 0.0702, "step": 2698 }, { "epoch": 0.72, "learning_rate": 2.788024592354985e-08, "loss": 0.096, "step": 2699 }, { "epoch": 0.72, "learning_rate": 2.785351510291366e-08, "loss": 0.0943, "step": 2700 }, { "epoch": 0.72, "learning_rate": 2.7826784282277464e-08, "loss": 0.0634, "step": 2701 }, { "epoch": 0.72, "learning_rate": 2.7800053461641272e-08, "loss": 0.0834, "step": 2702 }, { "epoch": 0.72, "learning_rate": 2.7773322641005077e-08, "loss": 0.0582, "step": 2703 }, { "epoch": 0.72, "learning_rate": 2.7746591820368885e-08, "loss": 0.0836, "step": 2704 }, { "epoch": 0.72, "learning_rate": 2.7719860999732693e-08, "loss": 0.0961, "step": 2705 }, { "epoch": 0.72, "learning_rate": 2.7693130179096498e-08, "loss": 0.0726, "step": 2706 }, { "epoch": 0.72, "learning_rate": 2.7666399358460303e-08, "loss": 0.0666, "step": 2707 }, { "epoch": 0.72, "learning_rate": 2.7639668537824114e-08, "loss": 0.097, "step": 2708 }, { "epoch": 0.72, "learning_rate": 2.761293771718792e-08, "loss": 0.0748, "step": 2709 }, { "epoch": 0.72, "learning_rate": 2.7586206896551723e-08, "loss": 0.0855, "step": 2710 }, { "epoch": 0.72, "learning_rate": 2.7559476075915528e-08, "loss": 0.0894, "step": 2711 }, { "epoch": 0.72, "learning_rate": 2.753274525527934e-08, "loss": 0.0708, "step": 2712 }, { "epoch": 0.73, "learning_rate": 2.7506014434643144e-08, "loss": 0.0808, "step": 2713 }, { "epoch": 0.73, "learning_rate": 2.747928361400695e-08, "loss": 0.0646, "step": 2714 }, { "epoch": 0.73, "learning_rate": 2.7452552793370754e-08, "loss": 0.0818, "step": 2715 }, { "epoch": 0.73, "learning_rate": 2.7425821972734565e-08, "loss": 0.1105, "step": 2716 }, { "epoch": 0.73, "learning_rate": 2.739909115209837e-08, "loss": 0.0886, "step": 2717 }, { "epoch": 0.73, "learning_rate": 2.7372360331462174e-08, "loss": 0.0967, "step": 2718 }, { "epoch": 0.73, "learning_rate": 2.734562951082598e-08, "loss": 0.0855, "step": 2719 }, { "epoch": 0.73, "learning_rate": 2.731889869018979e-08, "loss": 0.0831, "step": 2720 }, { "epoch": 0.73, "learning_rate": 2.7292167869553595e-08, "loss": 0.082, "step": 2721 }, { "epoch": 0.73, "learning_rate": 2.72654370489174e-08, "loss": 0.0769, "step": 2722 }, { "epoch": 0.73, "learning_rate": 2.7238706228281205e-08, "loss": 0.0756, "step": 2723 }, { "epoch": 0.73, "learning_rate": 2.7211975407645016e-08, "loss": 0.0902, "step": 2724 }, { "epoch": 0.73, "learning_rate": 2.718524458700882e-08, "loss": 0.0727, "step": 2725 }, { "epoch": 0.73, "learning_rate": 2.7158513766372626e-08, "loss": 0.0752, "step": 2726 }, { "epoch": 0.73, "learning_rate": 2.713178294573643e-08, "loss": 0.0835, "step": 2727 }, { "epoch": 0.73, "learning_rate": 2.7105052125100242e-08, "loss": 0.0653, "step": 2728 }, { "epoch": 0.73, "learning_rate": 2.7078321304464046e-08, "loss": 0.0876, "step": 2729 }, { "epoch": 0.73, "learning_rate": 2.705159048382785e-08, "loss": 0.0977, "step": 2730 }, { "epoch": 0.73, "learning_rate": 2.7024859663191656e-08, "loss": 0.0677, "step": 2731 }, { "epoch": 0.73, "learning_rate": 2.6998128842555467e-08, "loss": 0.0927, "step": 2732 }, { "epoch": 0.73, "learning_rate": 2.6971398021919272e-08, "loss": 0.0997, "step": 2733 }, { "epoch": 0.73, "learning_rate": 2.6944667201283077e-08, "loss": 0.0788, "step": 2734 }, { "epoch": 0.73, "learning_rate": 2.691793638064688e-08, "loss": 0.0789, "step": 2735 }, { "epoch": 0.73, "learning_rate": 2.6891205560010693e-08, "loss": 0.0638, "step": 2736 }, { "epoch": 0.73, "learning_rate": 2.6864474739374498e-08, "loss": 0.0646, "step": 2737 }, { "epoch": 0.73, "learning_rate": 2.6837743918738302e-08, "loss": 0.0716, "step": 2738 }, { "epoch": 0.73, "learning_rate": 2.681101309810211e-08, "loss": 0.0506, "step": 2739 }, { "epoch": 0.73, "learning_rate": 2.678428227746592e-08, "loss": 0.0788, "step": 2740 }, { "epoch": 0.73, "learning_rate": 2.6757551456829723e-08, "loss": 0.0724, "step": 2741 }, { "epoch": 0.73, "learning_rate": 2.6730820636193528e-08, "loss": 0.0971, "step": 2742 }, { "epoch": 0.73, "learning_rate": 2.6704089815557336e-08, "loss": 0.0634, "step": 2743 }, { "epoch": 0.73, "learning_rate": 2.6677358994921144e-08, "loss": 0.0691, "step": 2744 }, { "epoch": 0.73, "learning_rate": 2.665062817428495e-08, "loss": 0.0591, "step": 2745 }, { "epoch": 0.73, "learning_rate": 2.6623897353648757e-08, "loss": 0.0848, "step": 2746 }, { "epoch": 0.73, "learning_rate": 2.659716653301256e-08, "loss": 0.0599, "step": 2747 }, { "epoch": 0.73, "learning_rate": 2.657043571237637e-08, "loss": 0.0794, "step": 2748 }, { "epoch": 0.73, "learning_rate": 2.6543704891740178e-08, "loss": 0.0754, "step": 2749 }, { "epoch": 0.73, "learning_rate": 2.6516974071103982e-08, "loss": 0.0793, "step": 2750 }, { "epoch": 0.74, "learning_rate": 2.6490243250467787e-08, "loss": 0.079, "step": 2751 }, { "epoch": 0.74, "learning_rate": 2.64635124298316e-08, "loss": 0.072, "step": 2752 }, { "epoch": 0.74, "learning_rate": 2.6436781609195403e-08, "loss": 0.0766, "step": 2753 }, { "epoch": 0.74, "learning_rate": 2.6410050788559208e-08, "loss": 0.0885, "step": 2754 }, { "epoch": 0.74, "learning_rate": 2.6383319967923013e-08, "loss": 0.0809, "step": 2755 }, { "epoch": 0.74, "learning_rate": 2.6356589147286824e-08, "loss": 0.0782, "step": 2756 }, { "epoch": 0.74, "learning_rate": 2.632985832665063e-08, "loss": 0.0872, "step": 2757 }, { "epoch": 0.74, "learning_rate": 2.6303127506014434e-08, "loss": 0.1107, "step": 2758 }, { "epoch": 0.74, "learning_rate": 2.6276396685378238e-08, "loss": 0.0558, "step": 2759 }, { "epoch": 0.74, "learning_rate": 2.624966586474205e-08, "loss": 0.0578, "step": 2760 }, { "epoch": 0.74, "learning_rate": 2.6222935044105854e-08, "loss": 0.0677, "step": 2761 }, { "epoch": 0.74, "learning_rate": 2.619620422346966e-08, "loss": 0.0706, "step": 2762 }, { "epoch": 0.74, "learning_rate": 2.6169473402833464e-08, "loss": 0.0703, "step": 2763 }, { "epoch": 0.74, "learning_rate": 2.6142742582197275e-08, "loss": 0.073, "step": 2764 }, { "epoch": 0.74, "learning_rate": 2.611601176156108e-08, "loss": 0.0781, "step": 2765 }, { "epoch": 0.74, "learning_rate": 2.6089280940924885e-08, "loss": 0.09, "step": 2766 }, { "epoch": 0.74, "learning_rate": 2.606255012028869e-08, "loss": 0.1038, "step": 2767 }, { "epoch": 0.74, "learning_rate": 2.60358192996525e-08, "loss": 0.0846, "step": 2768 }, { "epoch": 0.74, "learning_rate": 2.6009088479016305e-08, "loss": 0.0731, "step": 2769 }, { "epoch": 0.74, "learning_rate": 2.598235765838011e-08, "loss": 0.0635, "step": 2770 }, { "epoch": 0.74, "learning_rate": 2.5955626837743915e-08, "loss": 0.0774, "step": 2771 }, { "epoch": 0.74, "learning_rate": 2.5928896017107726e-08, "loss": 0.083, "step": 2772 }, { "epoch": 0.74, "learning_rate": 2.590216519647153e-08, "loss": 0.0823, "step": 2773 }, { "epoch": 0.74, "learning_rate": 2.5875434375835336e-08, "loss": 0.0797, "step": 2774 }, { "epoch": 0.74, "learning_rate": 2.584870355519914e-08, "loss": 0.0651, "step": 2775 }, { "epoch": 0.74, "learning_rate": 2.5821972734562952e-08, "loss": 0.087, "step": 2776 }, { "epoch": 0.74, "learning_rate": 2.5795241913926757e-08, "loss": 0.0959, "step": 2777 }, { "epoch": 0.74, "learning_rate": 2.576851109329056e-08, "loss": 0.0665, "step": 2778 }, { "epoch": 0.74, "learning_rate": 2.5741780272654366e-08, "loss": 0.0909, "step": 2779 }, { "epoch": 0.74, "learning_rate": 2.5715049452018177e-08, "loss": 0.0783, "step": 2780 }, { "epoch": 0.74, "learning_rate": 2.5688318631381982e-08, "loss": 0.0719, "step": 2781 }, { "epoch": 0.74, "learning_rate": 2.5661587810745787e-08, "loss": 0.1025, "step": 2782 }, { "epoch": 0.74, "learning_rate": 2.5634856990109595e-08, "loss": 0.1081, "step": 2783 }, { "epoch": 0.74, "learning_rate": 2.5608126169473403e-08, "loss": 0.0788, "step": 2784 }, { "epoch": 0.74, "learning_rate": 2.5581395348837208e-08, "loss": 0.1034, "step": 2785 }, { "epoch": 0.74, "learning_rate": 2.5554664528201016e-08, "loss": 0.0598, "step": 2786 }, { "epoch": 0.74, "learning_rate": 2.552793370756482e-08, "loss": 0.0762, "step": 2787 }, { "epoch": 0.75, "learning_rate": 2.550120288692863e-08, "loss": 0.0676, "step": 2788 }, { "epoch": 0.75, "learning_rate": 2.5474472066292437e-08, "loss": 0.0881, "step": 2789 }, { "epoch": 0.75, "learning_rate": 2.544774124565624e-08, "loss": 0.06, "step": 2790 }, { "epoch": 0.75, "learning_rate": 2.5421010425020046e-08, "loss": 0.0716, "step": 2791 }, { "epoch": 0.75, "learning_rate": 2.5394279604383857e-08, "loss": 0.0706, "step": 2792 }, { "epoch": 0.75, "learning_rate": 2.5367548783747662e-08, "loss": 0.0588, "step": 2793 }, { "epoch": 0.75, "learning_rate": 2.5340817963111467e-08, "loss": 0.0507, "step": 2794 }, { "epoch": 0.75, "learning_rate": 2.5314087142475272e-08, "loss": 0.0785, "step": 2795 }, { "epoch": 0.75, "learning_rate": 2.5287356321839083e-08, "loss": 0.0754, "step": 2796 }, { "epoch": 0.75, "learning_rate": 2.5260625501202888e-08, "loss": 0.0662, "step": 2797 }, { "epoch": 0.75, "learning_rate": 2.5233894680566693e-08, "loss": 0.0635, "step": 2798 }, { "epoch": 0.75, "learning_rate": 2.5207163859930497e-08, "loss": 0.0519, "step": 2799 }, { "epoch": 0.75, "learning_rate": 2.518043303929431e-08, "loss": 0.0658, "step": 2800 }, { "epoch": 0.75, "learning_rate": 2.5153702218658113e-08, "loss": 0.093, "step": 2801 }, { "epoch": 0.75, "learning_rate": 2.5126971398021918e-08, "loss": 0.1008, "step": 2802 }, { "epoch": 0.75, "learning_rate": 2.5100240577385723e-08, "loss": 0.0632, "step": 2803 }, { "epoch": 0.75, "learning_rate": 2.5073509756749534e-08, "loss": 0.0927, "step": 2804 }, { "epoch": 0.75, "learning_rate": 2.504677893611334e-08, "loss": 0.0584, "step": 2805 }, { "epoch": 0.75, "learning_rate": 2.5020048115477144e-08, "loss": 0.0752, "step": 2806 }, { "epoch": 0.75, "learning_rate": 2.4993317294840952e-08, "loss": 0.0748, "step": 2807 }, { "epoch": 0.75, "learning_rate": 2.4966586474204756e-08, "loss": 0.0837, "step": 2808 }, { "epoch": 0.75, "learning_rate": 2.4939855653568564e-08, "loss": 0.0682, "step": 2809 }, { "epoch": 0.75, "learning_rate": 2.491312483293237e-08, "loss": 0.0674, "step": 2810 }, { "epoch": 0.75, "learning_rate": 2.4886394012296177e-08, "loss": 0.0751, "step": 2811 }, { "epoch": 0.75, "learning_rate": 2.4859663191659982e-08, "loss": 0.1195, "step": 2812 }, { "epoch": 0.75, "learning_rate": 2.483293237102379e-08, "loss": 0.0694, "step": 2813 }, { "epoch": 0.75, "learning_rate": 2.4806201550387595e-08, "loss": 0.0936, "step": 2814 }, { "epoch": 0.75, "learning_rate": 2.4779470729751403e-08, "loss": 0.0701, "step": 2815 }, { "epoch": 0.75, "learning_rate": 2.4752739909115208e-08, "loss": 0.0509, "step": 2816 }, { "epoch": 0.75, "learning_rate": 2.4726009088479016e-08, "loss": 0.0782, "step": 2817 }, { "epoch": 0.75, "learning_rate": 2.469927826784282e-08, "loss": 0.0822, "step": 2818 }, { "epoch": 0.75, "learning_rate": 2.467254744720663e-08, "loss": 0.0825, "step": 2819 }, { "epoch": 0.75, "learning_rate": 2.4645816626570433e-08, "loss": 0.0838, "step": 2820 }, { "epoch": 0.75, "learning_rate": 2.461908580593424e-08, "loss": 0.076, "step": 2821 }, { "epoch": 0.75, "learning_rate": 2.4592354985298046e-08, "loss": 0.1028, "step": 2822 }, { "epoch": 0.75, "learning_rate": 2.4565624164661854e-08, "loss": 0.083, "step": 2823 }, { "epoch": 0.75, "learning_rate": 2.453889334402566e-08, "loss": 0.0613, "step": 2824 }, { "epoch": 0.76, "learning_rate": 2.4512162523389467e-08, "loss": 0.0938, "step": 2825 }, { "epoch": 0.76, "learning_rate": 2.448543170275327e-08, "loss": 0.0668, "step": 2826 }, { "epoch": 0.76, "learning_rate": 2.445870088211708e-08, "loss": 0.1029, "step": 2827 }, { "epoch": 0.76, "learning_rate": 2.4431970061480888e-08, "loss": 0.088, "step": 2828 }, { "epoch": 0.76, "learning_rate": 2.4405239240844692e-08, "loss": 0.0935, "step": 2829 }, { "epoch": 0.76, "learning_rate": 2.43785084202085e-08, "loss": 0.0847, "step": 2830 }, { "epoch": 0.76, "learning_rate": 2.435177759957231e-08, "loss": 0.0742, "step": 2831 }, { "epoch": 0.76, "learning_rate": 2.4325046778936113e-08, "loss": 0.064, "step": 2832 }, { "epoch": 0.76, "learning_rate": 2.429831595829992e-08, "loss": 0.1087, "step": 2833 }, { "epoch": 0.76, "learning_rate": 2.4271585137663726e-08, "loss": 0.078, "step": 2834 }, { "epoch": 0.76, "learning_rate": 2.4244854317027534e-08, "loss": 0.0901, "step": 2835 }, { "epoch": 0.76, "learning_rate": 2.421812349639134e-08, "loss": 0.0987, "step": 2836 }, { "epoch": 0.76, "learning_rate": 2.4191392675755147e-08, "loss": 0.0585, "step": 2837 }, { "epoch": 0.76, "learning_rate": 2.416466185511895e-08, "loss": 0.0645, "step": 2838 }, { "epoch": 0.76, "learning_rate": 2.413793103448276e-08, "loss": 0.0748, "step": 2839 }, { "epoch": 0.76, "learning_rate": 2.4111200213846564e-08, "loss": 0.0647, "step": 2840 }, { "epoch": 0.76, "learning_rate": 2.4084469393210372e-08, "loss": 0.0658, "step": 2841 }, { "epoch": 0.76, "learning_rate": 2.4057738572574177e-08, "loss": 0.0589, "step": 2842 }, { "epoch": 0.76, "learning_rate": 2.4031007751937985e-08, "loss": 0.0658, "step": 2843 }, { "epoch": 0.76, "learning_rate": 2.400427693130179e-08, "loss": 0.0816, "step": 2844 }, { "epoch": 0.76, "learning_rate": 2.3977546110665598e-08, "loss": 0.0763, "step": 2845 }, { "epoch": 0.76, "learning_rate": 2.3950815290029403e-08, "loss": 0.0618, "step": 2846 }, { "epoch": 0.76, "learning_rate": 2.392408446939321e-08, "loss": 0.0775, "step": 2847 }, { "epoch": 0.76, "learning_rate": 2.3897353648757015e-08, "loss": 0.0692, "step": 2848 }, { "epoch": 0.76, "learning_rate": 2.3870622828120824e-08, "loss": 0.0925, "step": 2849 }, { "epoch": 0.76, "learning_rate": 2.3843892007484628e-08, "loss": 0.0891, "step": 2850 }, { "epoch": 0.76, "learning_rate": 2.3817161186848436e-08, "loss": 0.0709, "step": 2851 }, { "epoch": 0.76, "learning_rate": 2.379043036621224e-08, "loss": 0.0915, "step": 2852 }, { "epoch": 0.76, "learning_rate": 2.376369954557605e-08, "loss": 0.0768, "step": 2853 }, { "epoch": 0.76, "learning_rate": 2.3736968724939854e-08, "loss": 0.0663, "step": 2854 }, { "epoch": 0.76, "learning_rate": 2.3710237904303662e-08, "loss": 0.0731, "step": 2855 }, { "epoch": 0.76, "learning_rate": 2.3683507083667467e-08, "loss": 0.0659, "step": 2856 }, { "epoch": 0.76, "learning_rate": 2.3656776263031275e-08, "loss": 0.0796, "step": 2857 }, { "epoch": 0.76, "learning_rate": 2.363004544239508e-08, "loss": 0.0725, "step": 2858 }, { "epoch": 0.76, "learning_rate": 2.3603314621758887e-08, "loss": 0.0765, "step": 2859 }, { "epoch": 0.76, "learning_rate": 2.3576583801122692e-08, "loss": 0.0908, "step": 2860 }, { "epoch": 0.76, "learning_rate": 2.35498529804865e-08, "loss": 0.0711, "step": 2861 }, { "epoch": 0.76, "learning_rate": 2.3523122159850305e-08, "loss": 0.0721, "step": 2862 }, { "epoch": 0.77, "learning_rate": 2.3496391339214113e-08, "loss": 0.0611, "step": 2863 }, { "epoch": 0.77, "learning_rate": 2.3469660518577918e-08, "loss": 0.0795, "step": 2864 }, { "epoch": 0.77, "learning_rate": 2.3442929697941726e-08, "loss": 0.0745, "step": 2865 }, { "epoch": 0.77, "learning_rate": 2.341619887730553e-08, "loss": 0.0713, "step": 2866 }, { "epoch": 0.77, "learning_rate": 2.338946805666934e-08, "loss": 0.0649, "step": 2867 }, { "epoch": 0.77, "learning_rate": 2.3362737236033143e-08, "loss": 0.0623, "step": 2868 }, { "epoch": 0.77, "learning_rate": 2.333600641539695e-08, "loss": 0.0734, "step": 2869 }, { "epoch": 0.77, "learning_rate": 2.330927559476076e-08, "loss": 0.0779, "step": 2870 }, { "epoch": 0.77, "learning_rate": 2.3282544774124564e-08, "loss": 0.0659, "step": 2871 }, { "epoch": 0.77, "learning_rate": 2.3255813953488372e-08, "loss": 0.0851, "step": 2872 }, { "epoch": 0.77, "learning_rate": 2.3229083132852177e-08, "loss": 0.0583, "step": 2873 }, { "epoch": 0.77, "learning_rate": 2.3202352312215985e-08, "loss": 0.0994, "step": 2874 }, { "epoch": 0.77, "learning_rate": 2.3175621491579793e-08, "loss": 0.0735, "step": 2875 }, { "epoch": 0.77, "learning_rate": 2.3148890670943598e-08, "loss": 0.0938, "step": 2876 }, { "epoch": 0.77, "learning_rate": 2.3122159850307406e-08, "loss": 0.082, "step": 2877 }, { "epoch": 0.77, "learning_rate": 2.309542902967121e-08, "loss": 0.0609, "step": 2878 }, { "epoch": 0.77, "learning_rate": 2.306869820903502e-08, "loss": 0.0668, "step": 2879 }, { "epoch": 0.77, "learning_rate": 2.3041967388398823e-08, "loss": 0.0619, "step": 2880 }, { "epoch": 0.77, "learning_rate": 2.301523656776263e-08, "loss": 0.0829, "step": 2881 }, { "epoch": 0.77, "learning_rate": 2.2988505747126436e-08, "loss": 0.0662, "step": 2882 }, { "epoch": 0.77, "learning_rate": 2.2961774926490244e-08, "loss": 0.0477, "step": 2883 }, { "epoch": 0.77, "learning_rate": 2.293504410585405e-08, "loss": 0.0549, "step": 2884 }, { "epoch": 0.77, "learning_rate": 2.2908313285217857e-08, "loss": 0.0833, "step": 2885 }, { "epoch": 0.77, "learning_rate": 2.2881582464581662e-08, "loss": 0.0788, "step": 2886 }, { "epoch": 0.77, "learning_rate": 2.285485164394547e-08, "loss": 0.069, "step": 2887 }, { "epoch": 0.77, "learning_rate": 2.2828120823309274e-08, "loss": 0.0847, "step": 2888 }, { "epoch": 0.77, "learning_rate": 2.2801390002673083e-08, "loss": 0.0765, "step": 2889 }, { "epoch": 0.77, "learning_rate": 2.2774659182036887e-08, "loss": 0.0839, "step": 2890 }, { "epoch": 0.77, "learning_rate": 2.2747928361400695e-08, "loss": 0.0599, "step": 2891 }, { "epoch": 0.77, "learning_rate": 2.27211975407645e-08, "loss": 0.0741, "step": 2892 }, { "epoch": 0.77, "learning_rate": 2.2694466720128308e-08, "loss": 0.0656, "step": 2893 }, { "epoch": 0.77, "learning_rate": 2.2667735899492113e-08, "loss": 0.0753, "step": 2894 }, { "epoch": 0.77, "learning_rate": 2.264100507885592e-08, "loss": 0.0875, "step": 2895 }, { "epoch": 0.77, "learning_rate": 2.2614274258219726e-08, "loss": 0.0941, "step": 2896 }, { "epoch": 0.77, "learning_rate": 2.2587543437583534e-08, "loss": 0.0672, "step": 2897 }, { "epoch": 0.77, "learning_rate": 2.256081261694734e-08, "loss": 0.076, "step": 2898 }, { "epoch": 0.77, "learning_rate": 2.2534081796311146e-08, "loss": 0.0666, "step": 2899 }, { "epoch": 0.78, "learning_rate": 2.250735097567495e-08, "loss": 0.0844, "step": 2900 }, { "epoch": 0.78, "learning_rate": 2.248062015503876e-08, "loss": 0.0535, "step": 2901 }, { "epoch": 0.78, "learning_rate": 2.2453889334402564e-08, "loss": 0.061, "step": 2902 }, { "epoch": 0.78, "learning_rate": 2.2427158513766372e-08, "loss": 0.0928, "step": 2903 }, { "epoch": 0.78, "learning_rate": 2.2400427693130177e-08, "loss": 0.0664, "step": 2904 }, { "epoch": 0.78, "learning_rate": 2.2373696872493985e-08, "loss": 0.067, "step": 2905 }, { "epoch": 0.78, "learning_rate": 2.234696605185779e-08, "loss": 0.0714, "step": 2906 }, { "epoch": 0.78, "learning_rate": 2.2320235231221598e-08, "loss": 0.0706, "step": 2907 }, { "epoch": 0.78, "learning_rate": 2.2293504410585402e-08, "loss": 0.0782, "step": 2908 }, { "epoch": 0.78, "learning_rate": 2.226677358994921e-08, "loss": 0.0482, "step": 2909 }, { "epoch": 0.78, "learning_rate": 2.2240042769313015e-08, "loss": 0.0884, "step": 2910 }, { "epoch": 0.78, "learning_rate": 2.2213311948676823e-08, "loss": 0.0727, "step": 2911 }, { "epoch": 0.78, "learning_rate": 2.2186581128040628e-08, "loss": 0.0702, "step": 2912 }, { "epoch": 0.78, "learning_rate": 2.2159850307404436e-08, "loss": 0.069, "step": 2913 }, { "epoch": 0.78, "learning_rate": 2.2133119486768244e-08, "loss": 0.0959, "step": 2914 }, { "epoch": 0.78, "learning_rate": 2.210638866613205e-08, "loss": 0.1036, "step": 2915 }, { "epoch": 0.78, "learning_rate": 2.2079657845495857e-08, "loss": 0.0847, "step": 2916 }, { "epoch": 0.78, "learning_rate": 2.2052927024859665e-08, "loss": 0.0659, "step": 2917 }, { "epoch": 0.78, "learning_rate": 2.202619620422347e-08, "loss": 0.0795, "step": 2918 }, { "epoch": 0.78, "learning_rate": 2.1999465383587278e-08, "loss": 0.0692, "step": 2919 }, { "epoch": 0.78, "learning_rate": 2.1972734562951082e-08, "loss": 0.0847, "step": 2920 }, { "epoch": 0.78, "learning_rate": 2.194600374231489e-08, "loss": 0.0557, "step": 2921 }, { "epoch": 0.78, "learning_rate": 2.1919272921678695e-08, "loss": 0.0777, "step": 2922 }, { "epoch": 0.78, "learning_rate": 2.1892542101042503e-08, "loss": 0.0901, "step": 2923 }, { "epoch": 0.78, "learning_rate": 2.1865811280406308e-08, "loss": 0.0932, "step": 2924 }, { "epoch": 0.78, "learning_rate": 2.1839080459770116e-08, "loss": 0.097, "step": 2925 }, { "epoch": 0.78, "learning_rate": 2.181234963913392e-08, "loss": 0.0662, "step": 2926 }, { "epoch": 0.78, "learning_rate": 2.178561881849773e-08, "loss": 0.063, "step": 2927 }, { "epoch": 0.78, "learning_rate": 2.1758887997861533e-08, "loss": 0.0618, "step": 2928 }, { "epoch": 0.78, "learning_rate": 2.173215717722534e-08, "loss": 0.069, "step": 2929 }, { "epoch": 0.78, "learning_rate": 2.1705426356589146e-08, "loss": 0.0805, "step": 2930 }, { "epoch": 0.78, "learning_rate": 2.1678695535952954e-08, "loss": 0.0665, "step": 2931 }, { "epoch": 0.78, "learning_rate": 2.165196471531676e-08, "loss": 0.0685, "step": 2932 }, { "epoch": 0.78, "learning_rate": 2.1625233894680567e-08, "loss": 0.0992, "step": 2933 }, { "epoch": 0.78, "learning_rate": 2.1598503074044372e-08, "loss": 0.0779, "step": 2934 }, { "epoch": 0.78, "learning_rate": 2.157177225340818e-08, "loss": 0.0858, "step": 2935 }, { "epoch": 0.78, "learning_rate": 2.1545041432771985e-08, "loss": 0.0851, "step": 2936 }, { "epoch": 0.78, "learning_rate": 2.1518310612135793e-08, "loss": 0.0627, "step": 2937 }, { "epoch": 0.79, "learning_rate": 2.1491579791499597e-08, "loss": 0.0761, "step": 2938 }, { "epoch": 0.79, "learning_rate": 2.1464848970863405e-08, "loss": 0.0745, "step": 2939 }, { "epoch": 0.79, "learning_rate": 2.143811815022721e-08, "loss": 0.0912, "step": 2940 }, { "epoch": 0.79, "learning_rate": 2.1411387329591018e-08, "loss": 0.0835, "step": 2941 }, { "epoch": 0.79, "learning_rate": 2.1384656508954823e-08, "loss": 0.0807, "step": 2942 }, { "epoch": 0.79, "learning_rate": 2.135792568831863e-08, "loss": 0.0749, "step": 2943 }, { "epoch": 0.79, "learning_rate": 2.1331194867682436e-08, "loss": 0.0661, "step": 2944 }, { "epoch": 0.79, "learning_rate": 2.1304464047046244e-08, "loss": 0.0885, "step": 2945 }, { "epoch": 0.79, "learning_rate": 2.127773322641005e-08, "loss": 0.1206, "step": 2946 }, { "epoch": 0.79, "learning_rate": 2.1251002405773857e-08, "loss": 0.0971, "step": 2947 }, { "epoch": 0.79, "learning_rate": 2.122427158513766e-08, "loss": 0.0989, "step": 2948 }, { "epoch": 0.79, "learning_rate": 2.119754076450147e-08, "loss": 0.0855, "step": 2949 }, { "epoch": 0.79, "learning_rate": 2.1170809943865274e-08, "loss": 0.0798, "step": 2950 }, { "epoch": 0.79, "learning_rate": 2.1144079123229082e-08, "loss": 0.0747, "step": 2951 }, { "epoch": 0.79, "learning_rate": 2.1117348302592887e-08, "loss": 0.0501, "step": 2952 }, { "epoch": 0.79, "learning_rate": 2.1090617481956695e-08, "loss": 0.0577, "step": 2953 }, { "epoch": 0.79, "learning_rate": 2.10638866613205e-08, "loss": 0.0828, "step": 2954 }, { "epoch": 0.79, "learning_rate": 2.1037155840684308e-08, "loss": 0.063, "step": 2955 }, { "epoch": 0.79, "learning_rate": 2.1010425020048116e-08, "loss": 0.053, "step": 2956 }, { "epoch": 0.79, "learning_rate": 2.098369419941192e-08, "loss": 0.0954, "step": 2957 }, { "epoch": 0.79, "learning_rate": 2.095696337877573e-08, "loss": 0.074, "step": 2958 }, { "epoch": 0.79, "learning_rate": 2.0930232558139537e-08, "loss": 0.0774, "step": 2959 }, { "epoch": 0.79, "learning_rate": 2.090350173750334e-08, "loss": 0.0671, "step": 2960 }, { "epoch": 0.79, "learning_rate": 2.087677091686715e-08, "loss": 0.0718, "step": 2961 }, { "epoch": 0.79, "learning_rate": 2.0850040096230954e-08, "loss": 0.0858, "step": 2962 }, { "epoch": 0.79, "learning_rate": 2.0823309275594762e-08, "loss": 0.0577, "step": 2963 }, { "epoch": 0.79, "learning_rate": 2.0796578454958567e-08, "loss": 0.0658, "step": 2964 }, { "epoch": 0.79, "learning_rate": 2.0769847634322375e-08, "loss": 0.0685, "step": 2965 }, { "epoch": 0.79, "learning_rate": 2.074311681368618e-08, "loss": 0.0748, "step": 2966 }, { "epoch": 0.79, "learning_rate": 2.0716385993049988e-08, "loss": 0.0755, "step": 2967 }, { "epoch": 0.79, "learning_rate": 2.0689655172413793e-08, "loss": 0.095, "step": 2968 }, { "epoch": 0.79, "learning_rate": 2.06629243517776e-08, "loss": 0.061, "step": 2969 }, { "epoch": 0.79, "learning_rate": 2.0636193531141405e-08, "loss": 0.0937, "step": 2970 }, { "epoch": 0.79, "learning_rate": 2.0609462710505213e-08, "loss": 0.0784, "step": 2971 }, { "epoch": 0.79, "learning_rate": 2.0582731889869018e-08, "loss": 0.0673, "step": 2972 }, { "epoch": 0.79, "learning_rate": 2.0556001069232826e-08, "loss": 0.0639, "step": 2973 }, { "epoch": 0.79, "learning_rate": 2.052927024859663e-08, "loss": 0.104, "step": 2974 }, { "epoch": 0.8, "learning_rate": 2.050253942796044e-08, "loss": 0.0833, "step": 2975 }, { "epoch": 0.8, "learning_rate": 2.0475808607324244e-08, "loss": 0.0654, "step": 2976 }, { "epoch": 0.8, "learning_rate": 2.0449077786688052e-08, "loss": 0.0814, "step": 2977 }, { "epoch": 0.8, "learning_rate": 2.0422346966051856e-08, "loss": 0.0662, "step": 2978 }, { "epoch": 0.8, "learning_rate": 2.0395616145415664e-08, "loss": 0.0753, "step": 2979 }, { "epoch": 0.8, "learning_rate": 2.036888532477947e-08, "loss": 0.0914, "step": 2980 }, { "epoch": 0.8, "learning_rate": 2.0342154504143277e-08, "loss": 0.0716, "step": 2981 }, { "epoch": 0.8, "learning_rate": 2.0315423683507082e-08, "loss": 0.0714, "step": 2982 }, { "epoch": 0.8, "learning_rate": 2.028869286287089e-08, "loss": 0.0535, "step": 2983 }, { "epoch": 0.8, "learning_rate": 2.0261962042234695e-08, "loss": 0.0912, "step": 2984 }, { "epoch": 0.8, "learning_rate": 2.0235231221598503e-08, "loss": 0.0579, "step": 2985 }, { "epoch": 0.8, "learning_rate": 2.0208500400962308e-08, "loss": 0.0942, "step": 2986 }, { "epoch": 0.8, "learning_rate": 2.0181769580326116e-08, "loss": 0.0833, "step": 2987 }, { "epoch": 0.8, "learning_rate": 2.015503875968992e-08, "loss": 0.0767, "step": 2988 }, { "epoch": 0.8, "learning_rate": 2.012830793905373e-08, "loss": 0.0634, "step": 2989 }, { "epoch": 0.8, "learning_rate": 2.0101577118417533e-08, "loss": 0.0819, "step": 2990 }, { "epoch": 0.8, "learning_rate": 2.007484629778134e-08, "loss": 0.0569, "step": 2991 }, { "epoch": 0.8, "learning_rate": 2.0048115477145146e-08, "loss": 0.0994, "step": 2992 }, { "epoch": 0.8, "learning_rate": 2.0021384656508954e-08, "loss": 0.0624, "step": 2993 }, { "epoch": 0.8, "learning_rate": 1.999465383587276e-08, "loss": 0.0675, "step": 2994 }, { "epoch": 0.8, "learning_rate": 1.9967923015236567e-08, "loss": 0.0609, "step": 2995 }, { "epoch": 0.8, "learning_rate": 1.994119219460037e-08, "loss": 0.064, "step": 2996 }, { "epoch": 0.8, "learning_rate": 1.991446137396418e-08, "loss": 0.057, "step": 2997 }, { "epoch": 0.8, "learning_rate": 1.9887730553327988e-08, "loss": 0.0616, "step": 2998 }, { "epoch": 0.8, "learning_rate": 1.9860999732691792e-08, "loss": 0.0847, "step": 2999 }, { "epoch": 0.8, "learning_rate": 1.98342689120556e-08, "loss": 0.0755, "step": 3000 }, { "epoch": 0.8, "learning_rate": 1.980753809141941e-08, "loss": 0.0834, "step": 3001 }, { "epoch": 0.8, "learning_rate": 1.9780807270783213e-08, "loss": 0.066, "step": 3002 }, { "epoch": 0.8, "learning_rate": 1.975407645014702e-08, "loss": 0.0707, "step": 3003 }, { "epoch": 0.8, "learning_rate": 1.9727345629510826e-08, "loss": 0.0705, "step": 3004 }, { "epoch": 0.8, "learning_rate": 1.9700614808874634e-08, "loss": 0.0787, "step": 3005 }, { "epoch": 0.8, "learning_rate": 1.967388398823844e-08, "loss": 0.0562, "step": 3006 }, { "epoch": 0.8, "learning_rate": 1.9647153167602247e-08, "loss": 0.087, "step": 3007 }, { "epoch": 0.8, "learning_rate": 1.962042234696605e-08, "loss": 0.0917, "step": 3008 }, { "epoch": 0.8, "learning_rate": 1.959369152632986e-08, "loss": 0.0689, "step": 3009 }, { "epoch": 0.8, "learning_rate": 1.9566960705693664e-08, "loss": 0.0601, "step": 3010 }, { "epoch": 0.8, "learning_rate": 1.9540229885057472e-08, "loss": 0.0694, "step": 3011 }, { "epoch": 0.81, "learning_rate": 1.9513499064421277e-08, "loss": 0.0712, "step": 3012 }, { "epoch": 0.81, "learning_rate": 1.9486768243785085e-08, "loss": 0.0414, "step": 3013 }, { "epoch": 0.81, "learning_rate": 1.946003742314889e-08, "loss": 0.0764, "step": 3014 }, { "epoch": 0.81, "learning_rate": 1.9433306602512698e-08, "loss": 0.0731, "step": 3015 }, { "epoch": 0.81, "learning_rate": 1.9406575781876503e-08, "loss": 0.0788, "step": 3016 }, { "epoch": 0.81, "learning_rate": 1.937984496124031e-08, "loss": 0.0928, "step": 3017 }, { "epoch": 0.81, "learning_rate": 1.9353114140604115e-08, "loss": 0.0881, "step": 3018 }, { "epoch": 0.81, "learning_rate": 1.9326383319967923e-08, "loss": 0.0848, "step": 3019 }, { "epoch": 0.81, "learning_rate": 1.9299652499331728e-08, "loss": 0.0664, "step": 3020 }, { "epoch": 0.81, "learning_rate": 1.9272921678695536e-08, "loss": 0.0796, "step": 3021 }, { "epoch": 0.81, "learning_rate": 1.924619085805934e-08, "loss": 0.0546, "step": 3022 }, { "epoch": 0.81, "learning_rate": 1.921946003742315e-08, "loss": 0.0968, "step": 3023 }, { "epoch": 0.81, "learning_rate": 1.9192729216786954e-08, "loss": 0.0622, "step": 3024 }, { "epoch": 0.81, "learning_rate": 1.9165998396150762e-08, "loss": 0.0712, "step": 3025 }, { "epoch": 0.81, "learning_rate": 1.9139267575514567e-08, "loss": 0.0604, "step": 3026 }, { "epoch": 0.81, "learning_rate": 1.9112536754878375e-08, "loss": 0.0765, "step": 3027 }, { "epoch": 0.81, "learning_rate": 1.908580593424218e-08, "loss": 0.0906, "step": 3028 }, { "epoch": 0.81, "learning_rate": 1.9059075113605987e-08, "loss": 0.077, "step": 3029 }, { "epoch": 0.81, "learning_rate": 1.9032344292969792e-08, "loss": 0.0701, "step": 3030 }, { "epoch": 0.81, "learning_rate": 1.90056134723336e-08, "loss": 0.0733, "step": 3031 }, { "epoch": 0.81, "learning_rate": 1.8978882651697405e-08, "loss": 0.0651, "step": 3032 }, { "epoch": 0.81, "learning_rate": 1.8952151831061213e-08, "loss": 0.0891, "step": 3033 }, { "epoch": 0.81, "learning_rate": 1.8925421010425018e-08, "loss": 0.0816, "step": 3034 }, { "epoch": 0.81, "learning_rate": 1.8898690189788826e-08, "loss": 0.0988, "step": 3035 }, { "epoch": 0.81, "learning_rate": 1.887195936915263e-08, "loss": 0.0628, "step": 3036 }, { "epoch": 0.81, "learning_rate": 1.884522854851644e-08, "loss": 0.1031, "step": 3037 }, { "epoch": 0.81, "learning_rate": 1.8818497727880243e-08, "loss": 0.0624, "step": 3038 }, { "epoch": 0.81, "learning_rate": 1.879176690724405e-08, "loss": 0.0836, "step": 3039 }, { "epoch": 0.81, "learning_rate": 1.876503608660786e-08, "loss": 0.0718, "step": 3040 }, { "epoch": 0.81, "learning_rate": 1.8738305265971664e-08, "loss": 0.0699, "step": 3041 }, { "epoch": 0.81, "learning_rate": 1.8711574445335472e-08, "loss": 0.0808, "step": 3042 }, { "epoch": 0.81, "learning_rate": 1.8684843624699277e-08, "loss": 0.0669, "step": 3043 }, { "epoch": 0.81, "learning_rate": 1.8658112804063085e-08, "loss": 0.065, "step": 3044 }, { "epoch": 0.81, "learning_rate": 1.8631381983426893e-08, "loss": 0.0779, "step": 3045 }, { "epoch": 0.81, "learning_rate": 1.8604651162790698e-08, "loss": 0.0829, "step": 3046 }, { "epoch": 0.81, "learning_rate": 1.8577920342154506e-08, "loss": 0.0547, "step": 3047 }, { "epoch": 0.81, "learning_rate": 1.855118952151831e-08, "loss": 0.1116, "step": 3048 }, { "epoch": 0.81, "learning_rate": 1.852445870088212e-08, "loss": 0.0813, "step": 3049 }, { "epoch": 0.82, "learning_rate": 1.8497727880245923e-08, "loss": 0.0738, "step": 3050 }, { "epoch": 0.82, "learning_rate": 1.847099705960973e-08, "loss": 0.098, "step": 3051 }, { "epoch": 0.82, "learning_rate": 1.8444266238973536e-08, "loss": 0.0684, "step": 3052 }, { "epoch": 0.82, "learning_rate": 1.8417535418337344e-08, "loss": 0.0734, "step": 3053 }, { "epoch": 0.82, "learning_rate": 1.839080459770115e-08, "loss": 0.0946, "step": 3054 }, { "epoch": 0.82, "learning_rate": 1.8364073777064957e-08, "loss": 0.0625, "step": 3055 }, { "epoch": 0.82, "learning_rate": 1.8337342956428762e-08, "loss": 0.0724, "step": 3056 }, { "epoch": 0.82, "learning_rate": 1.831061213579257e-08, "loss": 0.0819, "step": 3057 }, { "epoch": 0.82, "learning_rate": 1.8283881315156374e-08, "loss": 0.0621, "step": 3058 }, { "epoch": 0.82, "learning_rate": 1.8257150494520183e-08, "loss": 0.0688, "step": 3059 }, { "epoch": 0.82, "learning_rate": 1.8230419673883987e-08, "loss": 0.0751, "step": 3060 }, { "epoch": 0.82, "learning_rate": 1.8203688853247795e-08, "loss": 0.0636, "step": 3061 }, { "epoch": 0.82, "learning_rate": 1.81769580326116e-08, "loss": 0.0677, "step": 3062 }, { "epoch": 0.82, "learning_rate": 1.8150227211975408e-08, "loss": 0.0993, "step": 3063 }, { "epoch": 0.82, "learning_rate": 1.8123496391339213e-08, "loss": 0.0733, "step": 3064 }, { "epoch": 0.82, "learning_rate": 1.809676557070302e-08, "loss": 0.0771, "step": 3065 }, { "epoch": 0.82, "learning_rate": 1.8070034750066826e-08, "loss": 0.0826, "step": 3066 }, { "epoch": 0.82, "learning_rate": 1.8043303929430634e-08, "loss": 0.0787, "step": 3067 }, { "epoch": 0.82, "learning_rate": 1.801657310879444e-08, "loss": 0.0922, "step": 3068 }, { "epoch": 0.82, "learning_rate": 1.7989842288158246e-08, "loss": 0.068, "step": 3069 }, { "epoch": 0.82, "learning_rate": 1.796311146752205e-08, "loss": 0.1025, "step": 3070 }, { "epoch": 0.82, "learning_rate": 1.793638064688586e-08, "loss": 0.3658, "step": 3071 }, { "epoch": 0.82, "learning_rate": 1.7909649826249664e-08, "loss": 0.0808, "step": 3072 }, { "epoch": 0.82, "learning_rate": 1.7882919005613472e-08, "loss": 0.0586, "step": 3073 }, { "epoch": 0.82, "learning_rate": 1.7856188184977277e-08, "loss": 0.0826, "step": 3074 }, { "epoch": 0.82, "learning_rate": 1.7829457364341085e-08, "loss": 0.0902, "step": 3075 }, { "epoch": 0.82, "learning_rate": 1.780272654370489e-08, "loss": 0.0703, "step": 3076 }, { "epoch": 0.82, "learning_rate": 1.7775995723068698e-08, "loss": 0.069, "step": 3077 }, { "epoch": 0.82, "learning_rate": 1.7749264902432502e-08, "loss": 0.0885, "step": 3078 }, { "epoch": 0.82, "learning_rate": 1.772253408179631e-08, "loss": 0.057, "step": 3079 }, { "epoch": 0.82, "learning_rate": 1.7695803261160115e-08, "loss": 0.0802, "step": 3080 }, { "epoch": 0.82, "learning_rate": 1.7669072440523923e-08, "loss": 0.0784, "step": 3081 }, { "epoch": 0.82, "learning_rate": 1.7642341619887728e-08, "loss": 0.0718, "step": 3082 }, { "epoch": 0.82, "learning_rate": 1.7615610799251536e-08, "loss": 0.0596, "step": 3083 }, { "epoch": 0.82, "learning_rate": 1.7588879978615344e-08, "loss": 0.0595, "step": 3084 }, { "epoch": 0.82, "learning_rate": 1.756214915797915e-08, "loss": 0.0885, "step": 3085 }, { "epoch": 0.82, "learning_rate": 1.7535418337342957e-08, "loss": 0.0767, "step": 3086 }, { "epoch": 0.83, "learning_rate": 1.7508687516706765e-08, "loss": 0.079, "step": 3087 }, { "epoch": 0.83, "learning_rate": 1.748195669607057e-08, "loss": 0.0842, "step": 3088 }, { "epoch": 0.83, "learning_rate": 1.7455225875434378e-08, "loss": 0.0748, "step": 3089 }, { "epoch": 0.83, "learning_rate": 1.7428495054798182e-08, "loss": 0.0603, "step": 3090 }, { "epoch": 0.83, "learning_rate": 1.740176423416199e-08, "loss": 0.0604, "step": 3091 }, { "epoch": 0.83, "learning_rate": 1.7375033413525795e-08, "loss": 0.0728, "step": 3092 }, { "epoch": 0.83, "learning_rate": 1.7348302592889603e-08, "loss": 0.0745, "step": 3093 }, { "epoch": 0.83, "learning_rate": 1.7321571772253408e-08, "loss": 0.0932, "step": 3094 }, { "epoch": 0.83, "learning_rate": 1.7294840951617216e-08, "loss": 0.0729, "step": 3095 }, { "epoch": 0.83, "learning_rate": 1.726811013098102e-08, "loss": 0.0706, "step": 3096 }, { "epoch": 0.83, "learning_rate": 1.724137931034483e-08, "loss": 0.0908, "step": 3097 }, { "epoch": 0.83, "learning_rate": 1.7214648489708633e-08, "loss": 0.0653, "step": 3098 }, { "epoch": 0.83, "learning_rate": 1.718791766907244e-08, "loss": 0.0625, "step": 3099 }, { "epoch": 0.83, "learning_rate": 1.7161186848436246e-08, "loss": 0.0655, "step": 3100 }, { "epoch": 0.83, "learning_rate": 1.7134456027800054e-08, "loss": 0.07, "step": 3101 }, { "epoch": 0.83, "learning_rate": 1.710772520716386e-08, "loss": 0.103, "step": 3102 }, { "epoch": 0.83, "learning_rate": 1.7080994386527667e-08, "loss": 0.0873, "step": 3103 }, { "epoch": 0.83, "learning_rate": 1.7054263565891472e-08, "loss": 0.068, "step": 3104 }, { "epoch": 0.83, "learning_rate": 1.702753274525528e-08, "loss": 0.0703, "step": 3105 }, { "epoch": 0.83, "learning_rate": 1.7000801924619085e-08, "loss": 0.0939, "step": 3106 }, { "epoch": 0.83, "learning_rate": 1.6974071103982893e-08, "loss": 0.0776, "step": 3107 }, { "epoch": 0.83, "learning_rate": 1.6947340283346697e-08, "loss": 0.063, "step": 3108 }, { "epoch": 0.83, "learning_rate": 1.6920609462710505e-08, "loss": 0.1031, "step": 3109 }, { "epoch": 0.83, "learning_rate": 1.689387864207431e-08, "loss": 0.0921, "step": 3110 }, { "epoch": 0.83, "learning_rate": 1.6867147821438118e-08, "loss": 0.0717, "step": 3111 }, { "epoch": 0.83, "learning_rate": 1.6840417000801923e-08, "loss": 0.0746, "step": 3112 }, { "epoch": 0.83, "learning_rate": 1.681368618016573e-08, "loss": 0.075, "step": 3113 }, { "epoch": 0.83, "learning_rate": 1.6786955359529536e-08, "loss": 0.083, "step": 3114 }, { "epoch": 0.83, "learning_rate": 1.6760224538893344e-08, "loss": 0.0656, "step": 3115 }, { "epoch": 0.83, "learning_rate": 1.673349371825715e-08, "loss": 0.0886, "step": 3116 }, { "epoch": 0.83, "learning_rate": 1.6706762897620957e-08, "loss": 0.0706, "step": 3117 }, { "epoch": 0.83, "learning_rate": 1.668003207698476e-08, "loss": 0.1026, "step": 3118 }, { "epoch": 0.83, "learning_rate": 1.665330125634857e-08, "loss": 0.0845, "step": 3119 }, { "epoch": 0.83, "learning_rate": 1.6626570435712374e-08, "loss": 0.0845, "step": 3120 }, { "epoch": 0.83, "learning_rate": 1.6599839615076182e-08, "loss": 0.1157, "step": 3121 }, { "epoch": 0.83, "learning_rate": 1.6573108794439987e-08, "loss": 0.0781, "step": 3122 }, { "epoch": 0.83, "learning_rate": 1.6546377973803795e-08, "loss": 0.0744, "step": 3123 }, { "epoch": 0.83, "learning_rate": 1.65196471531676e-08, "loss": 0.0659, "step": 3124 }, { "epoch": 0.84, "learning_rate": 1.6492916332531408e-08, "loss": 0.0729, "step": 3125 }, { "epoch": 0.84, "learning_rate": 1.6466185511895216e-08, "loss": 0.0876, "step": 3126 }, { "epoch": 0.84, "learning_rate": 1.643945469125902e-08, "loss": 0.0804, "step": 3127 }, { "epoch": 0.84, "learning_rate": 1.641272387062283e-08, "loss": 0.0847, "step": 3128 }, { "epoch": 0.84, "learning_rate": 1.6385993049986637e-08, "loss": 0.084, "step": 3129 }, { "epoch": 0.84, "learning_rate": 1.635926222935044e-08, "loss": 0.0566, "step": 3130 }, { "epoch": 0.84, "learning_rate": 1.633253140871425e-08, "loss": 0.0663, "step": 3131 }, { "epoch": 0.84, "learning_rate": 1.6305800588078054e-08, "loss": 0.0759, "step": 3132 }, { "epoch": 0.84, "learning_rate": 1.6279069767441862e-08, "loss": 0.0925, "step": 3133 }, { "epoch": 0.84, "learning_rate": 1.6252338946805667e-08, "loss": 0.0606, "step": 3134 }, { "epoch": 0.84, "learning_rate": 1.6225608126169475e-08, "loss": 0.0863, "step": 3135 }, { "epoch": 0.84, "learning_rate": 1.619887730553328e-08, "loss": 0.0705, "step": 3136 }, { "epoch": 0.84, "learning_rate": 1.6172146484897088e-08, "loss": 0.088, "step": 3137 }, { "epoch": 0.84, "learning_rate": 1.6145415664260892e-08, "loss": 0.0863, "step": 3138 }, { "epoch": 0.84, "learning_rate": 1.61186848436247e-08, "loss": 0.0979, "step": 3139 }, { "epoch": 0.84, "learning_rate": 1.6091954022988505e-08, "loss": 0.057, "step": 3140 }, { "epoch": 0.84, "learning_rate": 1.6065223202352313e-08, "loss": 0.101, "step": 3141 }, { "epoch": 0.84, "learning_rate": 1.6038492381716118e-08, "loss": 0.0651, "step": 3142 }, { "epoch": 0.84, "learning_rate": 1.6011761561079926e-08, "loss": 0.0676, "step": 3143 }, { "epoch": 0.84, "learning_rate": 1.598503074044373e-08, "loss": 0.1006, "step": 3144 }, { "epoch": 0.84, "learning_rate": 1.595829991980754e-08, "loss": 0.0763, "step": 3145 }, { "epoch": 0.84, "learning_rate": 1.5931569099171344e-08, "loss": 0.0675, "step": 3146 }, { "epoch": 0.84, "learning_rate": 1.5904838278535152e-08, "loss": 0.0942, "step": 3147 }, { "epoch": 0.84, "learning_rate": 1.5878107457898956e-08, "loss": 0.078, "step": 3148 }, { "epoch": 0.84, "learning_rate": 1.5851376637262764e-08, "loss": 0.0796, "step": 3149 }, { "epoch": 0.84, "learning_rate": 1.582464581662657e-08, "loss": 0.0728, "step": 3150 }, { "epoch": 0.84, "learning_rate": 1.5797914995990377e-08, "loss": 0.074, "step": 3151 }, { "epoch": 0.84, "learning_rate": 1.5771184175354182e-08, "loss": 0.1165, "step": 3152 }, { "epoch": 0.84, "learning_rate": 1.574445335471799e-08, "loss": 0.0796, "step": 3153 }, { "epoch": 0.84, "learning_rate": 1.5717722534081795e-08, "loss": 0.0796, "step": 3154 }, { "epoch": 0.84, "learning_rate": 1.5690991713445603e-08, "loss": 0.0778, "step": 3155 }, { "epoch": 0.84, "learning_rate": 1.5664260892809408e-08, "loss": 0.0794, "step": 3156 }, { "epoch": 0.84, "learning_rate": 1.5637530072173216e-08, "loss": 0.0642, "step": 3157 }, { "epoch": 0.84, "learning_rate": 1.561079925153702e-08, "loss": 0.0787, "step": 3158 }, { "epoch": 0.84, "learning_rate": 1.558406843090083e-08, "loss": 0.0916, "step": 3159 }, { "epoch": 0.84, "learning_rate": 1.5557337610264633e-08, "loss": 0.0867, "step": 3160 }, { "epoch": 0.84, "learning_rate": 1.553060678962844e-08, "loss": 0.0679, "step": 3161 }, { "epoch": 0.85, "learning_rate": 1.5503875968992246e-08, "loss": 0.0663, "step": 3162 }, { "epoch": 0.85, "learning_rate": 1.5477145148356054e-08, "loss": 0.0906, "step": 3163 }, { "epoch": 0.85, "learning_rate": 1.545041432771986e-08, "loss": 0.0684, "step": 3164 }, { "epoch": 0.85, "learning_rate": 1.5423683507083667e-08, "loss": 0.0933, "step": 3165 }, { "epoch": 0.85, "learning_rate": 1.539695268644747e-08, "loss": 0.0854, "step": 3166 }, { "epoch": 0.85, "learning_rate": 1.537022186581128e-08, "loss": 0.0802, "step": 3167 }, { "epoch": 0.85, "learning_rate": 1.5343491045175088e-08, "loss": 0.0687, "step": 3168 }, { "epoch": 0.85, "learning_rate": 1.5316760224538892e-08, "loss": 0.095, "step": 3169 }, { "epoch": 0.85, "learning_rate": 1.52900294039027e-08, "loss": 0.0928, "step": 3170 }, { "epoch": 0.85, "learning_rate": 1.526329858326651e-08, "loss": 0.0724, "step": 3171 }, { "epoch": 0.85, "learning_rate": 1.5236567762630313e-08, "loss": 0.0711, "step": 3172 }, { "epoch": 0.85, "learning_rate": 1.520983694199412e-08, "loss": 0.0492, "step": 3173 }, { "epoch": 0.85, "learning_rate": 1.5183106121357926e-08, "loss": 0.0649, "step": 3174 }, { "epoch": 0.85, "learning_rate": 1.5156375300721734e-08, "loss": 0.0547, "step": 3175 }, { "epoch": 0.85, "learning_rate": 1.512964448008554e-08, "loss": 0.095, "step": 3176 }, { "epoch": 0.85, "learning_rate": 1.5102913659449347e-08, "loss": 0.0649, "step": 3177 }, { "epoch": 0.85, "learning_rate": 1.507618283881315e-08, "loss": 0.101, "step": 3178 }, { "epoch": 0.85, "learning_rate": 1.504945201817696e-08, "loss": 0.0742, "step": 3179 }, { "epoch": 0.85, "learning_rate": 1.5022721197540764e-08, "loss": 0.082, "step": 3180 }, { "epoch": 0.85, "learning_rate": 1.4995990376904572e-08, "loss": 0.0648, "step": 3181 }, { "epoch": 0.85, "learning_rate": 1.4969259556268377e-08, "loss": 0.0766, "step": 3182 }, { "epoch": 0.85, "learning_rate": 1.4942528735632185e-08, "loss": 0.0803, "step": 3183 }, { "epoch": 0.85, "learning_rate": 1.491579791499599e-08, "loss": 0.1003, "step": 3184 }, { "epoch": 0.85, "learning_rate": 1.4889067094359798e-08, "loss": 0.0792, "step": 3185 }, { "epoch": 0.85, "learning_rate": 1.4862336273723603e-08, "loss": 0.0634, "step": 3186 }, { "epoch": 0.85, "learning_rate": 1.483560545308741e-08, "loss": 0.0821, "step": 3187 }, { "epoch": 0.85, "learning_rate": 1.4808874632451215e-08, "loss": 0.0643, "step": 3188 }, { "epoch": 0.85, "learning_rate": 1.4782143811815023e-08, "loss": 0.091, "step": 3189 }, { "epoch": 0.85, "learning_rate": 1.4755412991178828e-08, "loss": 0.0913, "step": 3190 }, { "epoch": 0.85, "learning_rate": 1.4728682170542636e-08, "loss": 0.0801, "step": 3191 }, { "epoch": 0.85, "learning_rate": 1.4701951349906441e-08, "loss": 0.0843, "step": 3192 }, { "epoch": 0.85, "learning_rate": 1.4675220529270249e-08, "loss": 0.0551, "step": 3193 }, { "epoch": 0.85, "learning_rate": 1.4648489708634054e-08, "loss": 0.0786, "step": 3194 }, { "epoch": 0.85, "learning_rate": 1.4621758887997862e-08, "loss": 0.0631, "step": 3195 }, { "epoch": 0.85, "learning_rate": 1.4595028067361667e-08, "loss": 0.09, "step": 3196 }, { "epoch": 0.85, "learning_rate": 1.4568297246725475e-08, "loss": 0.0581, "step": 3197 }, { "epoch": 0.85, "learning_rate": 1.454156642608928e-08, "loss": 0.0827, "step": 3198 }, { "epoch": 0.86, "learning_rate": 1.4514835605453087e-08, "loss": 0.0778, "step": 3199 }, { "epoch": 0.86, "learning_rate": 1.4488104784816894e-08, "loss": 0.0769, "step": 3200 }, { "epoch": 0.86, "learning_rate": 1.44613739641807e-08, "loss": 0.0723, "step": 3201 }, { "epoch": 0.86, "learning_rate": 1.4434643143544507e-08, "loss": 0.0774, "step": 3202 }, { "epoch": 0.86, "learning_rate": 1.4407912322908315e-08, "loss": 0.0843, "step": 3203 }, { "epoch": 0.86, "learning_rate": 1.438118150227212e-08, "loss": 0.0668, "step": 3204 }, { "epoch": 0.86, "learning_rate": 1.4354450681635927e-08, "loss": 0.0828, "step": 3205 }, { "epoch": 0.86, "learning_rate": 1.4327719860999732e-08, "loss": 0.0815, "step": 3206 }, { "epoch": 0.86, "learning_rate": 1.430098904036354e-08, "loss": 0.0736, "step": 3207 }, { "epoch": 0.86, "learning_rate": 1.4274258219727345e-08, "loss": 0.078, "step": 3208 }, { "epoch": 0.86, "learning_rate": 1.4247527399091153e-08, "loss": 0.06, "step": 3209 }, { "epoch": 0.86, "learning_rate": 1.4220796578454958e-08, "loss": 0.0529, "step": 3210 }, { "epoch": 0.86, "learning_rate": 1.4194065757818766e-08, "loss": 0.0849, "step": 3211 }, { "epoch": 0.86, "learning_rate": 1.416733493718257e-08, "loss": 0.0763, "step": 3212 }, { "epoch": 0.86, "learning_rate": 1.4140604116546379e-08, "loss": 0.0736, "step": 3213 }, { "epoch": 0.86, "learning_rate": 1.4113873295910183e-08, "loss": 0.0724, "step": 3214 }, { "epoch": 0.86, "learning_rate": 1.4087142475273991e-08, "loss": 0.0524, "step": 3215 }, { "epoch": 0.86, "learning_rate": 1.4060411654637796e-08, "loss": 0.0834, "step": 3216 }, { "epoch": 0.86, "learning_rate": 1.4033680834001604e-08, "loss": 0.0976, "step": 3217 }, { "epoch": 0.86, "learning_rate": 1.4006950013365409e-08, "loss": 0.0814, "step": 3218 }, { "epoch": 0.86, "learning_rate": 1.3980219192729217e-08, "loss": 0.0807, "step": 3219 }, { "epoch": 0.86, "learning_rate": 1.3953488372093022e-08, "loss": 0.0554, "step": 3220 }, { "epoch": 0.86, "learning_rate": 1.392675755145683e-08, "loss": 0.0821, "step": 3221 }, { "epoch": 0.86, "learning_rate": 1.3900026730820636e-08, "loss": 0.0847, "step": 3222 }, { "epoch": 0.86, "learning_rate": 1.3873295910184442e-08, "loss": 0.0663, "step": 3223 }, { "epoch": 0.86, "learning_rate": 1.3846565089548249e-08, "loss": 0.0656, "step": 3224 }, { "epoch": 0.86, "learning_rate": 1.3819834268912057e-08, "loss": 0.0773, "step": 3225 }, { "epoch": 0.86, "learning_rate": 1.3793103448275862e-08, "loss": 0.0871, "step": 3226 }, { "epoch": 0.86, "learning_rate": 1.376637262763967e-08, "loss": 0.0712, "step": 3227 }, { "epoch": 0.86, "learning_rate": 1.3739641807003474e-08, "loss": 0.0748, "step": 3228 }, { "epoch": 0.86, "learning_rate": 1.3712910986367283e-08, "loss": 0.0698, "step": 3229 }, { "epoch": 0.86, "learning_rate": 1.3686180165731087e-08, "loss": 0.0646, "step": 3230 }, { "epoch": 0.86, "learning_rate": 1.3659449345094895e-08, "loss": 0.0854, "step": 3231 }, { "epoch": 0.86, "learning_rate": 1.36327185244587e-08, "loss": 0.0677, "step": 3232 }, { "epoch": 0.86, "learning_rate": 1.3605987703822508e-08, "loss": 0.0825, "step": 3233 }, { "epoch": 0.86, "learning_rate": 1.3579256883186313e-08, "loss": 0.0541, "step": 3234 }, { "epoch": 0.86, "learning_rate": 1.3552526062550121e-08, "loss": 0.0679, "step": 3235 }, { "epoch": 0.86, "learning_rate": 1.3525795241913926e-08, "loss": 0.0735, "step": 3236 }, { "epoch": 0.87, "learning_rate": 1.3499064421277734e-08, "loss": 0.0566, "step": 3237 }, { "epoch": 0.87, "learning_rate": 1.3472333600641538e-08, "loss": 0.05, "step": 3238 }, { "epoch": 0.87, "learning_rate": 1.3445602780005346e-08, "loss": 0.0844, "step": 3239 }, { "epoch": 0.87, "learning_rate": 1.3418871959369151e-08, "loss": 0.0677, "step": 3240 }, { "epoch": 0.87, "learning_rate": 1.339214113873296e-08, "loss": 0.103, "step": 3241 }, { "epoch": 0.87, "learning_rate": 1.3365410318096764e-08, "loss": 0.0873, "step": 3242 }, { "epoch": 0.87, "learning_rate": 1.3338679497460572e-08, "loss": 0.0599, "step": 3243 }, { "epoch": 0.87, "learning_rate": 1.3311948676824378e-08, "loss": 0.071, "step": 3244 }, { "epoch": 0.87, "learning_rate": 1.3285217856188185e-08, "loss": 0.071, "step": 3245 }, { "epoch": 0.87, "learning_rate": 1.3258487035551991e-08, "loss": 0.07, "step": 3246 }, { "epoch": 0.87, "learning_rate": 1.32317562149158e-08, "loss": 0.0767, "step": 3247 }, { "epoch": 0.87, "learning_rate": 1.3205025394279604e-08, "loss": 0.0786, "step": 3248 }, { "epoch": 0.87, "learning_rate": 1.3178294573643412e-08, "loss": 0.0787, "step": 3249 }, { "epoch": 0.87, "learning_rate": 1.3151563753007217e-08, "loss": 0.0758, "step": 3250 }, { "epoch": 0.87, "learning_rate": 1.3124832932371025e-08, "loss": 0.1167, "step": 3251 }, { "epoch": 0.87, "learning_rate": 1.309810211173483e-08, "loss": 0.0675, "step": 3252 }, { "epoch": 0.87, "learning_rate": 1.3071371291098638e-08, "loss": 0.0742, "step": 3253 }, { "epoch": 0.87, "learning_rate": 1.3044640470462442e-08, "loss": 0.0967, "step": 3254 }, { "epoch": 0.87, "learning_rate": 1.301790964982625e-08, "loss": 0.0906, "step": 3255 }, { "epoch": 0.87, "learning_rate": 1.2991178829190055e-08, "loss": 0.0726, "step": 3256 }, { "epoch": 0.87, "learning_rate": 1.2964448008553863e-08, "loss": 0.054, "step": 3257 }, { "epoch": 0.87, "learning_rate": 1.2937717187917668e-08, "loss": 0.1168, "step": 3258 }, { "epoch": 0.87, "learning_rate": 1.2910986367281476e-08, "loss": 0.0789, "step": 3259 }, { "epoch": 0.87, "learning_rate": 1.288425554664528e-08, "loss": 0.0731, "step": 3260 }, { "epoch": 0.87, "learning_rate": 1.2857524726009089e-08, "loss": 0.0854, "step": 3261 }, { "epoch": 0.87, "learning_rate": 1.2830793905372893e-08, "loss": 0.0596, "step": 3262 }, { "epoch": 0.87, "learning_rate": 1.2804063084736702e-08, "loss": 0.057, "step": 3263 }, { "epoch": 0.87, "learning_rate": 1.2777332264100508e-08, "loss": 0.0804, "step": 3264 }, { "epoch": 0.87, "learning_rate": 1.2750601443464314e-08, "loss": 0.0648, "step": 3265 }, { "epoch": 0.87, "learning_rate": 1.272387062282812e-08, "loss": 0.0612, "step": 3266 }, { "epoch": 0.87, "learning_rate": 1.2697139802191929e-08, "loss": 0.0917, "step": 3267 }, { "epoch": 0.87, "learning_rate": 1.2670408981555733e-08, "loss": 0.0616, "step": 3268 }, { "epoch": 0.87, "learning_rate": 1.2643678160919542e-08, "loss": 0.0673, "step": 3269 }, { "epoch": 0.87, "learning_rate": 1.2616947340283346e-08, "loss": 0.0719, "step": 3270 }, { "epoch": 0.87, "learning_rate": 1.2590216519647154e-08, "loss": 0.091, "step": 3271 }, { "epoch": 0.87, "learning_rate": 1.2563485699010959e-08, "loss": 0.073, "step": 3272 }, { "epoch": 0.87, "learning_rate": 1.2536754878374767e-08, "loss": 0.0715, "step": 3273 }, { "epoch": 0.88, "learning_rate": 1.2510024057738572e-08, "loss": 0.0727, "step": 3274 }, { "epoch": 0.88, "learning_rate": 1.2483293237102378e-08, "loss": 0.0829, "step": 3275 }, { "epoch": 0.88, "learning_rate": 1.2456562416466185e-08, "loss": 0.0932, "step": 3276 }, { "epoch": 0.88, "learning_rate": 1.2429831595829991e-08, "loss": 0.0632, "step": 3277 }, { "epoch": 0.88, "learning_rate": 1.2403100775193797e-08, "loss": 0.0843, "step": 3278 }, { "epoch": 0.88, "learning_rate": 1.2376369954557604e-08, "loss": 0.0582, "step": 3279 }, { "epoch": 0.88, "learning_rate": 1.234963913392141e-08, "loss": 0.0865, "step": 3280 }, { "epoch": 0.88, "learning_rate": 1.2322908313285217e-08, "loss": 0.0557, "step": 3281 }, { "epoch": 0.88, "learning_rate": 1.2296177492649023e-08, "loss": 0.0721, "step": 3282 }, { "epoch": 0.88, "learning_rate": 1.226944667201283e-08, "loss": 0.0691, "step": 3283 }, { "epoch": 0.88, "learning_rate": 1.2242715851376636e-08, "loss": 0.0661, "step": 3284 }, { "epoch": 0.88, "learning_rate": 1.2215985030740444e-08, "loss": 0.072, "step": 3285 }, { "epoch": 0.88, "learning_rate": 1.218925421010425e-08, "loss": 0.0727, "step": 3286 }, { "epoch": 0.88, "learning_rate": 1.2162523389468057e-08, "loss": 0.0826, "step": 3287 }, { "epoch": 0.88, "learning_rate": 1.2135792568831863e-08, "loss": 0.0798, "step": 3288 }, { "epoch": 0.88, "learning_rate": 1.210906174819567e-08, "loss": 0.0551, "step": 3289 }, { "epoch": 0.88, "learning_rate": 1.2082330927559476e-08, "loss": 0.0784, "step": 3290 }, { "epoch": 0.88, "learning_rate": 1.2055600106923282e-08, "loss": 0.0581, "step": 3291 }, { "epoch": 0.88, "learning_rate": 1.2028869286287089e-08, "loss": 0.1039, "step": 3292 }, { "epoch": 0.88, "learning_rate": 1.2002138465650895e-08, "loss": 0.0733, "step": 3293 }, { "epoch": 0.88, "learning_rate": 1.1975407645014701e-08, "loss": 0.0751, "step": 3294 }, { "epoch": 0.88, "learning_rate": 1.1948676824378508e-08, "loss": 0.0879, "step": 3295 }, { "epoch": 0.88, "learning_rate": 1.1921946003742314e-08, "loss": 0.0878, "step": 3296 }, { "epoch": 0.88, "learning_rate": 1.189521518310612e-08, "loss": 0.0966, "step": 3297 }, { "epoch": 0.88, "learning_rate": 1.1868484362469927e-08, "loss": 0.087, "step": 3298 }, { "epoch": 0.88, "learning_rate": 1.1841753541833733e-08, "loss": 0.0867, "step": 3299 }, { "epoch": 0.88, "learning_rate": 1.181502272119754e-08, "loss": 0.138, "step": 3300 }, { "epoch": 0.88, "learning_rate": 1.1788291900561346e-08, "loss": 0.094, "step": 3301 }, { "epoch": 0.88, "learning_rate": 1.1761561079925152e-08, "loss": 0.0668, "step": 3302 }, { "epoch": 0.88, "learning_rate": 1.1734830259288959e-08, "loss": 0.0759, "step": 3303 }, { "epoch": 0.88, "learning_rate": 1.1708099438652765e-08, "loss": 0.0674, "step": 3304 }, { "epoch": 0.88, "learning_rate": 1.1681368618016572e-08, "loss": 0.0996, "step": 3305 }, { "epoch": 0.88, "learning_rate": 1.165463779738038e-08, "loss": 0.0904, "step": 3306 }, { "epoch": 0.88, "learning_rate": 1.1627906976744186e-08, "loss": 0.0625, "step": 3307 }, { "epoch": 0.88, "learning_rate": 1.1601176156107992e-08, "loss": 0.0674, "step": 3308 }, { "epoch": 0.88, "learning_rate": 1.1574445335471799e-08, "loss": 0.0815, "step": 3309 }, { "epoch": 0.88, "learning_rate": 1.1547714514835605e-08, "loss": 0.0706, "step": 3310 }, { "epoch": 0.88, "learning_rate": 1.1520983694199412e-08, "loss": 0.0657, "step": 3311 }, { "epoch": 0.89, "learning_rate": 1.1494252873563218e-08, "loss": 0.0914, "step": 3312 }, { "epoch": 0.89, "learning_rate": 1.1467522052927024e-08, "loss": 0.0675, "step": 3313 }, { "epoch": 0.89, "learning_rate": 1.1440791232290831e-08, "loss": 0.0571, "step": 3314 }, { "epoch": 0.89, "learning_rate": 1.1414060411654637e-08, "loss": 0.0741, "step": 3315 }, { "epoch": 0.89, "learning_rate": 1.1387329591018444e-08, "loss": 0.0797, "step": 3316 }, { "epoch": 0.89, "learning_rate": 1.136059877038225e-08, "loss": 0.0852, "step": 3317 }, { "epoch": 0.89, "learning_rate": 1.1333867949746056e-08, "loss": 0.0862, "step": 3318 }, { "epoch": 0.89, "learning_rate": 1.1307137129109863e-08, "loss": 0.0612, "step": 3319 }, { "epoch": 0.89, "learning_rate": 1.128040630847367e-08, "loss": 0.0684, "step": 3320 }, { "epoch": 0.89, "learning_rate": 1.1253675487837476e-08, "loss": 0.0739, "step": 3321 }, { "epoch": 0.89, "learning_rate": 1.1226944667201282e-08, "loss": 0.0843, "step": 3322 }, { "epoch": 0.89, "learning_rate": 1.1200213846565088e-08, "loss": 0.0706, "step": 3323 }, { "epoch": 0.89, "learning_rate": 1.1173483025928895e-08, "loss": 0.0711, "step": 3324 }, { "epoch": 0.89, "learning_rate": 1.1146752205292701e-08, "loss": 0.0734, "step": 3325 }, { "epoch": 0.89, "learning_rate": 1.1120021384656508e-08, "loss": 0.0682, "step": 3326 }, { "epoch": 0.89, "learning_rate": 1.1093290564020314e-08, "loss": 0.0749, "step": 3327 }, { "epoch": 0.89, "learning_rate": 1.1066559743384122e-08, "loss": 0.0694, "step": 3328 }, { "epoch": 0.89, "learning_rate": 1.1039828922747928e-08, "loss": 0.0648, "step": 3329 }, { "epoch": 0.89, "learning_rate": 1.1013098102111735e-08, "loss": 0.0669, "step": 3330 }, { "epoch": 0.89, "learning_rate": 1.0986367281475541e-08, "loss": 0.0612, "step": 3331 }, { "epoch": 0.89, "learning_rate": 1.0959636460839348e-08, "loss": 0.0629, "step": 3332 }, { "epoch": 0.89, "learning_rate": 1.0932905640203154e-08, "loss": 0.0812, "step": 3333 }, { "epoch": 0.89, "learning_rate": 1.090617481956696e-08, "loss": 0.0815, "step": 3334 }, { "epoch": 0.89, "learning_rate": 1.0879443998930767e-08, "loss": 0.0688, "step": 3335 }, { "epoch": 0.89, "learning_rate": 1.0852713178294573e-08, "loss": 0.0879, "step": 3336 }, { "epoch": 0.89, "learning_rate": 1.082598235765838e-08, "loss": 0.0867, "step": 3337 }, { "epoch": 0.89, "learning_rate": 1.0799251537022186e-08, "loss": 0.0837, "step": 3338 }, { "epoch": 0.89, "learning_rate": 1.0772520716385992e-08, "loss": 0.0559, "step": 3339 }, { "epoch": 0.89, "learning_rate": 1.0745789895749799e-08, "loss": 0.0648, "step": 3340 }, { "epoch": 0.89, "learning_rate": 1.0719059075113605e-08, "loss": 0.0578, "step": 3341 }, { "epoch": 0.89, "learning_rate": 1.0692328254477411e-08, "loss": 0.092, "step": 3342 }, { "epoch": 0.89, "learning_rate": 1.0665597433841218e-08, "loss": 0.0636, "step": 3343 }, { "epoch": 0.89, "learning_rate": 1.0638866613205024e-08, "loss": 0.0933, "step": 3344 }, { "epoch": 0.89, "learning_rate": 1.061213579256883e-08, "loss": 0.0675, "step": 3345 }, { "epoch": 0.89, "learning_rate": 1.0585404971932637e-08, "loss": 0.0847, "step": 3346 }, { "epoch": 0.89, "learning_rate": 1.0558674151296443e-08, "loss": 0.0972, "step": 3347 }, { "epoch": 0.89, "learning_rate": 1.053194333066025e-08, "loss": 0.0873, "step": 3348 }, { "epoch": 0.9, "learning_rate": 1.0505212510024058e-08, "loss": 0.0595, "step": 3349 }, { "epoch": 0.9, "learning_rate": 1.0478481689387864e-08, "loss": 0.0627, "step": 3350 }, { "epoch": 0.9, "learning_rate": 1.045175086875167e-08, "loss": 0.0732, "step": 3351 }, { "epoch": 0.9, "learning_rate": 1.0425020048115477e-08, "loss": 0.0794, "step": 3352 }, { "epoch": 0.9, "learning_rate": 1.0398289227479283e-08, "loss": 0.0506, "step": 3353 }, { "epoch": 0.9, "learning_rate": 1.037155840684309e-08, "loss": 0.084, "step": 3354 }, { "epoch": 0.9, "learning_rate": 1.0344827586206896e-08, "loss": 0.0703, "step": 3355 }, { "epoch": 0.9, "learning_rate": 1.0318096765570703e-08, "loss": 0.0773, "step": 3356 }, { "epoch": 0.9, "learning_rate": 1.0291365944934509e-08, "loss": 0.0869, "step": 3357 }, { "epoch": 0.9, "learning_rate": 1.0264635124298315e-08, "loss": 0.1081, "step": 3358 }, { "epoch": 0.9, "learning_rate": 1.0237904303662122e-08, "loss": 0.0569, "step": 3359 }, { "epoch": 0.9, "learning_rate": 1.0211173483025928e-08, "loss": 0.0615, "step": 3360 }, { "epoch": 0.9, "learning_rate": 1.0184442662389735e-08, "loss": 0.0668, "step": 3361 }, { "epoch": 0.9, "learning_rate": 1.0157711841753541e-08, "loss": 0.0997, "step": 3362 }, { "epoch": 0.9, "learning_rate": 1.0130981021117347e-08, "loss": 0.0684, "step": 3363 }, { "epoch": 0.9, "learning_rate": 1.0104250200481154e-08, "loss": 0.0894, "step": 3364 }, { "epoch": 0.9, "learning_rate": 1.007751937984496e-08, "loss": 0.0802, "step": 3365 }, { "epoch": 0.9, "learning_rate": 1.0050788559208767e-08, "loss": 0.0692, "step": 3366 }, { "epoch": 0.9, "learning_rate": 1.0024057738572573e-08, "loss": 0.0802, "step": 3367 }, { "epoch": 0.9, "learning_rate": 9.99732691793638e-09, "loss": 0.0882, "step": 3368 }, { "epoch": 0.9, "learning_rate": 9.970596097300186e-09, "loss": 0.0561, "step": 3369 }, { "epoch": 0.9, "learning_rate": 9.943865276663994e-09, "loss": 0.059, "step": 3370 }, { "epoch": 0.9, "learning_rate": 9.9171344560278e-09, "loss": 0.0706, "step": 3371 }, { "epoch": 0.9, "learning_rate": 9.890403635391607e-09, "loss": 0.052, "step": 3372 }, { "epoch": 0.9, "learning_rate": 9.863672814755413e-09, "loss": 0.0876, "step": 3373 }, { "epoch": 0.9, "learning_rate": 9.83694199411922e-09, "loss": 0.0804, "step": 3374 }, { "epoch": 0.9, "learning_rate": 9.810211173483026e-09, "loss": 0.0677, "step": 3375 }, { "epoch": 0.9, "learning_rate": 9.783480352846832e-09, "loss": 0.0823, "step": 3376 }, { "epoch": 0.9, "learning_rate": 9.756749532210639e-09, "loss": 0.1272, "step": 3377 }, { "epoch": 0.9, "learning_rate": 9.730018711574445e-09, "loss": 0.0834, "step": 3378 }, { "epoch": 0.9, "learning_rate": 9.703287890938251e-09, "loss": 0.0782, "step": 3379 }, { "epoch": 0.9, "learning_rate": 9.676557070302058e-09, "loss": 0.0438, "step": 3380 }, { "epoch": 0.9, "learning_rate": 9.649826249665864e-09, "loss": 0.0745, "step": 3381 }, { "epoch": 0.9, "learning_rate": 9.62309542902967e-09, "loss": 0.0801, "step": 3382 }, { "epoch": 0.9, "learning_rate": 9.596364608393477e-09, "loss": 0.0597, "step": 3383 }, { "epoch": 0.9, "learning_rate": 9.569633787757283e-09, "loss": 0.0723, "step": 3384 }, { "epoch": 0.9, "learning_rate": 9.54290296712109e-09, "loss": 0.0742, "step": 3385 }, { "epoch": 0.9, "learning_rate": 9.516172146484896e-09, "loss": 0.0994, "step": 3386 }, { "epoch": 0.91, "learning_rate": 9.489441325848702e-09, "loss": 0.0646, "step": 3387 }, { "epoch": 0.91, "learning_rate": 9.462710505212509e-09, "loss": 0.0756, "step": 3388 }, { "epoch": 0.91, "learning_rate": 9.435979684576315e-09, "loss": 0.0854, "step": 3389 }, { "epoch": 0.91, "learning_rate": 9.409248863940122e-09, "loss": 0.0649, "step": 3390 }, { "epoch": 0.91, "learning_rate": 9.38251804330393e-09, "loss": 0.0583, "step": 3391 }, { "epoch": 0.91, "learning_rate": 9.355787222667736e-09, "loss": 0.0875, "step": 3392 }, { "epoch": 0.91, "learning_rate": 9.329056402031542e-09, "loss": 0.088, "step": 3393 }, { "epoch": 0.91, "learning_rate": 9.302325581395349e-09, "loss": 0.0551, "step": 3394 }, { "epoch": 0.91, "learning_rate": 9.275594760759155e-09, "loss": 0.0812, "step": 3395 }, { "epoch": 0.91, "learning_rate": 9.248863940122962e-09, "loss": 0.0523, "step": 3396 }, { "epoch": 0.91, "learning_rate": 9.222133119486768e-09, "loss": 0.0962, "step": 3397 }, { "epoch": 0.91, "learning_rate": 9.195402298850574e-09, "loss": 0.0784, "step": 3398 }, { "epoch": 0.91, "learning_rate": 9.168671478214381e-09, "loss": 0.0603, "step": 3399 }, { "epoch": 0.91, "learning_rate": 9.141940657578187e-09, "loss": 0.0606, "step": 3400 }, { "epoch": 0.91, "learning_rate": 9.115209836941994e-09, "loss": 0.0671, "step": 3401 }, { "epoch": 0.91, "learning_rate": 9.0884790163058e-09, "loss": 0.0513, "step": 3402 }, { "epoch": 0.91, "learning_rate": 9.061748195669606e-09, "loss": 0.052, "step": 3403 }, { "epoch": 0.91, "learning_rate": 9.035017375033413e-09, "loss": 0.0486, "step": 3404 }, { "epoch": 0.91, "learning_rate": 9.00828655439722e-09, "loss": 0.074, "step": 3405 }, { "epoch": 0.91, "learning_rate": 8.981555733761026e-09, "loss": 0.074, "step": 3406 }, { "epoch": 0.91, "learning_rate": 8.954824913124832e-09, "loss": 0.0616, "step": 3407 }, { "epoch": 0.91, "learning_rate": 8.928094092488638e-09, "loss": 0.0715, "step": 3408 }, { "epoch": 0.91, "learning_rate": 8.901363271852445e-09, "loss": 0.0891, "step": 3409 }, { "epoch": 0.91, "learning_rate": 8.874632451216251e-09, "loss": 0.0756, "step": 3410 }, { "epoch": 0.91, "learning_rate": 8.847901630580058e-09, "loss": 0.083, "step": 3411 }, { "epoch": 0.91, "learning_rate": 8.821170809943864e-09, "loss": 0.0781, "step": 3412 }, { "epoch": 0.91, "learning_rate": 8.794439989307672e-09, "loss": 0.0948, "step": 3413 }, { "epoch": 0.91, "learning_rate": 8.767709168671478e-09, "loss": 0.0625, "step": 3414 }, { "epoch": 0.91, "learning_rate": 8.740978348035285e-09, "loss": 0.0888, "step": 3415 }, { "epoch": 0.91, "learning_rate": 8.714247527399091e-09, "loss": 0.047, "step": 3416 }, { "epoch": 0.91, "learning_rate": 8.687516706762898e-09, "loss": 0.0888, "step": 3417 }, { "epoch": 0.91, "learning_rate": 8.660785886126704e-09, "loss": 0.0682, "step": 3418 }, { "epoch": 0.91, "learning_rate": 8.63405506549051e-09, "loss": 0.0778, "step": 3419 }, { "epoch": 0.91, "learning_rate": 8.607324244854317e-09, "loss": 0.0721, "step": 3420 }, { "epoch": 0.91, "learning_rate": 8.580593424218123e-09, "loss": 0.0755, "step": 3421 }, { "epoch": 0.91, "learning_rate": 8.55386260358193e-09, "loss": 0.0741, "step": 3422 }, { "epoch": 0.91, "learning_rate": 8.527131782945736e-09, "loss": 0.0704, "step": 3423 }, { "epoch": 0.92, "learning_rate": 8.500400962309542e-09, "loss": 0.0644, "step": 3424 }, { "epoch": 0.92, "learning_rate": 8.473670141673349e-09, "loss": 0.069, "step": 3425 }, { "epoch": 0.92, "learning_rate": 8.446939321037155e-09, "loss": 0.0742, "step": 3426 }, { "epoch": 0.92, "learning_rate": 8.420208500400961e-09, "loss": 0.1084, "step": 3427 }, { "epoch": 0.92, "learning_rate": 8.393477679764768e-09, "loss": 0.0998, "step": 3428 }, { "epoch": 0.92, "learning_rate": 8.366746859128574e-09, "loss": 0.0805, "step": 3429 }, { "epoch": 0.92, "learning_rate": 8.34001603849238e-09, "loss": 0.078, "step": 3430 }, { "epoch": 0.92, "learning_rate": 8.313285217856187e-09, "loss": 0.0962, "step": 3431 }, { "epoch": 0.92, "learning_rate": 8.286554397219993e-09, "loss": 0.0612, "step": 3432 }, { "epoch": 0.92, "learning_rate": 8.2598235765838e-09, "loss": 0.0683, "step": 3433 }, { "epoch": 0.92, "learning_rate": 8.233092755947608e-09, "loss": 0.0748, "step": 3434 }, { "epoch": 0.92, "learning_rate": 8.206361935311414e-09, "loss": 0.0675, "step": 3435 }, { "epoch": 0.92, "learning_rate": 8.17963111467522e-09, "loss": 0.0701, "step": 3436 }, { "epoch": 0.92, "learning_rate": 8.152900294039027e-09, "loss": 0.0763, "step": 3437 }, { "epoch": 0.92, "learning_rate": 8.126169473402833e-09, "loss": 0.1042, "step": 3438 }, { "epoch": 0.92, "learning_rate": 8.09943865276664e-09, "loss": 0.0911, "step": 3439 }, { "epoch": 0.92, "learning_rate": 8.072707832130446e-09, "loss": 0.0598, "step": 3440 }, { "epoch": 0.92, "learning_rate": 8.045977011494253e-09, "loss": 0.0592, "step": 3441 }, { "epoch": 0.92, "learning_rate": 8.019246190858059e-09, "loss": 0.076, "step": 3442 }, { "epoch": 0.92, "learning_rate": 7.992515370221865e-09, "loss": 0.0559, "step": 3443 }, { "epoch": 0.92, "learning_rate": 7.965784549585672e-09, "loss": 0.0845, "step": 3444 }, { "epoch": 0.92, "learning_rate": 7.939053728949478e-09, "loss": 0.0772, "step": 3445 }, { "epoch": 0.92, "learning_rate": 7.912322908313285e-09, "loss": 0.098, "step": 3446 }, { "epoch": 0.92, "learning_rate": 7.885592087677091e-09, "loss": 0.0839, "step": 3447 }, { "epoch": 0.92, "learning_rate": 7.858861267040897e-09, "loss": 0.0694, "step": 3448 }, { "epoch": 0.92, "learning_rate": 7.832130446404704e-09, "loss": 0.0737, "step": 3449 }, { "epoch": 0.92, "learning_rate": 7.80539962576851e-09, "loss": 0.0789, "step": 3450 }, { "epoch": 0.92, "learning_rate": 7.778668805132317e-09, "loss": 0.0758, "step": 3451 }, { "epoch": 0.92, "learning_rate": 7.751937984496123e-09, "loss": 0.0602, "step": 3452 }, { "epoch": 0.92, "learning_rate": 7.72520716385993e-09, "loss": 0.0571, "step": 3453 }, { "epoch": 0.92, "learning_rate": 7.698476343223736e-09, "loss": 0.0867, "step": 3454 }, { "epoch": 0.92, "learning_rate": 7.671745522587544e-09, "loss": 0.0648, "step": 3455 }, { "epoch": 0.92, "learning_rate": 7.64501470195135e-09, "loss": 0.1383, "step": 3456 }, { "epoch": 0.92, "learning_rate": 7.618283881315157e-09, "loss": 0.0808, "step": 3457 }, { "epoch": 0.92, "learning_rate": 7.591553060678963e-09, "loss": 0.1029, "step": 3458 }, { "epoch": 0.92, "learning_rate": 7.56482224004277e-09, "loss": 0.0787, "step": 3459 }, { "epoch": 0.92, "learning_rate": 7.538091419406576e-09, "loss": 0.0895, "step": 3460 }, { "epoch": 0.93, "learning_rate": 7.511360598770382e-09, "loss": 0.0636, "step": 3461 }, { "epoch": 0.93, "learning_rate": 7.484629778134189e-09, "loss": 0.0785, "step": 3462 }, { "epoch": 0.93, "learning_rate": 7.457898957497995e-09, "loss": 0.0533, "step": 3463 }, { "epoch": 0.93, "learning_rate": 7.431168136861801e-09, "loss": 0.0612, "step": 3464 }, { "epoch": 0.93, "learning_rate": 7.404437316225608e-09, "loss": 0.0607, "step": 3465 }, { "epoch": 0.93, "learning_rate": 7.377706495589414e-09, "loss": 0.09, "step": 3466 }, { "epoch": 0.93, "learning_rate": 7.3509756749532205e-09, "loss": 0.0691, "step": 3467 }, { "epoch": 0.93, "learning_rate": 7.324244854317027e-09, "loss": 0.0736, "step": 3468 }, { "epoch": 0.93, "learning_rate": 7.297514033680833e-09, "loss": 0.0551, "step": 3469 }, { "epoch": 0.93, "learning_rate": 7.27078321304464e-09, "loss": 0.0658, "step": 3470 }, { "epoch": 0.93, "learning_rate": 7.244052392408447e-09, "loss": 0.0862, "step": 3471 }, { "epoch": 0.93, "learning_rate": 7.217321571772253e-09, "loss": 0.0703, "step": 3472 }, { "epoch": 0.93, "learning_rate": 7.19059075113606e-09, "loss": 0.0704, "step": 3473 }, { "epoch": 0.93, "learning_rate": 7.163859930499866e-09, "loss": 0.0949, "step": 3474 }, { "epoch": 0.93, "learning_rate": 7.1371291098636725e-09, "loss": 0.0991, "step": 3475 }, { "epoch": 0.93, "learning_rate": 7.110398289227479e-09, "loss": 0.08, "step": 3476 }, { "epoch": 0.93, "learning_rate": 7.083667468591285e-09, "loss": 0.0655, "step": 3477 }, { "epoch": 0.93, "learning_rate": 7.056936647955092e-09, "loss": 0.0729, "step": 3478 }, { "epoch": 0.93, "learning_rate": 7.030205827318898e-09, "loss": 0.0733, "step": 3479 }, { "epoch": 0.93, "learning_rate": 7.0034750066827044e-09, "loss": 0.0904, "step": 3480 }, { "epoch": 0.93, "learning_rate": 6.976744186046511e-09, "loss": 0.0895, "step": 3481 }, { "epoch": 0.93, "learning_rate": 6.950013365410318e-09, "loss": 0.0862, "step": 3482 }, { "epoch": 0.93, "learning_rate": 6.9232825447741244e-09, "loss": 0.0892, "step": 3483 }, { "epoch": 0.93, "learning_rate": 6.896551724137931e-09, "loss": 0.0897, "step": 3484 }, { "epoch": 0.93, "learning_rate": 6.869820903501737e-09, "loss": 0.0718, "step": 3485 }, { "epoch": 0.93, "learning_rate": 6.843090082865544e-09, "loss": 0.0751, "step": 3486 }, { "epoch": 0.93, "learning_rate": 6.81635926222935e-09, "loss": 0.073, "step": 3487 }, { "epoch": 0.93, "learning_rate": 6.789628441593156e-09, "loss": 0.0606, "step": 3488 }, { "epoch": 0.93, "learning_rate": 6.762897620956963e-09, "loss": 0.0697, "step": 3489 }, { "epoch": 0.93, "learning_rate": 6.736166800320769e-09, "loss": 0.0663, "step": 3490 }, { "epoch": 0.93, "learning_rate": 6.7094359796845756e-09, "loss": 0.0761, "step": 3491 }, { "epoch": 0.93, "learning_rate": 6.682705159048382e-09, "loss": 0.0711, "step": 3492 }, { "epoch": 0.93, "learning_rate": 6.655974338412189e-09, "loss": 0.081, "step": 3493 }, { "epoch": 0.93, "learning_rate": 6.629243517775996e-09, "loss": 0.0799, "step": 3494 }, { "epoch": 0.93, "learning_rate": 6.602512697139802e-09, "loss": 0.0996, "step": 3495 }, { "epoch": 0.93, "learning_rate": 6.575781876503608e-09, "loss": 0.0801, "step": 3496 }, { "epoch": 0.93, "learning_rate": 6.549051055867415e-09, "loss": 0.0687, "step": 3497 }, { "epoch": 0.93, "learning_rate": 6.522320235231221e-09, "loss": 0.0835, "step": 3498 }, { "epoch": 0.94, "learning_rate": 6.4955894145950276e-09, "loss": 0.0572, "step": 3499 }, { "epoch": 0.94, "learning_rate": 6.468858593958834e-09, "loss": 0.0697, "step": 3500 }, { "epoch": 0.94, "learning_rate": 6.44212777332264e-09, "loss": 0.1098, "step": 3501 }, { "epoch": 0.94, "learning_rate": 6.415396952686447e-09, "loss": 0.0535, "step": 3502 }, { "epoch": 0.94, "learning_rate": 6.388666132050254e-09, "loss": 0.0569, "step": 3503 }, { "epoch": 0.94, "learning_rate": 6.36193531141406e-09, "loss": 0.0887, "step": 3504 }, { "epoch": 0.94, "learning_rate": 6.335204490777867e-09, "loss": 0.0588, "step": 3505 }, { "epoch": 0.94, "learning_rate": 6.308473670141673e-09, "loss": 0.0624, "step": 3506 }, { "epoch": 0.94, "learning_rate": 6.2817428495054795e-09, "loss": 0.0659, "step": 3507 }, { "epoch": 0.94, "learning_rate": 6.255012028869286e-09, "loss": 0.0683, "step": 3508 }, { "epoch": 0.94, "learning_rate": 6.228281208233092e-09, "loss": 0.0951, "step": 3509 }, { "epoch": 0.94, "learning_rate": 6.201550387596899e-09, "loss": 0.0745, "step": 3510 }, { "epoch": 0.94, "learning_rate": 6.174819566960705e-09, "loss": 0.0946, "step": 3511 }, { "epoch": 0.94, "learning_rate": 6.1480887463245115e-09, "loss": 0.0651, "step": 3512 }, { "epoch": 0.94, "learning_rate": 6.121357925688318e-09, "loss": 0.0652, "step": 3513 }, { "epoch": 0.94, "learning_rate": 6.094627105052125e-09, "loss": 0.1453, "step": 3514 }, { "epoch": 0.94, "learning_rate": 6.0678962844159315e-09, "loss": 0.087, "step": 3515 }, { "epoch": 0.94, "learning_rate": 6.041165463779738e-09, "loss": 0.0678, "step": 3516 }, { "epoch": 0.94, "learning_rate": 6.014434643143544e-09, "loss": 0.0777, "step": 3517 }, { "epoch": 0.94, "learning_rate": 5.987703822507351e-09, "loss": 0.0771, "step": 3518 }, { "epoch": 0.94, "learning_rate": 5.960973001871157e-09, "loss": 0.0582, "step": 3519 }, { "epoch": 0.94, "learning_rate": 5.9342421812349635e-09, "loss": 0.0766, "step": 3520 }, { "epoch": 0.94, "learning_rate": 5.90751136059877e-09, "loss": 0.0801, "step": 3521 }, { "epoch": 0.94, "learning_rate": 5.880780539962576e-09, "loss": 0.0595, "step": 3522 }, { "epoch": 0.94, "learning_rate": 5.854049719326383e-09, "loss": 0.0789, "step": 3523 }, { "epoch": 0.94, "learning_rate": 5.82731889869019e-09, "loss": 0.0865, "step": 3524 }, { "epoch": 0.94, "learning_rate": 5.800588078053996e-09, "loss": 0.07, "step": 3525 }, { "epoch": 0.94, "learning_rate": 5.773857257417803e-09, "loss": 0.053, "step": 3526 }, { "epoch": 0.94, "learning_rate": 5.747126436781609e-09, "loss": 0.0625, "step": 3527 }, { "epoch": 0.94, "learning_rate": 5.7203956161454154e-09, "loss": 0.0752, "step": 3528 }, { "epoch": 0.94, "learning_rate": 5.693664795509222e-09, "loss": 0.0781, "step": 3529 }, { "epoch": 0.94, "learning_rate": 5.666933974873028e-09, "loss": 0.0843, "step": 3530 }, { "epoch": 0.94, "learning_rate": 5.640203154236835e-09, "loss": 0.0674, "step": 3531 }, { "epoch": 0.94, "learning_rate": 5.613472333600641e-09, "loss": 0.0718, "step": 3532 }, { "epoch": 0.94, "learning_rate": 5.586741512964447e-09, "loss": 0.0602, "step": 3533 }, { "epoch": 0.94, "learning_rate": 5.560010692328254e-09, "loss": 0.0519, "step": 3534 }, { "epoch": 0.94, "learning_rate": 5.533279871692061e-09, "loss": 0.0753, "step": 3535 }, { "epoch": 0.95, "learning_rate": 5.506549051055867e-09, "loss": 0.0917, "step": 3536 }, { "epoch": 0.95, "learning_rate": 5.479818230419674e-09, "loss": 0.0495, "step": 3537 }, { "epoch": 0.95, "learning_rate": 5.45308740978348e-09, "loss": 0.0682, "step": 3538 }, { "epoch": 0.95, "learning_rate": 5.4263565891472866e-09, "loss": 0.0628, "step": 3539 }, { "epoch": 0.95, "learning_rate": 5.399625768511093e-09, "loss": 0.0869, "step": 3540 }, { "epoch": 0.95, "learning_rate": 5.372894947874899e-09, "loss": 0.1013, "step": 3541 }, { "epoch": 0.95, "learning_rate": 5.346164127238706e-09, "loss": 0.0527, "step": 3542 }, { "epoch": 0.95, "learning_rate": 5.319433306602512e-09, "loss": 0.0658, "step": 3543 }, { "epoch": 0.95, "learning_rate": 5.2927024859663185e-09, "loss": 0.0734, "step": 3544 }, { "epoch": 0.95, "learning_rate": 5.265971665330125e-09, "loss": 0.0704, "step": 3545 }, { "epoch": 0.95, "learning_rate": 5.239240844693932e-09, "loss": 0.0964, "step": 3546 }, { "epoch": 0.95, "learning_rate": 5.2125100240577385e-09, "loss": 0.1267, "step": 3547 }, { "epoch": 0.95, "learning_rate": 5.185779203421545e-09, "loss": 0.072, "step": 3548 }, { "epoch": 0.95, "learning_rate": 5.159048382785351e-09, "loss": 0.0767, "step": 3549 }, { "epoch": 0.95, "learning_rate": 5.132317562149158e-09, "loss": 0.0602, "step": 3550 }, { "epoch": 0.95, "learning_rate": 5.105586741512964e-09, "loss": 0.0828, "step": 3551 }, { "epoch": 0.95, "learning_rate": 5.0788559208767705e-09, "loss": 0.0735, "step": 3552 }, { "epoch": 0.95, "learning_rate": 5.052125100240577e-09, "loss": 0.0866, "step": 3553 }, { "epoch": 0.95, "learning_rate": 5.025394279604383e-09, "loss": 0.0878, "step": 3554 }, { "epoch": 0.95, "learning_rate": 4.99866345896819e-09, "loss": 0.0801, "step": 3555 }, { "epoch": 0.95, "learning_rate": 4.971932638331997e-09, "loss": 0.0658, "step": 3556 }, { "epoch": 0.95, "learning_rate": 4.945201817695803e-09, "loss": 0.0609, "step": 3557 }, { "epoch": 0.95, "learning_rate": 4.91847099705961e-09, "loss": 0.071, "step": 3558 }, { "epoch": 0.95, "learning_rate": 4.891740176423416e-09, "loss": 0.0532, "step": 3559 }, { "epoch": 0.95, "learning_rate": 4.8650093557872225e-09, "loss": 0.0972, "step": 3560 }, { "epoch": 0.95, "learning_rate": 4.838278535151029e-09, "loss": 0.0848, "step": 3561 }, { "epoch": 0.95, "learning_rate": 4.811547714514835e-09, "loss": 0.0835, "step": 3562 }, { "epoch": 0.95, "learning_rate": 4.784816893878642e-09, "loss": 0.0418, "step": 3563 }, { "epoch": 0.95, "learning_rate": 4.758086073242448e-09, "loss": 0.0707, "step": 3564 }, { "epoch": 0.95, "learning_rate": 4.7313552526062544e-09, "loss": 0.075, "step": 3565 }, { "epoch": 0.95, "learning_rate": 4.704624431970061e-09, "loss": 0.0787, "step": 3566 }, { "epoch": 0.95, "learning_rate": 4.677893611333868e-09, "loss": 0.0918, "step": 3567 }, { "epoch": 0.95, "learning_rate": 4.6511627906976744e-09, "loss": 0.1884, "step": 3568 }, { "epoch": 0.95, "learning_rate": 4.624431970061481e-09, "loss": 0.0683, "step": 3569 }, { "epoch": 0.95, "learning_rate": 4.597701149425287e-09, "loss": 0.0845, "step": 3570 }, { "epoch": 0.95, "learning_rate": 4.570970328789094e-09, "loss": 0.0654, "step": 3571 }, { "epoch": 0.95, "learning_rate": 4.5442395081529e-09, "loss": 0.0806, "step": 3572 }, { "epoch": 0.95, "learning_rate": 4.517508687516706e-09, "loss": 0.0528, "step": 3573 }, { "epoch": 0.96, "learning_rate": 4.490777866880513e-09, "loss": 0.0812, "step": 3574 }, { "epoch": 0.96, "learning_rate": 4.464047046244319e-09, "loss": 0.0827, "step": 3575 }, { "epoch": 0.96, "learning_rate": 4.4373162256081256e-09, "loss": 0.0653, "step": 3576 }, { "epoch": 0.96, "learning_rate": 4.410585404971932e-09, "loss": 0.0735, "step": 3577 }, { "epoch": 0.96, "learning_rate": 4.383854584335739e-09, "loss": 0.0699, "step": 3578 }, { "epoch": 0.96, "learning_rate": 4.3571237636995456e-09, "loss": 0.069, "step": 3579 }, { "epoch": 0.96, "learning_rate": 4.330392943063352e-09, "loss": 0.0842, "step": 3580 }, { "epoch": 0.96, "learning_rate": 4.303662122427158e-09, "loss": 0.0603, "step": 3581 }, { "epoch": 0.96, "learning_rate": 4.276931301790965e-09, "loss": 0.0945, "step": 3582 }, { "epoch": 0.96, "learning_rate": 4.250200481154771e-09, "loss": 0.1058, "step": 3583 }, { "epoch": 0.96, "learning_rate": 4.2234696605185775e-09, "loss": 0.0598, "step": 3584 }, { "epoch": 0.96, "learning_rate": 4.196738839882384e-09, "loss": 0.089, "step": 3585 }, { "epoch": 0.96, "learning_rate": 4.17000801924619e-09, "loss": 0.0701, "step": 3586 }, { "epoch": 0.96, "learning_rate": 4.143277198609997e-09, "loss": 0.0836, "step": 3587 }, { "epoch": 0.96, "learning_rate": 4.116546377973804e-09, "loss": 0.0969, "step": 3588 }, { "epoch": 0.96, "learning_rate": 4.08981555733761e-09, "loss": 0.0653, "step": 3589 }, { "epoch": 0.96, "learning_rate": 4.063084736701417e-09, "loss": 0.077, "step": 3590 }, { "epoch": 0.96, "learning_rate": 4.036353916065223e-09, "loss": 0.0652, "step": 3591 }, { "epoch": 0.96, "learning_rate": 4.0096230954290295e-09, "loss": 0.0636, "step": 3592 }, { "epoch": 0.96, "learning_rate": 3.982892274792836e-09, "loss": 0.0643, "step": 3593 }, { "epoch": 0.96, "learning_rate": 3.956161454156642e-09, "loss": 0.084, "step": 3594 }, { "epoch": 0.96, "learning_rate": 3.929430633520449e-09, "loss": 0.0664, "step": 3595 }, { "epoch": 0.96, "learning_rate": 3.902699812884255e-09, "loss": 0.0669, "step": 3596 }, { "epoch": 0.96, "learning_rate": 3.8759689922480615e-09, "loss": 0.0588, "step": 3597 }, { "epoch": 0.96, "learning_rate": 3.849238171611868e-09, "loss": 0.0798, "step": 3598 }, { "epoch": 0.96, "learning_rate": 3.822507350975675e-09, "loss": 0.0775, "step": 3599 }, { "epoch": 0.96, "learning_rate": 3.7957765303394815e-09, "loss": 0.0749, "step": 3600 }, { "epoch": 0.96, "learning_rate": 3.769045709703288e-09, "loss": 0.0732, "step": 3601 }, { "epoch": 0.96, "learning_rate": 3.742314889067094e-09, "loss": 0.059, "step": 3602 }, { "epoch": 0.96, "learning_rate": 3.7155840684309007e-09, "loss": 0.063, "step": 3603 }, { "epoch": 0.96, "learning_rate": 3.688853247794707e-09, "loss": 0.0991, "step": 3604 }, { "epoch": 0.96, "learning_rate": 3.6621224271585134e-09, "loss": 0.0989, "step": 3605 }, { "epoch": 0.96, "learning_rate": 3.63539160652232e-09, "loss": 0.0705, "step": 3606 }, { "epoch": 0.96, "learning_rate": 3.6086607858861266e-09, "loss": 0.1064, "step": 3607 }, { "epoch": 0.96, "learning_rate": 3.581929965249933e-09, "loss": 0.059, "step": 3608 }, { "epoch": 0.96, "learning_rate": 3.5551991446137394e-09, "loss": 0.065, "step": 3609 }, { "epoch": 0.96, "learning_rate": 3.528468323977546e-09, "loss": 0.0821, "step": 3610 }, { "epoch": 0.97, "learning_rate": 3.5017375033413522e-09, "loss": 0.0499, "step": 3611 }, { "epoch": 0.97, "learning_rate": 3.475006682705159e-09, "loss": 0.0952, "step": 3612 }, { "epoch": 0.97, "learning_rate": 3.4482758620689654e-09, "loss": 0.059, "step": 3613 }, { "epoch": 0.97, "learning_rate": 3.421545041432772e-09, "loss": 0.0556, "step": 3614 }, { "epoch": 0.97, "learning_rate": 3.394814220796578e-09, "loss": 0.0818, "step": 3615 }, { "epoch": 0.97, "learning_rate": 3.3680834001603846e-09, "loss": 0.0779, "step": 3616 }, { "epoch": 0.97, "learning_rate": 3.341352579524191e-09, "loss": 0.0754, "step": 3617 }, { "epoch": 0.97, "learning_rate": 3.314621758887998e-09, "loss": 0.0743, "step": 3618 }, { "epoch": 0.97, "learning_rate": 3.287890938251804e-09, "loss": 0.0771, "step": 3619 }, { "epoch": 0.97, "learning_rate": 3.2611601176156106e-09, "loss": 0.0801, "step": 3620 }, { "epoch": 0.97, "learning_rate": 3.234429296979417e-09, "loss": 0.0761, "step": 3621 }, { "epoch": 0.97, "learning_rate": 3.2076984763432234e-09, "loss": 0.0736, "step": 3622 }, { "epoch": 0.97, "learning_rate": 3.18096765570703e-09, "loss": 0.0682, "step": 3623 }, { "epoch": 0.97, "learning_rate": 3.1542368350708366e-09, "loss": 0.0675, "step": 3624 }, { "epoch": 0.97, "learning_rate": 3.127506014434643e-09, "loss": 0.067, "step": 3625 }, { "epoch": 0.97, "learning_rate": 3.1007751937984494e-09, "loss": 0.0701, "step": 3626 }, { "epoch": 0.97, "learning_rate": 3.0740443731622557e-09, "loss": 0.056, "step": 3627 }, { "epoch": 0.97, "learning_rate": 3.0473135525260625e-09, "loss": 0.0857, "step": 3628 }, { "epoch": 0.97, "learning_rate": 3.020582731889869e-09, "loss": 0.0741, "step": 3629 }, { "epoch": 0.97, "learning_rate": 2.9938519112536753e-09, "loss": 0.0708, "step": 3630 }, { "epoch": 0.97, "learning_rate": 2.9671210906174817e-09, "loss": 0.0801, "step": 3631 }, { "epoch": 0.97, "learning_rate": 2.940390269981288e-09, "loss": 0.065, "step": 3632 }, { "epoch": 0.97, "learning_rate": 2.913659449345095e-09, "loss": 0.0764, "step": 3633 }, { "epoch": 0.97, "learning_rate": 2.8869286287089013e-09, "loss": 0.0771, "step": 3634 }, { "epoch": 0.97, "learning_rate": 2.8601978080727077e-09, "loss": 0.0609, "step": 3635 }, { "epoch": 0.97, "learning_rate": 2.833466987436514e-09, "loss": 0.0747, "step": 3636 }, { "epoch": 0.97, "learning_rate": 2.8067361668003205e-09, "loss": 0.0691, "step": 3637 }, { "epoch": 0.97, "learning_rate": 2.780005346164127e-09, "loss": 0.0799, "step": 3638 }, { "epoch": 0.97, "learning_rate": 2.7532745255279337e-09, "loss": 0.0728, "step": 3639 }, { "epoch": 0.97, "learning_rate": 2.72654370489174e-09, "loss": 0.0849, "step": 3640 }, { "epoch": 0.97, "learning_rate": 2.6998128842555465e-09, "loss": 0.09, "step": 3641 }, { "epoch": 0.97, "learning_rate": 2.673082063619353e-09, "loss": 0.075, "step": 3642 }, { "epoch": 0.97, "learning_rate": 2.6463512429831593e-09, "loss": 0.086, "step": 3643 }, { "epoch": 0.97, "learning_rate": 2.619620422346966e-09, "loss": 0.0648, "step": 3644 }, { "epoch": 0.97, "learning_rate": 2.5928896017107725e-09, "loss": 0.0761, "step": 3645 }, { "epoch": 0.97, "learning_rate": 2.566158781074579e-09, "loss": 0.0647, "step": 3646 }, { "epoch": 0.97, "learning_rate": 2.5394279604383853e-09, "loss": 0.0523, "step": 3647 }, { "epoch": 0.98, "learning_rate": 2.5126971398021916e-09, "loss": 0.0803, "step": 3648 }, { "epoch": 0.98, "learning_rate": 2.4859663191659984e-09, "loss": 0.0692, "step": 3649 }, { "epoch": 0.98, "learning_rate": 2.459235498529805e-09, "loss": 0.0764, "step": 3650 }, { "epoch": 0.98, "learning_rate": 2.4325046778936112e-09, "loss": 0.0869, "step": 3651 }, { "epoch": 0.98, "learning_rate": 2.4057738572574176e-09, "loss": 0.0716, "step": 3652 }, { "epoch": 0.98, "learning_rate": 2.379043036621224e-09, "loss": 0.0727, "step": 3653 }, { "epoch": 0.98, "learning_rate": 2.3523122159850304e-09, "loss": 0.0704, "step": 3654 }, { "epoch": 0.98, "learning_rate": 2.3255813953488372e-09, "loss": 0.0785, "step": 3655 }, { "epoch": 0.98, "learning_rate": 2.2988505747126436e-09, "loss": 0.0867, "step": 3656 }, { "epoch": 0.98, "learning_rate": 2.27211975407645e-09, "loss": 0.0616, "step": 3657 }, { "epoch": 0.98, "learning_rate": 2.2453889334402564e-09, "loss": 0.0732, "step": 3658 }, { "epoch": 0.98, "learning_rate": 2.2186581128040628e-09, "loss": 0.0724, "step": 3659 }, { "epoch": 0.98, "learning_rate": 2.1919272921678696e-09, "loss": 0.0995, "step": 3660 }, { "epoch": 0.98, "learning_rate": 2.165196471531676e-09, "loss": 0.0718, "step": 3661 }, { "epoch": 0.98, "learning_rate": 2.1384656508954824e-09, "loss": 0.0816, "step": 3662 }, { "epoch": 0.98, "learning_rate": 2.1117348302592888e-09, "loss": 0.0839, "step": 3663 }, { "epoch": 0.98, "learning_rate": 2.085004009623095e-09, "loss": 0.0709, "step": 3664 }, { "epoch": 0.98, "learning_rate": 2.058273188986902e-09, "loss": 0.0601, "step": 3665 }, { "epoch": 0.98, "learning_rate": 2.0315423683507084e-09, "loss": 0.0796, "step": 3666 }, { "epoch": 0.98, "learning_rate": 2.0048115477145148e-09, "loss": 0.0845, "step": 3667 }, { "epoch": 0.98, "learning_rate": 1.978080727078321e-09, "loss": 0.0801, "step": 3668 }, { "epoch": 0.98, "learning_rate": 1.9513499064421275e-09, "loss": 0.0685, "step": 3669 }, { "epoch": 0.98, "learning_rate": 1.924619085805934e-09, "loss": 0.0645, "step": 3670 }, { "epoch": 0.98, "learning_rate": 1.8978882651697407e-09, "loss": 0.064, "step": 3671 }, { "epoch": 0.98, "learning_rate": 1.871157444533547e-09, "loss": 0.1019, "step": 3672 }, { "epoch": 0.98, "learning_rate": 1.8444266238973535e-09, "loss": 0.0675, "step": 3673 }, { "epoch": 0.98, "learning_rate": 1.81769580326116e-09, "loss": 0.0761, "step": 3674 }, { "epoch": 0.98, "learning_rate": 1.7909649826249665e-09, "loss": 0.0549, "step": 3675 }, { "epoch": 0.98, "learning_rate": 1.764234161988773e-09, "loss": 0.0751, "step": 3676 }, { "epoch": 0.98, "learning_rate": 1.7375033413525795e-09, "loss": 0.0671, "step": 3677 }, { "epoch": 0.98, "learning_rate": 1.710772520716386e-09, "loss": 0.1178, "step": 3678 }, { "epoch": 0.98, "learning_rate": 1.6840417000801923e-09, "loss": 0.0655, "step": 3679 }, { "epoch": 0.98, "learning_rate": 1.657310879443999e-09, "loss": 0.0647, "step": 3680 }, { "epoch": 0.98, "learning_rate": 1.6305800588078053e-09, "loss": 0.0757, "step": 3681 }, { "epoch": 0.98, "learning_rate": 1.6038492381716117e-09, "loss": 0.0786, "step": 3682 }, { "epoch": 0.98, "learning_rate": 1.5771184175354183e-09, "loss": 0.0534, "step": 3683 }, { "epoch": 0.98, "learning_rate": 1.5503875968992247e-09, "loss": 0.094, "step": 3684 }, { "epoch": 0.98, "learning_rate": 1.5236567762630313e-09, "loss": 0.0682, "step": 3685 }, { "epoch": 0.99, "learning_rate": 1.4969259556268377e-09, "loss": 0.0731, "step": 3686 }, { "epoch": 0.99, "learning_rate": 1.470195134990644e-09, "loss": 0.0674, "step": 3687 }, { "epoch": 0.99, "learning_rate": 1.4434643143544507e-09, "loss": 0.0735, "step": 3688 }, { "epoch": 0.99, "learning_rate": 1.416733493718257e-09, "loss": 0.0543, "step": 3689 }, { "epoch": 0.99, "learning_rate": 1.3900026730820634e-09, "loss": 0.092, "step": 3690 }, { "epoch": 0.99, "learning_rate": 1.36327185244587e-09, "loss": 0.0774, "step": 3691 }, { "epoch": 0.99, "learning_rate": 1.3365410318096764e-09, "loss": 0.0727, "step": 3692 }, { "epoch": 0.99, "learning_rate": 1.309810211173483e-09, "loss": 0.0715, "step": 3693 }, { "epoch": 0.99, "learning_rate": 1.2830793905372894e-09, "loss": 0.1124, "step": 3694 }, { "epoch": 0.99, "learning_rate": 1.2563485699010958e-09, "loss": 0.0508, "step": 3695 }, { "epoch": 0.99, "learning_rate": 1.2296177492649024e-09, "loss": 0.0935, "step": 3696 }, { "epoch": 0.99, "learning_rate": 1.2028869286287088e-09, "loss": 0.0906, "step": 3697 }, { "epoch": 0.99, "learning_rate": 1.1761561079925152e-09, "loss": 0.0966, "step": 3698 }, { "epoch": 0.99, "learning_rate": 1.1494252873563218e-09, "loss": 0.0653, "step": 3699 }, { "epoch": 0.99, "learning_rate": 1.1226944667201282e-09, "loss": 0.0747, "step": 3700 }, { "epoch": 0.99, "learning_rate": 1.0959636460839348e-09, "loss": 0.0582, "step": 3701 }, { "epoch": 0.99, "learning_rate": 1.0692328254477412e-09, "loss": 0.0635, "step": 3702 }, { "epoch": 0.99, "learning_rate": 1.0425020048115476e-09, "loss": 0.0592, "step": 3703 }, { "epoch": 0.99, "learning_rate": 1.0157711841753542e-09, "loss": 0.0881, "step": 3704 }, { "epoch": 0.99, "learning_rate": 9.890403635391606e-10, "loss": 0.0797, "step": 3705 }, { "epoch": 0.99, "learning_rate": 9.62309542902967e-10, "loss": 0.0956, "step": 3706 }, { "epoch": 0.99, "learning_rate": 9.355787222667736e-10, "loss": 0.0649, "step": 3707 }, { "epoch": 0.99, "learning_rate": 9.0884790163058e-10, "loss": 0.0792, "step": 3708 }, { "epoch": 0.99, "learning_rate": 8.821170809943865e-10, "loss": 0.102, "step": 3709 }, { "epoch": 0.99, "learning_rate": 8.55386260358193e-10, "loss": 0.066, "step": 3710 }, { "epoch": 0.99, "learning_rate": 8.286554397219994e-10, "loss": 0.0749, "step": 3711 }, { "epoch": 0.99, "learning_rate": 8.019246190858058e-10, "loss": 0.0648, "step": 3712 }, { "epoch": 0.99, "learning_rate": 7.751937984496123e-10, "loss": 0.064, "step": 3713 }, { "epoch": 0.99, "learning_rate": 7.484629778134188e-10, "loss": 0.0713, "step": 3714 }, { "epoch": 0.99, "learning_rate": 7.217321571772253e-10, "loss": 0.0676, "step": 3715 }, { "epoch": 0.99, "learning_rate": 6.950013365410317e-10, "loss": 0.0673, "step": 3716 }, { "epoch": 0.99, "learning_rate": 6.682705159048382e-10, "loss": 0.08, "step": 3717 }, { "epoch": 0.99, "learning_rate": 6.415396952686447e-10, "loss": 0.0702, "step": 3718 }, { "epoch": 0.99, "learning_rate": 6.148088746324512e-10, "loss": 0.0634, "step": 3719 }, { "epoch": 0.99, "learning_rate": 5.880780539962576e-10, "loss": 0.0777, "step": 3720 }, { "epoch": 0.99, "learning_rate": 5.613472333600641e-10, "loss": 0.0549, "step": 3721 }, { "epoch": 0.99, "learning_rate": 5.346164127238706e-10, "loss": 0.0578, "step": 3722 }, { "epoch": 1.0, "learning_rate": 5.078855920876771e-10, "loss": 0.0688, "step": 3723 }, { "epoch": 1.0, "learning_rate": 4.811547714514835e-10, "loss": 0.0753, "step": 3724 }, { "epoch": 1.0, "learning_rate": 4.5442395081529e-10, "loss": 0.0649, "step": 3725 }, { "epoch": 1.0, "learning_rate": 4.276931301790965e-10, "loss": 0.0799, "step": 3726 }, { "epoch": 1.0, "learning_rate": 4.009623095429029e-10, "loss": 0.0654, "step": 3727 }, { "epoch": 1.0, "learning_rate": 3.742314889067094e-10, "loss": 0.065, "step": 3728 }, { "epoch": 1.0, "learning_rate": 3.4750066827051586e-10, "loss": 0.0871, "step": 3729 }, { "epoch": 1.0, "learning_rate": 3.2076984763432236e-10, "loss": 0.0647, "step": 3730 }, { "epoch": 1.0, "learning_rate": 2.940390269981288e-10, "loss": 0.0897, "step": 3731 }, { "epoch": 1.0, "learning_rate": 2.673082063619353e-10, "loss": 0.0919, "step": 3732 }, { "epoch": 1.0, "learning_rate": 2.4057738572574174e-10, "loss": 0.0646, "step": 3733 }, { "epoch": 1.0, "learning_rate": 2.1384656508954824e-10, "loss": 0.0766, "step": 3734 }, { "epoch": 1.0, "learning_rate": 1.871157444533547e-10, "loss": 0.1045, "step": 3735 }, { "epoch": 1.0, "learning_rate": 1.6038492381716118e-10, "loss": 0.0835, "step": 3736 }, { "epoch": 1.0, "learning_rate": 1.3365410318096765e-10, "loss": 0.0589, "step": 3737 }, { "epoch": 1.0, "learning_rate": 1.0692328254477412e-10, "loss": 0.0535, "step": 3738 }, { "epoch": 1.0, "learning_rate": 8.019246190858059e-11, "loss": 0.0816, "step": 3739 }, { "epoch": 1.0, "learning_rate": 5.346164127238706e-11, "loss": 0.0767, "step": 3740 }, { "epoch": 1.0, "learning_rate": 2.673082063619353e-11, "loss": 0.0665, "step": 3741 }, { "epoch": 1.0, "eval_loss": 0.15256968140602112, "eval_runtime": 4078.9054, "eval_samples_per_second": 3.937, "eval_steps_per_second": 0.123, "eval_wer": 0.20562431991294886, "step": 3741 }, { "epoch": 1.0, "step": 3741, "total_flos": 3.094314205418291e+20, "train_loss": 0.06809680537185991, "train_runtime": 120164.0372, "train_samples_per_second": 3.985, "train_steps_per_second": 0.031 } ], "max_steps": 3741, "num_train_epochs": 1, "total_flos": 3.094314205418291e+20, "trial_name": null, "trial_params": null }