{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "global_step": 7483, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 9.99866363757851e-08, "loss": 0.0316, "step": 1 }, { "epoch": 0.0, "learning_rate": 9.997327275157023e-08, "loss": 0.0345, "step": 2 }, { "epoch": 0.0, "learning_rate": 9.995990912735534e-08, "loss": 0.0698, "step": 3 }, { "epoch": 0.0, "learning_rate": 9.994654550314045e-08, "loss": 0.0304, "step": 4 }, { "epoch": 0.0, "learning_rate": 9.993318187892557e-08, "loss": 0.0403, "step": 5 }, { "epoch": 0.0, "learning_rate": 9.991981825471068e-08, "loss": 0.0412, "step": 6 }, { "epoch": 0.0, "learning_rate": 9.990645463049579e-08, "loss": 0.0463, "step": 7 }, { "epoch": 0.0, "learning_rate": 9.98930910062809e-08, "loss": 0.0469, "step": 8 }, { "epoch": 0.0, "learning_rate": 9.987972738206601e-08, "loss": 0.0579, "step": 9 }, { "epoch": 0.0, "learning_rate": 9.986636375785112e-08, "loss": 0.0582, "step": 10 }, { "epoch": 0.0, "learning_rate": 9.985300013363623e-08, "loss": 0.0386, "step": 11 }, { "epoch": 0.0, "learning_rate": 9.983963650942135e-08, "loss": 0.0509, "step": 12 }, { "epoch": 0.0, "learning_rate": 9.982627288520646e-08, "loss": 0.0345, "step": 13 }, { "epoch": 0.0, "learning_rate": 9.981290926099157e-08, "loss": 0.0477, "step": 14 }, { "epoch": 0.0, "learning_rate": 9.97995456367767e-08, "loss": 0.0652, "step": 15 }, { "epoch": 0.0, "learning_rate": 9.97861820125618e-08, "loss": 0.0681, "step": 16 }, { "epoch": 0.0, "learning_rate": 9.977281838834692e-08, "loss": 0.0609, "step": 17 }, { "epoch": 0.0, "learning_rate": 9.975945476413203e-08, "loss": 0.0528, "step": 18 }, { "epoch": 0.0, "learning_rate": 9.974609113991715e-08, "loss": 0.062, "step": 19 }, { "epoch": 0.0, "learning_rate": 9.973272751570226e-08, "loss": 0.0435, "step": 20 }, { "epoch": 0.0, "learning_rate": 9.971936389148737e-08, "loss": 0.0378, "step": 21 }, { "epoch": 0.0, "learning_rate": 9.970600026727248e-08, "loss": 0.0403, "step": 22 }, { "epoch": 0.0, "learning_rate": 9.969263664305759e-08, "loss": 0.0698, "step": 23 }, { "epoch": 0.0, "learning_rate": 9.96792730188427e-08, "loss": 0.07, "step": 24 }, { "epoch": 0.0, "learning_rate": 9.966590939462782e-08, "loss": 0.0548, "step": 25 }, { "epoch": 0.0, "learning_rate": 9.965254577041293e-08, "loss": 0.0258, "step": 26 }, { "epoch": 0.0, "learning_rate": 9.963918214619804e-08, "loss": 0.0523, "step": 27 }, { "epoch": 0.0, "learning_rate": 9.962581852198316e-08, "loss": 0.0383, "step": 28 }, { "epoch": 0.0, "learning_rate": 9.961245489776827e-08, "loss": 0.0508, "step": 29 }, { "epoch": 0.0, "learning_rate": 9.959909127355338e-08, "loss": 0.0455, "step": 30 }, { "epoch": 0.0, "learning_rate": 9.958572764933849e-08, "loss": 0.0421, "step": 31 }, { "epoch": 0.0, "learning_rate": 9.957236402512362e-08, "loss": 0.0547, "step": 32 }, { "epoch": 0.0, "learning_rate": 9.955900040090873e-08, "loss": 0.0603, "step": 33 }, { "epoch": 0.0, "learning_rate": 9.954563677669384e-08, "loss": 0.0491, "step": 34 }, { "epoch": 0.0, "learning_rate": 9.953227315247895e-08, "loss": 0.0537, "step": 35 }, { "epoch": 0.0, "learning_rate": 9.951890952826406e-08, "loss": 0.0741, "step": 36 }, { "epoch": 0.0, "learning_rate": 9.950554590404917e-08, "loss": 0.0307, "step": 37 }, { "epoch": 0.01, "learning_rate": 9.949218227983429e-08, "loss": 0.0427, "step": 38 }, { "epoch": 0.01, "learning_rate": 9.94788186556194e-08, "loss": 0.0727, "step": 39 }, { "epoch": 0.01, "learning_rate": 9.946545503140451e-08, "loss": 0.0526, "step": 40 }, { "epoch": 0.01, "learning_rate": 9.945209140718962e-08, "loss": 0.0554, "step": 41 }, { "epoch": 0.01, "learning_rate": 9.943872778297474e-08, "loss": 0.0405, "step": 42 }, { "epoch": 0.01, "learning_rate": 9.942536415875985e-08, "loss": 0.0781, "step": 43 }, { "epoch": 0.01, "learning_rate": 9.941200053454496e-08, "loss": 0.0606, "step": 44 }, { "epoch": 0.01, "learning_rate": 9.939863691033008e-08, "loss": 0.0438, "step": 45 }, { "epoch": 0.01, "learning_rate": 9.93852732861152e-08, "loss": 0.0474, "step": 46 }, { "epoch": 0.01, "learning_rate": 9.93719096619003e-08, "loss": 0.0417, "step": 47 }, { "epoch": 0.01, "learning_rate": 9.935854603768543e-08, "loss": 0.0484, "step": 48 }, { "epoch": 0.01, "learning_rate": 9.934518241347054e-08, "loss": 0.0512, "step": 49 }, { "epoch": 0.01, "learning_rate": 9.933181878925563e-08, "loss": 0.036, "step": 50 }, { "epoch": 0.01, "learning_rate": 9.931845516504076e-08, "loss": 0.06, "step": 51 }, { "epoch": 0.01, "learning_rate": 9.930509154082587e-08, "loss": 0.0489, "step": 52 }, { "epoch": 0.01, "learning_rate": 9.929172791661098e-08, "loss": 0.0444, "step": 53 }, { "epoch": 0.01, "learning_rate": 9.927836429239609e-08, "loss": 0.048, "step": 54 }, { "epoch": 0.01, "learning_rate": 9.926500066818121e-08, "loss": 0.0463, "step": 55 }, { "epoch": 0.01, "learning_rate": 9.925163704396632e-08, "loss": 0.0665, "step": 56 }, { "epoch": 0.01, "learning_rate": 9.923827341975143e-08, "loss": 0.0743, "step": 57 }, { "epoch": 0.01, "learning_rate": 9.922490979553655e-08, "loss": 0.049, "step": 58 }, { "epoch": 0.01, "learning_rate": 9.921154617132166e-08, "loss": 0.053, "step": 59 }, { "epoch": 0.01, "learning_rate": 9.919818254710677e-08, "loss": 0.042, "step": 60 }, { "epoch": 0.01, "learning_rate": 9.91848189228919e-08, "loss": 0.056, "step": 61 }, { "epoch": 0.01, "learning_rate": 9.9171455298677e-08, "loss": 0.043, "step": 62 }, { "epoch": 0.01, "learning_rate": 9.915809167446211e-08, "loss": 0.0535, "step": 63 }, { "epoch": 0.01, "learning_rate": 9.914472805024722e-08, "loss": 0.0613, "step": 64 }, { "epoch": 0.01, "learning_rate": 9.913136442603233e-08, "loss": 0.0528, "step": 65 }, { "epoch": 0.01, "learning_rate": 9.911800080181744e-08, "loss": 0.042, "step": 66 }, { "epoch": 0.01, "learning_rate": 9.910463717760255e-08, "loss": 0.0874, "step": 67 }, { "epoch": 0.01, "learning_rate": 9.909127355338768e-08, "loss": 0.038, "step": 68 }, { "epoch": 0.01, "learning_rate": 9.907790992917279e-08, "loss": 0.0547, "step": 69 }, { "epoch": 0.01, "learning_rate": 9.90645463049579e-08, "loss": 0.0326, "step": 70 }, { "epoch": 0.01, "learning_rate": 9.905118268074302e-08, "loss": 0.0493, "step": 71 }, { "epoch": 0.01, "learning_rate": 9.903781905652813e-08, "loss": 0.0551, "step": 72 }, { "epoch": 0.01, "learning_rate": 9.902445543231324e-08, "loss": 0.0476, "step": 73 }, { "epoch": 0.01, "learning_rate": 9.901109180809836e-08, "loss": 0.0416, "step": 74 }, { "epoch": 0.01, "learning_rate": 9.899772818388347e-08, "loss": 0.033, "step": 75 }, { "epoch": 0.01, "learning_rate": 9.898436455966858e-08, "loss": 0.0368, "step": 76 }, { "epoch": 0.01, "learning_rate": 9.897100093545369e-08, "loss": 0.0434, "step": 77 }, { "epoch": 0.01, "learning_rate": 9.89576373112388e-08, "loss": 0.0536, "step": 78 }, { "epoch": 0.01, "learning_rate": 9.894427368702391e-08, "loss": 0.0534, "step": 79 }, { "epoch": 0.01, "learning_rate": 9.893091006280902e-08, "loss": 0.0547, "step": 80 }, { "epoch": 0.01, "learning_rate": 9.891754643859414e-08, "loss": 0.0605, "step": 81 }, { "epoch": 0.01, "learning_rate": 9.890418281437925e-08, "loss": 0.0465, "step": 82 }, { "epoch": 0.01, "learning_rate": 9.889081919016436e-08, "loss": 0.0627, "step": 83 }, { "epoch": 0.01, "learning_rate": 9.887745556594949e-08, "loss": 0.0381, "step": 84 }, { "epoch": 0.01, "learning_rate": 9.88640919417346e-08, "loss": 0.0215, "step": 85 }, { "epoch": 0.01, "learning_rate": 9.88507283175197e-08, "loss": 0.0451, "step": 86 }, { "epoch": 0.01, "learning_rate": 9.883736469330482e-08, "loss": 0.0383, "step": 87 }, { "epoch": 0.01, "learning_rate": 9.882400106908994e-08, "loss": 0.0426, "step": 88 }, { "epoch": 0.01, "learning_rate": 9.881063744487505e-08, "loss": 0.0421, "step": 89 }, { "epoch": 0.01, "learning_rate": 9.879727382066016e-08, "loss": 0.036, "step": 90 }, { "epoch": 0.01, "learning_rate": 9.878391019644527e-08, "loss": 0.0638, "step": 91 }, { "epoch": 0.01, "learning_rate": 9.877054657223038e-08, "loss": 0.0499, "step": 92 }, { "epoch": 0.01, "learning_rate": 9.875718294801549e-08, "loss": 0.0348, "step": 93 }, { "epoch": 0.01, "learning_rate": 9.874381932380061e-08, "loss": 0.0415, "step": 94 }, { "epoch": 0.01, "learning_rate": 9.873045569958572e-08, "loss": 0.0597, "step": 95 }, { "epoch": 0.01, "learning_rate": 9.871709207537083e-08, "loss": 0.0352, "step": 96 }, { "epoch": 0.01, "learning_rate": 9.870372845115595e-08, "loss": 0.0425, "step": 97 }, { "epoch": 0.01, "learning_rate": 9.869036482694106e-08, "loss": 0.0908, "step": 98 }, { "epoch": 0.01, "learning_rate": 9.867700120272617e-08, "loss": 0.0565, "step": 99 }, { "epoch": 0.01, "learning_rate": 9.866363757851128e-08, "loss": 0.0466, "step": 100 }, { "epoch": 0.01, "learning_rate": 9.86502739542964e-08, "loss": 0.1067, "step": 101 }, { "epoch": 0.01, "learning_rate": 9.863691033008152e-08, "loss": 0.0629, "step": 102 }, { "epoch": 0.01, "learning_rate": 9.862354670586663e-08, "loss": 0.0631, "step": 103 }, { "epoch": 0.01, "learning_rate": 9.861018308165175e-08, "loss": 0.0829, "step": 104 }, { "epoch": 0.01, "learning_rate": 9.859681945743686e-08, "loss": 0.0417, "step": 105 }, { "epoch": 0.01, "learning_rate": 9.858345583322197e-08, "loss": 0.0369, "step": 106 }, { "epoch": 0.01, "learning_rate": 9.857009220900708e-08, "loss": 0.04, "step": 107 }, { "epoch": 0.01, "learning_rate": 9.855672858479219e-08, "loss": 0.0556, "step": 108 }, { "epoch": 0.01, "learning_rate": 9.85433649605773e-08, "loss": 0.0669, "step": 109 }, { "epoch": 0.01, "learning_rate": 9.853000133636241e-08, "loss": 0.0444, "step": 110 }, { "epoch": 0.01, "learning_rate": 9.851663771214753e-08, "loss": 0.0477, "step": 111 }, { "epoch": 0.01, "learning_rate": 9.850327408793264e-08, "loss": 0.0495, "step": 112 }, { "epoch": 0.02, "learning_rate": 9.848991046371775e-08, "loss": 0.0831, "step": 113 }, { "epoch": 0.02, "learning_rate": 9.847654683950287e-08, "loss": 0.0597, "step": 114 }, { "epoch": 0.02, "learning_rate": 9.846318321528798e-08, "loss": 0.0714, "step": 115 }, { "epoch": 0.02, "learning_rate": 9.84498195910731e-08, "loss": 0.0528, "step": 116 }, { "epoch": 0.02, "learning_rate": 9.843645596685822e-08, "loss": 0.0552, "step": 117 }, { "epoch": 0.02, "learning_rate": 9.842309234264333e-08, "loss": 0.0473, "step": 118 }, { "epoch": 0.02, "learning_rate": 9.840972871842844e-08, "loss": 0.0491, "step": 119 }, { "epoch": 0.02, "learning_rate": 9.839636509421355e-08, "loss": 0.0271, "step": 120 }, { "epoch": 0.02, "learning_rate": 9.838300146999866e-08, "loss": 0.0474, "step": 121 }, { "epoch": 0.02, "learning_rate": 9.836963784578377e-08, "loss": 0.0318, "step": 122 }, { "epoch": 0.02, "learning_rate": 9.835627422156888e-08, "loss": 0.074, "step": 123 }, { "epoch": 0.02, "learning_rate": 9.8342910597354e-08, "loss": 0.0584, "step": 124 }, { "epoch": 0.02, "learning_rate": 9.832954697313911e-08, "loss": 0.046, "step": 125 }, { "epoch": 0.02, "learning_rate": 9.831618334892422e-08, "loss": 0.0343, "step": 126 }, { "epoch": 0.02, "learning_rate": 9.830281972470934e-08, "loss": 0.0515, "step": 127 }, { "epoch": 0.02, "learning_rate": 9.828945610049445e-08, "loss": 0.0584, "step": 128 }, { "epoch": 0.02, "learning_rate": 9.827609247627956e-08, "loss": 0.0625, "step": 129 }, { "epoch": 0.02, "learning_rate": 9.826272885206468e-08, "loss": 0.0513, "step": 130 }, { "epoch": 0.02, "learning_rate": 9.82493652278498e-08, "loss": 0.0606, "step": 131 }, { "epoch": 0.02, "learning_rate": 9.82360016036349e-08, "loss": 0.0472, "step": 132 }, { "epoch": 0.02, "learning_rate": 9.822263797942001e-08, "loss": 0.0447, "step": 133 }, { "epoch": 0.02, "learning_rate": 9.820927435520512e-08, "loss": 0.0473, "step": 134 }, { "epoch": 0.02, "learning_rate": 9.819591073099023e-08, "loss": 0.0532, "step": 135 }, { "epoch": 0.02, "learning_rate": 9.818254710677534e-08, "loss": 0.0698, "step": 136 }, { "epoch": 0.02, "learning_rate": 9.816918348256047e-08, "loss": 0.0662, "step": 137 }, { "epoch": 0.02, "learning_rate": 9.815581985834558e-08, "loss": 0.0804, "step": 138 }, { "epoch": 0.02, "learning_rate": 9.814245623413069e-08, "loss": 0.0547, "step": 139 }, { "epoch": 0.02, "learning_rate": 9.812909260991581e-08, "loss": 0.0713, "step": 140 }, { "epoch": 0.02, "learning_rate": 9.811572898570092e-08, "loss": 0.0473, "step": 141 }, { "epoch": 0.02, "learning_rate": 9.810236536148603e-08, "loss": 0.047, "step": 142 }, { "epoch": 0.02, "learning_rate": 9.808900173727115e-08, "loss": 0.042, "step": 143 }, { "epoch": 0.02, "learning_rate": 9.807563811305626e-08, "loss": 0.0691, "step": 144 }, { "epoch": 0.02, "learning_rate": 9.806227448884137e-08, "loss": 0.0568, "step": 145 }, { "epoch": 0.02, "learning_rate": 9.804891086462648e-08, "loss": 0.049, "step": 146 }, { "epoch": 0.02, "learning_rate": 9.80355472404116e-08, "loss": 0.0431, "step": 147 }, { "epoch": 0.02, "learning_rate": 9.802218361619671e-08, "loss": 0.0383, "step": 148 }, { "epoch": 0.02, "learning_rate": 9.800881999198181e-08, "loss": 0.0502, "step": 149 }, { "epoch": 0.02, "learning_rate": 9.799545636776693e-08, "loss": 0.0435, "step": 150 }, { "epoch": 0.02, "learning_rate": 9.798209274355204e-08, "loss": 0.0662, "step": 151 }, { "epoch": 0.02, "learning_rate": 9.796872911933715e-08, "loss": 0.0675, "step": 152 }, { "epoch": 0.02, "learning_rate": 9.795536549512228e-08, "loss": 0.0292, "step": 153 }, { "epoch": 0.02, "learning_rate": 9.794200187090739e-08, "loss": 0.0517, "step": 154 }, { "epoch": 0.02, "learning_rate": 9.79286382466925e-08, "loss": 0.0455, "step": 155 }, { "epoch": 0.02, "learning_rate": 9.79152746224776e-08, "loss": 0.0704, "step": 156 }, { "epoch": 0.02, "learning_rate": 9.790191099826273e-08, "loss": 0.0496, "step": 157 }, { "epoch": 0.02, "learning_rate": 9.788854737404784e-08, "loss": 0.0615, "step": 158 }, { "epoch": 0.02, "learning_rate": 9.787518374983295e-08, "loss": 0.0561, "step": 159 }, { "epoch": 0.02, "learning_rate": 9.786182012561807e-08, "loss": 0.0428, "step": 160 }, { "epoch": 0.02, "learning_rate": 9.784845650140318e-08, "loss": 0.0484, "step": 161 }, { "epoch": 0.02, "learning_rate": 9.783509287718829e-08, "loss": 0.057, "step": 162 }, { "epoch": 0.02, "learning_rate": 9.78217292529734e-08, "loss": 0.0883, "step": 163 }, { "epoch": 0.02, "learning_rate": 9.780836562875851e-08, "loss": 0.0462, "step": 164 }, { "epoch": 0.02, "learning_rate": 9.779500200454362e-08, "loss": 0.0497, "step": 165 }, { "epoch": 0.02, "learning_rate": 9.778163838032874e-08, "loss": 0.0469, "step": 166 }, { "epoch": 0.02, "learning_rate": 9.776827475611385e-08, "loss": 0.0484, "step": 167 }, { "epoch": 0.02, "learning_rate": 9.775491113189896e-08, "loss": 0.0782, "step": 168 }, { "epoch": 0.02, "learning_rate": 9.774154750768407e-08, "loss": 0.0774, "step": 169 }, { "epoch": 0.02, "learning_rate": 9.77281838834692e-08, "loss": 0.0297, "step": 170 }, { "epoch": 0.02, "learning_rate": 9.77148202592543e-08, "loss": 0.0459, "step": 171 }, { "epoch": 0.02, "learning_rate": 9.770145663503942e-08, "loss": 0.0485, "step": 172 }, { "epoch": 0.02, "learning_rate": 9.768809301082454e-08, "loss": 0.0644, "step": 173 }, { "epoch": 0.02, "learning_rate": 9.767472938660965e-08, "loss": 0.0725, "step": 174 }, { "epoch": 0.02, "learning_rate": 9.766136576239476e-08, "loss": 0.0368, "step": 175 }, { "epoch": 0.02, "learning_rate": 9.764800213817987e-08, "loss": 0.0369, "step": 176 }, { "epoch": 0.02, "learning_rate": 9.763463851396498e-08, "loss": 0.0517, "step": 177 }, { "epoch": 0.02, "learning_rate": 9.762127488975009e-08, "loss": 0.0395, "step": 178 }, { "epoch": 0.02, "learning_rate": 9.76079112655352e-08, "loss": 0.0537, "step": 179 }, { "epoch": 0.02, "learning_rate": 9.759454764132032e-08, "loss": 0.0531, "step": 180 }, { "epoch": 0.02, "learning_rate": 9.758118401710543e-08, "loss": 0.0608, "step": 181 }, { "epoch": 0.02, "learning_rate": 9.756782039289054e-08, "loss": 0.0545, "step": 182 }, { "epoch": 0.02, "learning_rate": 9.755445676867566e-08, "loss": 0.0437, "step": 183 }, { "epoch": 0.02, "learning_rate": 9.754109314446077e-08, "loss": 0.0654, "step": 184 }, { "epoch": 0.02, "learning_rate": 9.752772952024588e-08, "loss": 0.0436, "step": 185 }, { "epoch": 0.02, "learning_rate": 9.751436589603101e-08, "loss": 0.0522, "step": 186 }, { "epoch": 0.02, "learning_rate": 9.750100227181612e-08, "loss": 0.0341, "step": 187 }, { "epoch": 0.03, "learning_rate": 9.748763864760123e-08, "loss": 0.0329, "step": 188 }, { "epoch": 0.03, "learning_rate": 9.747427502338635e-08, "loss": 0.0551, "step": 189 }, { "epoch": 0.03, "learning_rate": 9.746091139917145e-08, "loss": 0.072, "step": 190 }, { "epoch": 0.03, "learning_rate": 9.744754777495656e-08, "loss": 0.0461, "step": 191 }, { "epoch": 0.03, "learning_rate": 9.743418415074167e-08, "loss": 0.036, "step": 192 }, { "epoch": 0.03, "learning_rate": 9.742082052652679e-08, "loss": 0.0429, "step": 193 }, { "epoch": 0.03, "learning_rate": 9.74074569023119e-08, "loss": 0.0245, "step": 194 }, { "epoch": 0.03, "learning_rate": 9.739409327809701e-08, "loss": 0.0469, "step": 195 }, { "epoch": 0.03, "learning_rate": 9.738072965388213e-08, "loss": 0.0483, "step": 196 }, { "epoch": 0.03, "learning_rate": 9.736736602966724e-08, "loss": 0.0451, "step": 197 }, { "epoch": 0.03, "learning_rate": 9.735400240545235e-08, "loss": 0.0739, "step": 198 }, { "epoch": 0.03, "learning_rate": 9.734063878123747e-08, "loss": 0.0501, "step": 199 }, { "epoch": 0.03, "learning_rate": 9.732727515702258e-08, "loss": 0.0561, "step": 200 }, { "epoch": 0.03, "learning_rate": 9.73139115328077e-08, "loss": 0.0399, "step": 201 }, { "epoch": 0.03, "learning_rate": 9.73005479085928e-08, "loss": 0.0602, "step": 202 }, { "epoch": 0.03, "learning_rate": 9.728718428437793e-08, "loss": 0.0505, "step": 203 }, { "epoch": 0.03, "learning_rate": 9.727382066016304e-08, "loss": 0.0491, "step": 204 }, { "epoch": 0.03, "learning_rate": 9.726045703594815e-08, "loss": 0.044, "step": 205 }, { "epoch": 0.03, "learning_rate": 9.724709341173326e-08, "loss": 0.0459, "step": 206 }, { "epoch": 0.03, "learning_rate": 9.723372978751837e-08, "loss": 0.0394, "step": 207 }, { "epoch": 0.03, "learning_rate": 9.722036616330348e-08, "loss": 0.0529, "step": 208 }, { "epoch": 0.03, "learning_rate": 9.72070025390886e-08, "loss": 0.0387, "step": 209 }, { "epoch": 0.03, "learning_rate": 9.719363891487371e-08, "loss": 0.0502, "step": 210 }, { "epoch": 0.03, "learning_rate": 9.718027529065882e-08, "loss": 0.0695, "step": 211 }, { "epoch": 0.03, "learning_rate": 9.716691166644394e-08, "loss": 0.0583, "step": 212 }, { "epoch": 0.03, "learning_rate": 9.715354804222905e-08, "loss": 0.0362, "step": 213 }, { "epoch": 0.03, "learning_rate": 9.714018441801416e-08, "loss": 0.0558, "step": 214 }, { "epoch": 0.03, "learning_rate": 9.712682079379927e-08, "loss": 0.0418, "step": 215 }, { "epoch": 0.03, "learning_rate": 9.71134571695844e-08, "loss": 0.0397, "step": 216 }, { "epoch": 0.03, "learning_rate": 9.71000935453695e-08, "loss": 0.0618, "step": 217 }, { "epoch": 0.03, "learning_rate": 9.708672992115461e-08, "loss": 0.0511, "step": 218 }, { "epoch": 0.03, "learning_rate": 9.707336629693972e-08, "loss": 0.0525, "step": 219 }, { "epoch": 0.03, "learning_rate": 9.706000267272483e-08, "loss": 0.0806, "step": 220 }, { "epoch": 0.03, "learning_rate": 9.704663904850994e-08, "loss": 0.0642, "step": 221 }, { "epoch": 0.03, "learning_rate": 9.703327542429507e-08, "loss": 0.0505, "step": 222 }, { "epoch": 0.03, "learning_rate": 9.701991180008018e-08, "loss": 0.0708, "step": 223 }, { "epoch": 0.03, "learning_rate": 9.700654817586529e-08, "loss": 0.0541, "step": 224 }, { "epoch": 0.03, "learning_rate": 9.69931845516504e-08, "loss": 0.0955, "step": 225 }, { "epoch": 0.03, "learning_rate": 9.697982092743552e-08, "loss": 0.0577, "step": 226 }, { "epoch": 0.03, "learning_rate": 9.696645730322063e-08, "loss": 0.0614, "step": 227 }, { "epoch": 0.03, "learning_rate": 9.695309367900574e-08, "loss": 0.0489, "step": 228 }, { "epoch": 0.03, "learning_rate": 9.693973005479086e-08, "loss": 0.0601, "step": 229 }, { "epoch": 0.03, "learning_rate": 9.692636643057597e-08, "loss": 0.0456, "step": 230 }, { "epoch": 0.03, "learning_rate": 9.691300280636108e-08, "loss": 0.0614, "step": 231 }, { "epoch": 0.03, "learning_rate": 9.689963918214619e-08, "loss": 0.0596, "step": 232 }, { "epoch": 0.03, "learning_rate": 9.68862755579313e-08, "loss": 0.0369, "step": 233 }, { "epoch": 0.03, "learning_rate": 9.687291193371641e-08, "loss": 0.0398, "step": 234 }, { "epoch": 0.03, "learning_rate": 9.685954830950153e-08, "loss": 0.0487, "step": 235 }, { "epoch": 0.03, "learning_rate": 9.684618468528664e-08, "loss": 0.0522, "step": 236 }, { "epoch": 0.03, "learning_rate": 9.683282106107175e-08, "loss": 0.0782, "step": 237 }, { "epoch": 0.03, "learning_rate": 9.681945743685686e-08, "loss": 0.0416, "step": 238 }, { "epoch": 0.03, "learning_rate": 9.680609381264199e-08, "loss": 0.0314, "step": 239 }, { "epoch": 0.03, "learning_rate": 9.67927301884271e-08, "loss": 0.0348, "step": 240 }, { "epoch": 0.03, "learning_rate": 9.67793665642122e-08, "loss": 0.0603, "step": 241 }, { "epoch": 0.03, "learning_rate": 9.676600293999733e-08, "loss": 0.0622, "step": 242 }, { "epoch": 0.03, "learning_rate": 9.675263931578244e-08, "loss": 0.0453, "step": 243 }, { "epoch": 0.03, "learning_rate": 9.673927569156755e-08, "loss": 0.0459, "step": 244 }, { "epoch": 0.03, "learning_rate": 9.672591206735267e-08, "loss": 0.0389, "step": 245 }, { "epoch": 0.03, "learning_rate": 9.671254844313778e-08, "loss": 0.0484, "step": 246 }, { "epoch": 0.03, "learning_rate": 9.669918481892289e-08, "loss": 0.0662, "step": 247 }, { "epoch": 0.03, "learning_rate": 9.668582119470799e-08, "loss": 0.0456, "step": 248 }, { "epoch": 0.03, "learning_rate": 9.667245757049311e-08, "loss": 0.0405, "step": 249 }, { "epoch": 0.03, "learning_rate": 9.665909394627822e-08, "loss": 0.048, "step": 250 }, { "epoch": 0.03, "learning_rate": 9.664573032206333e-08, "loss": 0.0455, "step": 251 }, { "epoch": 0.03, "learning_rate": 9.663236669784845e-08, "loss": 0.0574, "step": 252 }, { "epoch": 0.03, "learning_rate": 9.661900307363356e-08, "loss": 0.0622, "step": 253 }, { "epoch": 0.03, "learning_rate": 9.660563944941867e-08, "loss": 0.0314, "step": 254 }, { "epoch": 0.03, "learning_rate": 9.65922758252038e-08, "loss": 0.0434, "step": 255 }, { "epoch": 0.03, "learning_rate": 9.657891220098891e-08, "loss": 0.0522, "step": 256 }, { "epoch": 0.03, "learning_rate": 9.656554857677402e-08, "loss": 0.0382, "step": 257 }, { "epoch": 0.03, "learning_rate": 9.655218495255914e-08, "loss": 0.0376, "step": 258 }, { "epoch": 0.03, "learning_rate": 9.653882132834425e-08, "loss": 0.0481, "step": 259 }, { "epoch": 0.03, "learning_rate": 9.652545770412936e-08, "loss": 0.0411, "step": 260 }, { "epoch": 0.03, "learning_rate": 9.651209407991447e-08, "loss": 0.0552, "step": 261 }, { "epoch": 0.04, "learning_rate": 9.649873045569958e-08, "loss": 0.0351, "step": 262 }, { "epoch": 0.04, "learning_rate": 9.648536683148469e-08, "loss": 0.065, "step": 263 }, { "epoch": 0.04, "learning_rate": 9.64720032072698e-08, "loss": 0.0729, "step": 264 }, { "epoch": 0.04, "learning_rate": 9.645863958305492e-08, "loss": 0.0544, "step": 265 }, { "epoch": 0.04, "learning_rate": 9.644527595884003e-08, "loss": 0.0685, "step": 266 }, { "epoch": 0.04, "learning_rate": 9.643191233462514e-08, "loss": 0.0844, "step": 267 }, { "epoch": 0.04, "learning_rate": 9.641854871041026e-08, "loss": 0.0605, "step": 268 }, { "epoch": 0.04, "learning_rate": 9.640518508619537e-08, "loss": 0.0471, "step": 269 }, { "epoch": 0.04, "learning_rate": 9.639182146198048e-08, "loss": 0.0311, "step": 270 }, { "epoch": 0.04, "learning_rate": 9.63784578377656e-08, "loss": 0.0242, "step": 271 }, { "epoch": 0.04, "learning_rate": 9.636509421355072e-08, "loss": 0.0583, "step": 272 }, { "epoch": 0.04, "learning_rate": 9.635173058933583e-08, "loss": 0.0492, "step": 273 }, { "epoch": 0.04, "learning_rate": 9.633836696512094e-08, "loss": 0.0678, "step": 274 }, { "epoch": 0.04, "learning_rate": 9.632500334090605e-08, "loss": 0.0491, "step": 275 }, { "epoch": 0.04, "learning_rate": 9.631163971669116e-08, "loss": 0.0562, "step": 276 }, { "epoch": 0.04, "learning_rate": 9.629827609247627e-08, "loss": 0.0693, "step": 277 }, { "epoch": 0.04, "learning_rate": 9.628491246826139e-08, "loss": 0.0376, "step": 278 }, { "epoch": 0.04, "learning_rate": 9.62715488440465e-08, "loss": 0.0533, "step": 279 }, { "epoch": 0.04, "learning_rate": 9.625818521983161e-08, "loss": 0.0603, "step": 280 }, { "epoch": 0.04, "learning_rate": 9.624482159561673e-08, "loss": 0.0542, "step": 281 }, { "epoch": 0.04, "learning_rate": 9.623145797140184e-08, "loss": 0.0606, "step": 282 }, { "epoch": 0.04, "learning_rate": 9.621809434718695e-08, "loss": 0.0462, "step": 283 }, { "epoch": 0.04, "learning_rate": 9.620473072297206e-08, "loss": 0.0504, "step": 284 }, { "epoch": 0.04, "learning_rate": 9.619136709875718e-08, "loss": 0.0323, "step": 285 }, { "epoch": 0.04, "learning_rate": 9.61780034745423e-08, "loss": 0.0321, "step": 286 }, { "epoch": 0.04, "learning_rate": 9.61646398503274e-08, "loss": 0.0362, "step": 287 }, { "epoch": 0.04, "learning_rate": 9.615127622611253e-08, "loss": 0.0584, "step": 288 }, { "epoch": 0.04, "learning_rate": 9.613791260189762e-08, "loss": 0.0382, "step": 289 }, { "epoch": 0.04, "learning_rate": 9.612454897768273e-08, "loss": 0.0636, "step": 290 }, { "epoch": 0.04, "learning_rate": 9.611118535346786e-08, "loss": 0.0491, "step": 291 }, { "epoch": 0.04, "learning_rate": 9.609782172925297e-08, "loss": 0.0631, "step": 292 }, { "epoch": 0.04, "learning_rate": 9.608445810503808e-08, "loss": 0.0353, "step": 293 }, { "epoch": 0.04, "learning_rate": 9.60710944808232e-08, "loss": 0.0651, "step": 294 }, { "epoch": 0.04, "learning_rate": 9.605773085660831e-08, "loss": 0.0712, "step": 295 }, { "epoch": 0.04, "learning_rate": 9.604436723239342e-08, "loss": 0.0449, "step": 296 }, { "epoch": 0.04, "learning_rate": 9.603100360817853e-08, "loss": 0.0333, "step": 297 }, { "epoch": 0.04, "learning_rate": 9.601763998396365e-08, "loss": 0.0505, "step": 298 }, { "epoch": 0.04, "learning_rate": 9.600427635974876e-08, "loss": 0.0405, "step": 299 }, { "epoch": 0.04, "learning_rate": 9.599091273553387e-08, "loss": 0.033, "step": 300 }, { "epoch": 0.04, "learning_rate": 9.5977549111319e-08, "loss": 0.0869, "step": 301 }, { "epoch": 0.04, "learning_rate": 9.59641854871041e-08, "loss": 0.0463, "step": 302 }, { "epoch": 0.04, "learning_rate": 9.595082186288921e-08, "loss": 0.0493, "step": 303 }, { "epoch": 0.04, "learning_rate": 9.593745823867432e-08, "loss": 0.0526, "step": 304 }, { "epoch": 0.04, "learning_rate": 9.592409461445943e-08, "loss": 0.0573, "step": 305 }, { "epoch": 0.04, "learning_rate": 9.591073099024454e-08, "loss": 0.0494, "step": 306 }, { "epoch": 0.04, "learning_rate": 9.589736736602965e-08, "loss": 0.0608, "step": 307 }, { "epoch": 0.04, "learning_rate": 9.588400374181478e-08, "loss": 0.0478, "step": 308 }, { "epoch": 0.04, "learning_rate": 9.587064011759989e-08, "loss": 0.031, "step": 309 }, { "epoch": 0.04, "learning_rate": 9.5857276493385e-08, "loss": 0.0531, "step": 310 }, { "epoch": 0.04, "learning_rate": 9.584391286917012e-08, "loss": 0.0309, "step": 311 }, { "epoch": 0.04, "learning_rate": 9.583054924495523e-08, "loss": 0.0533, "step": 312 }, { "epoch": 0.04, "learning_rate": 9.581718562074034e-08, "loss": 0.0752, "step": 313 }, { "epoch": 0.04, "learning_rate": 9.580382199652546e-08, "loss": 0.0464, "step": 314 }, { "epoch": 0.04, "learning_rate": 9.579045837231057e-08, "loss": 0.0647, "step": 315 }, { "epoch": 0.04, "learning_rate": 9.577709474809568e-08, "loss": 0.0516, "step": 316 }, { "epoch": 0.04, "learning_rate": 9.576373112388079e-08, "loss": 0.0438, "step": 317 }, { "epoch": 0.04, "learning_rate": 9.57503674996659e-08, "loss": 0.0456, "step": 318 }, { "epoch": 0.04, "learning_rate": 9.573700387545101e-08, "loss": 0.0457, "step": 319 }, { "epoch": 0.04, "learning_rate": 9.572364025123612e-08, "loss": 0.0383, "step": 320 }, { "epoch": 0.04, "learning_rate": 9.571027662702124e-08, "loss": 0.0412, "step": 321 }, { "epoch": 0.04, "learning_rate": 9.569691300280635e-08, "loss": 0.0427, "step": 322 }, { "epoch": 0.04, "learning_rate": 9.568354937859146e-08, "loss": 0.0401, "step": 323 }, { "epoch": 0.04, "learning_rate": 9.567018575437659e-08, "loss": 0.0452, "step": 324 }, { "epoch": 0.04, "learning_rate": 9.56568221301617e-08, "loss": 0.0806, "step": 325 }, { "epoch": 0.04, "learning_rate": 9.564345850594681e-08, "loss": 0.0457, "step": 326 }, { "epoch": 0.04, "learning_rate": 9.563009488173193e-08, "loss": 0.0645, "step": 327 }, { "epoch": 0.04, "learning_rate": 9.561673125751704e-08, "loss": 0.0582, "step": 328 }, { "epoch": 0.04, "learning_rate": 9.560336763330215e-08, "loss": 0.0534, "step": 329 }, { "epoch": 0.04, "learning_rate": 9.559000400908726e-08, "loss": 0.0516, "step": 330 }, { "epoch": 0.04, "learning_rate": 9.557664038487237e-08, "loss": 0.0403, "step": 331 }, { "epoch": 0.04, "learning_rate": 9.556327676065748e-08, "loss": 0.0528, "step": 332 }, { "epoch": 0.04, "learning_rate": 9.554991313644259e-08, "loss": 0.0583, "step": 333 }, { "epoch": 0.04, "learning_rate": 9.553654951222771e-08, "loss": 0.0726, "step": 334 }, { "epoch": 0.04, "learning_rate": 9.552318588801282e-08, "loss": 0.0434, "step": 335 }, { "epoch": 0.04, "learning_rate": 9.550982226379793e-08, "loss": 0.039, "step": 336 }, { "epoch": 0.05, "learning_rate": 9.549645863958305e-08, "loss": 0.0559, "step": 337 }, { "epoch": 0.05, "learning_rate": 9.548309501536816e-08, "loss": 0.0556, "step": 338 }, { "epoch": 0.05, "learning_rate": 9.546973139115327e-08, "loss": 0.0464, "step": 339 }, { "epoch": 0.05, "learning_rate": 9.54563677669384e-08, "loss": 0.038, "step": 340 }, { "epoch": 0.05, "learning_rate": 9.544300414272351e-08, "loss": 0.0485, "step": 341 }, { "epoch": 0.05, "learning_rate": 9.542964051850862e-08, "loss": 0.0334, "step": 342 }, { "epoch": 0.05, "learning_rate": 9.541627689429373e-08, "loss": 0.0562, "step": 343 }, { "epoch": 0.05, "learning_rate": 9.540291327007885e-08, "loss": 0.0804, "step": 344 }, { "epoch": 0.05, "learning_rate": 9.538954964586396e-08, "loss": 0.0574, "step": 345 }, { "epoch": 0.05, "learning_rate": 9.537618602164907e-08, "loss": 0.064, "step": 346 }, { "epoch": 0.05, "learning_rate": 9.536282239743418e-08, "loss": 0.0447, "step": 347 }, { "epoch": 0.05, "learning_rate": 9.534945877321929e-08, "loss": 0.0417, "step": 348 }, { "epoch": 0.05, "learning_rate": 9.53360951490044e-08, "loss": 0.0505, "step": 349 }, { "epoch": 0.05, "learning_rate": 9.532273152478952e-08, "loss": 0.0699, "step": 350 }, { "epoch": 0.05, "learning_rate": 9.530936790057463e-08, "loss": 0.0389, "step": 351 }, { "epoch": 0.05, "learning_rate": 9.529600427635974e-08, "loss": 0.0459, "step": 352 }, { "epoch": 0.05, "learning_rate": 9.528264065214485e-08, "loss": 0.0419, "step": 353 }, { "epoch": 0.05, "learning_rate": 9.526927702792997e-08, "loss": 0.0664, "step": 354 }, { "epoch": 0.05, "learning_rate": 9.525591340371508e-08, "loss": 0.0835, "step": 355 }, { "epoch": 0.05, "learning_rate": 9.52425497795002e-08, "loss": 0.0499, "step": 356 }, { "epoch": 0.05, "learning_rate": 9.522918615528532e-08, "loss": 0.0467, "step": 357 }, { "epoch": 0.05, "learning_rate": 9.521582253107043e-08, "loss": 0.0338, "step": 358 }, { "epoch": 0.05, "learning_rate": 9.520245890685554e-08, "loss": 0.0433, "step": 359 }, { "epoch": 0.05, "learning_rate": 9.518909528264065e-08, "loss": 0.0539, "step": 360 }, { "epoch": 0.05, "learning_rate": 9.517573165842576e-08, "loss": 0.0539, "step": 361 }, { "epoch": 0.05, "learning_rate": 9.516236803421087e-08, "loss": 0.0456, "step": 362 }, { "epoch": 0.05, "learning_rate": 9.514900440999599e-08, "loss": 0.0614, "step": 363 }, { "epoch": 0.05, "learning_rate": 9.51356407857811e-08, "loss": 0.0625, "step": 364 }, { "epoch": 0.05, "learning_rate": 9.512227716156621e-08, "loss": 0.0337, "step": 365 }, { "epoch": 0.05, "learning_rate": 9.510891353735132e-08, "loss": 0.0436, "step": 366 }, { "epoch": 0.05, "learning_rate": 9.509554991313644e-08, "loss": 0.0412, "step": 367 }, { "epoch": 0.05, "learning_rate": 9.508218628892155e-08, "loss": 0.0456, "step": 368 }, { "epoch": 0.05, "learning_rate": 9.506882266470666e-08, "loss": 0.0597, "step": 369 }, { "epoch": 0.05, "learning_rate": 9.505545904049179e-08, "loss": 0.059, "step": 370 }, { "epoch": 0.05, "learning_rate": 9.50420954162769e-08, "loss": 0.0596, "step": 371 }, { "epoch": 0.05, "learning_rate": 9.5028731792062e-08, "loss": 0.0503, "step": 372 }, { "epoch": 0.05, "learning_rate": 9.501536816784711e-08, "loss": 0.0508, "step": 373 }, { "epoch": 0.05, "learning_rate": 9.500200454363222e-08, "loss": 0.0434, "step": 374 }, { "epoch": 0.05, "learning_rate": 9.498864091941733e-08, "loss": 0.0351, "step": 375 }, { "epoch": 0.05, "learning_rate": 9.497527729520244e-08, "loss": 0.0776, "step": 376 }, { "epoch": 0.05, "learning_rate": 9.496191367098757e-08, "loss": 0.0546, "step": 377 }, { "epoch": 0.05, "learning_rate": 9.494855004677268e-08, "loss": 0.0542, "step": 378 }, { "epoch": 0.05, "learning_rate": 9.493518642255779e-08, "loss": 0.0469, "step": 379 }, { "epoch": 0.05, "learning_rate": 9.492182279834291e-08, "loss": 0.0607, "step": 380 }, { "epoch": 0.05, "learning_rate": 9.490845917412802e-08, "loss": 0.0508, "step": 381 }, { "epoch": 0.05, "learning_rate": 9.489509554991313e-08, "loss": 0.0524, "step": 382 }, { "epoch": 0.05, "learning_rate": 9.488173192569825e-08, "loss": 0.0388, "step": 383 }, { "epoch": 0.05, "learning_rate": 9.486836830148336e-08, "loss": 0.0585, "step": 384 }, { "epoch": 0.05, "learning_rate": 9.485500467726847e-08, "loss": 0.0676, "step": 385 }, { "epoch": 0.05, "learning_rate": 9.48416410530536e-08, "loss": 0.0539, "step": 386 }, { "epoch": 0.05, "learning_rate": 9.48282774288387e-08, "loss": 0.0563, "step": 387 }, { "epoch": 0.05, "learning_rate": 9.48149138046238e-08, "loss": 0.0699, "step": 388 }, { "epoch": 0.05, "learning_rate": 9.480155018040891e-08, "loss": 0.0352, "step": 389 }, { "epoch": 0.05, "learning_rate": 9.478818655619403e-08, "loss": 0.056, "step": 390 }, { "epoch": 0.05, "learning_rate": 9.477482293197914e-08, "loss": 0.0848, "step": 391 }, { "epoch": 0.05, "learning_rate": 9.476145930776425e-08, "loss": 0.0403, "step": 392 }, { "epoch": 0.05, "learning_rate": 9.474809568354938e-08, "loss": 0.0425, "step": 393 }, { "epoch": 0.05, "learning_rate": 9.473473205933449e-08, "loss": 0.093, "step": 394 }, { "epoch": 0.05, "learning_rate": 9.47213684351196e-08, "loss": 0.0483, "step": 395 }, { "epoch": 0.05, "learning_rate": 9.470800481090472e-08, "loss": 0.0755, "step": 396 }, { "epoch": 0.05, "learning_rate": 9.469464118668983e-08, "loss": 0.0579, "step": 397 }, { "epoch": 0.05, "learning_rate": 9.468127756247494e-08, "loss": 0.0589, "step": 398 }, { "epoch": 0.05, "learning_rate": 9.466791393826005e-08, "loss": 0.0815, "step": 399 }, { "epoch": 0.05, "learning_rate": 9.465455031404517e-08, "loss": 0.034, "step": 400 }, { "epoch": 0.05, "learning_rate": 9.464118668983028e-08, "loss": 0.0472, "step": 401 }, { "epoch": 0.05, "learning_rate": 9.462782306561539e-08, "loss": 0.0664, "step": 402 }, { "epoch": 0.05, "learning_rate": 9.46144594414005e-08, "loss": 0.0399, "step": 403 }, { "epoch": 0.05, "learning_rate": 9.460109581718561e-08, "loss": 0.0569, "step": 404 }, { "epoch": 0.05, "learning_rate": 9.458773219297072e-08, "loss": 0.0498, "step": 405 }, { "epoch": 0.05, "learning_rate": 9.457436856875585e-08, "loss": 0.0509, "step": 406 }, { "epoch": 0.05, "learning_rate": 9.456100494454095e-08, "loss": 0.0614, "step": 407 }, { "epoch": 0.05, "learning_rate": 9.454764132032606e-08, "loss": 0.0619, "step": 408 }, { "epoch": 0.05, "learning_rate": 9.453427769611119e-08, "loss": 0.0606, "step": 409 }, { "epoch": 0.05, "learning_rate": 9.45209140718963e-08, "loss": 0.0406, "step": 410 }, { "epoch": 0.05, "learning_rate": 9.450755044768141e-08, "loss": 0.0339, "step": 411 }, { "epoch": 0.06, "learning_rate": 9.449418682346652e-08, "loss": 0.0352, "step": 412 }, { "epoch": 0.06, "learning_rate": 9.448082319925164e-08, "loss": 0.0621, "step": 413 }, { "epoch": 0.06, "learning_rate": 9.446745957503675e-08, "loss": 0.0368, "step": 414 }, { "epoch": 0.06, "learning_rate": 9.445409595082186e-08, "loss": 0.045, "step": 415 }, { "epoch": 0.06, "learning_rate": 9.444073232660697e-08, "loss": 0.0364, "step": 416 }, { "epoch": 0.06, "learning_rate": 9.442736870239208e-08, "loss": 0.0427, "step": 417 }, { "epoch": 0.06, "learning_rate": 9.441400507817719e-08, "loss": 0.0693, "step": 418 }, { "epoch": 0.06, "learning_rate": 9.440064145396231e-08, "loss": 0.0372, "step": 419 }, { "epoch": 0.06, "learning_rate": 9.438727782974742e-08, "loss": 0.0369, "step": 420 }, { "epoch": 0.06, "learning_rate": 9.437391420553253e-08, "loss": 0.029, "step": 421 }, { "epoch": 0.06, "learning_rate": 9.436055058131764e-08, "loss": 0.0678, "step": 422 }, { "epoch": 0.06, "learning_rate": 9.434718695710277e-08, "loss": 0.0467, "step": 423 }, { "epoch": 0.06, "learning_rate": 9.433382333288788e-08, "loss": 0.0799, "step": 424 }, { "epoch": 0.06, "learning_rate": 9.432045970867298e-08, "loss": 0.0401, "step": 425 }, { "epoch": 0.06, "learning_rate": 9.430709608445811e-08, "loss": 0.0384, "step": 426 }, { "epoch": 0.06, "learning_rate": 9.429373246024322e-08, "loss": 0.0382, "step": 427 }, { "epoch": 0.06, "learning_rate": 9.428036883602833e-08, "loss": 0.046, "step": 428 }, { "epoch": 0.06, "learning_rate": 9.426700521181344e-08, "loss": 0.0628, "step": 429 }, { "epoch": 0.06, "learning_rate": 9.425364158759855e-08, "loss": 0.0742, "step": 430 }, { "epoch": 0.06, "learning_rate": 9.424027796338366e-08, "loss": 0.0497, "step": 431 }, { "epoch": 0.06, "learning_rate": 9.422691433916878e-08, "loss": 0.0838, "step": 432 }, { "epoch": 0.06, "learning_rate": 9.421355071495389e-08, "loss": 0.0436, "step": 433 }, { "epoch": 0.06, "learning_rate": 9.4200187090739e-08, "loss": 0.054, "step": 434 }, { "epoch": 0.06, "learning_rate": 9.418682346652411e-08, "loss": 0.0453, "step": 435 }, { "epoch": 0.06, "learning_rate": 9.417345984230923e-08, "loss": 0.043, "step": 436 }, { "epoch": 0.06, "learning_rate": 9.416009621809434e-08, "loss": 0.0372, "step": 437 }, { "epoch": 0.06, "learning_rate": 9.414673259387945e-08, "loss": 0.0344, "step": 438 }, { "epoch": 0.06, "learning_rate": 9.413336896966458e-08, "loss": 0.0408, "step": 439 }, { "epoch": 0.06, "learning_rate": 9.412000534544969e-08, "loss": 0.0458, "step": 440 }, { "epoch": 0.06, "learning_rate": 9.41066417212348e-08, "loss": 0.0447, "step": 441 }, { "epoch": 0.06, "learning_rate": 9.409327809701992e-08, "loss": 0.0543, "step": 442 }, { "epoch": 0.06, "learning_rate": 9.407991447280503e-08, "loss": 0.0591, "step": 443 }, { "epoch": 0.06, "learning_rate": 9.406655084859014e-08, "loss": 0.068, "step": 444 }, { "epoch": 0.06, "learning_rate": 9.405318722437523e-08, "loss": 0.0591, "step": 445 }, { "epoch": 0.06, "learning_rate": 9.403982360016036e-08, "loss": 0.0825, "step": 446 }, { "epoch": 0.06, "learning_rate": 9.402645997594547e-08, "loss": 0.0719, "step": 447 }, { "epoch": 0.06, "learning_rate": 9.401309635173058e-08, "loss": 0.0588, "step": 448 }, { "epoch": 0.06, "learning_rate": 9.39997327275157e-08, "loss": 0.0528, "step": 449 }, { "epoch": 0.06, "learning_rate": 9.398636910330081e-08, "loss": 0.0534, "step": 450 }, { "epoch": 0.06, "learning_rate": 9.397300547908592e-08, "loss": 0.0394, "step": 451 }, { "epoch": 0.06, "learning_rate": 9.395964185487104e-08, "loss": 0.0355, "step": 452 }, { "epoch": 0.06, "learning_rate": 9.394627823065615e-08, "loss": 0.0452, "step": 453 }, { "epoch": 0.06, "learning_rate": 9.393291460644126e-08, "loss": 0.0406, "step": 454 }, { "epoch": 0.06, "learning_rate": 9.391955098222639e-08, "loss": 0.0351, "step": 455 }, { "epoch": 0.06, "learning_rate": 9.39061873580115e-08, "loss": 0.0547, "step": 456 }, { "epoch": 0.06, "learning_rate": 9.38928237337966e-08, "loss": 0.049, "step": 457 }, { "epoch": 0.06, "learning_rate": 9.387946010958172e-08, "loss": 0.0824, "step": 458 }, { "epoch": 0.06, "learning_rate": 9.386609648536683e-08, "loss": 0.0405, "step": 459 }, { "epoch": 0.06, "learning_rate": 9.385273286115193e-08, "loss": 0.0657, "step": 460 }, { "epoch": 0.06, "learning_rate": 9.383936923693704e-08, "loss": 0.0446, "step": 461 }, { "epoch": 0.06, "learning_rate": 9.382600561272217e-08, "loss": 0.0439, "step": 462 }, { "epoch": 0.06, "learning_rate": 9.381264198850728e-08, "loss": 0.0372, "step": 463 }, { "epoch": 0.06, "learning_rate": 9.379927836429239e-08, "loss": 0.0556, "step": 464 }, { "epoch": 0.06, "learning_rate": 9.378591474007751e-08, "loss": 0.0593, "step": 465 }, { "epoch": 0.06, "learning_rate": 9.377255111586262e-08, "loss": 0.0457, "step": 466 }, { "epoch": 0.06, "learning_rate": 9.375918749164773e-08, "loss": 0.0478, "step": 467 }, { "epoch": 0.06, "learning_rate": 9.374582386743284e-08, "loss": 0.0485, "step": 468 }, { "epoch": 0.06, "learning_rate": 9.373246024321796e-08, "loss": 0.0649, "step": 469 }, { "epoch": 0.06, "learning_rate": 9.371909661900307e-08, "loss": 0.0502, "step": 470 }, { "epoch": 0.06, "learning_rate": 9.370573299478818e-08, "loss": 0.0608, "step": 471 }, { "epoch": 0.06, "learning_rate": 9.369236937057329e-08, "loss": 0.0554, "step": 472 }, { "epoch": 0.06, "learning_rate": 9.36790057463584e-08, "loss": 0.0447, "step": 473 }, { "epoch": 0.06, "learning_rate": 9.366564212214351e-08, "loss": 0.0584, "step": 474 }, { "epoch": 0.06, "learning_rate": 9.365227849792864e-08, "loss": 0.0342, "step": 475 }, { "epoch": 0.06, "learning_rate": 9.363891487371375e-08, "loss": 0.0433, "step": 476 }, { "epoch": 0.06, "learning_rate": 9.362555124949886e-08, "loss": 0.0652, "step": 477 }, { "epoch": 0.06, "learning_rate": 9.361218762528398e-08, "loss": 0.0525, "step": 478 }, { "epoch": 0.06, "learning_rate": 9.359882400106909e-08, "loss": 0.0443, "step": 479 }, { "epoch": 0.06, "learning_rate": 9.35854603768542e-08, "loss": 0.0374, "step": 480 }, { "epoch": 0.06, "learning_rate": 9.357209675263931e-08, "loss": 0.0599, "step": 481 }, { "epoch": 0.06, "learning_rate": 9.355873312842443e-08, "loss": 0.0565, "step": 482 }, { "epoch": 0.06, "learning_rate": 9.354536950420954e-08, "loss": 0.0495, "step": 483 }, { "epoch": 0.06, "learning_rate": 9.353200587999465e-08, "loss": 0.0336, "step": 484 }, { "epoch": 0.06, "learning_rate": 9.351864225577977e-08, "loss": 0.0495, "step": 485 }, { "epoch": 0.06, "learning_rate": 9.350527863156488e-08, "loss": 0.0605, "step": 486 }, { "epoch": 0.07, "learning_rate": 9.349191500734998e-08, "loss": 0.0559, "step": 487 }, { "epoch": 0.07, "learning_rate": 9.34785513831351e-08, "loss": 0.0437, "step": 488 }, { "epoch": 0.07, "learning_rate": 9.346518775892021e-08, "loss": 0.0591, "step": 489 }, { "epoch": 0.07, "learning_rate": 9.345182413470532e-08, "loss": 0.0432, "step": 490 }, { "epoch": 0.07, "learning_rate": 9.343846051049043e-08, "loss": 0.0391, "step": 491 }, { "epoch": 0.07, "learning_rate": 9.342509688627556e-08, "loss": 0.0794, "step": 492 }, { "epoch": 0.07, "learning_rate": 9.341173326206067e-08, "loss": 0.0464, "step": 493 }, { "epoch": 0.07, "learning_rate": 9.339836963784578e-08, "loss": 0.032, "step": 494 }, { "epoch": 0.07, "learning_rate": 9.33850060136309e-08, "loss": 0.0554, "step": 495 }, { "epoch": 0.07, "learning_rate": 9.337164238941601e-08, "loss": 0.04, "step": 496 }, { "epoch": 0.07, "learning_rate": 9.335827876520112e-08, "loss": 0.0601, "step": 497 }, { "epoch": 0.07, "learning_rate": 9.334491514098624e-08, "loss": 0.0489, "step": 498 }, { "epoch": 0.07, "learning_rate": 9.333155151677135e-08, "loss": 0.0373, "step": 499 }, { "epoch": 0.07, "learning_rate": 9.331818789255646e-08, "loss": 0.0451, "step": 500 }, { "epoch": 0.07, "learning_rate": 9.330482426834157e-08, "loss": 0.0577, "step": 501 }, { "epoch": 0.07, "learning_rate": 9.329146064412668e-08, "loss": 0.0634, "step": 502 }, { "epoch": 0.07, "learning_rate": 9.327809701991179e-08, "loss": 0.0364, "step": 503 }, { "epoch": 0.07, "learning_rate": 9.32647333956969e-08, "loss": 0.0566, "step": 504 }, { "epoch": 0.07, "learning_rate": 9.325136977148202e-08, "loss": 0.0406, "step": 505 }, { "epoch": 0.07, "learning_rate": 9.323800614726713e-08, "loss": 0.0424, "step": 506 }, { "epoch": 0.07, "learning_rate": 9.322464252305224e-08, "loss": 0.046, "step": 507 }, { "epoch": 0.07, "learning_rate": 9.321127889883737e-08, "loss": 0.0476, "step": 508 }, { "epoch": 0.07, "learning_rate": 9.319791527462248e-08, "loss": 0.0513, "step": 509 }, { "epoch": 0.07, "learning_rate": 9.318455165040759e-08, "loss": 0.057, "step": 510 }, { "epoch": 0.07, "learning_rate": 9.317118802619271e-08, "loss": 0.0552, "step": 511 }, { "epoch": 0.07, "learning_rate": 9.315782440197782e-08, "loss": 0.0728, "step": 512 }, { "epoch": 0.07, "learning_rate": 9.314446077776293e-08, "loss": 0.0449, "step": 513 }, { "epoch": 0.07, "learning_rate": 9.313109715354804e-08, "loss": 0.055, "step": 514 }, { "epoch": 0.07, "learning_rate": 9.311773352933315e-08, "loss": 0.056, "step": 515 }, { "epoch": 0.07, "learning_rate": 9.310436990511826e-08, "loss": 0.0431, "step": 516 }, { "epoch": 0.07, "learning_rate": 9.309100628090337e-08, "loss": 0.0502, "step": 517 }, { "epoch": 0.07, "learning_rate": 9.307764265668849e-08, "loss": 0.0363, "step": 518 }, { "epoch": 0.07, "learning_rate": 9.30642790324736e-08, "loss": 0.0553, "step": 519 }, { "epoch": 0.07, "learning_rate": 9.305091540825871e-08, "loss": 0.0547, "step": 520 }, { "epoch": 0.07, "learning_rate": 9.303755178404383e-08, "loss": 0.0619, "step": 521 }, { "epoch": 0.07, "learning_rate": 9.302418815982894e-08, "loss": 0.0458, "step": 522 }, { "epoch": 0.07, "learning_rate": 9.301082453561405e-08, "loss": 0.0497, "step": 523 }, { "epoch": 0.07, "learning_rate": 9.299746091139918e-08, "loss": 0.0365, "step": 524 }, { "epoch": 0.07, "learning_rate": 9.298409728718429e-08, "loss": 0.0432, "step": 525 }, { "epoch": 0.07, "learning_rate": 9.29707336629694e-08, "loss": 0.0461, "step": 526 }, { "epoch": 0.07, "learning_rate": 9.29573700387545e-08, "loss": 0.0661, "step": 527 }, { "epoch": 0.07, "learning_rate": 9.294400641453962e-08, "loss": 0.0484, "step": 528 }, { "epoch": 0.07, "learning_rate": 9.293064279032473e-08, "loss": 0.054, "step": 529 }, { "epoch": 0.07, "learning_rate": 9.291727916610984e-08, "loss": 0.0449, "step": 530 }, { "epoch": 0.07, "learning_rate": 9.290391554189496e-08, "loss": 0.0579, "step": 531 }, { "epoch": 0.07, "learning_rate": 9.289055191768007e-08, "loss": 0.0455, "step": 532 }, { "epoch": 0.07, "learning_rate": 9.287718829346518e-08, "loss": 0.0453, "step": 533 }, { "epoch": 0.07, "learning_rate": 9.28638246692503e-08, "loss": 0.0411, "step": 534 }, { "epoch": 0.07, "learning_rate": 9.285046104503541e-08, "loss": 0.0512, "step": 535 }, { "epoch": 0.07, "learning_rate": 9.283709742082052e-08, "loss": 0.0411, "step": 536 }, { "epoch": 0.07, "learning_rate": 9.282373379660563e-08, "loss": 0.0478, "step": 537 }, { "epoch": 0.07, "learning_rate": 9.281037017239075e-08, "loss": 0.0531, "step": 538 }, { "epoch": 0.07, "learning_rate": 9.279700654817586e-08, "loss": 0.0439, "step": 539 }, { "epoch": 0.07, "learning_rate": 9.278364292396097e-08, "loss": 0.038, "step": 540 }, { "epoch": 0.07, "learning_rate": 9.27702792997461e-08, "loss": 0.0661, "step": 541 }, { "epoch": 0.07, "learning_rate": 9.27569156755312e-08, "loss": 0.0641, "step": 542 }, { "epoch": 0.07, "learning_rate": 9.274355205131632e-08, "loss": 0.0427, "step": 543 }, { "epoch": 0.07, "learning_rate": 9.273018842710143e-08, "loss": 0.0474, "step": 544 }, { "epoch": 0.07, "learning_rate": 9.271682480288654e-08, "loss": 0.049, "step": 545 }, { "epoch": 0.07, "learning_rate": 9.270346117867165e-08, "loss": 0.0491, "step": 546 }, { "epoch": 0.07, "learning_rate": 9.269009755445677e-08, "loss": 0.0447, "step": 547 }, { "epoch": 0.07, "learning_rate": 9.267673393024188e-08, "loss": 0.0627, "step": 548 }, { "epoch": 0.07, "learning_rate": 9.266337030602699e-08, "loss": 0.0321, "step": 549 }, { "epoch": 0.07, "learning_rate": 9.26500066818121e-08, "loss": 0.0555, "step": 550 }, { "epoch": 0.07, "learning_rate": 9.263664305759722e-08, "loss": 0.0398, "step": 551 }, { "epoch": 0.07, "learning_rate": 9.262327943338233e-08, "loss": 0.0703, "step": 552 }, { "epoch": 0.07, "learning_rate": 9.260991580916744e-08, "loss": 0.0503, "step": 553 }, { "epoch": 0.07, "learning_rate": 9.259655218495256e-08, "loss": 0.0419, "step": 554 }, { "epoch": 0.07, "learning_rate": 9.258318856073767e-08, "loss": 0.0386, "step": 555 }, { "epoch": 0.07, "learning_rate": 9.256982493652278e-08, "loss": 0.0509, "step": 556 }, { "epoch": 0.07, "learning_rate": 9.255646131230789e-08, "loss": 0.0643, "step": 557 }, { "epoch": 0.07, "learning_rate": 9.2543097688093e-08, "loss": 0.0608, "step": 558 }, { "epoch": 0.07, "learning_rate": 9.252973406387811e-08, "loss": 0.0426, "step": 559 }, { "epoch": 0.07, "learning_rate": 9.251637043966322e-08, "loss": 0.0601, "step": 560 }, { "epoch": 0.07, "learning_rate": 9.250300681544835e-08, "loss": 0.0328, "step": 561 }, { "epoch": 0.08, "learning_rate": 9.248964319123346e-08, "loss": 0.0726, "step": 562 }, { "epoch": 0.08, "learning_rate": 9.247627956701857e-08, "loss": 0.0877, "step": 563 }, { "epoch": 0.08, "learning_rate": 9.246291594280369e-08, "loss": 0.0558, "step": 564 }, { "epoch": 0.08, "learning_rate": 9.24495523185888e-08, "loss": 0.0375, "step": 565 }, { "epoch": 0.08, "learning_rate": 9.243618869437391e-08, "loss": 0.0343, "step": 566 }, { "epoch": 0.08, "learning_rate": 9.242282507015903e-08, "loss": 0.0433, "step": 567 }, { "epoch": 0.08, "learning_rate": 9.240946144594414e-08, "loss": 0.0645, "step": 568 }, { "epoch": 0.08, "learning_rate": 9.239609782172925e-08, "loss": 0.0492, "step": 569 }, { "epoch": 0.08, "learning_rate": 9.238273419751436e-08, "loss": 0.0674, "step": 570 }, { "epoch": 0.08, "learning_rate": 9.236937057329947e-08, "loss": 0.0778, "step": 571 }, { "epoch": 0.08, "learning_rate": 9.235600694908458e-08, "loss": 0.0525, "step": 572 }, { "epoch": 0.08, "learning_rate": 9.234264332486969e-08, "loss": 0.0319, "step": 573 }, { "epoch": 0.08, "learning_rate": 9.232927970065481e-08, "loss": 0.042, "step": 574 }, { "epoch": 0.08, "learning_rate": 9.231591607643992e-08, "loss": 0.0285, "step": 575 }, { "epoch": 0.08, "learning_rate": 9.230255245222503e-08, "loss": 0.032, "step": 576 }, { "epoch": 0.08, "learning_rate": 9.228918882801016e-08, "loss": 0.053, "step": 577 }, { "epoch": 0.08, "learning_rate": 9.227582520379527e-08, "loss": 0.0408, "step": 578 }, { "epoch": 0.08, "learning_rate": 9.226246157958038e-08, "loss": 0.0448, "step": 579 }, { "epoch": 0.08, "learning_rate": 9.22490979553655e-08, "loss": 0.0596, "step": 580 }, { "epoch": 0.08, "learning_rate": 9.223573433115061e-08, "loss": 0.0663, "step": 581 }, { "epoch": 0.08, "learning_rate": 9.222237070693572e-08, "loss": 0.0559, "step": 582 }, { "epoch": 0.08, "learning_rate": 9.220900708272083e-08, "loss": 0.0432, "step": 583 }, { "epoch": 0.08, "learning_rate": 9.219564345850595e-08, "loss": 0.0484, "step": 584 }, { "epoch": 0.08, "learning_rate": 9.218227983429106e-08, "loss": 0.0353, "step": 585 }, { "epoch": 0.08, "learning_rate": 9.216891621007616e-08, "loss": 0.0447, "step": 586 }, { "epoch": 0.08, "learning_rate": 9.215555258586128e-08, "loss": 0.0636, "step": 587 }, { "epoch": 0.08, "learning_rate": 9.214218896164639e-08, "loss": 0.0451, "step": 588 }, { "epoch": 0.08, "learning_rate": 9.21288253374315e-08, "loss": 0.0626, "step": 589 }, { "epoch": 0.08, "learning_rate": 9.211546171321662e-08, "loss": 0.0553, "step": 590 }, { "epoch": 0.08, "learning_rate": 9.210209808900173e-08, "loss": 0.0383, "step": 591 }, { "epoch": 0.08, "learning_rate": 9.208873446478684e-08, "loss": 0.03, "step": 592 }, { "epoch": 0.08, "learning_rate": 9.207537084057197e-08, "loss": 0.0745, "step": 593 }, { "epoch": 0.08, "learning_rate": 9.206200721635708e-08, "loss": 0.0507, "step": 594 }, { "epoch": 0.08, "learning_rate": 9.204864359214219e-08, "loss": 0.0252, "step": 595 }, { "epoch": 0.08, "learning_rate": 9.20352799679273e-08, "loss": 0.039, "step": 596 }, { "epoch": 0.08, "learning_rate": 9.202191634371242e-08, "loss": 0.0583, "step": 597 }, { "epoch": 0.08, "learning_rate": 9.200855271949753e-08, "loss": 0.0493, "step": 598 }, { "epoch": 0.08, "learning_rate": 9.199518909528264e-08, "loss": 0.0379, "step": 599 }, { "epoch": 0.08, "learning_rate": 9.198182547106775e-08, "loss": 0.0657, "step": 600 }, { "epoch": 0.08, "learning_rate": 9.196846184685286e-08, "loss": 0.0462, "step": 601 }, { "epoch": 0.08, "learning_rate": 9.195509822263797e-08, "loss": 0.0722, "step": 602 }, { "epoch": 0.08, "learning_rate": 9.194173459842309e-08, "loss": 0.0603, "step": 603 }, { "epoch": 0.08, "learning_rate": 9.19283709742082e-08, "loss": 0.0445, "step": 604 }, { "epoch": 0.08, "learning_rate": 9.191500734999331e-08, "loss": 0.0719, "step": 605 }, { "epoch": 0.08, "learning_rate": 9.190164372577842e-08, "loss": 0.0447, "step": 606 }, { "epoch": 0.08, "learning_rate": 9.188828010156354e-08, "loss": 0.0742, "step": 607 }, { "epoch": 0.08, "learning_rate": 9.187491647734865e-08, "loss": 0.0764, "step": 608 }, { "epoch": 0.08, "learning_rate": 9.186155285313376e-08, "loss": 0.0708, "step": 609 }, { "epoch": 0.08, "learning_rate": 9.184818922891889e-08, "loss": 0.0489, "step": 610 }, { "epoch": 0.08, "learning_rate": 9.1834825604704e-08, "loss": 0.0444, "step": 611 }, { "epoch": 0.08, "learning_rate": 9.18214619804891e-08, "loss": 0.0531, "step": 612 }, { "epoch": 0.08, "learning_rate": 9.180809835627422e-08, "loss": 0.0565, "step": 613 }, { "epoch": 0.08, "learning_rate": 9.179473473205933e-08, "loss": 0.0825, "step": 614 }, { "epoch": 0.08, "learning_rate": 9.178137110784444e-08, "loss": 0.041, "step": 615 }, { "epoch": 0.08, "learning_rate": 9.176800748362956e-08, "loss": 0.0451, "step": 616 }, { "epoch": 0.08, "learning_rate": 9.175464385941467e-08, "loss": 0.0417, "step": 617 }, { "epoch": 0.08, "learning_rate": 9.174128023519978e-08, "loss": 0.0749, "step": 618 }, { "epoch": 0.08, "learning_rate": 9.172791661098489e-08, "loss": 0.0505, "step": 619 }, { "epoch": 0.08, "learning_rate": 9.171455298677001e-08, "loss": 0.05, "step": 620 }, { "epoch": 0.08, "learning_rate": 9.170118936255512e-08, "loss": 0.046, "step": 621 }, { "epoch": 0.08, "learning_rate": 9.168782573834023e-08, "loss": 0.0595, "step": 622 }, { "epoch": 0.08, "learning_rate": 9.167446211412535e-08, "loss": 0.0386, "step": 623 }, { "epoch": 0.08, "learning_rate": 9.166109848991046e-08, "loss": 0.0553, "step": 624 }, { "epoch": 0.08, "learning_rate": 9.164773486569557e-08, "loss": 0.0539, "step": 625 }, { "epoch": 0.08, "learning_rate": 9.16343712414807e-08, "loss": 0.0466, "step": 626 }, { "epoch": 0.08, "learning_rate": 9.162100761726579e-08, "loss": 0.0425, "step": 627 }, { "epoch": 0.08, "learning_rate": 9.16076439930509e-08, "loss": 0.0329, "step": 628 }, { "epoch": 0.08, "learning_rate": 9.159428036883601e-08, "loss": 0.0472, "step": 629 }, { "epoch": 0.08, "learning_rate": 9.158091674462114e-08, "loss": 0.041, "step": 630 }, { "epoch": 0.08, "learning_rate": 9.156755312040625e-08, "loss": 0.0713, "step": 631 }, { "epoch": 0.08, "learning_rate": 9.155418949619136e-08, "loss": 0.0521, "step": 632 }, { "epoch": 0.08, "learning_rate": 9.154082587197648e-08, "loss": 0.0498, "step": 633 }, { "epoch": 0.08, "learning_rate": 9.152746224776159e-08, "loss": 0.0416, "step": 634 }, { "epoch": 0.08, "learning_rate": 9.15140986235467e-08, "loss": 0.0463, "step": 635 }, { "epoch": 0.08, "learning_rate": 9.150073499933182e-08, "loss": 0.0846, "step": 636 }, { "epoch": 0.09, "learning_rate": 9.148737137511693e-08, "loss": 0.0633, "step": 637 }, { "epoch": 0.09, "learning_rate": 9.147400775090204e-08, "loss": 0.0579, "step": 638 }, { "epoch": 0.09, "learning_rate": 9.146064412668716e-08, "loss": 0.0454, "step": 639 }, { "epoch": 0.09, "learning_rate": 9.144728050247227e-08, "loss": 0.0714, "step": 640 }, { "epoch": 0.09, "learning_rate": 9.143391687825738e-08, "loss": 0.0569, "step": 641 }, { "epoch": 0.09, "learning_rate": 9.14205532540425e-08, "loss": 0.0531, "step": 642 }, { "epoch": 0.09, "learning_rate": 9.14071896298276e-08, "loss": 0.0332, "step": 643 }, { "epoch": 0.09, "learning_rate": 9.139382600561271e-08, "loss": 0.0704, "step": 644 }, { "epoch": 0.09, "learning_rate": 9.138046238139782e-08, "loss": 0.0427, "step": 645 }, { "epoch": 0.09, "learning_rate": 9.136709875718295e-08, "loss": 0.0446, "step": 646 }, { "epoch": 0.09, "learning_rate": 9.135373513296806e-08, "loss": 0.0418, "step": 647 }, { "epoch": 0.09, "learning_rate": 9.134037150875317e-08, "loss": 0.0421, "step": 648 }, { "epoch": 0.09, "learning_rate": 9.132700788453829e-08, "loss": 0.0664, "step": 649 }, { "epoch": 0.09, "learning_rate": 9.13136442603234e-08, "loss": 0.0509, "step": 650 }, { "epoch": 0.09, "learning_rate": 9.130028063610851e-08, "loss": 0.0402, "step": 651 }, { "epoch": 0.09, "learning_rate": 9.128691701189362e-08, "loss": 0.0434, "step": 652 }, { "epoch": 0.09, "learning_rate": 9.127355338767874e-08, "loss": 0.0528, "step": 653 }, { "epoch": 0.09, "learning_rate": 9.126018976346385e-08, "loss": 0.055, "step": 654 }, { "epoch": 0.09, "learning_rate": 9.124682613924896e-08, "loss": 0.063, "step": 655 }, { "epoch": 0.09, "learning_rate": 9.123346251503407e-08, "loss": 0.0594, "step": 656 }, { "epoch": 0.09, "learning_rate": 9.122009889081918e-08, "loss": 0.0543, "step": 657 }, { "epoch": 0.09, "learning_rate": 9.120673526660429e-08, "loss": 0.0459, "step": 658 }, { "epoch": 0.09, "learning_rate": 9.119337164238941e-08, "loss": 0.0409, "step": 659 }, { "epoch": 0.09, "learning_rate": 9.118000801817452e-08, "loss": 0.0554, "step": 660 }, { "epoch": 0.09, "learning_rate": 9.116664439395963e-08, "loss": 0.0518, "step": 661 }, { "epoch": 0.09, "learning_rate": 9.115328076974476e-08, "loss": 0.0452, "step": 662 }, { "epoch": 0.09, "learning_rate": 9.113991714552987e-08, "loss": 0.04, "step": 663 }, { "epoch": 0.09, "learning_rate": 9.112655352131498e-08, "loss": 0.0473, "step": 664 }, { "epoch": 0.09, "learning_rate": 9.111318989710009e-08, "loss": 0.0656, "step": 665 }, { "epoch": 0.09, "learning_rate": 9.109982627288521e-08, "loss": 0.035, "step": 666 }, { "epoch": 0.09, "learning_rate": 9.108646264867032e-08, "loss": 0.0306, "step": 667 }, { "epoch": 0.09, "learning_rate": 9.107309902445543e-08, "loss": 0.0481, "step": 668 }, { "epoch": 0.09, "learning_rate": 9.105973540024054e-08, "loss": 0.0381, "step": 669 }, { "epoch": 0.09, "learning_rate": 9.104637177602565e-08, "loss": 0.063, "step": 670 }, { "epoch": 0.09, "learning_rate": 9.103300815181076e-08, "loss": 0.039, "step": 671 }, { "epoch": 0.09, "learning_rate": 9.101964452759588e-08, "loss": 0.0447, "step": 672 }, { "epoch": 0.09, "learning_rate": 9.100628090338099e-08, "loss": 0.0754, "step": 673 }, { "epoch": 0.09, "learning_rate": 9.09929172791661e-08, "loss": 0.0682, "step": 674 }, { "epoch": 0.09, "learning_rate": 9.097955365495121e-08, "loss": 0.0602, "step": 675 }, { "epoch": 0.09, "learning_rate": 9.096619003073633e-08, "loss": 0.0607, "step": 676 }, { "epoch": 0.09, "learning_rate": 9.095282640652144e-08, "loss": 0.0366, "step": 677 }, { "epoch": 0.09, "learning_rate": 9.093946278230655e-08, "loss": 0.0294, "step": 678 }, { "epoch": 0.09, "learning_rate": 9.092609915809168e-08, "loss": 0.0604, "step": 679 }, { "epoch": 0.09, "learning_rate": 9.091273553387679e-08, "loss": 0.0363, "step": 680 }, { "epoch": 0.09, "learning_rate": 9.08993719096619e-08, "loss": 0.0342, "step": 681 }, { "epoch": 0.09, "learning_rate": 9.088600828544702e-08, "loss": 0.056, "step": 682 }, { "epoch": 0.09, "learning_rate": 9.087264466123213e-08, "loss": 0.0572, "step": 683 }, { "epoch": 0.09, "learning_rate": 9.085928103701723e-08, "loss": 0.0674, "step": 684 }, { "epoch": 0.09, "learning_rate": 9.084591741280235e-08, "loss": 0.0543, "step": 685 }, { "epoch": 0.09, "learning_rate": 9.083255378858746e-08, "loss": 0.1, "step": 686 }, { "epoch": 0.09, "learning_rate": 9.081919016437257e-08, "loss": 0.049, "step": 687 }, { "epoch": 0.09, "learning_rate": 9.080582654015768e-08, "loss": 0.0424, "step": 688 }, { "epoch": 0.09, "learning_rate": 9.07924629159428e-08, "loss": 0.0526, "step": 689 }, { "epoch": 0.09, "learning_rate": 9.077909929172791e-08, "loss": 0.074, "step": 690 }, { "epoch": 0.09, "learning_rate": 9.076573566751302e-08, "loss": 0.0499, "step": 691 }, { "epoch": 0.09, "learning_rate": 9.075237204329814e-08, "loss": 0.0485, "step": 692 }, { "epoch": 0.09, "learning_rate": 9.073900841908325e-08, "loss": 0.0416, "step": 693 }, { "epoch": 0.09, "learning_rate": 9.072564479486836e-08, "loss": 0.0612, "step": 694 }, { "epoch": 0.09, "learning_rate": 9.071228117065349e-08, "loss": 0.0409, "step": 695 }, { "epoch": 0.09, "learning_rate": 9.06989175464386e-08, "loss": 0.0422, "step": 696 }, { "epoch": 0.09, "learning_rate": 9.068555392222371e-08, "loss": 0.0571, "step": 697 }, { "epoch": 0.09, "learning_rate": 9.067219029800882e-08, "loss": 0.0413, "step": 698 }, { "epoch": 0.09, "learning_rate": 9.065882667379393e-08, "loss": 0.0374, "step": 699 }, { "epoch": 0.09, "learning_rate": 9.064546304957904e-08, "loss": 0.0312, "step": 700 }, { "epoch": 0.09, "learning_rate": 9.063209942536415e-08, "loss": 0.0476, "step": 701 }, { "epoch": 0.09, "learning_rate": 9.061873580114927e-08, "loss": 0.0518, "step": 702 }, { "epoch": 0.09, "learning_rate": 9.060537217693438e-08, "loss": 0.0805, "step": 703 }, { "epoch": 0.09, "learning_rate": 9.059200855271949e-08, "loss": 0.0386, "step": 704 }, { "epoch": 0.09, "learning_rate": 9.057864492850461e-08, "loss": 0.047, "step": 705 }, { "epoch": 0.09, "learning_rate": 9.056528130428972e-08, "loss": 0.0614, "step": 706 }, { "epoch": 0.09, "learning_rate": 9.055191768007483e-08, "loss": 0.0512, "step": 707 }, { "epoch": 0.09, "learning_rate": 9.053855405585995e-08, "loss": 0.038, "step": 708 }, { "epoch": 0.09, "learning_rate": 9.052519043164506e-08, "loss": 0.0561, "step": 709 }, { "epoch": 0.09, "learning_rate": 9.051182680743017e-08, "loss": 0.0419, "step": 710 }, { "epoch": 0.1, "learning_rate": 9.049846318321528e-08, "loss": 0.0372, "step": 711 }, { "epoch": 0.1, "learning_rate": 9.04850995590004e-08, "loss": 0.0507, "step": 712 }, { "epoch": 0.1, "learning_rate": 9.04717359347855e-08, "loss": 0.0404, "step": 713 }, { "epoch": 0.1, "learning_rate": 9.045837231057061e-08, "loss": 0.0349, "step": 714 }, { "epoch": 0.1, "learning_rate": 9.044500868635574e-08, "loss": 0.0449, "step": 715 }, { "epoch": 0.1, "learning_rate": 9.043164506214085e-08, "loss": 0.0461, "step": 716 }, { "epoch": 0.1, "learning_rate": 9.041828143792596e-08, "loss": 0.0537, "step": 717 }, { "epoch": 0.1, "learning_rate": 9.040491781371108e-08, "loss": 0.0609, "step": 718 }, { "epoch": 0.1, "learning_rate": 9.039155418949619e-08, "loss": 0.0451, "step": 719 }, { "epoch": 0.1, "learning_rate": 9.03781905652813e-08, "loss": 0.0737, "step": 720 }, { "epoch": 0.1, "learning_rate": 9.036482694106641e-08, "loss": 0.0606, "step": 721 }, { "epoch": 0.1, "learning_rate": 9.035146331685153e-08, "loss": 0.0279, "step": 722 }, { "epoch": 0.1, "learning_rate": 9.033809969263664e-08, "loss": 0.0678, "step": 723 }, { "epoch": 0.1, "learning_rate": 9.032473606842175e-08, "loss": 0.0368, "step": 724 }, { "epoch": 0.1, "learning_rate": 9.031137244420687e-08, "loss": 0.0454, "step": 725 }, { "epoch": 0.1, "learning_rate": 9.029800881999197e-08, "loss": 0.053, "step": 726 }, { "epoch": 0.1, "learning_rate": 9.028464519577708e-08, "loss": 0.0739, "step": 727 }, { "epoch": 0.1, "learning_rate": 9.02712815715622e-08, "loss": 0.0533, "step": 728 }, { "epoch": 0.1, "learning_rate": 9.025791794734731e-08, "loss": 0.0871, "step": 729 }, { "epoch": 0.1, "learning_rate": 9.024455432313242e-08, "loss": 0.0539, "step": 730 }, { "epoch": 0.1, "learning_rate": 9.023119069891755e-08, "loss": 0.0752, "step": 731 }, { "epoch": 0.1, "learning_rate": 9.021782707470266e-08, "loss": 0.0595, "step": 732 }, { "epoch": 0.1, "learning_rate": 9.020446345048777e-08, "loss": 0.0482, "step": 733 }, { "epoch": 0.1, "learning_rate": 9.019109982627288e-08, "loss": 0.0503, "step": 734 }, { "epoch": 0.1, "learning_rate": 9.0177736202058e-08, "loss": 0.0869, "step": 735 }, { "epoch": 0.1, "learning_rate": 9.016437257784311e-08, "loss": 0.0551, "step": 736 }, { "epoch": 0.1, "learning_rate": 9.015100895362822e-08, "loss": 0.0724, "step": 737 }, { "epoch": 0.1, "learning_rate": 9.013764532941334e-08, "loss": 0.0517, "step": 738 }, { "epoch": 0.1, "learning_rate": 9.012428170519845e-08, "loss": 0.0367, "step": 739 }, { "epoch": 0.1, "learning_rate": 9.011091808098356e-08, "loss": 0.074, "step": 740 }, { "epoch": 0.1, "learning_rate": 9.009755445676867e-08, "loss": 0.0642, "step": 741 }, { "epoch": 0.1, "learning_rate": 9.008419083255378e-08, "loss": 0.0403, "step": 742 }, { "epoch": 0.1, "learning_rate": 9.007082720833889e-08, "loss": 0.0366, "step": 743 }, { "epoch": 0.1, "learning_rate": 9.0057463584124e-08, "loss": 0.0466, "step": 744 }, { "epoch": 0.1, "learning_rate": 9.004409995990912e-08, "loss": 0.0425, "step": 745 }, { "epoch": 0.1, "learning_rate": 9.003073633569423e-08, "loss": 0.0554, "step": 746 }, { "epoch": 0.1, "learning_rate": 9.001737271147934e-08, "loss": 0.0468, "step": 747 }, { "epoch": 0.1, "learning_rate": 9.000400908726447e-08, "loss": 0.0463, "step": 748 }, { "epoch": 0.1, "learning_rate": 8.999064546304958e-08, "loss": 0.0443, "step": 749 }, { "epoch": 0.1, "learning_rate": 8.997728183883469e-08, "loss": 0.0717, "step": 750 }, { "epoch": 0.1, "learning_rate": 8.996391821461981e-08, "loss": 0.0565, "step": 751 }, { "epoch": 0.1, "learning_rate": 8.995055459040492e-08, "loss": 0.0659, "step": 752 }, { "epoch": 0.1, "learning_rate": 8.993719096619003e-08, "loss": 0.0416, "step": 753 }, { "epoch": 0.1, "learning_rate": 8.992382734197514e-08, "loss": 0.0424, "step": 754 }, { "epoch": 0.1, "learning_rate": 8.991046371776025e-08, "loss": 0.0564, "step": 755 }, { "epoch": 0.1, "learning_rate": 8.989710009354536e-08, "loss": 0.0367, "step": 756 }, { "epoch": 0.1, "learning_rate": 8.988373646933047e-08, "loss": 0.052, "step": 757 }, { "epoch": 0.1, "learning_rate": 8.987037284511559e-08, "loss": 0.0848, "step": 758 }, { "epoch": 0.1, "learning_rate": 8.98570092209007e-08, "loss": 0.0373, "step": 759 }, { "epoch": 0.1, "learning_rate": 8.984364559668581e-08, "loss": 0.0419, "step": 760 }, { "epoch": 0.1, "learning_rate": 8.983028197247093e-08, "loss": 0.0514, "step": 761 }, { "epoch": 0.1, "learning_rate": 8.981691834825604e-08, "loss": 0.0679, "step": 762 }, { "epoch": 0.1, "learning_rate": 8.980355472404115e-08, "loss": 0.0439, "step": 763 }, { "epoch": 0.1, "learning_rate": 8.979019109982628e-08, "loss": 0.0476, "step": 764 }, { "epoch": 0.1, "learning_rate": 8.977682747561139e-08, "loss": 0.0496, "step": 765 }, { "epoch": 0.1, "learning_rate": 8.97634638513965e-08, "loss": 0.0481, "step": 766 }, { "epoch": 0.1, "learning_rate": 8.975010022718161e-08, "loss": 0.0529, "step": 767 }, { "epoch": 0.1, "learning_rate": 8.973673660296672e-08, "loss": 0.0461, "step": 768 }, { "epoch": 0.1, "learning_rate": 8.972337297875183e-08, "loss": 0.0326, "step": 769 }, { "epoch": 0.1, "learning_rate": 8.971000935453694e-08, "loss": 0.0438, "step": 770 }, { "epoch": 0.1, "learning_rate": 8.969664573032206e-08, "loss": 0.0385, "step": 771 }, { "epoch": 0.1, "learning_rate": 8.968328210610717e-08, "loss": 0.0462, "step": 772 }, { "epoch": 0.1, "learning_rate": 8.966991848189228e-08, "loss": 0.0404, "step": 773 }, { "epoch": 0.1, "learning_rate": 8.96565548576774e-08, "loss": 0.0444, "step": 774 }, { "epoch": 0.1, "learning_rate": 8.964319123346251e-08, "loss": 0.0407, "step": 775 }, { "epoch": 0.1, "learning_rate": 8.962982760924762e-08, "loss": 0.0825, "step": 776 }, { "epoch": 0.1, "learning_rate": 8.961646398503274e-08, "loss": 0.0562, "step": 777 }, { "epoch": 0.1, "learning_rate": 8.960310036081785e-08, "loss": 0.0361, "step": 778 }, { "epoch": 0.1, "learning_rate": 8.958973673660296e-08, "loss": 0.0472, "step": 779 }, { "epoch": 0.1, "learning_rate": 8.957637311238807e-08, "loss": 0.0498, "step": 780 }, { "epoch": 0.1, "learning_rate": 8.95630094881732e-08, "loss": 0.0548, "step": 781 }, { "epoch": 0.1, "learning_rate": 8.954964586395831e-08, "loss": 0.044, "step": 782 }, { "epoch": 0.1, "learning_rate": 8.95362822397434e-08, "loss": 0.0715, "step": 783 }, { "epoch": 0.1, "learning_rate": 8.952291861552853e-08, "loss": 0.043, "step": 784 }, { "epoch": 0.1, "learning_rate": 8.950955499131364e-08, "loss": 0.0608, "step": 785 }, { "epoch": 0.11, "learning_rate": 8.949619136709875e-08, "loss": 0.091, "step": 786 }, { "epoch": 0.11, "learning_rate": 8.948282774288387e-08, "loss": 0.0367, "step": 787 }, { "epoch": 0.11, "learning_rate": 8.946946411866898e-08, "loss": 0.1016, "step": 788 }, { "epoch": 0.11, "learning_rate": 8.945610049445409e-08, "loss": 0.0495, "step": 789 }, { "epoch": 0.11, "learning_rate": 8.94427368702392e-08, "loss": 0.0542, "step": 790 }, { "epoch": 0.11, "learning_rate": 8.942937324602432e-08, "loss": 0.0486, "step": 791 }, { "epoch": 0.11, "learning_rate": 8.941600962180943e-08, "loss": 0.0537, "step": 792 }, { "epoch": 0.11, "learning_rate": 8.940264599759454e-08, "loss": 0.0533, "step": 793 }, { "epoch": 0.11, "learning_rate": 8.938928237337966e-08, "loss": 0.0563, "step": 794 }, { "epoch": 0.11, "learning_rate": 8.937591874916477e-08, "loss": 0.0587, "step": 795 }, { "epoch": 0.11, "learning_rate": 8.936255512494988e-08, "loss": 0.0494, "step": 796 }, { "epoch": 0.11, "learning_rate": 8.9349191500735e-08, "loss": 0.0444, "step": 797 }, { "epoch": 0.11, "learning_rate": 8.93358278765201e-08, "loss": 0.0698, "step": 798 }, { "epoch": 0.11, "learning_rate": 8.932246425230521e-08, "loss": 0.0575, "step": 799 }, { "epoch": 0.11, "learning_rate": 8.930910062809034e-08, "loss": 0.0516, "step": 800 }, { "epoch": 0.11, "learning_rate": 8.929573700387545e-08, "loss": 0.0582, "step": 801 }, { "epoch": 0.11, "learning_rate": 8.928237337966056e-08, "loss": 0.0543, "step": 802 }, { "epoch": 0.11, "learning_rate": 8.926900975544567e-08, "loss": 0.0362, "step": 803 }, { "epoch": 0.11, "learning_rate": 8.925564613123079e-08, "loss": 0.0381, "step": 804 }, { "epoch": 0.11, "learning_rate": 8.92422825070159e-08, "loss": 0.0592, "step": 805 }, { "epoch": 0.11, "learning_rate": 8.922891888280101e-08, "loss": 0.0522, "step": 806 }, { "epoch": 0.11, "learning_rate": 8.921555525858613e-08, "loss": 0.0658, "step": 807 }, { "epoch": 0.11, "learning_rate": 8.920219163437124e-08, "loss": 0.0407, "step": 808 }, { "epoch": 0.11, "learning_rate": 8.918882801015635e-08, "loss": 0.0463, "step": 809 }, { "epoch": 0.11, "learning_rate": 8.917546438594146e-08, "loss": 0.1152, "step": 810 }, { "epoch": 0.11, "learning_rate": 8.916210076172657e-08, "loss": 0.0475, "step": 811 }, { "epoch": 0.11, "learning_rate": 8.914873713751168e-08, "loss": 0.0595, "step": 812 }, { "epoch": 0.11, "learning_rate": 8.913537351329679e-08, "loss": 0.0416, "step": 813 }, { "epoch": 0.11, "learning_rate": 8.912200988908191e-08, "loss": 0.0382, "step": 814 }, { "epoch": 0.11, "learning_rate": 8.910864626486702e-08, "loss": 0.0523, "step": 815 }, { "epoch": 0.11, "learning_rate": 8.909528264065213e-08, "loss": 0.0588, "step": 816 }, { "epoch": 0.11, "learning_rate": 8.908191901643726e-08, "loss": 0.0488, "step": 817 }, { "epoch": 0.11, "learning_rate": 8.906855539222237e-08, "loss": 0.0622, "step": 818 }, { "epoch": 0.11, "learning_rate": 8.905519176800748e-08, "loss": 0.0582, "step": 819 }, { "epoch": 0.11, "learning_rate": 8.90418281437926e-08, "loss": 0.0628, "step": 820 }, { "epoch": 0.11, "learning_rate": 8.902846451957771e-08, "loss": 0.0345, "step": 821 }, { "epoch": 0.11, "learning_rate": 8.901510089536282e-08, "loss": 0.061, "step": 822 }, { "epoch": 0.11, "learning_rate": 8.900173727114794e-08, "loss": 0.052, "step": 823 }, { "epoch": 0.11, "learning_rate": 8.898837364693305e-08, "loss": 0.0381, "step": 824 }, { "epoch": 0.11, "learning_rate": 8.897501002271815e-08, "loss": 0.0555, "step": 825 }, { "epoch": 0.11, "learning_rate": 8.896164639850326e-08, "loss": 0.0459, "step": 826 }, { "epoch": 0.11, "learning_rate": 8.894828277428838e-08, "loss": 0.0408, "step": 827 }, { "epoch": 0.11, "learning_rate": 8.893491915007349e-08, "loss": 0.0623, "step": 828 }, { "epoch": 0.11, "learning_rate": 8.89215555258586e-08, "loss": 0.0434, "step": 829 }, { "epoch": 0.11, "learning_rate": 8.890819190164372e-08, "loss": 0.0731, "step": 830 }, { "epoch": 0.11, "learning_rate": 8.889482827742883e-08, "loss": 0.0318, "step": 831 }, { "epoch": 0.11, "learning_rate": 8.888146465321394e-08, "loss": 0.0338, "step": 832 }, { "epoch": 0.11, "learning_rate": 8.886810102899907e-08, "loss": 0.0349, "step": 833 }, { "epoch": 0.11, "learning_rate": 8.885473740478418e-08, "loss": 0.0609, "step": 834 }, { "epoch": 0.11, "learning_rate": 8.884137378056929e-08, "loss": 0.0408, "step": 835 }, { "epoch": 0.11, "learning_rate": 8.88280101563544e-08, "loss": 0.0864, "step": 836 }, { "epoch": 0.11, "learning_rate": 8.881464653213952e-08, "loss": 0.07, "step": 837 }, { "epoch": 0.11, "learning_rate": 8.880128290792463e-08, "loss": 0.0704, "step": 838 }, { "epoch": 0.11, "learning_rate": 8.878791928370974e-08, "loss": 0.0559, "step": 839 }, { "epoch": 0.11, "learning_rate": 8.877455565949485e-08, "loss": 0.0657, "step": 840 }, { "epoch": 0.11, "learning_rate": 8.876119203527996e-08, "loss": 0.0349, "step": 841 }, { "epoch": 0.11, "learning_rate": 8.874782841106507e-08, "loss": 0.029, "step": 842 }, { "epoch": 0.11, "learning_rate": 8.873446478685019e-08, "loss": 0.0429, "step": 843 }, { "epoch": 0.11, "learning_rate": 8.87211011626353e-08, "loss": 0.0477, "step": 844 }, { "epoch": 0.11, "learning_rate": 8.870773753842041e-08, "loss": 0.0541, "step": 845 }, { "epoch": 0.11, "learning_rate": 8.869437391420554e-08, "loss": 0.0467, "step": 846 }, { "epoch": 0.11, "learning_rate": 8.868101028999064e-08, "loss": 0.0423, "step": 847 }, { "epoch": 0.11, "learning_rate": 8.866764666577575e-08, "loss": 0.0437, "step": 848 }, { "epoch": 0.11, "learning_rate": 8.865428304156086e-08, "loss": 0.0715, "step": 849 }, { "epoch": 0.11, "learning_rate": 8.864091941734599e-08, "loss": 0.0498, "step": 850 }, { "epoch": 0.11, "learning_rate": 8.86275557931311e-08, "loss": 0.0833, "step": 851 }, { "epoch": 0.11, "learning_rate": 8.861419216891621e-08, "loss": 0.0354, "step": 852 }, { "epoch": 0.11, "learning_rate": 8.860082854470132e-08, "loss": 0.0693, "step": 853 }, { "epoch": 0.11, "learning_rate": 8.858746492048643e-08, "loss": 0.0418, "step": 854 }, { "epoch": 0.11, "learning_rate": 8.857410129627154e-08, "loss": 0.0405, "step": 855 }, { "epoch": 0.11, "learning_rate": 8.856073767205666e-08, "loss": 0.0868, "step": 856 }, { "epoch": 0.11, "learning_rate": 8.854737404784177e-08, "loss": 0.0482, "step": 857 }, { "epoch": 0.11, "learning_rate": 8.853401042362688e-08, "loss": 0.0513, "step": 858 }, { "epoch": 0.11, "learning_rate": 8.852064679941199e-08, "loss": 0.0514, "step": 859 }, { "epoch": 0.11, "learning_rate": 8.850728317519711e-08, "loss": 0.0491, "step": 860 }, { "epoch": 0.12, "learning_rate": 8.849391955098222e-08, "loss": 0.0345, "step": 861 }, { "epoch": 0.12, "learning_rate": 8.848055592676733e-08, "loss": 0.0411, "step": 862 }, { "epoch": 0.12, "learning_rate": 8.846719230255246e-08, "loss": 0.0539, "step": 863 }, { "epoch": 0.12, "learning_rate": 8.845382867833756e-08, "loss": 0.0422, "step": 864 }, { "epoch": 0.12, "learning_rate": 8.844046505412267e-08, "loss": 0.0439, "step": 865 }, { "epoch": 0.12, "learning_rate": 8.842710142990778e-08, "loss": 0.0304, "step": 866 }, { "epoch": 0.12, "learning_rate": 8.84137378056929e-08, "loss": 0.057, "step": 867 }, { "epoch": 0.12, "learning_rate": 8.8400374181478e-08, "loss": 0.0462, "step": 868 }, { "epoch": 0.12, "learning_rate": 8.838701055726313e-08, "loss": 0.0573, "step": 869 }, { "epoch": 0.12, "learning_rate": 8.837364693304824e-08, "loss": 0.0538, "step": 870 }, { "epoch": 0.12, "learning_rate": 8.836028330883335e-08, "loss": 0.0604, "step": 871 }, { "epoch": 0.12, "learning_rate": 8.834691968461846e-08, "loss": 0.0349, "step": 872 }, { "epoch": 0.12, "learning_rate": 8.833355606040358e-08, "loss": 0.0445, "step": 873 }, { "epoch": 0.12, "learning_rate": 8.832019243618869e-08, "loss": 0.0431, "step": 874 }, { "epoch": 0.12, "learning_rate": 8.83068288119738e-08, "loss": 0.05, "step": 875 }, { "epoch": 0.12, "learning_rate": 8.829346518775892e-08, "loss": 0.0512, "step": 876 }, { "epoch": 0.12, "learning_rate": 8.828010156354403e-08, "loss": 0.0514, "step": 877 }, { "epoch": 0.12, "learning_rate": 8.826673793932914e-08, "loss": 0.0373, "step": 878 }, { "epoch": 0.12, "learning_rate": 8.825337431511427e-08, "loss": 0.042, "step": 879 }, { "epoch": 0.12, "learning_rate": 8.824001069089938e-08, "loss": 0.0422, "step": 880 }, { "epoch": 0.12, "learning_rate": 8.822664706668449e-08, "loss": 0.0527, "step": 881 }, { "epoch": 0.12, "learning_rate": 8.82132834424696e-08, "loss": 0.0521, "step": 882 }, { "epoch": 0.12, "learning_rate": 8.81999198182547e-08, "loss": 0.0464, "step": 883 }, { "epoch": 0.12, "learning_rate": 8.818655619403981e-08, "loss": 0.0472, "step": 884 }, { "epoch": 0.12, "learning_rate": 8.817319256982492e-08, "loss": 0.0557, "step": 885 }, { "epoch": 0.12, "learning_rate": 8.815982894561005e-08, "loss": 0.028, "step": 886 }, { "epoch": 0.12, "learning_rate": 8.814646532139516e-08, "loss": 0.0708, "step": 887 }, { "epoch": 0.12, "learning_rate": 8.813310169718027e-08, "loss": 0.0371, "step": 888 }, { "epoch": 0.12, "learning_rate": 8.811973807296539e-08, "loss": 0.0294, "step": 889 }, { "epoch": 0.12, "learning_rate": 8.81063744487505e-08, "loss": 0.0417, "step": 890 }, { "epoch": 0.12, "learning_rate": 8.809301082453561e-08, "loss": 0.0601, "step": 891 }, { "epoch": 0.12, "learning_rate": 8.807964720032073e-08, "loss": 0.0513, "step": 892 }, { "epoch": 0.12, "learning_rate": 8.806628357610584e-08, "loss": 0.0567, "step": 893 }, { "epoch": 0.12, "learning_rate": 8.805291995189095e-08, "loss": 0.0512, "step": 894 }, { "epoch": 0.12, "learning_rate": 8.803955632767606e-08, "loss": 0.0532, "step": 895 }, { "epoch": 0.12, "learning_rate": 8.802619270346117e-08, "loss": 0.0352, "step": 896 }, { "epoch": 0.12, "learning_rate": 8.801282907924628e-08, "loss": 0.0627, "step": 897 }, { "epoch": 0.12, "learning_rate": 8.799946545503139e-08, "loss": 0.0545, "step": 898 }, { "epoch": 0.12, "learning_rate": 8.798610183081652e-08, "loss": 0.0388, "step": 899 }, { "epoch": 0.12, "learning_rate": 8.797273820660162e-08, "loss": 0.0541, "step": 900 }, { "epoch": 0.12, "learning_rate": 8.795937458238673e-08, "loss": 0.0446, "step": 901 }, { "epoch": 0.12, "learning_rate": 8.794601095817186e-08, "loss": 0.0824, "step": 902 }, { "epoch": 0.12, "learning_rate": 8.793264733395697e-08, "loss": 0.0381, "step": 903 }, { "epoch": 0.12, "learning_rate": 8.791928370974208e-08, "loss": 0.0353, "step": 904 }, { "epoch": 0.12, "learning_rate": 8.79059200855272e-08, "loss": 0.0556, "step": 905 }, { "epoch": 0.12, "learning_rate": 8.789255646131231e-08, "loss": 0.0605, "step": 906 }, { "epoch": 0.12, "learning_rate": 8.787919283709742e-08, "loss": 0.0514, "step": 907 }, { "epoch": 0.12, "learning_rate": 8.786582921288253e-08, "loss": 0.0469, "step": 908 }, { "epoch": 0.12, "learning_rate": 8.785246558866764e-08, "loss": 0.0358, "step": 909 }, { "epoch": 0.12, "learning_rate": 8.783910196445275e-08, "loss": 0.0539, "step": 910 }, { "epoch": 0.12, "learning_rate": 8.782573834023786e-08, "loss": 0.0657, "step": 911 }, { "epoch": 0.12, "learning_rate": 8.781237471602298e-08, "loss": 0.0572, "step": 912 }, { "epoch": 0.12, "learning_rate": 8.779901109180809e-08, "loss": 0.0453, "step": 913 }, { "epoch": 0.12, "learning_rate": 8.77856474675932e-08, "loss": 0.0633, "step": 914 }, { "epoch": 0.12, "learning_rate": 8.777228384337833e-08, "loss": 0.0588, "step": 915 }, { "epoch": 0.12, "learning_rate": 8.775892021916344e-08, "loss": 0.0616, "step": 916 }, { "epoch": 0.12, "learning_rate": 8.774555659494854e-08, "loss": 0.0382, "step": 917 }, { "epoch": 0.12, "learning_rate": 8.773219297073365e-08, "loss": 0.0591, "step": 918 }, { "epoch": 0.12, "learning_rate": 8.771882934651878e-08, "loss": 0.0659, "step": 919 }, { "epoch": 0.12, "learning_rate": 8.770546572230389e-08, "loss": 0.041, "step": 920 }, { "epoch": 0.12, "learning_rate": 8.7692102098089e-08, "loss": 0.0371, "step": 921 }, { "epoch": 0.12, "learning_rate": 8.767873847387412e-08, "loss": 0.0382, "step": 922 }, { "epoch": 0.12, "learning_rate": 8.766537484965923e-08, "loss": 0.0624, "step": 923 }, { "epoch": 0.12, "learning_rate": 8.765201122544433e-08, "loss": 0.0556, "step": 924 }, { "epoch": 0.12, "learning_rate": 8.763864760122945e-08, "loss": 0.081, "step": 925 }, { "epoch": 0.12, "learning_rate": 8.762528397701456e-08, "loss": 0.0719, "step": 926 }, { "epoch": 0.12, "learning_rate": 8.761192035279967e-08, "loss": 0.0603, "step": 927 }, { "epoch": 0.12, "learning_rate": 8.759855672858479e-08, "loss": 0.0298, "step": 928 }, { "epoch": 0.12, "learning_rate": 8.75851931043699e-08, "loss": 0.0658, "step": 929 }, { "epoch": 0.12, "learning_rate": 8.757182948015501e-08, "loss": 0.0691, "step": 930 }, { "epoch": 0.12, "learning_rate": 8.755846585594012e-08, "loss": 0.0508, "step": 931 }, { "epoch": 0.12, "learning_rate": 8.754510223172525e-08, "loss": 0.0392, "step": 932 }, { "epoch": 0.12, "learning_rate": 8.753173860751036e-08, "loss": 0.0461, "step": 933 }, { "epoch": 0.12, "learning_rate": 8.751837498329547e-08, "loss": 0.0435, "step": 934 }, { "epoch": 0.12, "learning_rate": 8.750501135908059e-08, "loss": 0.0568, "step": 935 }, { "epoch": 0.13, "learning_rate": 8.74916477348657e-08, "loss": 0.0607, "step": 936 }, { "epoch": 0.13, "learning_rate": 8.747828411065081e-08, "loss": 0.059, "step": 937 }, { "epoch": 0.13, "learning_rate": 8.746492048643592e-08, "loss": 0.0652, "step": 938 }, { "epoch": 0.13, "learning_rate": 8.745155686222103e-08, "loss": 0.0572, "step": 939 }, { "epoch": 0.13, "learning_rate": 8.743819323800614e-08, "loss": 0.0896, "step": 940 }, { "epoch": 0.13, "learning_rate": 8.742482961379125e-08, "loss": 0.0437, "step": 941 }, { "epoch": 0.13, "learning_rate": 8.741146598957637e-08, "loss": 0.0373, "step": 942 }, { "epoch": 0.13, "learning_rate": 8.739810236536148e-08, "loss": 0.0326, "step": 943 }, { "epoch": 0.13, "learning_rate": 8.738473874114659e-08, "loss": 0.0721, "step": 944 }, { "epoch": 0.13, "learning_rate": 8.737137511693171e-08, "loss": 0.0581, "step": 945 }, { "epoch": 0.13, "learning_rate": 8.735801149271682e-08, "loss": 0.0574, "step": 946 }, { "epoch": 0.13, "learning_rate": 8.734464786850193e-08, "loss": 0.0425, "step": 947 }, { "epoch": 0.13, "learning_rate": 8.733128424428706e-08, "loss": 0.0312, "step": 948 }, { "epoch": 0.13, "learning_rate": 8.731792062007217e-08, "loss": 0.073, "step": 949 }, { "epoch": 0.13, "learning_rate": 8.730455699585728e-08, "loss": 0.0545, "step": 950 }, { "epoch": 0.13, "learning_rate": 8.729119337164239e-08, "loss": 0.0296, "step": 951 }, { "epoch": 0.13, "learning_rate": 8.72778297474275e-08, "loss": 0.0671, "step": 952 }, { "epoch": 0.13, "learning_rate": 8.72644661232126e-08, "loss": 0.0515, "step": 953 }, { "epoch": 0.13, "learning_rate": 8.725110249899771e-08, "loss": 0.0665, "step": 954 }, { "epoch": 0.13, "learning_rate": 8.723773887478284e-08, "loss": 0.0594, "step": 955 }, { "epoch": 0.13, "learning_rate": 8.722437525056795e-08, "loss": 0.0705, "step": 956 }, { "epoch": 0.13, "learning_rate": 8.721101162635306e-08, "loss": 0.0519, "step": 957 }, { "epoch": 0.13, "learning_rate": 8.719764800213818e-08, "loss": 0.0734, "step": 958 }, { "epoch": 0.13, "learning_rate": 8.718428437792329e-08, "loss": 0.0389, "step": 959 }, { "epoch": 0.13, "learning_rate": 8.71709207537084e-08, "loss": 0.0543, "step": 960 }, { "epoch": 0.13, "learning_rate": 8.715755712949352e-08, "loss": 0.0564, "step": 961 }, { "epoch": 0.13, "learning_rate": 8.714419350527863e-08, "loss": 0.0351, "step": 962 }, { "epoch": 0.13, "learning_rate": 8.713082988106374e-08, "loss": 0.0453, "step": 963 }, { "epoch": 0.13, "learning_rate": 8.711746625684885e-08, "loss": 0.0364, "step": 964 }, { "epoch": 0.13, "learning_rate": 8.710410263263396e-08, "loss": 0.0667, "step": 965 }, { "epoch": 0.13, "learning_rate": 8.709073900841907e-08, "loss": 0.0598, "step": 966 }, { "epoch": 0.13, "learning_rate": 8.707737538420418e-08, "loss": 0.0534, "step": 967 }, { "epoch": 0.13, "learning_rate": 8.70640117599893e-08, "loss": 0.0371, "step": 968 }, { "epoch": 0.13, "learning_rate": 8.705064813577442e-08, "loss": 0.062, "step": 969 }, { "epoch": 0.13, "learning_rate": 8.703728451155952e-08, "loss": 0.0466, "step": 970 }, { "epoch": 0.13, "learning_rate": 8.702392088734465e-08, "loss": 0.0548, "step": 971 }, { "epoch": 0.13, "learning_rate": 8.701055726312976e-08, "loss": 0.0571, "step": 972 }, { "epoch": 0.13, "learning_rate": 8.699719363891487e-08, "loss": 0.0322, "step": 973 }, { "epoch": 0.13, "learning_rate": 8.698383001469999e-08, "loss": 0.066, "step": 974 }, { "epoch": 0.13, "learning_rate": 8.69704663904851e-08, "loss": 0.0433, "step": 975 }, { "epoch": 0.13, "learning_rate": 8.695710276627021e-08, "loss": 0.0481, "step": 976 }, { "epoch": 0.13, "learning_rate": 8.694373914205532e-08, "loss": 0.0533, "step": 977 }, { "epoch": 0.13, "learning_rate": 8.693037551784044e-08, "loss": 0.0599, "step": 978 }, { "epoch": 0.13, "learning_rate": 8.691701189362555e-08, "loss": 0.0668, "step": 979 }, { "epoch": 0.13, "learning_rate": 8.690364826941066e-08, "loss": 0.0452, "step": 980 }, { "epoch": 0.13, "learning_rate": 8.689028464519577e-08, "loss": 0.0516, "step": 981 }, { "epoch": 0.13, "learning_rate": 8.687692102098088e-08, "loss": 0.0713, "step": 982 }, { "epoch": 0.13, "learning_rate": 8.686355739676599e-08, "loss": 0.0471, "step": 983 }, { "epoch": 0.13, "learning_rate": 8.685019377255112e-08, "loss": 0.0331, "step": 984 }, { "epoch": 0.13, "learning_rate": 8.683683014833623e-08, "loss": 0.0327, "step": 985 }, { "epoch": 0.13, "learning_rate": 8.682346652412134e-08, "loss": 0.057, "step": 986 }, { "epoch": 0.13, "learning_rate": 8.681010289990645e-08, "loss": 0.0336, "step": 987 }, { "epoch": 0.13, "learning_rate": 8.679673927569157e-08, "loss": 0.0367, "step": 988 }, { "epoch": 0.13, "learning_rate": 8.678337565147668e-08, "loss": 0.0395, "step": 989 }, { "epoch": 0.13, "learning_rate": 8.677001202726179e-08, "loss": 0.0657, "step": 990 }, { "epoch": 0.13, "learning_rate": 8.675664840304691e-08, "loss": 0.0863, "step": 991 }, { "epoch": 0.13, "learning_rate": 8.674328477883202e-08, "loss": 0.0439, "step": 992 }, { "epoch": 0.13, "learning_rate": 8.672992115461713e-08, "loss": 0.0525, "step": 993 }, { "epoch": 0.13, "learning_rate": 8.671655753040224e-08, "loss": 0.0539, "step": 994 }, { "epoch": 0.13, "learning_rate": 8.670319390618735e-08, "loss": 0.0427, "step": 995 }, { "epoch": 0.13, "learning_rate": 8.668983028197246e-08, "loss": 0.052, "step": 996 }, { "epoch": 0.13, "learning_rate": 8.667646665775758e-08, "loss": 0.0603, "step": 997 }, { "epoch": 0.13, "learning_rate": 8.666310303354269e-08, "loss": 0.0454, "step": 998 }, { "epoch": 0.13, "learning_rate": 8.66497394093278e-08, "loss": 0.0652, "step": 999 }, { "epoch": 0.13, "learning_rate": 8.663637578511291e-08, "loss": 0.0519, "step": 1000 }, { "epoch": 0.13, "learning_rate": 8.662301216089804e-08, "loss": 0.0664, "step": 1001 }, { "epoch": 0.13, "learning_rate": 8.660964853668315e-08, "loss": 0.0555, "step": 1002 }, { "epoch": 0.13, "learning_rate": 8.659628491246826e-08, "loss": 0.0888, "step": 1003 }, { "epoch": 0.13, "learning_rate": 8.658292128825338e-08, "loss": 0.0307, "step": 1004 }, { "epoch": 0.13, "learning_rate": 8.656955766403849e-08, "loss": 0.0422, "step": 1005 }, { "epoch": 0.13, "learning_rate": 8.65561940398236e-08, "loss": 0.0596, "step": 1006 }, { "epoch": 0.13, "learning_rate": 8.654283041560871e-08, "loss": 0.0713, "step": 1007 }, { "epoch": 0.13, "learning_rate": 8.652946679139382e-08, "loss": 0.0534, "step": 1008 }, { "epoch": 0.13, "learning_rate": 8.651610316717893e-08, "loss": 0.0596, "step": 1009 }, { "epoch": 0.13, "learning_rate": 8.650273954296404e-08, "loss": 0.0432, "step": 1010 }, { "epoch": 0.14, "learning_rate": 8.648937591874916e-08, "loss": 0.067, "step": 1011 }, { "epoch": 0.14, "learning_rate": 8.647601229453427e-08, "loss": 0.0476, "step": 1012 }, { "epoch": 0.14, "learning_rate": 8.646264867031938e-08, "loss": 0.037, "step": 1013 }, { "epoch": 0.14, "learning_rate": 8.64492850461045e-08, "loss": 0.0656, "step": 1014 }, { "epoch": 0.14, "learning_rate": 8.643592142188961e-08, "loss": 0.0672, "step": 1015 }, { "epoch": 0.14, "learning_rate": 8.642255779767472e-08, "loss": 0.0667, "step": 1016 }, { "epoch": 0.14, "learning_rate": 8.640919417345985e-08, "loss": 0.0467, "step": 1017 }, { "epoch": 0.14, "learning_rate": 8.639583054924496e-08, "loss": 0.0458, "step": 1018 }, { "epoch": 0.14, "learning_rate": 8.638246692503007e-08, "loss": 0.0604, "step": 1019 }, { "epoch": 0.14, "learning_rate": 8.636910330081519e-08, "loss": 0.0428, "step": 1020 }, { "epoch": 0.14, "learning_rate": 8.63557396766003e-08, "loss": 0.0648, "step": 1021 }, { "epoch": 0.14, "learning_rate": 8.63423760523854e-08, "loss": 0.049, "step": 1022 }, { "epoch": 0.14, "learning_rate": 8.63290124281705e-08, "loss": 0.047, "step": 1023 }, { "epoch": 0.14, "learning_rate": 8.631564880395563e-08, "loss": 0.0402, "step": 1024 }, { "epoch": 0.14, "learning_rate": 8.630228517974074e-08, "loss": 0.0572, "step": 1025 }, { "epoch": 0.14, "learning_rate": 8.628892155552585e-08, "loss": 0.0521, "step": 1026 }, { "epoch": 0.14, "learning_rate": 8.627555793131097e-08, "loss": 0.0414, "step": 1027 }, { "epoch": 0.14, "learning_rate": 8.626219430709608e-08, "loss": 0.0477, "step": 1028 }, { "epoch": 0.14, "learning_rate": 8.624883068288119e-08, "loss": 0.0496, "step": 1029 }, { "epoch": 0.14, "learning_rate": 8.623546705866631e-08, "loss": 0.029, "step": 1030 }, { "epoch": 0.14, "learning_rate": 8.622210343445142e-08, "loss": 0.0566, "step": 1031 }, { "epoch": 0.14, "learning_rate": 8.620873981023653e-08, "loss": 0.0448, "step": 1032 }, { "epoch": 0.14, "learning_rate": 8.619537618602164e-08, "loss": 0.0487, "step": 1033 }, { "epoch": 0.14, "learning_rate": 8.618201256180677e-08, "loss": 0.0614, "step": 1034 }, { "epoch": 0.14, "learning_rate": 8.616864893759188e-08, "loss": 0.034, "step": 1035 }, { "epoch": 0.14, "learning_rate": 8.615528531337699e-08, "loss": 0.0441, "step": 1036 }, { "epoch": 0.14, "learning_rate": 8.61419216891621e-08, "loss": 0.0349, "step": 1037 }, { "epoch": 0.14, "learning_rate": 8.61285580649472e-08, "loss": 0.1038, "step": 1038 }, { "epoch": 0.14, "learning_rate": 8.611519444073232e-08, "loss": 0.0711, "step": 1039 }, { "epoch": 0.14, "learning_rate": 8.610183081651744e-08, "loss": 0.0464, "step": 1040 }, { "epoch": 0.14, "learning_rate": 8.608846719230255e-08, "loss": 0.0716, "step": 1041 }, { "epoch": 0.14, "learning_rate": 8.607510356808766e-08, "loss": 0.0424, "step": 1042 }, { "epoch": 0.14, "learning_rate": 8.606173994387278e-08, "loss": 0.0666, "step": 1043 }, { "epoch": 0.14, "learning_rate": 8.604837631965789e-08, "loss": 0.0483, "step": 1044 }, { "epoch": 0.14, "learning_rate": 8.6035012695443e-08, "loss": 0.0447, "step": 1045 }, { "epoch": 0.14, "learning_rate": 8.602164907122811e-08, "loss": 0.051, "step": 1046 }, { "epoch": 0.14, "learning_rate": 8.600828544701323e-08, "loss": 0.0744, "step": 1047 }, { "epoch": 0.14, "learning_rate": 8.599492182279834e-08, "loss": 0.0456, "step": 1048 }, { "epoch": 0.14, "learning_rate": 8.598155819858345e-08, "loss": 0.0934, "step": 1049 }, { "epoch": 0.14, "learning_rate": 8.596819457436856e-08, "loss": 0.0422, "step": 1050 }, { "epoch": 0.14, "learning_rate": 8.595483095015367e-08, "loss": 0.0392, "step": 1051 }, { "epoch": 0.14, "learning_rate": 8.594146732593878e-08, "loss": 0.0573, "step": 1052 }, { "epoch": 0.14, "learning_rate": 8.59281037017239e-08, "loss": 0.0407, "step": 1053 }, { "epoch": 0.14, "learning_rate": 8.591474007750902e-08, "loss": 0.0494, "step": 1054 }, { "epoch": 0.14, "learning_rate": 8.590137645329413e-08, "loss": 0.0385, "step": 1055 }, { "epoch": 0.14, "learning_rate": 8.588801282907924e-08, "loss": 0.0495, "step": 1056 }, { "epoch": 0.14, "learning_rate": 8.587464920486436e-08, "loss": 0.0554, "step": 1057 }, { "epoch": 0.14, "learning_rate": 8.586128558064947e-08, "loss": 0.0848, "step": 1058 }, { "epoch": 0.14, "learning_rate": 8.584792195643458e-08, "loss": 0.1001, "step": 1059 }, { "epoch": 0.14, "learning_rate": 8.58345583322197e-08, "loss": 0.0677, "step": 1060 }, { "epoch": 0.14, "learning_rate": 8.582119470800481e-08, "loss": 0.0698, "step": 1061 }, { "epoch": 0.14, "learning_rate": 8.580783108378992e-08, "loss": 0.0407, "step": 1062 }, { "epoch": 0.14, "learning_rate": 8.579446745957504e-08, "loss": 0.0659, "step": 1063 }, { "epoch": 0.14, "learning_rate": 8.578110383536014e-08, "loss": 0.0486, "step": 1064 }, { "epoch": 0.14, "learning_rate": 8.576774021114525e-08, "loss": 0.0384, "step": 1065 }, { "epoch": 0.14, "learning_rate": 8.575437658693037e-08, "loss": 0.0697, "step": 1066 }, { "epoch": 0.14, "learning_rate": 8.574101296271548e-08, "loss": 0.0633, "step": 1067 }, { "epoch": 0.14, "learning_rate": 8.572764933850059e-08, "loss": 0.0466, "step": 1068 }, { "epoch": 0.14, "learning_rate": 8.57142857142857e-08, "loss": 0.0472, "step": 1069 }, { "epoch": 0.14, "learning_rate": 8.570092209007083e-08, "loss": 0.0458, "step": 1070 }, { "epoch": 0.14, "learning_rate": 8.568755846585594e-08, "loss": 0.0534, "step": 1071 }, { "epoch": 0.14, "learning_rate": 8.567419484164105e-08, "loss": 0.0402, "step": 1072 }, { "epoch": 0.14, "learning_rate": 8.566083121742617e-08, "loss": 0.0479, "step": 1073 }, { "epoch": 0.14, "learning_rate": 8.564746759321128e-08, "loss": 0.0367, "step": 1074 }, { "epoch": 0.14, "learning_rate": 8.563410396899639e-08, "loss": 0.0388, "step": 1075 }, { "epoch": 0.14, "learning_rate": 8.562074034478151e-08, "loss": 0.0513, "step": 1076 }, { "epoch": 0.14, "learning_rate": 8.560737672056662e-08, "loss": 0.0434, "step": 1077 }, { "epoch": 0.14, "learning_rate": 8.559401309635173e-08, "loss": 0.0466, "step": 1078 }, { "epoch": 0.14, "learning_rate": 8.558064947213684e-08, "loss": 0.0551, "step": 1079 }, { "epoch": 0.14, "learning_rate": 8.556728584792195e-08, "loss": 0.0617, "step": 1080 }, { "epoch": 0.14, "learning_rate": 8.555392222370706e-08, "loss": 0.0552, "step": 1081 }, { "epoch": 0.14, "learning_rate": 8.554055859949217e-08, "loss": 0.0622, "step": 1082 }, { "epoch": 0.14, "learning_rate": 8.55271949752773e-08, "loss": 0.0572, "step": 1083 }, { "epoch": 0.14, "learning_rate": 8.55138313510624e-08, "loss": 0.0517, "step": 1084 }, { "epoch": 0.14, "learning_rate": 8.550046772684751e-08, "loss": 0.0641, "step": 1085 }, { "epoch": 0.15, "learning_rate": 8.548710410263264e-08, "loss": 0.043, "step": 1086 }, { "epoch": 0.15, "learning_rate": 8.547374047841775e-08, "loss": 0.049, "step": 1087 }, { "epoch": 0.15, "learning_rate": 8.546037685420286e-08, "loss": 0.0662, "step": 1088 }, { "epoch": 0.15, "learning_rate": 8.544701322998798e-08, "loss": 0.0576, "step": 1089 }, { "epoch": 0.15, "learning_rate": 8.543364960577309e-08, "loss": 0.0303, "step": 1090 }, { "epoch": 0.15, "learning_rate": 8.54202859815582e-08, "loss": 0.0553, "step": 1091 }, { "epoch": 0.15, "learning_rate": 8.540692235734331e-08, "loss": 0.0398, "step": 1092 }, { "epoch": 0.15, "learning_rate": 8.539355873312842e-08, "loss": 0.063, "step": 1093 }, { "epoch": 0.15, "learning_rate": 8.538019510891353e-08, "loss": 0.0621, "step": 1094 }, { "epoch": 0.15, "learning_rate": 8.536683148469864e-08, "loss": 0.0448, "step": 1095 }, { "epoch": 0.15, "learning_rate": 8.535346786048376e-08, "loss": 0.0905, "step": 1096 }, { "epoch": 0.15, "learning_rate": 8.534010423626887e-08, "loss": 0.0616, "step": 1097 }, { "epoch": 0.15, "learning_rate": 8.532674061205398e-08, "loss": 0.0519, "step": 1098 }, { "epoch": 0.15, "learning_rate": 8.53133769878391e-08, "loss": 0.0811, "step": 1099 }, { "epoch": 0.15, "learning_rate": 8.530001336362421e-08, "loss": 0.0873, "step": 1100 }, { "epoch": 0.15, "learning_rate": 8.528664973940932e-08, "loss": 0.0824, "step": 1101 }, { "epoch": 0.15, "learning_rate": 8.527328611519443e-08, "loss": 0.0405, "step": 1102 }, { "epoch": 0.15, "learning_rate": 8.525992249097956e-08, "loss": 0.0382, "step": 1103 }, { "epoch": 0.15, "learning_rate": 8.524655886676467e-08, "loss": 0.0492, "step": 1104 }, { "epoch": 0.15, "learning_rate": 8.523319524254978e-08, "loss": 0.0421, "step": 1105 }, { "epoch": 0.15, "learning_rate": 8.521983161833489e-08, "loss": 0.0382, "step": 1106 }, { "epoch": 0.15, "learning_rate": 8.520646799412e-08, "loss": 0.0324, "step": 1107 }, { "epoch": 0.15, "learning_rate": 8.51931043699051e-08, "loss": 0.0548, "step": 1108 }, { "epoch": 0.15, "learning_rate": 8.517974074569023e-08, "loss": 0.0399, "step": 1109 }, { "epoch": 0.15, "learning_rate": 8.516637712147534e-08, "loss": 0.0435, "step": 1110 }, { "epoch": 0.15, "learning_rate": 8.515301349726045e-08, "loss": 0.059, "step": 1111 }, { "epoch": 0.15, "learning_rate": 8.513964987304557e-08, "loss": 0.0567, "step": 1112 }, { "epoch": 0.15, "learning_rate": 8.512628624883068e-08, "loss": 0.0365, "step": 1113 }, { "epoch": 0.15, "learning_rate": 8.511292262461579e-08, "loss": 0.0389, "step": 1114 }, { "epoch": 0.15, "learning_rate": 8.50995590004009e-08, "loss": 0.049, "step": 1115 }, { "epoch": 0.15, "learning_rate": 8.508619537618602e-08, "loss": 0.0496, "step": 1116 }, { "epoch": 0.15, "learning_rate": 8.507283175197113e-08, "loss": 0.1104, "step": 1117 }, { "epoch": 0.15, "learning_rate": 8.505946812775624e-08, "loss": 0.0359, "step": 1118 }, { "epoch": 0.15, "learning_rate": 8.504610450354137e-08, "loss": 0.0618, "step": 1119 }, { "epoch": 0.15, "learning_rate": 8.503274087932648e-08, "loss": 0.0809, "step": 1120 }, { "epoch": 0.15, "learning_rate": 8.501937725511157e-08, "loss": 0.0312, "step": 1121 }, { "epoch": 0.15, "learning_rate": 8.50060136308967e-08, "loss": 0.0581, "step": 1122 }, { "epoch": 0.15, "learning_rate": 8.49926500066818e-08, "loss": 0.047, "step": 1123 }, { "epoch": 0.15, "learning_rate": 8.497928638246692e-08, "loss": 0.0543, "step": 1124 }, { "epoch": 0.15, "learning_rate": 8.496592275825203e-08, "loss": 0.082, "step": 1125 }, { "epoch": 0.15, "learning_rate": 8.495255913403715e-08, "loss": 0.0534, "step": 1126 }, { "epoch": 0.15, "learning_rate": 8.493919550982226e-08, "loss": 0.0441, "step": 1127 }, { "epoch": 0.15, "learning_rate": 8.492583188560737e-08, "loss": 0.0489, "step": 1128 }, { "epoch": 0.15, "learning_rate": 8.491246826139249e-08, "loss": 0.0299, "step": 1129 }, { "epoch": 0.15, "learning_rate": 8.48991046371776e-08, "loss": 0.05, "step": 1130 }, { "epoch": 0.15, "learning_rate": 8.488574101296271e-08, "loss": 0.0378, "step": 1131 }, { "epoch": 0.15, "learning_rate": 8.487237738874783e-08, "loss": 0.0233, "step": 1132 }, { "epoch": 0.15, "learning_rate": 8.485901376453294e-08, "loss": 0.0589, "step": 1133 }, { "epoch": 0.15, "learning_rate": 8.484565014031805e-08, "loss": 0.0394, "step": 1134 }, { "epoch": 0.15, "learning_rate": 8.483228651610316e-08, "loss": 0.0584, "step": 1135 }, { "epoch": 0.15, "learning_rate": 8.481892289188827e-08, "loss": 0.0564, "step": 1136 }, { "epoch": 0.15, "learning_rate": 8.480555926767338e-08, "loss": 0.0452, "step": 1137 }, { "epoch": 0.15, "learning_rate": 8.479219564345849e-08, "loss": 0.0468, "step": 1138 }, { "epoch": 0.15, "learning_rate": 8.477883201924362e-08, "loss": 0.0519, "step": 1139 }, { "epoch": 0.15, "learning_rate": 8.476546839502873e-08, "loss": 0.155, "step": 1140 }, { "epoch": 0.15, "learning_rate": 8.475210477081384e-08, "loss": 0.0612, "step": 1141 }, { "epoch": 0.15, "learning_rate": 8.473874114659896e-08, "loss": 0.0661, "step": 1142 }, { "epoch": 0.15, "learning_rate": 8.472537752238407e-08, "loss": 0.0698, "step": 1143 }, { "epoch": 0.15, "learning_rate": 8.471201389816918e-08, "loss": 0.0458, "step": 1144 }, { "epoch": 0.15, "learning_rate": 8.46986502739543e-08, "loss": 0.0405, "step": 1145 }, { "epoch": 0.15, "learning_rate": 8.468528664973941e-08, "loss": 0.0494, "step": 1146 }, { "epoch": 0.15, "learning_rate": 8.467192302552452e-08, "loss": 0.0377, "step": 1147 }, { "epoch": 0.15, "learning_rate": 8.465855940130963e-08, "loss": 0.056, "step": 1148 }, { "epoch": 0.15, "learning_rate": 8.464519577709474e-08, "loss": 0.0357, "step": 1149 }, { "epoch": 0.15, "learning_rate": 8.463183215287985e-08, "loss": 0.059, "step": 1150 }, { "epoch": 0.15, "learning_rate": 8.461846852866496e-08, "loss": 0.0633, "step": 1151 }, { "epoch": 0.15, "learning_rate": 8.460510490445008e-08, "loss": 0.0732, "step": 1152 }, { "epoch": 0.15, "learning_rate": 8.45917412802352e-08, "loss": 0.074, "step": 1153 }, { "epoch": 0.15, "learning_rate": 8.45783776560203e-08, "loss": 0.0709, "step": 1154 }, { "epoch": 0.15, "learning_rate": 8.456501403180543e-08, "loss": 0.0294, "step": 1155 }, { "epoch": 0.15, "learning_rate": 8.455165040759054e-08, "loss": 0.0573, "step": 1156 }, { "epoch": 0.15, "learning_rate": 8.453828678337565e-08, "loss": 0.0619, "step": 1157 }, { "epoch": 0.15, "learning_rate": 8.452492315916077e-08, "loss": 0.0904, "step": 1158 }, { "epoch": 0.15, "learning_rate": 8.451155953494588e-08, "loss": 0.0697, "step": 1159 }, { "epoch": 0.16, "learning_rate": 8.449819591073099e-08, "loss": 0.064, "step": 1160 }, { "epoch": 0.16, "learning_rate": 8.44848322865161e-08, "loss": 0.0474, "step": 1161 }, { "epoch": 0.16, "learning_rate": 8.447146866230122e-08, "loss": 0.0574, "step": 1162 }, { "epoch": 0.16, "learning_rate": 8.445810503808632e-08, "loss": 0.0347, "step": 1163 }, { "epoch": 0.16, "learning_rate": 8.444474141387143e-08, "loss": 0.0607, "step": 1164 }, { "epoch": 0.16, "learning_rate": 8.443137778965655e-08, "loss": 0.0334, "step": 1165 }, { "epoch": 0.16, "learning_rate": 8.441801416544166e-08, "loss": 0.039, "step": 1166 }, { "epoch": 0.16, "learning_rate": 8.440465054122677e-08, "loss": 0.0545, "step": 1167 }, { "epoch": 0.16, "learning_rate": 8.43912869170119e-08, "loss": 0.0713, "step": 1168 }, { "epoch": 0.16, "learning_rate": 8.4377923292797e-08, "loss": 0.0446, "step": 1169 }, { "epoch": 0.16, "learning_rate": 8.436455966858211e-08, "loss": 0.0738, "step": 1170 }, { "epoch": 0.16, "learning_rate": 8.435119604436722e-08, "loss": 0.09, "step": 1171 }, { "epoch": 0.16, "learning_rate": 8.433783242015235e-08, "loss": 0.0345, "step": 1172 }, { "epoch": 0.16, "learning_rate": 8.432446879593746e-08, "loss": 0.07, "step": 1173 }, { "epoch": 0.16, "learning_rate": 8.431110517172257e-08, "loss": 0.0493, "step": 1174 }, { "epoch": 0.16, "learning_rate": 8.429774154750769e-08, "loss": 0.0571, "step": 1175 }, { "epoch": 0.16, "learning_rate": 8.42843779232928e-08, "loss": 0.121, "step": 1176 }, { "epoch": 0.16, "learning_rate": 8.427101429907791e-08, "loss": 0.0417, "step": 1177 }, { "epoch": 0.16, "learning_rate": 8.425765067486302e-08, "loss": 0.0589, "step": 1178 }, { "epoch": 0.16, "learning_rate": 8.424428705064813e-08, "loss": 0.0397, "step": 1179 }, { "epoch": 0.16, "learning_rate": 8.423092342643324e-08, "loss": 0.0586, "step": 1180 }, { "epoch": 0.16, "learning_rate": 8.421755980221836e-08, "loss": 0.0521, "step": 1181 }, { "epoch": 0.16, "learning_rate": 8.420419617800347e-08, "loss": 0.0412, "step": 1182 }, { "epoch": 0.16, "learning_rate": 8.419083255378858e-08, "loss": 0.0485, "step": 1183 }, { "epoch": 0.16, "learning_rate": 8.417746892957369e-08, "loss": 0.0498, "step": 1184 }, { "epoch": 0.16, "learning_rate": 8.416410530535881e-08, "loss": 0.0446, "step": 1185 }, { "epoch": 0.16, "learning_rate": 8.415074168114392e-08, "loss": 0.046, "step": 1186 }, { "epoch": 0.16, "learning_rate": 8.413737805692903e-08, "loss": 0.0423, "step": 1187 }, { "epoch": 0.16, "learning_rate": 8.412401443271416e-08, "loss": 0.0557, "step": 1188 }, { "epoch": 0.16, "learning_rate": 8.411065080849927e-08, "loss": 0.0449, "step": 1189 }, { "epoch": 0.16, "learning_rate": 8.409728718428438e-08, "loss": 0.0416, "step": 1190 }, { "epoch": 0.16, "learning_rate": 8.408392356006949e-08, "loss": 0.0458, "step": 1191 }, { "epoch": 0.16, "learning_rate": 8.40705599358546e-08, "loss": 0.0553, "step": 1192 }, { "epoch": 0.16, "learning_rate": 8.40571963116397e-08, "loss": 0.0558, "step": 1193 }, { "epoch": 0.16, "learning_rate": 8.404383268742482e-08, "loss": 0.05, "step": 1194 }, { "epoch": 0.16, "learning_rate": 8.403046906320994e-08, "loss": 0.0461, "step": 1195 }, { "epoch": 0.16, "learning_rate": 8.401710543899505e-08, "loss": 0.0528, "step": 1196 }, { "epoch": 0.16, "learning_rate": 8.400374181478016e-08, "loss": 0.0507, "step": 1197 }, { "epoch": 0.16, "learning_rate": 8.399037819056528e-08, "loss": 0.0422, "step": 1198 }, { "epoch": 0.16, "learning_rate": 8.397701456635039e-08, "loss": 0.0601, "step": 1199 }, { "epoch": 0.16, "learning_rate": 8.39636509421355e-08, "loss": 0.0437, "step": 1200 }, { "epoch": 0.16, "learning_rate": 8.395028731792062e-08, "loss": 0.0666, "step": 1201 }, { "epoch": 0.16, "learning_rate": 8.393692369370573e-08, "loss": 0.0574, "step": 1202 }, { "epoch": 0.16, "learning_rate": 8.392356006949084e-08, "loss": 0.0573, "step": 1203 }, { "epoch": 0.16, "learning_rate": 8.391019644527595e-08, "loss": 0.0458, "step": 1204 }, { "epoch": 0.16, "learning_rate": 8.389683282106106e-08, "loss": 0.0461, "step": 1205 }, { "epoch": 0.16, "learning_rate": 8.388346919684617e-08, "loss": 0.0603, "step": 1206 }, { "epoch": 0.16, "learning_rate": 8.387010557263128e-08, "loss": 0.0614, "step": 1207 }, { "epoch": 0.16, "learning_rate": 8.38567419484164e-08, "loss": 0.0555, "step": 1208 }, { "epoch": 0.16, "learning_rate": 8.384337832420152e-08, "loss": 0.0575, "step": 1209 }, { "epoch": 0.16, "learning_rate": 8.383001469998663e-08, "loss": 0.0607, "step": 1210 }, { "epoch": 0.16, "learning_rate": 8.381665107577175e-08, "loss": 0.0672, "step": 1211 }, { "epoch": 0.16, "learning_rate": 8.380328745155686e-08, "loss": 0.0777, "step": 1212 }, { "epoch": 0.16, "learning_rate": 8.378992382734197e-08, "loss": 0.0648, "step": 1213 }, { "epoch": 0.16, "learning_rate": 8.377656020312709e-08, "loss": 0.055, "step": 1214 }, { "epoch": 0.16, "learning_rate": 8.37631965789122e-08, "loss": 0.0783, "step": 1215 }, { "epoch": 0.16, "learning_rate": 8.374983295469731e-08, "loss": 0.0675, "step": 1216 }, { "epoch": 0.16, "learning_rate": 8.373646933048242e-08, "loss": 0.0506, "step": 1217 }, { "epoch": 0.16, "learning_rate": 8.372310570626754e-08, "loss": 0.0416, "step": 1218 }, { "epoch": 0.16, "learning_rate": 8.370974208205265e-08, "loss": 0.0632, "step": 1219 }, { "epoch": 0.16, "learning_rate": 8.369637845783775e-08, "loss": 0.058, "step": 1220 }, { "epoch": 0.16, "learning_rate": 8.368301483362287e-08, "loss": 0.0572, "step": 1221 }, { "epoch": 0.16, "learning_rate": 8.366965120940798e-08, "loss": 0.055, "step": 1222 }, { "epoch": 0.16, "learning_rate": 8.36562875851931e-08, "loss": 0.0564, "step": 1223 }, { "epoch": 0.16, "learning_rate": 8.364292396097822e-08, "loss": 0.0543, "step": 1224 }, { "epoch": 0.16, "learning_rate": 8.362956033676333e-08, "loss": 0.0403, "step": 1225 }, { "epoch": 0.16, "learning_rate": 8.361619671254844e-08, "loss": 0.0603, "step": 1226 }, { "epoch": 0.16, "learning_rate": 8.360283308833356e-08, "loss": 0.0481, "step": 1227 }, { "epoch": 0.16, "learning_rate": 8.358946946411867e-08, "loss": 0.0722, "step": 1228 }, { "epoch": 0.16, "learning_rate": 8.357610583990378e-08, "loss": 0.0414, "step": 1229 }, { "epoch": 0.16, "learning_rate": 8.356274221568889e-08, "loss": 0.0486, "step": 1230 }, { "epoch": 0.16, "learning_rate": 8.354937859147401e-08, "loss": 0.0504, "step": 1231 }, { "epoch": 0.16, "learning_rate": 8.353601496725912e-08, "loss": 0.064, "step": 1232 }, { "epoch": 0.16, "learning_rate": 8.352265134304423e-08, "loss": 0.0573, "step": 1233 }, { "epoch": 0.16, "learning_rate": 8.350928771882934e-08, "loss": 0.0396, "step": 1234 }, { "epoch": 0.17, "learning_rate": 8.349592409461445e-08, "loss": 0.0336, "step": 1235 }, { "epoch": 0.17, "learning_rate": 8.348256047039956e-08, "loss": 0.0568, "step": 1236 }, { "epoch": 0.17, "learning_rate": 8.346919684618468e-08, "loss": 0.0535, "step": 1237 }, { "epoch": 0.17, "learning_rate": 8.34558332219698e-08, "loss": 0.0438, "step": 1238 }, { "epoch": 0.17, "learning_rate": 8.34424695977549e-08, "loss": 0.0457, "step": 1239 }, { "epoch": 0.17, "learning_rate": 8.342910597354001e-08, "loss": 0.0469, "step": 1240 }, { "epoch": 0.17, "learning_rate": 8.341574234932514e-08, "loss": 0.0562, "step": 1241 }, { "epoch": 0.17, "learning_rate": 8.340237872511025e-08, "loss": 0.0623, "step": 1242 }, { "epoch": 0.17, "learning_rate": 8.338901510089536e-08, "loss": 0.0472, "step": 1243 }, { "epoch": 0.17, "learning_rate": 8.337565147668048e-08, "loss": 0.0287, "step": 1244 }, { "epoch": 0.17, "learning_rate": 8.336228785246559e-08, "loss": 0.0629, "step": 1245 }, { "epoch": 0.17, "learning_rate": 8.33489242282507e-08, "loss": 0.0433, "step": 1246 }, { "epoch": 0.17, "learning_rate": 8.333556060403581e-08, "loss": 0.0409, "step": 1247 }, { "epoch": 0.17, "learning_rate": 8.332219697982092e-08, "loss": 0.0508, "step": 1248 }, { "epoch": 0.17, "learning_rate": 8.330883335560603e-08, "loss": 0.0854, "step": 1249 }, { "epoch": 0.17, "learning_rate": 8.329546973139115e-08, "loss": 0.0559, "step": 1250 }, { "epoch": 0.17, "learning_rate": 8.328210610717626e-08, "loss": 0.0437, "step": 1251 }, { "epoch": 0.17, "learning_rate": 8.326874248296137e-08, "loss": 0.0594, "step": 1252 }, { "epoch": 0.17, "learning_rate": 8.325537885874648e-08, "loss": 0.0333, "step": 1253 }, { "epoch": 0.17, "learning_rate": 8.32420152345316e-08, "loss": 0.0429, "step": 1254 }, { "epoch": 0.17, "learning_rate": 8.322865161031671e-08, "loss": 0.0661, "step": 1255 }, { "epoch": 0.17, "learning_rate": 8.321528798610182e-08, "loss": 0.0363, "step": 1256 }, { "epoch": 0.17, "learning_rate": 8.320192436188695e-08, "loss": 0.0547, "step": 1257 }, { "epoch": 0.17, "learning_rate": 8.318856073767206e-08, "loss": 0.0493, "step": 1258 }, { "epoch": 0.17, "learning_rate": 8.317519711345717e-08, "loss": 0.062, "step": 1259 }, { "epoch": 0.17, "learning_rate": 8.316183348924229e-08, "loss": 0.0578, "step": 1260 }, { "epoch": 0.17, "learning_rate": 8.314846986502739e-08, "loss": 0.0387, "step": 1261 }, { "epoch": 0.17, "learning_rate": 8.31351062408125e-08, "loss": 0.0589, "step": 1262 }, { "epoch": 0.17, "learning_rate": 8.31217426165976e-08, "loss": 0.064, "step": 1263 }, { "epoch": 0.17, "learning_rate": 8.310837899238273e-08, "loss": 0.0651, "step": 1264 }, { "epoch": 0.17, "learning_rate": 8.309501536816784e-08, "loss": 0.0501, "step": 1265 }, { "epoch": 0.17, "learning_rate": 8.308165174395295e-08, "loss": 0.0602, "step": 1266 }, { "epoch": 0.17, "learning_rate": 8.306828811973807e-08, "loss": 0.0659, "step": 1267 }, { "epoch": 0.17, "learning_rate": 8.305492449552318e-08, "loss": 0.0296, "step": 1268 }, { "epoch": 0.17, "learning_rate": 8.304156087130829e-08, "loss": 0.0472, "step": 1269 }, { "epoch": 0.17, "learning_rate": 8.302819724709341e-08, "loss": 0.056, "step": 1270 }, { "epoch": 0.17, "learning_rate": 8.301483362287852e-08, "loss": 0.0511, "step": 1271 }, { "epoch": 0.17, "learning_rate": 8.300146999866363e-08, "loss": 0.0465, "step": 1272 }, { "epoch": 0.17, "learning_rate": 8.298810637444876e-08, "loss": 0.0351, "step": 1273 }, { "epoch": 0.17, "learning_rate": 8.297474275023387e-08, "loss": 0.062, "step": 1274 }, { "epoch": 0.17, "learning_rate": 8.296137912601898e-08, "loss": 0.035, "step": 1275 }, { "epoch": 0.17, "learning_rate": 8.294801550180409e-08, "loss": 0.0545, "step": 1276 }, { "epoch": 0.17, "learning_rate": 8.29346518775892e-08, "loss": 0.0363, "step": 1277 }, { "epoch": 0.17, "learning_rate": 8.29212882533743e-08, "loss": 0.0391, "step": 1278 }, { "epoch": 0.17, "learning_rate": 8.290792462915942e-08, "loss": 0.083, "step": 1279 }, { "epoch": 0.17, "learning_rate": 8.289456100494454e-08, "loss": 0.0726, "step": 1280 }, { "epoch": 0.17, "learning_rate": 8.288119738072965e-08, "loss": 0.0521, "step": 1281 }, { "epoch": 0.17, "learning_rate": 8.286783375651476e-08, "loss": 0.075, "step": 1282 }, { "epoch": 0.17, "learning_rate": 8.285447013229988e-08, "loss": 0.044, "step": 1283 }, { "epoch": 0.17, "learning_rate": 8.284110650808499e-08, "loss": 0.0619, "step": 1284 }, { "epoch": 0.17, "learning_rate": 8.28277428838701e-08, "loss": 0.059, "step": 1285 }, { "epoch": 0.17, "learning_rate": 8.281437925965521e-08, "loss": 0.0863, "step": 1286 }, { "epoch": 0.17, "learning_rate": 8.280101563544033e-08, "loss": 0.0622, "step": 1287 }, { "epoch": 0.17, "learning_rate": 8.278765201122544e-08, "loss": 0.0458, "step": 1288 }, { "epoch": 0.17, "learning_rate": 8.277428838701055e-08, "loss": 0.0542, "step": 1289 }, { "epoch": 0.17, "learning_rate": 8.276092476279566e-08, "loss": 0.0607, "step": 1290 }, { "epoch": 0.17, "learning_rate": 8.274756113858077e-08, "loss": 0.0421, "step": 1291 }, { "epoch": 0.17, "learning_rate": 8.273419751436588e-08, "loss": 0.0409, "step": 1292 }, { "epoch": 0.17, "learning_rate": 8.272083389015101e-08, "loss": 0.0683, "step": 1293 }, { "epoch": 0.17, "learning_rate": 8.270747026593612e-08, "loss": 0.0538, "step": 1294 }, { "epoch": 0.17, "learning_rate": 8.269410664172123e-08, "loss": 0.0568, "step": 1295 }, { "epoch": 0.17, "learning_rate": 8.268074301750635e-08, "loss": 0.0451, "step": 1296 }, { "epoch": 0.17, "learning_rate": 8.266737939329146e-08, "loss": 0.0399, "step": 1297 }, { "epoch": 0.17, "learning_rate": 8.265401576907657e-08, "loss": 0.0628, "step": 1298 }, { "epoch": 0.17, "learning_rate": 8.264065214486168e-08, "loss": 0.049, "step": 1299 }, { "epoch": 0.17, "learning_rate": 8.26272885206468e-08, "loss": 0.0557, "step": 1300 }, { "epoch": 0.17, "learning_rate": 8.261392489643191e-08, "loss": 0.0456, "step": 1301 }, { "epoch": 0.17, "learning_rate": 8.260056127221702e-08, "loss": 0.0609, "step": 1302 }, { "epoch": 0.17, "learning_rate": 8.258719764800213e-08, "loss": 0.0596, "step": 1303 }, { "epoch": 0.17, "learning_rate": 8.257383402378724e-08, "loss": 0.0369, "step": 1304 }, { "epoch": 0.17, "learning_rate": 8.256047039957235e-08, "loss": 0.1043, "step": 1305 }, { "epoch": 0.17, "learning_rate": 8.254710677535747e-08, "loss": 0.0511, "step": 1306 }, { "epoch": 0.17, "learning_rate": 8.253374315114258e-08, "loss": 0.0728, "step": 1307 }, { "epoch": 0.17, "learning_rate": 8.25203795269277e-08, "loss": 0.0611, "step": 1308 }, { "epoch": 0.17, "learning_rate": 8.25070159027128e-08, "loss": 0.0675, "step": 1309 }, { "epoch": 0.18, "learning_rate": 8.249365227849793e-08, "loss": 0.0477, "step": 1310 }, { "epoch": 0.18, "learning_rate": 8.248028865428304e-08, "loss": 0.0543, "step": 1311 }, { "epoch": 0.18, "learning_rate": 8.246692503006815e-08, "loss": 0.0632, "step": 1312 }, { "epoch": 0.18, "learning_rate": 8.245356140585327e-08, "loss": 0.0549, "step": 1313 }, { "epoch": 0.18, "learning_rate": 8.244019778163838e-08, "loss": 0.0479, "step": 1314 }, { "epoch": 0.18, "learning_rate": 8.242683415742349e-08, "loss": 0.0717, "step": 1315 }, { "epoch": 0.18, "learning_rate": 8.241347053320861e-08, "loss": 0.0574, "step": 1316 }, { "epoch": 0.18, "learning_rate": 8.240010690899372e-08, "loss": 0.0452, "step": 1317 }, { "epoch": 0.18, "learning_rate": 8.238674328477883e-08, "loss": 0.0354, "step": 1318 }, { "epoch": 0.18, "learning_rate": 8.237337966056394e-08, "loss": 0.0351, "step": 1319 }, { "epoch": 0.18, "learning_rate": 8.236001603634905e-08, "loss": 0.0383, "step": 1320 }, { "epoch": 0.18, "learning_rate": 8.234665241213416e-08, "loss": 0.0341, "step": 1321 }, { "epoch": 0.18, "learning_rate": 8.233328878791927e-08, "loss": 0.0699, "step": 1322 }, { "epoch": 0.18, "learning_rate": 8.23199251637044e-08, "loss": 0.0699, "step": 1323 }, { "epoch": 0.18, "learning_rate": 8.23065615394895e-08, "loss": 0.0629, "step": 1324 }, { "epoch": 0.18, "learning_rate": 8.229319791527461e-08, "loss": 0.0608, "step": 1325 }, { "epoch": 0.18, "learning_rate": 8.227983429105974e-08, "loss": 0.0439, "step": 1326 }, { "epoch": 0.18, "learning_rate": 8.226647066684485e-08, "loss": 0.0501, "step": 1327 }, { "epoch": 0.18, "learning_rate": 8.225310704262996e-08, "loss": 0.0365, "step": 1328 }, { "epoch": 0.18, "learning_rate": 8.223974341841508e-08, "loss": 0.0397, "step": 1329 }, { "epoch": 0.18, "learning_rate": 8.222637979420019e-08, "loss": 0.0519, "step": 1330 }, { "epoch": 0.18, "learning_rate": 8.22130161699853e-08, "loss": 0.0617, "step": 1331 }, { "epoch": 0.18, "learning_rate": 8.219965254577041e-08, "loss": 0.06, "step": 1332 }, { "epoch": 0.18, "learning_rate": 8.218628892155552e-08, "loss": 0.0457, "step": 1333 }, { "epoch": 0.18, "learning_rate": 8.217292529734063e-08, "loss": 0.0458, "step": 1334 }, { "epoch": 0.18, "learning_rate": 8.215956167312574e-08, "loss": 0.0599, "step": 1335 }, { "epoch": 0.18, "learning_rate": 8.214619804891086e-08, "loss": 0.0778, "step": 1336 }, { "epoch": 0.18, "learning_rate": 8.213283442469597e-08, "loss": 0.0515, "step": 1337 }, { "epoch": 0.18, "learning_rate": 8.211947080048108e-08, "loss": 0.0569, "step": 1338 }, { "epoch": 0.18, "learning_rate": 8.21061071762662e-08, "loss": 0.0401, "step": 1339 }, { "epoch": 0.18, "learning_rate": 8.209274355205131e-08, "loss": 0.0602, "step": 1340 }, { "epoch": 0.18, "learning_rate": 8.207937992783642e-08, "loss": 0.0439, "step": 1341 }, { "epoch": 0.18, "learning_rate": 8.206601630362155e-08, "loss": 0.0649, "step": 1342 }, { "epoch": 0.18, "learning_rate": 8.205265267940666e-08, "loss": 0.0409, "step": 1343 }, { "epoch": 0.18, "learning_rate": 8.203928905519177e-08, "loss": 0.0614, "step": 1344 }, { "epoch": 0.18, "learning_rate": 8.202592543097688e-08, "loss": 0.0533, "step": 1345 }, { "epoch": 0.18, "learning_rate": 8.201256180676199e-08, "loss": 0.0536, "step": 1346 }, { "epoch": 0.18, "learning_rate": 8.19991981825471e-08, "loss": 0.0499, "step": 1347 }, { "epoch": 0.18, "learning_rate": 8.198583455833221e-08, "loss": 0.0457, "step": 1348 }, { "epoch": 0.18, "learning_rate": 8.197247093411733e-08, "loss": 0.0334, "step": 1349 }, { "epoch": 0.18, "learning_rate": 8.195910730990244e-08, "loss": 0.0685, "step": 1350 }, { "epoch": 0.18, "learning_rate": 8.194574368568755e-08, "loss": 0.0716, "step": 1351 }, { "epoch": 0.18, "learning_rate": 8.193238006147267e-08, "loss": 0.0516, "step": 1352 }, { "epoch": 0.18, "learning_rate": 8.191901643725778e-08, "loss": 0.0488, "step": 1353 }, { "epoch": 0.18, "learning_rate": 8.190565281304289e-08, "loss": 0.0469, "step": 1354 }, { "epoch": 0.18, "learning_rate": 8.1892289188828e-08, "loss": 0.0496, "step": 1355 }, { "epoch": 0.18, "learning_rate": 8.187892556461312e-08, "loss": 0.0564, "step": 1356 }, { "epoch": 0.18, "learning_rate": 8.186556194039823e-08, "loss": 0.0511, "step": 1357 }, { "epoch": 0.18, "learning_rate": 8.185219831618334e-08, "loss": 0.0315, "step": 1358 }, { "epoch": 0.18, "learning_rate": 8.183883469196847e-08, "loss": 0.0497, "step": 1359 }, { "epoch": 0.18, "learning_rate": 8.182547106775356e-08, "loss": 0.0515, "step": 1360 }, { "epoch": 0.18, "learning_rate": 8.181210744353867e-08, "loss": 0.0412, "step": 1361 }, { "epoch": 0.18, "learning_rate": 8.17987438193238e-08, "loss": 0.0505, "step": 1362 }, { "epoch": 0.18, "learning_rate": 8.178538019510891e-08, "loss": 0.0535, "step": 1363 }, { "epoch": 0.18, "learning_rate": 8.177201657089402e-08, "loss": 0.0636, "step": 1364 }, { "epoch": 0.18, "learning_rate": 8.175865294667914e-08, "loss": 0.0565, "step": 1365 }, { "epoch": 0.18, "learning_rate": 8.174528932246425e-08, "loss": 0.0899, "step": 1366 }, { "epoch": 0.18, "learning_rate": 8.173192569824936e-08, "loss": 0.0363, "step": 1367 }, { "epoch": 0.18, "learning_rate": 8.171856207403447e-08, "loss": 0.056, "step": 1368 }, { "epoch": 0.18, "learning_rate": 8.170519844981959e-08, "loss": 0.0753, "step": 1369 }, { "epoch": 0.18, "learning_rate": 8.16918348256047e-08, "loss": 0.0375, "step": 1370 }, { "epoch": 0.18, "learning_rate": 8.167847120138981e-08, "loss": 0.059, "step": 1371 }, { "epoch": 0.18, "learning_rate": 8.166510757717494e-08, "loss": 0.0904, "step": 1372 }, { "epoch": 0.18, "learning_rate": 8.165174395296005e-08, "loss": 0.0784, "step": 1373 }, { "epoch": 0.18, "learning_rate": 8.163838032874515e-08, "loss": 0.0672, "step": 1374 }, { "epoch": 0.18, "learning_rate": 8.162501670453026e-08, "loss": 0.085, "step": 1375 }, { "epoch": 0.18, "learning_rate": 8.161165308031537e-08, "loss": 0.0377, "step": 1376 }, { "epoch": 0.18, "learning_rate": 8.159828945610048e-08, "loss": 0.028, "step": 1377 }, { "epoch": 0.18, "learning_rate": 8.15849258318856e-08, "loss": 0.0353, "step": 1378 }, { "epoch": 0.18, "learning_rate": 8.157156220767072e-08, "loss": 0.0442, "step": 1379 }, { "epoch": 0.18, "learning_rate": 8.155819858345583e-08, "loss": 0.0633, "step": 1380 }, { "epoch": 0.18, "learning_rate": 8.154483495924094e-08, "loss": 0.0383, "step": 1381 }, { "epoch": 0.18, "learning_rate": 8.153147133502606e-08, "loss": 0.0706, "step": 1382 }, { "epoch": 0.18, "learning_rate": 8.151810771081117e-08, "loss": 0.0417, "step": 1383 }, { "epoch": 0.18, "learning_rate": 8.150474408659628e-08, "loss": 0.0477, "step": 1384 }, { "epoch": 0.19, "learning_rate": 8.14913804623814e-08, "loss": 0.0414, "step": 1385 }, { "epoch": 0.19, "learning_rate": 8.147801683816651e-08, "loss": 0.0406, "step": 1386 }, { "epoch": 0.19, "learning_rate": 8.146465321395162e-08, "loss": 0.0478, "step": 1387 }, { "epoch": 0.19, "learning_rate": 8.145128958973673e-08, "loss": 0.0423, "step": 1388 }, { "epoch": 0.19, "learning_rate": 8.143792596552184e-08, "loss": 0.0284, "step": 1389 }, { "epoch": 0.19, "learning_rate": 8.142456234130695e-08, "loss": 0.0457, "step": 1390 }, { "epoch": 0.19, "learning_rate": 8.141119871709206e-08, "loss": 0.0539, "step": 1391 }, { "epoch": 0.19, "learning_rate": 8.139783509287718e-08, "loss": 0.0581, "step": 1392 }, { "epoch": 0.19, "learning_rate": 8.13844714686623e-08, "loss": 0.0522, "step": 1393 }, { "epoch": 0.19, "learning_rate": 8.13711078444474e-08, "loss": 0.0754, "step": 1394 }, { "epoch": 0.19, "learning_rate": 8.135774422023253e-08, "loss": 0.0477, "step": 1395 }, { "epoch": 0.19, "learning_rate": 8.134438059601764e-08, "loss": 0.0765, "step": 1396 }, { "epoch": 0.19, "learning_rate": 8.133101697180275e-08, "loss": 0.0625, "step": 1397 }, { "epoch": 0.19, "learning_rate": 8.131765334758787e-08, "loss": 0.0362, "step": 1398 }, { "epoch": 0.19, "learning_rate": 8.130428972337298e-08, "loss": 0.0339, "step": 1399 }, { "epoch": 0.19, "learning_rate": 8.129092609915809e-08, "loss": 0.0612, "step": 1400 }, { "epoch": 0.19, "learning_rate": 8.12775624749432e-08, "loss": 0.0495, "step": 1401 }, { "epoch": 0.19, "learning_rate": 8.126419885072831e-08, "loss": 0.0414, "step": 1402 }, { "epoch": 0.19, "learning_rate": 8.125083522651342e-08, "loss": 0.0538, "step": 1403 }, { "epoch": 0.19, "learning_rate": 8.123747160229853e-08, "loss": 0.0783, "step": 1404 }, { "epoch": 0.19, "learning_rate": 8.122410797808365e-08, "loss": 0.0538, "step": 1405 }, { "epoch": 0.19, "learning_rate": 8.121074435386876e-08, "loss": 0.0622, "step": 1406 }, { "epoch": 0.19, "learning_rate": 8.119738072965387e-08, "loss": 0.0582, "step": 1407 }, { "epoch": 0.19, "learning_rate": 8.1184017105439e-08, "loss": 0.0351, "step": 1408 }, { "epoch": 0.19, "learning_rate": 8.11706534812241e-08, "loss": 0.0484, "step": 1409 }, { "epoch": 0.19, "learning_rate": 8.115728985700921e-08, "loss": 0.055, "step": 1410 }, { "epoch": 0.19, "learning_rate": 8.114392623279434e-08, "loss": 0.0736, "step": 1411 }, { "epoch": 0.19, "learning_rate": 8.113056260857945e-08, "loss": 0.0663, "step": 1412 }, { "epoch": 0.19, "learning_rate": 8.111719898436456e-08, "loss": 0.0493, "step": 1413 }, { "epoch": 0.19, "learning_rate": 8.110383536014967e-08, "loss": 0.0558, "step": 1414 }, { "epoch": 0.19, "learning_rate": 8.109047173593479e-08, "loss": 0.042, "step": 1415 }, { "epoch": 0.19, "learning_rate": 8.10771081117199e-08, "loss": 0.059, "step": 1416 }, { "epoch": 0.19, "learning_rate": 8.106374448750501e-08, "loss": 0.0399, "step": 1417 }, { "epoch": 0.19, "learning_rate": 8.105038086329012e-08, "loss": 0.0615, "step": 1418 }, { "epoch": 0.19, "learning_rate": 8.103701723907523e-08, "loss": 0.0583, "step": 1419 }, { "epoch": 0.19, "learning_rate": 8.102365361486034e-08, "loss": 0.0442, "step": 1420 }, { "epoch": 0.19, "learning_rate": 8.101028999064546e-08, "loss": 0.0513, "step": 1421 }, { "epoch": 0.19, "learning_rate": 8.099692636643057e-08, "loss": 0.0483, "step": 1422 }, { "epoch": 0.19, "learning_rate": 8.098356274221568e-08, "loss": 0.0395, "step": 1423 }, { "epoch": 0.19, "learning_rate": 8.097019911800079e-08, "loss": 0.084, "step": 1424 }, { "epoch": 0.19, "learning_rate": 8.095683549378592e-08, "loss": 0.0654, "step": 1425 }, { "epoch": 0.19, "learning_rate": 8.094347186957103e-08, "loss": 0.0408, "step": 1426 }, { "epoch": 0.19, "learning_rate": 8.093010824535613e-08, "loss": 0.0921, "step": 1427 }, { "epoch": 0.19, "learning_rate": 8.091674462114126e-08, "loss": 0.0299, "step": 1428 }, { "epoch": 0.19, "learning_rate": 8.090338099692637e-08, "loss": 0.0601, "step": 1429 }, { "epoch": 0.19, "learning_rate": 8.089001737271148e-08, "loss": 0.0724, "step": 1430 }, { "epoch": 0.19, "learning_rate": 8.087665374849659e-08, "loss": 0.0624, "step": 1431 }, { "epoch": 0.19, "learning_rate": 8.08632901242817e-08, "loss": 0.0884, "step": 1432 }, { "epoch": 0.19, "learning_rate": 8.084992650006681e-08, "loss": 0.0645, "step": 1433 }, { "epoch": 0.19, "learning_rate": 8.083656287585193e-08, "loss": 0.0625, "step": 1434 }, { "epoch": 0.19, "learning_rate": 8.082319925163704e-08, "loss": 0.0548, "step": 1435 }, { "epoch": 0.19, "learning_rate": 8.080983562742215e-08, "loss": 0.0537, "step": 1436 }, { "epoch": 0.19, "learning_rate": 8.079647200320726e-08, "loss": 0.0494, "step": 1437 }, { "epoch": 0.19, "learning_rate": 8.078310837899238e-08, "loss": 0.0456, "step": 1438 }, { "epoch": 0.19, "learning_rate": 8.076974475477749e-08, "loss": 0.0476, "step": 1439 }, { "epoch": 0.19, "learning_rate": 8.07563811305626e-08, "loss": 0.0355, "step": 1440 }, { "epoch": 0.19, "learning_rate": 8.074301750634773e-08, "loss": 0.0585, "step": 1441 }, { "epoch": 0.19, "learning_rate": 8.072965388213284e-08, "loss": 0.0409, "step": 1442 }, { "epoch": 0.19, "learning_rate": 8.071629025791795e-08, "loss": 0.0421, "step": 1443 }, { "epoch": 0.19, "learning_rate": 8.070292663370306e-08, "loss": 0.0662, "step": 1444 }, { "epoch": 0.19, "learning_rate": 8.068956300948816e-08, "loss": 0.0607, "step": 1445 }, { "epoch": 0.19, "learning_rate": 8.067619938527327e-08, "loss": 0.0391, "step": 1446 }, { "epoch": 0.19, "learning_rate": 8.06628357610584e-08, "loss": 0.0515, "step": 1447 }, { "epoch": 0.19, "learning_rate": 8.064947213684351e-08, "loss": 0.0427, "step": 1448 }, { "epoch": 0.19, "learning_rate": 8.063610851262862e-08, "loss": 0.0422, "step": 1449 }, { "epoch": 0.19, "learning_rate": 8.062274488841373e-08, "loss": 0.0431, "step": 1450 }, { "epoch": 0.19, "learning_rate": 8.060938126419885e-08, "loss": 0.0381, "step": 1451 }, { "epoch": 0.19, "learning_rate": 8.059601763998396e-08, "loss": 0.0557, "step": 1452 }, { "epoch": 0.19, "learning_rate": 8.058265401576907e-08, "loss": 0.044, "step": 1453 }, { "epoch": 0.19, "learning_rate": 8.056929039155419e-08, "loss": 0.054, "step": 1454 }, { "epoch": 0.19, "learning_rate": 8.05559267673393e-08, "loss": 0.0486, "step": 1455 }, { "epoch": 0.19, "learning_rate": 8.054256314312441e-08, "loss": 0.0332, "step": 1456 }, { "epoch": 0.19, "learning_rate": 8.052919951890954e-08, "loss": 0.0466, "step": 1457 }, { "epoch": 0.19, "learning_rate": 8.051583589469465e-08, "loss": 0.0883, "step": 1458 }, { "epoch": 0.19, "learning_rate": 8.050247227047974e-08, "loss": 0.0606, "step": 1459 }, { "epoch": 0.2, "learning_rate": 8.048910864626485e-08, "loss": 0.0491, "step": 1460 }, { "epoch": 0.2, "learning_rate": 8.047574502204998e-08, "loss": 0.0458, "step": 1461 }, { "epoch": 0.2, "learning_rate": 8.046238139783508e-08, "loss": 0.0625, "step": 1462 }, { "epoch": 0.2, "learning_rate": 8.04490177736202e-08, "loss": 0.0417, "step": 1463 }, { "epoch": 0.2, "learning_rate": 8.043565414940532e-08, "loss": 0.075, "step": 1464 }, { "epoch": 0.2, "learning_rate": 8.042229052519043e-08, "loss": 0.037, "step": 1465 }, { "epoch": 0.2, "learning_rate": 8.040892690097554e-08, "loss": 0.0445, "step": 1466 }, { "epoch": 0.2, "learning_rate": 8.039556327676066e-08, "loss": 0.0447, "step": 1467 }, { "epoch": 0.2, "learning_rate": 8.038219965254577e-08, "loss": 0.0486, "step": 1468 }, { "epoch": 0.2, "learning_rate": 8.036883602833088e-08, "loss": 0.0512, "step": 1469 }, { "epoch": 0.2, "learning_rate": 8.0355472404116e-08, "loss": 0.0436, "step": 1470 }, { "epoch": 0.2, "learning_rate": 8.034210877990111e-08, "loss": 0.0425, "step": 1471 }, { "epoch": 0.2, "learning_rate": 8.032874515568622e-08, "loss": 0.0471, "step": 1472 }, { "epoch": 0.2, "learning_rate": 8.031538153147133e-08, "loss": 0.0588, "step": 1473 }, { "epoch": 0.2, "learning_rate": 8.030201790725644e-08, "loss": 0.0721, "step": 1474 }, { "epoch": 0.2, "learning_rate": 8.028865428304155e-08, "loss": 0.0504, "step": 1475 }, { "epoch": 0.2, "learning_rate": 8.027529065882666e-08, "loss": 0.0749, "step": 1476 }, { "epoch": 0.2, "learning_rate": 8.026192703461179e-08, "loss": 0.0466, "step": 1477 }, { "epoch": 0.2, "learning_rate": 8.02485634103969e-08, "loss": 0.0421, "step": 1478 }, { "epoch": 0.2, "learning_rate": 8.0235199786182e-08, "loss": 0.0402, "step": 1479 }, { "epoch": 0.2, "learning_rate": 8.022183616196713e-08, "loss": 0.0439, "step": 1480 }, { "epoch": 0.2, "learning_rate": 8.020847253775224e-08, "loss": 0.0492, "step": 1481 }, { "epoch": 0.2, "learning_rate": 8.019510891353735e-08, "loss": 0.0508, "step": 1482 }, { "epoch": 0.2, "learning_rate": 8.018174528932246e-08, "loss": 0.076, "step": 1483 }, { "epoch": 0.2, "learning_rate": 8.016838166510758e-08, "loss": 0.0474, "step": 1484 }, { "epoch": 0.2, "learning_rate": 8.015501804089269e-08, "loss": 0.0401, "step": 1485 }, { "epoch": 0.2, "learning_rate": 8.01416544166778e-08, "loss": 0.0493, "step": 1486 }, { "epoch": 0.2, "learning_rate": 8.012829079246291e-08, "loss": 0.0585, "step": 1487 }, { "epoch": 0.2, "learning_rate": 8.011492716824802e-08, "loss": 0.0843, "step": 1488 }, { "epoch": 0.2, "learning_rate": 8.010156354403313e-08, "loss": 0.054, "step": 1489 }, { "epoch": 0.2, "learning_rate": 8.008819991981825e-08, "loss": 0.059, "step": 1490 }, { "epoch": 0.2, "learning_rate": 8.007483629560336e-08, "loss": 0.0426, "step": 1491 }, { "epoch": 0.2, "learning_rate": 8.006147267138847e-08, "loss": 0.0445, "step": 1492 }, { "epoch": 0.2, "learning_rate": 8.00481090471736e-08, "loss": 0.0452, "step": 1493 }, { "epoch": 0.2, "learning_rate": 8.00347454229587e-08, "loss": 0.1081, "step": 1494 }, { "epoch": 0.2, "learning_rate": 8.002138179874382e-08, "loss": 0.0635, "step": 1495 }, { "epoch": 0.2, "learning_rate": 8.000801817452893e-08, "loss": 0.0438, "step": 1496 }, { "epoch": 0.2, "learning_rate": 7.999465455031405e-08, "loss": 0.0502, "step": 1497 }, { "epoch": 0.2, "learning_rate": 7.998129092609916e-08, "loss": 0.0346, "step": 1498 }, { "epoch": 0.2, "learning_rate": 7.996792730188427e-08, "loss": 0.0493, "step": 1499 }, { "epoch": 0.2, "learning_rate": 7.995456367766938e-08, "loss": 0.0593, "step": 1500 }, { "epoch": 0.2, "learning_rate": 7.994120005345449e-08, "loss": 0.0686, "step": 1501 }, { "epoch": 0.2, "learning_rate": 7.99278364292396e-08, "loss": 0.0449, "step": 1502 }, { "epoch": 0.2, "learning_rate": 7.991447280502472e-08, "loss": 0.0497, "step": 1503 }, { "epoch": 0.2, "learning_rate": 7.990110918080983e-08, "loss": 0.0689, "step": 1504 }, { "epoch": 0.2, "learning_rate": 7.988774555659494e-08, "loss": 0.0569, "step": 1505 }, { "epoch": 0.2, "learning_rate": 7.987438193238005e-08, "loss": 0.0398, "step": 1506 }, { "epoch": 0.2, "learning_rate": 7.986101830816517e-08, "loss": 0.06, "step": 1507 }, { "epoch": 0.2, "learning_rate": 7.984765468395028e-08, "loss": 0.0316, "step": 1508 }, { "epoch": 0.2, "learning_rate": 7.983429105973539e-08, "loss": 0.0347, "step": 1509 }, { "epoch": 0.2, "learning_rate": 7.982092743552052e-08, "loss": 0.0525, "step": 1510 }, { "epoch": 0.2, "learning_rate": 7.980756381130563e-08, "loss": 0.0369, "step": 1511 }, { "epoch": 0.2, "learning_rate": 7.979420018709074e-08, "loss": 0.0436, "step": 1512 }, { "epoch": 0.2, "learning_rate": 7.978083656287586e-08, "loss": 0.038, "step": 1513 }, { "epoch": 0.2, "learning_rate": 7.976747293866097e-08, "loss": 0.0671, "step": 1514 }, { "epoch": 0.2, "learning_rate": 7.975410931444608e-08, "loss": 0.051, "step": 1515 }, { "epoch": 0.2, "learning_rate": 7.974074569023119e-08, "loss": 0.0545, "step": 1516 }, { "epoch": 0.2, "learning_rate": 7.97273820660163e-08, "loss": 0.0398, "step": 1517 }, { "epoch": 0.2, "learning_rate": 7.971401844180141e-08, "loss": 0.0759, "step": 1518 }, { "epoch": 0.2, "learning_rate": 7.970065481758652e-08, "loss": 0.0506, "step": 1519 }, { "epoch": 0.2, "learning_rate": 7.968729119337164e-08, "loss": 0.0424, "step": 1520 }, { "epoch": 0.2, "learning_rate": 7.967392756915675e-08, "loss": 0.0553, "step": 1521 }, { "epoch": 0.2, "learning_rate": 7.966056394494186e-08, "loss": 0.0572, "step": 1522 }, { "epoch": 0.2, "learning_rate": 7.964720032072698e-08, "loss": 0.0371, "step": 1523 }, { "epoch": 0.2, "learning_rate": 7.963383669651209e-08, "loss": 0.0603, "step": 1524 }, { "epoch": 0.2, "learning_rate": 7.96204730722972e-08, "loss": 0.0815, "step": 1525 }, { "epoch": 0.2, "learning_rate": 7.960710944808233e-08, "loss": 0.0495, "step": 1526 }, { "epoch": 0.2, "learning_rate": 7.959374582386744e-08, "loss": 0.0593, "step": 1527 }, { "epoch": 0.2, "learning_rate": 7.958038219965255e-08, "loss": 0.0691, "step": 1528 }, { "epoch": 0.2, "learning_rate": 7.956701857543766e-08, "loss": 0.0288, "step": 1529 }, { "epoch": 0.2, "learning_rate": 7.955365495122277e-08, "loss": 0.0607, "step": 1530 }, { "epoch": 0.2, "learning_rate": 7.954029132700788e-08, "loss": 0.0454, "step": 1531 }, { "epoch": 0.2, "learning_rate": 7.952692770279299e-08, "loss": 0.054, "step": 1532 }, { "epoch": 0.2, "learning_rate": 7.951356407857811e-08, "loss": 0.0373, "step": 1533 }, { "epoch": 0.2, "learning_rate": 7.950020045436322e-08, "loss": 0.0556, "step": 1534 }, { "epoch": 0.21, "learning_rate": 7.948683683014833e-08, "loss": 0.0636, "step": 1535 }, { "epoch": 0.21, "learning_rate": 7.947347320593345e-08, "loss": 0.0671, "step": 1536 }, { "epoch": 0.21, "learning_rate": 7.946010958171856e-08, "loss": 0.0412, "step": 1537 }, { "epoch": 0.21, "learning_rate": 7.944674595750367e-08, "loss": 0.0435, "step": 1538 }, { "epoch": 0.21, "learning_rate": 7.94333823332888e-08, "loss": 0.0498, "step": 1539 }, { "epoch": 0.21, "learning_rate": 7.94200187090739e-08, "loss": 0.0538, "step": 1540 }, { "epoch": 0.21, "learning_rate": 7.940665508485901e-08, "loss": 0.0648, "step": 1541 }, { "epoch": 0.21, "learning_rate": 7.939329146064412e-08, "loss": 0.0591, "step": 1542 }, { "epoch": 0.21, "learning_rate": 7.937992783642923e-08, "loss": 0.0374, "step": 1543 }, { "epoch": 0.21, "learning_rate": 7.936656421221434e-08, "loss": 0.0425, "step": 1544 }, { "epoch": 0.21, "learning_rate": 7.935320058799945e-08, "loss": 0.0408, "step": 1545 }, { "epoch": 0.21, "learning_rate": 7.933983696378458e-08, "loss": 0.0703, "step": 1546 }, { "epoch": 0.21, "learning_rate": 7.932647333956969e-08, "loss": 0.0458, "step": 1547 }, { "epoch": 0.21, "learning_rate": 7.93131097153548e-08, "loss": 0.0454, "step": 1548 }, { "epoch": 0.21, "learning_rate": 7.929974609113992e-08, "loss": 0.0745, "step": 1549 }, { "epoch": 0.21, "learning_rate": 7.928638246692503e-08, "loss": 0.0533, "step": 1550 }, { "epoch": 0.21, "learning_rate": 7.927301884271014e-08, "loss": 0.0539, "step": 1551 }, { "epoch": 0.21, "learning_rate": 7.925965521849525e-08, "loss": 0.0366, "step": 1552 }, { "epoch": 0.21, "learning_rate": 7.924629159428037e-08, "loss": 0.0544, "step": 1553 }, { "epoch": 0.21, "learning_rate": 7.923292797006548e-08, "loss": 0.0367, "step": 1554 }, { "epoch": 0.21, "learning_rate": 7.921956434585059e-08, "loss": 0.0542, "step": 1555 }, { "epoch": 0.21, "learning_rate": 7.920620072163571e-08, "loss": 0.0478, "step": 1556 }, { "epoch": 0.21, "learning_rate": 7.919283709742082e-08, "loss": 0.0393, "step": 1557 }, { "epoch": 0.21, "learning_rate": 7.917947347320592e-08, "loss": 0.0555, "step": 1558 }, { "epoch": 0.21, "learning_rate": 7.916610984899104e-08, "loss": 0.0321, "step": 1559 }, { "epoch": 0.21, "learning_rate": 7.915274622477615e-08, "loss": 0.0339, "step": 1560 }, { "epoch": 0.21, "learning_rate": 7.913938260056126e-08, "loss": 0.0592, "step": 1561 }, { "epoch": 0.21, "learning_rate": 7.912601897634639e-08, "loss": 0.0567, "step": 1562 }, { "epoch": 0.21, "learning_rate": 7.91126553521315e-08, "loss": 0.0368, "step": 1563 }, { "epoch": 0.21, "learning_rate": 7.90992917279166e-08, "loss": 0.0552, "step": 1564 }, { "epoch": 0.21, "learning_rate": 7.908592810370172e-08, "loss": 0.0406, "step": 1565 }, { "epoch": 0.21, "learning_rate": 7.907256447948684e-08, "loss": 0.0695, "step": 1566 }, { "epoch": 0.21, "learning_rate": 7.905920085527195e-08, "loss": 0.0655, "step": 1567 }, { "epoch": 0.21, "learning_rate": 7.904583723105706e-08, "loss": 0.0622, "step": 1568 }, { "epoch": 0.21, "learning_rate": 7.903247360684218e-08, "loss": 0.0488, "step": 1569 }, { "epoch": 0.21, "learning_rate": 7.901910998262729e-08, "loss": 0.0419, "step": 1570 }, { "epoch": 0.21, "learning_rate": 7.90057463584124e-08, "loss": 0.0485, "step": 1571 }, { "epoch": 0.21, "learning_rate": 7.899238273419751e-08, "loss": 0.1277, "step": 1572 }, { "epoch": 0.21, "learning_rate": 7.897901910998262e-08, "loss": 0.0698, "step": 1573 }, { "epoch": 0.21, "learning_rate": 7.896565548576773e-08, "loss": 0.0952, "step": 1574 }, { "epoch": 0.21, "learning_rate": 7.895229186155284e-08, "loss": 0.0686, "step": 1575 }, { "epoch": 0.21, "learning_rate": 7.893892823733796e-08, "loss": 0.0639, "step": 1576 }, { "epoch": 0.21, "learning_rate": 7.892556461312307e-08, "loss": 0.0548, "step": 1577 }, { "epoch": 0.21, "learning_rate": 7.891220098890818e-08, "loss": 0.0351, "step": 1578 }, { "epoch": 0.21, "learning_rate": 7.88988373646933e-08, "loss": 0.0446, "step": 1579 }, { "epoch": 0.21, "learning_rate": 7.888547374047842e-08, "loss": 0.0464, "step": 1580 }, { "epoch": 0.21, "learning_rate": 7.887211011626353e-08, "loss": 0.0543, "step": 1581 }, { "epoch": 0.21, "learning_rate": 7.885874649204865e-08, "loss": 0.0507, "step": 1582 }, { "epoch": 0.21, "learning_rate": 7.884538286783376e-08, "loss": 0.0499, "step": 1583 }, { "epoch": 0.21, "learning_rate": 7.883201924361887e-08, "loss": 0.0473, "step": 1584 }, { "epoch": 0.21, "learning_rate": 7.881865561940398e-08, "loss": 0.0457, "step": 1585 }, { "epoch": 0.21, "learning_rate": 7.880529199518909e-08, "loss": 0.0955, "step": 1586 }, { "epoch": 0.21, "learning_rate": 7.87919283709742e-08, "loss": 0.068, "step": 1587 }, { "epoch": 0.21, "learning_rate": 7.877856474675931e-08, "loss": 0.0318, "step": 1588 }, { "epoch": 0.21, "learning_rate": 7.876520112254443e-08, "loss": 0.0526, "step": 1589 }, { "epoch": 0.21, "learning_rate": 7.875183749832954e-08, "loss": 0.0457, "step": 1590 }, { "epoch": 0.21, "learning_rate": 7.873847387411465e-08, "loss": 0.0523, "step": 1591 }, { "epoch": 0.21, "learning_rate": 7.872511024989977e-08, "loss": 0.0557, "step": 1592 }, { "epoch": 0.21, "learning_rate": 7.871174662568488e-08, "loss": 0.0488, "step": 1593 }, { "epoch": 0.21, "learning_rate": 7.869838300146999e-08, "loss": 0.0553, "step": 1594 }, { "epoch": 0.21, "learning_rate": 7.868501937725512e-08, "loss": 0.0799, "step": 1595 }, { "epoch": 0.21, "learning_rate": 7.867165575304023e-08, "loss": 0.0541, "step": 1596 }, { "epoch": 0.21, "learning_rate": 7.865829212882534e-08, "loss": 0.0508, "step": 1597 }, { "epoch": 0.21, "learning_rate": 7.864492850461045e-08, "loss": 0.0484, "step": 1598 }, { "epoch": 0.21, "learning_rate": 7.863156488039556e-08, "loss": 0.0484, "step": 1599 }, { "epoch": 0.21, "learning_rate": 7.861820125618067e-08, "loss": 0.0499, "step": 1600 }, { "epoch": 0.21, "learning_rate": 7.860483763196578e-08, "loss": 0.0648, "step": 1601 }, { "epoch": 0.21, "learning_rate": 7.85914740077509e-08, "loss": 0.0329, "step": 1602 }, { "epoch": 0.21, "learning_rate": 7.857811038353601e-08, "loss": 0.0434, "step": 1603 }, { "epoch": 0.21, "learning_rate": 7.856474675932112e-08, "loss": 0.1754, "step": 1604 }, { "epoch": 0.21, "learning_rate": 7.855138313510624e-08, "loss": 0.0415, "step": 1605 }, { "epoch": 0.21, "learning_rate": 7.853801951089135e-08, "loss": 0.0595, "step": 1606 }, { "epoch": 0.21, "learning_rate": 7.852465588667646e-08, "loss": 0.0527, "step": 1607 }, { "epoch": 0.21, "learning_rate": 7.851129226246158e-08, "loss": 0.0576, "step": 1608 }, { "epoch": 0.22, "learning_rate": 7.84979286382467e-08, "loss": 0.0375, "step": 1609 }, { "epoch": 0.22, "learning_rate": 7.84845650140318e-08, "loss": 0.0556, "step": 1610 }, { "epoch": 0.22, "learning_rate": 7.847120138981691e-08, "loss": 0.0607, "step": 1611 }, { "epoch": 0.22, "learning_rate": 7.845783776560204e-08, "loss": 0.0806, "step": 1612 }, { "epoch": 0.22, "learning_rate": 7.844447414138715e-08, "loss": 0.0238, "step": 1613 }, { "epoch": 0.22, "learning_rate": 7.843111051717226e-08, "loss": 0.0366, "step": 1614 }, { "epoch": 0.22, "learning_rate": 7.841774689295737e-08, "loss": 0.0663, "step": 1615 }, { "epoch": 0.22, "learning_rate": 7.840438326874248e-08, "loss": 0.0511, "step": 1616 }, { "epoch": 0.22, "learning_rate": 7.839101964452759e-08, "loss": 0.0529, "step": 1617 }, { "epoch": 0.22, "learning_rate": 7.837765602031271e-08, "loss": 0.0296, "step": 1618 }, { "epoch": 0.22, "learning_rate": 7.836429239609782e-08, "loss": 0.0481, "step": 1619 }, { "epoch": 0.22, "learning_rate": 7.835092877188293e-08, "loss": 0.057, "step": 1620 }, { "epoch": 0.22, "learning_rate": 7.833756514766804e-08, "loss": 0.0449, "step": 1621 }, { "epoch": 0.22, "learning_rate": 7.832420152345316e-08, "loss": 0.0435, "step": 1622 }, { "epoch": 0.22, "learning_rate": 7.831083789923827e-08, "loss": 0.0522, "step": 1623 }, { "epoch": 0.22, "learning_rate": 7.829747427502338e-08, "loss": 0.0478, "step": 1624 }, { "epoch": 0.22, "learning_rate": 7.82841106508085e-08, "loss": 0.0538, "step": 1625 }, { "epoch": 0.22, "learning_rate": 7.827074702659361e-08, "loss": 0.0408, "step": 1626 }, { "epoch": 0.22, "learning_rate": 7.825738340237872e-08, "loss": 0.0417, "step": 1627 }, { "epoch": 0.22, "learning_rate": 7.824401977816383e-08, "loss": 0.0371, "step": 1628 }, { "epoch": 0.22, "learning_rate": 7.823065615394894e-08, "loss": 0.0639, "step": 1629 }, { "epoch": 0.22, "learning_rate": 7.821729252973405e-08, "loss": 0.0426, "step": 1630 }, { "epoch": 0.22, "learning_rate": 7.820392890551918e-08, "loss": 0.0442, "step": 1631 }, { "epoch": 0.22, "learning_rate": 7.819056528130429e-08, "loss": 0.0574, "step": 1632 }, { "epoch": 0.22, "learning_rate": 7.81772016570894e-08, "loss": 0.0899, "step": 1633 }, { "epoch": 0.22, "learning_rate": 7.81638380328745e-08, "loss": 0.0575, "step": 1634 }, { "epoch": 0.22, "learning_rate": 7.815047440865963e-08, "loss": 0.0583, "step": 1635 }, { "epoch": 0.22, "learning_rate": 7.813711078444474e-08, "loss": 0.0422, "step": 1636 }, { "epoch": 0.22, "learning_rate": 7.812374716022985e-08, "loss": 0.0495, "step": 1637 }, { "epoch": 0.22, "learning_rate": 7.811038353601497e-08, "loss": 0.0525, "step": 1638 }, { "epoch": 0.22, "learning_rate": 7.809701991180008e-08, "loss": 0.0478, "step": 1639 }, { "epoch": 0.22, "learning_rate": 7.808365628758519e-08, "loss": 0.0403, "step": 1640 }, { "epoch": 0.22, "learning_rate": 7.80702926633703e-08, "loss": 0.0502, "step": 1641 }, { "epoch": 0.22, "learning_rate": 7.805692903915541e-08, "loss": 0.0514, "step": 1642 }, { "epoch": 0.22, "learning_rate": 7.804356541494052e-08, "loss": 0.0436, "step": 1643 }, { "epoch": 0.22, "learning_rate": 7.803020179072563e-08, "loss": 0.0654, "step": 1644 }, { "epoch": 0.22, "learning_rate": 7.801683816651075e-08, "loss": 0.0602, "step": 1645 }, { "epoch": 0.22, "learning_rate": 7.800347454229586e-08, "loss": 0.0433, "step": 1646 }, { "epoch": 0.22, "learning_rate": 7.799011091808097e-08, "loss": 0.0368, "step": 1647 }, { "epoch": 0.22, "learning_rate": 7.79767472938661e-08, "loss": 0.0496, "step": 1648 }, { "epoch": 0.22, "learning_rate": 7.79633836696512e-08, "loss": 0.0595, "step": 1649 }, { "epoch": 0.22, "learning_rate": 7.795002004543632e-08, "loss": 0.0822, "step": 1650 }, { "epoch": 0.22, "learning_rate": 7.793665642122144e-08, "loss": 0.0437, "step": 1651 }, { "epoch": 0.22, "learning_rate": 7.792329279700655e-08, "loss": 0.0513, "step": 1652 }, { "epoch": 0.22, "learning_rate": 7.790992917279166e-08, "loss": 0.0689, "step": 1653 }, { "epoch": 0.22, "learning_rate": 7.789656554857678e-08, "loss": 0.0497, "step": 1654 }, { "epoch": 0.22, "learning_rate": 7.788320192436189e-08, "loss": 0.0474, "step": 1655 }, { "epoch": 0.22, "learning_rate": 7.7869838300147e-08, "loss": 0.0432, "step": 1656 }, { "epoch": 0.22, "learning_rate": 7.78564746759321e-08, "loss": 0.0428, "step": 1657 }, { "epoch": 0.22, "learning_rate": 7.784311105171722e-08, "loss": 0.045, "step": 1658 }, { "epoch": 0.22, "learning_rate": 7.782974742750233e-08, "loss": 0.0518, "step": 1659 }, { "epoch": 0.22, "learning_rate": 7.781638380328744e-08, "loss": 0.0616, "step": 1660 }, { "epoch": 0.22, "learning_rate": 7.780302017907256e-08, "loss": 0.0407, "step": 1661 }, { "epoch": 0.22, "learning_rate": 7.778965655485767e-08, "loss": 0.0346, "step": 1662 }, { "epoch": 0.22, "learning_rate": 7.777629293064278e-08, "loss": 0.0494, "step": 1663 }, { "epoch": 0.22, "learning_rate": 7.77629293064279e-08, "loss": 0.0567, "step": 1664 }, { "epoch": 0.22, "learning_rate": 7.774956568221302e-08, "loss": 0.0348, "step": 1665 }, { "epoch": 0.22, "learning_rate": 7.773620205799813e-08, "loss": 0.0623, "step": 1666 }, { "epoch": 0.22, "learning_rate": 7.772283843378324e-08, "loss": 0.0388, "step": 1667 }, { "epoch": 0.22, "learning_rate": 7.770947480956836e-08, "loss": 0.0495, "step": 1668 }, { "epoch": 0.22, "learning_rate": 7.769611118535347e-08, "loss": 0.03, "step": 1669 }, { "epoch": 0.22, "learning_rate": 7.768274756113858e-08, "loss": 0.0614, "step": 1670 }, { "epoch": 0.22, "learning_rate": 7.766938393692369e-08, "loss": 0.0779, "step": 1671 }, { "epoch": 0.22, "learning_rate": 7.76560203127088e-08, "loss": 0.0384, "step": 1672 }, { "epoch": 0.22, "learning_rate": 7.764265668849391e-08, "loss": 0.0624, "step": 1673 }, { "epoch": 0.22, "learning_rate": 7.762929306427903e-08, "loss": 0.0509, "step": 1674 }, { "epoch": 0.22, "learning_rate": 7.761592944006414e-08, "loss": 0.0482, "step": 1675 }, { "epoch": 0.22, "learning_rate": 7.760256581584925e-08, "loss": 0.048, "step": 1676 }, { "epoch": 0.22, "learning_rate": 7.758920219163437e-08, "loss": 0.0596, "step": 1677 }, { "epoch": 0.22, "learning_rate": 7.757583856741948e-08, "loss": 0.0451, "step": 1678 }, { "epoch": 0.22, "learning_rate": 7.75624749432046e-08, "loss": 0.0322, "step": 1679 }, { "epoch": 0.22, "learning_rate": 7.75491113189897e-08, "loss": 0.0525, "step": 1680 }, { "epoch": 0.22, "learning_rate": 7.753574769477483e-08, "loss": 0.0541, "step": 1681 }, { "epoch": 0.22, "learning_rate": 7.752238407055994e-08, "loss": 0.0502, "step": 1682 }, { "epoch": 0.22, "learning_rate": 7.750902044634505e-08, "loss": 0.0607, "step": 1683 }, { "epoch": 0.23, "learning_rate": 7.749565682213016e-08, "loss": 0.0488, "step": 1684 }, { "epoch": 0.23, "learning_rate": 7.748229319791527e-08, "loss": 0.0483, "step": 1685 }, { "epoch": 0.23, "learning_rate": 7.746892957370038e-08, "loss": 0.0588, "step": 1686 }, { "epoch": 0.23, "learning_rate": 7.74555659494855e-08, "loss": 0.0339, "step": 1687 }, { "epoch": 0.23, "learning_rate": 7.744220232527061e-08, "loss": 0.0798, "step": 1688 }, { "epoch": 0.23, "learning_rate": 7.742883870105572e-08, "loss": 0.0607, "step": 1689 }, { "epoch": 0.23, "learning_rate": 7.741547507684083e-08, "loss": 0.044, "step": 1690 }, { "epoch": 0.23, "learning_rate": 7.740211145262595e-08, "loss": 0.0584, "step": 1691 }, { "epoch": 0.23, "learning_rate": 7.738874782841106e-08, "loss": 0.0396, "step": 1692 }, { "epoch": 0.23, "learning_rate": 7.737538420419617e-08, "loss": 0.05, "step": 1693 }, { "epoch": 0.23, "learning_rate": 7.73620205799813e-08, "loss": 0.0265, "step": 1694 }, { "epoch": 0.23, "learning_rate": 7.73486569557664e-08, "loss": 0.0459, "step": 1695 }, { "epoch": 0.23, "learning_rate": 7.733529333155151e-08, "loss": 0.0316, "step": 1696 }, { "epoch": 0.23, "learning_rate": 7.732192970733664e-08, "loss": 0.0772, "step": 1697 }, { "epoch": 0.23, "learning_rate": 7.730856608312173e-08, "loss": 0.056, "step": 1698 }, { "epoch": 0.23, "learning_rate": 7.729520245890684e-08, "loss": 0.0453, "step": 1699 }, { "epoch": 0.23, "learning_rate": 7.728183883469197e-08, "loss": 0.0548, "step": 1700 }, { "epoch": 0.23, "learning_rate": 7.726847521047708e-08, "loss": 0.0404, "step": 1701 }, { "epoch": 0.23, "learning_rate": 7.725511158626219e-08, "loss": 0.0444, "step": 1702 }, { "epoch": 0.23, "learning_rate": 7.72417479620473e-08, "loss": 0.0744, "step": 1703 }, { "epoch": 0.23, "learning_rate": 7.722838433783242e-08, "loss": 0.0492, "step": 1704 }, { "epoch": 0.23, "learning_rate": 7.721502071361753e-08, "loss": 0.0521, "step": 1705 }, { "epoch": 0.23, "learning_rate": 7.720165708940264e-08, "loss": 0.0704, "step": 1706 }, { "epoch": 0.23, "learning_rate": 7.718829346518776e-08, "loss": 0.0511, "step": 1707 }, { "epoch": 0.23, "learning_rate": 7.717492984097287e-08, "loss": 0.0459, "step": 1708 }, { "epoch": 0.23, "learning_rate": 7.716156621675798e-08, "loss": 0.0683, "step": 1709 }, { "epoch": 0.23, "learning_rate": 7.71482025925431e-08, "loss": 0.0473, "step": 1710 }, { "epoch": 0.23, "learning_rate": 7.713483896832821e-08, "loss": 0.0471, "step": 1711 }, { "epoch": 0.23, "learning_rate": 7.712147534411332e-08, "loss": 0.0498, "step": 1712 }, { "epoch": 0.23, "learning_rate": 7.710811171989843e-08, "loss": 0.039, "step": 1713 }, { "epoch": 0.23, "learning_rate": 7.709474809568354e-08, "loss": 0.0752, "step": 1714 }, { "epoch": 0.23, "learning_rate": 7.708138447146865e-08, "loss": 0.0653, "step": 1715 }, { "epoch": 0.23, "learning_rate": 7.706802084725376e-08, "loss": 0.0449, "step": 1716 }, { "epoch": 0.23, "learning_rate": 7.705465722303889e-08, "loss": 0.0516, "step": 1717 }, { "epoch": 0.23, "learning_rate": 7.7041293598824e-08, "loss": 0.0362, "step": 1718 }, { "epoch": 0.23, "learning_rate": 7.70279299746091e-08, "loss": 0.0455, "step": 1719 }, { "epoch": 0.23, "learning_rate": 7.701456635039423e-08, "loss": 0.049, "step": 1720 }, { "epoch": 0.23, "learning_rate": 7.700120272617934e-08, "loss": 0.0418, "step": 1721 }, { "epoch": 0.23, "learning_rate": 7.698783910196445e-08, "loss": 0.0639, "step": 1722 }, { "epoch": 0.23, "learning_rate": 7.697447547774957e-08, "loss": 0.048, "step": 1723 }, { "epoch": 0.23, "learning_rate": 7.696111185353468e-08, "loss": 0.0581, "step": 1724 }, { "epoch": 0.23, "learning_rate": 7.694774822931979e-08, "loss": 0.035, "step": 1725 }, { "epoch": 0.23, "learning_rate": 7.69343846051049e-08, "loss": 0.1265, "step": 1726 }, { "epoch": 0.23, "learning_rate": 7.692102098089001e-08, "loss": 0.0586, "step": 1727 }, { "epoch": 0.23, "learning_rate": 7.690765735667512e-08, "loss": 0.0558, "step": 1728 }, { "epoch": 0.23, "learning_rate": 7.689429373246023e-08, "loss": 0.0502, "step": 1729 }, { "epoch": 0.23, "learning_rate": 7.688093010824535e-08, "loss": 0.0701, "step": 1730 }, { "epoch": 0.23, "learning_rate": 7.686756648403046e-08, "loss": 0.0438, "step": 1731 }, { "epoch": 0.23, "learning_rate": 7.685420285981557e-08, "loss": 0.0575, "step": 1732 }, { "epoch": 0.23, "learning_rate": 7.68408392356007e-08, "loss": 0.0664, "step": 1733 }, { "epoch": 0.23, "learning_rate": 7.682747561138581e-08, "loss": 0.0452, "step": 1734 }, { "epoch": 0.23, "learning_rate": 7.681411198717092e-08, "loss": 0.0539, "step": 1735 }, { "epoch": 0.23, "learning_rate": 7.680074836295603e-08, "loss": 0.0339, "step": 1736 }, { "epoch": 0.23, "learning_rate": 7.678738473874115e-08, "loss": 0.0504, "step": 1737 }, { "epoch": 0.23, "learning_rate": 7.677402111452626e-08, "loss": 0.0357, "step": 1738 }, { "epoch": 0.23, "learning_rate": 7.676065749031137e-08, "loss": 0.0527, "step": 1739 }, { "epoch": 0.23, "learning_rate": 7.674729386609648e-08, "loss": 0.0539, "step": 1740 }, { "epoch": 0.23, "learning_rate": 7.673393024188159e-08, "loss": 0.0639, "step": 1741 }, { "epoch": 0.23, "learning_rate": 7.67205666176667e-08, "loss": 0.0552, "step": 1742 }, { "epoch": 0.23, "learning_rate": 7.670720299345182e-08, "loss": 0.0533, "step": 1743 }, { "epoch": 0.23, "learning_rate": 7.669383936923693e-08, "loss": 0.0554, "step": 1744 }, { "epoch": 0.23, "learning_rate": 7.668047574502204e-08, "loss": 0.0332, "step": 1745 }, { "epoch": 0.23, "learning_rate": 7.666711212080716e-08, "loss": 0.0534, "step": 1746 }, { "epoch": 0.23, "learning_rate": 7.665374849659227e-08, "loss": 0.0601, "step": 1747 }, { "epoch": 0.23, "learning_rate": 7.664038487237738e-08, "loss": 0.0326, "step": 1748 }, { "epoch": 0.23, "learning_rate": 7.66270212481625e-08, "loss": 0.0516, "step": 1749 }, { "epoch": 0.23, "learning_rate": 7.661365762394762e-08, "loss": 0.0326, "step": 1750 }, { "epoch": 0.23, "learning_rate": 7.660029399973273e-08, "loss": 0.0529, "step": 1751 }, { "epoch": 0.23, "learning_rate": 7.658693037551784e-08, "loss": 0.0542, "step": 1752 }, { "epoch": 0.23, "learning_rate": 7.657356675130296e-08, "loss": 0.0392, "step": 1753 }, { "epoch": 0.23, "learning_rate": 7.656020312708807e-08, "loss": 0.0464, "step": 1754 }, { "epoch": 0.23, "learning_rate": 7.654683950287318e-08, "loss": 0.0478, "step": 1755 }, { "epoch": 0.23, "learning_rate": 7.653347587865829e-08, "loss": 0.0424, "step": 1756 }, { "epoch": 0.23, "learning_rate": 7.65201122544434e-08, "loss": 0.0339, "step": 1757 }, { "epoch": 0.23, "learning_rate": 7.650674863022851e-08, "loss": 0.0446, "step": 1758 }, { "epoch": 0.24, "learning_rate": 7.649338500601362e-08, "loss": 0.0714, "step": 1759 }, { "epoch": 0.24, "learning_rate": 7.648002138179874e-08, "loss": 0.05, "step": 1760 }, { "epoch": 0.24, "learning_rate": 7.646665775758385e-08, "loss": 0.0591, "step": 1761 }, { "epoch": 0.24, "learning_rate": 7.645329413336896e-08, "loss": 0.0471, "step": 1762 }, { "epoch": 0.24, "learning_rate": 7.643993050915408e-08, "loss": 0.0514, "step": 1763 }, { "epoch": 0.24, "learning_rate": 7.64265668849392e-08, "loss": 0.0543, "step": 1764 }, { "epoch": 0.24, "learning_rate": 7.64132032607243e-08, "loss": 0.063, "step": 1765 }, { "epoch": 0.24, "learning_rate": 7.639983963650943e-08, "loss": 0.0498, "step": 1766 }, { "epoch": 0.24, "learning_rate": 7.638647601229454e-08, "loss": 0.0369, "step": 1767 }, { "epoch": 0.24, "learning_rate": 7.637311238807965e-08, "loss": 0.0267, "step": 1768 }, { "epoch": 0.24, "learning_rate": 7.635974876386476e-08, "loss": 0.0643, "step": 1769 }, { "epoch": 0.24, "learning_rate": 7.634638513964987e-08, "loss": 0.049, "step": 1770 }, { "epoch": 0.24, "learning_rate": 7.633302151543498e-08, "loss": 0.0348, "step": 1771 }, { "epoch": 0.24, "learning_rate": 7.631965789122009e-08, "loss": 0.0486, "step": 1772 }, { "epoch": 0.24, "learning_rate": 7.630629426700521e-08, "loss": 0.0815, "step": 1773 }, { "epoch": 0.24, "learning_rate": 7.629293064279032e-08, "loss": 0.0314, "step": 1774 }, { "epoch": 0.24, "learning_rate": 7.627956701857543e-08, "loss": 0.0509, "step": 1775 }, { "epoch": 0.24, "learning_rate": 7.626620339436055e-08, "loss": 0.043, "step": 1776 }, { "epoch": 0.24, "learning_rate": 7.625283977014566e-08, "loss": 0.066, "step": 1777 }, { "epoch": 0.24, "learning_rate": 7.623947614593077e-08, "loss": 0.0566, "step": 1778 }, { "epoch": 0.24, "learning_rate": 7.62261125217159e-08, "loss": 0.0656, "step": 1779 }, { "epoch": 0.24, "learning_rate": 7.6212748897501e-08, "loss": 0.052, "step": 1780 }, { "epoch": 0.24, "learning_rate": 7.619938527328611e-08, "loss": 0.0527, "step": 1781 }, { "epoch": 0.24, "learning_rate": 7.618602164907122e-08, "loss": 0.0744, "step": 1782 }, { "epoch": 0.24, "learning_rate": 7.617265802485633e-08, "loss": 0.0301, "step": 1783 }, { "epoch": 0.24, "learning_rate": 7.615929440064144e-08, "loss": 0.0585, "step": 1784 }, { "epoch": 0.24, "learning_rate": 7.614593077642655e-08, "loss": 0.0624, "step": 1785 }, { "epoch": 0.24, "learning_rate": 7.613256715221168e-08, "loss": 0.063, "step": 1786 }, { "epoch": 0.24, "learning_rate": 7.611920352799679e-08, "loss": 0.0436, "step": 1787 }, { "epoch": 0.24, "learning_rate": 7.61058399037819e-08, "loss": 0.0398, "step": 1788 }, { "epoch": 0.24, "learning_rate": 7.609247627956702e-08, "loss": 0.0501, "step": 1789 }, { "epoch": 0.24, "learning_rate": 7.607911265535213e-08, "loss": 0.0528, "step": 1790 }, { "epoch": 0.24, "learning_rate": 7.606574903113724e-08, "loss": 0.0641, "step": 1791 }, { "epoch": 0.24, "learning_rate": 7.605238540692236e-08, "loss": 0.044, "step": 1792 }, { "epoch": 0.24, "learning_rate": 7.603902178270747e-08, "loss": 0.0475, "step": 1793 }, { "epoch": 0.24, "learning_rate": 7.602565815849258e-08, "loss": 0.0577, "step": 1794 }, { "epoch": 0.24, "learning_rate": 7.601229453427769e-08, "loss": 0.038, "step": 1795 }, { "epoch": 0.24, "learning_rate": 7.599893091006281e-08, "loss": 0.0336, "step": 1796 }, { "epoch": 0.24, "learning_rate": 7.598556728584791e-08, "loss": 0.0496, "step": 1797 }, { "epoch": 0.24, "learning_rate": 7.597220366163302e-08, "loss": 0.0519, "step": 1798 }, { "epoch": 0.24, "learning_rate": 7.595884003741814e-08, "loss": 0.0441, "step": 1799 }, { "epoch": 0.24, "learning_rate": 7.594547641320325e-08, "loss": 0.0792, "step": 1800 }, { "epoch": 0.24, "learning_rate": 7.593211278898836e-08, "loss": 0.054, "step": 1801 }, { "epoch": 0.24, "learning_rate": 7.591874916477349e-08, "loss": 0.0419, "step": 1802 }, { "epoch": 0.24, "learning_rate": 7.59053855405586e-08, "loss": 0.0553, "step": 1803 }, { "epoch": 0.24, "learning_rate": 7.589202191634371e-08, "loss": 0.0602, "step": 1804 }, { "epoch": 0.24, "learning_rate": 7.587865829212882e-08, "loss": 0.055, "step": 1805 }, { "epoch": 0.24, "learning_rate": 7.586529466791394e-08, "loss": 0.055, "step": 1806 }, { "epoch": 0.24, "learning_rate": 7.585193104369905e-08, "loss": 0.0294, "step": 1807 }, { "epoch": 0.24, "learning_rate": 7.583856741948416e-08, "loss": 0.0445, "step": 1808 }, { "epoch": 0.24, "learning_rate": 7.582520379526928e-08, "loss": 0.032, "step": 1809 }, { "epoch": 0.24, "learning_rate": 7.581184017105439e-08, "loss": 0.0393, "step": 1810 }, { "epoch": 0.24, "learning_rate": 7.57984765468395e-08, "loss": 0.0594, "step": 1811 }, { "epoch": 0.24, "learning_rate": 7.578511292262461e-08, "loss": 0.0437, "step": 1812 }, { "epoch": 0.24, "learning_rate": 7.577174929840972e-08, "loss": 0.0626, "step": 1813 }, { "epoch": 0.24, "learning_rate": 7.575838567419483e-08, "loss": 0.0352, "step": 1814 }, { "epoch": 0.24, "learning_rate": 7.574502204997995e-08, "loss": 0.0602, "step": 1815 }, { "epoch": 0.24, "learning_rate": 7.573165842576506e-08, "loss": 0.0335, "step": 1816 }, { "epoch": 0.24, "learning_rate": 7.571829480155017e-08, "loss": 0.0565, "step": 1817 }, { "epoch": 0.24, "learning_rate": 7.570493117733528e-08, "loss": 0.0445, "step": 1818 }, { "epoch": 0.24, "learning_rate": 7.569156755312041e-08, "loss": 0.0666, "step": 1819 }, { "epoch": 0.24, "learning_rate": 7.567820392890552e-08, "loss": 0.0393, "step": 1820 }, { "epoch": 0.24, "learning_rate": 7.566484030469063e-08, "loss": 0.0755, "step": 1821 }, { "epoch": 0.24, "learning_rate": 7.565147668047575e-08, "loss": 0.0481, "step": 1822 }, { "epoch": 0.24, "learning_rate": 7.563811305626086e-08, "loss": 0.028, "step": 1823 }, { "epoch": 0.24, "learning_rate": 7.562474943204597e-08, "loss": 0.0346, "step": 1824 }, { "epoch": 0.24, "learning_rate": 7.561138580783108e-08, "loss": 0.0461, "step": 1825 }, { "epoch": 0.24, "learning_rate": 7.559802218361619e-08, "loss": 0.0478, "step": 1826 }, { "epoch": 0.24, "learning_rate": 7.55846585594013e-08, "loss": 0.0325, "step": 1827 }, { "epoch": 0.24, "learning_rate": 7.557129493518641e-08, "loss": 0.0432, "step": 1828 }, { "epoch": 0.24, "learning_rate": 7.555793131097153e-08, "loss": 0.0426, "step": 1829 }, { "epoch": 0.24, "learning_rate": 7.554456768675664e-08, "loss": 0.0439, "step": 1830 }, { "epoch": 0.24, "learning_rate": 7.553120406254175e-08, "loss": 0.0969, "step": 1831 }, { "epoch": 0.24, "learning_rate": 7.551784043832687e-08, "loss": 0.0558, "step": 1832 }, { "epoch": 0.24, "learning_rate": 7.550447681411198e-08, "loss": 0.0542, "step": 1833 }, { "epoch": 0.25, "learning_rate": 7.54911131898971e-08, "loss": 0.0613, "step": 1834 }, { "epoch": 0.25, "learning_rate": 7.547774956568222e-08, "loss": 0.1175, "step": 1835 }, { "epoch": 0.25, "learning_rate": 7.546438594146733e-08, "loss": 0.0498, "step": 1836 }, { "epoch": 0.25, "learning_rate": 7.545102231725244e-08, "loss": 0.0358, "step": 1837 }, { "epoch": 0.25, "learning_rate": 7.543765869303755e-08, "loss": 0.0433, "step": 1838 }, { "epoch": 0.25, "learning_rate": 7.542429506882266e-08, "loss": 0.1052, "step": 1839 }, { "epoch": 0.25, "learning_rate": 7.541093144460777e-08, "loss": 0.0596, "step": 1840 }, { "epoch": 0.25, "learning_rate": 7.539756782039288e-08, "loss": 0.0661, "step": 1841 }, { "epoch": 0.25, "learning_rate": 7.5384204196178e-08, "loss": 0.0488, "step": 1842 }, { "epoch": 0.25, "learning_rate": 7.537084057196311e-08, "loss": 0.0411, "step": 1843 }, { "epoch": 0.25, "learning_rate": 7.535747694774822e-08, "loss": 0.0399, "step": 1844 }, { "epoch": 0.25, "learning_rate": 7.534411332353334e-08, "loss": 0.0572, "step": 1845 }, { "epoch": 0.25, "learning_rate": 7.533074969931845e-08, "loss": 0.0566, "step": 1846 }, { "epoch": 0.25, "learning_rate": 7.531738607510356e-08, "loss": 0.0587, "step": 1847 }, { "epoch": 0.25, "learning_rate": 7.530402245088869e-08, "loss": 0.0379, "step": 1848 }, { "epoch": 0.25, "learning_rate": 7.52906588266738e-08, "loss": 0.0745, "step": 1849 }, { "epoch": 0.25, "learning_rate": 7.52772952024589e-08, "loss": 0.0492, "step": 1850 }, { "epoch": 0.25, "learning_rate": 7.526393157824401e-08, "loss": 0.0485, "step": 1851 }, { "epoch": 0.25, "learning_rate": 7.525056795402914e-08, "loss": 0.0584, "step": 1852 }, { "epoch": 0.25, "learning_rate": 7.523720432981425e-08, "loss": 0.0557, "step": 1853 }, { "epoch": 0.25, "learning_rate": 7.522384070559936e-08, "loss": 0.0694, "step": 1854 }, { "epoch": 0.25, "learning_rate": 7.521047708138447e-08, "loss": 0.0497, "step": 1855 }, { "epoch": 0.25, "learning_rate": 7.519711345716958e-08, "loss": 0.0474, "step": 1856 }, { "epoch": 0.25, "learning_rate": 7.518374983295469e-08, "loss": 0.0651, "step": 1857 }, { "epoch": 0.25, "learning_rate": 7.517038620873981e-08, "loss": 0.0692, "step": 1858 }, { "epoch": 0.25, "learning_rate": 7.515702258452492e-08, "loss": 0.046, "step": 1859 }, { "epoch": 0.25, "learning_rate": 7.514365896031003e-08, "loss": 0.0618, "step": 1860 }, { "epoch": 0.25, "learning_rate": 7.513029533609515e-08, "loss": 0.0479, "step": 1861 }, { "epoch": 0.25, "learning_rate": 7.511693171188026e-08, "loss": 0.0319, "step": 1862 }, { "epoch": 0.25, "learning_rate": 7.510356808766537e-08, "loss": 0.0692, "step": 1863 }, { "epoch": 0.25, "learning_rate": 7.509020446345048e-08, "loss": 0.0534, "step": 1864 }, { "epoch": 0.25, "learning_rate": 7.50768408392356e-08, "loss": 0.0457, "step": 1865 }, { "epoch": 0.25, "learning_rate": 7.506347721502071e-08, "loss": 0.0515, "step": 1866 }, { "epoch": 0.25, "learning_rate": 7.505011359080582e-08, "loss": 0.0435, "step": 1867 }, { "epoch": 0.25, "learning_rate": 7.503674996659093e-08, "loss": 0.0516, "step": 1868 }, { "epoch": 0.25, "learning_rate": 7.502338634237604e-08, "loss": 0.0384, "step": 1869 }, { "epoch": 0.25, "learning_rate": 7.501002271816115e-08, "loss": 0.0586, "step": 1870 }, { "epoch": 0.25, "learning_rate": 7.499665909394628e-08, "loss": 0.0513, "step": 1871 }, { "epoch": 0.25, "learning_rate": 7.498329546973139e-08, "loss": 0.0636, "step": 1872 }, { "epoch": 0.25, "learning_rate": 7.49699318455165e-08, "loss": 0.0591, "step": 1873 }, { "epoch": 0.25, "learning_rate": 7.495656822130161e-08, "loss": 0.0553, "step": 1874 }, { "epoch": 0.25, "learning_rate": 7.494320459708673e-08, "loss": 0.0562, "step": 1875 }, { "epoch": 0.25, "learning_rate": 7.492984097287184e-08, "loss": 0.0288, "step": 1876 }, { "epoch": 0.25, "learning_rate": 7.491647734865695e-08, "loss": 0.0398, "step": 1877 }, { "epoch": 0.25, "learning_rate": 7.490311372444207e-08, "loss": 0.0763, "step": 1878 }, { "epoch": 0.25, "learning_rate": 7.488975010022718e-08, "loss": 0.0394, "step": 1879 }, { "epoch": 0.25, "learning_rate": 7.487638647601229e-08, "loss": 0.0285, "step": 1880 }, { "epoch": 0.25, "learning_rate": 7.48630228517974e-08, "loss": 0.0523, "step": 1881 }, { "epoch": 0.25, "learning_rate": 7.484965922758251e-08, "loss": 0.0751, "step": 1882 }, { "epoch": 0.25, "learning_rate": 7.483629560336762e-08, "loss": 0.0416, "step": 1883 }, { "epoch": 0.25, "learning_rate": 7.482293197915274e-08, "loss": 0.0537, "step": 1884 }, { "epoch": 0.25, "learning_rate": 7.480956835493785e-08, "loss": 0.0569, "step": 1885 }, { "epoch": 0.25, "learning_rate": 7.479620473072296e-08, "loss": 0.0557, "step": 1886 }, { "epoch": 0.25, "learning_rate": 7.478284110650807e-08, "loss": 0.0508, "step": 1887 }, { "epoch": 0.25, "learning_rate": 7.47694774822932e-08, "loss": 0.0592, "step": 1888 }, { "epoch": 0.25, "learning_rate": 7.475611385807831e-08, "loss": 0.0383, "step": 1889 }, { "epoch": 0.25, "learning_rate": 7.474275023386342e-08, "loss": 0.0555, "step": 1890 }, { "epoch": 0.25, "learning_rate": 7.472938660964854e-08, "loss": 0.0391, "step": 1891 }, { "epoch": 0.25, "learning_rate": 7.471602298543365e-08, "loss": 0.0452, "step": 1892 }, { "epoch": 0.25, "learning_rate": 7.470265936121876e-08, "loss": 0.043, "step": 1893 }, { "epoch": 0.25, "learning_rate": 7.468929573700388e-08, "loss": 0.0618, "step": 1894 }, { "epoch": 0.25, "learning_rate": 7.467593211278899e-08, "loss": 0.0515, "step": 1895 }, { "epoch": 0.25, "learning_rate": 7.466256848857409e-08, "loss": 0.0943, "step": 1896 }, { "epoch": 0.25, "learning_rate": 7.46492048643592e-08, "loss": 0.0652, "step": 1897 }, { "epoch": 0.25, "learning_rate": 7.463584124014432e-08, "loss": 0.0556, "step": 1898 }, { "epoch": 0.25, "learning_rate": 7.462247761592943e-08, "loss": 0.0444, "step": 1899 }, { "epoch": 0.25, "learning_rate": 7.460911399171454e-08, "loss": 0.0519, "step": 1900 }, { "epoch": 0.25, "learning_rate": 7.459575036749967e-08, "loss": 0.0533, "step": 1901 }, { "epoch": 0.25, "learning_rate": 7.458238674328477e-08, "loss": 0.0556, "step": 1902 }, { "epoch": 0.25, "learning_rate": 7.456902311906988e-08, "loss": 0.0402, "step": 1903 }, { "epoch": 0.25, "learning_rate": 7.455565949485501e-08, "loss": 0.0544, "step": 1904 }, { "epoch": 0.25, "learning_rate": 7.454229587064012e-08, "loss": 0.0337, "step": 1905 }, { "epoch": 0.25, "learning_rate": 7.452893224642523e-08, "loss": 0.0363, "step": 1906 }, { "epoch": 0.25, "learning_rate": 7.451556862221035e-08, "loss": 0.0561, "step": 1907 }, { "epoch": 0.25, "learning_rate": 7.450220499799546e-08, "loss": 0.0205, "step": 1908 }, { "epoch": 0.26, "learning_rate": 7.448884137378057e-08, "loss": 0.0291, "step": 1909 }, { "epoch": 0.26, "learning_rate": 7.447547774956568e-08, "loss": 0.0328, "step": 1910 }, { "epoch": 0.26, "learning_rate": 7.446211412535079e-08, "loss": 0.0658, "step": 1911 }, { "epoch": 0.26, "learning_rate": 7.44487505011359e-08, "loss": 0.0541, "step": 1912 }, { "epoch": 0.26, "learning_rate": 7.443538687692101e-08, "loss": 0.05, "step": 1913 }, { "epoch": 0.26, "learning_rate": 7.442202325270613e-08, "loss": 0.0586, "step": 1914 }, { "epoch": 0.26, "learning_rate": 7.440865962849124e-08, "loss": 0.0513, "step": 1915 }, { "epoch": 0.26, "learning_rate": 7.439529600427635e-08, "loss": 0.0467, "step": 1916 }, { "epoch": 0.26, "learning_rate": 7.438193238006148e-08, "loss": 0.0375, "step": 1917 }, { "epoch": 0.26, "learning_rate": 7.436856875584659e-08, "loss": 0.0417, "step": 1918 }, { "epoch": 0.26, "learning_rate": 7.43552051316317e-08, "loss": 0.0368, "step": 1919 }, { "epoch": 0.26, "learning_rate": 7.43418415074168e-08, "loss": 0.0264, "step": 1920 }, { "epoch": 0.26, "learning_rate": 7.432847788320193e-08, "loss": 0.035, "step": 1921 }, { "epoch": 0.26, "learning_rate": 7.431511425898704e-08, "loss": 0.0359, "step": 1922 }, { "epoch": 0.26, "learning_rate": 7.430175063477215e-08, "loss": 0.0893, "step": 1923 }, { "epoch": 0.26, "learning_rate": 7.428838701055726e-08, "loss": 0.0565, "step": 1924 }, { "epoch": 0.26, "learning_rate": 7.427502338634237e-08, "loss": 0.0524, "step": 1925 }, { "epoch": 0.26, "learning_rate": 7.426165976212748e-08, "loss": 0.0428, "step": 1926 }, { "epoch": 0.26, "learning_rate": 7.42482961379126e-08, "loss": 0.0461, "step": 1927 }, { "epoch": 0.26, "learning_rate": 7.423493251369771e-08, "loss": 0.0429, "step": 1928 }, { "epoch": 0.26, "learning_rate": 7.422156888948282e-08, "loss": 0.0627, "step": 1929 }, { "epoch": 0.26, "learning_rate": 7.420820526526794e-08, "loss": 0.0421, "step": 1930 }, { "epoch": 0.26, "learning_rate": 7.419484164105305e-08, "loss": 0.0385, "step": 1931 }, { "epoch": 0.26, "learning_rate": 7.418147801683816e-08, "loss": 0.0394, "step": 1932 }, { "epoch": 0.26, "learning_rate": 7.416811439262327e-08, "loss": 0.0479, "step": 1933 }, { "epoch": 0.26, "learning_rate": 7.41547507684084e-08, "loss": 0.0574, "step": 1934 }, { "epoch": 0.26, "learning_rate": 7.41413871441935e-08, "loss": 0.0466, "step": 1935 }, { "epoch": 0.26, "learning_rate": 7.412802351997862e-08, "loss": 0.0606, "step": 1936 }, { "epoch": 0.26, "learning_rate": 7.411465989576372e-08, "loss": 0.0761, "step": 1937 }, { "epoch": 0.26, "learning_rate": 7.410129627154883e-08, "loss": 0.0298, "step": 1938 }, { "epoch": 0.26, "learning_rate": 7.408793264733394e-08, "loss": 0.0494, "step": 1939 }, { "epoch": 0.26, "learning_rate": 7.407456902311907e-08, "loss": 0.0518, "step": 1940 }, { "epoch": 0.26, "learning_rate": 7.406120539890418e-08, "loss": 0.0537, "step": 1941 }, { "epoch": 0.26, "learning_rate": 7.404784177468929e-08, "loss": 0.0469, "step": 1942 }, { "epoch": 0.26, "learning_rate": 7.40344781504744e-08, "loss": 0.0315, "step": 1943 }, { "epoch": 0.26, "learning_rate": 7.402111452625952e-08, "loss": 0.0537, "step": 1944 }, { "epoch": 0.26, "learning_rate": 7.400775090204463e-08, "loss": 0.0562, "step": 1945 }, { "epoch": 0.26, "learning_rate": 7.399438727782974e-08, "loss": 0.0902, "step": 1946 }, { "epoch": 0.26, "learning_rate": 7.398102365361486e-08, "loss": 0.0471, "step": 1947 }, { "epoch": 0.26, "learning_rate": 7.396766002939997e-08, "loss": 0.0651, "step": 1948 }, { "epoch": 0.26, "learning_rate": 7.395429640518508e-08, "loss": 0.0393, "step": 1949 }, { "epoch": 0.26, "learning_rate": 7.39409327809702e-08, "loss": 0.0379, "step": 1950 }, { "epoch": 0.26, "learning_rate": 7.392756915675532e-08, "loss": 0.0425, "step": 1951 }, { "epoch": 0.26, "learning_rate": 7.391420553254043e-08, "loss": 0.0524, "step": 1952 }, { "epoch": 0.26, "learning_rate": 7.390084190832554e-08, "loss": 0.0489, "step": 1953 }, { "epoch": 0.26, "learning_rate": 7.388747828411065e-08, "loss": 0.0412, "step": 1954 }, { "epoch": 0.26, "learning_rate": 7.387411465989575e-08, "loss": 0.095, "step": 1955 }, { "epoch": 0.26, "learning_rate": 7.386075103568086e-08, "loss": 0.0418, "step": 1956 }, { "epoch": 0.26, "learning_rate": 7.384738741146599e-08, "loss": 0.0434, "step": 1957 }, { "epoch": 0.26, "learning_rate": 7.38340237872511e-08, "loss": 0.0562, "step": 1958 }, { "epoch": 0.26, "learning_rate": 7.382066016303621e-08, "loss": 0.0624, "step": 1959 }, { "epoch": 0.26, "learning_rate": 7.380729653882133e-08, "loss": 0.0874, "step": 1960 }, { "epoch": 0.26, "learning_rate": 7.379393291460644e-08, "loss": 0.0364, "step": 1961 }, { "epoch": 0.26, "learning_rate": 7.378056929039155e-08, "loss": 0.0384, "step": 1962 }, { "epoch": 0.26, "learning_rate": 7.376720566617667e-08, "loss": 0.0479, "step": 1963 }, { "epoch": 0.26, "learning_rate": 7.375384204196178e-08, "loss": 0.0427, "step": 1964 }, { "epoch": 0.26, "learning_rate": 7.374047841774689e-08, "loss": 0.0642, "step": 1965 }, { "epoch": 0.26, "learning_rate": 7.3727114793532e-08, "loss": 0.0299, "step": 1966 }, { "epoch": 0.26, "learning_rate": 7.371375116931711e-08, "loss": 0.03, "step": 1967 }, { "epoch": 0.26, "learning_rate": 7.370038754510222e-08, "loss": 0.0495, "step": 1968 }, { "epoch": 0.26, "learning_rate": 7.368702392088733e-08, "loss": 0.0484, "step": 1969 }, { "epoch": 0.26, "learning_rate": 7.367366029667246e-08, "loss": 0.0602, "step": 1970 }, { "epoch": 0.26, "learning_rate": 7.366029667245757e-08, "loss": 0.067, "step": 1971 }, { "epoch": 0.26, "learning_rate": 7.364693304824267e-08, "loss": 0.0524, "step": 1972 }, { "epoch": 0.26, "learning_rate": 7.36335694240278e-08, "loss": 0.0305, "step": 1973 }, { "epoch": 0.26, "learning_rate": 7.362020579981291e-08, "loss": 0.0596, "step": 1974 }, { "epoch": 0.26, "learning_rate": 7.360684217559802e-08, "loss": 0.0367, "step": 1975 }, { "epoch": 0.26, "learning_rate": 7.359347855138314e-08, "loss": 0.049, "step": 1976 }, { "epoch": 0.26, "learning_rate": 7.358011492716825e-08, "loss": 0.0353, "step": 1977 }, { "epoch": 0.26, "learning_rate": 7.356675130295336e-08, "loss": 0.0399, "step": 1978 }, { "epoch": 0.26, "learning_rate": 7.355338767873847e-08, "loss": 0.0538, "step": 1979 }, { "epoch": 0.26, "learning_rate": 7.354002405452358e-08, "loss": 0.0702, "step": 1980 }, { "epoch": 0.26, "learning_rate": 7.352666043030869e-08, "loss": 0.0729, "step": 1981 }, { "epoch": 0.26, "learning_rate": 7.35132968060938e-08, "loss": 0.0765, "step": 1982 }, { "epoch": 0.27, "learning_rate": 7.349993318187892e-08, "loss": 0.0416, "step": 1983 }, { "epoch": 0.27, "learning_rate": 7.348656955766403e-08, "loss": 0.039, "step": 1984 }, { "epoch": 0.27, "learning_rate": 7.347320593344914e-08, "loss": 0.0404, "step": 1985 }, { "epoch": 0.27, "learning_rate": 7.345984230923427e-08, "loss": 0.058, "step": 1986 }, { "epoch": 0.27, "learning_rate": 7.344647868501938e-08, "loss": 0.0605, "step": 1987 }, { "epoch": 0.27, "learning_rate": 7.343311506080449e-08, "loss": 0.0548, "step": 1988 }, { "epoch": 0.27, "learning_rate": 7.34197514365896e-08, "loss": 0.0527, "step": 1989 }, { "epoch": 0.27, "learning_rate": 7.340638781237472e-08, "loss": 0.0665, "step": 1990 }, { "epoch": 0.27, "learning_rate": 7.339302418815983e-08, "loss": 0.0313, "step": 1991 }, { "epoch": 0.27, "learning_rate": 7.337966056394494e-08, "loss": 0.0393, "step": 1992 }, { "epoch": 0.27, "learning_rate": 7.336629693973006e-08, "loss": 0.0476, "step": 1993 }, { "epoch": 0.27, "learning_rate": 7.335293331551517e-08, "loss": 0.0359, "step": 1994 }, { "epoch": 0.27, "learning_rate": 7.333956969130027e-08, "loss": 0.0695, "step": 1995 }, { "epoch": 0.27, "learning_rate": 7.332620606708539e-08, "loss": 0.032, "step": 1996 }, { "epoch": 0.27, "learning_rate": 7.33128424428705e-08, "loss": 0.041, "step": 1997 }, { "epoch": 0.27, "learning_rate": 7.329947881865561e-08, "loss": 0.0642, "step": 1998 }, { "epoch": 0.27, "learning_rate": 7.328611519444073e-08, "loss": 0.0477, "step": 1999 }, { "epoch": 0.27, "learning_rate": 7.327275157022584e-08, "loss": 0.0522, "step": 2000 }, { "epoch": 0.27, "learning_rate": 7.325938794601095e-08, "loss": 0.0318, "step": 2001 }, { "epoch": 0.27, "learning_rate": 7.324602432179606e-08, "loss": 0.0467, "step": 2002 }, { "epoch": 0.27, "learning_rate": 7.323266069758119e-08, "loss": 0.0466, "step": 2003 }, { "epoch": 0.27, "learning_rate": 7.32192970733663e-08, "loss": 0.044, "step": 2004 }, { "epoch": 0.27, "learning_rate": 7.32059334491514e-08, "loss": 0.0437, "step": 2005 }, { "epoch": 0.27, "learning_rate": 7.319256982493653e-08, "loss": 0.0614, "step": 2006 }, { "epoch": 0.27, "learning_rate": 7.317920620072164e-08, "loss": 0.0394, "step": 2007 }, { "epoch": 0.27, "learning_rate": 7.316584257650675e-08, "loss": 0.0527, "step": 2008 }, { "epoch": 0.27, "learning_rate": 7.315247895229186e-08, "loss": 0.0945, "step": 2009 }, { "epoch": 0.27, "learning_rate": 7.313911532807697e-08, "loss": 0.0376, "step": 2010 }, { "epoch": 0.27, "learning_rate": 7.312575170386208e-08, "loss": 0.0947, "step": 2011 }, { "epoch": 0.27, "learning_rate": 7.311238807964719e-08, "loss": 0.0541, "step": 2012 }, { "epoch": 0.27, "learning_rate": 7.309902445543231e-08, "loss": 0.0486, "step": 2013 }, { "epoch": 0.27, "learning_rate": 7.308566083121742e-08, "loss": 0.0599, "step": 2014 }, { "epoch": 0.27, "learning_rate": 7.308566083121742e-08, "loss": 0.0627, "step": 2015 }, { "epoch": 0.27, "learning_rate": 7.307229720700253e-08, "loss": 0.0597, "step": 2016 }, { "epoch": 0.27, "learning_rate": 7.305893358278765e-08, "loss": 0.039, "step": 2017 }, { "epoch": 0.27, "learning_rate": 7.304556995857276e-08, "loss": 0.0531, "step": 2018 }, { "epoch": 0.27, "learning_rate": 7.303220633435787e-08, "loss": 0.1045, "step": 2019 }, { "epoch": 0.27, "learning_rate": 7.3018842710143e-08, "loss": 0.0632, "step": 2020 }, { "epoch": 0.27, "learning_rate": 7.30054790859281e-08, "loss": 0.0652, "step": 2021 }, { "epoch": 0.27, "learning_rate": 7.299211546171322e-08, "loss": 0.056, "step": 2022 }, { "epoch": 0.27, "learning_rate": 7.297875183749833e-08, "loss": 0.0607, "step": 2023 }, { "epoch": 0.27, "learning_rate": 7.296538821328344e-08, "loss": 0.059, "step": 2024 }, { "epoch": 0.27, "learning_rate": 7.295202458906855e-08, "loss": 0.0523, "step": 2025 }, { "epoch": 0.27, "learning_rate": 7.293866096485365e-08, "loss": 0.0429, "step": 2026 }, { "epoch": 0.27, "learning_rate": 7.292529734063878e-08, "loss": 0.0643, "step": 2027 }, { "epoch": 0.27, "learning_rate": 7.291193371642389e-08, "loss": 0.0397, "step": 2028 }, { "epoch": 0.27, "learning_rate": 7.2898570092209e-08, "loss": 0.0637, "step": 2029 }, { "epoch": 0.27, "learning_rate": 7.288520646799412e-08, "loss": 0.0663, "step": 2030 }, { "epoch": 0.27, "learning_rate": 7.287184284377923e-08, "loss": 0.0576, "step": 2031 }, { "epoch": 0.27, "learning_rate": 7.285847921956434e-08, "loss": 0.0392, "step": 2032 }, { "epoch": 0.27, "learning_rate": 7.284511559534946e-08, "loss": 0.0462, "step": 2033 }, { "epoch": 0.27, "learning_rate": 7.283175197113457e-08, "loss": 0.057, "step": 2034 }, { "epoch": 0.27, "learning_rate": 7.281838834691968e-08, "loss": 0.0614, "step": 2035 }, { "epoch": 0.27, "learning_rate": 7.28050247227048e-08, "loss": 0.0584, "step": 2036 }, { "epoch": 0.27, "learning_rate": 7.27916610984899e-08, "loss": 0.0496, "step": 2037 }, { "epoch": 0.27, "learning_rate": 7.277829747427501e-08, "loss": 0.0653, "step": 2038 }, { "epoch": 0.27, "learning_rate": 7.276493385006012e-08, "loss": 0.0397, "step": 2039 }, { "epoch": 0.27, "learning_rate": 7.275157022584525e-08, "loss": 0.067, "step": 2040 }, { "epoch": 0.27, "learning_rate": 7.273820660163036e-08, "loss": 0.0329, "step": 2041 }, { "epoch": 0.27, "learning_rate": 7.272484297741547e-08, "loss": 0.0548, "step": 2042 }, { "epoch": 0.27, "learning_rate": 7.271147935320059e-08, "loss": 0.0405, "step": 2043 }, { "epoch": 0.27, "learning_rate": 7.26981157289857e-08, "loss": 0.0522, "step": 2044 }, { "epoch": 0.27, "learning_rate": 7.268475210477081e-08, "loss": 0.0476, "step": 2045 }, { "epoch": 0.27, "learning_rate": 7.267138848055593e-08, "loss": 0.0334, "step": 2046 }, { "epoch": 0.27, "learning_rate": 7.265802485634104e-08, "loss": 0.0592, "step": 2047 }, { "epoch": 0.27, "learning_rate": 7.264466123212615e-08, "loss": 0.0516, "step": 2048 }, { "epoch": 0.27, "learning_rate": 7.263129760791126e-08, "loss": 0.0383, "step": 2049 }, { "epoch": 0.27, "learning_rate": 7.261793398369638e-08, "loss": 0.0556, "step": 2050 }, { "epoch": 0.27, "learning_rate": 7.26045703594815e-08, "loss": 0.0506, "step": 2051 }, { "epoch": 0.27, "learning_rate": 7.25912067352666e-08, "loss": 0.0585, "step": 2052 }, { "epoch": 0.27, "learning_rate": 7.257784311105171e-08, "loss": 0.0555, "step": 2053 }, { "epoch": 0.27, "learning_rate": 7.256447948683682e-08, "loss": 0.0478, "step": 2054 }, { "epoch": 0.27, "learning_rate": 7.255111586262193e-08, "loss": 0.0357, "step": 2055 }, { "epoch": 0.27, "learning_rate": 7.253775223840706e-08, "loss": 0.0536, "step": 2056 }, { "epoch": 0.27, "learning_rate": 7.252438861419217e-08, "loss": 0.0349, "step": 2057 }, { "epoch": 0.28, "learning_rate": 7.251102498997728e-08, "loss": 0.0444, "step": 2058 }, { "epoch": 0.28, "learning_rate": 7.24976613657624e-08, "loss": 0.0377, "step": 2059 }, { "epoch": 0.28, "learning_rate": 7.248429774154751e-08, "loss": 0.0551, "step": 2060 }, { "epoch": 0.28, "learning_rate": 7.247093411733262e-08, "loss": 0.0421, "step": 2061 }, { "epoch": 0.28, "learning_rate": 7.245757049311773e-08, "loss": 0.0545, "step": 2062 }, { "epoch": 0.28, "learning_rate": 7.244420686890285e-08, "loss": 0.0551, "step": 2063 }, { "epoch": 0.28, "learning_rate": 7.243084324468796e-08, "loss": 0.0462, "step": 2064 }, { "epoch": 0.28, "learning_rate": 7.241747962047307e-08, "loss": 0.0808, "step": 2065 }, { "epoch": 0.28, "learning_rate": 7.240411599625818e-08, "loss": 0.0557, "step": 2066 }, { "epoch": 0.28, "learning_rate": 7.239075237204329e-08, "loss": 0.0563, "step": 2067 }, { "epoch": 0.28, "learning_rate": 7.23773887478284e-08, "loss": 0.046, "step": 2068 }, { "epoch": 0.28, "learning_rate": 7.236402512361352e-08, "loss": 0.0475, "step": 2069 }, { "epoch": 0.28, "learning_rate": 7.235066149939863e-08, "loss": 0.0416, "step": 2070 }, { "epoch": 0.28, "learning_rate": 7.233729787518374e-08, "loss": 0.0749, "step": 2071 }, { "epoch": 0.28, "learning_rate": 7.232393425096885e-08, "loss": 0.0464, "step": 2072 }, { "epoch": 0.28, "learning_rate": 7.231057062675398e-08, "loss": 0.0337, "step": 2073 }, { "epoch": 0.28, "learning_rate": 7.229720700253909e-08, "loss": 0.0663, "step": 2074 }, { "epoch": 0.28, "learning_rate": 7.22838433783242e-08, "loss": 0.0676, "step": 2075 }, { "epoch": 0.28, "learning_rate": 7.227047975410932e-08, "loss": 0.0405, "step": 2076 }, { "epoch": 0.28, "learning_rate": 7.225711612989443e-08, "loss": 0.1091, "step": 2077 }, { "epoch": 0.28, "learning_rate": 7.224375250567954e-08, "loss": 0.0565, "step": 2078 }, { "epoch": 0.28, "learning_rate": 7.223038888146465e-08, "loss": 0.0801, "step": 2079 }, { "epoch": 0.28, "learning_rate": 7.221702525724976e-08, "loss": 0.0509, "step": 2080 }, { "epoch": 0.28, "learning_rate": 7.220366163303487e-08, "loss": 0.0637, "step": 2081 }, { "epoch": 0.28, "learning_rate": 7.219029800881999e-08, "loss": 0.0567, "step": 2082 }, { "epoch": 0.28, "learning_rate": 7.21769343846051e-08, "loss": 0.056, "step": 2083 }, { "epoch": 0.28, "learning_rate": 7.216357076039021e-08, "loss": 0.0614, "step": 2084 }, { "epoch": 0.28, "learning_rate": 7.215020713617532e-08, "loss": 0.0664, "step": 2085 }, { "epoch": 0.28, "learning_rate": 7.213684351196044e-08, "loss": 0.0508, "step": 2086 }, { "epoch": 0.28, "learning_rate": 7.212347988774555e-08, "loss": 0.0551, "step": 2087 }, { "epoch": 0.28, "learning_rate": 7.211011626353066e-08, "loss": 0.0673, "step": 2088 }, { "epoch": 0.28, "learning_rate": 7.209675263931579e-08, "loss": 0.052, "step": 2089 }, { "epoch": 0.28, "learning_rate": 7.20833890151009e-08, "loss": 0.043, "step": 2090 }, { "epoch": 0.28, "learning_rate": 7.2070025390886e-08, "loss": 0.054, "step": 2091 }, { "epoch": 0.28, "learning_rate": 7.205666176667113e-08, "loss": 0.0288, "step": 2092 }, { "epoch": 0.28, "learning_rate": 7.204329814245624e-08, "loss": 0.0454, "step": 2093 }, { "epoch": 0.28, "learning_rate": 7.202993451824135e-08, "loss": 0.0349, "step": 2094 }, { "epoch": 0.28, "learning_rate": 7.201657089402645e-08, "loss": 0.0552, "step": 2095 }, { "epoch": 0.28, "learning_rate": 7.200320726981157e-08, "loss": 0.0402, "step": 2096 }, { "epoch": 0.28, "learning_rate": 7.198984364559668e-08, "loss": 0.0537, "step": 2097 }, { "epoch": 0.28, "learning_rate": 7.197648002138179e-08, "loss": 0.0512, "step": 2098 }, { "epoch": 0.28, "learning_rate": 7.196311639716691e-08, "loss": 0.0327, "step": 2099 }, { "epoch": 0.28, "learning_rate": 7.194975277295202e-08, "loss": 0.0652, "step": 2100 }, { "epoch": 0.28, "learning_rate": 7.193638914873713e-08, "loss": 0.082, "step": 2101 }, { "epoch": 0.28, "learning_rate": 7.192302552452225e-08, "loss": 0.0592, "step": 2102 }, { "epoch": 0.28, "learning_rate": 7.190966190030736e-08, "loss": 0.0383, "step": 2103 }, { "epoch": 0.28, "learning_rate": 7.189629827609247e-08, "loss": 0.0765, "step": 2104 }, { "epoch": 0.28, "learning_rate": 7.18829346518776e-08, "loss": 0.0393, "step": 2105 }, { "epoch": 0.28, "learning_rate": 7.18695710276627e-08, "loss": 0.0526, "step": 2106 }, { "epoch": 0.28, "learning_rate": 7.185620740344782e-08, "loss": 0.0481, "step": 2107 }, { "epoch": 0.28, "learning_rate": 7.184284377923293e-08, "loss": 0.0646, "step": 2108 }, { "epoch": 0.28, "learning_rate": 7.182948015501804e-08, "loss": 0.0553, "step": 2109 }, { "epoch": 0.28, "learning_rate": 7.181611653080315e-08, "loss": 0.056, "step": 2110 }, { "epoch": 0.28, "learning_rate": 7.180275290658826e-08, "loss": 0.0548, "step": 2111 }, { "epoch": 0.28, "learning_rate": 7.178938928237338e-08, "loss": 0.0462, "step": 2112 }, { "epoch": 0.28, "learning_rate": 7.177602565815849e-08, "loss": 0.0645, "step": 2113 }, { "epoch": 0.28, "learning_rate": 7.17626620339436e-08, "loss": 0.0583, "step": 2114 }, { "epoch": 0.28, "learning_rate": 7.174929840972872e-08, "loss": 0.0588, "step": 2115 }, { "epoch": 0.28, "learning_rate": 7.173593478551383e-08, "loss": 0.0514, "step": 2116 }, { "epoch": 0.28, "learning_rate": 7.172257116129894e-08, "loss": 0.0435, "step": 2117 }, { "epoch": 0.28, "learning_rate": 7.170920753708405e-08, "loss": 0.0505, "step": 2118 }, { "epoch": 0.28, "learning_rate": 7.169584391286917e-08, "loss": 0.0395, "step": 2119 }, { "epoch": 0.28, "learning_rate": 7.168248028865428e-08, "loss": 0.0516, "step": 2120 }, { "epoch": 0.28, "learning_rate": 7.16691166644394e-08, "loss": 0.0493, "step": 2121 }, { "epoch": 0.28, "learning_rate": 7.16557530402245e-08, "loss": 0.0444, "step": 2122 }, { "epoch": 0.28, "learning_rate": 7.164238941600961e-08, "loss": 0.0388, "step": 2123 }, { "epoch": 0.28, "learning_rate": 7.162902579179472e-08, "loss": 0.0442, "step": 2124 }, { "epoch": 0.28, "learning_rate": 7.161566216757985e-08, "loss": 0.0427, "step": 2125 }, { "epoch": 0.28, "learning_rate": 7.160229854336496e-08, "loss": 0.063, "step": 2126 }, { "epoch": 0.28, "learning_rate": 7.158893491915007e-08, "loss": 0.041, "step": 2127 }, { "epoch": 0.28, "learning_rate": 7.157557129493519e-08, "loss": 0.0637, "step": 2128 }, { "epoch": 0.28, "learning_rate": 7.15622076707203e-08, "loss": 0.0323, "step": 2129 }, { "epoch": 0.28, "learning_rate": 7.154884404650541e-08, "loss": 0.044, "step": 2130 }, { "epoch": 0.28, "learning_rate": 7.153548042229052e-08, "loss": 0.0447, "step": 2131 }, { "epoch": 0.28, "learning_rate": 7.152211679807564e-08, "loss": 0.0431, "step": 2132 }, { "epoch": 0.29, "learning_rate": 7.150875317386075e-08, "loss": 0.0628, "step": 2133 }, { "epoch": 0.29, "learning_rate": 7.149538954964586e-08, "loss": 0.046, "step": 2134 }, { "epoch": 0.29, "learning_rate": 7.148202592543098e-08, "loss": 0.0292, "step": 2135 }, { "epoch": 0.29, "learning_rate": 7.146866230121608e-08, "loss": 0.065, "step": 2136 }, { "epoch": 0.29, "learning_rate": 7.145529867700119e-08, "loss": 0.0476, "step": 2137 }, { "epoch": 0.29, "learning_rate": 7.144193505278631e-08, "loss": 0.0789, "step": 2138 }, { "epoch": 0.29, "learning_rate": 7.142857142857142e-08, "loss": 0.0499, "step": 2139 }, { "epoch": 0.29, "learning_rate": 7.141520780435653e-08, "loss": 0.0948, "step": 2140 }, { "epoch": 0.29, "learning_rate": 7.140184418014164e-08, "loss": 0.0269, "step": 2141 }, { "epoch": 0.29, "learning_rate": 7.138848055592677e-08, "loss": 0.0586, "step": 2142 }, { "epoch": 0.29, "learning_rate": 7.137511693171188e-08, "loss": 0.0766, "step": 2143 }, { "epoch": 0.29, "learning_rate": 7.136175330749699e-08, "loss": 0.0521, "step": 2144 }, { "epoch": 0.29, "learning_rate": 7.134838968328211e-08, "loss": 0.0698, "step": 2145 }, { "epoch": 0.29, "learning_rate": 7.133502605906722e-08, "loss": 0.0604, "step": 2146 }, { "epoch": 0.29, "learning_rate": 7.132166243485233e-08, "loss": 0.0519, "step": 2147 }, { "epoch": 0.29, "learning_rate": 7.130829881063745e-08, "loss": 0.0451, "step": 2148 }, { "epoch": 0.29, "learning_rate": 7.129493518642256e-08, "loss": 0.0616, "step": 2149 }, { "epoch": 0.29, "learning_rate": 7.128157156220767e-08, "loss": 0.0471, "step": 2150 }, { "epoch": 0.29, "learning_rate": 7.126820793799278e-08, "loss": 0.0198, "step": 2151 }, { "epoch": 0.29, "learning_rate": 7.125484431377789e-08, "loss": 0.0687, "step": 2152 }, { "epoch": 0.29, "learning_rate": 7.1241480689563e-08, "loss": 0.0704, "step": 2153 }, { "epoch": 0.29, "learning_rate": 7.122811706534811e-08, "loss": 0.059, "step": 2154 }, { "epoch": 0.29, "learning_rate": 7.121475344113323e-08, "loss": 0.0485, "step": 2155 }, { "epoch": 0.29, "learning_rate": 7.120138981691834e-08, "loss": 0.0358, "step": 2156 }, { "epoch": 0.29, "learning_rate": 7.118802619270345e-08, "loss": 0.0439, "step": 2157 }, { "epoch": 0.29, "learning_rate": 7.117466256848858e-08, "loss": 0.0293, "step": 2158 }, { "epoch": 0.29, "learning_rate": 7.116129894427369e-08, "loss": 0.0616, "step": 2159 }, { "epoch": 0.29, "learning_rate": 7.11479353200588e-08, "loss": 0.0598, "step": 2160 }, { "epoch": 0.29, "learning_rate": 7.113457169584392e-08, "loss": 0.0411, "step": 2161 }, { "epoch": 0.29, "learning_rate": 7.112120807162903e-08, "loss": 0.0384, "step": 2162 }, { "epoch": 0.29, "learning_rate": 7.110784444741414e-08, "loss": 0.0479, "step": 2163 }, { "epoch": 0.29, "learning_rate": 7.109448082319925e-08, "loss": 0.053, "step": 2164 }, { "epoch": 0.29, "learning_rate": 7.108111719898436e-08, "loss": 0.0567, "step": 2165 }, { "epoch": 0.29, "learning_rate": 7.106775357476947e-08, "loss": 0.0575, "step": 2166 }, { "epoch": 0.29, "learning_rate": 7.105438995055458e-08, "loss": 0.0419, "step": 2167 }, { "epoch": 0.29, "learning_rate": 7.10410263263397e-08, "loss": 0.0665, "step": 2168 }, { "epoch": 0.29, "learning_rate": 7.102766270212481e-08, "loss": 0.0493, "step": 2169 }, { "epoch": 0.29, "learning_rate": 7.101429907790992e-08, "loss": 0.0577, "step": 2170 }, { "epoch": 0.29, "learning_rate": 7.100093545369504e-08, "loss": 0.0686, "step": 2171 }, { "epoch": 0.29, "learning_rate": 7.098757182948015e-08, "loss": 0.0685, "step": 2172 }, { "epoch": 0.29, "learning_rate": 7.097420820526526e-08, "loss": 0.0608, "step": 2173 }, { "epoch": 0.29, "learning_rate": 7.096084458105039e-08, "loss": 0.0432, "step": 2174 }, { "epoch": 0.29, "learning_rate": 7.09474809568355e-08, "loss": 0.0522, "step": 2175 }, { "epoch": 0.29, "learning_rate": 7.09341173326206e-08, "loss": 0.0871, "step": 2176 }, { "epoch": 0.29, "learning_rate": 7.092075370840572e-08, "loss": 0.0426, "step": 2177 }, { "epoch": 0.29, "learning_rate": 7.090739008419083e-08, "loss": 0.064, "step": 2178 }, { "epoch": 0.29, "learning_rate": 7.089402645997594e-08, "loss": 0.0583, "step": 2179 }, { "epoch": 0.29, "learning_rate": 7.088066283576105e-08, "loss": 0.0781, "step": 2180 }, { "epoch": 0.29, "learning_rate": 7.086729921154617e-08, "loss": 0.0388, "step": 2181 }, { "epoch": 0.29, "learning_rate": 7.085393558733128e-08, "loss": 0.0492, "step": 2182 }, { "epoch": 0.29, "learning_rate": 7.084057196311639e-08, "loss": 0.0514, "step": 2183 }, { "epoch": 0.29, "learning_rate": 7.082720833890151e-08, "loss": 0.0577, "step": 2184 }, { "epoch": 0.29, "learning_rate": 7.081384471468662e-08, "loss": 0.0477, "step": 2185 }, { "epoch": 0.29, "learning_rate": 7.080048109047173e-08, "loss": 0.068, "step": 2186 }, { "epoch": 0.29, "learning_rate": 7.078711746625684e-08, "loss": 0.0422, "step": 2187 }, { "epoch": 0.29, "learning_rate": 7.077375384204196e-08, "loss": 0.0572, "step": 2188 }, { "epoch": 0.29, "learning_rate": 7.076039021782707e-08, "loss": 0.0767, "step": 2189 }, { "epoch": 0.29, "learning_rate": 7.074702659361218e-08, "loss": 0.0466, "step": 2190 }, { "epoch": 0.29, "learning_rate": 7.073366296939731e-08, "loss": 0.0459, "step": 2191 }, { "epoch": 0.29, "learning_rate": 7.072029934518242e-08, "loss": 0.0541, "step": 2192 }, { "epoch": 0.29, "learning_rate": 7.070693572096753e-08, "loss": 0.0433, "step": 2193 }, { "epoch": 0.29, "learning_rate": 7.069357209675264e-08, "loss": 0.0468, "step": 2194 }, { "epoch": 0.29, "learning_rate": 7.068020847253775e-08, "loss": 0.0404, "step": 2195 }, { "epoch": 0.29, "learning_rate": 7.066684484832286e-08, "loss": 0.0487, "step": 2196 }, { "epoch": 0.29, "learning_rate": 7.065348122410798e-08, "loss": 0.0364, "step": 2197 }, { "epoch": 0.29, "learning_rate": 7.064011759989309e-08, "loss": 0.0459, "step": 2198 }, { "epoch": 0.29, "learning_rate": 7.06267539756782e-08, "loss": 0.0533, "step": 2199 }, { "epoch": 0.29, "learning_rate": 7.061339035146331e-08, "loss": 0.065, "step": 2200 }, { "epoch": 0.29, "learning_rate": 7.060002672724843e-08, "loss": 0.0471, "step": 2201 }, { "epoch": 0.29, "learning_rate": 7.058666310303354e-08, "loss": 0.0425, "step": 2202 }, { "epoch": 0.29, "learning_rate": 7.057329947881865e-08, "loss": 0.0409, "step": 2203 }, { "epoch": 0.29, "learning_rate": 7.055993585460377e-08, "loss": 0.0379, "step": 2204 }, { "epoch": 0.29, "learning_rate": 7.054657223038888e-08, "loss": 0.055, "step": 2205 }, { "epoch": 0.29, "learning_rate": 7.0533208606174e-08, "loss": 0.0474, "step": 2206 }, { "epoch": 0.29, "learning_rate": 7.05198449819591e-08, "loss": 0.048, "step": 2207 }, { "epoch": 0.3, "learning_rate": 7.050648135774421e-08, "loss": 0.0561, "step": 2208 }, { "epoch": 0.3, "learning_rate": 7.049311773352932e-08, "loss": 0.0525, "step": 2209 }, { "epoch": 0.3, "learning_rate": 7.047975410931443e-08, "loss": 0.0485, "step": 2210 }, { "epoch": 0.3, "learning_rate": 7.046639048509956e-08, "loss": 0.0497, "step": 2211 }, { "epoch": 0.3, "learning_rate": 7.045302686088467e-08, "loss": 0.0406, "step": 2212 }, { "epoch": 0.3, "learning_rate": 7.043966323666978e-08, "loss": 0.0564, "step": 2213 }, { "epoch": 0.3, "learning_rate": 7.04262996124549e-08, "loss": 0.0641, "step": 2214 }, { "epoch": 0.3, "learning_rate": 7.041293598824001e-08, "loss": 0.0494, "step": 2215 }, { "epoch": 0.3, "learning_rate": 7.039957236402512e-08, "loss": 0.0566, "step": 2216 }, { "epoch": 0.3, "learning_rate": 7.038620873981024e-08, "loss": 0.0804, "step": 2217 }, { "epoch": 0.3, "learning_rate": 7.037284511559535e-08, "loss": 0.0402, "step": 2218 }, { "epoch": 0.3, "learning_rate": 7.035948149138046e-08, "loss": 0.0372, "step": 2219 }, { "epoch": 0.3, "learning_rate": 7.034611786716557e-08, "loss": 0.052, "step": 2220 }, { "epoch": 0.3, "learning_rate": 7.033275424295068e-08, "loss": 0.05, "step": 2221 }, { "epoch": 0.3, "learning_rate": 7.031939061873579e-08, "loss": 0.0459, "step": 2222 }, { "epoch": 0.3, "learning_rate": 7.03060269945209e-08, "loss": 0.0442, "step": 2223 }, { "epoch": 0.3, "learning_rate": 7.029266337030602e-08, "loss": 0.039, "step": 2224 }, { "epoch": 0.3, "learning_rate": 7.027929974609113e-08, "loss": 0.0424, "step": 2225 }, { "epoch": 0.3, "learning_rate": 7.026593612187624e-08, "loss": 0.0411, "step": 2226 }, { "epoch": 0.3, "learning_rate": 7.025257249766137e-08, "loss": 0.0502, "step": 2227 }, { "epoch": 0.3, "learning_rate": 7.023920887344648e-08, "loss": 0.0503, "step": 2228 }, { "epoch": 0.3, "learning_rate": 7.022584524923159e-08, "loss": 0.0516, "step": 2229 }, { "epoch": 0.3, "learning_rate": 7.021248162501671e-08, "loss": 0.0414, "step": 2230 }, { "epoch": 0.3, "learning_rate": 7.019911800080182e-08, "loss": 0.057, "step": 2231 }, { "epoch": 0.3, "learning_rate": 7.018575437658693e-08, "loss": 0.0794, "step": 2232 }, { "epoch": 0.3, "learning_rate": 7.017239075237204e-08, "loss": 0.0617, "step": 2233 }, { "epoch": 0.3, "learning_rate": 7.015902712815716e-08, "loss": 0.0551, "step": 2234 }, { "epoch": 0.3, "learning_rate": 7.014566350394226e-08, "loss": 0.0393, "step": 2235 }, { "epoch": 0.3, "learning_rate": 7.013229987972737e-08, "loss": 0.0707, "step": 2236 }, { "epoch": 0.3, "learning_rate": 7.011893625551249e-08, "loss": 0.0538, "step": 2237 }, { "epoch": 0.3, "learning_rate": 7.01055726312976e-08, "loss": 0.0743, "step": 2238 }, { "epoch": 0.3, "learning_rate": 7.009220900708271e-08, "loss": 0.0466, "step": 2239 }, { "epoch": 0.3, "learning_rate": 7.007884538286783e-08, "loss": 0.0578, "step": 2240 }, { "epoch": 0.3, "learning_rate": 7.006548175865294e-08, "loss": 0.0278, "step": 2241 }, { "epoch": 0.3, "learning_rate": 7.005211813443805e-08, "loss": 0.0525, "step": 2242 }, { "epoch": 0.3, "learning_rate": 7.003875451022318e-08, "loss": 0.0473, "step": 2243 }, { "epoch": 0.3, "learning_rate": 7.002539088600829e-08, "loss": 0.0782, "step": 2244 }, { "epoch": 0.3, "learning_rate": 7.00120272617934e-08, "loss": 0.0535, "step": 2245 }, { "epoch": 0.3, "learning_rate": 6.99986636375785e-08, "loss": 0.0545, "step": 2246 }, { "epoch": 0.3, "learning_rate": 6.998530001336363e-08, "loss": 0.0395, "step": 2247 }, { "epoch": 0.3, "learning_rate": 6.997193638914874e-08, "loss": 0.0635, "step": 2248 }, { "epoch": 0.3, "learning_rate": 6.995857276493385e-08, "loss": 0.0613, "step": 2249 }, { "epoch": 0.3, "learning_rate": 6.994520914071896e-08, "loss": 0.0498, "step": 2250 }, { "epoch": 0.3, "learning_rate": 6.993184551650407e-08, "loss": 0.0404, "step": 2251 }, { "epoch": 0.3, "learning_rate": 6.991848189228918e-08, "loss": 0.0788, "step": 2252 }, { "epoch": 0.3, "learning_rate": 6.99051182680743e-08, "loss": 0.0672, "step": 2253 }, { "epoch": 0.3, "learning_rate": 6.989175464385941e-08, "loss": 0.0434, "step": 2254 }, { "epoch": 0.3, "learning_rate": 6.987839101964452e-08, "loss": 0.0437, "step": 2255 }, { "epoch": 0.3, "learning_rate": 6.986502739542963e-08, "loss": 0.045, "step": 2256 }, { "epoch": 0.3, "learning_rate": 6.985166377121475e-08, "loss": 0.0476, "step": 2257 }, { "epoch": 0.3, "learning_rate": 6.983830014699986e-08, "loss": 0.0377, "step": 2258 }, { "epoch": 0.3, "learning_rate": 6.982493652278497e-08, "loss": 0.0514, "step": 2259 }, { "epoch": 0.3, "learning_rate": 6.98115728985701e-08, "loss": 0.0391, "step": 2260 }, { "epoch": 0.3, "learning_rate": 6.979820927435521e-08, "loss": 0.047, "step": 2261 }, { "epoch": 0.3, "learning_rate": 6.978484565014032e-08, "loss": 0.0261, "step": 2262 }, { "epoch": 0.3, "learning_rate": 6.977148202592543e-08, "loss": 0.0432, "step": 2263 }, { "epoch": 0.3, "learning_rate": 6.975811840171054e-08, "loss": 0.0687, "step": 2264 }, { "epoch": 0.3, "learning_rate": 6.974475477749565e-08, "loss": 0.0887, "step": 2265 }, { "epoch": 0.3, "learning_rate": 6.973139115328077e-08, "loss": 0.0582, "step": 2266 }, { "epoch": 0.3, "learning_rate": 6.971802752906588e-08, "loss": 0.0693, "step": 2267 }, { "epoch": 0.3, "learning_rate": 6.970466390485099e-08, "loss": 0.0636, "step": 2268 }, { "epoch": 0.3, "learning_rate": 6.96913002806361e-08, "loss": 0.0554, "step": 2269 }, { "epoch": 0.3, "learning_rate": 6.967793665642122e-08, "loss": 0.0373, "step": 2270 }, { "epoch": 0.3, "learning_rate": 6.966457303220633e-08, "loss": 0.0496, "step": 2271 }, { "epoch": 0.3, "learning_rate": 6.965120940799144e-08, "loss": 0.0427, "step": 2272 }, { "epoch": 0.3, "learning_rate": 6.963784578377656e-08, "loss": 0.0556, "step": 2273 }, { "epoch": 0.3, "learning_rate": 6.962448215956167e-08, "loss": 0.0559, "step": 2274 }, { "epoch": 0.3, "learning_rate": 6.961111853534678e-08, "loss": 0.0466, "step": 2275 }, { "epoch": 0.3, "learning_rate": 6.95977549111319e-08, "loss": 0.0391, "step": 2276 }, { "epoch": 0.3, "learning_rate": 6.9584391286917e-08, "loss": 0.0471, "step": 2277 }, { "epoch": 0.3, "learning_rate": 6.957102766270211e-08, "loss": 0.0786, "step": 2278 }, { "epoch": 0.3, "learning_rate": 6.955766403848722e-08, "loss": 0.0484, "step": 2279 }, { "epoch": 0.3, "learning_rate": 6.954430041427235e-08, "loss": 0.0605, "step": 2280 }, { "epoch": 0.3, "learning_rate": 6.953093679005746e-08, "loss": 0.0309, "step": 2281 }, { "epoch": 0.3, "learning_rate": 6.951757316584257e-08, "loss": 0.0694, "step": 2282 }, { "epoch": 0.31, "learning_rate": 6.950420954162769e-08, "loss": 0.0616, "step": 2283 }, { "epoch": 0.31, "learning_rate": 6.94908459174128e-08, "loss": 0.0466, "step": 2284 }, { "epoch": 0.31, "learning_rate": 6.947748229319791e-08, "loss": 0.0357, "step": 2285 }, { "epoch": 0.31, "learning_rate": 6.946411866898303e-08, "loss": 0.0543, "step": 2286 }, { "epoch": 0.31, "learning_rate": 6.945075504476814e-08, "loss": 0.0422, "step": 2287 }, { "epoch": 0.31, "learning_rate": 6.943739142055325e-08, "loss": 0.0658, "step": 2288 }, { "epoch": 0.31, "learning_rate": 6.942402779633837e-08, "loss": 0.0484, "step": 2289 }, { "epoch": 0.31, "learning_rate": 6.941066417212348e-08, "loss": 0.0507, "step": 2290 }, { "epoch": 0.31, "learning_rate": 6.93973005479086e-08, "loss": 0.0518, "step": 2291 }, { "epoch": 0.31, "learning_rate": 6.938393692369369e-08, "loss": 0.0744, "step": 2292 }, { "epoch": 0.31, "learning_rate": 6.937057329947881e-08, "loss": 0.0603, "step": 2293 }, { "epoch": 0.31, "learning_rate": 6.935720967526392e-08, "loss": 0.0403, "step": 2294 }, { "epoch": 0.31, "learning_rate": 6.934384605104903e-08, "loss": 0.0689, "step": 2295 }, { "epoch": 0.31, "learning_rate": 6.933048242683416e-08, "loss": 0.1037, "step": 2296 }, { "epoch": 0.31, "learning_rate": 6.931711880261927e-08, "loss": 0.0393, "step": 2297 }, { "epoch": 0.31, "learning_rate": 6.930375517840438e-08, "loss": 0.0408, "step": 2298 }, { "epoch": 0.31, "learning_rate": 6.92903915541895e-08, "loss": 0.0432, "step": 2299 }, { "epoch": 0.31, "learning_rate": 6.927702792997461e-08, "loss": 0.0504, "step": 2300 }, { "epoch": 0.31, "learning_rate": 6.926366430575972e-08, "loss": 0.0362, "step": 2301 }, { "epoch": 0.31, "learning_rate": 6.925030068154483e-08, "loss": 0.0456, "step": 2302 }, { "epoch": 0.31, "learning_rate": 6.923693705732995e-08, "loss": 0.0601, "step": 2303 }, { "epoch": 0.31, "learning_rate": 6.922357343311506e-08, "loss": 0.0462, "step": 2304 }, { "epoch": 0.31, "learning_rate": 6.921020980890017e-08, "loss": 0.0493, "step": 2305 }, { "epoch": 0.31, "learning_rate": 6.919684618468528e-08, "loss": 0.0404, "step": 2306 }, { "epoch": 0.31, "learning_rate": 6.918348256047039e-08, "loss": 0.057, "step": 2307 }, { "epoch": 0.31, "learning_rate": 6.91701189362555e-08, "loss": 0.052, "step": 2308 }, { "epoch": 0.31, "learning_rate": 6.915675531204062e-08, "loss": 0.051, "step": 2309 }, { "epoch": 0.31, "learning_rate": 6.914339168782573e-08, "loss": 0.0516, "step": 2310 }, { "epoch": 0.31, "learning_rate": 6.913002806361084e-08, "loss": 0.0376, "step": 2311 }, { "epoch": 0.31, "learning_rate": 6.911666443939597e-08, "loss": 0.0775, "step": 2312 }, { "epoch": 0.31, "learning_rate": 6.910330081518108e-08, "loss": 0.0627, "step": 2313 }, { "epoch": 0.31, "learning_rate": 6.908993719096619e-08, "loss": 0.03, "step": 2314 }, { "epoch": 0.31, "learning_rate": 6.90765735667513e-08, "loss": 0.0506, "step": 2315 }, { "epoch": 0.31, "learning_rate": 6.906320994253642e-08, "loss": 0.0529, "step": 2316 }, { "epoch": 0.31, "learning_rate": 6.904984631832153e-08, "loss": 0.0457, "step": 2317 }, { "epoch": 0.31, "learning_rate": 6.903648269410664e-08, "loss": 0.0411, "step": 2318 }, { "epoch": 0.31, "learning_rate": 6.902311906989175e-08, "loss": 0.0302, "step": 2319 }, { "epoch": 0.31, "learning_rate": 6.900975544567686e-08, "loss": 0.0354, "step": 2320 }, { "epoch": 0.31, "learning_rate": 6.899639182146197e-08, "loss": 0.0836, "step": 2321 }, { "epoch": 0.31, "learning_rate": 6.898302819724709e-08, "loss": 0.0448, "step": 2322 }, { "epoch": 0.31, "learning_rate": 6.89696645730322e-08, "loss": 0.0638, "step": 2323 }, { "epoch": 0.31, "learning_rate": 6.895630094881731e-08, "loss": 0.0365, "step": 2324 }, { "epoch": 0.31, "learning_rate": 6.894293732460242e-08, "loss": 0.0508, "step": 2325 }, { "epoch": 0.31, "learning_rate": 6.892957370038754e-08, "loss": 0.0566, "step": 2326 }, { "epoch": 0.31, "learning_rate": 6.891621007617265e-08, "loss": 0.0351, "step": 2327 }, { "epoch": 0.31, "learning_rate": 6.890284645195776e-08, "loss": 0.0457, "step": 2328 }, { "epoch": 0.31, "learning_rate": 6.888948282774289e-08, "loss": 0.0513, "step": 2329 }, { "epoch": 0.31, "learning_rate": 6.8876119203528e-08, "loss": 0.0465, "step": 2330 }, { "epoch": 0.31, "learning_rate": 6.886275557931311e-08, "loss": 0.0667, "step": 2331 }, { "epoch": 0.31, "learning_rate": 6.884939195509823e-08, "loss": 0.0475, "step": 2332 }, { "epoch": 0.31, "learning_rate": 6.883602833088334e-08, "loss": 0.0602, "step": 2333 }, { "epoch": 0.31, "learning_rate": 6.882266470666844e-08, "loss": 0.0664, "step": 2334 }, { "epoch": 0.31, "learning_rate": 6.880930108245356e-08, "loss": 0.0658, "step": 2335 }, { "epoch": 0.31, "learning_rate": 6.879593745823867e-08, "loss": 0.0682, "step": 2336 }, { "epoch": 0.31, "learning_rate": 6.878257383402378e-08, "loss": 0.0642, "step": 2337 }, { "epoch": 0.31, "learning_rate": 6.876921020980889e-08, "loss": 0.078, "step": 2338 }, { "epoch": 0.31, "learning_rate": 6.875584658559401e-08, "loss": 0.0685, "step": 2339 }, { "epoch": 0.31, "learning_rate": 6.874248296137912e-08, "loss": 0.0482, "step": 2340 }, { "epoch": 0.31, "learning_rate": 6.872911933716423e-08, "loss": 0.0457, "step": 2341 }, { "epoch": 0.31, "learning_rate": 6.871575571294935e-08, "loss": 0.0253, "step": 2342 }, { "epoch": 0.31, "learning_rate": 6.870239208873446e-08, "loss": 0.043, "step": 2343 }, { "epoch": 0.31, "learning_rate": 6.868902846451957e-08, "loss": 0.0593, "step": 2344 }, { "epoch": 0.31, "learning_rate": 6.86756648403047e-08, "loss": 0.0443, "step": 2345 }, { "epoch": 0.31, "learning_rate": 6.866230121608981e-08, "loss": 0.0347, "step": 2346 }, { "epoch": 0.31, "learning_rate": 6.864893759187492e-08, "loss": 0.04, "step": 2347 }, { "epoch": 0.31, "learning_rate": 6.863557396766003e-08, "loss": 0.0418, "step": 2348 }, { "epoch": 0.31, "learning_rate": 6.862221034344514e-08, "loss": 0.0414, "step": 2349 }, { "epoch": 0.31, "learning_rate": 6.860884671923025e-08, "loss": 0.044, "step": 2350 }, { "epoch": 0.31, "learning_rate": 6.859548309501536e-08, "loss": 0.0298, "step": 2351 }, { "epoch": 0.31, "learning_rate": 6.858211947080048e-08, "loss": 0.039, "step": 2352 }, { "epoch": 0.31, "learning_rate": 6.856875584658559e-08, "loss": 0.0656, "step": 2353 }, { "epoch": 0.31, "learning_rate": 6.85553922223707e-08, "loss": 0.085, "step": 2354 }, { "epoch": 0.31, "learning_rate": 6.854202859815582e-08, "loss": 0.0413, "step": 2355 }, { "epoch": 0.31, "learning_rate": 6.852866497394093e-08, "loss": 0.072, "step": 2356 }, { "epoch": 0.31, "learning_rate": 6.851530134972604e-08, "loss": 0.0658, "step": 2357 }, { "epoch": 0.32, "learning_rate": 6.850193772551117e-08, "loss": 0.051, "step": 2358 }, { "epoch": 0.32, "learning_rate": 6.848857410129628e-08, "loss": 0.0302, "step": 2359 }, { "epoch": 0.32, "learning_rate": 6.847521047708138e-08, "loss": 0.056, "step": 2360 }, { "epoch": 0.32, "learning_rate": 6.84618468528665e-08, "loss": 0.0369, "step": 2361 }, { "epoch": 0.32, "learning_rate": 6.84484832286516e-08, "loss": 0.0554, "step": 2362 }, { "epoch": 0.32, "learning_rate": 6.843511960443671e-08, "loss": 0.0637, "step": 2363 }, { "epoch": 0.32, "learning_rate": 6.842175598022182e-08, "loss": 0.0306, "step": 2364 }, { "epoch": 0.32, "learning_rate": 6.840839235600695e-08, "loss": 0.0425, "step": 2365 }, { "epoch": 0.32, "learning_rate": 6.839502873179206e-08, "loss": 0.0533, "step": 2366 }, { "epoch": 0.32, "learning_rate": 6.838166510757717e-08, "loss": 0.0598, "step": 2367 }, { "epoch": 0.32, "learning_rate": 6.836830148336229e-08, "loss": 0.0628, "step": 2368 }, { "epoch": 0.32, "learning_rate": 6.83549378591474e-08, "loss": 0.0736, "step": 2369 }, { "epoch": 0.32, "learning_rate": 6.834157423493251e-08, "loss": 0.0561, "step": 2370 }, { "epoch": 0.32, "learning_rate": 6.832821061071762e-08, "loss": 0.0524, "step": 2371 }, { "epoch": 0.32, "learning_rate": 6.831484698650274e-08, "loss": 0.0517, "step": 2372 }, { "epoch": 0.32, "learning_rate": 6.830148336228785e-08, "loss": 0.0416, "step": 2373 }, { "epoch": 0.32, "learning_rate": 6.828811973807296e-08, "loss": 0.0452, "step": 2374 }, { "epoch": 0.32, "learning_rate": 6.827475611385807e-08, "loss": 0.0322, "step": 2375 }, { "epoch": 0.32, "learning_rate": 6.826139248964318e-08, "loss": 0.053, "step": 2376 }, { "epoch": 0.32, "learning_rate": 6.824802886542829e-08, "loss": 0.053, "step": 2377 }, { "epoch": 0.32, "learning_rate": 6.823466524121341e-08, "loss": 0.0508, "step": 2378 }, { "epoch": 0.32, "learning_rate": 6.822130161699852e-08, "loss": 0.0507, "step": 2379 }, { "epoch": 0.32, "learning_rate": 6.820793799278363e-08, "loss": 0.0578, "step": 2380 }, { "epoch": 0.32, "learning_rate": 6.819457436856876e-08, "loss": 0.0539, "step": 2381 }, { "epoch": 0.32, "learning_rate": 6.818121074435387e-08, "loss": 0.059, "step": 2382 }, { "epoch": 0.32, "learning_rate": 6.816784712013898e-08, "loss": 0.0749, "step": 2383 }, { "epoch": 0.32, "learning_rate": 6.815448349592409e-08, "loss": 0.0383, "step": 2384 }, { "epoch": 0.32, "learning_rate": 6.814111987170921e-08, "loss": 0.0651, "step": 2385 }, { "epoch": 0.32, "learning_rate": 6.812775624749432e-08, "loss": 0.046, "step": 2386 }, { "epoch": 0.32, "learning_rate": 6.811439262327943e-08, "loss": 0.0684, "step": 2387 }, { "epoch": 0.32, "learning_rate": 6.810102899906455e-08, "loss": 0.0438, "step": 2388 }, { "epoch": 0.32, "learning_rate": 6.808766537484966e-08, "loss": 0.0408, "step": 2389 }, { "epoch": 0.32, "learning_rate": 6.807430175063477e-08, "loss": 0.0365, "step": 2390 }, { "epoch": 0.32, "learning_rate": 6.806093812641988e-08, "loss": 0.0342, "step": 2391 }, { "epoch": 0.32, "learning_rate": 6.804757450220499e-08, "loss": 0.0468, "step": 2392 }, { "epoch": 0.32, "learning_rate": 6.80342108779901e-08, "loss": 0.0567, "step": 2393 }, { "epoch": 0.32, "learning_rate": 6.802084725377521e-08, "loss": 0.0802, "step": 2394 }, { "epoch": 0.32, "learning_rate": 6.800748362956033e-08, "loss": 0.0333, "step": 2395 }, { "epoch": 0.32, "learning_rate": 6.799412000534544e-08, "loss": 0.0632, "step": 2396 }, { "epoch": 0.32, "learning_rate": 6.798075638113055e-08, "loss": 0.0612, "step": 2397 }, { "epoch": 0.32, "learning_rate": 6.796739275691568e-08, "loss": 0.0499, "step": 2398 }, { "epoch": 0.32, "learning_rate": 6.795402913270079e-08, "loss": 0.0455, "step": 2399 }, { "epoch": 0.32, "learning_rate": 6.79406655084859e-08, "loss": 0.0285, "step": 2400 }, { "epoch": 0.32, "learning_rate": 6.792730188427102e-08, "loss": 0.0519, "step": 2401 }, { "epoch": 0.32, "learning_rate": 6.791393826005613e-08, "loss": 0.0484, "step": 2402 }, { "epoch": 0.32, "learning_rate": 6.790057463584124e-08, "loss": 0.0525, "step": 2403 }, { "epoch": 0.32, "learning_rate": 6.788721101162635e-08, "loss": 0.0675, "step": 2404 }, { "epoch": 0.32, "learning_rate": 6.787384738741146e-08, "loss": 0.0382, "step": 2405 }, { "epoch": 0.32, "learning_rate": 6.786048376319657e-08, "loss": 0.0667, "step": 2406 }, { "epoch": 0.32, "learning_rate": 6.784712013898168e-08, "loss": 0.0457, "step": 2407 }, { "epoch": 0.32, "learning_rate": 6.78337565147668e-08, "loss": 0.0363, "step": 2408 }, { "epoch": 0.32, "learning_rate": 6.782039289055191e-08, "loss": 0.0402, "step": 2409 }, { "epoch": 0.32, "learning_rate": 6.780702926633702e-08, "loss": 0.066, "step": 2410 }, { "epoch": 0.32, "learning_rate": 6.779366564212215e-08, "loss": 0.0376, "step": 2411 }, { "epoch": 0.32, "learning_rate": 6.778030201790725e-08, "loss": 0.0439, "step": 2412 }, { "epoch": 0.32, "learning_rate": 6.776693839369236e-08, "loss": 0.0496, "step": 2413 }, { "epoch": 0.32, "learning_rate": 6.775357476947749e-08, "loss": 0.0522, "step": 2414 }, { "epoch": 0.32, "learning_rate": 6.77402111452626e-08, "loss": 0.0388, "step": 2415 }, { "epoch": 0.32, "learning_rate": 6.772684752104771e-08, "loss": 0.0535, "step": 2416 }, { "epoch": 0.32, "learning_rate": 6.771348389683282e-08, "loss": 0.0431, "step": 2417 }, { "epoch": 0.32, "learning_rate": 6.770012027261793e-08, "loss": 0.0665, "step": 2418 }, { "epoch": 0.32, "learning_rate": 6.768675664840304e-08, "loss": 0.0561, "step": 2419 }, { "epoch": 0.32, "learning_rate": 6.767339302418815e-08, "loss": 0.0391, "step": 2420 }, { "epoch": 0.32, "learning_rate": 6.766002939997327e-08, "loss": 0.0449, "step": 2421 }, { "epoch": 0.32, "learning_rate": 6.764666577575838e-08, "loss": 0.0609, "step": 2422 }, { "epoch": 0.32, "learning_rate": 6.763330215154349e-08, "loss": 0.029, "step": 2423 }, { "epoch": 0.32, "learning_rate": 6.761993852732861e-08, "loss": 0.0384, "step": 2424 }, { "epoch": 0.32, "learning_rate": 6.760657490311372e-08, "loss": 0.0502, "step": 2425 }, { "epoch": 0.32, "learning_rate": 6.759321127889883e-08, "loss": 0.0422, "step": 2426 }, { "epoch": 0.32, "learning_rate": 6.757984765468396e-08, "loss": 0.0494, "step": 2427 }, { "epoch": 0.32, "learning_rate": 6.756648403046907e-08, "loss": 0.0453, "step": 2428 }, { "epoch": 0.32, "learning_rate": 6.755312040625418e-08, "loss": 0.0517, "step": 2429 }, { "epoch": 0.32, "learning_rate": 6.753975678203928e-08, "loss": 0.0486, "step": 2430 }, { "epoch": 0.32, "learning_rate": 6.752639315782441e-08, "loss": 0.0541, "step": 2431 }, { "epoch": 0.33, "learning_rate": 6.751302953360952e-08, "loss": 0.0639, "step": 2432 }, { "epoch": 0.33, "learning_rate": 6.749966590939461e-08, "loss": 0.0451, "step": 2433 }, { "epoch": 0.33, "learning_rate": 6.748630228517974e-08, "loss": 0.0534, "step": 2434 }, { "epoch": 0.33, "learning_rate": 6.747293866096485e-08, "loss": 0.0641, "step": 2435 }, { "epoch": 0.33, "learning_rate": 6.745957503674996e-08, "loss": 0.0368, "step": 2436 }, { "epoch": 0.33, "learning_rate": 6.744621141253508e-08, "loss": 0.0527, "step": 2437 }, { "epoch": 0.33, "learning_rate": 6.743284778832019e-08, "loss": 0.0507, "step": 2438 }, { "epoch": 0.33, "learning_rate": 6.74194841641053e-08, "loss": 0.026, "step": 2439 }, { "epoch": 0.33, "learning_rate": 6.740612053989041e-08, "loss": 0.0527, "step": 2440 }, { "epoch": 0.33, "learning_rate": 6.739275691567553e-08, "loss": 0.0315, "step": 2441 }, { "epoch": 0.33, "learning_rate": 6.737939329146064e-08, "loss": 0.0408, "step": 2442 }, { "epoch": 0.33, "learning_rate": 6.736602966724575e-08, "loss": 0.0293, "step": 2443 }, { "epoch": 0.33, "learning_rate": 6.735266604303088e-08, "loss": 0.0636, "step": 2444 }, { "epoch": 0.33, "learning_rate": 6.733930241881599e-08, "loss": 0.0423, "step": 2445 }, { "epoch": 0.33, "learning_rate": 6.73259387946011e-08, "loss": 0.0401, "step": 2446 }, { "epoch": 0.33, "learning_rate": 6.73125751703862e-08, "loss": 0.0604, "step": 2447 }, { "epoch": 0.33, "learning_rate": 6.729921154617131e-08, "loss": 0.0541, "step": 2448 }, { "epoch": 0.33, "learning_rate": 6.728584792195642e-08, "loss": 0.0444, "step": 2449 }, { "epoch": 0.33, "learning_rate": 6.727248429774155e-08, "loss": 0.0381, "step": 2450 }, { "epoch": 0.33, "learning_rate": 6.725912067352666e-08, "loss": 0.036, "step": 2451 }, { "epoch": 0.33, "learning_rate": 6.724575704931177e-08, "loss": 0.0506, "step": 2452 }, { "epoch": 0.33, "learning_rate": 6.723239342509688e-08, "loss": 0.0332, "step": 2453 }, { "epoch": 0.33, "learning_rate": 6.7219029800882e-08, "loss": 0.0588, "step": 2454 }, { "epoch": 0.33, "learning_rate": 6.720566617666711e-08, "loss": 0.0517, "step": 2455 }, { "epoch": 0.33, "learning_rate": 6.719230255245222e-08, "loss": 0.0538, "step": 2456 }, { "epoch": 0.33, "learning_rate": 6.717893892823734e-08, "loss": 0.0429, "step": 2457 }, { "epoch": 0.33, "learning_rate": 6.716557530402245e-08, "loss": 0.0335, "step": 2458 }, { "epoch": 0.33, "learning_rate": 6.715221167980756e-08, "loss": 0.0623, "step": 2459 }, { "epoch": 0.33, "learning_rate": 6.713884805559267e-08, "loss": 0.0417, "step": 2460 }, { "epoch": 0.33, "learning_rate": 6.712548443137778e-08, "loss": 0.0392, "step": 2461 }, { "epoch": 0.33, "learning_rate": 6.711212080716289e-08, "loss": 0.0428, "step": 2462 }, { "epoch": 0.33, "learning_rate": 6.7098757182948e-08, "loss": 0.0532, "step": 2463 }, { "epoch": 0.33, "learning_rate": 6.708539355873313e-08, "loss": 0.0728, "step": 2464 }, { "epoch": 0.33, "learning_rate": 6.707202993451823e-08, "loss": 0.0862, "step": 2465 }, { "epoch": 0.33, "learning_rate": 6.705866631030334e-08, "loss": 0.0465, "step": 2466 }, { "epoch": 0.33, "learning_rate": 6.704530268608847e-08, "loss": 0.0508, "step": 2467 }, { "epoch": 0.33, "learning_rate": 6.703193906187358e-08, "loss": 0.0452, "step": 2468 }, { "epoch": 0.33, "learning_rate": 6.701857543765869e-08, "loss": 0.0963, "step": 2469 }, { "epoch": 0.33, "learning_rate": 6.700521181344381e-08, "loss": 0.0426, "step": 2470 }, { "epoch": 0.33, "learning_rate": 6.699184818922892e-08, "loss": 0.065, "step": 2471 }, { "epoch": 0.33, "learning_rate": 6.697848456501403e-08, "loss": 0.0496, "step": 2472 }, { "epoch": 0.33, "learning_rate": 6.696512094079915e-08, "loss": 0.0423, "step": 2473 }, { "epoch": 0.33, "learning_rate": 6.695175731658425e-08, "loss": 0.0365, "step": 2474 }, { "epoch": 0.33, "learning_rate": 6.693839369236936e-08, "loss": 0.0449, "step": 2475 }, { "epoch": 0.33, "learning_rate": 6.692503006815447e-08, "loss": 0.0526, "step": 2476 }, { "epoch": 0.33, "learning_rate": 6.691166644393959e-08, "loss": 0.0598, "step": 2477 }, { "epoch": 0.33, "learning_rate": 6.68983028197247e-08, "loss": 0.0539, "step": 2478 }, { "epoch": 0.33, "learning_rate": 6.688493919550981e-08, "loss": 0.0414, "step": 2479 }, { "epoch": 0.33, "learning_rate": 6.687157557129494e-08, "loss": 0.0555, "step": 2480 }, { "epoch": 0.33, "learning_rate": 6.685821194708005e-08, "loss": 0.0763, "step": 2481 }, { "epoch": 0.33, "learning_rate": 6.684484832286516e-08, "loss": 0.0538, "step": 2482 }, { "epoch": 0.33, "learning_rate": 6.683148469865028e-08, "loss": 0.0371, "step": 2483 }, { "epoch": 0.33, "learning_rate": 6.681812107443539e-08, "loss": 0.0424, "step": 2484 }, { "epoch": 0.33, "learning_rate": 6.68047574502205e-08, "loss": 0.0363, "step": 2485 }, { "epoch": 0.33, "learning_rate": 6.679139382600561e-08, "loss": 0.0554, "step": 2486 }, { "epoch": 0.33, "learning_rate": 6.677803020179073e-08, "loss": 0.0517, "step": 2487 }, { "epoch": 0.33, "learning_rate": 6.676466657757584e-08, "loss": 0.0256, "step": 2488 }, { "epoch": 0.33, "learning_rate": 6.675130295336095e-08, "loss": 0.0675, "step": 2489 }, { "epoch": 0.33, "learning_rate": 6.673793932914606e-08, "loss": 0.0807, "step": 2490 }, { "epoch": 0.33, "learning_rate": 6.672457570493117e-08, "loss": 0.043, "step": 2491 }, { "epoch": 0.33, "learning_rate": 6.671121208071628e-08, "loss": 0.0268, "step": 2492 }, { "epoch": 0.33, "learning_rate": 6.66978484565014e-08, "loss": 0.0377, "step": 2493 }, { "epoch": 0.33, "learning_rate": 6.668448483228651e-08, "loss": 0.0503, "step": 2494 }, { "epoch": 0.33, "learning_rate": 6.667112120807162e-08, "loss": 0.0954, "step": 2495 }, { "epoch": 0.33, "learning_rate": 6.665775758385675e-08, "loss": 0.0543, "step": 2496 }, { "epoch": 0.33, "learning_rate": 6.664439395964186e-08, "loss": 0.0444, "step": 2497 }, { "epoch": 0.33, "learning_rate": 6.663103033542697e-08, "loss": 0.0516, "step": 2498 }, { "epoch": 0.33, "learning_rate": 6.661766671121208e-08, "loss": 0.0466, "step": 2499 }, { "epoch": 0.33, "learning_rate": 6.66043030869972e-08, "loss": 0.0428, "step": 2500 }, { "epoch": 0.33, "learning_rate": 6.659093946278231e-08, "loss": 0.0514, "step": 2501 }, { "epoch": 0.33, "learning_rate": 6.657757583856742e-08, "loss": 0.0548, "step": 2502 }, { "epoch": 0.33, "learning_rate": 6.656421221435253e-08, "loss": 0.0574, "step": 2503 }, { "epoch": 0.33, "learning_rate": 6.655084859013764e-08, "loss": 0.0919, "step": 2504 }, { "epoch": 0.33, "learning_rate": 6.653748496592275e-08, "loss": 0.0593, "step": 2505 }, { "epoch": 0.33, "learning_rate": 6.652412134170787e-08, "loss": 0.035, "step": 2506 }, { "epoch": 0.34, "learning_rate": 6.651075771749298e-08, "loss": 0.0399, "step": 2507 }, { "epoch": 0.34, "learning_rate": 6.649739409327809e-08, "loss": 0.0344, "step": 2508 }, { "epoch": 0.34, "learning_rate": 6.64840304690632e-08, "loss": 0.0529, "step": 2509 }, { "epoch": 0.34, "learning_rate": 6.647066684484832e-08, "loss": 0.0686, "step": 2510 }, { "epoch": 0.34, "learning_rate": 6.645730322063343e-08, "loss": 0.0531, "step": 2511 }, { "epoch": 0.34, "learning_rate": 6.644393959641854e-08, "loss": 0.0517, "step": 2512 }, { "epoch": 0.34, "learning_rate": 6.643057597220367e-08, "loss": 0.054, "step": 2513 }, { "epoch": 0.34, "learning_rate": 6.641721234798878e-08, "loss": 0.0404, "step": 2514 }, { "epoch": 0.34, "learning_rate": 6.640384872377389e-08, "loss": 0.1119, "step": 2515 }, { "epoch": 0.34, "learning_rate": 6.6390485099559e-08, "loss": 0.0419, "step": 2516 }, { "epoch": 0.34, "learning_rate": 6.63771214753441e-08, "loss": 0.0771, "step": 2517 }, { "epoch": 0.34, "learning_rate": 6.636375785112921e-08, "loss": 0.05, "step": 2518 }, { "epoch": 0.34, "learning_rate": 6.635039422691434e-08, "loss": 0.0465, "step": 2519 }, { "epoch": 0.34, "learning_rate": 6.633703060269945e-08, "loss": 0.0542, "step": 2520 }, { "epoch": 0.34, "learning_rate": 6.632366697848456e-08, "loss": 0.0665, "step": 2521 }, { "epoch": 0.34, "learning_rate": 6.631030335426967e-08, "loss": 0.027, "step": 2522 }, { "epoch": 0.34, "learning_rate": 6.629693973005479e-08, "loss": 0.0541, "step": 2523 }, { "epoch": 0.34, "learning_rate": 6.62835761058399e-08, "loss": 0.0586, "step": 2524 }, { "epoch": 0.34, "learning_rate": 6.627021248162501e-08, "loss": 0.0403, "step": 2525 }, { "epoch": 0.34, "learning_rate": 6.625684885741013e-08, "loss": 0.0416, "step": 2526 }, { "epoch": 0.34, "learning_rate": 6.624348523319524e-08, "loss": 0.0891, "step": 2527 }, { "epoch": 0.34, "learning_rate": 6.623012160898035e-08, "loss": 0.0395, "step": 2528 }, { "epoch": 0.34, "learning_rate": 6.621675798476548e-08, "loss": 0.0266, "step": 2529 }, { "epoch": 0.34, "learning_rate": 6.620339436055059e-08, "loss": 0.0454, "step": 2530 }, { "epoch": 0.34, "learning_rate": 6.619003073633568e-08, "loss": 0.0542, "step": 2531 }, { "epoch": 0.34, "learning_rate": 6.617666711212079e-08, "loss": 0.0659, "step": 2532 }, { "epoch": 0.34, "learning_rate": 6.616330348790592e-08, "loss": 0.0523, "step": 2533 }, { "epoch": 0.34, "learning_rate": 6.614993986369103e-08, "loss": 0.0606, "step": 2534 }, { "epoch": 0.34, "learning_rate": 6.613657623947614e-08, "loss": 0.0466, "step": 2535 }, { "epoch": 0.34, "learning_rate": 6.612321261526126e-08, "loss": 0.0576, "step": 2536 }, { "epoch": 0.34, "learning_rate": 6.610984899104637e-08, "loss": 0.0277, "step": 2537 }, { "epoch": 0.34, "learning_rate": 6.609648536683148e-08, "loss": 0.0497, "step": 2538 }, { "epoch": 0.34, "learning_rate": 6.60831217426166e-08, "loss": 0.044, "step": 2539 }, { "epoch": 0.34, "learning_rate": 6.606975811840171e-08, "loss": 0.0618, "step": 2540 }, { "epoch": 0.34, "learning_rate": 6.605639449418682e-08, "loss": 0.041, "step": 2541 }, { "epoch": 0.34, "learning_rate": 6.604303086997194e-08, "loss": 0.0488, "step": 2542 }, { "epoch": 0.34, "learning_rate": 6.602966724575705e-08, "loss": 0.0383, "step": 2543 }, { "epoch": 0.34, "learning_rate": 6.601630362154216e-08, "loss": 0.0621, "step": 2544 }, { "epoch": 0.34, "learning_rate": 6.600293999732727e-08, "loss": 0.0354, "step": 2545 }, { "epoch": 0.34, "learning_rate": 6.598957637311238e-08, "loss": 0.0477, "step": 2546 }, { "epoch": 0.34, "learning_rate": 6.597621274889749e-08, "loss": 0.0448, "step": 2547 }, { "epoch": 0.34, "learning_rate": 6.59628491246826e-08, "loss": 0.0422, "step": 2548 }, { "epoch": 0.34, "learning_rate": 6.594948550046773e-08, "loss": 0.057, "step": 2549 }, { "epoch": 0.34, "learning_rate": 6.593612187625284e-08, "loss": 0.0266, "step": 2550 }, { "epoch": 0.34, "learning_rate": 6.592275825203795e-08, "loss": 0.0539, "step": 2551 }, { "epoch": 0.34, "learning_rate": 6.590939462782307e-08, "loss": 0.0403, "step": 2552 }, { "epoch": 0.34, "learning_rate": 6.589603100360818e-08, "loss": 0.0494, "step": 2553 }, { "epoch": 0.34, "learning_rate": 6.588266737939329e-08, "loss": 0.0698, "step": 2554 }, { "epoch": 0.34, "learning_rate": 6.58693037551784e-08, "loss": 0.0553, "step": 2555 }, { "epoch": 0.34, "learning_rate": 6.585594013096352e-08, "loss": 0.0575, "step": 2556 }, { "epoch": 0.34, "learning_rate": 6.584257650674863e-08, "loss": 0.0456, "step": 2557 }, { "epoch": 0.34, "learning_rate": 6.582921288253374e-08, "loss": 0.0469, "step": 2558 }, { "epoch": 0.34, "learning_rate": 6.581584925831885e-08, "loss": 0.0669, "step": 2559 }, { "epoch": 0.34, "learning_rate": 6.580248563410396e-08, "loss": 0.0704, "step": 2560 }, { "epoch": 0.34, "learning_rate": 6.578912200988907e-08, "loss": 0.0607, "step": 2561 }, { "epoch": 0.34, "learning_rate": 6.577575838567419e-08, "loss": 0.0463, "step": 2562 }, { "epoch": 0.34, "learning_rate": 6.57623947614593e-08, "loss": 0.0657, "step": 2563 }, { "epoch": 0.34, "learning_rate": 6.574903113724441e-08, "loss": 0.0596, "step": 2564 }, { "epoch": 0.34, "learning_rate": 6.573566751302954e-08, "loss": 0.0436, "step": 2565 }, { "epoch": 0.34, "learning_rate": 6.572230388881465e-08, "loss": 0.0509, "step": 2566 }, { "epoch": 0.34, "learning_rate": 6.570894026459976e-08, "loss": 0.0952, "step": 2567 }, { "epoch": 0.34, "learning_rate": 6.569557664038487e-08, "loss": 0.0541, "step": 2568 }, { "epoch": 0.34, "learning_rate": 6.568221301616999e-08, "loss": 0.0484, "step": 2569 }, { "epoch": 0.34, "learning_rate": 6.56688493919551e-08, "loss": 0.0396, "step": 2570 }, { "epoch": 0.34, "learning_rate": 6.565548576774021e-08, "loss": 0.0502, "step": 2571 }, { "epoch": 0.34, "learning_rate": 6.564212214352533e-08, "loss": 0.0542, "step": 2572 }, { "epoch": 0.34, "learning_rate": 6.562875851931043e-08, "loss": 0.039, "step": 2573 }, { "epoch": 0.34, "learning_rate": 6.561539489509554e-08, "loss": 0.0574, "step": 2574 }, { "epoch": 0.34, "learning_rate": 6.560203127088066e-08, "loss": 0.0506, "step": 2575 }, { "epoch": 0.34, "learning_rate": 6.558866764666577e-08, "loss": 0.0375, "step": 2576 }, { "epoch": 0.34, "learning_rate": 6.557530402245088e-08, "loss": 0.0527, "step": 2577 }, { "epoch": 0.34, "learning_rate": 6.556194039823599e-08, "loss": 0.0659, "step": 2578 }, { "epoch": 0.34, "learning_rate": 6.554857677402111e-08, "loss": 0.054, "step": 2579 }, { "epoch": 0.34, "learning_rate": 6.553521314980622e-08, "loss": 0.0474, "step": 2580 }, { "epoch": 0.34, "learning_rate": 6.552184952559133e-08, "loss": 0.0577, "step": 2581 }, { "epoch": 0.35, "learning_rate": 6.550848590137646e-08, "loss": 0.0529, "step": 2582 }, { "epoch": 0.35, "learning_rate": 6.549512227716157e-08, "loss": 0.0494, "step": 2583 }, { "epoch": 0.35, "learning_rate": 6.548175865294668e-08, "loss": 0.0413, "step": 2584 }, { "epoch": 0.35, "learning_rate": 6.54683950287318e-08, "loss": 0.0616, "step": 2585 }, { "epoch": 0.35, "learning_rate": 6.545503140451691e-08, "loss": 0.0526, "step": 2586 }, { "epoch": 0.35, "learning_rate": 6.544166778030202e-08, "loss": 0.049, "step": 2587 }, { "epoch": 0.35, "learning_rate": 6.542830415608713e-08, "loss": 0.0359, "step": 2588 }, { "epoch": 0.35, "learning_rate": 6.541494053187224e-08, "loss": 0.0473, "step": 2589 }, { "epoch": 0.35, "learning_rate": 6.540157690765735e-08, "loss": 0.0471, "step": 2590 }, { "epoch": 0.35, "learning_rate": 6.538821328344246e-08, "loss": 0.0555, "step": 2591 }, { "epoch": 0.35, "learning_rate": 6.537484965922758e-08, "loss": 0.069, "step": 2592 }, { "epoch": 0.35, "learning_rate": 6.536148603501269e-08, "loss": 0.0656, "step": 2593 }, { "epoch": 0.35, "learning_rate": 6.53481224107978e-08, "loss": 0.0608, "step": 2594 }, { "epoch": 0.35, "learning_rate": 6.533475878658292e-08, "loss": 0.0515, "step": 2595 }, { "epoch": 0.35, "learning_rate": 6.532139516236803e-08, "loss": 0.0324, "step": 2596 }, { "epoch": 0.35, "learning_rate": 6.530803153815314e-08, "loss": 0.0501, "step": 2597 }, { "epoch": 0.35, "learning_rate": 6.529466791393827e-08, "loss": 0.0483, "step": 2598 }, { "epoch": 0.35, "learning_rate": 6.528130428972338e-08, "loss": 0.0324, "step": 2599 }, { "epoch": 0.35, "learning_rate": 6.526794066550849e-08, "loss": 0.0337, "step": 2600 }, { "epoch": 0.35, "learning_rate": 6.52545770412936e-08, "loss": 0.0475, "step": 2601 }, { "epoch": 0.35, "learning_rate": 6.52412134170787e-08, "loss": 0.0379, "step": 2602 }, { "epoch": 0.35, "learning_rate": 6.522784979286382e-08, "loss": 0.042, "step": 2603 }, { "epoch": 0.35, "learning_rate": 6.521448616864893e-08, "loss": 0.0432, "step": 2604 }, { "epoch": 0.35, "learning_rate": 6.520112254443405e-08, "loss": 0.061, "step": 2605 }, { "epoch": 0.35, "learning_rate": 6.518775892021916e-08, "loss": 0.0672, "step": 2606 }, { "epoch": 0.35, "learning_rate": 6.517439529600427e-08, "loss": 0.0556, "step": 2607 }, { "epoch": 0.35, "learning_rate": 6.516103167178939e-08, "loss": 0.0438, "step": 2608 }, { "epoch": 0.35, "learning_rate": 6.51476680475745e-08, "loss": 0.0388, "step": 2609 }, { "epoch": 0.35, "learning_rate": 6.513430442335961e-08, "loss": 0.0504, "step": 2610 }, { "epoch": 0.35, "learning_rate": 6.512094079914473e-08, "loss": 0.0507, "step": 2611 }, { "epoch": 0.35, "learning_rate": 6.510757717492984e-08, "loss": 0.0543, "step": 2612 }, { "epoch": 0.35, "learning_rate": 6.509421355071495e-08, "loss": 0.0588, "step": 2613 }, { "epoch": 0.35, "learning_rate": 6.508084992650006e-08, "loss": 0.0609, "step": 2614 }, { "epoch": 0.35, "learning_rate": 6.506748630228517e-08, "loss": 0.0543, "step": 2615 }, { "epoch": 0.35, "learning_rate": 6.505412267807028e-08, "loss": 0.0306, "step": 2616 }, { "epoch": 0.35, "learning_rate": 6.504075905385539e-08, "loss": 0.053, "step": 2617 }, { "epoch": 0.35, "learning_rate": 6.504075905385539e-08, "loss": 0.0688, "step": 2618 }, { "epoch": 0.35, "learning_rate": 6.502739542964052e-08, "loss": 0.0431, "step": 2619 }, { "epoch": 0.35, "learning_rate": 6.501403180542563e-08, "loss": 0.046, "step": 2620 }, { "epoch": 0.35, "learning_rate": 6.500066818121074e-08, "loss": 0.0419, "step": 2621 }, { "epoch": 0.35, "learning_rate": 6.498730455699586e-08, "loss": 0.0509, "step": 2622 }, { "epoch": 0.35, "learning_rate": 6.497394093278097e-08, "loss": 0.0665, "step": 2623 }, { "epoch": 0.35, "learning_rate": 6.496057730856608e-08, "loss": 0.0475, "step": 2624 }, { "epoch": 0.35, "learning_rate": 6.49472136843512e-08, "loss": 0.0544, "step": 2625 }, { "epoch": 0.35, "learning_rate": 6.493385006013631e-08, "loss": 0.0489, "step": 2626 }, { "epoch": 0.35, "learning_rate": 6.492048643592142e-08, "loss": 0.0325, "step": 2627 }, { "epoch": 0.35, "learning_rate": 6.490712281170653e-08, "loss": 0.0636, "step": 2628 }, { "epoch": 0.35, "learning_rate": 6.489375918749165e-08, "loss": 0.0608, "step": 2629 }, { "epoch": 0.35, "learning_rate": 6.488039556327676e-08, "loss": 0.0461, "step": 2630 }, { "epoch": 0.35, "learning_rate": 6.486703193906186e-08, "loss": 0.0395, "step": 2631 }, { "epoch": 0.35, "learning_rate": 6.485366831484698e-08, "loss": 0.0906, "step": 2632 }, { "epoch": 0.35, "learning_rate": 6.484030469063209e-08, "loss": 0.042, "step": 2633 }, { "epoch": 0.35, "learning_rate": 6.48269410664172e-08, "loss": 0.0435, "step": 2634 }, { "epoch": 0.35, "learning_rate": 6.481357744220233e-08, "loss": 0.0593, "step": 2635 }, { "epoch": 0.35, "learning_rate": 6.480021381798744e-08, "loss": 0.0715, "step": 2636 }, { "epoch": 0.35, "learning_rate": 6.478685019377255e-08, "loss": 0.0464, "step": 2637 }, { "epoch": 0.35, "learning_rate": 6.477348656955766e-08, "loss": 0.066, "step": 2638 }, { "epoch": 0.35, "learning_rate": 6.476012294534278e-08, "loss": 0.0537, "step": 2639 }, { "epoch": 0.35, "learning_rate": 6.474675932112789e-08, "loss": 0.0553, "step": 2640 }, { "epoch": 0.35, "learning_rate": 6.4733395696913e-08, "loss": 0.0398, "step": 2641 }, { "epoch": 0.35, "learning_rate": 6.472003207269812e-08, "loss": 0.0439, "step": 2642 }, { "epoch": 0.35, "learning_rate": 6.470666844848323e-08, "loss": 0.0737, "step": 2643 }, { "epoch": 0.35, "learning_rate": 6.469330482426834e-08, "loss": 0.0606, "step": 2644 }, { "epoch": 0.35, "learning_rate": 6.467994120005345e-08, "loss": 0.0474, "step": 2645 }, { "epoch": 0.35, "learning_rate": 6.466657757583856e-08, "loss": 0.047, "step": 2646 }, { "epoch": 0.35, "learning_rate": 6.465321395162367e-08, "loss": 0.047, "step": 2647 }, { "epoch": 0.35, "learning_rate": 6.46398503274088e-08, "loss": 0.035, "step": 2648 }, { "epoch": 0.35, "learning_rate": 6.46264867031939e-08, "loss": 0.0543, "step": 2649 }, { "epoch": 0.35, "learning_rate": 6.461312307897901e-08, "loss": 0.0615, "step": 2650 }, { "epoch": 0.35, "learning_rate": 6.459975945476412e-08, "loss": 0.0512, "step": 2651 }, { "epoch": 0.35, "learning_rate": 6.458639583054925e-08, "loss": 0.0568, "step": 2652 }, { "epoch": 0.35, "learning_rate": 6.457303220633436e-08, "loss": 0.039, "step": 2653 }, { "epoch": 0.35, "learning_rate": 6.455966858211947e-08, "loss": 0.044, "step": 2654 }, { "epoch": 0.35, "learning_rate": 6.454630495790459e-08, "loss": 0.0776, "step": 2655 }, { "epoch": 0.35, "learning_rate": 6.45329413336897e-08, "loss": 0.0663, "step": 2656 }, { "epoch": 0.36, "learning_rate": 6.451957770947481e-08, "loss": 0.0411, "step": 2657 }, { "epoch": 0.36, "learning_rate": 6.450621408525992e-08, "loss": 0.0289, "step": 2658 }, { "epoch": 0.36, "learning_rate": 6.449285046104503e-08, "loss": 0.0482, "step": 2659 }, { "epoch": 0.36, "learning_rate": 6.447948683683014e-08, "loss": 0.0629, "step": 2660 }, { "epoch": 0.36, "learning_rate": 6.446612321261525e-08, "loss": 0.0806, "step": 2661 }, { "epoch": 0.36, "learning_rate": 6.445275958840037e-08, "loss": 0.0431, "step": 2662 }, { "epoch": 0.36, "learning_rate": 6.443939596418548e-08, "loss": 0.0452, "step": 2663 }, { "epoch": 0.36, "learning_rate": 6.442603233997059e-08, "loss": 0.0453, "step": 2664 }, { "epoch": 0.36, "learning_rate": 6.441266871575571e-08, "loss": 0.0451, "step": 2665 }, { "epoch": 0.36, "learning_rate": 6.439930509154082e-08, "loss": 0.0389, "step": 2666 }, { "epoch": 0.36, "learning_rate": 6.438594146732593e-08, "loss": 0.0692, "step": 2667 }, { "epoch": 0.36, "learning_rate": 6.437257784311106e-08, "loss": 0.0648, "step": 2668 }, { "epoch": 0.36, "learning_rate": 6.435921421889617e-08, "loss": 0.0379, "step": 2669 }, { "epoch": 0.36, "learning_rate": 6.434585059468128e-08, "loss": 0.0411, "step": 2670 }, { "epoch": 0.36, "learning_rate": 6.43324869704664e-08, "loss": 0.048, "step": 2671 }, { "epoch": 0.36, "learning_rate": 6.431912334625151e-08, "loss": 0.0365, "step": 2672 }, { "epoch": 0.36, "learning_rate": 6.43057597220366e-08, "loss": 0.0476, "step": 2673 }, { "epoch": 0.36, "learning_rate": 6.429239609782172e-08, "loss": 0.047, "step": 2674 }, { "epoch": 0.36, "learning_rate": 6.427903247360684e-08, "loss": 0.072, "step": 2675 }, { "epoch": 0.36, "learning_rate": 6.426566884939195e-08, "loss": 0.0582, "step": 2676 }, { "epoch": 0.36, "learning_rate": 6.425230522517706e-08, "loss": 0.0509, "step": 2677 }, { "epoch": 0.36, "learning_rate": 6.423894160096218e-08, "loss": 0.0511, "step": 2678 }, { "epoch": 0.36, "learning_rate": 6.422557797674729e-08, "loss": 0.0575, "step": 2679 }, { "epoch": 0.36, "learning_rate": 6.42122143525324e-08, "loss": 0.0565, "step": 2680 }, { "epoch": 0.36, "learning_rate": 6.419885072831752e-08, "loss": 0.0458, "step": 2681 }, { "epoch": 0.36, "learning_rate": 6.418548710410263e-08, "loss": 0.0445, "step": 2682 }, { "epoch": 0.36, "learning_rate": 6.417212347988774e-08, "loss": 0.0642, "step": 2683 }, { "epoch": 0.36, "learning_rate": 6.415875985567285e-08, "loss": 0.0527, "step": 2684 }, { "epoch": 0.36, "learning_rate": 6.414539623145798e-08, "loss": 0.0522, "step": 2685 }, { "epoch": 0.36, "learning_rate": 6.413203260724309e-08, "loss": 0.0562, "step": 2686 }, { "epoch": 0.36, "learning_rate": 6.41186689830282e-08, "loss": 0.0588, "step": 2687 }, { "epoch": 0.36, "learning_rate": 6.41053053588133e-08, "loss": 0.0751, "step": 2688 }, { "epoch": 0.36, "learning_rate": 6.409194173459842e-08, "loss": 0.0427, "step": 2689 }, { "epoch": 0.36, "learning_rate": 6.407857811038353e-08, "loss": 0.0529, "step": 2690 }, { "epoch": 0.36, "learning_rate": 6.406521448616865e-08, "loss": 0.036, "step": 2691 }, { "epoch": 0.36, "learning_rate": 6.405185086195376e-08, "loss": 0.0373, "step": 2692 }, { "epoch": 0.36, "learning_rate": 6.403848723773887e-08, "loss": 0.0565, "step": 2693 }, { "epoch": 0.36, "learning_rate": 6.402512361352399e-08, "loss": 0.0525, "step": 2694 }, { "epoch": 0.36, "learning_rate": 6.40117599893091e-08, "loss": 0.0413, "step": 2695 }, { "epoch": 0.36, "learning_rate": 6.399839636509421e-08, "loss": 0.048, "step": 2696 }, { "epoch": 0.36, "learning_rate": 6.398503274087932e-08, "loss": 0.0502, "step": 2697 }, { "epoch": 0.36, "learning_rate": 6.397166911666444e-08, "loss": 0.0631, "step": 2698 }, { "epoch": 0.36, "learning_rate": 6.395830549244955e-08, "loss": 0.0593, "step": 2699 }, { "epoch": 0.36, "learning_rate": 6.394494186823466e-08, "loss": 0.099, "step": 2700 }, { "epoch": 0.36, "learning_rate": 6.393157824401977e-08, "loss": 0.0543, "step": 2701 }, { "epoch": 0.36, "learning_rate": 6.391821461980488e-08, "loss": 0.0398, "step": 2702 }, { "epoch": 0.36, "learning_rate": 6.390485099559e-08, "loss": 0.035, "step": 2703 }, { "epoch": 0.36, "learning_rate": 6.389148737137512e-08, "loss": 0.044, "step": 2704 }, { "epoch": 0.36, "learning_rate": 6.387812374716023e-08, "loss": 0.0458, "step": 2705 }, { "epoch": 0.36, "learning_rate": 6.386476012294534e-08, "loss": 0.0632, "step": 2706 }, { "epoch": 0.36, "learning_rate": 6.385139649873045e-08, "loss": 0.0506, "step": 2707 }, { "epoch": 0.36, "learning_rate": 6.383803287451557e-08, "loss": 0.0488, "step": 2708 }, { "epoch": 0.36, "learning_rate": 6.382466925030068e-08, "loss": 0.0677, "step": 2709 }, { "epoch": 0.36, "learning_rate": 6.381130562608579e-08, "loss": 0.0436, "step": 2710 }, { "epoch": 0.36, "learning_rate": 6.379794200187091e-08, "loss": 0.0412, "step": 2711 }, { "epoch": 0.36, "learning_rate": 6.378457837765602e-08, "loss": 0.0553, "step": 2712 }, { "epoch": 0.36, "learning_rate": 6.377121475344113e-08, "loss": 0.0524, "step": 2713 }, { "epoch": 0.36, "learning_rate": 6.375785112922624e-08, "loss": 0.067, "step": 2714 }, { "epoch": 0.36, "learning_rate": 6.374448750501135e-08, "loss": 0.0413, "step": 2715 }, { "epoch": 0.36, "learning_rate": 6.373112388079646e-08, "loss": 0.0415, "step": 2716 }, { "epoch": 0.36, "learning_rate": 6.371776025658158e-08, "loss": 0.0475, "step": 2717 }, { "epoch": 0.36, "learning_rate": 6.37043966323667e-08, "loss": 0.0484, "step": 2718 }, { "epoch": 0.36, "learning_rate": 6.36910330081518e-08, "loss": 0.0632, "step": 2719 }, { "epoch": 0.36, "learning_rate": 6.367766938393691e-08, "loss": 0.0432, "step": 2720 }, { "epoch": 0.36, "learning_rate": 6.366430575972204e-08, "loss": 0.0499, "step": 2721 }, { "epoch": 0.36, "learning_rate": 6.365094213550715e-08, "loss": 0.06, "step": 2722 }, { "epoch": 0.36, "learning_rate": 6.363757851129226e-08, "loss": 0.0453, "step": 2723 }, { "epoch": 0.36, "learning_rate": 6.362421488707738e-08, "loss": 0.077, "step": 2724 }, { "epoch": 0.36, "learning_rate": 6.361085126286249e-08, "loss": 0.064, "step": 2725 }, { "epoch": 0.36, "learning_rate": 6.35974876386476e-08, "loss": 0.0507, "step": 2726 }, { "epoch": 0.36, "learning_rate": 6.358412401443272e-08, "loss": 0.0589, "step": 2727 }, { "epoch": 0.36, "learning_rate": 6.357076039021783e-08, "loss": 0.0812, "step": 2728 }, { "epoch": 0.36, "learning_rate": 6.355739676600294e-08, "loss": 0.0611, "step": 2729 }, { "epoch": 0.36, "learning_rate": 6.354403314178804e-08, "loss": 0.0538, "step": 2730 }, { "epoch": 0.36, "learning_rate": 6.353066951757316e-08, "loss": 0.0443, "step": 2731 }, { "epoch": 0.37, "learning_rate": 6.351730589335827e-08, "loss": 0.0975, "step": 2732 }, { "epoch": 0.37, "learning_rate": 6.350394226914338e-08, "loss": 0.0594, "step": 2733 }, { "epoch": 0.37, "learning_rate": 6.34905786449285e-08, "loss": 0.0463, "step": 2734 }, { "epoch": 0.37, "learning_rate": 6.347721502071361e-08, "loss": 0.0657, "step": 2735 }, { "epoch": 0.37, "learning_rate": 6.346385139649872e-08, "loss": 0.0629, "step": 2736 }, { "epoch": 0.37, "learning_rate": 6.345048777228385e-08, "loss": 0.0544, "step": 2737 }, { "epoch": 0.37, "learning_rate": 6.343712414806896e-08, "loss": 0.0484, "step": 2738 }, { "epoch": 0.37, "learning_rate": 6.342376052385407e-08, "loss": 0.0478, "step": 2739 }, { "epoch": 0.37, "learning_rate": 6.341039689963919e-08, "loss": 0.0678, "step": 2740 }, { "epoch": 0.37, "learning_rate": 6.33970332754243e-08, "loss": 0.0276, "step": 2741 }, { "epoch": 0.37, "learning_rate": 6.338366965120941e-08, "loss": 0.0639, "step": 2742 }, { "epoch": 0.37, "learning_rate": 6.337030602699452e-08, "loss": 0.0586, "step": 2743 }, { "epoch": 0.37, "learning_rate": 6.335694240277963e-08, "loss": 0.0468, "step": 2744 }, { "epoch": 0.37, "learning_rate": 6.334357877856474e-08, "loss": 0.0641, "step": 2745 }, { "epoch": 0.37, "learning_rate": 6.333021515434985e-08, "loss": 0.0378, "step": 2746 }, { "epoch": 0.37, "learning_rate": 6.331685153013497e-08, "loss": 0.0497, "step": 2747 }, { "epoch": 0.37, "learning_rate": 6.330348790592008e-08, "loss": 0.0387, "step": 2748 }, { "epoch": 0.37, "learning_rate": 6.329012428170519e-08, "loss": 0.0458, "step": 2749 }, { "epoch": 0.37, "learning_rate": 6.327676065749031e-08, "loss": 0.0727, "step": 2750 }, { "epoch": 0.37, "learning_rate": 6.326339703327542e-08, "loss": 0.0497, "step": 2751 }, { "epoch": 0.37, "learning_rate": 6.325003340906053e-08, "loss": 0.0657, "step": 2752 }, { "epoch": 0.37, "learning_rate": 6.323666978484564e-08, "loss": 0.063, "step": 2753 }, { "epoch": 0.37, "learning_rate": 6.322330616063077e-08, "loss": 0.0397, "step": 2754 }, { "epoch": 0.37, "learning_rate": 6.320994253641588e-08, "loss": 0.0665, "step": 2755 }, { "epoch": 0.37, "learning_rate": 6.319657891220099e-08, "loss": 0.0334, "step": 2756 }, { "epoch": 0.37, "learning_rate": 6.31832152879861e-08, "loss": 0.0511, "step": 2757 }, { "epoch": 0.37, "learning_rate": 6.31698516637712e-08, "loss": 0.0399, "step": 2758 }, { "epoch": 0.37, "learning_rate": 6.315648803955632e-08, "loss": 0.0274, "step": 2759 }, { "epoch": 0.37, "learning_rate": 6.314312441534144e-08, "loss": 0.0445, "step": 2760 }, { "epoch": 0.37, "learning_rate": 6.312976079112655e-08, "loss": 0.0755, "step": 2761 }, { "epoch": 0.37, "learning_rate": 6.311639716691166e-08, "loss": 0.0472, "step": 2762 }, { "epoch": 0.37, "learning_rate": 6.310303354269678e-08, "loss": 0.0802, "step": 2763 }, { "epoch": 0.37, "learning_rate": 6.308966991848189e-08, "loss": 0.075, "step": 2764 }, { "epoch": 0.37, "learning_rate": 6.3076306294267e-08, "loss": 0.0512, "step": 2765 }, { "epoch": 0.37, "learning_rate": 6.306294267005211e-08, "loss": 0.0658, "step": 2766 }, { "epoch": 0.37, "learning_rate": 6.304957904583723e-08, "loss": 0.0437, "step": 2767 }, { "epoch": 0.37, "learning_rate": 6.303621542162234e-08, "loss": 0.0337, "step": 2768 }, { "epoch": 0.37, "learning_rate": 6.302285179740745e-08, "loss": 0.0339, "step": 2769 }, { "epoch": 0.37, "learning_rate": 6.300948817319258e-08, "loss": 0.0355, "step": 2770 }, { "epoch": 0.37, "learning_rate": 6.299612454897769e-08, "loss": 0.0578, "step": 2771 }, { "epoch": 0.37, "learning_rate": 6.298276092476278e-08, "loss": 0.0927, "step": 2772 }, { "epoch": 0.37, "learning_rate": 6.296939730054791e-08, "loss": 0.0398, "step": 2773 }, { "epoch": 0.37, "learning_rate": 6.295603367633302e-08, "loss": 0.0388, "step": 2774 }, { "epoch": 0.37, "learning_rate": 6.294267005211813e-08, "loss": 0.0512, "step": 2775 }, { "epoch": 0.37, "learning_rate": 6.292930642790324e-08, "loss": 0.0458, "step": 2776 }, { "epoch": 0.37, "learning_rate": 6.291594280368836e-08, "loss": 0.0418, "step": 2777 }, { "epoch": 0.37, "learning_rate": 6.290257917947347e-08, "loss": 0.0341, "step": 2778 }, { "epoch": 0.37, "learning_rate": 6.288921555525858e-08, "loss": 0.0484, "step": 2779 }, { "epoch": 0.37, "learning_rate": 6.28758519310437e-08, "loss": 0.0389, "step": 2780 }, { "epoch": 0.37, "learning_rate": 6.286248830682881e-08, "loss": 0.0409, "step": 2781 }, { "epoch": 0.37, "learning_rate": 6.284912468261392e-08, "loss": 0.0628, "step": 2782 }, { "epoch": 0.37, "learning_rate": 6.283576105839904e-08, "loss": 0.0403, "step": 2783 }, { "epoch": 0.37, "learning_rate": 6.282239743418415e-08, "loss": 0.0439, "step": 2784 }, { "epoch": 0.37, "learning_rate": 6.280903380996926e-08, "loss": 0.0675, "step": 2785 }, { "epoch": 0.37, "learning_rate": 6.279567018575437e-08, "loss": 0.0536, "step": 2786 }, { "epoch": 0.37, "learning_rate": 6.278230656153948e-08, "loss": 0.0298, "step": 2787 }, { "epoch": 0.37, "learning_rate": 6.27689429373246e-08, "loss": 0.0584, "step": 2788 }, { "epoch": 0.37, "learning_rate": 6.27555793131097e-08, "loss": 0.059, "step": 2789 }, { "epoch": 0.37, "learning_rate": 6.274221568889483e-08, "loss": 0.0418, "step": 2790 }, { "epoch": 0.37, "learning_rate": 6.272885206467994e-08, "loss": 0.0355, "step": 2791 }, { "epoch": 0.37, "learning_rate": 6.271548844046505e-08, "loss": 0.0428, "step": 2792 }, { "epoch": 0.37, "learning_rate": 6.270212481625017e-08, "loss": 0.0575, "step": 2793 }, { "epoch": 0.37, "learning_rate": 6.268876119203528e-08, "loss": 0.0405, "step": 2794 }, { "epoch": 0.37, "learning_rate": 6.267539756782039e-08, "loss": 0.046, "step": 2795 }, { "epoch": 0.37, "learning_rate": 6.266203394360551e-08, "loss": 0.0375, "step": 2796 }, { "epoch": 0.37, "learning_rate": 6.264867031939062e-08, "loss": 0.0546, "step": 2797 }, { "epoch": 0.37, "learning_rate": 6.263530669517573e-08, "loss": 0.0433, "step": 2798 }, { "epoch": 0.37, "learning_rate": 6.262194307096084e-08, "loss": 0.0814, "step": 2799 }, { "epoch": 0.37, "learning_rate": 6.260857944674595e-08, "loss": 0.042, "step": 2800 }, { "epoch": 0.37, "learning_rate": 6.259521582253106e-08, "loss": 0.0509, "step": 2801 }, { "epoch": 0.37, "learning_rate": 6.258185219831617e-08, "loss": 0.0599, "step": 2802 }, { "epoch": 0.37, "learning_rate": 6.25684885741013e-08, "loss": 0.0436, "step": 2803 }, { "epoch": 0.37, "learning_rate": 6.25551249498864e-08, "loss": 0.0534, "step": 2804 }, { "epoch": 0.37, "learning_rate": 6.254176132567151e-08, "loss": 0.0482, "step": 2805 }, { "epoch": 0.37, "learning_rate": 6.252839770145664e-08, "loss": 0.0365, "step": 2806 }, { "epoch": 0.38, "learning_rate": 6.251503407724175e-08, "loss": 0.0429, "step": 2807 }, { "epoch": 0.38, "learning_rate": 6.250167045302686e-08, "loss": 0.0467, "step": 2808 }, { "epoch": 0.38, "learning_rate": 6.248830682881198e-08, "loss": 0.0521, "step": 2809 }, { "epoch": 0.38, "learning_rate": 6.247494320459709e-08, "loss": 0.0513, "step": 2810 }, { "epoch": 0.38, "learning_rate": 6.24615795803822e-08, "loss": 0.0557, "step": 2811 }, { "epoch": 0.38, "learning_rate": 6.244821595616731e-08, "loss": 0.0526, "step": 2812 }, { "epoch": 0.38, "learning_rate": 6.243485233195242e-08, "loss": 0.0445, "step": 2813 }, { "epoch": 0.38, "learning_rate": 6.242148870773753e-08, "loss": 0.0401, "step": 2814 }, { "epoch": 0.38, "learning_rate": 6.240812508352264e-08, "loss": 0.0535, "step": 2815 }, { "epoch": 0.38, "learning_rate": 6.239476145930776e-08, "loss": 0.0702, "step": 2816 }, { "epoch": 0.38, "learning_rate": 6.238139783509287e-08, "loss": 0.0574, "step": 2817 }, { "epoch": 0.38, "learning_rate": 6.236803421087798e-08, "loss": 0.0733, "step": 2818 }, { "epoch": 0.38, "learning_rate": 6.23546705866631e-08, "loss": 0.055, "step": 2819 }, { "epoch": 0.38, "learning_rate": 6.234130696244821e-08, "loss": 0.0274, "step": 2820 }, { "epoch": 0.38, "learning_rate": 6.232794333823332e-08, "loss": 0.0492, "step": 2821 }, { "epoch": 0.38, "learning_rate": 6.231457971401843e-08, "loss": 0.0611, "step": 2822 }, { "epoch": 0.38, "learning_rate": 6.230121608980356e-08, "loss": 0.065, "step": 2823 }, { "epoch": 0.38, "learning_rate": 6.228785246558867e-08, "loss": 0.0333, "step": 2824 }, { "epoch": 0.38, "learning_rate": 6.227448884137378e-08, "loss": 0.0597, "step": 2825 }, { "epoch": 0.38, "learning_rate": 6.22611252171589e-08, "loss": 0.0507, "step": 2826 }, { "epoch": 0.38, "learning_rate": 6.224776159294401e-08, "loss": 0.0518, "step": 2827 }, { "epoch": 0.38, "learning_rate": 6.223439796872912e-08, "loss": 0.0419, "step": 2828 }, { "epoch": 0.38, "learning_rate": 6.222103434451423e-08, "loss": 0.0538, "step": 2829 }, { "epoch": 0.38, "learning_rate": 6.220767072029934e-08, "loss": 0.0521, "step": 2830 }, { "epoch": 0.38, "learning_rate": 6.219430709608445e-08, "loss": 0.048, "step": 2831 }, { "epoch": 0.38, "learning_rate": 6.218094347186957e-08, "loss": 0.0603, "step": 2832 }, { "epoch": 0.38, "learning_rate": 6.216757984765468e-08, "loss": 0.0429, "step": 2833 }, { "epoch": 0.38, "learning_rate": 6.215421622343979e-08, "loss": 0.0607, "step": 2834 }, { "epoch": 0.38, "learning_rate": 6.21408525992249e-08, "loss": 0.041, "step": 2835 }, { "epoch": 0.38, "learning_rate": 6.212748897501002e-08, "loss": 0.0508, "step": 2836 }, { "epoch": 0.38, "learning_rate": 6.211412535079513e-08, "loss": 0.0483, "step": 2837 }, { "epoch": 0.38, "learning_rate": 6.210076172658024e-08, "loss": 0.0428, "step": 2838 }, { "epoch": 0.38, "learning_rate": 6.208739810236537e-08, "loss": 0.0485, "step": 2839 }, { "epoch": 0.38, "learning_rate": 6.207403447815048e-08, "loss": 0.0609, "step": 2840 }, { "epoch": 0.38, "learning_rate": 6.206067085393559e-08, "loss": 0.0562, "step": 2841 }, { "epoch": 0.38, "learning_rate": 6.20473072297207e-08, "loss": 0.0532, "step": 2842 }, { "epoch": 0.38, "learning_rate": 6.203394360550581e-08, "loss": 0.0431, "step": 2843 }, { "epoch": 0.38, "learning_rate": 6.202057998129092e-08, "loss": 0.0421, "step": 2844 }, { "epoch": 0.38, "learning_rate": 6.200721635707603e-08, "loss": 0.0429, "step": 2845 }, { "epoch": 0.38, "learning_rate": 6.199385273286115e-08, "loss": 0.0631, "step": 2846 }, { "epoch": 0.38, "learning_rate": 6.198048910864626e-08, "loss": 0.0252, "step": 2847 }, { "epoch": 0.38, "learning_rate": 6.196712548443137e-08, "loss": 0.043, "step": 2848 }, { "epoch": 0.38, "learning_rate": 6.195376186021649e-08, "loss": 0.0476, "step": 2849 }, { "epoch": 0.38, "learning_rate": 6.19403982360016e-08, "loss": 0.0301, "step": 2850 }, { "epoch": 0.38, "learning_rate": 6.192703461178671e-08, "loss": 0.0582, "step": 2851 }, { "epoch": 0.38, "learning_rate": 6.191367098757184e-08, "loss": 0.0563, "step": 2852 }, { "epoch": 0.38, "learning_rate": 6.190030736335694e-08, "loss": 0.0554, "step": 2853 }, { "epoch": 0.38, "learning_rate": 6.188694373914205e-08, "loss": 0.0679, "step": 2854 }, { "epoch": 0.38, "learning_rate": 6.187358011492716e-08, "loss": 0.0987, "step": 2855 }, { "epoch": 0.38, "learning_rate": 6.186021649071227e-08, "loss": 0.0302, "step": 2856 }, { "epoch": 0.38, "learning_rate": 6.184685286649738e-08, "loss": 0.0612, "step": 2857 }, { "epoch": 0.38, "learning_rate": 6.18334892422825e-08, "loss": 0.0649, "step": 2858 }, { "epoch": 0.38, "learning_rate": 6.182012561806762e-08, "loss": 0.0486, "step": 2859 }, { "epoch": 0.38, "learning_rate": 6.180676199385273e-08, "loss": 0.0459, "step": 2860 }, { "epoch": 0.38, "learning_rate": 6.179339836963784e-08, "loss": 0.068, "step": 2861 }, { "epoch": 0.38, "learning_rate": 6.178003474542296e-08, "loss": 0.0598, "step": 2862 }, { "epoch": 0.38, "learning_rate": 6.176667112120807e-08, "loss": 0.0477, "step": 2863 }, { "epoch": 0.38, "learning_rate": 6.175330749699318e-08, "loss": 0.0504, "step": 2864 }, { "epoch": 0.38, "learning_rate": 6.17399438727783e-08, "loss": 0.0336, "step": 2865 }, { "epoch": 0.38, "learning_rate": 6.172658024856341e-08, "loss": 0.0364, "step": 2866 }, { "epoch": 0.38, "learning_rate": 6.171321662434852e-08, "loss": 0.0331, "step": 2867 }, { "epoch": 0.38, "learning_rate": 6.169985300013363e-08, "loss": 0.0502, "step": 2868 }, { "epoch": 0.38, "learning_rate": 6.168648937591876e-08, "loss": 0.0844, "step": 2869 }, { "epoch": 0.38, "learning_rate": 6.167312575170385e-08, "loss": 0.0385, "step": 2870 }, { "epoch": 0.38, "learning_rate": 6.165976212748896e-08, "loss": 0.0694, "step": 2871 }, { "epoch": 0.38, "learning_rate": 6.164639850327408e-08, "loss": 0.0431, "step": 2872 }, { "epoch": 0.38, "learning_rate": 6.16330348790592e-08, "loss": 0.0401, "step": 2873 }, { "epoch": 0.38, "learning_rate": 6.16196712548443e-08, "loss": 0.0466, "step": 2874 }, { "epoch": 0.38, "learning_rate": 6.160630763062943e-08, "loss": 0.071, "step": 2875 }, { "epoch": 0.38, "learning_rate": 6.159294400641454e-08, "loss": 0.0538, "step": 2876 }, { "epoch": 0.38, "learning_rate": 6.157958038219965e-08, "loss": 0.0555, "step": 2877 }, { "epoch": 0.38, "learning_rate": 6.156621675798477e-08, "loss": 0.047, "step": 2878 }, { "epoch": 0.38, "learning_rate": 6.155285313376988e-08, "loss": 0.036, "step": 2879 }, { "epoch": 0.38, "learning_rate": 6.153948950955499e-08, "loss": 0.0367, "step": 2880 }, { "epoch": 0.39, "learning_rate": 6.15261258853401e-08, "loss": 0.0461, "step": 2881 }, { "epoch": 0.39, "learning_rate": 6.151276226112522e-08, "loss": 0.0501, "step": 2882 }, { "epoch": 0.39, "learning_rate": 6.149939863691033e-08, "loss": 0.0566, "step": 2883 }, { "epoch": 0.39, "learning_rate": 6.148603501269544e-08, "loss": 0.0432, "step": 2884 }, { "epoch": 0.39, "learning_rate": 6.147267138848055e-08, "loss": 0.0947, "step": 2885 }, { "epoch": 0.39, "learning_rate": 6.145930776426566e-08, "loss": 0.0436, "step": 2886 }, { "epoch": 0.39, "learning_rate": 6.144594414005077e-08, "loss": 0.0606, "step": 2887 }, { "epoch": 0.39, "learning_rate": 6.14325805158359e-08, "loss": 0.0384, "step": 2888 }, { "epoch": 0.39, "learning_rate": 6.1419216891621e-08, "loss": 0.0443, "step": 2889 }, { "epoch": 0.39, "learning_rate": 6.140585326740611e-08, "loss": 0.0838, "step": 2890 }, { "epoch": 0.39, "learning_rate": 6.139248964319122e-08, "loss": 0.0487, "step": 2891 }, { "epoch": 0.39, "learning_rate": 6.137912601897635e-08, "loss": 0.0563, "step": 2892 }, { "epoch": 0.39, "learning_rate": 6.136576239476146e-08, "loss": 0.0687, "step": 2893 }, { "epoch": 0.39, "learning_rate": 6.135239877054657e-08, "loss": 0.0461, "step": 2894 }, { "epoch": 0.39, "learning_rate": 6.133903514633169e-08, "loss": 0.0419, "step": 2895 }, { "epoch": 0.39, "learning_rate": 6.13256715221168e-08, "loss": 0.0506, "step": 2896 }, { "epoch": 0.39, "learning_rate": 6.131230789790191e-08, "loss": 0.051, "step": 2897 }, { "epoch": 0.39, "learning_rate": 6.129894427368702e-08, "loss": 0.0492, "step": 2898 }, { "epoch": 0.39, "learning_rate": 6.128558064947213e-08, "loss": 0.0616, "step": 2899 }, { "epoch": 0.39, "learning_rate": 6.127221702525724e-08, "loss": 0.0498, "step": 2900 }, { "epoch": 0.39, "learning_rate": 6.125885340104236e-08, "loss": 0.0451, "step": 2901 }, { "epoch": 0.39, "learning_rate": 6.124548977682747e-08, "loss": 0.0667, "step": 2902 }, { "epoch": 0.39, "learning_rate": 6.123212615261258e-08, "loss": 0.0769, "step": 2903 }, { "epoch": 0.39, "learning_rate": 6.121876252839769e-08, "loss": 0.0453, "step": 2904 }, { "epoch": 0.39, "learning_rate": 6.120539890418282e-08, "loss": 0.0504, "step": 2905 }, { "epoch": 0.39, "learning_rate": 6.119203527996792e-08, "loss": 0.0517, "step": 2906 }, { "epoch": 0.39, "learning_rate": 6.117867165575303e-08, "loss": 0.0223, "step": 2907 }, { "epoch": 0.39, "learning_rate": 6.116530803153816e-08, "loss": 0.0427, "step": 2908 }, { "epoch": 0.39, "learning_rate": 6.115194440732327e-08, "loss": 0.0697, "step": 2909 }, { "epoch": 0.39, "learning_rate": 6.113858078310838e-08, "loss": 0.0352, "step": 2910 }, { "epoch": 0.39, "learning_rate": 6.11252171588935e-08, "loss": 0.0493, "step": 2911 }, { "epoch": 0.39, "learning_rate": 6.11118535346786e-08, "loss": 0.0486, "step": 2912 }, { "epoch": 0.39, "learning_rate": 6.109848991046371e-08, "loss": 0.0379, "step": 2913 }, { "epoch": 0.39, "learning_rate": 6.108512628624882e-08, "loss": 0.051, "step": 2914 }, { "epoch": 0.39, "learning_rate": 6.107176266203394e-08, "loss": 0.0692, "step": 2915 }, { "epoch": 0.39, "learning_rate": 6.105839903781905e-08, "loss": 0.0654, "step": 2916 }, { "epoch": 0.39, "learning_rate": 6.104503541360416e-08, "loss": 0.0669, "step": 2917 }, { "epoch": 0.39, "learning_rate": 6.103167178938928e-08, "loss": 0.0439, "step": 2918 }, { "epoch": 0.39, "learning_rate": 6.101830816517439e-08, "loss": 0.0692, "step": 2919 }, { "epoch": 0.39, "learning_rate": 6.10049445409595e-08, "loss": 0.0313, "step": 2920 }, { "epoch": 0.39, "learning_rate": 6.099158091674463e-08, "loss": 0.0842, "step": 2921 }, { "epoch": 0.39, "learning_rate": 6.097821729252974e-08, "loss": 0.052, "step": 2922 }, { "epoch": 0.39, "learning_rate": 6.096485366831484e-08, "loss": 0.0703, "step": 2923 }, { "epoch": 0.39, "learning_rate": 6.095149004409997e-08, "loss": 0.0541, "step": 2924 }, { "epoch": 0.39, "learning_rate": 6.093812641988508e-08, "loss": 0.0469, "step": 2925 }, { "epoch": 0.39, "learning_rate": 6.092476279567019e-08, "loss": 0.0521, "step": 2926 }, { "epoch": 0.39, "learning_rate": 6.09113991714553e-08, "loss": 0.0508, "step": 2927 }, { "epoch": 0.39, "learning_rate": 6.089803554724041e-08, "loss": 0.049, "step": 2928 }, { "epoch": 0.39, "learning_rate": 6.088467192302552e-08, "loss": 0.0773, "step": 2929 }, { "epoch": 0.39, "learning_rate": 6.087130829881063e-08, "loss": 0.0381, "step": 2930 }, { "epoch": 0.39, "learning_rate": 6.085794467459575e-08, "loss": 0.0754, "step": 2931 }, { "epoch": 0.39, "learning_rate": 6.084458105038086e-08, "loss": 0.055, "step": 2932 }, { "epoch": 0.39, "learning_rate": 6.083121742616597e-08, "loss": 0.0472, "step": 2933 }, { "epoch": 0.39, "learning_rate": 6.081785380195109e-08, "loss": 0.0534, "step": 2934 }, { "epoch": 0.39, "learning_rate": 6.08044901777362e-08, "loss": 0.0653, "step": 2935 }, { "epoch": 0.39, "learning_rate": 6.079112655352131e-08, "loss": 0.0694, "step": 2936 }, { "epoch": 0.39, "learning_rate": 6.077776292930642e-08, "loss": 0.0361, "step": 2937 }, { "epoch": 0.39, "learning_rate": 6.076439930509155e-08, "loss": 0.0763, "step": 2938 }, { "epoch": 0.39, "learning_rate": 6.075103568087666e-08, "loss": 0.055, "step": 2939 }, { "epoch": 0.39, "learning_rate": 6.073767205666177e-08, "loss": 0.0506, "step": 2940 }, { "epoch": 0.39, "learning_rate": 6.072430843244687e-08, "loss": 0.0562, "step": 2941 }, { "epoch": 0.39, "learning_rate": 6.071094480823198e-08, "loss": 0.0784, "step": 2942 }, { "epoch": 0.39, "learning_rate": 6.06975811840171e-08, "loss": 0.0428, "step": 2943 }, { "epoch": 0.39, "learning_rate": 6.068421755980222e-08, "loss": 0.0948, "step": 2944 }, { "epoch": 0.39, "learning_rate": 6.067085393558733e-08, "loss": 0.0433, "step": 2945 }, { "epoch": 0.39, "learning_rate": 6.065749031137244e-08, "loss": 0.0393, "step": 2946 }, { "epoch": 0.39, "learning_rate": 6.064412668715756e-08, "loss": 0.0571, "step": 2947 }, { "epoch": 0.39, "learning_rate": 6.063076306294267e-08, "loss": 0.0477, "step": 2948 }, { "epoch": 0.39, "learning_rate": 6.061739943872778e-08, "loss": 0.0627, "step": 2949 }, { "epoch": 0.39, "learning_rate": 6.060403581451289e-08, "loss": 0.0771, "step": 2950 }, { "epoch": 0.39, "learning_rate": 6.059067219029801e-08, "loss": 0.0237, "step": 2951 }, { "epoch": 0.39, "learning_rate": 6.057730856608312e-08, "loss": 0.0443, "step": 2952 }, { "epoch": 0.39, "learning_rate": 6.056394494186823e-08, "loss": 0.0493, "step": 2953 }, { "epoch": 0.39, "learning_rate": 6.055058131765334e-08, "loss": 0.0491, "step": 2954 }, { "epoch": 0.39, "learning_rate": 6.053721769343845e-08, "loss": 0.0318, "step": 2955 }, { "epoch": 0.4, "learning_rate": 6.052385406922356e-08, "loss": 0.0401, "step": 2956 }, { "epoch": 0.4, "learning_rate": 6.051049044500869e-08, "loss": 0.052, "step": 2957 }, { "epoch": 0.4, "learning_rate": 6.04971268207938e-08, "loss": 0.0817, "step": 2958 }, { "epoch": 0.4, "learning_rate": 6.04837631965789e-08, "loss": 0.0357, "step": 2959 }, { "epoch": 0.4, "learning_rate": 6.047039957236401e-08, "loss": 0.0498, "step": 2960 }, { "epoch": 0.4, "learning_rate": 6.045703594814914e-08, "loss": 0.052, "step": 2961 }, { "epoch": 0.4, "learning_rate": 6.044367232393425e-08, "loss": 0.055, "step": 2962 }, { "epoch": 0.4, "learning_rate": 6.043030869971936e-08, "loss": 0.0315, "step": 2963 }, { "epoch": 0.4, "learning_rate": 6.041694507550448e-08, "loss": 0.0603, "step": 2964 }, { "epoch": 0.4, "learning_rate": 6.040358145128959e-08, "loss": 0.0381, "step": 2965 }, { "epoch": 0.4, "learning_rate": 6.03902178270747e-08, "loss": 0.0455, "step": 2966 }, { "epoch": 0.4, "learning_rate": 6.037685420285982e-08, "loss": 0.0512, "step": 2967 }, { "epoch": 0.4, "learning_rate": 6.036349057864493e-08, "loss": 0.0543, "step": 2968 }, { "epoch": 0.4, "learning_rate": 6.035012695443003e-08, "loss": 0.0373, "step": 2969 }, { "epoch": 0.4, "learning_rate": 6.033676333021515e-08, "loss": 0.0603, "step": 2970 }, { "epoch": 0.4, "learning_rate": 6.032339970600026e-08, "loss": 0.051, "step": 2971 }, { "epoch": 0.4, "learning_rate": 6.031003608178537e-08, "loss": 0.1098, "step": 2972 }, { "epoch": 0.4, "learning_rate": 6.029667245757048e-08, "loss": 0.0374, "step": 2973 }, { "epoch": 0.4, "learning_rate": 6.02833088333556e-08, "loss": 0.0409, "step": 2974 }, { "epoch": 0.4, "learning_rate": 6.026994520914072e-08, "loss": 0.0471, "step": 2975 }, { "epoch": 0.4, "learning_rate": 6.025658158492582e-08, "loss": 0.0634, "step": 2976 }, { "epoch": 0.4, "learning_rate": 6.024321796071095e-08, "loss": 0.0347, "step": 2977 }, { "epoch": 0.4, "learning_rate": 6.022985433649606e-08, "loss": 0.0668, "step": 2978 }, { "epoch": 0.4, "learning_rate": 6.021649071228117e-08, "loss": 0.0846, "step": 2979 }, { "epoch": 0.4, "learning_rate": 6.020312708806629e-08, "loss": 0.0714, "step": 2980 }, { "epoch": 0.4, "learning_rate": 6.01897634638514e-08, "loss": 0.0592, "step": 2981 }, { "epoch": 0.4, "learning_rate": 6.017639983963651e-08, "loss": 0.0563, "step": 2982 }, { "epoch": 0.4, "learning_rate": 6.016303621542162e-08, "loss": 0.0569, "step": 2983 }, { "epoch": 0.4, "learning_rate": 6.014967259120673e-08, "loss": 0.0508, "step": 2984 }, { "epoch": 0.4, "learning_rate": 6.013630896699184e-08, "loss": 0.0548, "step": 2985 }, { "epoch": 0.4, "learning_rate": 6.012294534277695e-08, "loss": 0.0657, "step": 2986 }, { "epoch": 0.4, "learning_rate": 6.010958171856207e-08, "loss": 0.0476, "step": 2987 }, { "epoch": 0.4, "learning_rate": 6.009621809434718e-08, "loss": 0.0578, "step": 2988 }, { "epoch": 0.4, "learning_rate": 6.008285447013229e-08, "loss": 0.053, "step": 2989 }, { "epoch": 0.4, "learning_rate": 6.006949084591742e-08, "loss": 0.0525, "step": 2990 }, { "epoch": 0.4, "learning_rate": 6.005612722170253e-08, "loss": 0.0363, "step": 2991 }, { "epoch": 0.4, "learning_rate": 6.004276359748764e-08, "loss": 0.0743, "step": 2992 }, { "epoch": 0.4, "learning_rate": 6.002939997327276e-08, "loss": 0.0313, "step": 2993 }, { "epoch": 0.4, "learning_rate": 6.001603634905787e-08, "loss": 0.048, "step": 2994 }, { "epoch": 0.4, "learning_rate": 6.000267272484298e-08, "loss": 0.0446, "step": 2995 }, { "epoch": 0.4, "learning_rate": 5.998930910062809e-08, "loss": 0.0381, "step": 2996 }, { "epoch": 0.4, "learning_rate": 5.99759454764132e-08, "loss": 0.084, "step": 2997 }, { "epoch": 0.4, "learning_rate": 5.996258185219831e-08, "loss": 0.0553, "step": 2998 }, { "epoch": 0.4, "learning_rate": 5.994921822798342e-08, "loss": 0.0378, "step": 2999 }, { "epoch": 0.4, "learning_rate": 5.993585460376854e-08, "loss": 0.0461, "step": 3000 }, { "epoch": 0.4, "learning_rate": 5.992249097955365e-08, "loss": 0.0686, "step": 3001 }, { "epoch": 0.4, "learning_rate": 5.990912735533876e-08, "loss": 0.0651, "step": 3002 }, { "epoch": 0.4, "learning_rate": 5.989576373112388e-08, "loss": 0.0419, "step": 3003 }, { "epoch": 0.4, "learning_rate": 5.988240010690899e-08, "loss": 0.0439, "step": 3004 }, { "epoch": 0.4, "learning_rate": 5.98690364826941e-08, "loss": 0.0567, "step": 3005 }, { "epoch": 0.4, "learning_rate": 5.985567285847921e-08, "loss": 0.0547, "step": 3006 }, { "epoch": 0.4, "learning_rate": 5.984230923426434e-08, "loss": 0.0562, "step": 3007 }, { "epoch": 0.4, "learning_rate": 5.982894561004945e-08, "loss": 0.0254, "step": 3008 }, { "epoch": 0.4, "learning_rate": 5.981558198583456e-08, "loss": 0.046, "step": 3009 }, { "epoch": 0.4, "learning_rate": 5.980221836161968e-08, "loss": 0.0648, "step": 3010 }, { "epoch": 0.4, "learning_rate": 5.978885473740478e-08, "loss": 0.0508, "step": 3011 }, { "epoch": 0.4, "learning_rate": 5.977549111318988e-08, "loss": 0.0539, "step": 3012 }, { "epoch": 0.4, "learning_rate": 5.976212748897501e-08, "loss": 0.0426, "step": 3013 }, { "epoch": 0.4, "learning_rate": 5.974876386476012e-08, "loss": 0.0724, "step": 3014 }, { "epoch": 0.4, "learning_rate": 5.973540024054523e-08, "loss": 0.0575, "step": 3015 }, { "epoch": 0.4, "learning_rate": 5.972203661633035e-08, "loss": 0.0353, "step": 3016 }, { "epoch": 0.4, "learning_rate": 5.970867299211546e-08, "loss": 0.0534, "step": 3017 }, { "epoch": 0.4, "learning_rate": 5.969530936790057e-08, "loss": 0.0512, "step": 3018 }, { "epoch": 0.4, "learning_rate": 5.968194574368568e-08, "loss": 0.0395, "step": 3019 }, { "epoch": 0.4, "learning_rate": 5.96685821194708e-08, "loss": 0.0457, "step": 3020 }, { "epoch": 0.4, "learning_rate": 5.965521849525591e-08, "loss": 0.0427, "step": 3021 }, { "epoch": 0.4, "learning_rate": 5.964185487104102e-08, "loss": 0.045, "step": 3022 }, { "epoch": 0.4, "learning_rate": 5.962849124682615e-08, "loss": 0.0819, "step": 3023 }, { "epoch": 0.4, "learning_rate": 5.961512762261126e-08, "loss": 0.042, "step": 3024 }, { "epoch": 0.4, "learning_rate": 5.960176399839636e-08, "loss": 0.0547, "step": 3025 }, { "epoch": 0.4, "learning_rate": 5.958840037418148e-08, "loss": 0.0351, "step": 3026 }, { "epoch": 0.4, "learning_rate": 5.957503674996659e-08, "loss": 0.0391, "step": 3027 }, { "epoch": 0.4, "learning_rate": 5.95616731257517e-08, "loss": 0.0747, "step": 3028 }, { "epoch": 0.4, "learning_rate": 5.954830950153681e-08, "loss": 0.05, "step": 3029 }, { "epoch": 0.4, "learning_rate": 5.953494587732193e-08, "loss": 0.0592, "step": 3030 }, { "epoch": 0.41, "learning_rate": 5.952158225310704e-08, "loss": 0.0468, "step": 3031 }, { "epoch": 0.41, "learning_rate": 5.950821862889215e-08, "loss": 0.064, "step": 3032 }, { "epoch": 0.41, "learning_rate": 5.949485500467727e-08, "loss": 0.0413, "step": 3033 }, { "epoch": 0.41, "learning_rate": 5.948149138046238e-08, "loss": 0.0328, "step": 3034 }, { "epoch": 0.41, "learning_rate": 5.946812775624749e-08, "loss": 0.0398, "step": 3035 }, { "epoch": 0.41, "learning_rate": 5.945476413203261e-08, "loss": 0.0747, "step": 3036 }, { "epoch": 0.41, "learning_rate": 5.944140050781772e-08, "loss": 0.0808, "step": 3037 }, { "epoch": 0.41, "learning_rate": 5.9428036883602827e-08, "loss": 0.053, "step": 3038 }, { "epoch": 0.41, "learning_rate": 5.941467325938795e-08, "loss": 0.0769, "step": 3039 }, { "epoch": 0.41, "learning_rate": 5.940130963517306e-08, "loss": 0.0545, "step": 3040 }, { "epoch": 0.41, "learning_rate": 5.938794601095817e-08, "loss": 0.0448, "step": 3041 }, { "epoch": 0.41, "learning_rate": 5.937458238674328e-08, "loss": 0.0457, "step": 3042 }, { "epoch": 0.41, "learning_rate": 5.9361218762528396e-08, "loss": 0.0285, "step": 3043 }, { "epoch": 0.41, "learning_rate": 5.9347855138313505e-08, "loss": 0.0476, "step": 3044 }, { "epoch": 0.41, "learning_rate": 5.9334491514098615e-08, "loss": 0.0565, "step": 3045 }, { "epoch": 0.41, "learning_rate": 5.932112788988374e-08, "loss": 0.0399, "step": 3046 }, { "epoch": 0.41, "learning_rate": 5.930776426566885e-08, "loss": 0.0452, "step": 3047 }, { "epoch": 0.41, "learning_rate": 5.929440064145396e-08, "loss": 0.0634, "step": 3048 }, { "epoch": 0.41, "learning_rate": 5.9281037017239074e-08, "loss": 0.0507, "step": 3049 }, { "epoch": 0.41, "learning_rate": 5.9267673393024184e-08, "loss": 0.0421, "step": 3050 }, { "epoch": 0.41, "learning_rate": 5.9254309768809294e-08, "loss": 0.0359, "step": 3051 }, { "epoch": 0.41, "learning_rate": 5.9240946144594404e-08, "loss": 0.0358, "step": 3052 }, { "epoch": 0.41, "learning_rate": 5.922758252037953e-08, "loss": 0.047, "step": 3053 }, { "epoch": 0.41, "learning_rate": 5.921421889616464e-08, "loss": 0.077, "step": 3054 }, { "epoch": 0.41, "learning_rate": 5.9200855271949747e-08, "loss": 0.0499, "step": 3055 }, { "epoch": 0.41, "learning_rate": 5.918749164773486e-08, "loss": 0.0578, "step": 3056 }, { "epoch": 0.41, "learning_rate": 5.917412802351997e-08, "loss": 0.0336, "step": 3057 }, { "epoch": 0.41, "learning_rate": 5.916076439930508e-08, "loss": 0.0565, "step": 3058 }, { "epoch": 0.41, "learning_rate": 5.9147400775090206e-08, "loss": 0.0437, "step": 3059 }, { "epoch": 0.41, "learning_rate": 5.9134037150875316e-08, "loss": 0.0498, "step": 3060 }, { "epoch": 0.41, "learning_rate": 5.9120673526660425e-08, "loss": 0.0608, "step": 3061 }, { "epoch": 0.41, "learning_rate": 5.910730990244555e-08, "loss": 0.0641, "step": 3062 }, { "epoch": 0.41, "learning_rate": 5.909394627823066e-08, "loss": 0.0473, "step": 3063 }, { "epoch": 0.41, "learning_rate": 5.908058265401576e-08, "loss": 0.0356, "step": 3064 }, { "epoch": 0.41, "learning_rate": 5.906721902980087e-08, "loss": 0.0327, "step": 3065 }, { "epoch": 0.41, "learning_rate": 5.9053855405585995e-08, "loss": 0.0627, "step": 3066 }, { "epoch": 0.41, "learning_rate": 5.9040491781371104e-08, "loss": 0.0627, "step": 3067 }, { "epoch": 0.41, "learning_rate": 5.9027128157156214e-08, "loss": 0.0505, "step": 3068 }, { "epoch": 0.41, "learning_rate": 5.901376453294134e-08, "loss": 0.0609, "step": 3069 }, { "epoch": 0.41, "learning_rate": 5.900040090872645e-08, "loss": 0.0558, "step": 3070 }, { "epoch": 0.41, "learning_rate": 5.898703728451156e-08, "loss": 0.04, "step": 3071 }, { "epoch": 0.41, "learning_rate": 5.8973673660296673e-08, "loss": 0.0403, "step": 3072 }, { "epoch": 0.41, "learning_rate": 5.896031003608178e-08, "loss": 0.0619, "step": 3073 }, { "epoch": 0.41, "learning_rate": 5.894694641186689e-08, "loss": 0.0795, "step": 3074 }, { "epoch": 0.41, "learning_rate": 5.8933582787652e-08, "loss": 0.0793, "step": 3075 }, { "epoch": 0.41, "learning_rate": 5.8920219163437126e-08, "loss": 0.0545, "step": 3076 }, { "epoch": 0.41, "learning_rate": 5.8906855539222236e-08, "loss": 0.0498, "step": 3077 }, { "epoch": 0.41, "learning_rate": 5.8893491915007346e-08, "loss": 0.058, "step": 3078 }, { "epoch": 0.41, "learning_rate": 5.888012829079246e-08, "loss": 0.0348, "step": 3079 }, { "epoch": 0.41, "learning_rate": 5.886676466657757e-08, "loss": 0.0455, "step": 3080 }, { "epoch": 0.41, "learning_rate": 5.885340104236268e-08, "loss": 0.08, "step": 3081 }, { "epoch": 0.41, "learning_rate": 5.8840037418147805e-08, "loss": 0.0467, "step": 3082 }, { "epoch": 0.41, "learning_rate": 5.8826673793932915e-08, "loss": 0.0441, "step": 3083 }, { "epoch": 0.41, "learning_rate": 5.8813310169718024e-08, "loss": 0.0397, "step": 3084 }, { "epoch": 0.41, "learning_rate": 5.879994654550314e-08, "loss": 0.0479, "step": 3085 }, { "epoch": 0.41, "learning_rate": 5.878658292128825e-08, "loss": 0.0531, "step": 3086 }, { "epoch": 0.41, "learning_rate": 5.877321929707336e-08, "loss": 0.0556, "step": 3087 }, { "epoch": 0.41, "learning_rate": 5.875985567285847e-08, "loss": 0.0465, "step": 3088 }, { "epoch": 0.41, "learning_rate": 5.8746492048643593e-08, "loss": 0.0474, "step": 3089 }, { "epoch": 0.41, "learning_rate": 5.87331284244287e-08, "loss": 0.0563, "step": 3090 }, { "epoch": 0.41, "learning_rate": 5.871976480021381e-08, "loss": 0.0606, "step": 3091 }, { "epoch": 0.41, "learning_rate": 5.870640117599893e-08, "loss": 0.0517, "step": 3092 }, { "epoch": 0.41, "learning_rate": 5.869303755178404e-08, "loss": 0.0556, "step": 3093 }, { "epoch": 0.41, "learning_rate": 5.867967392756915e-08, "loss": 0.0456, "step": 3094 }, { "epoch": 0.41, "learning_rate": 5.866631030335427e-08, "loss": 0.0412, "step": 3095 }, { "epoch": 0.41, "learning_rate": 5.865294667913938e-08, "loss": 0.1025, "step": 3096 }, { "epoch": 0.41, "learning_rate": 5.863958305492449e-08, "loss": 0.045, "step": 3097 }, { "epoch": 0.41, "learning_rate": 5.86262194307096e-08, "loss": 0.089, "step": 3098 }, { "epoch": 0.41, "learning_rate": 5.861285580649472e-08, "loss": 0.0493, "step": 3099 }, { "epoch": 0.41, "learning_rate": 5.859949218227983e-08, "loss": 0.0431, "step": 3100 }, { "epoch": 0.41, "learning_rate": 5.858612855806494e-08, "loss": 0.0692, "step": 3101 }, { "epoch": 0.41, "learning_rate": 5.857276493385006e-08, "loss": 0.0624, "step": 3102 }, { "epoch": 0.41, "learning_rate": 5.855940130963517e-08, "loss": 0.0402, "step": 3103 }, { "epoch": 0.41, "learning_rate": 5.854603768542028e-08, "loss": 0.0734, "step": 3104 }, { "epoch": 0.41, "learning_rate": 5.85326740612054e-08, "loss": 0.0381, "step": 3105 }, { "epoch": 0.42, "learning_rate": 5.851931043699051e-08, "loss": 0.0326, "step": 3106 }, { "epoch": 0.42, "learning_rate": 5.850594681277562e-08, "loss": 0.0375, "step": 3107 }, { "epoch": 0.42, "learning_rate": 5.849258318856074e-08, "loss": 0.0287, "step": 3108 }, { "epoch": 0.42, "learning_rate": 5.847921956434585e-08, "loss": 0.0571, "step": 3109 }, { "epoch": 0.42, "learning_rate": 5.846585594013096e-08, "loss": 0.0521, "step": 3110 }, { "epoch": 0.42, "learning_rate": 5.845249231591607e-08, "loss": 0.0683, "step": 3111 }, { "epoch": 0.42, "learning_rate": 5.843912869170119e-08, "loss": 0.0267, "step": 3112 }, { "epoch": 0.42, "learning_rate": 5.84257650674863e-08, "loss": 0.0545, "step": 3113 }, { "epoch": 0.42, "learning_rate": 5.8412401443271405e-08, "loss": 0.0309, "step": 3114 }, { "epoch": 0.42, "learning_rate": 5.839903781905653e-08, "loss": 0.0718, "step": 3115 }, { "epoch": 0.42, "learning_rate": 5.838567419484164e-08, "loss": 0.0416, "step": 3116 }, { "epoch": 0.42, "learning_rate": 5.837231057062675e-08, "loss": 0.0355, "step": 3117 }, { "epoch": 0.42, "learning_rate": 5.835894694641187e-08, "loss": 0.0361, "step": 3118 }, { "epoch": 0.42, "learning_rate": 5.834558332219698e-08, "loss": 0.0554, "step": 3119 }, { "epoch": 0.42, "learning_rate": 5.833221969798209e-08, "loss": 0.0481, "step": 3120 }, { "epoch": 0.42, "learning_rate": 5.83188560737672e-08, "loss": 0.0658, "step": 3121 }, { "epoch": 0.42, "learning_rate": 5.830549244955232e-08, "loss": 0.0512, "step": 3122 }, { "epoch": 0.42, "learning_rate": 5.829212882533743e-08, "loss": 0.0444, "step": 3123 }, { "epoch": 0.42, "learning_rate": 5.827876520112254e-08, "loss": 0.0661, "step": 3124 }, { "epoch": 0.42, "learning_rate": 5.826540157690766e-08, "loss": 0.047, "step": 3125 }, { "epoch": 0.42, "learning_rate": 5.825203795269277e-08, "loss": 0.0622, "step": 3126 }, { "epoch": 0.42, "learning_rate": 5.823867432847788e-08, "loss": 0.0224, "step": 3127 }, { "epoch": 0.42, "learning_rate": 5.8225310704262996e-08, "loss": 0.038, "step": 3128 }, { "epoch": 0.42, "learning_rate": 5.8211947080048106e-08, "loss": 0.0957, "step": 3129 }, { "epoch": 0.42, "learning_rate": 5.8198583455833216e-08, "loss": 0.046, "step": 3130 }, { "epoch": 0.42, "learning_rate": 5.818521983161834e-08, "loss": 0.0505, "step": 3131 }, { "epoch": 0.42, "learning_rate": 5.817185620740345e-08, "loss": 0.0504, "step": 3132 }, { "epoch": 0.42, "learning_rate": 5.815849258318856e-08, "loss": 0.0397, "step": 3133 }, { "epoch": 0.42, "learning_rate": 5.814512895897367e-08, "loss": 0.0625, "step": 3134 }, { "epoch": 0.42, "learning_rate": 5.8131765334758785e-08, "loss": 0.0542, "step": 3135 }, { "epoch": 0.42, "learning_rate": 5.8118401710543895e-08, "loss": 0.0804, "step": 3136 }, { "epoch": 0.42, "learning_rate": 5.8105038086329004e-08, "loss": 0.0506, "step": 3137 }, { "epoch": 0.42, "learning_rate": 5.809167446211413e-08, "loss": 0.0657, "step": 3138 }, { "epoch": 0.42, "learning_rate": 5.807831083789924e-08, "loss": 0.0482, "step": 3139 }, { "epoch": 0.42, "learning_rate": 5.806494721368435e-08, "loss": 0.0401, "step": 3140 }, { "epoch": 0.42, "learning_rate": 5.8051583589469464e-08, "loss": 0.0554, "step": 3141 }, { "epoch": 0.42, "learning_rate": 5.8038219965254573e-08, "loss": 0.0278, "step": 3142 }, { "epoch": 0.42, "learning_rate": 5.802485634103968e-08, "loss": 0.0439, "step": 3143 }, { "epoch": 0.42, "learning_rate": 5.801149271682479e-08, "loss": 0.0411, "step": 3144 }, { "epoch": 0.42, "learning_rate": 5.7998129092609916e-08, "loss": 0.0731, "step": 3145 }, { "epoch": 0.42, "learning_rate": 5.7984765468395026e-08, "loss": 0.0438, "step": 3146 }, { "epoch": 0.42, "learning_rate": 5.7971401844180136e-08, "loss": 0.0528, "step": 3147 }, { "epoch": 0.42, "learning_rate": 5.795803821996525e-08, "loss": 0.0444, "step": 3148 }, { "epoch": 0.42, "learning_rate": 5.794467459575036e-08, "loss": 0.0425, "step": 3149 }, { "epoch": 0.42, "learning_rate": 5.793131097153547e-08, "loss": 0.044, "step": 3150 }, { "epoch": 0.42, "learning_rate": 5.7917947347320595e-08, "loss": 0.0558, "step": 3151 }, { "epoch": 0.42, "learning_rate": 5.7904583723105705e-08, "loss": 0.0843, "step": 3152 }, { "epoch": 0.42, "learning_rate": 5.7891220098890815e-08, "loss": 0.0593, "step": 3153 }, { "epoch": 0.42, "learning_rate": 5.787785647467594e-08, "loss": 0.0376, "step": 3154 }, { "epoch": 0.42, "learning_rate": 5.786449285046104e-08, "loss": 0.0384, "step": 3155 }, { "epoch": 0.42, "learning_rate": 5.785112922624615e-08, "loss": 0.042, "step": 3156 }, { "epoch": 0.42, "learning_rate": 5.783776560203126e-08, "loss": 0.0615, "step": 3157 }, { "epoch": 0.42, "learning_rate": 5.7824401977816384e-08, "loss": 0.0436, "step": 3158 }, { "epoch": 0.42, "learning_rate": 5.7811038353601494e-08, "loss": 0.0475, "step": 3159 }, { "epoch": 0.42, "learning_rate": 5.7797674729386603e-08, "loss": 0.0364, "step": 3160 }, { "epoch": 0.42, "learning_rate": 5.7784311105171726e-08, "loss": 0.0382, "step": 3161 }, { "epoch": 0.42, "learning_rate": 5.7770947480956836e-08, "loss": 0.0553, "step": 3162 }, { "epoch": 0.42, "learning_rate": 5.775758385674194e-08, "loss": 0.0619, "step": 3163 }, { "epoch": 0.42, "learning_rate": 5.774422023252706e-08, "loss": 0.0832, "step": 3164 }, { "epoch": 0.42, "learning_rate": 5.773085660831217e-08, "loss": 0.048, "step": 3165 }, { "epoch": 0.42, "learning_rate": 5.771749298409728e-08, "loss": 0.0369, "step": 3166 }, { "epoch": 0.42, "learning_rate": 5.770412935988239e-08, "loss": 0.0469, "step": 3167 }, { "epoch": 0.42, "learning_rate": 5.7690765735667515e-08, "loss": 0.0325, "step": 3168 }, { "epoch": 0.42, "learning_rate": 5.7677402111452625e-08, "loss": 0.0426, "step": 3169 }, { "epoch": 0.42, "learning_rate": 5.7664038487237735e-08, "loss": 0.0429, "step": 3170 }, { "epoch": 0.42, "learning_rate": 5.765067486302285e-08, "loss": 0.0414, "step": 3171 }, { "epoch": 0.42, "learning_rate": 5.763731123880796e-08, "loss": 0.0577, "step": 3172 }, { "epoch": 0.42, "learning_rate": 5.762394761459307e-08, "loss": 0.0455, "step": 3173 }, { "epoch": 0.42, "learning_rate": 5.7610583990378194e-08, "loss": 0.0574, "step": 3174 }, { "epoch": 0.42, "learning_rate": 5.7597220366163304e-08, "loss": 0.0448, "step": 3175 }, { "epoch": 0.42, "learning_rate": 5.7583856741948414e-08, "loss": 0.036, "step": 3176 }, { "epoch": 0.42, "learning_rate": 5.757049311773353e-08, "loss": 0.0433, "step": 3177 }, { "epoch": 0.42, "learning_rate": 5.755712949351864e-08, "loss": 0.0456, "step": 3178 }, { "epoch": 0.42, "learning_rate": 5.754376586930375e-08, "loss": 0.0385, "step": 3179 }, { "epoch": 0.42, "learning_rate": 5.753040224508886e-08, "loss": 0.0754, "step": 3180 }, { "epoch": 0.43, "learning_rate": 5.751703862087398e-08, "loss": 0.0474, "step": 3181 }, { "epoch": 0.43, "learning_rate": 5.750367499665909e-08, "loss": 0.0777, "step": 3182 }, { "epoch": 0.43, "learning_rate": 5.74903113724442e-08, "loss": 0.0678, "step": 3183 }, { "epoch": 0.43, "learning_rate": 5.747694774822932e-08, "loss": 0.0553, "step": 3184 }, { "epoch": 0.43, "learning_rate": 5.746358412401443e-08, "loss": 0.0475, "step": 3185 }, { "epoch": 0.43, "learning_rate": 5.745022049979954e-08, "loss": 0.0481, "step": 3186 }, { "epoch": 0.43, "learning_rate": 5.743685687558466e-08, "loss": 0.0275, "step": 3187 }, { "epoch": 0.43, "learning_rate": 5.742349325136977e-08, "loss": 0.05, "step": 3188 }, { "epoch": 0.43, "learning_rate": 5.741012962715488e-08, "loss": 0.0597, "step": 3189 }, { "epoch": 0.43, "learning_rate": 5.739676600294e-08, "loss": 0.0731, "step": 3190 }, { "epoch": 0.43, "learning_rate": 5.738340237872511e-08, "loss": 0.0387, "step": 3191 }, { "epoch": 0.43, "learning_rate": 5.737003875451022e-08, "loss": 0.057, "step": 3192 }, { "epoch": 0.43, "learning_rate": 5.735667513029533e-08, "loss": 0.0644, "step": 3193 }, { "epoch": 0.43, "learning_rate": 5.734331150608045e-08, "loss": 0.0398, "step": 3194 }, { "epoch": 0.43, "learning_rate": 5.732994788186556e-08, "loss": 0.0516, "step": 3195 }, { "epoch": 0.43, "learning_rate": 5.731658425765067e-08, "loss": 0.0294, "step": 3196 }, { "epoch": 0.43, "learning_rate": 5.7303220633435786e-08, "loss": 0.0344, "step": 3197 }, { "epoch": 0.43, "learning_rate": 5.7289857009220896e-08, "loss": 0.0802, "step": 3198 }, { "epoch": 0.43, "learning_rate": 5.7276493385006006e-08, "loss": 0.0466, "step": 3199 }, { "epoch": 0.43, "learning_rate": 5.726312976079113e-08, "loss": 0.0596, "step": 3200 }, { "epoch": 0.43, "learning_rate": 5.724976613657624e-08, "loss": 0.0537, "step": 3201 }, { "epoch": 0.43, "learning_rate": 5.723640251236135e-08, "loss": 0.0386, "step": 3202 }, { "epoch": 0.43, "learning_rate": 5.722303888814646e-08, "loss": 0.0458, "step": 3203 }, { "epoch": 0.43, "learning_rate": 5.7209675263931575e-08, "loss": 0.0629, "step": 3204 }, { "epoch": 0.43, "learning_rate": 5.7196311639716685e-08, "loss": 0.0558, "step": 3205 }, { "epoch": 0.43, "learning_rate": 5.7182948015501795e-08, "loss": 0.0425, "step": 3206 }, { "epoch": 0.43, "learning_rate": 5.716958439128692e-08, "loss": 0.0389, "step": 3207 }, { "epoch": 0.43, "learning_rate": 5.715622076707203e-08, "loss": 0.0449, "step": 3208 }, { "epoch": 0.43, "learning_rate": 5.714285714285714e-08, "loss": 0.0389, "step": 3209 }, { "epoch": 0.43, "learning_rate": 5.712949351864226e-08, "loss": 0.0471, "step": 3210 }, { "epoch": 0.43, "learning_rate": 5.711612989442737e-08, "loss": 0.0415, "step": 3211 }, { "epoch": 0.43, "learning_rate": 5.7102766270212474e-08, "loss": 0.0665, "step": 3212 }, { "epoch": 0.43, "learning_rate": 5.7089402645997597e-08, "loss": 0.0506, "step": 3213 }, { "epoch": 0.43, "learning_rate": 5.7076039021782706e-08, "loss": 0.0556, "step": 3214 }, { "epoch": 0.43, "learning_rate": 5.7062675397567816e-08, "loss": 0.0589, "step": 3215 }, { "epoch": 0.43, "learning_rate": 5.7049311773352926e-08, "loss": 0.064, "step": 3216 }, { "epoch": 0.43, "learning_rate": 5.703594814913805e-08, "loss": 0.0291, "step": 3217 }, { "epoch": 0.43, "learning_rate": 5.702258452492316e-08, "loss": 0.0446, "step": 3218 }, { "epoch": 0.43, "learning_rate": 5.700922090070827e-08, "loss": 0.0581, "step": 3219 }, { "epoch": 0.43, "learning_rate": 5.6995857276493385e-08, "loss": 0.059, "step": 3220 }, { "epoch": 0.43, "learning_rate": 5.6982493652278495e-08, "loss": 0.041, "step": 3221 }, { "epoch": 0.43, "learning_rate": 5.6969130028063605e-08, "loss": 0.0653, "step": 3222 }, { "epoch": 0.43, "learning_rate": 5.695576640384873e-08, "loss": 0.0386, "step": 3223 }, { "epoch": 0.43, "learning_rate": 5.694240277963384e-08, "loss": 0.0578, "step": 3224 }, { "epoch": 0.43, "learning_rate": 5.692903915541895e-08, "loss": 0.0711, "step": 3225 }, { "epoch": 0.43, "learning_rate": 5.691567553120406e-08, "loss": 0.0403, "step": 3226 }, { "epoch": 0.43, "learning_rate": 5.6902311906989174e-08, "loss": 0.0671, "step": 3227 }, { "epoch": 0.43, "learning_rate": 5.6888948282774284e-08, "loss": 0.0668, "step": 3228 }, { "epoch": 0.43, "learning_rate": 5.6875584658559394e-08, "loss": 0.0384, "step": 3229 }, { "epoch": 0.43, "learning_rate": 5.686222103434452e-08, "loss": 0.0489, "step": 3230 }, { "epoch": 0.43, "learning_rate": 5.6848857410129627e-08, "loss": 0.0717, "step": 3231 }, { "epoch": 0.43, "learning_rate": 5.6835493785914736e-08, "loss": 0.0513, "step": 3232 }, { "epoch": 0.43, "learning_rate": 5.682213016169985e-08, "loss": 0.0613, "step": 3233 }, { "epoch": 0.43, "learning_rate": 5.680876653748496e-08, "loss": 0.0409, "step": 3234 }, { "epoch": 0.43, "learning_rate": 5.679540291327007e-08, "loss": 0.0567, "step": 3235 }, { "epoch": 0.43, "learning_rate": 5.6782039289055196e-08, "loss": 0.061, "step": 3236 }, { "epoch": 0.43, "learning_rate": 5.6768675664840305e-08, "loss": 0.0544, "step": 3237 }, { "epoch": 0.43, "learning_rate": 5.6755312040625415e-08, "loss": 0.0526, "step": 3238 }, { "epoch": 0.43, "learning_rate": 5.6741948416410525e-08, "loss": 0.026, "step": 3239 }, { "epoch": 0.43, "learning_rate": 5.672858479219564e-08, "loss": 0.0792, "step": 3240 }, { "epoch": 0.43, "learning_rate": 5.671522116798075e-08, "loss": 0.0643, "step": 3241 }, { "epoch": 0.43, "learning_rate": 5.670185754376586e-08, "loss": 0.0362, "step": 3242 }, { "epoch": 0.43, "learning_rate": 5.6688493919550984e-08, "loss": 0.0381, "step": 3243 }, { "epoch": 0.43, "learning_rate": 5.6675130295336094e-08, "loss": 0.0602, "step": 3244 }, { "epoch": 0.43, "learning_rate": 5.6661766671121204e-08, "loss": 0.0636, "step": 3245 }, { "epoch": 0.43, "learning_rate": 5.664840304690632e-08, "loss": 0.0413, "step": 3246 }, { "epoch": 0.43, "learning_rate": 5.663503942269143e-08, "loss": 0.0787, "step": 3247 }, { "epoch": 0.43, "learning_rate": 5.662167579847654e-08, "loss": 0.0616, "step": 3248 }, { "epoch": 0.43, "learning_rate": 5.660831217426165e-08, "loss": 0.0737, "step": 3249 }, { "epoch": 0.43, "learning_rate": 5.659494855004677e-08, "loss": 0.0424, "step": 3250 }, { "epoch": 0.43, "learning_rate": 5.658158492583188e-08, "loss": 0.0495, "step": 3251 }, { "epoch": 0.43, "learning_rate": 5.656822130161699e-08, "loss": 0.058, "step": 3252 }, { "epoch": 0.43, "learning_rate": 5.6554857677402116e-08, "loss": 0.0367, "step": 3253 }, { "epoch": 0.43, "learning_rate": 5.654149405318722e-08, "loss": 0.0559, "step": 3254 }, { "epoch": 0.43, "learning_rate": 5.652813042897233e-08, "loss": 0.0432, "step": 3255 }, { "epoch": 0.44, "learning_rate": 5.651476680475745e-08, "loss": 0.0576, "step": 3256 }, { "epoch": 0.44, "learning_rate": 5.650140318054256e-08, "loss": 0.0574, "step": 3257 }, { "epoch": 0.44, "learning_rate": 5.648803955632767e-08, "loss": 0.0438, "step": 3258 }, { "epoch": 0.44, "learning_rate": 5.6474675932112795e-08, "loss": 0.0478, "step": 3259 }, { "epoch": 0.44, "learning_rate": 5.6461312307897904e-08, "loss": 0.0489, "step": 3260 }, { "epoch": 0.44, "learning_rate": 5.6447948683683014e-08, "loss": 0.0712, "step": 3261 }, { "epoch": 0.44, "learning_rate": 5.643458505946812e-08, "loss": 0.0742, "step": 3262 }, { "epoch": 0.44, "learning_rate": 5.642122143525324e-08, "loss": 0.0455, "step": 3263 }, { "epoch": 0.44, "learning_rate": 5.640785781103835e-08, "loss": 0.0643, "step": 3264 }, { "epoch": 0.44, "learning_rate": 5.639449418682346e-08, "loss": 0.0421, "step": 3265 }, { "epoch": 0.44, "learning_rate": 5.638113056260858e-08, "loss": 0.0329, "step": 3266 }, { "epoch": 0.44, "learning_rate": 5.636776693839369e-08, "loss": 0.0622, "step": 3267 }, { "epoch": 0.44, "learning_rate": 5.63544033141788e-08, "loss": 0.0461, "step": 3268 }, { "epoch": 0.44, "learning_rate": 5.634103968996392e-08, "loss": 0.0645, "step": 3269 }, { "epoch": 0.44, "learning_rate": 5.632767606574903e-08, "loss": 0.0457, "step": 3270 }, { "epoch": 0.44, "learning_rate": 5.631431244153414e-08, "loss": 0.0585, "step": 3271 }, { "epoch": 0.44, "learning_rate": 5.630094881731925e-08, "loss": 0.047, "step": 3272 }, { "epoch": 0.44, "learning_rate": 5.628758519310437e-08, "loss": 0.0572, "step": 3273 }, { "epoch": 0.44, "learning_rate": 5.627422156888948e-08, "loss": 0.0672, "step": 3274 }, { "epoch": 0.44, "learning_rate": 5.626085794467459e-08, "loss": 0.0483, "step": 3275 }, { "epoch": 0.44, "learning_rate": 5.624749432045971e-08, "loss": 0.0464, "step": 3276 }, { "epoch": 0.44, "learning_rate": 5.623413069624482e-08, "loss": 0.0369, "step": 3277 }, { "epoch": 0.44, "learning_rate": 5.622076707202993e-08, "loss": 0.0782, "step": 3278 }, { "epoch": 0.44, "learning_rate": 5.620740344781505e-08, "loss": 0.0307, "step": 3279 }, { "epoch": 0.44, "learning_rate": 5.619403982360016e-08, "loss": 0.055, "step": 3280 }, { "epoch": 0.44, "learning_rate": 5.618067619938527e-08, "loss": 0.0381, "step": 3281 }, { "epoch": 0.44, "learning_rate": 5.616731257517039e-08, "loss": 0.0477, "step": 3282 }, { "epoch": 0.44, "learning_rate": 5.61539489509555e-08, "loss": 0.047, "step": 3283 }, { "epoch": 0.44, "learning_rate": 5.6140585326740607e-08, "loss": 0.0477, "step": 3284 }, { "epoch": 0.44, "learning_rate": 5.6127221702525716e-08, "loss": 0.0704, "step": 3285 }, { "epoch": 0.44, "learning_rate": 5.611385807831084e-08, "loss": 0.0794, "step": 3286 }, { "epoch": 0.44, "learning_rate": 5.610049445409595e-08, "loss": 0.0238, "step": 3287 }, { "epoch": 0.44, "learning_rate": 5.608713082988106e-08, "loss": 0.1039, "step": 3288 }, { "epoch": 0.44, "learning_rate": 5.6073767205666176e-08, "loss": 0.0483, "step": 3289 }, { "epoch": 0.44, "learning_rate": 5.6060403581451285e-08, "loss": 0.0313, "step": 3290 }, { "epoch": 0.44, "learning_rate": 5.6047039957236395e-08, "loss": 0.0634, "step": 3291 }, { "epoch": 0.44, "learning_rate": 5.603367633302152e-08, "loss": 0.0933, "step": 3292 }, { "epoch": 0.44, "learning_rate": 5.602031270880663e-08, "loss": 0.0759, "step": 3293 }, { "epoch": 0.44, "learning_rate": 5.600694908459174e-08, "loss": 0.0328, "step": 3294 }, { "epoch": 0.44, "learning_rate": 5.599358546037685e-08, "loss": 0.0548, "step": 3295 }, { "epoch": 0.44, "learning_rate": 5.5980221836161964e-08, "loss": 0.0569, "step": 3296 }, { "epoch": 0.44, "learning_rate": 5.5966858211947074e-08, "loss": 0.0378, "step": 3297 }, { "epoch": 0.44, "learning_rate": 5.5953494587732184e-08, "loss": 0.0518, "step": 3298 }, { "epoch": 0.44, "learning_rate": 5.594013096351731e-08, "loss": 0.0257, "step": 3299 }, { "epoch": 0.44, "learning_rate": 5.592676733930242e-08, "loss": 0.0604, "step": 3300 }, { "epoch": 0.44, "learning_rate": 5.5913403715087527e-08, "loss": 0.0471, "step": 3301 }, { "epoch": 0.44, "learning_rate": 5.590004009087265e-08, "loss": 0.0418, "step": 3302 }, { "epoch": 0.44, "learning_rate": 5.588667646665775e-08, "loss": 0.0596, "step": 3303 }, { "epoch": 0.44, "learning_rate": 5.587331284244286e-08, "loss": 0.0319, "step": 3304 }, { "epoch": 0.44, "learning_rate": 5.5859949218227986e-08, "loss": 0.0515, "step": 3305 }, { "epoch": 0.44, "learning_rate": 5.5846585594013096e-08, "loss": 0.0302, "step": 3306 }, { "epoch": 0.44, "learning_rate": 5.5833221969798206e-08, "loss": 0.0358, "step": 3307 }, { "epoch": 0.44, "learning_rate": 5.5819858345583315e-08, "loss": 0.0515, "step": 3308 }, { "epoch": 0.44, "learning_rate": 5.580649472136844e-08, "loss": 0.0452, "step": 3309 }, { "epoch": 0.44, "learning_rate": 5.579313109715355e-08, "loss": 0.062, "step": 3310 }, { "epoch": 0.44, "learning_rate": 5.577976747293865e-08, "loss": 0.0778, "step": 3311 }, { "epoch": 0.44, "learning_rate": 5.5766403848723775e-08, "loss": 0.0404, "step": 3312 }, { "epoch": 0.44, "learning_rate": 5.5753040224508884e-08, "loss": 0.0608, "step": 3313 }, { "epoch": 0.44, "learning_rate": 5.5739676600293994e-08, "loss": 0.0378, "step": 3314 }, { "epoch": 0.44, "learning_rate": 5.572631297607912e-08, "loss": 0.0764, "step": 3315 }, { "epoch": 0.44, "learning_rate": 5.571294935186423e-08, "loss": 0.0439, "step": 3316 }, { "epoch": 0.44, "learning_rate": 5.569958572764934e-08, "loss": 0.0458, "step": 3317 }, { "epoch": 0.44, "learning_rate": 5.568622210343445e-08, "loss": 0.0527, "step": 3318 }, { "epoch": 0.44, "learning_rate": 5.567285847921956e-08, "loss": 0.0523, "step": 3319 }, { "epoch": 0.44, "learning_rate": 5.565949485500467e-08, "loss": 0.0566, "step": 3320 }, { "epoch": 0.44, "learning_rate": 5.564613123078978e-08, "loss": 0.044, "step": 3321 }, { "epoch": 0.44, "learning_rate": 5.5632767606574906e-08, "loss": 0.0574, "step": 3322 }, { "epoch": 0.44, "learning_rate": 5.5619403982360016e-08, "loss": 0.0303, "step": 3323 }, { "epoch": 0.44, "learning_rate": 5.5606040358145126e-08, "loss": 0.048, "step": 3324 }, { "epoch": 0.44, "learning_rate": 5.559267673393024e-08, "loss": 0.0644, "step": 3325 }, { "epoch": 0.44, "learning_rate": 5.557931310971535e-08, "loss": 0.036, "step": 3326 }, { "epoch": 0.44, "learning_rate": 5.556594948550046e-08, "loss": 0.0519, "step": 3327 }, { "epoch": 0.44, "learning_rate": 5.5552585861285585e-08, "loss": 0.0516, "step": 3328 }, { "epoch": 0.44, "learning_rate": 5.5539222237070695e-08, "loss": 0.0506, "step": 3329 }, { "epoch": 0.45, "learning_rate": 5.5525858612855804e-08, "loss": 0.0367, "step": 3330 }, { "epoch": 0.45, "learning_rate": 5.5512494988640914e-08, "loss": 0.0451, "step": 3331 }, { "epoch": 0.45, "learning_rate": 5.549913136442603e-08, "loss": 0.0472, "step": 3332 }, { "epoch": 0.45, "learning_rate": 5.548576774021114e-08, "loss": 0.0694, "step": 3333 }, { "epoch": 0.45, "learning_rate": 5.547240411599625e-08, "loss": 0.0639, "step": 3334 }, { "epoch": 0.45, "learning_rate": 5.5459040491781373e-08, "loss": 0.0395, "step": 3335 }, { "epoch": 0.45, "learning_rate": 5.5445676867566483e-08, "loss": 0.0615, "step": 3336 }, { "epoch": 0.45, "learning_rate": 5.543231324335159e-08, "loss": 0.0721, "step": 3337 }, { "epoch": 0.45, "learning_rate": 5.541894961913671e-08, "loss": 0.0615, "step": 3338 }, { "epoch": 0.45, "learning_rate": 5.540558599492182e-08, "loss": 0.0387, "step": 3339 }, { "epoch": 0.45, "learning_rate": 5.539222237070693e-08, "loss": 0.061, "step": 3340 }, { "epoch": 0.45, "learning_rate": 5.537885874649204e-08, "loss": 0.055, "step": 3341 }, { "epoch": 0.45, "learning_rate": 5.536549512227716e-08, "loss": 0.0631, "step": 3342 }, { "epoch": 0.45, "learning_rate": 5.535213149806227e-08, "loss": 0.0722, "step": 3343 }, { "epoch": 0.45, "learning_rate": 5.533876787384738e-08, "loss": 0.0508, "step": 3344 }, { "epoch": 0.45, "learning_rate": 5.53254042496325e-08, "loss": 0.0376, "step": 3345 }, { "epoch": 0.45, "learning_rate": 5.531204062541761e-08, "loss": 0.0475, "step": 3346 }, { "epoch": 0.45, "learning_rate": 5.529867700120272e-08, "loss": 0.0428, "step": 3347 }, { "epoch": 0.45, "learning_rate": 5.528531337698784e-08, "loss": 0.0468, "step": 3348 }, { "epoch": 0.45, "learning_rate": 5.527194975277295e-08, "loss": 0.0469, "step": 3349 }, { "epoch": 0.45, "learning_rate": 5.525858612855806e-08, "loss": 0.0487, "step": 3350 }, { "epoch": 0.45, "learning_rate": 5.5245222504343184e-08, "loss": 0.0544, "step": 3351 }, { "epoch": 0.45, "learning_rate": 5.5231858880128294e-08, "loss": 0.06, "step": 3352 }, { "epoch": 0.45, "learning_rate": 5.52184952559134e-08, "loss": 0.0532, "step": 3353 }, { "epoch": 0.45, "learning_rate": 5.5205131631698507e-08, "loss": 0.0602, "step": 3354 }, { "epoch": 0.45, "learning_rate": 5.519176800748363e-08, "loss": 0.0807, "step": 3355 }, { "epoch": 0.45, "learning_rate": 5.517840438326874e-08, "loss": 0.0402, "step": 3356 }, { "epoch": 0.45, "learning_rate": 5.516504075905385e-08, "loss": 0.0713, "step": 3357 }, { "epoch": 0.45, "learning_rate": 5.515167713483897e-08, "loss": 0.0443, "step": 3358 }, { "epoch": 0.45, "learning_rate": 5.513831351062408e-08, "loss": 0.0571, "step": 3359 }, { "epoch": 0.45, "learning_rate": 5.512494988640919e-08, "loss": 0.0429, "step": 3360 }, { "epoch": 0.45, "learning_rate": 5.511158626219431e-08, "loss": 0.0528, "step": 3361 }, { "epoch": 0.45, "learning_rate": 5.509822263797942e-08, "loss": 0.0519, "step": 3362 }, { "epoch": 0.45, "learning_rate": 5.508485901376453e-08, "loss": 0.0376, "step": 3363 }, { "epoch": 0.45, "learning_rate": 5.507149538954964e-08, "loss": 0.0731, "step": 3364 }, { "epoch": 0.45, "learning_rate": 5.505813176533476e-08, "loss": 0.0384, "step": 3365 }, { "epoch": 0.45, "learning_rate": 5.504476814111987e-08, "loss": 0.0646, "step": 3366 }, { "epoch": 0.45, "learning_rate": 5.503140451690498e-08, "loss": 0.0564, "step": 3367 }, { "epoch": 0.45, "learning_rate": 5.50180408926901e-08, "loss": 0.0458, "step": 3368 }, { "epoch": 0.45, "learning_rate": 5.500467726847521e-08, "loss": 0.0588, "step": 3369 }, { "epoch": 0.45, "learning_rate": 5.499131364426032e-08, "loss": 0.0574, "step": 3370 }, { "epoch": 0.45, "learning_rate": 5.497795002004544e-08, "loss": 0.0708, "step": 3371 }, { "epoch": 0.45, "learning_rate": 5.496458639583055e-08, "loss": 0.0662, "step": 3372 }, { "epoch": 0.45, "learning_rate": 5.495122277161566e-08, "loss": 0.0549, "step": 3373 }, { "epoch": 0.45, "learning_rate": 5.4937859147400776e-08, "loss": 0.0495, "step": 3374 }, { "epoch": 0.45, "learning_rate": 5.4924495523185886e-08, "loss": 0.0624, "step": 3375 }, { "epoch": 0.45, "learning_rate": 5.4911131898970996e-08, "loss": 0.0512, "step": 3376 }, { "epoch": 0.45, "learning_rate": 5.4897768274756106e-08, "loss": 0.0632, "step": 3377 }, { "epoch": 0.45, "learning_rate": 5.488440465054123e-08, "loss": 0.0397, "step": 3378 }, { "epoch": 0.45, "learning_rate": 5.487104102632634e-08, "loss": 0.0492, "step": 3379 }, { "epoch": 0.45, "learning_rate": 5.485767740211145e-08, "loss": 0.0661, "step": 3380 }, { "epoch": 0.45, "learning_rate": 5.4844313777896565e-08, "loss": 0.0467, "step": 3381 }, { "epoch": 0.45, "learning_rate": 5.4830950153681675e-08, "loss": 0.0505, "step": 3382 }, { "epoch": 0.45, "learning_rate": 5.4817586529466784e-08, "loss": 0.033, "step": 3383 }, { "epoch": 0.45, "learning_rate": 5.480422290525191e-08, "loss": 0.0818, "step": 3384 }, { "epoch": 0.45, "learning_rate": 5.479085928103702e-08, "loss": 0.0309, "step": 3385 }, { "epoch": 0.45, "learning_rate": 5.477749565682213e-08, "loss": 0.0407, "step": 3386 }, { "epoch": 0.45, "learning_rate": 5.476413203260724e-08, "loss": 0.0405, "step": 3387 }, { "epoch": 0.45, "learning_rate": 5.4750768408392353e-08, "loss": 0.036, "step": 3388 }, { "epoch": 0.45, "learning_rate": 5.4737404784177463e-08, "loss": 0.0386, "step": 3389 }, { "epoch": 0.45, "learning_rate": 5.472404115996257e-08, "loss": 0.0395, "step": 3390 }, { "epoch": 0.45, "learning_rate": 5.4710677535747696e-08, "loss": 0.0321, "step": 3391 }, { "epoch": 0.45, "learning_rate": 5.4697313911532806e-08, "loss": 0.0566, "step": 3392 }, { "epoch": 0.45, "learning_rate": 5.4683950287317916e-08, "loss": 0.0497, "step": 3393 }, { "epoch": 0.45, "learning_rate": 5.467058666310303e-08, "loss": 0.0694, "step": 3394 }, { "epoch": 0.45, "learning_rate": 5.465722303888814e-08, "loss": 0.04, "step": 3395 }, { "epoch": 0.45, "learning_rate": 5.464385941467325e-08, "loss": 0.0407, "step": 3396 }, { "epoch": 0.45, "learning_rate": 5.4630495790458375e-08, "loss": 0.0864, "step": 3397 }, { "epoch": 0.45, "learning_rate": 5.4617132166243485e-08, "loss": 0.0552, "step": 3398 }, { "epoch": 0.45, "learning_rate": 5.4603768542028595e-08, "loss": 0.0548, "step": 3399 }, { "epoch": 0.45, "learning_rate": 5.4590404917813705e-08, "loss": 0.0428, "step": 3400 }, { "epoch": 0.45, "learning_rate": 5.457704129359883e-08, "loss": 0.0348, "step": 3401 }, { "epoch": 0.45, "learning_rate": 5.456367766938393e-08, "loss": 0.0462, "step": 3402 }, { "epoch": 0.45, "learning_rate": 5.455031404516904e-08, "loss": 0.0488, "step": 3403 }, { "epoch": 0.45, "learning_rate": 5.4536950420954164e-08, "loss": 0.0406, "step": 3404 }, { "epoch": 0.46, "learning_rate": 5.4523586796739274e-08, "loss": 0.0711, "step": 3405 }, { "epoch": 0.46, "learning_rate": 5.4510223172524383e-08, "loss": 0.0484, "step": 3406 }, { "epoch": 0.46, "learning_rate": 5.4496859548309506e-08, "loss": 0.0336, "step": 3407 }, { "epoch": 0.46, "learning_rate": 5.4483495924094616e-08, "loss": 0.0323, "step": 3408 }, { "epoch": 0.46, "learning_rate": 5.4470132299879726e-08, "loss": 0.0476, "step": 3409 }, { "epoch": 0.46, "learning_rate": 5.445676867566483e-08, "loss": 0.0559, "step": 3410 }, { "epoch": 0.46, "learning_rate": 5.444340505144995e-08, "loss": 0.0501, "step": 3411 }, { "epoch": 0.46, "learning_rate": 5.443004142723506e-08, "loss": 0.0327, "step": 3412 }, { "epoch": 0.46, "learning_rate": 5.441667780302017e-08, "loss": 0.0437, "step": 3413 }, { "epoch": 0.46, "learning_rate": 5.4403314178805295e-08, "loss": 0.0552, "step": 3414 }, { "epoch": 0.46, "learning_rate": 5.4389950554590405e-08, "loss": 0.0499, "step": 3415 }, { "epoch": 0.46, "learning_rate": 5.4376586930375515e-08, "loss": 0.0803, "step": 3416 }, { "epoch": 0.46, "learning_rate": 5.436322330616063e-08, "loss": 0.0534, "step": 3417 }, { "epoch": 0.46, "learning_rate": 5.434985968194574e-08, "loss": 0.0601, "step": 3418 }, { "epoch": 0.46, "learning_rate": 5.433649605773085e-08, "loss": 0.0566, "step": 3419 }, { "epoch": 0.46, "learning_rate": 5.4323132433515974e-08, "loss": 0.038, "step": 3420 }, { "epoch": 0.46, "learning_rate": 5.4309768809301084e-08, "loss": 0.0431, "step": 3421 }, { "epoch": 0.46, "learning_rate": 5.4296405185086194e-08, "loss": 0.0522, "step": 3422 }, { "epoch": 0.46, "learning_rate": 5.4283041560871304e-08, "loss": 0.0712, "step": 3423 }, { "epoch": 0.46, "learning_rate": 5.426967793665642e-08, "loss": 0.0794, "step": 3424 }, { "epoch": 0.46, "learning_rate": 5.425631431244153e-08, "loss": 0.0465, "step": 3425 }, { "epoch": 0.46, "learning_rate": 5.424295068822664e-08, "loss": 0.0539, "step": 3426 }, { "epoch": 0.46, "learning_rate": 5.422958706401176e-08, "loss": 0.0562, "step": 3427 }, { "epoch": 0.46, "learning_rate": 5.421622343979687e-08, "loss": 0.0193, "step": 3428 }, { "epoch": 0.46, "learning_rate": 5.420285981558198e-08, "loss": 0.0528, "step": 3429 }, { "epoch": 0.46, "learning_rate": 5.41894961913671e-08, "loss": 0.0377, "step": 3430 }, { "epoch": 0.46, "learning_rate": 5.417613256715221e-08, "loss": 0.0578, "step": 3431 }, { "epoch": 0.46, "learning_rate": 5.416276894293732e-08, "loss": 0.0262, "step": 3432 }, { "epoch": 0.46, "learning_rate": 5.414940531872243e-08, "loss": 0.0682, "step": 3433 }, { "epoch": 0.46, "learning_rate": 5.413604169450755e-08, "loss": 0.0313, "step": 3434 }, { "epoch": 0.46, "learning_rate": 5.412267807029266e-08, "loss": 0.08, "step": 3435 }, { "epoch": 0.46, "learning_rate": 5.410931444607777e-08, "loss": 0.0817, "step": 3436 }, { "epoch": 0.46, "learning_rate": 5.409595082186289e-08, "loss": 0.0363, "step": 3437 }, { "epoch": 0.46, "learning_rate": 5.4082587197648e-08, "loss": 0.0502, "step": 3438 }, { "epoch": 0.46, "learning_rate": 5.406922357343311e-08, "loss": 0.0422, "step": 3439 }, { "epoch": 0.46, "learning_rate": 5.405585994921823e-08, "loss": 0.0376, "step": 3440 }, { "epoch": 0.46, "learning_rate": 5.404249632500334e-08, "loss": 0.0584, "step": 3441 }, { "epoch": 0.46, "learning_rate": 5.402913270078845e-08, "loss": 0.0457, "step": 3442 }, { "epoch": 0.46, "learning_rate": 5.4015769076573566e-08, "loss": 0.0435, "step": 3443 }, { "epoch": 0.46, "learning_rate": 5.4002405452358676e-08, "loss": 0.0326, "step": 3444 }, { "epoch": 0.46, "learning_rate": 5.3989041828143786e-08, "loss": 0.0474, "step": 3445 }, { "epoch": 0.46, "learning_rate": 5.3975678203928896e-08, "loss": 0.0512, "step": 3446 }, { "epoch": 0.46, "learning_rate": 5.396231457971402e-08, "loss": 0.0608, "step": 3447 }, { "epoch": 0.46, "learning_rate": 5.394895095549913e-08, "loss": 0.0794, "step": 3448 }, { "epoch": 0.46, "learning_rate": 5.393558733128424e-08, "loss": 0.0751, "step": 3449 }, { "epoch": 0.46, "learning_rate": 5.392222370706936e-08, "loss": 0.0478, "step": 3450 }, { "epoch": 0.46, "learning_rate": 5.3908860082854465e-08, "loss": 0.0693, "step": 3451 }, { "epoch": 0.46, "learning_rate": 5.3895496458639575e-08, "loss": 0.0426, "step": 3452 }, { "epoch": 0.46, "learning_rate": 5.38821328344247e-08, "loss": 0.0518, "step": 3453 }, { "epoch": 0.46, "learning_rate": 5.386876921020981e-08, "loss": 0.0548, "step": 3454 }, { "epoch": 0.46, "learning_rate": 5.385540558599492e-08, "loss": 0.0715, "step": 3455 }, { "epoch": 0.46, "learning_rate": 5.384204196178003e-08, "loss": 0.0595, "step": 3456 }, { "epoch": 0.46, "learning_rate": 5.382867833756515e-08, "loss": 0.0637, "step": 3457 }, { "epoch": 0.46, "learning_rate": 5.381531471335026e-08, "loss": 0.0435, "step": 3458 }, { "epoch": 0.46, "learning_rate": 5.380195108913537e-08, "loss": 0.0509, "step": 3459 }, { "epoch": 0.46, "learning_rate": 5.3788587464920486e-08, "loss": 0.0556, "step": 3460 }, { "epoch": 0.46, "learning_rate": 5.3775223840705596e-08, "loss": 0.0725, "step": 3461 }, { "epoch": 0.46, "learning_rate": 5.3761860216490706e-08, "loss": 0.0574, "step": 3462 }, { "epoch": 0.46, "learning_rate": 5.374849659227583e-08, "loss": 0.0819, "step": 3463 }, { "epoch": 0.46, "learning_rate": 5.373513296806094e-08, "loss": 0.0519, "step": 3464 }, { "epoch": 0.46, "learning_rate": 5.372176934384605e-08, "loss": 0.0357, "step": 3465 }, { "epoch": 0.46, "learning_rate": 5.3708405719631165e-08, "loss": 0.0509, "step": 3466 }, { "epoch": 0.46, "learning_rate": 5.3695042095416275e-08, "loss": 0.0283, "step": 3467 }, { "epoch": 0.46, "learning_rate": 5.3681678471201385e-08, "loss": 0.0584, "step": 3468 }, { "epoch": 0.46, "learning_rate": 5.3668314846986495e-08, "loss": 0.0259, "step": 3469 }, { "epoch": 0.46, "learning_rate": 5.365495122277162e-08, "loss": 0.0821, "step": 3470 }, { "epoch": 0.46, "learning_rate": 5.364158759855673e-08, "loss": 0.0476, "step": 3471 }, { "epoch": 0.46, "learning_rate": 5.362822397434184e-08, "loss": 0.0668, "step": 3472 }, { "epoch": 0.46, "learning_rate": 5.3614860350126954e-08, "loss": 0.0608, "step": 3473 }, { "epoch": 0.46, "learning_rate": 5.3601496725912064e-08, "loss": 0.0747, "step": 3474 }, { "epoch": 0.46, "learning_rate": 5.3588133101697174e-08, "loss": 0.0571, "step": 3475 }, { "epoch": 0.46, "learning_rate": 5.35747694774823e-08, "loss": 0.0694, "step": 3476 }, { "epoch": 0.46, "learning_rate": 5.3561405853267407e-08, "loss": 0.041, "step": 3477 }, { "epoch": 0.46, "learning_rate": 5.3548042229052516e-08, "loss": 0.0452, "step": 3478 }, { "epoch": 0.46, "learning_rate": 5.3534678604837626e-08, "loss": 0.0506, "step": 3479 }, { "epoch": 0.47, "learning_rate": 5.352131498062274e-08, "loss": 0.0499, "step": 3480 }, { "epoch": 0.47, "learning_rate": 5.350795135640785e-08, "loss": 0.0413, "step": 3481 }, { "epoch": 0.47, "learning_rate": 5.349458773219296e-08, "loss": 0.0506, "step": 3482 }, { "epoch": 0.47, "learning_rate": 5.3481224107978085e-08, "loss": 0.0496, "step": 3483 }, { "epoch": 0.47, "learning_rate": 5.3467860483763195e-08, "loss": 0.07, "step": 3484 }, { "epoch": 0.47, "learning_rate": 5.3454496859548305e-08, "loss": 0.0472, "step": 3485 }, { "epoch": 0.47, "learning_rate": 5.344113323533342e-08, "loss": 0.0497, "step": 3486 }, { "epoch": 0.47, "learning_rate": 5.342776961111853e-08, "loss": 0.0438, "step": 3487 }, { "epoch": 0.47, "learning_rate": 5.341440598690364e-08, "loss": 0.0276, "step": 3488 }, { "epoch": 0.47, "learning_rate": 5.3401042362688764e-08, "loss": 0.0483, "step": 3489 }, { "epoch": 0.47, "learning_rate": 5.3387678738473874e-08, "loss": 0.0485, "step": 3490 }, { "epoch": 0.47, "learning_rate": 5.3374315114258984e-08, "loss": 0.0443, "step": 3491 }, { "epoch": 0.47, "learning_rate": 5.3360951490044094e-08, "loss": 0.0463, "step": 3492 }, { "epoch": 0.47, "learning_rate": 5.334758786582921e-08, "loss": 0.0508, "step": 3493 }, { "epoch": 0.47, "learning_rate": 5.333422424161432e-08, "loss": 0.0601, "step": 3494 }, { "epoch": 0.47, "learning_rate": 5.332086061739943e-08, "loss": 0.0507, "step": 3495 }, { "epoch": 0.47, "learning_rate": 5.330749699318455e-08, "loss": 0.0589, "step": 3496 }, { "epoch": 0.47, "learning_rate": 5.329413336896966e-08, "loss": 0.0628, "step": 3497 }, { "epoch": 0.47, "learning_rate": 5.328076974475477e-08, "loss": 0.0539, "step": 3498 }, { "epoch": 0.47, "learning_rate": 5.3267406120539896e-08, "loss": 0.0394, "step": 3499 }, { "epoch": 0.47, "learning_rate": 5.3254042496325006e-08, "loss": 0.0666, "step": 3500 }, { "epoch": 0.47, "learning_rate": 5.324067887211011e-08, "loss": 0.0493, "step": 3501 }, { "epoch": 0.47, "learning_rate": 5.322731524789522e-08, "loss": 0.0486, "step": 3502 }, { "epoch": 0.47, "learning_rate": 5.321395162368034e-08, "loss": 0.0389, "step": 3503 }, { "epoch": 0.47, "learning_rate": 5.320058799946545e-08, "loss": 0.0472, "step": 3504 }, { "epoch": 0.47, "learning_rate": 5.318722437525056e-08, "loss": 0.0425, "step": 3505 }, { "epoch": 0.47, "learning_rate": 5.3173860751035684e-08, "loss": 0.049, "step": 3506 }, { "epoch": 0.47, "learning_rate": 5.3160497126820794e-08, "loss": 0.048, "step": 3507 }, { "epoch": 0.47, "learning_rate": 5.3147133502605904e-08, "loss": 0.0369, "step": 3508 }, { "epoch": 0.47, "learning_rate": 5.313376987839102e-08, "loss": 0.0405, "step": 3509 }, { "epoch": 0.47, "learning_rate": 5.312040625417613e-08, "loss": 0.0656, "step": 3510 }, { "epoch": 0.47, "learning_rate": 5.310704262996124e-08, "loss": 0.0501, "step": 3511 }, { "epoch": 0.47, "learning_rate": 5.309367900574636e-08, "loss": 0.0627, "step": 3512 }, { "epoch": 0.47, "learning_rate": 5.308031538153147e-08, "loss": 0.0339, "step": 3513 }, { "epoch": 0.47, "learning_rate": 5.306695175731658e-08, "loss": 0.0644, "step": 3514 }, { "epoch": 0.47, "learning_rate": 5.305358813310169e-08, "loss": 0.0442, "step": 3515 }, { "epoch": 0.47, "learning_rate": 5.304022450888681e-08, "loss": 0.0289, "step": 3516 }, { "epoch": 0.47, "learning_rate": 5.302686088467192e-08, "loss": 0.0722, "step": 3517 }, { "epoch": 0.47, "learning_rate": 5.301349726045703e-08, "loss": 0.0501, "step": 3518 }, { "epoch": 0.47, "learning_rate": 5.300013363624215e-08, "loss": 0.0774, "step": 3519 }, { "epoch": 0.47, "learning_rate": 5.298677001202726e-08, "loss": 0.0522, "step": 3520 }, { "epoch": 0.47, "learning_rate": 5.297340638781237e-08, "loss": 0.0579, "step": 3521 }, { "epoch": 0.47, "learning_rate": 5.296004276359749e-08, "loss": 0.0373, "step": 3522 }, { "epoch": 0.47, "learning_rate": 5.29466791393826e-08, "loss": 0.0705, "step": 3523 }, { "epoch": 0.47, "learning_rate": 5.293331551516771e-08, "loss": 0.0548, "step": 3524 }, { "epoch": 0.47, "learning_rate": 5.291995189095282e-08, "loss": 0.0436, "step": 3525 }, { "epoch": 0.47, "learning_rate": 5.290658826673794e-08, "loss": 0.0496, "step": 3526 }, { "epoch": 0.47, "learning_rate": 5.289322464252305e-08, "loss": 0.0389, "step": 3527 }, { "epoch": 0.47, "learning_rate": 5.287986101830816e-08, "loss": 0.0246, "step": 3528 }, { "epoch": 0.47, "learning_rate": 5.286649739409328e-08, "loss": 0.0383, "step": 3529 }, { "epoch": 0.47, "learning_rate": 5.2853133769878387e-08, "loss": 0.0391, "step": 3530 }, { "epoch": 0.47, "learning_rate": 5.2839770145663496e-08, "loss": 0.0488, "step": 3531 }, { "epoch": 0.47, "learning_rate": 5.282640652144862e-08, "loss": 0.0577, "step": 3532 }, { "epoch": 0.47, "learning_rate": 5.281304289723373e-08, "loss": 0.06, "step": 3533 }, { "epoch": 0.47, "learning_rate": 5.279967927301884e-08, "loss": 0.0766, "step": 3534 }, { "epoch": 0.47, "learning_rate": 5.2786315648803956e-08, "loss": 0.0469, "step": 3535 }, { "epoch": 0.47, "learning_rate": 5.2772952024589065e-08, "loss": 0.0404, "step": 3536 }, { "epoch": 0.47, "learning_rate": 5.2759588400374175e-08, "loss": 0.0497, "step": 3537 }, { "epoch": 0.47, "learning_rate": 5.2746224776159285e-08, "loss": 0.0489, "step": 3538 }, { "epoch": 0.47, "learning_rate": 5.273286115194441e-08, "loss": 0.0412, "step": 3539 }, { "epoch": 0.47, "learning_rate": 5.271949752772952e-08, "loss": 0.0475, "step": 3540 }, { "epoch": 0.47, "learning_rate": 5.270613390351463e-08, "loss": 0.0329, "step": 3541 }, { "epoch": 0.47, "learning_rate": 5.2692770279299744e-08, "loss": 0.0381, "step": 3542 }, { "epoch": 0.47, "learning_rate": 5.2679406655084854e-08, "loss": 0.0519, "step": 3543 }, { "epoch": 0.47, "learning_rate": 5.2666043030869964e-08, "loss": 0.0464, "step": 3544 }, { "epoch": 0.47, "learning_rate": 5.265267940665509e-08, "loss": 0.0332, "step": 3545 }, { "epoch": 0.47, "learning_rate": 5.26393157824402e-08, "loss": 0.0574, "step": 3546 }, { "epoch": 0.47, "learning_rate": 5.2625952158225307e-08, "loss": 0.067, "step": 3547 }, { "epoch": 0.47, "learning_rate": 5.2612588534010417e-08, "loss": 0.0452, "step": 3548 }, { "epoch": 0.47, "learning_rate": 5.259922490979554e-08, "loss": 0.0574, "step": 3549 }, { "epoch": 0.47, "learning_rate": 5.258586128558064e-08, "loss": 0.0687, "step": 3550 }, { "epoch": 0.47, "learning_rate": 5.257249766136575e-08, "loss": 0.0426, "step": 3551 }, { "epoch": 0.47, "learning_rate": 5.2559134037150876e-08, "loss": 0.0416, "step": 3552 }, { "epoch": 0.47, "learning_rate": 5.2545770412935986e-08, "loss": 0.0438, "step": 3553 }, { "epoch": 0.47, "learning_rate": 5.2532406788721095e-08, "loss": 0.0433, "step": 3554 }, { "epoch": 0.48, "learning_rate": 5.251904316450622e-08, "loss": 0.0332, "step": 3555 }, { "epoch": 0.48, "learning_rate": 5.250567954029133e-08, "loss": 0.046, "step": 3556 }, { "epoch": 0.48, "learning_rate": 5.249231591607644e-08, "loss": 0.0682, "step": 3557 }, { "epoch": 0.48, "learning_rate": 5.2478952291861555e-08, "loss": 0.0308, "step": 3558 }, { "epoch": 0.48, "learning_rate": 5.2465588667646664e-08, "loss": 0.0453, "step": 3559 }, { "epoch": 0.48, "learning_rate": 5.2452225043431774e-08, "loss": 0.0578, "step": 3560 }, { "epoch": 0.48, "learning_rate": 5.2438861419216884e-08, "loss": 0.0328, "step": 3561 }, { "epoch": 0.48, "learning_rate": 5.242549779500201e-08, "loss": 0.0729, "step": 3562 }, { "epoch": 0.48, "learning_rate": 5.241213417078712e-08, "loss": 0.0704, "step": 3563 }, { "epoch": 0.48, "learning_rate": 5.239877054657223e-08, "loss": 0.0426, "step": 3564 }, { "epoch": 0.48, "learning_rate": 5.238540692235734e-08, "loss": 0.085, "step": 3565 }, { "epoch": 0.48, "learning_rate": 5.237204329814245e-08, "loss": 0.0417, "step": 3566 }, { "epoch": 0.48, "learning_rate": 5.235867967392756e-08, "loss": 0.0789, "step": 3567 }, { "epoch": 0.48, "learning_rate": 5.2345316049712686e-08, "loss": 0.0465, "step": 3568 }, { "epoch": 0.48, "learning_rate": 5.2331952425497796e-08, "loss": 0.0674, "step": 3569 }, { "epoch": 0.48, "learning_rate": 5.2318588801282906e-08, "loss": 0.0697, "step": 3570 }, { "epoch": 0.48, "learning_rate": 5.2305225177068015e-08, "loss": 0.0406, "step": 3571 }, { "epoch": 0.48, "learning_rate": 5.229186155285313e-08, "loss": 0.0502, "step": 3572 }, { "epoch": 0.48, "learning_rate": 5.227849792863824e-08, "loss": 0.0449, "step": 3573 }, { "epoch": 0.48, "learning_rate": 5.226513430442335e-08, "loss": 0.0403, "step": 3574 }, { "epoch": 0.48, "learning_rate": 5.2251770680208475e-08, "loss": 0.0308, "step": 3575 }, { "epoch": 0.48, "learning_rate": 5.2238407055993584e-08, "loss": 0.0528, "step": 3576 }, { "epoch": 0.48, "learning_rate": 5.2225043431778694e-08, "loss": 0.0489, "step": 3577 }, { "epoch": 0.48, "learning_rate": 5.221167980756381e-08, "loss": 0.0846, "step": 3578 }, { "epoch": 0.48, "learning_rate": 5.219831618334892e-08, "loss": 0.0624, "step": 3579 }, { "epoch": 0.48, "learning_rate": 5.218495255913403e-08, "loss": 0.0332, "step": 3580 }, { "epoch": 0.48, "learning_rate": 5.2171588934919154e-08, "loss": 0.0295, "step": 3581 }, { "epoch": 0.48, "learning_rate": 5.2158225310704263e-08, "loss": 0.0452, "step": 3582 }, { "epoch": 0.48, "learning_rate": 5.214486168648937e-08, "loss": 0.0529, "step": 3583 }, { "epoch": 0.48, "learning_rate": 5.213149806227448e-08, "loss": 0.0715, "step": 3584 }, { "epoch": 0.48, "learning_rate": 5.21181344380596e-08, "loss": 0.041, "step": 3585 }, { "epoch": 0.48, "learning_rate": 5.210477081384471e-08, "loss": 0.0874, "step": 3586 }, { "epoch": 0.48, "learning_rate": 5.209140718962982e-08, "loss": 0.0467, "step": 3587 }, { "epoch": 0.48, "learning_rate": 5.207804356541494e-08, "loss": 0.0547, "step": 3588 }, { "epoch": 0.48, "learning_rate": 5.206467994120005e-08, "loss": 0.0578, "step": 3589 }, { "epoch": 0.48, "learning_rate": 5.205131631698516e-08, "loss": 0.072, "step": 3590 }, { "epoch": 0.48, "learning_rate": 5.2037952692770285e-08, "loss": 0.0485, "step": 3591 }, { "epoch": 0.48, "learning_rate": 5.202458906855539e-08, "loss": 0.0537, "step": 3592 }, { "epoch": 0.48, "learning_rate": 5.20112254443405e-08, "loss": 0.0471, "step": 3593 }, { "epoch": 0.48, "learning_rate": 5.199786182012561e-08, "loss": 0.0616, "step": 3594 }, { "epoch": 0.48, "learning_rate": 5.198449819591073e-08, "loss": 0.053, "step": 3595 }, { "epoch": 0.48, "learning_rate": 5.197113457169584e-08, "loss": 0.0622, "step": 3596 }, { "epoch": 0.48, "learning_rate": 5.195777094748095e-08, "loss": 0.0451, "step": 3597 }, { "epoch": 0.48, "learning_rate": 5.1944407323266074e-08, "loss": 0.0313, "step": 3598 }, { "epoch": 0.48, "learning_rate": 5.1931043699051183e-08, "loss": 0.0447, "step": 3599 }, { "epoch": 0.48, "learning_rate": 5.1917680074836287e-08, "loss": 0.0455, "step": 3600 }, { "epoch": 0.48, "learning_rate": 5.190431645062141e-08, "loss": 0.0446, "step": 3601 }, { "epoch": 0.48, "learning_rate": 5.189095282640652e-08, "loss": 0.0416, "step": 3602 }, { "epoch": 0.48, "learning_rate": 5.187758920219163e-08, "loss": 0.0622, "step": 3603 }, { "epoch": 0.48, "learning_rate": 5.186422557797675e-08, "loss": 0.0651, "step": 3604 }, { "epoch": 0.48, "learning_rate": 5.185086195376186e-08, "loss": 0.0576, "step": 3605 }, { "epoch": 0.48, "learning_rate": 5.183749832954697e-08, "loss": 0.0492, "step": 3606 }, { "epoch": 0.48, "learning_rate": 5.182413470533208e-08, "loss": 0.0356, "step": 3607 }, { "epoch": 0.48, "learning_rate": 5.18107710811172e-08, "loss": 0.058, "step": 3608 }, { "epoch": 0.48, "learning_rate": 5.179740745690231e-08, "loss": 0.055, "step": 3609 }, { "epoch": 0.48, "learning_rate": 5.178404383268742e-08, "loss": 0.0379, "step": 3610 }, { "epoch": 0.48, "learning_rate": 5.177068020847254e-08, "loss": 0.0459, "step": 3611 }, { "epoch": 0.48, "learning_rate": 5.175731658425765e-08, "loss": 0.0299, "step": 3612 }, { "epoch": 0.48, "learning_rate": 5.174395296004276e-08, "loss": 0.0296, "step": 3613 }, { "epoch": 0.48, "learning_rate": 5.173058933582788e-08, "loss": 0.039, "step": 3614 }, { "epoch": 0.48, "learning_rate": 5.171722571161299e-08, "loss": 0.0353, "step": 3615 }, { "epoch": 0.48, "learning_rate": 5.17038620873981e-08, "loss": 0.053, "step": 3616 }, { "epoch": 0.48, "learning_rate": 5.169049846318321e-08, "loss": 0.0755, "step": 3617 }, { "epoch": 0.48, "learning_rate": 5.167713483896833e-08, "loss": 0.0463, "step": 3618 }, { "epoch": 0.48, "learning_rate": 5.166377121475344e-08, "loss": 0.0567, "step": 3619 }, { "epoch": 0.48, "learning_rate": 5.165040759053855e-08, "loss": 0.0473, "step": 3620 }, { "epoch": 0.48, "learning_rate": 5.1637043966323666e-08, "loss": 0.0288, "step": 3621 }, { "epoch": 0.48, "learning_rate": 5.1623680342108776e-08, "loss": 0.0539, "step": 3622 }, { "epoch": 0.48, "learning_rate": 5.1610316717893886e-08, "loss": 0.0448, "step": 3623 }, { "epoch": 0.48, "learning_rate": 5.159695309367901e-08, "loss": 0.0954, "step": 3624 }, { "epoch": 0.48, "learning_rate": 5.158358946946412e-08, "loss": 0.0461, "step": 3625 }, { "epoch": 0.48, "learning_rate": 5.157022584524923e-08, "loss": 0.08, "step": 3626 }, { "epoch": 0.48, "learning_rate": 5.1556862221034345e-08, "loss": 0.0455, "step": 3627 }, { "epoch": 0.48, "learning_rate": 5.1543498596819455e-08, "loss": 0.0525, "step": 3628 }, { "epoch": 0.48, "learning_rate": 5.1530134972604564e-08, "loss": 0.0419, "step": 3629 }, { "epoch": 0.49, "learning_rate": 5.1516771348389674e-08, "loss": 0.0427, "step": 3630 }, { "epoch": 0.49, "learning_rate": 5.15034077241748e-08, "loss": 0.059, "step": 3631 }, { "epoch": 0.49, "learning_rate": 5.149004409995991e-08, "loss": 0.042, "step": 3632 }, { "epoch": 0.49, "learning_rate": 5.147668047574502e-08, "loss": 0.0685, "step": 3633 }, { "epoch": 0.49, "learning_rate": 5.1463316851530133e-08, "loss": 0.0372, "step": 3634 }, { "epoch": 0.49, "learning_rate": 5.1449953227315243e-08, "loss": 0.1347, "step": 3635 }, { "epoch": 0.49, "learning_rate": 5.143658960310035e-08, "loss": 0.0406, "step": 3636 }, { "epoch": 0.49, "learning_rate": 5.1423225978885476e-08, "loss": 0.0508, "step": 3637 }, { "epoch": 0.49, "learning_rate": 5.1409862354670586e-08, "loss": 0.0352, "step": 3638 }, { "epoch": 0.49, "learning_rate": 5.1396498730455696e-08, "loss": 0.0393, "step": 3639 }, { "epoch": 0.49, "learning_rate": 5.1383135106240806e-08, "loss": 0.0593, "step": 3640 }, { "epoch": 0.49, "learning_rate": 5.136977148202592e-08, "loss": 0.0639, "step": 3641 }, { "epoch": 0.49, "learning_rate": 5.135640785781103e-08, "loss": 0.062, "step": 3642 }, { "epoch": 0.49, "learning_rate": 5.134304423359614e-08, "loss": 0.0415, "step": 3643 }, { "epoch": 0.49, "learning_rate": 5.1329680609381265e-08, "loss": 0.0486, "step": 3644 }, { "epoch": 0.49, "learning_rate": 5.1316316985166375e-08, "loss": 0.0409, "step": 3645 }, { "epoch": 0.49, "learning_rate": 5.1302953360951485e-08, "loss": 0.0612, "step": 3646 }, { "epoch": 0.49, "learning_rate": 5.128958973673661e-08, "loss": 0.0321, "step": 3647 }, { "epoch": 0.49, "learning_rate": 5.127622611252172e-08, "loss": 0.0845, "step": 3648 }, { "epoch": 0.49, "learning_rate": 5.126286248830682e-08, "loss": 0.0431, "step": 3649 }, { "epoch": 0.49, "learning_rate": 5.1249498864091944e-08, "loss": 0.078, "step": 3650 }, { "epoch": 0.49, "learning_rate": 5.1236135239877054e-08, "loss": 0.043, "step": 3651 }, { "epoch": 0.49, "learning_rate": 5.1222771615662163e-08, "loss": 0.0379, "step": 3652 }, { "epoch": 0.49, "learning_rate": 5.120940799144727e-08, "loss": 0.086, "step": 3653 }, { "epoch": 0.49, "learning_rate": 5.1196044367232396e-08, "loss": 0.0559, "step": 3654 }, { "epoch": 0.49, "learning_rate": 5.1182680743017506e-08, "loss": 0.0481, "step": 3655 }, { "epoch": 0.49, "learning_rate": 5.1169317118802616e-08, "loss": 0.0557, "step": 3656 }, { "epoch": 0.49, "learning_rate": 5.115595349458773e-08, "loss": 0.0495, "step": 3657 }, { "epoch": 0.49, "learning_rate": 5.114258987037284e-08, "loss": 0.086, "step": 3658 }, { "epoch": 0.49, "learning_rate": 5.112922624615795e-08, "loss": 0.0609, "step": 3659 }, { "epoch": 0.49, "learning_rate": 5.1115862621943075e-08, "loss": 0.0648, "step": 3660 }, { "epoch": 0.49, "learning_rate": 5.1102498997728185e-08, "loss": 0.0535, "step": 3661 }, { "epoch": 0.49, "learning_rate": 5.1089135373513295e-08, "loss": 0.0393, "step": 3662 }, { "epoch": 0.49, "learning_rate": 5.1075771749298405e-08, "loss": 0.0338, "step": 3663 }, { "epoch": 0.49, "learning_rate": 5.106240812508352e-08, "loss": 0.045, "step": 3664 }, { "epoch": 0.49, "learning_rate": 5.104904450086863e-08, "loss": 0.0793, "step": 3665 }, { "epoch": 0.49, "learning_rate": 5.103568087665374e-08, "loss": 0.0287, "step": 3666 }, { "epoch": 0.49, "learning_rate": 5.1022317252438864e-08, "loss": 0.0568, "step": 3667 }, { "epoch": 0.49, "learning_rate": 5.1008953628223974e-08, "loss": 0.0533, "step": 3668 }, { "epoch": 0.49, "learning_rate": 5.0995590004009084e-08, "loss": 0.0563, "step": 3669 }, { "epoch": 0.49, "learning_rate": 5.09822263797942e-08, "loss": 0.0452, "step": 3670 }, { "epoch": 0.49, "learning_rate": 5.096886275557931e-08, "loss": 0.0533, "step": 3671 }, { "epoch": 0.49, "learning_rate": 5.095549913136442e-08, "loss": 0.0461, "step": 3672 }, { "epoch": 0.49, "learning_rate": 5.094213550714954e-08, "loss": 0.0455, "step": 3673 }, { "epoch": 0.49, "learning_rate": 5.092877188293465e-08, "loss": 0.0365, "step": 3674 }, { "epoch": 0.49, "learning_rate": 5.091540825871976e-08, "loss": 0.0437, "step": 3675 }, { "epoch": 0.49, "learning_rate": 5.090204463450487e-08, "loss": 0.0355, "step": 3676 }, { "epoch": 0.49, "learning_rate": 5.088868101028999e-08, "loss": 0.0678, "step": 3677 }, { "epoch": 0.49, "learning_rate": 5.08753173860751e-08, "loss": 0.0725, "step": 3678 }, { "epoch": 0.49, "learning_rate": 5.086195376186021e-08, "loss": 0.0601, "step": 3679 }, { "epoch": 0.49, "learning_rate": 5.084859013764533e-08, "loss": 0.0534, "step": 3680 }, { "epoch": 0.49, "learning_rate": 5.083522651343044e-08, "loss": 0.0809, "step": 3681 }, { "epoch": 0.49, "learning_rate": 5.082186288921555e-08, "loss": 0.045, "step": 3682 }, { "epoch": 0.49, "learning_rate": 5.080849926500067e-08, "loss": 0.0567, "step": 3683 }, { "epoch": 0.49, "learning_rate": 5.079513564078578e-08, "loss": 0.0536, "step": 3684 }, { "epoch": 0.49, "learning_rate": 5.078177201657089e-08, "loss": 0.039, "step": 3685 }, { "epoch": 0.49, "learning_rate": 5.0768408392356e-08, "loss": 0.0505, "step": 3686 }, { "epoch": 0.49, "learning_rate": 5.075504476814112e-08, "loss": 0.0458, "step": 3687 }, { "epoch": 0.49, "learning_rate": 5.074168114392623e-08, "loss": 0.0316, "step": 3688 }, { "epoch": 0.49, "learning_rate": 5.072831751971134e-08, "loss": 0.0392, "step": 3689 }, { "epoch": 0.49, "learning_rate": 5.071495389549646e-08, "loss": 0.04, "step": 3690 }, { "epoch": 0.49, "learning_rate": 5.0701590271281566e-08, "loss": 0.0715, "step": 3691 }, { "epoch": 0.49, "learning_rate": 5.0688226647066676e-08, "loss": 0.0349, "step": 3692 }, { "epoch": 0.49, "learning_rate": 5.06748630228518e-08, "loss": 0.0533, "step": 3693 }, { "epoch": 0.49, "learning_rate": 5.066149939863691e-08, "loss": 0.0614, "step": 3694 }, { "epoch": 0.49, "learning_rate": 5.064813577442202e-08, "loss": 0.0388, "step": 3695 }, { "epoch": 0.49, "learning_rate": 5.063477215020714e-08, "loss": 0.0431, "step": 3696 }, { "epoch": 0.49, "learning_rate": 5.062140852599225e-08, "loss": 0.0557, "step": 3697 }, { "epoch": 0.49, "learning_rate": 5.060804490177736e-08, "loss": 0.0276, "step": 3698 }, { "epoch": 0.49, "learning_rate": 5.0594681277562465e-08, "loss": 0.0477, "step": 3699 }, { "epoch": 0.49, "learning_rate": 5.058131765334759e-08, "loss": 0.0324, "step": 3700 }, { "epoch": 0.49, "learning_rate": 5.05679540291327e-08, "loss": 0.0401, "step": 3701 }, { "epoch": 0.49, "learning_rate": 5.055459040491781e-08, "loss": 0.0498, "step": 3702 }, { "epoch": 0.49, "learning_rate": 5.054122678070293e-08, "loss": 0.0631, "step": 3703 }, { "epoch": 0.49, "learning_rate": 5.052786315648804e-08, "loss": 0.056, "step": 3704 }, { "epoch": 0.5, "learning_rate": 5.051449953227315e-08, "loss": 0.0373, "step": 3705 }, { "epoch": 0.5, "learning_rate": 5.0501135908058266e-08, "loss": 0.0483, "step": 3706 }, { "epoch": 0.5, "learning_rate": 5.0487772283843376e-08, "loss": 0.0368, "step": 3707 }, { "epoch": 0.5, "learning_rate": 5.0474408659628486e-08, "loss": 0.0544, "step": 3708 }, { "epoch": 0.5, "learning_rate": 5.0461045035413596e-08, "loss": 0.0438, "step": 3709 }, { "epoch": 0.5, "learning_rate": 5.044768141119872e-08, "loss": 0.0538, "step": 3710 }, { "epoch": 0.5, "learning_rate": 5.043431778698383e-08, "loss": 0.0333, "step": 3711 }, { "epoch": 0.5, "learning_rate": 5.042095416276894e-08, "loss": 0.0676, "step": 3712 }, { "epoch": 0.5, "learning_rate": 5.0407590538554055e-08, "loss": 0.0671, "step": 3713 }, { "epoch": 0.5, "learning_rate": 5.0394226914339165e-08, "loss": 0.0635, "step": 3714 }, { "epoch": 0.5, "learning_rate": 5.0380863290124275e-08, "loss": 0.0478, "step": 3715 }, { "epoch": 0.5, "learning_rate": 5.03674996659094e-08, "loss": 0.0805, "step": 3716 }, { "epoch": 0.5, "learning_rate": 5.035413604169451e-08, "loss": 0.0528, "step": 3717 }, { "epoch": 0.5, "learning_rate": 5.034077241747962e-08, "loss": 0.051, "step": 3718 }, { "epoch": 0.5, "learning_rate": 5.0327408793264734e-08, "loss": 0.0354, "step": 3719 }, { "epoch": 0.5, "learning_rate": 5.0314045169049844e-08, "loss": 0.0577, "step": 3720 }, { "epoch": 0.5, "learning_rate": 5.0300681544834954e-08, "loss": 0.0546, "step": 3721 }, { "epoch": 0.5, "learning_rate": 5.0287317920620064e-08, "loss": 0.0832, "step": 3722 }, { "epoch": 0.5, "learning_rate": 5.0273954296405187e-08, "loss": 0.041, "step": 3723 }, { "epoch": 0.5, "learning_rate": 5.0260590672190296e-08, "loss": 0.0548, "step": 3724 }, { "epoch": 0.5, "learning_rate": 5.0247227047975406e-08, "loss": 0.0656, "step": 3725 }, { "epoch": 0.5, "learning_rate": 5.023386342376052e-08, "loss": 0.0558, "step": 3726 }, { "epoch": 0.5, "learning_rate": 5.022049979954563e-08, "loss": 0.0432, "step": 3727 }, { "epoch": 0.5, "learning_rate": 5.020713617533074e-08, "loss": 0.0741, "step": 3728 }, { "epoch": 0.5, "learning_rate": 5.0193772551115865e-08, "loss": 0.0542, "step": 3729 }, { "epoch": 0.5, "learning_rate": 5.0180408926900975e-08, "loss": 0.0697, "step": 3730 }, { "epoch": 0.5, "learning_rate": 5.0167045302686085e-08, "loss": 0.0493, "step": 3731 }, { "epoch": 0.5, "learning_rate": 5.0153681678471195e-08, "loss": 0.0674, "step": 3732 }, { "epoch": 0.5, "learning_rate": 5.014031805425631e-08, "loss": 0.0366, "step": 3733 }, { "epoch": 0.5, "learning_rate": 5.012695443004142e-08, "loss": 0.0517, "step": 3734 }, { "epoch": 0.5, "learning_rate": 5.011359080582653e-08, "loss": 0.042, "step": 3735 }, { "epoch": 0.5, "learning_rate": 5.0100227181611654e-08, "loss": 0.0441, "step": 3736 }, { "epoch": 0.5, "learning_rate": 5.0086863557396764e-08, "loss": 0.0518, "step": 3737 }, { "epoch": 0.5, "learning_rate": 5.0073499933181874e-08, "loss": 0.0285, "step": 3738 }, { "epoch": 0.5, "learning_rate": 5.0060136308967e-08, "loss": 0.034, "step": 3739 }, { "epoch": 0.5, "learning_rate": 5.00467726847521e-08, "loss": 0.0511, "step": 3740 }, { "epoch": 0.5, "learning_rate": 5.003340906053721e-08, "loss": 0.0575, "step": 3741 }, { "epoch": 0.5, "learning_rate": 5.002004543632233e-08, "loss": 0.041, "step": 3742 }, { "epoch": 0.5, "learning_rate": 5.000668181210744e-08, "loss": 0.0518, "step": 3743 }, { "epoch": 0.5, "learning_rate": 4.999331818789255e-08, "loss": 0.0592, "step": 3744 }, { "epoch": 0.5, "learning_rate": 4.997995456367767e-08, "loss": 0.0377, "step": 3745 }, { "epoch": 0.5, "learning_rate": 4.9966590939462786e-08, "loss": 0.0501, "step": 3746 }, { "epoch": 0.5, "learning_rate": 4.9953227315247895e-08, "loss": 0.0625, "step": 3747 }, { "epoch": 0.5, "learning_rate": 4.9939863691033005e-08, "loss": 0.0718, "step": 3748 }, { "epoch": 0.5, "learning_rate": 4.9926500066818115e-08, "loss": 0.0498, "step": 3749 }, { "epoch": 0.5, "learning_rate": 4.991313644260323e-08, "loss": 0.0695, "step": 3750 }, { "epoch": 0.5, "learning_rate": 4.989977281838835e-08, "loss": 0.0479, "step": 3751 }, { "epoch": 0.5, "learning_rate": 4.988640919417346e-08, "loss": 0.0487, "step": 3752 }, { "epoch": 0.5, "learning_rate": 4.9873045569958574e-08, "loss": 0.0389, "step": 3753 }, { "epoch": 0.5, "learning_rate": 4.9859681945743684e-08, "loss": 0.0637, "step": 3754 }, { "epoch": 0.5, "learning_rate": 4.9846318321528794e-08, "loss": 0.0337, "step": 3755 }, { "epoch": 0.5, "learning_rate": 4.983295469731391e-08, "loss": 0.0383, "step": 3756 }, { "epoch": 0.5, "learning_rate": 4.981959107309902e-08, "loss": 0.0994, "step": 3757 }, { "epoch": 0.5, "learning_rate": 4.9806227448884137e-08, "loss": 0.0394, "step": 3758 }, { "epoch": 0.5, "learning_rate": 4.9792863824669246e-08, "loss": 0.05, "step": 3759 }, { "epoch": 0.5, "learning_rate": 4.977950020045436e-08, "loss": 0.0731, "step": 3760 }, { "epoch": 0.5, "learning_rate": 4.976613657623947e-08, "loss": 0.0582, "step": 3761 }, { "epoch": 0.5, "learning_rate": 4.975277295202458e-08, "loss": 0.0306, "step": 3762 }, { "epoch": 0.5, "learning_rate": 4.97394093278097e-08, "loss": 0.0682, "step": 3763 }, { "epoch": 0.5, "learning_rate": 4.972604570359481e-08, "loss": 0.0369, "step": 3764 }, { "epoch": 0.5, "learning_rate": 4.9712682079379925e-08, "loss": 0.0478, "step": 3765 }, { "epoch": 0.5, "learning_rate": 4.969931845516504e-08, "loss": 0.0715, "step": 3766 }, { "epoch": 0.5, "learning_rate": 4.968595483095015e-08, "loss": 0.0707, "step": 3767 }, { "epoch": 0.5, "learning_rate": 4.967259120673527e-08, "loss": 0.0327, "step": 3768 }, { "epoch": 0.5, "learning_rate": 4.965922758252038e-08, "loss": 0.045, "step": 3769 }, { "epoch": 0.5, "learning_rate": 4.964586395830549e-08, "loss": 0.079, "step": 3770 }, { "epoch": 0.5, "learning_rate": 4.9632500334090604e-08, "loss": 0.0638, "step": 3771 }, { "epoch": 0.5, "learning_rate": 4.9619136709875714e-08, "loss": 0.0507, "step": 3772 }, { "epoch": 0.5, "learning_rate": 4.960577308566083e-08, "loss": 0.062, "step": 3773 }, { "epoch": 0.5, "learning_rate": 4.959240946144595e-08, "loss": 0.0617, "step": 3774 }, { "epoch": 0.5, "learning_rate": 4.957904583723106e-08, "loss": 0.0589, "step": 3775 }, { "epoch": 0.5, "learning_rate": 4.9565682213016167e-08, "loss": 0.0779, "step": 3776 }, { "epoch": 0.5, "learning_rate": 4.9552318588801276e-08, "loss": 0.0674, "step": 3777 }, { "epoch": 0.5, "learning_rate": 4.953895496458639e-08, "loss": 0.0702, "step": 3778 }, { "epoch": 0.51, "learning_rate": 4.952559134037151e-08, "loss": 0.0614, "step": 3779 }, { "epoch": 0.51, "learning_rate": 4.951222771615662e-08, "loss": 0.0358, "step": 3780 }, { "epoch": 0.51, "learning_rate": 4.9498864091941736e-08, "loss": 0.0326, "step": 3781 }, { "epoch": 0.51, "learning_rate": 4.9485500467726845e-08, "loss": 0.0393, "step": 3782 }, { "epoch": 0.51, "learning_rate": 4.9472136843511955e-08, "loss": 0.0426, "step": 3783 }, { "epoch": 0.51, "learning_rate": 4.945877321929707e-08, "loss": 0.0503, "step": 3784 }, { "epoch": 0.51, "learning_rate": 4.944540959508218e-08, "loss": 0.058, "step": 3785 }, { "epoch": 0.51, "learning_rate": 4.94320459708673e-08, "loss": 0.0561, "step": 3786 }, { "epoch": 0.51, "learning_rate": 4.941868234665241e-08, "loss": 0.0587, "step": 3787 }, { "epoch": 0.51, "learning_rate": 4.9405318722437524e-08, "loss": 0.0186, "step": 3788 }, { "epoch": 0.51, "learning_rate": 4.9391955098222634e-08, "loss": 0.0499, "step": 3789 }, { "epoch": 0.51, "learning_rate": 4.9378591474007744e-08, "loss": 0.0435, "step": 3790 }, { "epoch": 0.51, "learning_rate": 4.936522784979286e-08, "loss": 0.0517, "step": 3791 }, { "epoch": 0.51, "learning_rate": 4.935186422557798e-08, "loss": 0.0417, "step": 3792 }, { "epoch": 0.51, "learning_rate": 4.9338500601363087e-08, "loss": 0.0504, "step": 3793 }, { "epoch": 0.51, "learning_rate": 4.93251369771482e-08, "loss": 0.0493, "step": 3794 }, { "epoch": 0.51, "learning_rate": 4.931177335293331e-08, "loss": 0.0306, "step": 3795 }, { "epoch": 0.51, "learning_rate": 4.929840972871843e-08, "loss": 0.0442, "step": 3796 }, { "epoch": 0.51, "learning_rate": 4.928504610450354e-08, "loss": 0.046, "step": 3797 }, { "epoch": 0.51, "learning_rate": 4.927168248028865e-08, "loss": 0.0687, "step": 3798 }, { "epoch": 0.51, "learning_rate": 4.9258318856073766e-08, "loss": 0.0612, "step": 3799 }, { "epoch": 0.51, "learning_rate": 4.9244955231858875e-08, "loss": 0.03, "step": 3800 }, { "epoch": 0.51, "learning_rate": 4.923159160764399e-08, "loss": 0.043, "step": 3801 }, { "epoch": 0.51, "learning_rate": 4.921822798342911e-08, "loss": 0.0425, "step": 3802 }, { "epoch": 0.51, "learning_rate": 4.920486435921422e-08, "loss": 0.0362, "step": 3803 }, { "epoch": 0.51, "learning_rate": 4.919150073499933e-08, "loss": 0.0468, "step": 3804 }, { "epoch": 0.51, "learning_rate": 4.917813711078444e-08, "loss": 0.085, "step": 3805 }, { "epoch": 0.51, "learning_rate": 4.9164773486569554e-08, "loss": 0.0629, "step": 3806 }, { "epoch": 0.51, "learning_rate": 4.915140986235467e-08, "loss": 0.0361, "step": 3807 }, { "epoch": 0.51, "learning_rate": 4.913804623813978e-08, "loss": 0.0363, "step": 3808 }, { "epoch": 0.51, "learning_rate": 4.91246826139249e-08, "loss": 0.0516, "step": 3809 }, { "epoch": 0.51, "learning_rate": 4.911131898971001e-08, "loss": 0.0476, "step": 3810 }, { "epoch": 0.51, "learning_rate": 4.9097955365495117e-08, "loss": 0.0362, "step": 3811 }, { "epoch": 0.51, "learning_rate": 4.908459174128023e-08, "loss": 0.045, "step": 3812 }, { "epoch": 0.51, "learning_rate": 4.907122811706534e-08, "loss": 0.0729, "step": 3813 }, { "epoch": 0.51, "learning_rate": 4.905786449285046e-08, "loss": 0.0421, "step": 3814 }, { "epoch": 0.51, "learning_rate": 4.9044500868635576e-08, "loss": 0.055, "step": 3815 }, { "epoch": 0.51, "learning_rate": 4.9031137244420686e-08, "loss": 0.0521, "step": 3816 }, { "epoch": 0.51, "learning_rate": 4.90177736202058e-08, "loss": 0.0334, "step": 3817 }, { "epoch": 0.51, "learning_rate": 4.9004409995990905e-08, "loss": 0.0569, "step": 3818 }, { "epoch": 0.51, "learning_rate": 4.899104637177602e-08, "loss": 0.0491, "step": 3819 }, { "epoch": 0.51, "learning_rate": 4.897768274756114e-08, "loss": 0.0692, "step": 3820 }, { "epoch": 0.51, "learning_rate": 4.896431912334625e-08, "loss": 0.0638, "step": 3821 }, { "epoch": 0.51, "learning_rate": 4.8950955499131365e-08, "loss": 0.0531, "step": 3822 }, { "epoch": 0.51, "learning_rate": 4.8937591874916474e-08, "loss": 0.055, "step": 3823 }, { "epoch": 0.51, "learning_rate": 4.892422825070159e-08, "loss": 0.0484, "step": 3824 }, { "epoch": 0.51, "learning_rate": 4.89108646264867e-08, "loss": 0.0575, "step": 3825 }, { "epoch": 0.51, "learning_rate": 4.889750100227181e-08, "loss": 0.0655, "step": 3826 }, { "epoch": 0.51, "learning_rate": 4.888413737805693e-08, "loss": 0.0373, "step": 3827 }, { "epoch": 0.51, "learning_rate": 4.887077375384204e-08, "loss": 0.0364, "step": 3828 }, { "epoch": 0.51, "learning_rate": 4.885741012962715e-08, "loss": 0.0405, "step": 3829 }, { "epoch": 0.51, "learning_rate": 4.884404650541227e-08, "loss": 0.053, "step": 3830 }, { "epoch": 0.51, "learning_rate": 4.883068288119738e-08, "loss": 0.0427, "step": 3831 }, { "epoch": 0.51, "learning_rate": 4.881731925698249e-08, "loss": 0.0378, "step": 3832 }, { "epoch": 0.51, "learning_rate": 4.88039556327676e-08, "loss": 0.0501, "step": 3833 }, { "epoch": 0.51, "learning_rate": 4.8790592008552716e-08, "loss": 0.0488, "step": 3834 }, { "epoch": 0.51, "learning_rate": 4.877722838433783e-08, "loss": 0.0504, "step": 3835 }, { "epoch": 0.51, "learning_rate": 4.876386476012294e-08, "loss": 0.074, "step": 3836 }, { "epoch": 0.51, "learning_rate": 4.875050113590806e-08, "loss": 0.0721, "step": 3837 }, { "epoch": 0.51, "learning_rate": 4.8737137511693175e-08, "loss": 0.0605, "step": 3838 }, { "epoch": 0.51, "learning_rate": 4.872377388747828e-08, "loss": 0.0899, "step": 3839 }, { "epoch": 0.51, "learning_rate": 4.8710410263263394e-08, "loss": 0.0472, "step": 3840 }, { "epoch": 0.51, "learning_rate": 4.8697046639048504e-08, "loss": 0.0466, "step": 3841 }, { "epoch": 0.51, "learning_rate": 4.868368301483362e-08, "loss": 0.0371, "step": 3842 }, { "epoch": 0.51, "learning_rate": 4.867031939061874e-08, "loss": 0.0443, "step": 3843 }, { "epoch": 0.51, "learning_rate": 4.865695576640385e-08, "loss": 0.0592, "step": 3844 }, { "epoch": 0.51, "learning_rate": 4.8643592142188963e-08, "loss": 0.0426, "step": 3845 }, { "epoch": 0.51, "learning_rate": 4.863022851797407e-08, "loss": 0.0847, "step": 3846 }, { "epoch": 0.51, "learning_rate": 4.861686489375918e-08, "loss": 0.0463, "step": 3847 }, { "epoch": 0.51, "learning_rate": 4.86035012695443e-08, "loss": 0.0535, "step": 3848 }, { "epoch": 0.51, "learning_rate": 4.859013764532941e-08, "loss": 0.0347, "step": 3849 }, { "epoch": 0.51, "learning_rate": 4.8576774021114526e-08, "loss": 0.0452, "step": 3850 }, { "epoch": 0.51, "learning_rate": 4.8563410396899636e-08, "loss": 0.048, "step": 3851 }, { "epoch": 0.51, "learning_rate": 4.855004677268475e-08, "loss": 0.0645, "step": 3852 }, { "epoch": 0.51, "learning_rate": 4.853668314846986e-08, "loss": 0.0426, "step": 3853 }, { "epoch": 0.52, "learning_rate": 4.852331952425497e-08, "loss": 0.0499, "step": 3854 }, { "epoch": 0.52, "learning_rate": 4.850995590004009e-08, "loss": 0.0489, "step": 3855 }, { "epoch": 0.52, "learning_rate": 4.84965922758252e-08, "loss": 0.0524, "step": 3856 }, { "epoch": 0.52, "learning_rate": 4.8483228651610315e-08, "loss": 0.0773, "step": 3857 }, { "epoch": 0.52, "learning_rate": 4.846986502739543e-08, "loss": 0.0828, "step": 3858 }, { "epoch": 0.52, "learning_rate": 4.845650140318054e-08, "loss": 0.0419, "step": 3859 }, { "epoch": 0.52, "learning_rate": 4.844313777896565e-08, "loss": 0.0322, "step": 3860 }, { "epoch": 0.52, "learning_rate": 4.842977415475077e-08, "loss": 0.0325, "step": 3861 }, { "epoch": 0.52, "learning_rate": 4.841641053053588e-08, "loss": 0.0435, "step": 3862 }, { "epoch": 0.52, "learning_rate": 4.8403046906320993e-08, "loss": 0.0496, "step": 3863 }, { "epoch": 0.52, "learning_rate": 4.83896832821061e-08, "loss": 0.073, "step": 3864 }, { "epoch": 0.52, "learning_rate": 4.837631965789122e-08, "loss": 0.0427, "step": 3865 }, { "epoch": 0.52, "learning_rate": 4.8362956033676336e-08, "loss": 0.0802, "step": 3866 }, { "epoch": 0.52, "learning_rate": 4.8349592409461446e-08, "loss": 0.1164, "step": 3867 }, { "epoch": 0.52, "learning_rate": 4.8336228785246556e-08, "loss": 0.0388, "step": 3868 }, { "epoch": 0.52, "learning_rate": 4.8322865161031666e-08, "loss": 0.0267, "step": 3869 }, { "epoch": 0.52, "learning_rate": 4.830950153681678e-08, "loss": 0.0441, "step": 3870 }, { "epoch": 0.52, "learning_rate": 4.82961379126019e-08, "loss": 0.0646, "step": 3871 }, { "epoch": 0.52, "learning_rate": 4.828277428838701e-08, "loss": 0.036, "step": 3872 }, { "epoch": 0.52, "learning_rate": 4.8269410664172125e-08, "loss": 0.0328, "step": 3873 }, { "epoch": 0.52, "learning_rate": 4.8256047039957235e-08, "loss": 0.0825, "step": 3874 }, { "epoch": 0.52, "learning_rate": 4.8242683415742344e-08, "loss": 0.0675, "step": 3875 }, { "epoch": 0.52, "learning_rate": 4.822931979152746e-08, "loss": 0.041, "step": 3876 }, { "epoch": 0.52, "learning_rate": 4.821595616731257e-08, "loss": 0.0506, "step": 3877 }, { "epoch": 0.52, "learning_rate": 4.820259254309769e-08, "loss": 0.0384, "step": 3878 }, { "epoch": 0.52, "learning_rate": 4.81892289188828e-08, "loss": 0.0864, "step": 3879 }, { "epoch": 0.52, "learning_rate": 4.8175865294667914e-08, "loss": 0.0511, "step": 3880 }, { "epoch": 0.52, "learning_rate": 4.8162501670453023e-08, "loss": 0.0469, "step": 3881 }, { "epoch": 0.52, "learning_rate": 4.814913804623813e-08, "loss": 0.0391, "step": 3882 }, { "epoch": 0.52, "learning_rate": 4.813577442202325e-08, "loss": 0.0571, "step": 3883 }, { "epoch": 0.52, "learning_rate": 4.8122410797808366e-08, "loss": 0.0465, "step": 3884 }, { "epoch": 0.52, "learning_rate": 4.8109047173593476e-08, "loss": 0.0515, "step": 3885 }, { "epoch": 0.52, "learning_rate": 4.809568354937859e-08, "loss": 0.0409, "step": 3886 }, { "epoch": 0.52, "learning_rate": 4.80823199251637e-08, "loss": 0.0423, "step": 3887 }, { "epoch": 0.52, "learning_rate": 4.806895630094881e-08, "loss": 0.0441, "step": 3888 }, { "epoch": 0.52, "learning_rate": 4.805559267673393e-08, "loss": 0.05, "step": 3889 }, { "epoch": 0.52, "learning_rate": 4.804222905251904e-08, "loss": 0.0701, "step": 3890 }, { "epoch": 0.52, "learning_rate": 4.8028865428304155e-08, "loss": 0.0576, "step": 3891 }, { "epoch": 0.52, "learning_rate": 4.8015501804089265e-08, "loss": 0.0489, "step": 3892 }, { "epoch": 0.52, "learning_rate": 4.800213817987438e-08, "loss": 0.0421, "step": 3893 }, { "epoch": 0.52, "learning_rate": 4.79887745556595e-08, "loss": 0.0606, "step": 3894 }, { "epoch": 0.52, "learning_rate": 4.797541093144461e-08, "loss": 0.0584, "step": 3895 }, { "epoch": 0.52, "learning_rate": 4.796204730722972e-08, "loss": 0.0405, "step": 3896 }, { "epoch": 0.52, "learning_rate": 4.794868368301483e-08, "loss": 0.0435, "step": 3897 }, { "epoch": 0.52, "learning_rate": 4.7935320058799943e-08, "loss": 0.043, "step": 3898 }, { "epoch": 0.52, "learning_rate": 4.792195643458506e-08, "loss": 0.0872, "step": 3899 }, { "epoch": 0.52, "learning_rate": 4.790859281037017e-08, "loss": 0.0354, "step": 3900 }, { "epoch": 0.52, "learning_rate": 4.7895229186155286e-08, "loss": 0.0452, "step": 3901 }, { "epoch": 0.52, "learning_rate": 4.7881865561940396e-08, "loss": 0.0504, "step": 3902 }, { "epoch": 0.52, "learning_rate": 4.7868501937725506e-08, "loss": 0.071, "step": 3903 }, { "epoch": 0.52, "learning_rate": 4.785513831351062e-08, "loss": 0.0422, "step": 3904 }, { "epoch": 0.52, "learning_rate": 4.784177468929573e-08, "loss": 0.0318, "step": 3905 }, { "epoch": 0.52, "learning_rate": 4.782841106508085e-08, "loss": 0.0333, "step": 3906 }, { "epoch": 0.52, "learning_rate": 4.7815047440865965e-08, "loss": 0.0601, "step": 3907 }, { "epoch": 0.52, "learning_rate": 4.7801683816651075e-08, "loss": 0.0497, "step": 3908 }, { "epoch": 0.52, "learning_rate": 4.7788320192436185e-08, "loss": 0.0567, "step": 3909 }, { "epoch": 0.52, "learning_rate": 4.7774956568221295e-08, "loss": 0.0458, "step": 3910 }, { "epoch": 0.52, "learning_rate": 4.776159294400641e-08, "loss": 0.0367, "step": 3911 }, { "epoch": 0.52, "learning_rate": 4.774822931979153e-08, "loss": 0.0613, "step": 3912 }, { "epoch": 0.52, "learning_rate": 4.773486569557664e-08, "loss": 0.0361, "step": 3913 }, { "epoch": 0.52, "learning_rate": 4.7721502071361754e-08, "loss": 0.054, "step": 3914 }, { "epoch": 0.52, "learning_rate": 4.7708138447146864e-08, "loss": 0.0449, "step": 3915 }, { "epoch": 0.52, "learning_rate": 4.769477482293198e-08, "loss": 0.036, "step": 3916 }, { "epoch": 0.52, "learning_rate": 4.768141119871709e-08, "loss": 0.0653, "step": 3917 }, { "epoch": 0.52, "learning_rate": 4.76680475745022e-08, "loss": 0.0404, "step": 3918 }, { "epoch": 0.52, "learning_rate": 4.7654683950287316e-08, "loss": 0.0571, "step": 3919 }, { "epoch": 0.52, "learning_rate": 4.7641320326072426e-08, "loss": 0.0681, "step": 3920 }, { "epoch": 0.52, "learning_rate": 4.762795670185754e-08, "loss": 0.0704, "step": 3921 }, { "epoch": 0.52, "learning_rate": 4.761459307764266e-08, "loss": 0.0558, "step": 3922 }, { "epoch": 0.52, "learning_rate": 4.760122945342777e-08, "loss": 0.0694, "step": 3923 }, { "epoch": 0.52, "learning_rate": 4.758786582921288e-08, "loss": 0.0557, "step": 3924 }, { "epoch": 0.52, "learning_rate": 4.7574502204997995e-08, "loss": 0.0525, "step": 3925 }, { "epoch": 0.52, "learning_rate": 4.7561138580783105e-08, "loss": 0.0473, "step": 3926 }, { "epoch": 0.52, "learning_rate": 4.754777495656822e-08, "loss": 0.0644, "step": 3927 }, { "epoch": 0.52, "learning_rate": 4.753441133235333e-08, "loss": 0.0592, "step": 3928 }, { "epoch": 0.53, "learning_rate": 4.752104770813845e-08, "loss": 0.0343, "step": 3929 }, { "epoch": 0.53, "learning_rate": 4.750768408392356e-08, "loss": 0.0356, "step": 3930 }, { "epoch": 0.53, "learning_rate": 4.749432045970867e-08, "loss": 0.0587, "step": 3931 }, { "epoch": 0.53, "learning_rate": 4.7480956835493784e-08, "loss": 0.0524, "step": 3932 }, { "epoch": 0.53, "learning_rate": 4.7467593211278893e-08, "loss": 0.0363, "step": 3933 }, { "epoch": 0.53, "learning_rate": 4.745422958706401e-08, "loss": 0.0525, "step": 3934 }, { "epoch": 0.53, "learning_rate": 4.7440865962849126e-08, "loss": 0.0669, "step": 3935 }, { "epoch": 0.53, "learning_rate": 4.7427502338634236e-08, "loss": 0.0343, "step": 3936 }, { "epoch": 0.53, "learning_rate": 4.741413871441935e-08, "loss": 0.0528, "step": 3937 }, { "epoch": 0.53, "learning_rate": 4.7400775090204456e-08, "loss": 0.0318, "step": 3938 }, { "epoch": 0.53, "learning_rate": 4.738741146598957e-08, "loss": 0.0427, "step": 3939 }, { "epoch": 0.53, "learning_rate": 4.737404784177469e-08, "loss": 0.0497, "step": 3940 }, { "epoch": 0.53, "learning_rate": 4.73606842175598e-08, "loss": 0.0283, "step": 3941 }, { "epoch": 0.53, "learning_rate": 4.7347320593344915e-08, "loss": 0.0719, "step": 3942 }, { "epoch": 0.53, "learning_rate": 4.7333956969130025e-08, "loss": 0.0284, "step": 3943 }, { "epoch": 0.53, "learning_rate": 4.732059334491514e-08, "loss": 0.0355, "step": 3944 }, { "epoch": 0.53, "learning_rate": 4.730722972070025e-08, "loss": 0.0613, "step": 3945 }, { "epoch": 0.53, "learning_rate": 4.729386609648536e-08, "loss": 0.0525, "step": 3946 }, { "epoch": 0.53, "learning_rate": 4.728050247227048e-08, "loss": 0.0449, "step": 3947 }, { "epoch": 0.53, "learning_rate": 4.7267138848055594e-08, "loss": 0.0561, "step": 3948 }, { "epoch": 0.53, "learning_rate": 4.7253775223840704e-08, "loss": 0.0591, "step": 3949 }, { "epoch": 0.53, "learning_rate": 4.724041159962582e-08, "loss": 0.0469, "step": 3950 }, { "epoch": 0.53, "learning_rate": 4.722704797541093e-08, "loss": 0.0434, "step": 3951 }, { "epoch": 0.53, "learning_rate": 4.721368435119604e-08, "loss": 0.0671, "step": 3952 }, { "epoch": 0.53, "learning_rate": 4.7200320726981156e-08, "loss": 0.0223, "step": 3953 }, { "epoch": 0.53, "learning_rate": 4.7186957102766266e-08, "loss": 0.0457, "step": 3954 }, { "epoch": 0.53, "learning_rate": 4.717359347855138e-08, "loss": 0.0483, "step": 3955 }, { "epoch": 0.53, "learning_rate": 4.716022985433649e-08, "loss": 0.0731, "step": 3956 }, { "epoch": 0.53, "learning_rate": 4.714686623012161e-08, "loss": 0.0415, "step": 3957 }, { "epoch": 0.53, "learning_rate": 4.713350260590672e-08, "loss": 0.0588, "step": 3958 }, { "epoch": 0.53, "learning_rate": 4.712013898169183e-08, "loss": 0.0603, "step": 3959 }, { "epoch": 0.53, "learning_rate": 4.7106775357476945e-08, "loss": 0.0442, "step": 3960 }, { "epoch": 0.53, "learning_rate": 4.7093411733262055e-08, "loss": 0.0342, "step": 3961 }, { "epoch": 0.53, "learning_rate": 4.708004810904717e-08, "loss": 0.0371, "step": 3962 }, { "epoch": 0.53, "learning_rate": 4.706668448483229e-08, "loss": 0.0422, "step": 3963 }, { "epoch": 0.53, "learning_rate": 4.70533208606174e-08, "loss": 0.0463, "step": 3964 }, { "epoch": 0.53, "learning_rate": 4.7039957236402514e-08, "loss": 0.0535, "step": 3965 }, { "epoch": 0.53, "learning_rate": 4.702659361218762e-08, "loss": 0.0599, "step": 3966 }, { "epoch": 0.53, "learning_rate": 4.7013229987972734e-08, "loss": 0.054, "step": 3967 }, { "epoch": 0.53, "learning_rate": 4.699986636375785e-08, "loss": 0.0483, "step": 3968 }, { "epoch": 0.53, "learning_rate": 4.698650273954296e-08, "loss": 0.0431, "step": 3969 }, { "epoch": 0.53, "learning_rate": 4.6973139115328076e-08, "loss": 0.055, "step": 3970 }, { "epoch": 0.53, "learning_rate": 4.695977549111319e-08, "loss": 0.0549, "step": 3971 }, { "epoch": 0.53, "learning_rate": 4.69464118668983e-08, "loss": 0.0686, "step": 3972 }, { "epoch": 0.53, "learning_rate": 4.693304824268341e-08, "loss": 0.0469, "step": 3973 }, { "epoch": 0.53, "learning_rate": 4.691968461846852e-08, "loss": 0.0614, "step": 3974 }, { "epoch": 0.53, "learning_rate": 4.690632099425364e-08, "loss": 0.0362, "step": 3975 }, { "epoch": 0.53, "learning_rate": 4.6892957370038755e-08, "loss": 0.0446, "step": 3976 }, { "epoch": 0.53, "learning_rate": 4.6879593745823865e-08, "loss": 0.0704, "step": 3977 }, { "epoch": 0.53, "learning_rate": 4.686623012160898e-08, "loss": 0.0529, "step": 3978 }, { "epoch": 0.53, "learning_rate": 4.685286649739409e-08, "loss": 0.0542, "step": 3979 }, { "epoch": 0.53, "learning_rate": 4.68395028731792e-08, "loss": 0.0394, "step": 3980 }, { "epoch": 0.53, "learning_rate": 4.682613924896432e-08, "loss": 0.0572, "step": 3981 }, { "epoch": 0.53, "learning_rate": 4.681277562474943e-08, "loss": 0.0575, "step": 3982 }, { "epoch": 0.53, "learning_rate": 4.6799412000534544e-08, "loss": 0.0531, "step": 3983 }, { "epoch": 0.53, "learning_rate": 4.6786048376319654e-08, "loss": 0.0649, "step": 3984 }, { "epoch": 0.53, "learning_rate": 4.677268475210477e-08, "loss": 0.0537, "step": 3985 }, { "epoch": 0.53, "learning_rate": 4.675932112788989e-08, "loss": 0.0461, "step": 3986 }, { "epoch": 0.53, "learning_rate": 4.674595750367499e-08, "loss": 0.071, "step": 3987 }, { "epoch": 0.53, "learning_rate": 4.6732593879460106e-08, "loss": 0.0447, "step": 3988 }, { "epoch": 0.53, "learning_rate": 4.6719230255245216e-08, "loss": 0.0468, "step": 3989 }, { "epoch": 0.53, "learning_rate": 4.670586663103033e-08, "loss": 0.0609, "step": 3990 }, { "epoch": 0.53, "learning_rate": 4.669250300681545e-08, "loss": 0.0446, "step": 3991 }, { "epoch": 0.53, "learning_rate": 4.667913938260056e-08, "loss": 0.0532, "step": 3992 }, { "epoch": 0.53, "learning_rate": 4.6665775758385675e-08, "loss": 0.0393, "step": 3993 }, { "epoch": 0.53, "learning_rate": 4.6652412134170785e-08, "loss": 0.0495, "step": 3994 }, { "epoch": 0.53, "learning_rate": 4.6639048509955895e-08, "loss": 0.046, "step": 3995 }, { "epoch": 0.53, "learning_rate": 4.662568488574101e-08, "loss": 0.0392, "step": 3996 }, { "epoch": 0.53, "learning_rate": 4.661232126152612e-08, "loss": 0.0599, "step": 3997 }, { "epoch": 0.53, "learning_rate": 4.659895763731124e-08, "loss": 0.0397, "step": 3998 }, { "epoch": 0.53, "learning_rate": 4.6585594013096354e-08, "loss": 0.0574, "step": 3999 }, { "epoch": 0.53, "learning_rate": 4.6572230388881464e-08, "loss": 0.0338, "step": 4000 }, { "epoch": 0.53, "learning_rate": 4.6558866764666574e-08, "loss": 0.0373, "step": 4001 }, { "epoch": 0.53, "learning_rate": 4.6545503140451684e-08, "loss": 0.0536, "step": 4002 }, { "epoch": 0.53, "learning_rate": 4.65321395162368e-08, "loss": 0.0575, "step": 4003 }, { "epoch": 0.54, "learning_rate": 4.6518775892021917e-08, "loss": 0.0591, "step": 4004 }, { "epoch": 0.54, "learning_rate": 4.6505412267807026e-08, "loss": 0.0433, "step": 4005 }, { "epoch": 0.54, "learning_rate": 4.649204864359214e-08, "loss": 0.0304, "step": 4006 }, { "epoch": 0.54, "learning_rate": 4.647868501937725e-08, "loss": 0.0625, "step": 4007 }, { "epoch": 0.54, "learning_rate": 4.646532139516236e-08, "loss": 0.0358, "step": 4008 }, { "epoch": 0.54, "learning_rate": 4.645195777094748e-08, "loss": 0.0425, "step": 4009 }, { "epoch": 0.54, "learning_rate": 4.643859414673259e-08, "loss": 0.0387, "step": 4010 }, { "epoch": 0.54, "learning_rate": 4.6425230522517705e-08, "loss": 0.0588, "step": 4011 }, { "epoch": 0.54, "learning_rate": 4.6411866898302815e-08, "loss": 0.0372, "step": 4012 }, { "epoch": 0.54, "learning_rate": 4.639850327408793e-08, "loss": 0.0373, "step": 4013 }, { "epoch": 0.54, "learning_rate": 4.638513964987305e-08, "loss": 0.0592, "step": 4014 }, { "epoch": 0.54, "learning_rate": 4.637177602565816e-08, "loss": 0.0491, "step": 4015 }, { "epoch": 0.54, "learning_rate": 4.635841240144327e-08, "loss": 0.0463, "step": 4016 }, { "epoch": 0.54, "learning_rate": 4.6345048777228384e-08, "loss": 0.0614, "step": 4017 }, { "epoch": 0.54, "learning_rate": 4.6331685153013494e-08, "loss": 0.0624, "step": 4018 }, { "epoch": 0.54, "learning_rate": 4.631832152879861e-08, "loss": 0.0428, "step": 4019 }, { "epoch": 0.54, "learning_rate": 4.630495790458372e-08, "loss": 0.043, "step": 4020 }, { "epoch": 0.54, "learning_rate": 4.629159428036884e-08, "loss": 0.046, "step": 4021 }, { "epoch": 0.54, "learning_rate": 4.6278230656153947e-08, "loss": 0.0751, "step": 4022 }, { "epoch": 0.54, "learning_rate": 4.6264867031939056e-08, "loss": 0.0501, "step": 4023 }, { "epoch": 0.54, "learning_rate": 4.625150340772417e-08, "loss": 0.0588, "step": 4024 }, { "epoch": 0.54, "learning_rate": 4.623813978350928e-08, "loss": 0.0473, "step": 4025 }, { "epoch": 0.54, "learning_rate": 4.62247761592944e-08, "loss": 0.0546, "step": 4026 }, { "epoch": 0.54, "learning_rate": 4.6211412535079516e-08, "loss": 0.069, "step": 4027 }, { "epoch": 0.54, "learning_rate": 4.6198048910864625e-08, "loss": 0.0385, "step": 4028 }, { "epoch": 0.54, "learning_rate": 4.6184685286649735e-08, "loss": 0.0383, "step": 4029 }, { "epoch": 0.54, "learning_rate": 4.6171321662434845e-08, "loss": 0.0466, "step": 4030 }, { "epoch": 0.54, "learning_rate": 4.615795803821996e-08, "loss": 0.0262, "step": 4031 }, { "epoch": 0.54, "learning_rate": 4.614459441400508e-08, "loss": 0.0359, "step": 4032 }, { "epoch": 0.54, "learning_rate": 4.613123078979019e-08, "loss": 0.0421, "step": 4033 }, { "epoch": 0.54, "learning_rate": 4.6117867165575304e-08, "loss": 0.0367, "step": 4034 }, { "epoch": 0.54, "learning_rate": 4.6104503541360414e-08, "loss": 0.0437, "step": 4035 }, { "epoch": 0.54, "learning_rate": 4.609113991714553e-08, "loss": 0.0615, "step": 4036 }, { "epoch": 0.54, "learning_rate": 4.607777629293064e-08, "loss": 0.0511, "step": 4037 }, { "epoch": 0.54, "learning_rate": 4.606441266871575e-08, "loss": 0.0714, "step": 4038 }, { "epoch": 0.54, "learning_rate": 4.605104904450087e-08, "loss": 0.0427, "step": 4039 }, { "epoch": 0.54, "learning_rate": 4.603768542028598e-08, "loss": 0.066, "step": 4040 }, { "epoch": 0.54, "learning_rate": 4.602432179607109e-08, "loss": 0.0378, "step": 4041 }, { "epoch": 0.54, "learning_rate": 4.601095817185621e-08, "loss": 0.0293, "step": 4042 }, { "epoch": 0.54, "learning_rate": 4.599759454764132e-08, "loss": 0.0441, "step": 4043 }, { "epoch": 0.54, "learning_rate": 4.598423092342643e-08, "loss": 0.048, "step": 4044 }, { "epoch": 0.54, "learning_rate": 4.5970867299211546e-08, "loss": 0.0508, "step": 4045 }, { "epoch": 0.54, "learning_rate": 4.5957503674996655e-08, "loss": 0.0511, "step": 4046 }, { "epoch": 0.54, "learning_rate": 4.594414005078177e-08, "loss": 0.058, "step": 4047 }, { "epoch": 0.54, "learning_rate": 4.593077642656688e-08, "loss": 0.0661, "step": 4048 }, { "epoch": 0.54, "learning_rate": 4.5917412802352e-08, "loss": 0.0528, "step": 4049 }, { "epoch": 0.54, "learning_rate": 4.590404917813711e-08, "loss": 0.0467, "step": 4050 }, { "epoch": 0.54, "learning_rate": 4.589068555392222e-08, "loss": 0.0404, "step": 4051 }, { "epoch": 0.54, "learning_rate": 4.5877321929707334e-08, "loss": 0.0509, "step": 4052 }, { "epoch": 0.54, "learning_rate": 4.5863958305492444e-08, "loss": 0.0681, "step": 4053 }, { "epoch": 0.54, "learning_rate": 4.585059468127756e-08, "loss": 0.0322, "step": 4054 }, { "epoch": 0.54, "learning_rate": 4.583723105706268e-08, "loss": 0.0341, "step": 4055 }, { "epoch": 0.54, "learning_rate": 4.582386743284779e-08, "loss": 0.0684, "step": 4056 }, { "epoch": 0.54, "learning_rate": 4.5810503808632897e-08, "loss": 0.0366, "step": 4057 }, { "epoch": 0.54, "learning_rate": 4.5797140184418006e-08, "loss": 0.055, "step": 4058 }, { "epoch": 0.54, "learning_rate": 4.578377656020312e-08, "loss": 0.0389, "step": 4059 }, { "epoch": 0.54, "learning_rate": 4.577041293598824e-08, "loss": 0.0457, "step": 4060 }, { "epoch": 0.54, "learning_rate": 4.575704931177335e-08, "loss": 0.0395, "step": 4061 }, { "epoch": 0.54, "learning_rate": 4.5743685687558466e-08, "loss": 0.0469, "step": 4062 }, { "epoch": 0.54, "learning_rate": 4.573032206334358e-08, "loss": 0.0486, "step": 4063 }, { "epoch": 0.54, "learning_rate": 4.571695843912869e-08, "loss": 0.0369, "step": 4064 }, { "epoch": 0.54, "learning_rate": 4.57035948149138e-08, "loss": 0.0544, "step": 4065 }, { "epoch": 0.54, "learning_rate": 4.569023119069891e-08, "loss": 0.034, "step": 4066 }, { "epoch": 0.54, "learning_rate": 4.567686756648403e-08, "loss": 0.0546, "step": 4067 }, { "epoch": 0.54, "learning_rate": 4.5663503942269145e-08, "loss": 0.0775, "step": 4068 }, { "epoch": 0.54, "learning_rate": 4.5650140318054254e-08, "loss": 0.0617, "step": 4069 }, { "epoch": 0.54, "learning_rate": 4.563677669383937e-08, "loss": 0.058, "step": 4070 }, { "epoch": 0.54, "learning_rate": 4.562341306962448e-08, "loss": 0.0517, "step": 4071 }, { "epoch": 0.54, "learning_rate": 4.561004944540959e-08, "loss": 0.0546, "step": 4072 }, { "epoch": 0.54, "learning_rate": 4.559668582119471e-08, "loss": 0.0393, "step": 4073 }, { "epoch": 0.54, "learning_rate": 4.558332219697982e-08, "loss": 0.0521, "step": 4074 }, { "epoch": 0.54, "learning_rate": 4.556995857276493e-08, "loss": 0.0504, "step": 4075 }, { "epoch": 0.54, "learning_rate": 4.555659494855004e-08, "loss": 0.0454, "step": 4076 }, { "epoch": 0.54, "learning_rate": 4.554323132433516e-08, "loss": 0.0553, "step": 4077 }, { "epoch": 0.54, "learning_rate": 4.552986770012027e-08, "loss": 0.0571, "step": 4078 }, { "epoch": 0.55, "learning_rate": 4.551650407590538e-08, "loss": 0.0459, "step": 4079 }, { "epoch": 0.55, "learning_rate": 4.5503140451690496e-08, "loss": 0.0317, "step": 4080 }, { "epoch": 0.55, "learning_rate": 4.5489776827475605e-08, "loss": 0.0742, "step": 4081 }, { "epoch": 0.55, "learning_rate": 4.547641320326072e-08, "loss": 0.0604, "step": 4082 }, { "epoch": 0.55, "learning_rate": 4.546304957904584e-08, "loss": 0.0322, "step": 4083 }, { "epoch": 0.55, "learning_rate": 4.544968595483095e-08, "loss": 0.0542, "step": 4084 }, { "epoch": 0.55, "learning_rate": 4.5436322330616065e-08, "loss": 0.0646, "step": 4085 }, { "epoch": 0.55, "learning_rate": 4.5422958706401174e-08, "loss": 0.0456, "step": 4086 }, { "epoch": 0.55, "learning_rate": 4.5409595082186284e-08, "loss": 0.0425, "step": 4087 }, { "epoch": 0.55, "learning_rate": 4.53962314579714e-08, "loss": 0.0542, "step": 4088 }, { "epoch": 0.55, "learning_rate": 4.538286783375651e-08, "loss": 0.0481, "step": 4089 }, { "epoch": 0.55, "learning_rate": 4.536950420954163e-08, "loss": 0.0345, "step": 4090 }, { "epoch": 0.55, "learning_rate": 4.5356140585326743e-08, "loss": 0.0535, "step": 4091 }, { "epoch": 0.55, "learning_rate": 4.5342776961111853e-08, "loss": 0.0449, "step": 4092 }, { "epoch": 0.55, "learning_rate": 4.532941333689696e-08, "loss": 0.0502, "step": 4093 }, { "epoch": 0.55, "learning_rate": 4.531604971268207e-08, "loss": 0.037, "step": 4094 }, { "epoch": 0.55, "learning_rate": 4.530268608846719e-08, "loss": 0.0327, "step": 4095 }, { "epoch": 0.55, "learning_rate": 4.5289322464252306e-08, "loss": 0.0415, "step": 4096 }, { "epoch": 0.55, "learning_rate": 4.5275958840037416e-08, "loss": 0.0282, "step": 4097 }, { "epoch": 0.55, "learning_rate": 4.526259521582253e-08, "loss": 0.0562, "step": 4098 }, { "epoch": 0.55, "learning_rate": 4.524923159160764e-08, "loss": 0.0375, "step": 4099 }, { "epoch": 0.55, "learning_rate": 4.523586796739275e-08, "loss": 0.0486, "step": 4100 }, { "epoch": 0.55, "learning_rate": 4.522250434317787e-08, "loss": 0.0754, "step": 4101 }, { "epoch": 0.55, "learning_rate": 4.520914071896298e-08, "loss": 0.0452, "step": 4102 }, { "epoch": 0.55, "learning_rate": 4.5195777094748095e-08, "loss": 0.0424, "step": 4103 }, { "epoch": 0.55, "learning_rate": 4.5182413470533204e-08, "loss": 0.0348, "step": 4104 }, { "epoch": 0.55, "learning_rate": 4.516904984631832e-08, "loss": 0.0402, "step": 4105 }, { "epoch": 0.55, "learning_rate": 4.515568622210344e-08, "loss": 0.049, "step": 4106 }, { "epoch": 0.55, "learning_rate": 4.514232259788854e-08, "loss": 0.0474, "step": 4107 }, { "epoch": 0.55, "learning_rate": 4.512895897367366e-08, "loss": 0.0413, "step": 4108 }, { "epoch": 0.55, "learning_rate": 4.5115595349458773e-08, "loss": 0.0465, "step": 4109 }, { "epoch": 0.55, "learning_rate": 4.510223172524388e-08, "loss": 0.0505, "step": 4110 }, { "epoch": 0.55, "learning_rate": 4.5088868101029e-08, "loss": 0.0762, "step": 4111 }, { "epoch": 0.55, "learning_rate": 4.507550447681411e-08, "loss": 0.0567, "step": 4112 }, { "epoch": 0.55, "learning_rate": 4.5062140852599226e-08, "loss": 0.0589, "step": 4113 }, { "epoch": 0.55, "learning_rate": 4.5048777228384336e-08, "loss": 0.0359, "step": 4114 }, { "epoch": 0.55, "learning_rate": 4.5035413604169446e-08, "loss": 0.058, "step": 4115 }, { "epoch": 0.55, "learning_rate": 4.502204997995456e-08, "loss": 0.0319, "step": 4116 }, { "epoch": 0.55, "learning_rate": 4.500868635573967e-08, "loss": 0.0519, "step": 4117 }, { "epoch": 0.55, "learning_rate": 4.499532273152479e-08, "loss": 0.0462, "step": 4118 }, { "epoch": 0.55, "learning_rate": 4.4981959107309905e-08, "loss": 0.0438, "step": 4119 }, { "epoch": 0.55, "learning_rate": 4.4968595483095015e-08, "loss": 0.0654, "step": 4120 }, { "epoch": 0.55, "learning_rate": 4.4955231858880125e-08, "loss": 0.0532, "step": 4121 }, { "epoch": 0.55, "learning_rate": 4.4941868234665234e-08, "loss": 0.0558, "step": 4122 }, { "epoch": 0.55, "learning_rate": 4.492850461045035e-08, "loss": 0.0434, "step": 4123 }, { "epoch": 0.55, "learning_rate": 4.491514098623547e-08, "loss": 0.0419, "step": 4124 }, { "epoch": 0.55, "learning_rate": 4.490177736202058e-08, "loss": 0.0576, "step": 4125 }, { "epoch": 0.55, "learning_rate": 4.4888413737805694e-08, "loss": 0.0443, "step": 4126 }, { "epoch": 0.55, "learning_rate": 4.4875050113590803e-08, "loss": 0.0313, "step": 4127 }, { "epoch": 0.55, "learning_rate": 4.486168648937591e-08, "loss": 0.0467, "step": 4128 }, { "epoch": 0.55, "learning_rate": 4.484832286516103e-08, "loss": 0.0464, "step": 4129 }, { "epoch": 0.55, "learning_rate": 4.483495924094614e-08, "loss": 0.0435, "step": 4130 }, { "epoch": 0.55, "learning_rate": 4.4821595616731256e-08, "loss": 0.0622, "step": 4131 }, { "epoch": 0.55, "learning_rate": 4.480823199251637e-08, "loss": 0.053, "step": 4132 }, { "epoch": 0.55, "learning_rate": 4.479486836830148e-08, "loss": 0.0339, "step": 4133 }, { "epoch": 0.55, "learning_rate": 4.47815047440866e-08, "loss": 0.0618, "step": 4134 }, { "epoch": 0.55, "learning_rate": 4.47681411198717e-08, "loss": 0.0453, "step": 4135 }, { "epoch": 0.55, "learning_rate": 4.475477749565682e-08, "loss": 0.0279, "step": 4136 }, { "epoch": 0.55, "learning_rate": 4.4741413871441935e-08, "loss": 0.0488, "step": 4137 }, { "epoch": 0.55, "learning_rate": 4.4728050247227045e-08, "loss": 0.0714, "step": 4138 }, { "epoch": 0.55, "learning_rate": 4.471468662301216e-08, "loss": 0.0491, "step": 4139 }, { "epoch": 0.55, "learning_rate": 4.470132299879727e-08, "loss": 0.0774, "step": 4140 }, { "epoch": 0.55, "learning_rate": 4.468795937458239e-08, "loss": 0.0609, "step": 4141 }, { "epoch": 0.55, "learning_rate": 4.46745957503675e-08, "loss": 0.0608, "step": 4142 }, { "epoch": 0.55, "learning_rate": 4.466123212615261e-08, "loss": 0.0456, "step": 4143 }, { "epoch": 0.55, "learning_rate": 4.4647868501937723e-08, "loss": 0.0451, "step": 4144 }, { "epoch": 0.55, "learning_rate": 4.463450487772283e-08, "loss": 0.0665, "step": 4145 }, { "epoch": 0.55, "learning_rate": 4.462114125350795e-08, "loss": 0.0651, "step": 4146 }, { "epoch": 0.55, "learning_rate": 4.4607777629293066e-08, "loss": 0.0372, "step": 4147 }, { "epoch": 0.55, "learning_rate": 4.4594414005078176e-08, "loss": 0.0364, "step": 4148 }, { "epoch": 0.55, "learning_rate": 4.4581050380863286e-08, "loss": 0.0642, "step": 4149 }, { "epoch": 0.55, "learning_rate": 4.4567686756648396e-08, "loss": 0.0663, "step": 4150 }, { "epoch": 0.55, "learning_rate": 4.455432313243351e-08, "loss": 0.0489, "step": 4151 }, { "epoch": 0.55, "learning_rate": 4.454095950821863e-08, "loss": 0.0524, "step": 4152 }, { "epoch": 0.55, "learning_rate": 4.452759588400374e-08, "loss": 0.0782, "step": 4153 }, { "epoch": 0.56, "learning_rate": 4.4514232259788855e-08, "loss": 0.0494, "step": 4154 }, { "epoch": 0.56, "learning_rate": 4.450086863557397e-08, "loss": 0.0329, "step": 4155 }, { "epoch": 0.56, "learning_rate": 4.4487505011359075e-08, "loss": 0.0566, "step": 4156 }, { "epoch": 0.56, "learning_rate": 4.447414138714419e-08, "loss": 0.0625, "step": 4157 }, { "epoch": 0.56, "learning_rate": 4.44607777629293e-08, "loss": 0.0461, "step": 4158 }, { "epoch": 0.56, "learning_rate": 4.444741413871442e-08, "loss": 0.0277, "step": 4159 }, { "epoch": 0.56, "learning_rate": 4.4434050514499534e-08, "loss": 0.0576, "step": 4160 }, { "epoch": 0.56, "learning_rate": 4.4420686890284644e-08, "loss": 0.0488, "step": 4161 }, { "epoch": 0.56, "learning_rate": 4.440732326606976e-08, "loss": 0.0651, "step": 4162 }, { "epoch": 0.56, "learning_rate": 4.439395964185487e-08, "loss": 0.0409, "step": 4163 }, { "epoch": 0.56, "learning_rate": 4.438059601763998e-08, "loss": 0.0507, "step": 4164 }, { "epoch": 0.56, "learning_rate": 4.4367232393425096e-08, "loss": 0.0674, "step": 4165 }, { "epoch": 0.56, "learning_rate": 4.4353868769210206e-08, "loss": 0.0716, "step": 4166 }, { "epoch": 0.56, "learning_rate": 4.434050514499532e-08, "loss": 0.0578, "step": 4167 }, { "epoch": 0.56, "learning_rate": 4.432714152078043e-08, "loss": 0.0485, "step": 4168 }, { "epoch": 0.56, "learning_rate": 4.431377789656555e-08, "loss": 0.0563, "step": 4169 }, { "epoch": 0.56, "learning_rate": 4.430041427235066e-08, "loss": 0.0556, "step": 4170 }, { "epoch": 0.56, "learning_rate": 4.428705064813577e-08, "loss": 0.0347, "step": 4171 }, { "epoch": 0.56, "learning_rate": 4.4273687023920885e-08, "loss": 0.0651, "step": 4172 }, { "epoch": 0.56, "learning_rate": 4.4260323399705995e-08, "loss": 0.0431, "step": 4173 }, { "epoch": 0.56, "learning_rate": 4.424695977549111e-08, "loss": 0.0599, "step": 4174 }, { "epoch": 0.56, "learning_rate": 4.423359615127623e-08, "loss": 0.0528, "step": 4175 }, { "epoch": 0.56, "learning_rate": 4.422023252706134e-08, "loss": 0.0429, "step": 4176 }, { "epoch": 0.56, "learning_rate": 4.420686890284645e-08, "loss": 0.0515, "step": 4177 }, { "epoch": 0.56, "learning_rate": 4.4193505278631564e-08, "loss": 0.0715, "step": 4178 }, { "epoch": 0.56, "learning_rate": 4.4180141654416674e-08, "loss": 0.0644, "step": 4179 }, { "epoch": 0.56, "learning_rate": 4.416677803020179e-08, "loss": 0.0355, "step": 4180 }, { "epoch": 0.56, "learning_rate": 4.41534144059869e-08, "loss": 0.0477, "step": 4181 }, { "epoch": 0.56, "learning_rate": 4.4140050781772016e-08, "loss": 0.0408, "step": 4182 }, { "epoch": 0.56, "learning_rate": 4.412668715755713e-08, "loss": 0.0401, "step": 4183 }, { "epoch": 0.56, "learning_rate": 4.411332353334224e-08, "loss": 0.0493, "step": 4184 }, { "epoch": 0.56, "learning_rate": 4.409995990912735e-08, "loss": 0.0663, "step": 4185 }, { "epoch": 0.56, "learning_rate": 4.408659628491246e-08, "loss": 0.0701, "step": 4186 }, { "epoch": 0.56, "learning_rate": 4.407323266069758e-08, "loss": 0.0487, "step": 4187 }, { "epoch": 0.56, "learning_rate": 4.4059869036482695e-08, "loss": 0.0687, "step": 4188 }, { "epoch": 0.56, "learning_rate": 4.4046505412267805e-08, "loss": 0.0404, "step": 4189 }, { "epoch": 0.56, "learning_rate": 4.403314178805292e-08, "loss": 0.0372, "step": 4190 }, { "epoch": 0.56, "learning_rate": 4.401977816383803e-08, "loss": 0.0561, "step": 4191 }, { "epoch": 0.56, "learning_rate": 4.400641453962314e-08, "loss": 0.0496, "step": 4192 }, { "epoch": 0.56, "learning_rate": 4.399305091540826e-08, "loss": 0.0611, "step": 4193 }, { "epoch": 0.56, "learning_rate": 4.397968729119337e-08, "loss": 0.0707, "step": 4194 }, { "epoch": 0.56, "learning_rate": 4.3966323666978484e-08, "loss": 0.0511, "step": 4195 }, { "epoch": 0.56, "learning_rate": 4.39529600427636e-08, "loss": 0.0417, "step": 4196 }, { "epoch": 0.56, "learning_rate": 4.393959641854871e-08, "loss": 0.0484, "step": 4197 }, { "epoch": 0.56, "learning_rate": 4.392623279433382e-08, "loss": 0.0525, "step": 4198 }, { "epoch": 0.56, "learning_rate": 4.391286917011893e-08, "loss": 0.0466, "step": 4199 }, { "epoch": 0.56, "learning_rate": 4.3899505545904046e-08, "loss": 0.0563, "step": 4200 }, { "epoch": 0.56, "learning_rate": 4.388614192168916e-08, "loss": 0.0741, "step": 4201 }, { "epoch": 0.56, "learning_rate": 4.387277829747427e-08, "loss": 0.0625, "step": 4202 }, { "epoch": 0.56, "learning_rate": 4.385941467325939e-08, "loss": 0.0364, "step": 4203 }, { "epoch": 0.56, "learning_rate": 4.38460510490445e-08, "loss": 0.0493, "step": 4204 }, { "epoch": 0.56, "learning_rate": 4.3832687424829615e-08, "loss": 0.0359, "step": 4205 }, { "epoch": 0.56, "learning_rate": 4.3819323800614725e-08, "loss": 0.0717, "step": 4206 }, { "epoch": 0.56, "learning_rate": 4.3805960176399835e-08, "loss": 0.0627, "step": 4207 }, { "epoch": 0.56, "learning_rate": 4.379259655218495e-08, "loss": 0.0429, "step": 4208 }, { "epoch": 0.56, "learning_rate": 4.377923292797006e-08, "loss": 0.0514, "step": 4209 }, { "epoch": 0.56, "learning_rate": 4.376586930375518e-08, "loss": 0.0391, "step": 4210 }, { "epoch": 0.56, "learning_rate": 4.3752505679540294e-08, "loss": 0.0478, "step": 4211 }, { "epoch": 0.56, "learning_rate": 4.3739142055325404e-08, "loss": 0.0699, "step": 4212 }, { "epoch": 0.56, "learning_rate": 4.3725778431110514e-08, "loss": 0.0646, "step": 4213 }, { "epoch": 0.56, "learning_rate": 4.3712414806895624e-08, "loss": 0.0646, "step": 4214 }, { "epoch": 0.56, "learning_rate": 4.369905118268074e-08, "loss": 0.0372, "step": 4215 }, { "epoch": 0.56, "learning_rate": 4.3685687558465856e-08, "loss": 0.0446, "step": 4216 }, { "epoch": 0.56, "learning_rate": 4.3672323934250966e-08, "loss": 0.0312, "step": 4217 }, { "epoch": 0.56, "learning_rate": 4.365896031003608e-08, "loss": 0.0387, "step": 4218 }, { "epoch": 0.56, "learning_rate": 4.364559668582119e-08, "loss": 0.0281, "step": 4219 }, { "epoch": 0.56, "learning_rate": 4.36322330616063e-08, "loss": 0.0644, "step": 4220 }, { "epoch": 0.56, "learning_rate": 4.361886943739142e-08, "loss": 0.0577, "step": 4221 }, { "epoch": 0.56, "learning_rate": 4.360550581317653e-08, "loss": 0.0692, "step": 4222 }, { "epoch": 0.56, "learning_rate": 4.3592142188961645e-08, "loss": 0.0322, "step": 4223 }, { "epoch": 0.56, "learning_rate": 4.357877856474676e-08, "loss": 0.0417, "step": 4224 }, { "epoch": 0.56, "learning_rate": 4.356541494053187e-08, "loss": 0.0533, "step": 4225 }, { "epoch": 0.56, "learning_rate": 4.355205131631698e-08, "loss": 0.0424, "step": 4226 }, { "epoch": 0.56, "learning_rate": 4.353868769210209e-08, "loss": 0.077, "step": 4227 }, { "epoch": 0.57, "learning_rate": 4.352532406788721e-08, "loss": 0.0547, "step": 4228 }, { "epoch": 0.57, "learning_rate": 4.3511960443672324e-08, "loss": 0.058, "step": 4229 }, { "epoch": 0.57, "learning_rate": 4.3498596819457434e-08, "loss": 0.077, "step": 4230 }, { "epoch": 0.57, "learning_rate": 4.348523319524255e-08, "loss": 0.0457, "step": 4231 }, { "epoch": 0.57, "learning_rate": 4.347186957102766e-08, "loss": 0.0496, "step": 4232 }, { "epoch": 0.57, "learning_rate": 4.3458505946812777e-08, "loss": 0.0645, "step": 4233 }, { "epoch": 0.57, "learning_rate": 4.3445142322597886e-08, "loss": 0.0386, "step": 4234 }, { "epoch": 0.57, "learning_rate": 4.3431778698382996e-08, "loss": 0.0435, "step": 4235 }, { "epoch": 0.57, "learning_rate": 4.341841507416811e-08, "loss": 0.0337, "step": 4236 }, { "epoch": 0.57, "learning_rate": 4.340505144995322e-08, "loss": 0.0516, "step": 4237 }, { "epoch": 0.57, "learning_rate": 4.339168782573834e-08, "loss": 0.0447, "step": 4238 }, { "epoch": 0.57, "learning_rate": 4.3378324201523455e-08, "loss": 0.0804, "step": 4239 }, { "epoch": 0.57, "learning_rate": 4.3364960577308565e-08, "loss": 0.0322, "step": 4240 }, { "epoch": 0.57, "learning_rate": 4.3351596953093675e-08, "loss": 0.0523, "step": 4241 }, { "epoch": 0.57, "learning_rate": 4.333823332887879e-08, "loss": 0.0321, "step": 4242 }, { "epoch": 0.57, "learning_rate": 4.33248697046639e-08, "loss": 0.0553, "step": 4243 }, { "epoch": 0.57, "learning_rate": 4.331150608044902e-08, "loss": 0.0474, "step": 4244 }, { "epoch": 0.57, "learning_rate": 4.329814245623413e-08, "loss": 0.0488, "step": 4245 }, { "epoch": 0.57, "learning_rate": 4.3284778832019244e-08, "loss": 0.0372, "step": 4246 }, { "epoch": 0.57, "learning_rate": 4.3271415207804354e-08, "loss": 0.0273, "step": 4247 }, { "epoch": 0.57, "learning_rate": 4.3258051583589464e-08, "loss": 0.0419, "step": 4248 }, { "epoch": 0.57, "learning_rate": 4.324468795937458e-08, "loss": 0.0484, "step": 4249 }, { "epoch": 0.57, "learning_rate": 4.323132433515969e-08, "loss": 0.0613, "step": 4250 }, { "epoch": 0.57, "learning_rate": 4.3217960710944807e-08, "loss": 0.033, "step": 4251 }, { "epoch": 0.57, "learning_rate": 4.320459708672992e-08, "loss": 0.0669, "step": 4252 }, { "epoch": 0.57, "learning_rate": 4.319123346251503e-08, "loss": 0.0418, "step": 4253 }, { "epoch": 0.57, "learning_rate": 4.317786983830015e-08, "loss": 0.0501, "step": 4254 }, { "epoch": 0.57, "learning_rate": 4.316450621408525e-08, "loss": 0.0517, "step": 4255 }, { "epoch": 0.57, "learning_rate": 4.315114258987037e-08, "loss": 0.0419, "step": 4256 }, { "epoch": 0.57, "learning_rate": 4.3137778965655485e-08, "loss": 0.0436, "step": 4257 }, { "epoch": 0.57, "learning_rate": 4.3124415341440595e-08, "loss": 0.0581, "step": 4258 }, { "epoch": 0.57, "learning_rate": 4.311105171722571e-08, "loss": 0.0524, "step": 4259 }, { "epoch": 0.57, "learning_rate": 4.309768809301082e-08, "loss": 0.0635, "step": 4260 }, { "epoch": 0.57, "learning_rate": 4.308432446879594e-08, "loss": 0.0581, "step": 4261 }, { "epoch": 0.57, "learning_rate": 4.307096084458105e-08, "loss": 0.0484, "step": 4262 }, { "epoch": 0.57, "learning_rate": 4.305759722036616e-08, "loss": 0.0426, "step": 4263 }, { "epoch": 0.57, "learning_rate": 4.3044233596151274e-08, "loss": 0.0608, "step": 4264 }, { "epoch": 0.57, "learning_rate": 4.303086997193639e-08, "loss": 0.0638, "step": 4265 }, { "epoch": 0.57, "learning_rate": 4.30175063477215e-08, "loss": 0.0415, "step": 4266 }, { "epoch": 0.57, "learning_rate": 4.300414272350662e-08, "loss": 0.0411, "step": 4267 }, { "epoch": 0.57, "learning_rate": 4.2990779099291727e-08, "loss": 0.0971, "step": 4268 }, { "epoch": 0.57, "learning_rate": 4.2977415475076836e-08, "loss": 0.0473, "step": 4269 }, { "epoch": 0.57, "learning_rate": 4.296405185086195e-08, "loss": 0.0454, "step": 4270 }, { "epoch": 0.57, "learning_rate": 4.295068822664706e-08, "loss": 0.0672, "step": 4271 }, { "epoch": 0.57, "learning_rate": 4.293732460243218e-08, "loss": 0.0763, "step": 4272 }, { "epoch": 0.57, "learning_rate": 4.292396097821729e-08, "loss": 0.058, "step": 4273 }, { "epoch": 0.57, "learning_rate": 4.2910597354002405e-08, "loss": 0.0367, "step": 4274 }, { "epoch": 0.57, "learning_rate": 4.289723372978752e-08, "loss": 0.086, "step": 4275 }, { "epoch": 0.57, "learning_rate": 4.2883870105572625e-08, "loss": 0.0532, "step": 4276 }, { "epoch": 0.57, "learning_rate": 4.287050648135774e-08, "loss": 0.0508, "step": 4277 }, { "epoch": 0.57, "learning_rate": 4.285714285714285e-08, "loss": 0.0601, "step": 4278 }, { "epoch": 0.57, "learning_rate": 4.284377923292797e-08, "loss": 0.0261, "step": 4279 }, { "epoch": 0.57, "learning_rate": 4.2830415608713084e-08, "loss": 0.0765, "step": 4280 }, { "epoch": 0.57, "learning_rate": 4.2817051984498194e-08, "loss": 0.0463, "step": 4281 }, { "epoch": 0.57, "learning_rate": 4.280368836028331e-08, "loss": 0.0432, "step": 4282 }, { "epoch": 0.57, "learning_rate": 4.279032473606842e-08, "loss": 0.0603, "step": 4283 }, { "epoch": 0.57, "learning_rate": 4.277696111185353e-08, "loss": 0.0676, "step": 4284 }, { "epoch": 0.57, "learning_rate": 4.276359748763865e-08, "loss": 0.0407, "step": 4285 }, { "epoch": 0.57, "learning_rate": 4.2750233863423757e-08, "loss": 0.0388, "step": 4286 }, { "epoch": 0.57, "learning_rate": 4.273687023920887e-08, "loss": 0.0576, "step": 4287 }, { "epoch": 0.57, "learning_rate": 4.272350661499399e-08, "loss": 0.0437, "step": 4288 }, { "epoch": 0.57, "learning_rate": 4.27101429907791e-08, "loss": 0.0817, "step": 4289 }, { "epoch": 0.57, "learning_rate": 4.269677936656421e-08, "loss": 0.0291, "step": 4290 }, { "epoch": 0.57, "learning_rate": 4.268341574234932e-08, "loss": 0.0632, "step": 4291 }, { "epoch": 0.57, "learning_rate": 4.2670052118134435e-08, "loss": 0.0588, "step": 4292 }, { "epoch": 0.57, "learning_rate": 4.265668849391955e-08, "loss": 0.037, "step": 4293 }, { "epoch": 0.57, "learning_rate": 4.264332486970466e-08, "loss": 0.0496, "step": 4294 }, { "epoch": 0.57, "learning_rate": 4.262996124548978e-08, "loss": 0.0547, "step": 4295 }, { "epoch": 0.57, "learning_rate": 4.261659762127489e-08, "loss": 0.0539, "step": 4296 }, { "epoch": 0.57, "learning_rate": 4.260323399706e-08, "loss": 0.0477, "step": 4297 }, { "epoch": 0.57, "learning_rate": 4.2589870372845114e-08, "loss": 0.0488, "step": 4298 }, { "epoch": 0.57, "learning_rate": 4.2576506748630224e-08, "loss": 0.053, "step": 4299 }, { "epoch": 0.57, "learning_rate": 4.256314312441534e-08, "loss": 0.0549, "step": 4300 }, { "epoch": 0.57, "learning_rate": 4.254977950020045e-08, "loss": 0.0367, "step": 4301 }, { "epoch": 0.57, "learning_rate": 4.253641587598557e-08, "loss": 0.0341, "step": 4302 }, { "epoch": 0.58, "learning_rate": 4.252305225177068e-08, "loss": 0.0315, "step": 4303 }, { "epoch": 0.58, "learning_rate": 4.2509688627555786e-08, "loss": 0.0872, "step": 4304 }, { "epoch": 0.58, "learning_rate": 4.24963250033409e-08, "loss": 0.0562, "step": 4305 }, { "epoch": 0.58, "learning_rate": 4.248296137912601e-08, "loss": 0.0375, "step": 4306 }, { "epoch": 0.58, "learning_rate": 4.246959775491113e-08, "loss": 0.041, "step": 4307 }, { "epoch": 0.58, "learning_rate": 4.2456234130696246e-08, "loss": 0.0326, "step": 4308 }, { "epoch": 0.58, "learning_rate": 4.2442870506481356e-08, "loss": 0.0554, "step": 4309 }, { "epoch": 0.58, "learning_rate": 4.242950688226647e-08, "loss": 0.0308, "step": 4310 }, { "epoch": 0.58, "learning_rate": 4.241614325805158e-08, "loss": 0.0823, "step": 4311 }, { "epoch": 0.58, "learning_rate": 4.240277963383669e-08, "loss": 0.0448, "step": 4312 }, { "epoch": 0.58, "learning_rate": 4.238941600962181e-08, "loss": 0.0454, "step": 4313 }, { "epoch": 0.58, "learning_rate": 4.237605238540692e-08, "loss": 0.0628, "step": 4314 }, { "epoch": 0.58, "learning_rate": 4.2362688761192034e-08, "loss": 0.0334, "step": 4315 }, { "epoch": 0.58, "learning_rate": 4.234932513697715e-08, "loss": 0.0581, "step": 4316 }, { "epoch": 0.58, "learning_rate": 4.233596151276226e-08, "loss": 0.0596, "step": 4317 }, { "epoch": 0.58, "learning_rate": 4.232259788854737e-08, "loss": 0.0545, "step": 4318 }, { "epoch": 0.58, "learning_rate": 4.230923426433248e-08, "loss": 0.0789, "step": 4319 }, { "epoch": 0.58, "learning_rate": 4.22958706401176e-08, "loss": 0.0467, "step": 4320 }, { "epoch": 0.58, "learning_rate": 4.228250701590271e-08, "loss": 0.062, "step": 4321 }, { "epoch": 0.58, "learning_rate": 4.226914339168782e-08, "loss": 0.058, "step": 4322 }, { "epoch": 0.58, "learning_rate": 4.225577976747294e-08, "loss": 0.0967, "step": 4323 }, { "epoch": 0.58, "learning_rate": 4.224241614325805e-08, "loss": 0.0357, "step": 4324 }, { "epoch": 0.58, "learning_rate": 4.222905251904316e-08, "loss": 0.0576, "step": 4325 }, { "epoch": 0.58, "learning_rate": 4.2215688894828276e-08, "loss": 0.0421, "step": 4326 }, { "epoch": 0.58, "learning_rate": 4.2202325270613385e-08, "loss": 0.0315, "step": 4327 }, { "epoch": 0.58, "learning_rate": 4.21889616463985e-08, "loss": 0.0729, "step": 4328 }, { "epoch": 0.58, "learning_rate": 4.217559802218361e-08, "loss": 0.0825, "step": 4329 }, { "epoch": 0.58, "learning_rate": 4.216223439796873e-08, "loss": 0.0686, "step": 4330 }, { "epoch": 0.58, "learning_rate": 4.2148870773753845e-08, "loss": 0.0428, "step": 4331 }, { "epoch": 0.58, "learning_rate": 4.2135507149538954e-08, "loss": 0.0544, "step": 4332 }, { "epoch": 0.58, "learning_rate": 4.2122143525324064e-08, "loss": 0.0383, "step": 4333 }, { "epoch": 0.58, "learning_rate": 4.210877990110918e-08, "loss": 0.0567, "step": 4334 }, { "epoch": 0.58, "learning_rate": 4.209541627689429e-08, "loss": 0.0378, "step": 4335 }, { "epoch": 0.58, "learning_rate": 4.208205265267941e-08, "loss": 0.0399, "step": 4336 }, { "epoch": 0.58, "learning_rate": 4.206868902846452e-08, "loss": 0.054, "step": 4337 }, { "epoch": 0.58, "learning_rate": 4.2055325404249633e-08, "loss": 0.0379, "step": 4338 }, { "epoch": 0.58, "learning_rate": 4.204196178003474e-08, "loss": 0.0624, "step": 4339 }, { "epoch": 0.58, "learning_rate": 4.202859815581985e-08, "loss": 0.0242, "step": 4340 }, { "epoch": 0.58, "learning_rate": 4.201523453160497e-08, "loss": 0.0646, "step": 4341 }, { "epoch": 0.58, "learning_rate": 4.200187090739008e-08, "loss": 0.0454, "step": 4342 }, { "epoch": 0.58, "learning_rate": 4.1988507283175196e-08, "loss": 0.0297, "step": 4343 }, { "epoch": 0.58, "learning_rate": 4.197514365896031e-08, "loss": 0.0623, "step": 4344 }, { "epoch": 0.58, "learning_rate": 4.196178003474542e-08, "loss": 0.0499, "step": 4345 }, { "epoch": 0.58, "learning_rate": 4.194841641053053e-08, "loss": 0.0372, "step": 4346 }, { "epoch": 0.58, "learning_rate": 4.193505278631564e-08, "loss": 0.0804, "step": 4347 }, { "epoch": 0.58, "learning_rate": 4.192168916210076e-08, "loss": 0.0626, "step": 4348 }, { "epoch": 0.58, "learning_rate": 4.1908325537885875e-08, "loss": 0.0818, "step": 4349 }, { "epoch": 0.58, "learning_rate": 4.1894961913670984e-08, "loss": 0.0381, "step": 4350 }, { "epoch": 0.58, "learning_rate": 4.18815982894561e-08, "loss": 0.0688, "step": 4351 }, { "epoch": 0.58, "learning_rate": 4.186823466524121e-08, "loss": 0.0524, "step": 4352 }, { "epoch": 0.58, "learning_rate": 4.185487104102633e-08, "loss": 0.0566, "step": 4353 }, { "epoch": 0.58, "learning_rate": 4.184150741681144e-08, "loss": 0.0568, "step": 4354 }, { "epoch": 0.58, "learning_rate": 4.182814379259655e-08, "loss": 0.0474, "step": 4355 }, { "epoch": 0.58, "learning_rate": 4.181478016838166e-08, "loss": 0.062, "step": 4356 }, { "epoch": 0.58, "learning_rate": 4.180141654416678e-08, "loss": 0.034, "step": 4357 }, { "epoch": 0.58, "learning_rate": 4.178805291995189e-08, "loss": 0.0559, "step": 4358 }, { "epoch": 0.58, "learning_rate": 4.1774689295737006e-08, "loss": 0.0624, "step": 4359 }, { "epoch": 0.58, "learning_rate": 4.1761325671522116e-08, "loss": 0.0413, "step": 4360 }, { "epoch": 0.58, "learning_rate": 4.1747962047307226e-08, "loss": 0.06, "step": 4361 }, { "epoch": 0.58, "learning_rate": 4.173459842309234e-08, "loss": 0.0456, "step": 4362 }, { "epoch": 0.58, "learning_rate": 4.172123479887745e-08, "loss": 0.0528, "step": 4363 }, { "epoch": 0.58, "learning_rate": 4.170787117466257e-08, "loss": 0.0368, "step": 4364 }, { "epoch": 0.58, "learning_rate": 4.169450755044768e-08, "loss": 0.0531, "step": 4365 }, { "epoch": 0.58, "learning_rate": 4.1681143926232795e-08, "loss": 0.0523, "step": 4366 }, { "epoch": 0.58, "learning_rate": 4.1667780302017905e-08, "loss": 0.0527, "step": 4367 }, { "epoch": 0.58, "learning_rate": 4.1654416677803014e-08, "loss": 0.0619, "step": 4368 }, { "epoch": 0.58, "learning_rate": 4.164105305358813e-08, "loss": 0.0587, "step": 4369 }, { "epoch": 0.58, "learning_rate": 4.162768942937324e-08, "loss": 0.0439, "step": 4370 }, { "epoch": 0.58, "learning_rate": 4.161432580515836e-08, "loss": 0.0683, "step": 4371 }, { "epoch": 0.58, "learning_rate": 4.1600962180943474e-08, "loss": 0.0243, "step": 4372 }, { "epoch": 0.58, "learning_rate": 4.1587598556728583e-08, "loss": 0.0529, "step": 4373 }, { "epoch": 0.58, "learning_rate": 4.157423493251369e-08, "loss": 0.0525, "step": 4374 }, { "epoch": 0.58, "learning_rate": 4.15608713082988e-08, "loss": 0.0739, "step": 4375 }, { "epoch": 0.58, "learning_rate": 4.154750768408392e-08, "loss": 0.0687, "step": 4376 }, { "epoch": 0.58, "learning_rate": 4.1534144059869036e-08, "loss": 0.0306, "step": 4377 }, { "epoch": 0.59, "learning_rate": 4.1520780435654146e-08, "loss": 0.0713, "step": 4378 }, { "epoch": 0.59, "learning_rate": 4.150741681143926e-08, "loss": 0.0452, "step": 4379 }, { "epoch": 0.59, "learning_rate": 4.149405318722438e-08, "loss": 0.0542, "step": 4380 }, { "epoch": 0.59, "learning_rate": 4.148068956300949e-08, "loss": 0.0486, "step": 4381 }, { "epoch": 0.59, "learning_rate": 4.14673259387946e-08, "loss": 0.0591, "step": 4382 }, { "epoch": 0.59, "learning_rate": 4.145396231457971e-08, "loss": 0.0399, "step": 4383 }, { "epoch": 0.59, "learning_rate": 4.1440598690364825e-08, "loss": 0.0469, "step": 4384 }, { "epoch": 0.59, "learning_rate": 4.142723506614994e-08, "loss": 0.0621, "step": 4385 }, { "epoch": 0.59, "learning_rate": 4.141387144193505e-08, "loss": 0.047, "step": 4386 }, { "epoch": 0.59, "learning_rate": 4.140050781772017e-08, "loss": 0.039, "step": 4387 }, { "epoch": 0.59, "learning_rate": 4.138714419350528e-08, "loss": 0.0286, "step": 4388 }, { "epoch": 0.59, "learning_rate": 4.137378056929039e-08, "loss": 0.0489, "step": 4389 }, { "epoch": 0.59, "learning_rate": 4.1360416945075503e-08, "loss": 0.0836, "step": 4390 }, { "epoch": 0.59, "learning_rate": 4.1347053320860613e-08, "loss": 0.0534, "step": 4391 }, { "epoch": 0.59, "learning_rate": 4.133368969664573e-08, "loss": 0.0477, "step": 4392 }, { "epoch": 0.59, "learning_rate": 4.132032607243084e-08, "loss": 0.0607, "step": 4393 }, { "epoch": 0.59, "learning_rate": 4.1306962448215956e-08, "loss": 0.0735, "step": 4394 }, { "epoch": 0.59, "learning_rate": 4.1293598824001066e-08, "loss": 0.0583, "step": 4395 }, { "epoch": 0.59, "learning_rate": 4.1280235199786176e-08, "loss": 0.0381, "step": 4396 }, { "epoch": 0.59, "learning_rate": 4.126687157557129e-08, "loss": 0.0693, "step": 4397 }, { "epoch": 0.59, "learning_rate": 4.12535079513564e-08, "loss": 0.0893, "step": 4398 }, { "epoch": 0.59, "learning_rate": 4.124014432714152e-08, "loss": 0.0419, "step": 4399 }, { "epoch": 0.59, "learning_rate": 4.1226780702926635e-08, "loss": 0.0543, "step": 4400 }, { "epoch": 0.59, "learning_rate": 4.1213417078711745e-08, "loss": 0.0441, "step": 4401 }, { "epoch": 0.59, "learning_rate": 4.120005345449686e-08, "loss": 0.0476, "step": 4402 }, { "epoch": 0.59, "learning_rate": 4.118668983028197e-08, "loss": 0.0528, "step": 4403 }, { "epoch": 0.59, "learning_rate": 4.117332620606708e-08, "loss": 0.0421, "step": 4404 }, { "epoch": 0.59, "learning_rate": 4.11599625818522e-08, "loss": 0.036, "step": 4405 }, { "epoch": 0.59, "learning_rate": 4.114659895763731e-08, "loss": 0.0481, "step": 4406 }, { "epoch": 0.59, "learning_rate": 4.1133235333422424e-08, "loss": 0.0641, "step": 4407 }, { "epoch": 0.59, "learning_rate": 4.111987170920754e-08, "loss": 0.0807, "step": 4408 }, { "epoch": 0.59, "learning_rate": 4.110650808499265e-08, "loss": 0.0566, "step": 4409 }, { "epoch": 0.59, "learning_rate": 4.109314446077776e-08, "loss": 0.0613, "step": 4410 }, { "epoch": 0.59, "learning_rate": 4.107978083656287e-08, "loss": 0.0428, "step": 4411 }, { "epoch": 0.59, "learning_rate": 4.1066417212347986e-08, "loss": 0.0383, "step": 4412 }, { "epoch": 0.59, "learning_rate": 4.10530535881331e-08, "loss": 0.0541, "step": 4413 }, { "epoch": 0.59, "learning_rate": 4.103968996391821e-08, "loss": 0.038, "step": 4414 }, { "epoch": 0.59, "learning_rate": 4.102632633970333e-08, "loss": 0.0629, "step": 4415 }, { "epoch": 0.59, "learning_rate": 4.101296271548844e-08, "loss": 0.0436, "step": 4416 }, { "epoch": 0.59, "learning_rate": 4.099959909127355e-08, "loss": 0.0337, "step": 4417 }, { "epoch": 0.59, "learning_rate": 4.0986235467058665e-08, "loss": 0.0363, "step": 4418 }, { "epoch": 0.59, "learning_rate": 4.0972871842843775e-08, "loss": 0.0402, "step": 4419 }, { "epoch": 0.59, "learning_rate": 4.095950821862889e-08, "loss": 0.0386, "step": 4420 }, { "epoch": 0.59, "learning_rate": 4.0946144594414e-08, "loss": 0.041, "step": 4421 }, { "epoch": 0.59, "learning_rate": 4.093278097019912e-08, "loss": 0.0459, "step": 4422 }, { "epoch": 0.59, "learning_rate": 4.0919417345984234e-08, "loss": 0.0841, "step": 4423 }, { "epoch": 0.59, "learning_rate": 4.090605372176934e-08, "loss": 0.0628, "step": 4424 }, { "epoch": 0.59, "learning_rate": 4.0892690097554454e-08, "loss": 0.0601, "step": 4425 }, { "epoch": 0.59, "learning_rate": 4.087932647333957e-08, "loss": 0.063, "step": 4426 }, { "epoch": 0.59, "learning_rate": 4.086596284912468e-08, "loss": 0.051, "step": 4427 }, { "epoch": 0.59, "learning_rate": 4.0852599224909796e-08, "loss": 0.0717, "step": 4428 }, { "epoch": 0.59, "learning_rate": 4.0839235600694906e-08, "loss": 0.071, "step": 4429 }, { "epoch": 0.59, "learning_rate": 4.082587197648002e-08, "loss": 0.0459, "step": 4430 }, { "epoch": 0.59, "learning_rate": 4.081250835226513e-08, "loss": 0.0527, "step": 4431 }, { "epoch": 0.59, "learning_rate": 4.079914472805024e-08, "loss": 0.036, "step": 4432 }, { "epoch": 0.59, "learning_rate": 4.078578110383536e-08, "loss": 0.0392, "step": 4433 }, { "epoch": 0.59, "learning_rate": 4.077241747962047e-08, "loss": 0.0411, "step": 4434 }, { "epoch": 0.59, "learning_rate": 4.0759053855405585e-08, "loss": 0.0342, "step": 4435 }, { "epoch": 0.59, "learning_rate": 4.07456902311907e-08, "loss": 0.0467, "step": 4436 }, { "epoch": 0.59, "learning_rate": 4.073232660697581e-08, "loss": 0.0588, "step": 4437 }, { "epoch": 0.59, "learning_rate": 4.071896298276092e-08, "loss": 0.0443, "step": 4438 }, { "epoch": 0.59, "learning_rate": 4.070559935854603e-08, "loss": 0.0604, "step": 4439 }, { "epoch": 0.59, "learning_rate": 4.069223573433115e-08, "loss": 0.0704, "step": 4440 }, { "epoch": 0.59, "learning_rate": 4.0678872110116264e-08, "loss": 0.0406, "step": 4441 }, { "epoch": 0.59, "learning_rate": 4.0665508485901374e-08, "loss": 0.0559, "step": 4442 }, { "epoch": 0.59, "learning_rate": 4.065214486168649e-08, "loss": 0.0786, "step": 4443 }, { "epoch": 0.59, "learning_rate": 4.06387812374716e-08, "loss": 0.0312, "step": 4444 }, { "epoch": 0.59, "learning_rate": 4.062541761325671e-08, "loss": 0.0639, "step": 4445 }, { "epoch": 0.59, "learning_rate": 4.0612053989041826e-08, "loss": 0.0452, "step": 4446 }, { "epoch": 0.59, "learning_rate": 4.0598690364826936e-08, "loss": 0.0516, "step": 4447 }, { "epoch": 0.59, "learning_rate": 4.058532674061205e-08, "loss": 0.0487, "step": 4448 }, { "epoch": 0.59, "learning_rate": 4.057196311639717e-08, "loss": 0.0831, "step": 4449 }, { "epoch": 0.59, "learning_rate": 4.055859949218228e-08, "loss": 0.045, "step": 4450 }, { "epoch": 0.59, "learning_rate": 4.0545235867967395e-08, "loss": 0.0616, "step": 4451 }, { "epoch": 0.59, "learning_rate": 4.0531872243752505e-08, "loss": 0.0355, "step": 4452 }, { "epoch": 0.6, "learning_rate": 4.0518508619537615e-08, "loss": 0.0465, "step": 4453 }, { "epoch": 0.6, "learning_rate": 4.050514499532273e-08, "loss": 0.0649, "step": 4454 }, { "epoch": 0.6, "learning_rate": 4.049178137110784e-08, "loss": 0.0386, "step": 4455 }, { "epoch": 0.6, "learning_rate": 4.047841774689296e-08, "loss": 0.0301, "step": 4456 }, { "epoch": 0.6, "learning_rate": 4.046505412267807e-08, "loss": 0.0396, "step": 4457 }, { "epoch": 0.6, "learning_rate": 4.0451690498463184e-08, "loss": 0.0424, "step": 4458 }, { "epoch": 0.6, "learning_rate": 4.0438326874248294e-08, "loss": 0.0459, "step": 4459 }, { "epoch": 0.6, "learning_rate": 4.0424963250033404e-08, "loss": 0.0637, "step": 4460 }, { "epoch": 0.6, "learning_rate": 4.041159962581852e-08, "loss": 0.0609, "step": 4461 }, { "epoch": 0.6, "learning_rate": 4.039823600160363e-08, "loss": 0.0502, "step": 4462 }, { "epoch": 0.6, "learning_rate": 4.0384872377388746e-08, "loss": 0.0368, "step": 4463 }, { "epoch": 0.6, "learning_rate": 4.037150875317386e-08, "loss": 0.0382, "step": 4464 }, { "epoch": 0.6, "learning_rate": 4.035814512895897e-08, "loss": 0.0439, "step": 4465 }, { "epoch": 0.6, "learning_rate": 4.034478150474408e-08, "loss": 0.0409, "step": 4466 }, { "epoch": 0.6, "learning_rate": 4.03314178805292e-08, "loss": 0.0526, "step": 4467 }, { "epoch": 0.6, "learning_rate": 4.031805425631431e-08, "loss": 0.0508, "step": 4468 }, { "epoch": 0.6, "learning_rate": 4.0304690632099425e-08, "loss": 0.0474, "step": 4469 }, { "epoch": 0.6, "learning_rate": 4.0291327007884535e-08, "loss": 0.0498, "step": 4470 }, { "epoch": 0.6, "learning_rate": 4.027796338366965e-08, "loss": 0.0391, "step": 4471 }, { "epoch": 0.6, "learning_rate": 4.026459975945477e-08, "loss": 0.0592, "step": 4472 }, { "epoch": 0.6, "learning_rate": 4.025123613523987e-08, "loss": 0.0558, "step": 4473 }, { "epoch": 0.6, "learning_rate": 4.023787251102499e-08, "loss": 0.0318, "step": 4474 }, { "epoch": 0.6, "learning_rate": 4.02245088868101e-08, "loss": 0.045, "step": 4475 }, { "epoch": 0.6, "learning_rate": 4.0211145262595214e-08, "loss": 0.0483, "step": 4476 }, { "epoch": 0.6, "learning_rate": 4.019778163838033e-08, "loss": 0.0486, "step": 4477 }, { "epoch": 0.6, "learning_rate": 4.018441801416544e-08, "loss": 0.0562, "step": 4478 }, { "epoch": 0.6, "learning_rate": 4.0171054389950557e-08, "loss": 0.0992, "step": 4479 }, { "epoch": 0.6, "learning_rate": 4.0157690765735666e-08, "loss": 0.061, "step": 4480 }, { "epoch": 0.6, "learning_rate": 4.0144327141520776e-08, "loss": 0.0598, "step": 4481 }, { "epoch": 0.6, "learning_rate": 4.013096351730589e-08, "loss": 0.0415, "step": 4482 }, { "epoch": 0.6, "learning_rate": 4.0117599893091e-08, "loss": 0.0466, "step": 4483 }, { "epoch": 0.6, "learning_rate": 4.010423626887612e-08, "loss": 0.0576, "step": 4484 }, { "epoch": 0.6, "learning_rate": 4.009087264466123e-08, "loss": 0.048, "step": 4485 }, { "epoch": 0.6, "learning_rate": 4.0077509020446345e-08, "loss": 0.0523, "step": 4486 }, { "epoch": 0.6, "learning_rate": 4.0064145396231455e-08, "loss": 0.0307, "step": 4487 }, { "epoch": 0.6, "learning_rate": 4.0050781772016565e-08, "loss": 0.0366, "step": 4488 }, { "epoch": 0.6, "learning_rate": 4.003741814780168e-08, "loss": 0.0658, "step": 4489 }, { "epoch": 0.6, "learning_rate": 4.00240545235868e-08, "loss": 0.0617, "step": 4490 }, { "epoch": 0.6, "learning_rate": 4.001069089937191e-08, "loss": 0.0397, "step": 4491 }, { "epoch": 0.6, "learning_rate": 3.9997327275157024e-08, "loss": 0.0458, "step": 4492 }, { "epoch": 0.6, "learning_rate": 3.9983963650942134e-08, "loss": 0.0809, "step": 4493 }, { "epoch": 0.6, "learning_rate": 3.9970600026727244e-08, "loss": 0.0321, "step": 4494 }, { "epoch": 0.6, "learning_rate": 3.995723640251236e-08, "loss": 0.043, "step": 4495 }, { "epoch": 0.6, "learning_rate": 3.994387277829747e-08, "loss": 0.0406, "step": 4496 }, { "epoch": 0.6, "learning_rate": 3.9930509154082587e-08, "loss": 0.1968, "step": 4497 }, { "epoch": 0.6, "learning_rate": 3.9917145529867696e-08, "loss": 0.0579, "step": 4498 }, { "epoch": 0.6, "learning_rate": 3.990378190565281e-08, "loss": 0.0435, "step": 4499 }, { "epoch": 0.6, "learning_rate": 3.989041828143793e-08, "loss": 0.0571, "step": 4500 }, { "epoch": 0.6, "learning_rate": 3.987705465722304e-08, "loss": 0.0709, "step": 4501 }, { "epoch": 0.6, "learning_rate": 3.986369103300815e-08, "loss": 0.0519, "step": 4502 }, { "epoch": 0.6, "learning_rate": 3.985032740879326e-08, "loss": 0.035, "step": 4503 }, { "epoch": 0.6, "learning_rate": 3.9836963784578375e-08, "loss": 0.0516, "step": 4504 }, { "epoch": 0.6, "learning_rate": 3.982360016036349e-08, "loss": 0.0564, "step": 4505 }, { "epoch": 0.6, "learning_rate": 3.98102365361486e-08, "loss": 0.0385, "step": 4506 }, { "epoch": 0.6, "learning_rate": 3.979687291193372e-08, "loss": 0.0345, "step": 4507 }, { "epoch": 0.6, "learning_rate": 3.978350928771883e-08, "loss": 0.0884, "step": 4508 }, { "epoch": 0.6, "learning_rate": 3.977014566350394e-08, "loss": 0.0498, "step": 4509 }, { "epoch": 0.6, "learning_rate": 3.9756782039289054e-08, "loss": 0.0668, "step": 4510 }, { "epoch": 0.6, "learning_rate": 3.9743418415074164e-08, "loss": 0.0707, "step": 4511 }, { "epoch": 0.6, "learning_rate": 3.973005479085928e-08, "loss": 0.0379, "step": 4512 }, { "epoch": 0.6, "learning_rate": 3.97166911666444e-08, "loss": 0.0599, "step": 4513 }, { "epoch": 0.6, "learning_rate": 3.9703327542429507e-08, "loss": 0.0524, "step": 4514 }, { "epoch": 0.6, "learning_rate": 3.9689963918214616e-08, "loss": 0.0705, "step": 4515 }, { "epoch": 0.6, "learning_rate": 3.9676600293999726e-08, "loss": 0.0466, "step": 4516 }, { "epoch": 0.6, "learning_rate": 3.966323666978484e-08, "loss": 0.0463, "step": 4517 }, { "epoch": 0.6, "learning_rate": 3.964987304556996e-08, "loss": 0.0455, "step": 4518 }, { "epoch": 0.6, "learning_rate": 3.963650942135507e-08, "loss": 0.0509, "step": 4519 }, { "epoch": 0.6, "learning_rate": 3.9623145797140185e-08, "loss": 0.0425, "step": 4520 }, { "epoch": 0.6, "learning_rate": 3.9609782172925295e-08, "loss": 0.0498, "step": 4521 }, { "epoch": 0.6, "learning_rate": 3.959641854871041e-08, "loss": 0.0298, "step": 4522 }, { "epoch": 0.6, "learning_rate": 3.958305492449552e-08, "loss": 0.0475, "step": 4523 }, { "epoch": 0.6, "learning_rate": 3.956969130028063e-08, "loss": 0.0521, "step": 4524 }, { "epoch": 0.6, "learning_rate": 3.955632767606575e-08, "loss": 0.0539, "step": 4525 }, { "epoch": 0.6, "learning_rate": 3.954296405185086e-08, "loss": 0.049, "step": 4526 }, { "epoch": 0.6, "learning_rate": 3.9529600427635974e-08, "loss": 0.0597, "step": 4527 }, { "epoch": 0.61, "learning_rate": 3.951623680342109e-08, "loss": 0.0324, "step": 4528 }, { "epoch": 0.61, "learning_rate": 3.95028731792062e-08, "loss": 0.0505, "step": 4529 }, { "epoch": 0.61, "learning_rate": 3.948950955499131e-08, "loss": 0.0424, "step": 4530 }, { "epoch": 0.61, "learning_rate": 3.947614593077642e-08, "loss": 0.0497, "step": 4531 }, { "epoch": 0.61, "learning_rate": 3.9462782306561537e-08, "loss": 0.0498, "step": 4532 }, { "epoch": 0.61, "learning_rate": 3.944941868234665e-08, "loss": 0.0547, "step": 4533 }, { "epoch": 0.61, "learning_rate": 3.943605505813176e-08, "loss": 0.0376, "step": 4534 }, { "epoch": 0.61, "learning_rate": 3.942269143391688e-08, "loss": 0.0487, "step": 4535 }, { "epoch": 0.61, "learning_rate": 3.940932780970199e-08, "loss": 0.063, "step": 4536 }, { "epoch": 0.61, "learning_rate": 3.93959641854871e-08, "loss": 0.0305, "step": 4537 }, { "epoch": 0.61, "learning_rate": 3.9382600561272215e-08, "loss": 0.0581, "step": 4538 }, { "epoch": 0.61, "learning_rate": 3.9369236937057325e-08, "loss": 0.0676, "step": 4539 }, { "epoch": 0.61, "learning_rate": 3.935587331284244e-08, "loss": 0.0534, "step": 4540 }, { "epoch": 0.61, "learning_rate": 3.934250968862756e-08, "loss": 0.0334, "step": 4541 }, { "epoch": 0.61, "learning_rate": 3.932914606441267e-08, "loss": 0.0498, "step": 4542 }, { "epoch": 0.61, "learning_rate": 3.931578244019778e-08, "loss": 0.0505, "step": 4543 }, { "epoch": 0.61, "learning_rate": 3.930241881598289e-08, "loss": 0.0626, "step": 4544 }, { "epoch": 0.61, "learning_rate": 3.9289055191768004e-08, "loss": 0.0431, "step": 4545 }, { "epoch": 0.61, "learning_rate": 3.927569156755312e-08, "loss": 0.0442, "step": 4546 }, { "epoch": 0.61, "learning_rate": 3.926232794333823e-08, "loss": 0.0325, "step": 4547 }, { "epoch": 0.61, "learning_rate": 3.924896431912335e-08, "loss": 0.071, "step": 4548 }, { "epoch": 0.61, "learning_rate": 3.9235600694908457e-08, "loss": 0.036, "step": 4549 }, { "epoch": 0.61, "learning_rate": 3.922223707069357e-08, "loss": 0.057, "step": 4550 }, { "epoch": 0.61, "learning_rate": 3.920887344647868e-08, "loss": 0.053, "step": 4551 }, { "epoch": 0.61, "learning_rate": 3.919550982226379e-08, "loss": 0.0718, "step": 4552 }, { "epoch": 0.61, "learning_rate": 3.918214619804891e-08, "loss": 0.0548, "step": 4553 }, { "epoch": 0.61, "learning_rate": 3.916878257383402e-08, "loss": 0.0464, "step": 4554 }, { "epoch": 0.61, "learning_rate": 3.9155418949619136e-08, "loss": 0.0491, "step": 4555 }, { "epoch": 0.61, "learning_rate": 3.914205532540425e-08, "loss": 0.0683, "step": 4556 }, { "epoch": 0.61, "learning_rate": 3.912869170118936e-08, "loss": 0.0437, "step": 4557 }, { "epoch": 0.61, "learning_rate": 3.911532807697447e-08, "loss": 0.0435, "step": 4558 }, { "epoch": 0.61, "learning_rate": 3.910196445275959e-08, "loss": 0.0375, "step": 4559 }, { "epoch": 0.61, "learning_rate": 3.90886008285447e-08, "loss": 0.0697, "step": 4560 }, { "epoch": 0.61, "learning_rate": 3.9075237204329814e-08, "loss": 0.0483, "step": 4561 }, { "epoch": 0.61, "learning_rate": 3.9061873580114924e-08, "loss": 0.0339, "step": 4562 }, { "epoch": 0.61, "learning_rate": 3.904850995590004e-08, "loss": 0.0402, "step": 4563 }, { "epoch": 0.61, "learning_rate": 3.903514633168515e-08, "loss": 0.075, "step": 4564 }, { "epoch": 0.61, "learning_rate": 3.902178270747026e-08, "loss": 0.0802, "step": 4565 }, { "epoch": 0.61, "learning_rate": 3.900841908325538e-08, "loss": 0.0343, "step": 4566 }, { "epoch": 0.61, "learning_rate": 3.8995055459040487e-08, "loss": 0.0406, "step": 4567 }, { "epoch": 0.61, "learning_rate": 3.89816918348256e-08, "loss": 0.0609, "step": 4568 }, { "epoch": 0.61, "learning_rate": 3.896832821061072e-08, "loss": 0.0578, "step": 4569 }, { "epoch": 0.61, "learning_rate": 3.895496458639583e-08, "loss": 0.0479, "step": 4570 }, { "epoch": 0.61, "learning_rate": 3.8941600962180946e-08, "loss": 0.0625, "step": 4571 }, { "epoch": 0.61, "learning_rate": 3.892823733796605e-08, "loss": 0.052, "step": 4572 }, { "epoch": 0.61, "learning_rate": 3.8914873713751165e-08, "loss": 0.038, "step": 4573 }, { "epoch": 0.61, "learning_rate": 3.890151008953628e-08, "loss": 0.0522, "step": 4574 }, { "epoch": 0.61, "learning_rate": 3.888814646532139e-08, "loss": 0.0467, "step": 4575 }, { "epoch": 0.61, "learning_rate": 3.887478284110651e-08, "loss": 0.0421, "step": 4576 }, { "epoch": 0.61, "learning_rate": 3.886141921689162e-08, "loss": 0.053, "step": 4577 }, { "epoch": 0.61, "learning_rate": 3.8848055592676734e-08, "loss": 0.0499, "step": 4578 }, { "epoch": 0.61, "learning_rate": 3.8834691968461844e-08, "loss": 0.0529, "step": 4579 }, { "epoch": 0.61, "learning_rate": 3.8821328344246954e-08, "loss": 0.0553, "step": 4580 }, { "epoch": 0.61, "learning_rate": 3.880796472003207e-08, "loss": 0.0411, "step": 4581 }, { "epoch": 0.61, "learning_rate": 3.879460109581719e-08, "loss": 0.0828, "step": 4582 }, { "epoch": 0.61, "learning_rate": 3.87812374716023e-08, "loss": 0.044, "step": 4583 }, { "epoch": 0.61, "learning_rate": 3.8767873847387413e-08, "loss": 0.057, "step": 4584 }, { "epoch": 0.61, "learning_rate": 3.875451022317252e-08, "loss": 0.0489, "step": 4585 }, { "epoch": 0.61, "learning_rate": 3.874114659895763e-08, "loss": 0.0556, "step": 4586 }, { "epoch": 0.61, "learning_rate": 3.872778297474275e-08, "loss": 0.0579, "step": 4587 }, { "epoch": 0.61, "learning_rate": 3.871441935052786e-08, "loss": 0.0476, "step": 4588 }, { "epoch": 0.61, "learning_rate": 3.8701055726312976e-08, "loss": 0.0534, "step": 4589 }, { "epoch": 0.61, "learning_rate": 3.8687692102098086e-08, "loss": 0.0533, "step": 4590 }, { "epoch": 0.61, "learning_rate": 3.86743284778832e-08, "loss": 0.0382, "step": 4591 }, { "epoch": 0.61, "learning_rate": 3.866096485366832e-08, "loss": 0.0305, "step": 4592 }, { "epoch": 0.61, "learning_rate": 3.864760122945342e-08, "loss": 0.0402, "step": 4593 }, { "epoch": 0.61, "learning_rate": 3.863423760523854e-08, "loss": 0.0509, "step": 4594 }, { "epoch": 0.61, "learning_rate": 3.862087398102365e-08, "loss": 0.0519, "step": 4595 }, { "epoch": 0.61, "learning_rate": 3.8607510356808764e-08, "loss": 0.0879, "step": 4596 }, { "epoch": 0.61, "learning_rate": 3.859414673259388e-08, "loss": 0.0508, "step": 4597 }, { "epoch": 0.61, "learning_rate": 3.858078310837899e-08, "loss": 0.0493, "step": 4598 }, { "epoch": 0.61, "learning_rate": 3.856741948416411e-08, "loss": 0.0362, "step": 4599 }, { "epoch": 0.61, "learning_rate": 3.855405585994922e-08, "loss": 0.0494, "step": 4600 }, { "epoch": 0.61, "learning_rate": 3.854069223573433e-08, "loss": 0.0517, "step": 4601 }, { "epoch": 0.61, "learning_rate": 3.852732861151944e-08, "loss": 0.0579, "step": 4602 }, { "epoch": 0.62, "learning_rate": 3.851396498730455e-08, "loss": 0.0318, "step": 4603 }, { "epoch": 0.62, "learning_rate": 3.850060136308967e-08, "loss": 0.045, "step": 4604 }, { "epoch": 0.62, "learning_rate": 3.8487237738874786e-08, "loss": 0.0328, "step": 4605 }, { "epoch": 0.62, "learning_rate": 3.8473874114659896e-08, "loss": 0.0687, "step": 4606 }, { "epoch": 0.62, "learning_rate": 3.8460510490445006e-08, "loss": 0.0461, "step": 4607 }, { "epoch": 0.62, "learning_rate": 3.8447146866230116e-08, "loss": 0.0445, "step": 4608 }, { "epoch": 0.62, "learning_rate": 3.843378324201523e-08, "loss": 0.0712, "step": 4609 }, { "epoch": 0.62, "learning_rate": 3.842041961780035e-08, "loss": 0.0248, "step": 4610 }, { "epoch": 0.62, "learning_rate": 3.840705599358546e-08, "loss": 0.0481, "step": 4611 }, { "epoch": 0.62, "learning_rate": 3.8393692369370575e-08, "loss": 0.0436, "step": 4612 }, { "epoch": 0.62, "learning_rate": 3.8380328745155685e-08, "loss": 0.0486, "step": 4613 }, { "epoch": 0.62, "learning_rate": 3.8366965120940794e-08, "loss": 0.0497, "step": 4614 }, { "epoch": 0.62, "learning_rate": 3.835360149672591e-08, "loss": 0.0595, "step": 4615 }, { "epoch": 0.62, "learning_rate": 3.834023787251102e-08, "loss": 0.0369, "step": 4616 }, { "epoch": 0.62, "learning_rate": 3.832687424829614e-08, "loss": 0.0604, "step": 4617 }, { "epoch": 0.62, "learning_rate": 3.831351062408125e-08, "loss": 0.0465, "step": 4618 }, { "epoch": 0.62, "learning_rate": 3.8300146999866363e-08, "loss": 0.0458, "step": 4619 }, { "epoch": 0.62, "learning_rate": 3.828678337565148e-08, "loss": 0.0563, "step": 4620 }, { "epoch": 0.62, "learning_rate": 3.827341975143659e-08, "loss": 0.0515, "step": 4621 }, { "epoch": 0.62, "learning_rate": 3.82600561272217e-08, "loss": 0.044, "step": 4622 }, { "epoch": 0.62, "learning_rate": 3.824669250300681e-08, "loss": 0.0477, "step": 4623 }, { "epoch": 0.62, "learning_rate": 3.8233328878791926e-08, "loss": 0.039, "step": 4624 }, { "epoch": 0.62, "learning_rate": 3.821996525457704e-08, "loss": 0.0447, "step": 4625 }, { "epoch": 0.62, "learning_rate": 3.820660163036215e-08, "loss": 0.0314, "step": 4626 }, { "epoch": 0.62, "learning_rate": 3.819323800614727e-08, "loss": 0.0486, "step": 4627 }, { "epoch": 0.62, "learning_rate": 3.817987438193238e-08, "loss": 0.0703, "step": 4628 }, { "epoch": 0.62, "learning_rate": 3.816651075771749e-08, "loss": 0.0513, "step": 4629 }, { "epoch": 0.62, "learning_rate": 3.8153147133502605e-08, "loss": 0.0335, "step": 4630 }, { "epoch": 0.62, "learning_rate": 3.8139783509287714e-08, "loss": 0.0343, "step": 4631 }, { "epoch": 0.62, "learning_rate": 3.812641988507283e-08, "loss": 0.0464, "step": 4632 }, { "epoch": 0.62, "learning_rate": 3.811305626085795e-08, "loss": 0.0513, "step": 4633 }, { "epoch": 0.62, "learning_rate": 3.809969263664306e-08, "loss": 0.0575, "step": 4634 }, { "epoch": 0.62, "learning_rate": 3.808632901242817e-08, "loss": 0.059, "step": 4635 }, { "epoch": 0.62, "learning_rate": 3.807296538821328e-08, "loss": 0.0563, "step": 4636 }, { "epoch": 0.62, "learning_rate": 3.8059601763998393e-08, "loss": 0.0408, "step": 4637 }, { "epoch": 0.62, "learning_rate": 3.804623813978351e-08, "loss": 0.0309, "step": 4638 }, { "epoch": 0.62, "learning_rate": 3.803287451556862e-08, "loss": 0.0463, "step": 4639 }, { "epoch": 0.62, "learning_rate": 3.8019510891353736e-08, "loss": 0.072, "step": 4640 }, { "epoch": 0.62, "learning_rate": 3.8006147267138846e-08, "loss": 0.048, "step": 4641 }, { "epoch": 0.62, "learning_rate": 3.7992783642923956e-08, "loss": 0.0449, "step": 4642 }, { "epoch": 0.62, "learning_rate": 3.797942001870907e-08, "loss": 0.0497, "step": 4643 }, { "epoch": 0.62, "learning_rate": 3.796605639449418e-08, "loss": 0.0426, "step": 4644 }, { "epoch": 0.62, "learning_rate": 3.79526927702793e-08, "loss": 0.0692, "step": 4645 }, { "epoch": 0.62, "learning_rate": 3.793932914606441e-08, "loss": 0.0703, "step": 4646 }, { "epoch": 0.62, "learning_rate": 3.7925965521849525e-08, "loss": 0.0598, "step": 4647 }, { "epoch": 0.62, "learning_rate": 3.791260189763464e-08, "loss": 0.0503, "step": 4648 }, { "epoch": 0.62, "learning_rate": 3.789923827341975e-08, "loss": 0.0535, "step": 4649 }, { "epoch": 0.62, "learning_rate": 3.788587464920486e-08, "loss": 0.0356, "step": 4650 }, { "epoch": 0.62, "learning_rate": 3.787251102498998e-08, "loss": 0.04, "step": 4651 }, { "epoch": 0.62, "learning_rate": 3.785914740077509e-08, "loss": 0.0425, "step": 4652 }, { "epoch": 0.62, "learning_rate": 3.7845783776560204e-08, "loss": 0.0388, "step": 4653 }, { "epoch": 0.62, "learning_rate": 3.7832420152345313e-08, "loss": 0.035, "step": 4654 }, { "epoch": 0.62, "learning_rate": 3.781905652813043e-08, "loss": 0.0458, "step": 4655 }, { "epoch": 0.62, "learning_rate": 3.780569290391554e-08, "loss": 0.0355, "step": 4656 }, { "epoch": 0.62, "learning_rate": 3.779232927970065e-08, "loss": 0.0458, "step": 4657 }, { "epoch": 0.62, "learning_rate": 3.7778965655485766e-08, "loss": 0.0694, "step": 4658 }, { "epoch": 0.62, "learning_rate": 3.7765602031270876e-08, "loss": 0.0503, "step": 4659 }, { "epoch": 0.62, "learning_rate": 3.775223840705599e-08, "loss": 0.0628, "step": 4660 }, { "epoch": 0.62, "learning_rate": 3.773887478284111e-08, "loss": 0.0685, "step": 4661 }, { "epoch": 0.62, "learning_rate": 3.772551115862622e-08, "loss": 0.0487, "step": 4662 }, { "epoch": 0.62, "learning_rate": 3.771214753441133e-08, "loss": 0.0533, "step": 4663 }, { "epoch": 0.62, "learning_rate": 3.769878391019644e-08, "loss": 0.0366, "step": 4664 }, { "epoch": 0.62, "learning_rate": 3.7685420285981555e-08, "loss": 0.0551, "step": 4665 }, { "epoch": 0.62, "learning_rate": 3.767205666176667e-08, "loss": 0.028, "step": 4666 }, { "epoch": 0.62, "learning_rate": 3.765869303755178e-08, "loss": 0.0769, "step": 4667 }, { "epoch": 0.62, "learning_rate": 3.76453294133369e-08, "loss": 0.0674, "step": 4668 }, { "epoch": 0.62, "learning_rate": 3.763196578912201e-08, "loss": 0.0306, "step": 4669 }, { "epoch": 0.62, "learning_rate": 3.7618602164907124e-08, "loss": 0.0368, "step": 4670 }, { "epoch": 0.62, "learning_rate": 3.7605238540692234e-08, "loss": 0.0636, "step": 4671 }, { "epoch": 0.62, "learning_rate": 3.7591874916477343e-08, "loss": 0.05, "step": 4672 }, { "epoch": 0.62, "learning_rate": 3.757851129226246e-08, "loss": 0.0477, "step": 4673 }, { "epoch": 0.62, "learning_rate": 3.7565147668047576e-08, "loss": 0.0356, "step": 4674 }, { "epoch": 0.62, "learning_rate": 3.7551784043832686e-08, "loss": 0.0691, "step": 4675 }, { "epoch": 0.62, "learning_rate": 3.75384204196178e-08, "loss": 0.0527, "step": 4676 }, { "epoch": 0.63, "learning_rate": 3.752505679540291e-08, "loss": 0.0454, "step": 4677 }, { "epoch": 0.63, "learning_rate": 3.751169317118802e-08, "loss": 0.0349, "step": 4678 }, { "epoch": 0.63, "learning_rate": 3.749832954697314e-08, "loss": 0.0322, "step": 4679 }, { "epoch": 0.63, "learning_rate": 3.748496592275825e-08, "loss": 0.0555, "step": 4680 }, { "epoch": 0.63, "learning_rate": 3.7471602298543365e-08, "loss": 0.0575, "step": 4681 }, { "epoch": 0.63, "learning_rate": 3.7458238674328475e-08, "loss": 0.0411, "step": 4682 }, { "epoch": 0.63, "learning_rate": 3.744487505011359e-08, "loss": 0.0528, "step": 4683 }, { "epoch": 0.63, "learning_rate": 3.74315114258987e-08, "loss": 0.0513, "step": 4684 }, { "epoch": 0.63, "learning_rate": 3.741814780168381e-08, "loss": 0.0612, "step": 4685 }, { "epoch": 0.63, "learning_rate": 3.740478417746893e-08, "loss": 0.0631, "step": 4686 }, { "epoch": 0.63, "learning_rate": 3.739142055325404e-08, "loss": 0.0546, "step": 4687 }, { "epoch": 0.63, "learning_rate": 3.7378056929039154e-08, "loss": 0.0333, "step": 4688 }, { "epoch": 0.63, "learning_rate": 3.736469330482427e-08, "loss": 0.0333, "step": 4689 }, { "epoch": 0.63, "learning_rate": 3.735132968060938e-08, "loss": 0.0725, "step": 4690 }, { "epoch": 0.63, "learning_rate": 3.7337966056394496e-08, "loss": 0.085, "step": 4691 }, { "epoch": 0.63, "learning_rate": 3.73246024321796e-08, "loss": 0.0503, "step": 4692 }, { "epoch": 0.63, "learning_rate": 3.7311238807964716e-08, "loss": 0.0522, "step": 4693 }, { "epoch": 0.63, "learning_rate": 3.729787518374983e-08, "loss": 0.0478, "step": 4694 }, { "epoch": 0.63, "learning_rate": 3.728451155953494e-08, "loss": 0.0389, "step": 4695 }, { "epoch": 0.63, "learning_rate": 3.727114793532006e-08, "loss": 0.0536, "step": 4696 }, { "epoch": 0.63, "learning_rate": 3.7257784311105175e-08, "loss": 0.0495, "step": 4697 }, { "epoch": 0.63, "learning_rate": 3.7244420686890285e-08, "loss": 0.0612, "step": 4698 }, { "epoch": 0.63, "learning_rate": 3.7231057062675395e-08, "loss": 0.0833, "step": 4699 }, { "epoch": 0.63, "learning_rate": 3.7217693438460505e-08, "loss": 0.0573, "step": 4700 }, { "epoch": 0.63, "learning_rate": 3.720432981424562e-08, "loss": 0.0711, "step": 4701 }, { "epoch": 0.63, "learning_rate": 3.719096619003074e-08, "loss": 0.0502, "step": 4702 }, { "epoch": 0.63, "learning_rate": 3.717760256581585e-08, "loss": 0.0291, "step": 4703 }, { "epoch": 0.63, "learning_rate": 3.7164238941600964e-08, "loss": 0.0333, "step": 4704 }, { "epoch": 0.63, "learning_rate": 3.7150875317386074e-08, "loss": 0.0447, "step": 4705 }, { "epoch": 0.63, "learning_rate": 3.7137511693171184e-08, "loss": 0.0382, "step": 4706 }, { "epoch": 0.63, "learning_rate": 3.71241480689563e-08, "loss": 0.0739, "step": 4707 }, { "epoch": 0.63, "learning_rate": 3.711078444474141e-08, "loss": 0.0492, "step": 4708 }, { "epoch": 0.63, "learning_rate": 3.7097420820526526e-08, "loss": 0.0502, "step": 4709 }, { "epoch": 0.63, "learning_rate": 3.7084057196311636e-08, "loss": 0.0583, "step": 4710 }, { "epoch": 0.63, "learning_rate": 3.707069357209675e-08, "loss": 0.067, "step": 4711 }, { "epoch": 0.63, "learning_rate": 3.705732994788186e-08, "loss": 0.0572, "step": 4712 }, { "epoch": 0.63, "learning_rate": 3.704396632366697e-08, "loss": 0.0311, "step": 4713 }, { "epoch": 0.63, "learning_rate": 3.703060269945209e-08, "loss": 0.0786, "step": 4714 }, { "epoch": 0.63, "learning_rate": 3.70172390752372e-08, "loss": 0.0501, "step": 4715 }, { "epoch": 0.63, "learning_rate": 3.7003875451022315e-08, "loss": 0.0404, "step": 4716 }, { "epoch": 0.63, "learning_rate": 3.699051182680743e-08, "loss": 0.0374, "step": 4717 }, { "epoch": 0.63, "learning_rate": 3.697714820259254e-08, "loss": 0.0495, "step": 4718 }, { "epoch": 0.63, "learning_rate": 3.696378457837766e-08, "loss": 0.0518, "step": 4719 }, { "epoch": 0.63, "learning_rate": 3.695042095416277e-08, "loss": 0.056, "step": 4720 }, { "epoch": 0.63, "learning_rate": 3.693705732994788e-08, "loss": 0.0467, "step": 4721 }, { "epoch": 0.63, "learning_rate": 3.6923693705732994e-08, "loss": 0.061, "step": 4722 }, { "epoch": 0.63, "learning_rate": 3.6910330081518104e-08, "loss": 0.0551, "step": 4723 }, { "epoch": 0.63, "learning_rate": 3.689696645730322e-08, "loss": 0.048, "step": 4724 }, { "epoch": 0.63, "learning_rate": 3.6883602833088337e-08, "loss": 0.0505, "step": 4725 }, { "epoch": 0.63, "learning_rate": 3.6870239208873446e-08, "loss": 0.0503, "step": 4726 }, { "epoch": 0.63, "learning_rate": 3.6856875584658556e-08, "loss": 0.0595, "step": 4727 }, { "epoch": 0.63, "learning_rate": 3.6843511960443666e-08, "loss": 0.0534, "step": 4728 }, { "epoch": 0.63, "learning_rate": 3.683014833622878e-08, "loss": 0.0749, "step": 4729 }, { "epoch": 0.63, "learning_rate": 3.68167847120139e-08, "loss": 0.0315, "step": 4730 }, { "epoch": 0.63, "learning_rate": 3.680342108779901e-08, "loss": 0.0386, "step": 4731 }, { "epoch": 0.63, "learning_rate": 3.6790057463584125e-08, "loss": 0.067, "step": 4732 }, { "epoch": 0.63, "learning_rate": 3.6776693839369235e-08, "loss": 0.0456, "step": 4733 }, { "epoch": 0.63, "learning_rate": 3.6763330215154345e-08, "loss": 0.0532, "step": 4734 }, { "epoch": 0.63, "learning_rate": 3.674996659093946e-08, "loss": 0.0549, "step": 4735 }, { "epoch": 0.63, "learning_rate": 3.673660296672457e-08, "loss": 0.0542, "step": 4736 }, { "epoch": 0.63, "learning_rate": 3.672323934250969e-08, "loss": 0.088, "step": 4737 }, { "epoch": 0.63, "learning_rate": 3.67098757182948e-08, "loss": 0.0421, "step": 4738 }, { "epoch": 0.63, "learning_rate": 3.6696512094079914e-08, "loss": 0.04, "step": 4739 }, { "epoch": 0.63, "learning_rate": 3.668314846986503e-08, "loss": 0.0581, "step": 4740 }, { "epoch": 0.63, "learning_rate": 3.6669784845650134e-08, "loss": 0.0674, "step": 4741 }, { "epoch": 0.63, "learning_rate": 3.665642122143525e-08, "loss": 0.0536, "step": 4742 }, { "epoch": 0.63, "learning_rate": 3.6643057597220367e-08, "loss": 0.0438, "step": 4743 }, { "epoch": 0.63, "learning_rate": 3.6629693973005476e-08, "loss": 0.0599, "step": 4744 }, { "epoch": 0.63, "learning_rate": 3.661633034879059e-08, "loss": 0.054, "step": 4745 }, { "epoch": 0.63, "learning_rate": 3.66029667245757e-08, "loss": 0.0309, "step": 4746 }, { "epoch": 0.63, "learning_rate": 3.658960310036082e-08, "loss": 0.054, "step": 4747 }, { "epoch": 0.63, "learning_rate": 3.657623947614593e-08, "loss": 0.0369, "step": 4748 }, { "epoch": 0.63, "learning_rate": 3.656287585193104e-08, "loss": 0.0617, "step": 4749 }, { "epoch": 0.63, "learning_rate": 3.6549512227716155e-08, "loss": 0.0284, "step": 4750 }, { "epoch": 0.63, "learning_rate": 3.6536148603501265e-08, "loss": 0.0528, "step": 4751 }, { "epoch": 0.64, "learning_rate": 3.652278497928638e-08, "loss": 0.0514, "step": 4752 }, { "epoch": 0.64, "learning_rate": 3.65094213550715e-08, "loss": 0.0396, "step": 4753 }, { "epoch": 0.64, "learning_rate": 3.649605773085661e-08, "loss": 0.0413, "step": 4754 }, { "epoch": 0.64, "learning_rate": 3.648269410664172e-08, "loss": 0.0582, "step": 4755 }, { "epoch": 0.64, "learning_rate": 3.646933048242683e-08, "loss": 0.0577, "step": 4756 }, { "epoch": 0.64, "learning_rate": 3.6455966858211944e-08, "loss": 0.0473, "step": 4757 }, { "epoch": 0.64, "learning_rate": 3.644260323399706e-08, "loss": 0.0511, "step": 4758 }, { "epoch": 0.64, "learning_rate": 3.642923960978217e-08, "loss": 0.0464, "step": 4759 }, { "epoch": 0.64, "learning_rate": 3.6415875985567287e-08, "loss": 0.039, "step": 4760 }, { "epoch": 0.64, "learning_rate": 3.64025123613524e-08, "loss": 0.0577, "step": 4761 }, { "epoch": 0.64, "learning_rate": 3.6389148737137506e-08, "loss": 0.0371, "step": 4762 }, { "epoch": 0.64, "learning_rate": 3.637578511292262e-08, "loss": 0.0613, "step": 4763 }, { "epoch": 0.64, "learning_rate": 3.636242148870773e-08, "loss": 0.0491, "step": 4764 }, { "epoch": 0.64, "learning_rate": 3.634905786449285e-08, "loss": 0.0546, "step": 4765 }, { "epoch": 0.64, "learning_rate": 3.6335694240277966e-08, "loss": 0.0408, "step": 4766 }, { "epoch": 0.64, "learning_rate": 3.6322330616063075e-08, "loss": 0.033, "step": 4767 }, { "epoch": 0.64, "learning_rate": 3.630896699184819e-08, "loss": 0.0631, "step": 4768 }, { "epoch": 0.64, "learning_rate": 3.62956033676333e-08, "loss": 0.0458, "step": 4769 }, { "epoch": 0.64, "learning_rate": 3.628223974341841e-08, "loss": 0.0657, "step": 4770 }, { "epoch": 0.64, "learning_rate": 3.626887611920353e-08, "loss": 0.0529, "step": 4771 }, { "epoch": 0.64, "learning_rate": 3.625551249498864e-08, "loss": 0.0532, "step": 4772 }, { "epoch": 0.64, "learning_rate": 3.6242148870773754e-08, "loss": 0.0448, "step": 4773 }, { "epoch": 0.64, "learning_rate": 3.6228785246558864e-08, "loss": 0.0475, "step": 4774 }, { "epoch": 0.64, "learning_rate": 3.621542162234398e-08, "loss": 0.0402, "step": 4775 }, { "epoch": 0.64, "learning_rate": 3.620205799812909e-08, "loss": 0.0489, "step": 4776 }, { "epoch": 0.64, "learning_rate": 3.61886943739142e-08, "loss": 0.0444, "step": 4777 }, { "epoch": 0.64, "learning_rate": 3.6175330749699317e-08, "loss": 0.0584, "step": 4778 }, { "epoch": 0.64, "learning_rate": 3.6161967125484426e-08, "loss": 0.0646, "step": 4779 }, { "epoch": 0.64, "learning_rate": 3.614860350126954e-08, "loss": 0.0407, "step": 4780 }, { "epoch": 0.64, "learning_rate": 3.613523987705466e-08, "loss": 0.0451, "step": 4781 }, { "epoch": 0.64, "learning_rate": 3.612187625283977e-08, "loss": 0.0386, "step": 4782 }, { "epoch": 0.64, "learning_rate": 3.610851262862488e-08, "loss": 0.0373, "step": 4783 }, { "epoch": 0.64, "learning_rate": 3.6095149004409995e-08, "loss": 0.0754, "step": 4784 }, { "epoch": 0.64, "learning_rate": 3.6081785380195105e-08, "loss": 0.0495, "step": 4785 }, { "epoch": 0.64, "learning_rate": 3.606842175598022e-08, "loss": 0.0634, "step": 4786 }, { "epoch": 0.64, "learning_rate": 3.605505813176533e-08, "loss": 0.0634, "step": 4787 }, { "epoch": 0.64, "learning_rate": 3.604169450755045e-08, "loss": 0.0451, "step": 4788 }, { "epoch": 0.64, "learning_rate": 3.6028330883335564e-08, "loss": 0.0641, "step": 4789 }, { "epoch": 0.64, "learning_rate": 3.6014967259120674e-08, "loss": 0.0554, "step": 4790 }, { "epoch": 0.64, "learning_rate": 3.6001603634905784e-08, "loss": 0.0693, "step": 4791 }, { "epoch": 0.64, "learning_rate": 3.5988240010690894e-08, "loss": 0.0627, "step": 4792 }, { "epoch": 0.64, "learning_rate": 3.597487638647601e-08, "loss": 0.0735, "step": 4793 }, { "epoch": 0.64, "learning_rate": 3.596151276226113e-08, "loss": 0.0407, "step": 4794 }, { "epoch": 0.64, "learning_rate": 3.594814913804624e-08, "loss": 0.0643, "step": 4795 }, { "epoch": 0.64, "learning_rate": 3.593478551383135e-08, "loss": 0.0833, "step": 4796 }, { "epoch": 0.64, "learning_rate": 3.592142188961646e-08, "loss": 0.0513, "step": 4797 }, { "epoch": 0.64, "learning_rate": 3.590805826540157e-08, "loss": 0.042, "step": 4798 }, { "epoch": 0.64, "learning_rate": 3.589469464118669e-08, "loss": 0.0681, "step": 4799 }, { "epoch": 0.64, "learning_rate": 3.58813310169718e-08, "loss": 0.0465, "step": 4800 }, { "epoch": 0.64, "learning_rate": 3.5867967392756916e-08, "loss": 0.0456, "step": 4801 }, { "epoch": 0.64, "learning_rate": 3.5854603768542025e-08, "loss": 0.0537, "step": 4802 }, { "epoch": 0.64, "learning_rate": 3.584124014432714e-08, "loss": 0.0561, "step": 4803 }, { "epoch": 0.64, "learning_rate": 3.582787652011225e-08, "loss": 0.0377, "step": 4804 }, { "epoch": 0.64, "learning_rate": 3.581451289589736e-08, "loss": 0.0434, "step": 4805 }, { "epoch": 0.64, "learning_rate": 3.580114927168248e-08, "loss": 0.0354, "step": 4806 }, { "epoch": 0.64, "learning_rate": 3.5787785647467594e-08, "loss": 0.0402, "step": 4807 }, { "epoch": 0.64, "learning_rate": 3.5774422023252704e-08, "loss": 0.0487, "step": 4808 }, { "epoch": 0.64, "learning_rate": 3.576105839903782e-08, "loss": 0.0661, "step": 4809 }, { "epoch": 0.64, "learning_rate": 3.574769477482293e-08, "loss": 0.066, "step": 4810 }, { "epoch": 0.64, "learning_rate": 3.573433115060804e-08, "loss": 0.0472, "step": 4811 }, { "epoch": 0.64, "learning_rate": 3.572096752639316e-08, "loss": 0.07, "step": 4812 }, { "epoch": 0.64, "learning_rate": 3.5707603902178267e-08, "loss": 0.0526, "step": 4813 }, { "epoch": 0.64, "learning_rate": 3.569424027796338e-08, "loss": 0.0515, "step": 4814 }, { "epoch": 0.64, "learning_rate": 3.568087665374849e-08, "loss": 0.0399, "step": 4815 }, { "epoch": 0.64, "learning_rate": 3.566751302953361e-08, "loss": 0.04, "step": 4816 }, { "epoch": 0.64, "learning_rate": 3.5654149405318726e-08, "loss": 0.0519, "step": 4817 }, { "epoch": 0.64, "learning_rate": 3.5640785781103836e-08, "loss": 0.0467, "step": 4818 }, { "epoch": 0.64, "learning_rate": 3.5627422156888945e-08, "loss": 0.0598, "step": 4819 }, { "epoch": 0.64, "learning_rate": 3.5614058532674055e-08, "loss": 0.0589, "step": 4820 }, { "epoch": 0.64, "learning_rate": 3.560069490845917e-08, "loss": 0.0383, "step": 4821 }, { "epoch": 0.64, "learning_rate": 3.558733128424429e-08, "loss": 0.0503, "step": 4822 }, { "epoch": 0.64, "learning_rate": 3.55739676600294e-08, "loss": 0.0449, "step": 4823 }, { "epoch": 0.64, "learning_rate": 3.5560604035814515e-08, "loss": 0.0531, "step": 4824 }, { "epoch": 0.64, "learning_rate": 3.5547240411599624e-08, "loss": 0.076, "step": 4825 }, { "epoch": 0.64, "learning_rate": 3.5533876787384734e-08, "loss": 0.0573, "step": 4826 }, { "epoch": 0.65, "learning_rate": 3.552051316316985e-08, "loss": 0.05, "step": 4827 }, { "epoch": 0.65, "learning_rate": 3.550714953895496e-08, "loss": 0.0437, "step": 4828 }, { "epoch": 0.65, "learning_rate": 3.549378591474008e-08, "loss": 0.0499, "step": 4829 }, { "epoch": 0.65, "learning_rate": 3.5480422290525193e-08, "loss": 0.0351, "step": 4830 }, { "epoch": 0.65, "learning_rate": 3.54670586663103e-08, "loss": 0.0435, "step": 4831 }, { "epoch": 0.65, "learning_rate": 3.545369504209541e-08, "loss": 0.0354, "step": 4832 }, { "epoch": 0.65, "learning_rate": 3.544033141788052e-08, "loss": 0.075, "step": 4833 }, { "epoch": 0.65, "learning_rate": 3.542696779366564e-08, "loss": 0.0481, "step": 4834 }, { "epoch": 0.65, "learning_rate": 3.5413604169450756e-08, "loss": 0.04, "step": 4835 }, { "epoch": 0.65, "learning_rate": 3.5400240545235866e-08, "loss": 0.0543, "step": 4836 }, { "epoch": 0.65, "learning_rate": 3.538687692102098e-08, "loss": 0.0637, "step": 4837 }, { "epoch": 0.65, "learning_rate": 3.537351329680609e-08, "loss": 0.0447, "step": 4838 }, { "epoch": 0.65, "learning_rate": 3.536014967259121e-08, "loss": 0.0573, "step": 4839 }, { "epoch": 0.65, "learning_rate": 3.534678604837632e-08, "loss": 0.0444, "step": 4840 }, { "epoch": 0.65, "learning_rate": 3.533342242416143e-08, "loss": 0.0486, "step": 4841 }, { "epoch": 0.65, "learning_rate": 3.5320058799946544e-08, "loss": 0.0462, "step": 4842 }, { "epoch": 0.65, "learning_rate": 3.5306695175731654e-08, "loss": 0.0558, "step": 4843 }, { "epoch": 0.65, "learning_rate": 3.529333155151677e-08, "loss": 0.0783, "step": 4844 }, { "epoch": 0.65, "learning_rate": 3.527996792730189e-08, "loss": 0.0609, "step": 4845 }, { "epoch": 0.65, "learning_rate": 3.5266604303087e-08, "loss": 0.0544, "step": 4846 }, { "epoch": 0.65, "learning_rate": 3.525324067887211e-08, "loss": 0.042, "step": 4847 }, { "epoch": 0.65, "learning_rate": 3.5239877054657217e-08, "loss": 0.0538, "step": 4848 }, { "epoch": 0.65, "learning_rate": 3.522651343044233e-08, "loss": 0.0329, "step": 4849 }, { "epoch": 0.65, "learning_rate": 3.521314980622745e-08, "loss": 0.0443, "step": 4850 }, { "epoch": 0.65, "learning_rate": 3.519978618201256e-08, "loss": 0.0509, "step": 4851 }, { "epoch": 0.65, "learning_rate": 3.5186422557797676e-08, "loss": 0.0589, "step": 4852 }, { "epoch": 0.65, "learning_rate": 3.5173058933582786e-08, "loss": 0.0588, "step": 4853 }, { "epoch": 0.65, "learning_rate": 3.5159695309367896e-08, "loss": 0.063, "step": 4854 }, { "epoch": 0.65, "learning_rate": 3.514633168515301e-08, "loss": 0.0542, "step": 4855 }, { "epoch": 0.65, "learning_rate": 3.513296806093812e-08, "loss": 0.052, "step": 4856 }, { "epoch": 0.65, "learning_rate": 3.511960443672324e-08, "loss": 0.0319, "step": 4857 }, { "epoch": 0.65, "learning_rate": 3.5106240812508355e-08, "loss": 0.0537, "step": 4858 }, { "epoch": 0.65, "learning_rate": 3.5092877188293465e-08, "loss": 0.0475, "step": 4859 }, { "epoch": 0.65, "learning_rate": 3.507951356407858e-08, "loss": 0.0636, "step": 4860 }, { "epoch": 0.65, "learning_rate": 3.5066149939863684e-08, "loss": 0.0874, "step": 4861 }, { "epoch": 0.65, "learning_rate": 3.50527863156488e-08, "loss": 0.0487, "step": 4862 }, { "epoch": 0.65, "learning_rate": 3.503942269143392e-08, "loss": 0.0682, "step": 4863 }, { "epoch": 0.65, "learning_rate": 3.502605906721903e-08, "loss": 0.0348, "step": 4864 }, { "epoch": 0.65, "learning_rate": 3.5012695443004143e-08, "loss": 0.0372, "step": 4865 }, { "epoch": 0.65, "learning_rate": 3.499933181878925e-08, "loss": 0.0381, "step": 4866 }, { "epoch": 0.65, "learning_rate": 3.498596819457437e-08, "loss": 0.0248, "step": 4867 }, { "epoch": 0.65, "learning_rate": 3.497260457035948e-08, "loss": 0.0792, "step": 4868 }, { "epoch": 0.65, "learning_rate": 3.495924094614459e-08, "loss": 0.0459, "step": 4869 }, { "epoch": 0.65, "learning_rate": 3.4945877321929706e-08, "loss": 0.0451, "step": 4870 }, { "epoch": 0.65, "learning_rate": 3.4932513697714816e-08, "loss": 0.0495, "step": 4871 }, { "epoch": 0.65, "learning_rate": 3.491915007349993e-08, "loss": 0.0483, "step": 4872 }, { "epoch": 0.65, "learning_rate": 3.490578644928505e-08, "loss": 0.046, "step": 4873 }, { "epoch": 0.65, "learning_rate": 3.489242282507016e-08, "loss": 0.0637, "step": 4874 }, { "epoch": 0.65, "learning_rate": 3.487905920085527e-08, "loss": 0.0265, "step": 4875 }, { "epoch": 0.65, "learning_rate": 3.4865695576640385e-08, "loss": 0.0518, "step": 4876 }, { "epoch": 0.65, "learning_rate": 3.4852331952425495e-08, "loss": 0.0396, "step": 4877 }, { "epoch": 0.65, "learning_rate": 3.483896832821061e-08, "loss": 0.0425, "step": 4878 }, { "epoch": 0.65, "learning_rate": 3.482560470399572e-08, "loss": 0.0902, "step": 4879 }, { "epoch": 0.65, "learning_rate": 3.481224107978084e-08, "loss": 0.0448, "step": 4880 }, { "epoch": 0.65, "learning_rate": 3.479887745556595e-08, "loss": 0.0412, "step": 4881 }, { "epoch": 0.65, "learning_rate": 3.478551383135106e-08, "loss": 0.0598, "step": 4882 }, { "epoch": 0.65, "learning_rate": 3.4772150207136173e-08, "loss": 0.0581, "step": 4883 }, { "epoch": 0.65, "learning_rate": 3.475878658292128e-08, "loss": 0.0367, "step": 4884 }, { "epoch": 0.65, "learning_rate": 3.47454229587064e-08, "loss": 0.0429, "step": 4885 }, { "epoch": 0.65, "learning_rate": 3.4732059334491516e-08, "loss": 0.0587, "step": 4886 }, { "epoch": 0.65, "learning_rate": 3.4718695710276626e-08, "loss": 0.0552, "step": 4887 }, { "epoch": 0.65, "learning_rate": 3.470533208606174e-08, "loss": 0.0601, "step": 4888 }, { "epoch": 0.65, "learning_rate": 3.4691968461846846e-08, "loss": 0.0425, "step": 4889 }, { "epoch": 0.65, "learning_rate": 3.467860483763196e-08, "loss": 0.0501, "step": 4890 }, { "epoch": 0.65, "learning_rate": 3.466524121341708e-08, "loss": 0.0502, "step": 4891 }, { "epoch": 0.65, "learning_rate": 3.465187758920219e-08, "loss": 0.0449, "step": 4892 }, { "epoch": 0.65, "learning_rate": 3.4638513964987305e-08, "loss": 0.0533, "step": 4893 }, { "epoch": 0.65, "learning_rate": 3.4625150340772415e-08, "loss": 0.0681, "step": 4894 }, { "epoch": 0.65, "learning_rate": 3.461178671655753e-08, "loss": 0.0371, "step": 4895 }, { "epoch": 0.65, "learning_rate": 3.459842309234264e-08, "loss": 0.0408, "step": 4896 }, { "epoch": 0.65, "learning_rate": 3.458505946812775e-08, "loss": 0.0508, "step": 4897 }, { "epoch": 0.65, "learning_rate": 3.457169584391287e-08, "loss": 0.0312, "step": 4898 }, { "epoch": 0.65, "learning_rate": 3.4558332219697984e-08, "loss": 0.0442, "step": 4899 }, { "epoch": 0.65, "learning_rate": 3.4544968595483093e-08, "loss": 0.0459, "step": 4900 }, { "epoch": 0.65, "learning_rate": 3.453160497126821e-08, "loss": 0.0636, "step": 4901 }, { "epoch": 0.66, "learning_rate": 3.451824134705332e-08, "loss": 0.0443, "step": 4902 }, { "epoch": 0.66, "learning_rate": 3.450487772283843e-08, "loss": 0.042, "step": 4903 }, { "epoch": 0.66, "learning_rate": 3.4491514098623546e-08, "loss": 0.0548, "step": 4904 }, { "epoch": 0.66, "learning_rate": 3.4478150474408656e-08, "loss": 0.0459, "step": 4905 }, { "epoch": 0.66, "learning_rate": 3.446478685019377e-08, "loss": 0.0442, "step": 4906 }, { "epoch": 0.66, "learning_rate": 3.445142322597888e-08, "loss": 0.0557, "step": 4907 }, { "epoch": 0.66, "learning_rate": 3.4438059601764e-08, "loss": 0.0496, "step": 4908 }, { "epoch": 0.66, "learning_rate": 3.4424695977549115e-08, "loss": 0.0521, "step": 4909 }, { "epoch": 0.66, "learning_rate": 3.441133235333422e-08, "loss": 0.0373, "step": 4910 }, { "epoch": 0.66, "learning_rate": 3.4397968729119335e-08, "loss": 0.0461, "step": 4911 }, { "epoch": 0.66, "learning_rate": 3.4384605104904445e-08, "loss": 0.051, "step": 4912 }, { "epoch": 0.66, "learning_rate": 3.437124148068956e-08, "loss": 0.0653, "step": 4913 }, { "epoch": 0.66, "learning_rate": 3.435787785647468e-08, "loss": 0.035, "step": 4914 }, { "epoch": 0.66, "learning_rate": 3.434451423225979e-08, "loss": 0.0468, "step": 4915 }, { "epoch": 0.66, "learning_rate": 3.4331150608044904e-08, "loss": 0.0537, "step": 4916 }, { "epoch": 0.66, "learning_rate": 3.4317786983830014e-08, "loss": 0.0445, "step": 4917 }, { "epoch": 0.66, "learning_rate": 3.4304423359615123e-08, "loss": 0.0529, "step": 4918 }, { "epoch": 0.66, "learning_rate": 3.429105973540024e-08, "loss": 0.0411, "step": 4919 }, { "epoch": 0.66, "learning_rate": 3.427769611118535e-08, "loss": 0.0544, "step": 4920 }, { "epoch": 0.66, "learning_rate": 3.4264332486970466e-08, "loss": 0.057, "step": 4921 }, { "epoch": 0.66, "learning_rate": 3.425096886275558e-08, "loss": 0.0423, "step": 4922 }, { "epoch": 0.66, "learning_rate": 3.423760523854069e-08, "loss": 0.0472, "step": 4923 }, { "epoch": 0.66, "learning_rate": 3.42242416143258e-08, "loss": 0.0489, "step": 4924 }, { "epoch": 0.66, "learning_rate": 3.421087799011091e-08, "loss": 0.0806, "step": 4925 }, { "epoch": 0.66, "learning_rate": 3.419751436589603e-08, "loss": 0.053, "step": 4926 }, { "epoch": 0.66, "learning_rate": 3.4184150741681145e-08, "loss": 0.0307, "step": 4927 }, { "epoch": 0.66, "learning_rate": 3.4170787117466255e-08, "loss": 0.0524, "step": 4928 }, { "epoch": 0.66, "learning_rate": 3.415742349325137e-08, "loss": 0.0377, "step": 4929 }, { "epoch": 0.66, "learning_rate": 3.414405986903648e-08, "loss": 0.032, "step": 4930 }, { "epoch": 0.66, "learning_rate": 3.413069624482159e-08, "loss": 0.0515, "step": 4931 }, { "epoch": 0.66, "learning_rate": 3.411733262060671e-08, "loss": 0.0638, "step": 4932 }, { "epoch": 0.66, "learning_rate": 3.410396899639182e-08, "loss": 0.0492, "step": 4933 }, { "epoch": 0.66, "learning_rate": 3.4090605372176934e-08, "loss": 0.0318, "step": 4934 }, { "epoch": 0.66, "learning_rate": 3.4077241747962044e-08, "loss": 0.0464, "step": 4935 }, { "epoch": 0.66, "learning_rate": 3.406387812374716e-08, "loss": 0.0659, "step": 4936 }, { "epoch": 0.66, "learning_rate": 3.4050514499532276e-08, "loss": 0.0594, "step": 4937 }, { "epoch": 0.66, "learning_rate": 3.4037150875317386e-08, "loss": 0.1181, "step": 4938 }, { "epoch": 0.66, "learning_rate": 3.4023787251102496e-08, "loss": 0.0439, "step": 4939 }, { "epoch": 0.66, "learning_rate": 3.4010423626887606e-08, "loss": 0.0442, "step": 4940 }, { "epoch": 0.66, "learning_rate": 3.399706000267272e-08, "loss": 0.0528, "step": 4941 }, { "epoch": 0.66, "learning_rate": 3.398369637845784e-08, "loss": 0.0437, "step": 4942 }, { "epoch": 0.66, "learning_rate": 3.397033275424295e-08, "loss": 0.0453, "step": 4943 }, { "epoch": 0.66, "learning_rate": 3.3956969130028065e-08, "loss": 0.0467, "step": 4944 }, { "epoch": 0.66, "learning_rate": 3.3943605505813175e-08, "loss": 0.0507, "step": 4945 }, { "epoch": 0.66, "learning_rate": 3.3930241881598285e-08, "loss": 0.0425, "step": 4946 }, { "epoch": 0.66, "learning_rate": 3.39168782573834e-08, "loss": 0.0395, "step": 4947 }, { "epoch": 0.66, "learning_rate": 3.390351463316851e-08, "loss": 0.0563, "step": 4948 }, { "epoch": 0.66, "learning_rate": 3.389015100895363e-08, "loss": 0.0448, "step": 4949 }, { "epoch": 0.66, "learning_rate": 3.3876787384738744e-08, "loss": 0.0585, "step": 4950 }, { "epoch": 0.66, "learning_rate": 3.3863423760523854e-08, "loss": 0.0381, "step": 4951 }, { "epoch": 0.66, "learning_rate": 3.3850060136308964e-08, "loss": 0.0521, "step": 4952 }, { "epoch": 0.66, "learning_rate": 3.3836696512094073e-08, "loss": 0.0307, "step": 4953 }, { "epoch": 0.66, "learning_rate": 3.382333288787919e-08, "loss": 0.066, "step": 4954 }, { "epoch": 0.66, "learning_rate": 3.3809969263664306e-08, "loss": 0.0597, "step": 4955 }, { "epoch": 0.66, "learning_rate": 3.3796605639449416e-08, "loss": 0.0494, "step": 4956 }, { "epoch": 0.66, "learning_rate": 3.378324201523453e-08, "loss": 0.0646, "step": 4957 }, { "epoch": 0.66, "learning_rate": 3.376987839101964e-08, "loss": 0.0503, "step": 4958 }, { "epoch": 0.66, "learning_rate": 3.375651476680476e-08, "loss": 0.0588, "step": 4959 }, { "epoch": 0.66, "learning_rate": 3.374315114258987e-08, "loss": 0.052, "step": 4960 }, { "epoch": 0.66, "learning_rate": 3.372978751837498e-08, "loss": 0.0517, "step": 4961 }, { "epoch": 0.66, "learning_rate": 3.3716423894160095e-08, "loss": 0.0779, "step": 4962 }, { "epoch": 0.66, "learning_rate": 3.3703060269945205e-08, "loss": 0.0624, "step": 4963 }, { "epoch": 0.66, "learning_rate": 3.368969664573032e-08, "loss": 0.0615, "step": 4964 }, { "epoch": 0.66, "learning_rate": 3.367633302151544e-08, "loss": 0.0499, "step": 4965 }, { "epoch": 0.66, "learning_rate": 3.366296939730055e-08, "loss": 0.0352, "step": 4966 }, { "epoch": 0.66, "learning_rate": 3.364960577308566e-08, "loss": 0.0564, "step": 4967 }, { "epoch": 0.66, "learning_rate": 3.3636242148870774e-08, "loss": 0.0514, "step": 4968 }, { "epoch": 0.66, "learning_rate": 3.3622878524655884e-08, "loss": 0.034, "step": 4969 }, { "epoch": 0.66, "learning_rate": 3.3609514900441e-08, "loss": 0.0571, "step": 4970 }, { "epoch": 0.66, "learning_rate": 3.359615127622611e-08, "loss": 0.0401, "step": 4971 }, { "epoch": 0.66, "learning_rate": 3.3582787652011226e-08, "loss": 0.0539, "step": 4972 }, { "epoch": 0.66, "learning_rate": 3.3569424027796336e-08, "loss": 0.0668, "step": 4973 }, { "epoch": 0.66, "learning_rate": 3.3556060403581446e-08, "loss": 0.0471, "step": 4974 }, { "epoch": 0.66, "learning_rate": 3.354269677936656e-08, "loss": 0.056, "step": 4975 }, { "epoch": 0.66, "learning_rate": 3.352933315515167e-08, "loss": 0.0599, "step": 4976 }, { "epoch": 0.67, "learning_rate": 3.351596953093679e-08, "loss": 0.0492, "step": 4977 }, { "epoch": 0.67, "learning_rate": 3.3502605906721905e-08, "loss": 0.0537, "step": 4978 }, { "epoch": 0.67, "learning_rate": 3.3489242282507015e-08, "loss": 0.0422, "step": 4979 }, { "epoch": 0.67, "learning_rate": 3.3475878658292125e-08, "loss": 0.0295, "step": 4980 }, { "epoch": 0.67, "learning_rate": 3.3462515034077235e-08, "loss": 0.0694, "step": 4981 }, { "epoch": 0.67, "learning_rate": 3.344915140986235e-08, "loss": 0.0394, "step": 4982 }, { "epoch": 0.67, "learning_rate": 3.343578778564747e-08, "loss": 0.0413, "step": 4983 }, { "epoch": 0.67, "learning_rate": 3.342242416143258e-08, "loss": 0.0513, "step": 4984 }, { "epoch": 0.67, "learning_rate": 3.3409060537217694e-08, "loss": 0.0455, "step": 4985 }, { "epoch": 0.67, "learning_rate": 3.3395696913002804e-08, "loss": 0.0357, "step": 4986 }, { "epoch": 0.67, "learning_rate": 3.338233328878792e-08, "loss": 0.069, "step": 4987 }, { "epoch": 0.67, "learning_rate": 3.336896966457303e-08, "loss": 0.0523, "step": 4988 }, { "epoch": 0.67, "learning_rate": 3.335560604035814e-08, "loss": 0.0579, "step": 4989 }, { "epoch": 0.67, "learning_rate": 3.3342242416143256e-08, "loss": 0.0597, "step": 4990 }, { "epoch": 0.67, "learning_rate": 3.332887879192837e-08, "loss": 0.0625, "step": 4991 }, { "epoch": 0.67, "learning_rate": 3.331551516771348e-08, "loss": 0.0791, "step": 4992 }, { "epoch": 0.67, "learning_rate": 3.33021515434986e-08, "loss": 0.0653, "step": 4993 }, { "epoch": 0.67, "learning_rate": 3.328878791928371e-08, "loss": 0.0519, "step": 4994 }, { "epoch": 0.67, "learning_rate": 3.327542429506882e-08, "loss": 0.0456, "step": 4995 }, { "epoch": 0.67, "learning_rate": 3.3262060670853935e-08, "loss": 0.0644, "step": 4996 }, { "epoch": 0.67, "learning_rate": 3.3248697046639045e-08, "loss": 0.0616, "step": 4997 }, { "epoch": 0.67, "learning_rate": 3.323533342242416e-08, "loss": 0.0753, "step": 4998 }, { "epoch": 0.67, "learning_rate": 3.322196979820927e-08, "loss": 0.0632, "step": 4999 }, { "epoch": 0.67, "learning_rate": 3.320860617399439e-08, "loss": 0.045, "step": 5000 }, { "epoch": 0.67, "learning_rate": 3.31952425497795e-08, "loss": 0.0511, "step": 5001 }, { "epoch": 0.67, "learning_rate": 3.318187892556461e-08, "loss": 0.052, "step": 5002 }, { "epoch": 0.67, "learning_rate": 3.3168515301349724e-08, "loss": 0.0856, "step": 5003 }, { "epoch": 0.67, "learning_rate": 3.3155151677134834e-08, "loss": 0.1088, "step": 5004 }, { "epoch": 0.67, "learning_rate": 3.314178805291995e-08, "loss": 0.0591, "step": 5005 }, { "epoch": 0.67, "learning_rate": 3.3128424428705067e-08, "loss": 0.0488, "step": 5006 }, { "epoch": 0.67, "learning_rate": 3.3115060804490177e-08, "loss": 0.0751, "step": 5007 }, { "epoch": 0.67, "learning_rate": 3.310169718027529e-08, "loss": 0.0514, "step": 5008 }, { "epoch": 0.67, "learning_rate": 3.3088333556060396e-08, "loss": 0.0994, "step": 5009 }, { "epoch": 0.67, "learning_rate": 3.307496993184551e-08, "loss": 0.0561, "step": 5010 }, { "epoch": 0.67, "learning_rate": 3.306160630763063e-08, "loss": 0.0444, "step": 5011 }, { "epoch": 0.67, "learning_rate": 3.304824268341574e-08, "loss": 0.0668, "step": 5012 }, { "epoch": 0.67, "learning_rate": 3.3034879059200855e-08, "loss": 0.084, "step": 5013 }, { "epoch": 0.67, "learning_rate": 3.302151543498597e-08, "loss": 0.0472, "step": 5014 }, { "epoch": 0.67, "learning_rate": 3.300815181077108e-08, "loss": 0.0513, "step": 5015 }, { "epoch": 0.67, "learning_rate": 3.299478818655619e-08, "loss": 0.0813, "step": 5016 }, { "epoch": 0.67, "learning_rate": 3.29814245623413e-08, "loss": 0.0647, "step": 5017 }, { "epoch": 0.67, "learning_rate": 3.296806093812642e-08, "loss": 0.0588, "step": 5018 }, { "epoch": 0.67, "learning_rate": 3.2954697313911534e-08, "loss": 0.048, "step": 5019 }, { "epoch": 0.67, "learning_rate": 3.2941333689696644e-08, "loss": 0.067, "step": 5020 }, { "epoch": 0.67, "learning_rate": 3.292797006548176e-08, "loss": 0.0915, "step": 5021 }, { "epoch": 0.67, "learning_rate": 3.291460644126687e-08, "loss": 0.0628, "step": 5022 }, { "epoch": 0.67, "learning_rate": 3.290124281705198e-08, "loss": 0.0635, "step": 5023 }, { "epoch": 0.67, "learning_rate": 3.2887879192837097e-08, "loss": 0.0538, "step": 5024 }, { "epoch": 0.67, "learning_rate": 3.2874515568622206e-08, "loss": 0.0558, "step": 5025 }, { "epoch": 0.67, "learning_rate": 3.286115194440732e-08, "loss": 0.0678, "step": 5026 }, { "epoch": 0.67, "learning_rate": 3.284778832019243e-08, "loss": 0.0632, "step": 5027 }, { "epoch": 0.67, "learning_rate": 3.283442469597755e-08, "loss": 0.0519, "step": 5028 }, { "epoch": 0.67, "learning_rate": 3.2821061071762666e-08, "loss": 0.0681, "step": 5029 }, { "epoch": 0.67, "learning_rate": 3.280769744754777e-08, "loss": 0.0625, "step": 5030 }, { "epoch": 0.67, "learning_rate": 3.2794333823332885e-08, "loss": 0.0851, "step": 5031 }, { "epoch": 0.67, "learning_rate": 3.2780970199117995e-08, "loss": 0.0767, "step": 5032 }, { "epoch": 0.67, "learning_rate": 3.276760657490311e-08, "loss": 0.0487, "step": 5033 }, { "epoch": 0.67, "learning_rate": 3.275424295068823e-08, "loss": 0.0651, "step": 5034 }, { "epoch": 0.67, "learning_rate": 3.274087932647334e-08, "loss": 0.0447, "step": 5035 }, { "epoch": 0.67, "learning_rate": 3.2727515702258454e-08, "loss": 0.0632, "step": 5036 }, { "epoch": 0.67, "learning_rate": 3.2714152078043564e-08, "loss": 0.0491, "step": 5037 }, { "epoch": 0.67, "learning_rate": 3.2700788453828674e-08, "loss": 0.1146, "step": 5038 }, { "epoch": 0.67, "learning_rate": 3.268742482961379e-08, "loss": 0.0596, "step": 5039 }, { "epoch": 0.67, "learning_rate": 3.26740612053989e-08, "loss": 0.1482, "step": 5040 }, { "epoch": 0.67, "learning_rate": 3.266069758118402e-08, "loss": 0.0698, "step": 5041 }, { "epoch": 0.67, "learning_rate": 3.264733395696913e-08, "loss": 0.0599, "step": 5042 }, { "epoch": 0.67, "learning_rate": 3.263397033275424e-08, "loss": 0.0563, "step": 5043 }, { "epoch": 0.67, "learning_rate": 3.262060670853935e-08, "loss": 0.0788, "step": 5044 }, { "epoch": 0.67, "learning_rate": 3.260724308432446e-08, "loss": 0.056, "step": 5045 }, { "epoch": 0.67, "learning_rate": 3.259387946010958e-08, "loss": 0.0916, "step": 5046 }, { "epoch": 0.67, "learning_rate": 3.2580515835894696e-08, "loss": 0.0408, "step": 5047 }, { "epoch": 0.67, "learning_rate": 3.2567152211679805e-08, "loss": 0.0607, "step": 5048 }, { "epoch": 0.67, "learning_rate": 3.255378858746492e-08, "loss": 0.0764, "step": 5049 }, { "epoch": 0.67, "learning_rate": 3.254042496325003e-08, "loss": 0.0686, "step": 5050 }, { "epoch": 0.67, "learning_rate": 3.252706133903514e-08, "loss": 0.1117, "step": 5051 }, { "epoch": 0.68, "learning_rate": 3.251369771482026e-08, "loss": 0.0713, "step": 5052 }, { "epoch": 0.68, "learning_rate": 3.250033409060537e-08, "loss": 0.0594, "step": 5053 }, { "epoch": 0.68, "learning_rate": 3.2486970466390484e-08, "loss": 0.0911, "step": 5054 }, { "epoch": 0.68, "learning_rate": 3.24736068421756e-08, "loss": 0.0682, "step": 5055 }, { "epoch": 0.68, "learning_rate": 3.246024321796071e-08, "loss": 0.0503, "step": 5056 }, { "epoch": 0.68, "learning_rate": 3.244687959374583e-08, "loss": 0.0465, "step": 5057 }, { "epoch": 0.68, "learning_rate": 3.243351596953093e-08, "loss": 0.0542, "step": 5058 }, { "epoch": 0.68, "learning_rate": 3.2420152345316047e-08, "loss": 0.0621, "step": 5059 }, { "epoch": 0.68, "learning_rate": 3.240678872110116e-08, "loss": 0.0704, "step": 5060 }, { "epoch": 0.68, "learning_rate": 3.239342509688627e-08, "loss": 0.0479, "step": 5061 }, { "epoch": 0.68, "learning_rate": 3.238006147267139e-08, "loss": 0.1124, "step": 5062 }, { "epoch": 0.68, "learning_rate": 3.23666978484565e-08, "loss": 0.0529, "step": 5063 }, { "epoch": 0.68, "learning_rate": 3.2353334224241616e-08, "loss": 0.0765, "step": 5064 }, { "epoch": 0.68, "learning_rate": 3.2339970600026726e-08, "loss": 0.0515, "step": 5065 }, { "epoch": 0.68, "learning_rate": 3.2326606975811835e-08, "loss": 0.0453, "step": 5066 }, { "epoch": 0.68, "learning_rate": 3.231324335159695e-08, "loss": 0.0689, "step": 5067 }, { "epoch": 0.68, "learning_rate": 3.229987972738206e-08, "loss": 0.0775, "step": 5068 }, { "epoch": 0.68, "learning_rate": 3.228651610316718e-08, "loss": 0.0663, "step": 5069 }, { "epoch": 0.68, "learning_rate": 3.2273152478952295e-08, "loss": 0.0554, "step": 5070 }, { "epoch": 0.68, "learning_rate": 3.2259788854737404e-08, "loss": 0.0509, "step": 5071 }, { "epoch": 0.68, "learning_rate": 3.2246425230522514e-08, "loss": 0.0553, "step": 5072 }, { "epoch": 0.68, "learning_rate": 3.2233061606307624e-08, "loss": 0.0537, "step": 5073 }, { "epoch": 0.68, "learning_rate": 3.221969798209274e-08, "loss": 0.0488, "step": 5074 }, { "epoch": 0.68, "learning_rate": 3.220633435787786e-08, "loss": 0.0861, "step": 5075 }, { "epoch": 0.68, "learning_rate": 3.219297073366297e-08, "loss": 0.0559, "step": 5076 }, { "epoch": 0.68, "learning_rate": 3.217960710944808e-08, "loss": 0.0573, "step": 5077 }, { "epoch": 0.68, "learning_rate": 3.21662434852332e-08, "loss": 0.0802, "step": 5078 }, { "epoch": 0.68, "learning_rate": 3.21528798610183e-08, "loss": 0.0645, "step": 5079 }, { "epoch": 0.68, "learning_rate": 3.213951623680342e-08, "loss": 0.0495, "step": 5080 }, { "epoch": 0.68, "learning_rate": 3.212615261258853e-08, "loss": 0.05, "step": 5081 }, { "epoch": 0.68, "learning_rate": 3.2112788988373646e-08, "loss": 0.0697, "step": 5082 }, { "epoch": 0.68, "learning_rate": 3.209942536415876e-08, "loss": 0.0939, "step": 5083 }, { "epoch": 0.68, "learning_rate": 3.208606173994387e-08, "loss": 0.0691, "step": 5084 }, { "epoch": 0.68, "learning_rate": 3.207269811572899e-08, "loss": 0.0711, "step": 5085 }, { "epoch": 0.68, "learning_rate": 3.20593344915141e-08, "loss": 0.0547, "step": 5086 }, { "epoch": 0.68, "learning_rate": 3.204597086729921e-08, "loss": 0.0749, "step": 5087 }, { "epoch": 0.68, "learning_rate": 3.2032607243084324e-08, "loss": 0.0736, "step": 5088 }, { "epoch": 0.68, "learning_rate": 3.2019243618869434e-08, "loss": 0.0571, "step": 5089 }, { "epoch": 0.68, "learning_rate": 3.200587999465455e-08, "loss": 0.072, "step": 5090 }, { "epoch": 0.68, "learning_rate": 3.199251637043966e-08, "loss": 0.0466, "step": 5091 }, { "epoch": 0.68, "learning_rate": 3.197915274622478e-08, "loss": 0.0535, "step": 5092 }, { "epoch": 0.68, "learning_rate": 3.196578912200989e-08, "loss": 0.054, "step": 5093 }, { "epoch": 0.68, "learning_rate": 3.1952425497795e-08, "loss": 0.055, "step": 5094 }, { "epoch": 0.68, "learning_rate": 3.193906187358011e-08, "loss": 0.0785, "step": 5095 }, { "epoch": 0.68, "learning_rate": 3.192569824936522e-08, "loss": 0.0382, "step": 5096 }, { "epoch": 0.68, "learning_rate": 3.191233462515034e-08, "loss": 0.0805, "step": 5097 }, { "epoch": 0.68, "learning_rate": 3.1898971000935456e-08, "loss": 0.0727, "step": 5098 }, { "epoch": 0.68, "learning_rate": 3.1885607376720566e-08, "loss": 0.0492, "step": 5099 }, { "epoch": 0.68, "learning_rate": 3.1872243752505676e-08, "loss": 0.067, "step": 5100 }, { "epoch": 0.68, "learning_rate": 3.185888012829079e-08, "loss": 0.0453, "step": 5101 }, { "epoch": 0.68, "learning_rate": 3.18455165040759e-08, "loss": 0.0544, "step": 5102 }, { "epoch": 0.68, "learning_rate": 3.183215287986102e-08, "loss": 0.0552, "step": 5103 }, { "epoch": 0.68, "learning_rate": 3.181878925564613e-08, "loss": 0.0425, "step": 5104 }, { "epoch": 0.68, "learning_rate": 3.1805425631431245e-08, "loss": 0.0573, "step": 5105 }, { "epoch": 0.68, "learning_rate": 3.179206200721636e-08, "loss": 0.0489, "step": 5106 }, { "epoch": 0.68, "learning_rate": 3.177869838300147e-08, "loss": 0.0715, "step": 5107 }, { "epoch": 0.68, "learning_rate": 3.176533475878658e-08, "loss": 0.0618, "step": 5108 }, { "epoch": 0.68, "learning_rate": 3.175197113457169e-08, "loss": 0.0564, "step": 5109 }, { "epoch": 0.68, "learning_rate": 3.173860751035681e-08, "loss": 0.0811, "step": 5110 }, { "epoch": 0.68, "learning_rate": 3.1725243886141923e-08, "loss": 0.0853, "step": 5111 }, { "epoch": 0.68, "learning_rate": 3.171188026192703e-08, "loss": 0.0806, "step": 5112 }, { "epoch": 0.68, "learning_rate": 3.169851663771215e-08, "loss": 0.0437, "step": 5113 }, { "epoch": 0.68, "learning_rate": 3.168515301349726e-08, "loss": 0.0694, "step": 5114 }, { "epoch": 0.68, "learning_rate": 3.167178938928237e-08, "loss": 0.0543, "step": 5115 }, { "epoch": 0.68, "learning_rate": 3.1658425765067486e-08, "loss": 0.0366, "step": 5116 }, { "epoch": 0.68, "learning_rate": 3.1645062140852596e-08, "loss": 0.0705, "step": 5117 }, { "epoch": 0.68, "learning_rate": 3.163169851663771e-08, "loss": 0.0551, "step": 5118 }, { "epoch": 0.68, "learning_rate": 3.161833489242282e-08, "loss": 0.0956, "step": 5119 }, { "epoch": 0.68, "learning_rate": 3.160497126820794e-08, "loss": 0.0903, "step": 5120 }, { "epoch": 0.68, "learning_rate": 3.159160764399305e-08, "loss": 0.1066, "step": 5121 }, { "epoch": 0.68, "learning_rate": 3.157824401977816e-08, "loss": 0.0658, "step": 5122 }, { "epoch": 0.68, "learning_rate": 3.1564880395563275e-08, "loss": 0.0616, "step": 5123 }, { "epoch": 0.68, "learning_rate": 3.155151677134839e-08, "loss": 0.0995, "step": 5124 }, { "epoch": 0.68, "learning_rate": 3.15381531471335e-08, "loss": 0.0543, "step": 5125 }, { "epoch": 0.69, "learning_rate": 3.152478952291862e-08, "loss": 0.0493, "step": 5126 }, { "epoch": 0.69, "learning_rate": 3.151142589870373e-08, "loss": 0.0437, "step": 5127 }, { "epoch": 0.69, "learning_rate": 3.1498062274488844e-08, "loss": 0.0967, "step": 5128 }, { "epoch": 0.69, "learning_rate": 3.1484698650273953e-08, "loss": 0.0931, "step": 5129 }, { "epoch": 0.69, "learning_rate": 3.147133502605906e-08, "loss": 0.0509, "step": 5130 }, { "epoch": 0.69, "learning_rate": 3.145797140184418e-08, "loss": 0.0601, "step": 5131 }, { "epoch": 0.69, "learning_rate": 3.144460777762929e-08, "loss": 0.0378, "step": 5132 }, { "epoch": 0.69, "learning_rate": 3.1431244153414406e-08, "loss": 0.0582, "step": 5133 }, { "epoch": 0.69, "learning_rate": 3.141788052919952e-08, "loss": 0.0551, "step": 5134 }, { "epoch": 0.69, "learning_rate": 3.140451690498463e-08, "loss": 0.0792, "step": 5135 }, { "epoch": 0.69, "learning_rate": 3.139115328076974e-08, "loss": 0.0791, "step": 5136 }, { "epoch": 0.69, "learning_rate": 3.137778965655485e-08, "loss": 0.0533, "step": 5137 }, { "epoch": 0.69, "learning_rate": 3.136442603233997e-08, "loss": 0.0449, "step": 5138 }, { "epoch": 0.69, "learning_rate": 3.1351062408125085e-08, "loss": 0.045, "step": 5139 }, { "epoch": 0.69, "learning_rate": 3.1337698783910195e-08, "loss": 0.0603, "step": 5140 }, { "epoch": 0.69, "learning_rate": 3.132433515969531e-08, "loss": 0.0441, "step": 5141 }, { "epoch": 0.69, "learning_rate": 3.131097153548042e-08, "loss": 0.0729, "step": 5142 }, { "epoch": 0.69, "learning_rate": 3.129760791126553e-08, "loss": 0.0705, "step": 5143 }, { "epoch": 0.69, "learning_rate": 3.128424428705065e-08, "loss": 0.0741, "step": 5144 }, { "epoch": 0.69, "learning_rate": 3.127088066283576e-08, "loss": 0.0639, "step": 5145 }, { "epoch": 0.69, "learning_rate": 3.1257517038620873e-08, "loss": 0.0427, "step": 5146 }, { "epoch": 0.69, "learning_rate": 3.124415341440599e-08, "loss": 0.0556, "step": 5147 }, { "epoch": 0.69, "learning_rate": 3.12307897901911e-08, "loss": 0.0741, "step": 5148 }, { "epoch": 0.69, "learning_rate": 3.121742616597621e-08, "loss": 0.0677, "step": 5149 }, { "epoch": 0.69, "learning_rate": 3.120406254176132e-08, "loss": 0.0539, "step": 5150 }, { "epoch": 0.69, "learning_rate": 3.1190698917546436e-08, "loss": 0.0527, "step": 5151 }, { "epoch": 0.69, "learning_rate": 3.117733529333155e-08, "loss": 0.0798, "step": 5152 }, { "epoch": 0.69, "learning_rate": 3.116397166911666e-08, "loss": 0.0838, "step": 5153 }, { "epoch": 0.69, "learning_rate": 3.115060804490178e-08, "loss": 0.0903, "step": 5154 }, { "epoch": 0.69, "learning_rate": 3.113724442068689e-08, "loss": 0.0639, "step": 5155 }, { "epoch": 0.69, "learning_rate": 3.1123880796472005e-08, "loss": 0.0635, "step": 5156 }, { "epoch": 0.69, "learning_rate": 3.1110517172257115e-08, "loss": 0.055, "step": 5157 }, { "epoch": 0.69, "learning_rate": 3.1097153548042225e-08, "loss": 0.056, "step": 5158 }, { "epoch": 0.69, "learning_rate": 3.108378992382734e-08, "loss": 0.0797, "step": 5159 }, { "epoch": 0.69, "learning_rate": 3.107042629961245e-08, "loss": 0.0563, "step": 5160 }, { "epoch": 0.69, "learning_rate": 3.105706267539757e-08, "loss": 0.0634, "step": 5161 }, { "epoch": 0.69, "learning_rate": 3.1043699051182684e-08, "loss": 0.0723, "step": 5162 }, { "epoch": 0.69, "learning_rate": 3.1030335426967794e-08, "loss": 0.0694, "step": 5163 }, { "epoch": 0.69, "learning_rate": 3.1016971802752903e-08, "loss": 0.0585, "step": 5164 }, { "epoch": 0.69, "learning_rate": 3.100360817853801e-08, "loss": 0.0712, "step": 5165 }, { "epoch": 0.69, "learning_rate": 3.099024455432313e-08, "loss": 0.0639, "step": 5166 }, { "epoch": 0.69, "learning_rate": 3.0976880930108246e-08, "loss": 0.0735, "step": 5167 }, { "epoch": 0.69, "learning_rate": 3.0963517305893356e-08, "loss": 0.0864, "step": 5168 }, { "epoch": 0.69, "learning_rate": 3.095015368167847e-08, "loss": 0.095, "step": 5169 }, { "epoch": 0.69, "learning_rate": 3.093679005746358e-08, "loss": 0.0559, "step": 5170 }, { "epoch": 0.69, "learning_rate": 3.092342643324869e-08, "loss": 0.0613, "step": 5171 }, { "epoch": 0.69, "learning_rate": 3.091006280903381e-08, "loss": 0.0446, "step": 5172 }, { "epoch": 0.69, "learning_rate": 3.089669918481892e-08, "loss": 0.0675, "step": 5173 }, { "epoch": 0.69, "learning_rate": 3.0883335560604035e-08, "loss": 0.1075, "step": 5174 }, { "epoch": 0.69, "learning_rate": 3.086997193638915e-08, "loss": 0.0804, "step": 5175 }, { "epoch": 0.69, "learning_rate": 3.085660831217426e-08, "loss": 0.0793, "step": 5176 }, { "epoch": 0.69, "learning_rate": 3.084324468795938e-08, "loss": 0.0588, "step": 5177 }, { "epoch": 0.69, "learning_rate": 3.082988106374448e-08, "loss": 0.0688, "step": 5178 }, { "epoch": 0.69, "learning_rate": 3.08165174395296e-08, "loss": 0.0691, "step": 5179 }, { "epoch": 0.69, "learning_rate": 3.0803153815314714e-08, "loss": 0.0672, "step": 5180 }, { "epoch": 0.69, "learning_rate": 3.0789790191099824e-08, "loss": 0.0786, "step": 5181 }, { "epoch": 0.69, "learning_rate": 3.077642656688494e-08, "loss": 0.0747, "step": 5182 }, { "epoch": 0.69, "learning_rate": 3.076306294267005e-08, "loss": 0.0526, "step": 5183 }, { "epoch": 0.69, "learning_rate": 3.0749699318455166e-08, "loss": 0.116, "step": 5184 }, { "epoch": 0.69, "learning_rate": 3.0736335694240276e-08, "loss": 0.0752, "step": 5185 }, { "epoch": 0.69, "learning_rate": 3.0722972070025386e-08, "loss": 0.0776, "step": 5186 }, { "epoch": 0.69, "learning_rate": 3.07096084458105e-08, "loss": 0.0581, "step": 5187 }, { "epoch": 0.69, "learning_rate": 3.069624482159561e-08, "loss": 0.0563, "step": 5188 }, { "epoch": 0.69, "learning_rate": 3.068288119738073e-08, "loss": 0.0597, "step": 5189 }, { "epoch": 0.69, "learning_rate": 3.0669517573165845e-08, "loss": 0.0322, "step": 5190 }, { "epoch": 0.69, "learning_rate": 3.0656153948950955e-08, "loss": 0.0434, "step": 5191 }, { "epoch": 0.69, "learning_rate": 3.0642790324736065e-08, "loss": 0.1044, "step": 5192 }, { "epoch": 0.69, "learning_rate": 3.062942670052118e-08, "loss": 0.0773, "step": 5193 }, { "epoch": 0.69, "learning_rate": 3.061606307630629e-08, "loss": 0.0775, "step": 5194 }, { "epoch": 0.69, "learning_rate": 3.060269945209141e-08, "loss": 0.099, "step": 5195 }, { "epoch": 0.69, "learning_rate": 3.058933582787652e-08, "loss": 0.0673, "step": 5196 }, { "epoch": 0.69, "learning_rate": 3.0575972203661634e-08, "loss": 0.0643, "step": 5197 }, { "epoch": 0.69, "learning_rate": 3.056260857944675e-08, "loss": 0.06, "step": 5198 }, { "epoch": 0.69, "learning_rate": 3.0549244955231853e-08, "loss": 0.0489, "step": 5199 }, { "epoch": 0.69, "learning_rate": 3.053588133101697e-08, "loss": 0.0712, "step": 5200 }, { "epoch": 0.7, "learning_rate": 3.052251770680208e-08, "loss": 0.1153, "step": 5201 }, { "epoch": 0.7, "learning_rate": 3.0509154082587196e-08, "loss": 0.0957, "step": 5202 }, { "epoch": 0.7, "learning_rate": 3.049579045837231e-08, "loss": 0.0793, "step": 5203 }, { "epoch": 0.7, "learning_rate": 3.048242683415742e-08, "loss": 0.0586, "step": 5204 }, { "epoch": 0.7, "learning_rate": 3.046906320994254e-08, "loss": 0.0788, "step": 5205 }, { "epoch": 0.7, "learning_rate": 3.045569958572765e-08, "loss": 0.0621, "step": 5206 }, { "epoch": 0.7, "learning_rate": 3.044233596151276e-08, "loss": 0.0381, "step": 5207 }, { "epoch": 0.7, "learning_rate": 3.0428972337297875e-08, "loss": 0.0687, "step": 5208 }, { "epoch": 0.7, "learning_rate": 3.0415608713082985e-08, "loss": 0.0678, "step": 5209 }, { "epoch": 0.7, "learning_rate": 3.04022450888681e-08, "loss": 0.0931, "step": 5210 }, { "epoch": 0.7, "learning_rate": 3.038888146465321e-08, "loss": 0.0533, "step": 5211 }, { "epoch": 0.7, "learning_rate": 3.037551784043833e-08, "loss": 0.0571, "step": 5212 }, { "epoch": 0.7, "learning_rate": 3.036215421622344e-08, "loss": 0.0588, "step": 5213 }, { "epoch": 0.7, "learning_rate": 3.034879059200855e-08, "loss": 0.0615, "step": 5214 }, { "epoch": 0.7, "learning_rate": 3.0335426967793664e-08, "loss": 0.067, "step": 5215 }, { "epoch": 0.7, "learning_rate": 3.032206334357878e-08, "loss": 0.0417, "step": 5216 }, { "epoch": 0.7, "learning_rate": 3.030869971936389e-08, "loss": 0.0556, "step": 5217 }, { "epoch": 0.7, "learning_rate": 3.0295336095149006e-08, "loss": 0.0633, "step": 5218 }, { "epoch": 0.7, "learning_rate": 3.0281972470934116e-08, "loss": 0.0573, "step": 5219 }, { "epoch": 0.7, "learning_rate": 3.0268608846719226e-08, "loss": 0.063, "step": 5220 }, { "epoch": 0.7, "learning_rate": 3.025524522250434e-08, "loss": 0.0594, "step": 5221 }, { "epoch": 0.7, "learning_rate": 3.024188159828945e-08, "loss": 0.1171, "step": 5222 }, { "epoch": 0.7, "learning_rate": 3.022851797407457e-08, "loss": 0.0734, "step": 5223 }, { "epoch": 0.7, "learning_rate": 3.021515434985968e-08, "loss": 0.0829, "step": 5224 }, { "epoch": 0.7, "learning_rate": 3.0201790725644795e-08, "loss": 0.0633, "step": 5225 }, { "epoch": 0.7, "learning_rate": 3.018842710142991e-08, "loss": 0.064, "step": 5226 }, { "epoch": 0.7, "learning_rate": 3.0175063477215015e-08, "loss": 0.0819, "step": 5227 }, { "epoch": 0.7, "learning_rate": 3.016169985300013e-08, "loss": 0.0467, "step": 5228 }, { "epoch": 0.7, "learning_rate": 3.014833622878524e-08, "loss": 0.0539, "step": 5229 }, { "epoch": 0.7, "learning_rate": 3.013497260457036e-08, "loss": 0.0793, "step": 5230 }, { "epoch": 0.7, "learning_rate": 3.0121608980355474e-08, "loss": 0.0723, "step": 5231 }, { "epoch": 0.7, "learning_rate": 3.0108245356140584e-08, "loss": 0.05, "step": 5232 }, { "epoch": 0.7, "learning_rate": 3.00948817319257e-08, "loss": 0.0923, "step": 5233 }, { "epoch": 0.7, "learning_rate": 3.008151810771081e-08, "loss": 0.0752, "step": 5234 }, { "epoch": 0.7, "learning_rate": 3.006815448349592e-08, "loss": 0.0415, "step": 5235 }, { "epoch": 0.7, "learning_rate": 3.0054790859281036e-08, "loss": 0.0603, "step": 5236 }, { "epoch": 0.7, "learning_rate": 3.0041427235066146e-08, "loss": 0.0799, "step": 5237 }, { "epoch": 0.7, "learning_rate": 3.002806361085126e-08, "loss": 0.0642, "step": 5238 }, { "epoch": 0.7, "learning_rate": 3.001469998663638e-08, "loss": 0.0537, "step": 5239 }, { "epoch": 0.7, "learning_rate": 3.000133636242149e-08, "loss": 0.056, "step": 5240 }, { "epoch": 0.7, "learning_rate": 2.99879727382066e-08, "loss": 0.0708, "step": 5241 }, { "epoch": 0.7, "learning_rate": 2.997460911399171e-08, "loss": 0.0568, "step": 5242 }, { "epoch": 0.7, "learning_rate": 2.9961245489776825e-08, "loss": 0.0806, "step": 5243 }, { "epoch": 0.7, "learning_rate": 2.994788186556194e-08, "loss": 0.0546, "step": 5244 }, { "epoch": 0.7, "learning_rate": 2.993451824134705e-08, "loss": 0.0698, "step": 5245 }, { "epoch": 0.7, "learning_rate": 2.992115461713217e-08, "loss": 0.0591, "step": 5246 }, { "epoch": 0.7, "learning_rate": 2.990779099291728e-08, "loss": 0.0437, "step": 5247 }, { "epoch": 0.7, "learning_rate": 2.989442736870239e-08, "loss": 0.0419, "step": 5248 }, { "epoch": 0.7, "learning_rate": 2.9881063744487504e-08, "loss": 0.0748, "step": 5249 }, { "epoch": 0.7, "learning_rate": 2.9867700120272614e-08, "loss": 0.0529, "step": 5250 }, { "epoch": 0.7, "learning_rate": 2.985433649605773e-08, "loss": 0.0884, "step": 5251 }, { "epoch": 0.7, "learning_rate": 2.984097287184284e-08, "loss": 0.0856, "step": 5252 }, { "epoch": 0.7, "learning_rate": 2.9827609247627957e-08, "loss": 0.0724, "step": 5253 }, { "epoch": 0.7, "learning_rate": 2.981424562341307e-08, "loss": 0.0566, "step": 5254 }, { "epoch": 0.7, "learning_rate": 2.980088199919818e-08, "loss": 0.0468, "step": 5255 }, { "epoch": 0.7, "learning_rate": 2.9787518374983296e-08, "loss": 0.057, "step": 5256 }, { "epoch": 0.7, "learning_rate": 2.9774154750768406e-08, "loss": 0.0505, "step": 5257 }, { "epoch": 0.7, "learning_rate": 2.976079112655352e-08, "loss": 0.0427, "step": 5258 }, { "epoch": 0.7, "learning_rate": 2.9747427502338635e-08, "loss": 0.0698, "step": 5259 }, { "epoch": 0.7, "learning_rate": 2.9734063878123745e-08, "loss": 0.0742, "step": 5260 }, { "epoch": 0.7, "learning_rate": 2.972070025390886e-08, "loss": 0.0819, "step": 5261 }, { "epoch": 0.7, "learning_rate": 2.9707336629693975e-08, "loss": 0.0713, "step": 5262 }, { "epoch": 0.7, "learning_rate": 2.9693973005479085e-08, "loss": 0.0598, "step": 5263 }, { "epoch": 0.7, "learning_rate": 2.9680609381264198e-08, "loss": 0.1702, "step": 5264 }, { "epoch": 0.7, "learning_rate": 2.9667245757049308e-08, "loss": 0.0481, "step": 5265 }, { "epoch": 0.7, "learning_rate": 2.9653882132834424e-08, "loss": 0.0782, "step": 5266 }, { "epoch": 0.7, "learning_rate": 2.9640518508619537e-08, "loss": 0.0914, "step": 5267 }, { "epoch": 0.7, "learning_rate": 2.9627154884404647e-08, "loss": 0.0847, "step": 5268 }, { "epoch": 0.7, "learning_rate": 2.9613791260189763e-08, "loss": 0.0818, "step": 5269 }, { "epoch": 0.7, "learning_rate": 2.9600427635974873e-08, "loss": 0.0481, "step": 5270 }, { "epoch": 0.7, "learning_rate": 2.9587064011759986e-08, "loss": 0.0566, "step": 5271 }, { "epoch": 0.7, "learning_rate": 2.9573700387545103e-08, "loss": 0.0961, "step": 5272 }, { "epoch": 0.7, "learning_rate": 2.9560336763330213e-08, "loss": 0.0504, "step": 5273 }, { "epoch": 0.7, "learning_rate": 2.954697313911533e-08, "loss": 0.0447, "step": 5274 }, { "epoch": 0.7, "learning_rate": 2.9533609514900436e-08, "loss": 0.0546, "step": 5275 }, { "epoch": 0.71, "learning_rate": 2.9520245890685552e-08, "loss": 0.0431, "step": 5276 }, { "epoch": 0.71, "learning_rate": 2.950688226647067e-08, "loss": 0.075, "step": 5277 }, { "epoch": 0.71, "learning_rate": 2.949351864225578e-08, "loss": 0.079, "step": 5278 }, { "epoch": 0.71, "learning_rate": 2.948015501804089e-08, "loss": 0.0497, "step": 5279 }, { "epoch": 0.71, "learning_rate": 2.9466791393826e-08, "loss": 0.0478, "step": 5280 }, { "epoch": 0.71, "learning_rate": 2.9453427769611118e-08, "loss": 0.0687, "step": 5281 }, { "epoch": 0.71, "learning_rate": 2.944006414539623e-08, "loss": 0.0512, "step": 5282 }, { "epoch": 0.71, "learning_rate": 2.942670052118134e-08, "loss": 0.0635, "step": 5283 }, { "epoch": 0.71, "learning_rate": 2.9413336896966457e-08, "loss": 0.0546, "step": 5284 }, { "epoch": 0.71, "learning_rate": 2.939997327275157e-08, "loss": 0.0585, "step": 5285 }, { "epoch": 0.71, "learning_rate": 2.938660964853668e-08, "loss": 0.0684, "step": 5286 }, { "epoch": 0.71, "learning_rate": 2.9373246024321797e-08, "loss": 0.0544, "step": 5287 }, { "epoch": 0.71, "learning_rate": 2.9359882400106907e-08, "loss": 0.0542, "step": 5288 }, { "epoch": 0.71, "learning_rate": 2.934651877589202e-08, "loss": 0.0569, "step": 5289 }, { "epoch": 0.71, "learning_rate": 2.9333155151677136e-08, "loss": 0.051, "step": 5290 }, { "epoch": 0.71, "learning_rate": 2.9319791527462246e-08, "loss": 0.0584, "step": 5291 }, { "epoch": 0.71, "learning_rate": 2.930642790324736e-08, "loss": 0.0617, "step": 5292 }, { "epoch": 0.71, "learning_rate": 2.929306427903247e-08, "loss": 0.0799, "step": 5293 }, { "epoch": 0.71, "learning_rate": 2.9279700654817585e-08, "loss": 0.0899, "step": 5294 }, { "epoch": 0.71, "learning_rate": 2.92663370306027e-08, "loss": 0.0828, "step": 5295 }, { "epoch": 0.71, "learning_rate": 2.925297340638781e-08, "loss": 0.0498, "step": 5296 }, { "epoch": 0.71, "learning_rate": 2.9239609782172925e-08, "loss": 0.0679, "step": 5297 }, { "epoch": 0.71, "learning_rate": 2.9226246157958035e-08, "loss": 0.0628, "step": 5298 }, { "epoch": 0.71, "learning_rate": 2.921288253374315e-08, "loss": 0.0677, "step": 5299 }, { "epoch": 0.71, "learning_rate": 2.9199518909528264e-08, "loss": 0.074, "step": 5300 }, { "epoch": 0.71, "learning_rate": 2.9186155285313374e-08, "loss": 0.0547, "step": 5301 }, { "epoch": 0.71, "learning_rate": 2.917279166109849e-08, "loss": 0.0638, "step": 5302 }, { "epoch": 0.71, "learning_rate": 2.91594280368836e-08, "loss": 0.0995, "step": 5303 }, { "epoch": 0.71, "learning_rate": 2.9146064412668714e-08, "loss": 0.0533, "step": 5304 }, { "epoch": 0.71, "learning_rate": 2.913270078845383e-08, "loss": 0.0629, "step": 5305 }, { "epoch": 0.71, "learning_rate": 2.911933716423894e-08, "loss": 0.0521, "step": 5306 }, { "epoch": 0.71, "learning_rate": 2.9105973540024053e-08, "loss": 0.0649, "step": 5307 }, { "epoch": 0.71, "learning_rate": 2.909260991580917e-08, "loss": 0.0701, "step": 5308 }, { "epoch": 0.71, "learning_rate": 2.907924629159428e-08, "loss": 0.0688, "step": 5309 }, { "epoch": 0.71, "learning_rate": 2.9065882667379392e-08, "loss": 0.0542, "step": 5310 }, { "epoch": 0.71, "learning_rate": 2.9052519043164502e-08, "loss": 0.0496, "step": 5311 }, { "epoch": 0.71, "learning_rate": 2.903915541894962e-08, "loss": 0.0717, "step": 5312 }, { "epoch": 0.71, "learning_rate": 2.9025791794734732e-08, "loss": 0.0695, "step": 5313 }, { "epoch": 0.71, "learning_rate": 2.901242817051984e-08, "loss": 0.0773, "step": 5314 }, { "epoch": 0.71, "learning_rate": 2.8999064546304958e-08, "loss": 0.0607, "step": 5315 }, { "epoch": 0.71, "learning_rate": 2.8985700922090068e-08, "loss": 0.0416, "step": 5316 }, { "epoch": 0.71, "learning_rate": 2.897233729787518e-08, "loss": 0.066, "step": 5317 }, { "epoch": 0.71, "learning_rate": 2.8958973673660298e-08, "loss": 0.0435, "step": 5318 }, { "epoch": 0.71, "learning_rate": 2.8945610049445407e-08, "loss": 0.0481, "step": 5319 }, { "epoch": 0.71, "learning_rate": 2.893224642523052e-08, "loss": 0.0654, "step": 5320 }, { "epoch": 0.71, "learning_rate": 2.891888280101563e-08, "loss": 0.0639, "step": 5321 }, { "epoch": 0.71, "learning_rate": 2.8905519176800747e-08, "loss": 0.0565, "step": 5322 }, { "epoch": 0.71, "learning_rate": 2.8892155552585863e-08, "loss": 0.072, "step": 5323 }, { "epoch": 0.71, "learning_rate": 2.887879192837097e-08, "loss": 0.0731, "step": 5324 }, { "epoch": 0.71, "learning_rate": 2.8865428304156086e-08, "loss": 0.0492, "step": 5325 }, { "epoch": 0.71, "learning_rate": 2.8852064679941196e-08, "loss": 0.0689, "step": 5326 }, { "epoch": 0.71, "learning_rate": 2.8838701055726312e-08, "loss": 0.0725, "step": 5327 }, { "epoch": 0.71, "learning_rate": 2.8825337431511426e-08, "loss": 0.0348, "step": 5328 }, { "epoch": 0.71, "learning_rate": 2.8811973807296535e-08, "loss": 0.0521, "step": 5329 }, { "epoch": 0.71, "learning_rate": 2.8798610183081652e-08, "loss": 0.0607, "step": 5330 }, { "epoch": 0.71, "learning_rate": 2.8785246558866765e-08, "loss": 0.063, "step": 5331 }, { "epoch": 0.71, "learning_rate": 2.8771882934651875e-08, "loss": 0.0621, "step": 5332 }, { "epoch": 0.71, "learning_rate": 2.875851931043699e-08, "loss": 0.0779, "step": 5333 }, { "epoch": 0.71, "learning_rate": 2.87451556862221e-08, "loss": 0.0861, "step": 5334 }, { "epoch": 0.71, "learning_rate": 2.8731792062007214e-08, "loss": 0.0508, "step": 5335 }, { "epoch": 0.71, "learning_rate": 2.871842843779233e-08, "loss": 0.0579, "step": 5336 }, { "epoch": 0.71, "learning_rate": 2.870506481357744e-08, "loss": 0.0493, "step": 5337 }, { "epoch": 0.71, "learning_rate": 2.8691701189362554e-08, "loss": 0.0628, "step": 5338 }, { "epoch": 0.71, "learning_rate": 2.8678337565147664e-08, "loss": 0.0722, "step": 5339 }, { "epoch": 0.71, "learning_rate": 2.866497394093278e-08, "loss": 0.0596, "step": 5340 }, { "epoch": 0.71, "learning_rate": 2.8651610316717893e-08, "loss": 0.0529, "step": 5341 }, { "epoch": 0.71, "learning_rate": 2.8638246692503003e-08, "loss": 0.0523, "step": 5342 }, { "epoch": 0.71, "learning_rate": 2.862488306828812e-08, "loss": 0.0531, "step": 5343 }, { "epoch": 0.71, "learning_rate": 2.861151944407323e-08, "loss": 0.048, "step": 5344 }, { "epoch": 0.71, "learning_rate": 2.8598155819858342e-08, "loss": 0.0645, "step": 5345 }, { "epoch": 0.71, "learning_rate": 2.858479219564346e-08, "loss": 0.0613, "step": 5346 }, { "epoch": 0.71, "learning_rate": 2.857142857142857e-08, "loss": 0.066, "step": 5347 }, { "epoch": 0.71, "learning_rate": 2.8558064947213685e-08, "loss": 0.0835, "step": 5348 }, { "epoch": 0.71, "learning_rate": 2.8544701322998798e-08, "loss": 0.0935, "step": 5349 }, { "epoch": 0.71, "learning_rate": 2.8531337698783908e-08, "loss": 0.0405, "step": 5350 }, { "epoch": 0.72, "learning_rate": 2.8517974074569025e-08, "loss": 0.0718, "step": 5351 }, { "epoch": 0.72, "learning_rate": 2.8504610450354134e-08, "loss": 0.0536, "step": 5352 }, { "epoch": 0.72, "learning_rate": 2.8491246826139248e-08, "loss": 0.0591, "step": 5353 }, { "epoch": 0.72, "learning_rate": 2.8477883201924364e-08, "loss": 0.0645, "step": 5354 }, { "epoch": 0.72, "learning_rate": 2.8464519577709474e-08, "loss": 0.0671, "step": 5355 }, { "epoch": 0.72, "learning_rate": 2.8451155953494587e-08, "loss": 0.0629, "step": 5356 }, { "epoch": 0.72, "learning_rate": 2.8437792329279697e-08, "loss": 0.0669, "step": 5357 }, { "epoch": 0.72, "learning_rate": 2.8424428705064813e-08, "loss": 0.0582, "step": 5358 }, { "epoch": 0.72, "learning_rate": 2.8411065080849926e-08, "loss": 0.056, "step": 5359 }, { "epoch": 0.72, "learning_rate": 2.8397701456635036e-08, "loss": 0.0648, "step": 5360 }, { "epoch": 0.72, "learning_rate": 2.8384337832420153e-08, "loss": 0.0536, "step": 5361 }, { "epoch": 0.72, "learning_rate": 2.8370974208205263e-08, "loss": 0.0644, "step": 5362 }, { "epoch": 0.72, "learning_rate": 2.8357610583990376e-08, "loss": 0.075, "step": 5363 }, { "epoch": 0.72, "learning_rate": 2.8344246959775492e-08, "loss": 0.0697, "step": 5364 }, { "epoch": 0.72, "learning_rate": 2.8330883335560602e-08, "loss": 0.0495, "step": 5365 }, { "epoch": 0.72, "learning_rate": 2.8317519711345715e-08, "loss": 0.0667, "step": 5366 }, { "epoch": 0.72, "learning_rate": 2.8304156087130825e-08, "loss": 0.0584, "step": 5367 }, { "epoch": 0.72, "learning_rate": 2.829079246291594e-08, "loss": 0.0549, "step": 5368 }, { "epoch": 0.72, "learning_rate": 2.8277428838701058e-08, "loss": 0.0997, "step": 5369 }, { "epoch": 0.72, "learning_rate": 2.8264065214486164e-08, "loss": 0.0534, "step": 5370 }, { "epoch": 0.72, "learning_rate": 2.825070159027128e-08, "loss": 0.0759, "step": 5371 }, { "epoch": 0.72, "learning_rate": 2.8237337966056397e-08, "loss": 0.0731, "step": 5372 }, { "epoch": 0.72, "learning_rate": 2.8223974341841507e-08, "loss": 0.0662, "step": 5373 }, { "epoch": 0.72, "learning_rate": 2.821061071762662e-08, "loss": 0.0535, "step": 5374 }, { "epoch": 0.72, "learning_rate": 2.819724709341173e-08, "loss": 0.0536, "step": 5375 }, { "epoch": 0.72, "learning_rate": 2.8183883469196847e-08, "loss": 0.0694, "step": 5376 }, { "epoch": 0.72, "learning_rate": 2.817051984498196e-08, "loss": 0.0519, "step": 5377 }, { "epoch": 0.72, "learning_rate": 2.815715622076707e-08, "loss": 0.0603, "step": 5378 }, { "epoch": 0.72, "learning_rate": 2.8143792596552186e-08, "loss": 0.0772, "step": 5379 }, { "epoch": 0.72, "learning_rate": 2.8130428972337296e-08, "loss": 0.0478, "step": 5380 }, { "epoch": 0.72, "learning_rate": 2.811706534812241e-08, "loss": 0.0618, "step": 5381 }, { "epoch": 0.72, "learning_rate": 2.8103701723907525e-08, "loss": 0.0586, "step": 5382 }, { "epoch": 0.72, "learning_rate": 2.8090338099692635e-08, "loss": 0.0397, "step": 5383 }, { "epoch": 0.72, "learning_rate": 2.807697447547775e-08, "loss": 0.0652, "step": 5384 }, { "epoch": 0.72, "learning_rate": 2.8063610851262858e-08, "loss": 0.0627, "step": 5385 }, { "epoch": 0.72, "learning_rate": 2.8050247227047975e-08, "loss": 0.066, "step": 5386 }, { "epoch": 0.72, "learning_rate": 2.8036883602833088e-08, "loss": 0.0395, "step": 5387 }, { "epoch": 0.72, "learning_rate": 2.8023519978618198e-08, "loss": 0.0598, "step": 5388 }, { "epoch": 0.72, "learning_rate": 2.8010156354403314e-08, "loss": 0.0543, "step": 5389 }, { "epoch": 0.72, "learning_rate": 2.7996792730188424e-08, "loss": 0.0648, "step": 5390 }, { "epoch": 0.72, "learning_rate": 2.7983429105973537e-08, "loss": 0.0713, "step": 5391 }, { "epoch": 0.72, "learning_rate": 2.7970065481758653e-08, "loss": 0.0453, "step": 5392 }, { "epoch": 0.72, "learning_rate": 2.7956701857543763e-08, "loss": 0.0411, "step": 5393 }, { "epoch": 0.72, "learning_rate": 2.7943338233328876e-08, "loss": 0.065, "step": 5394 }, { "epoch": 0.72, "learning_rate": 2.7929974609113993e-08, "loss": 0.0692, "step": 5395 }, { "epoch": 0.72, "learning_rate": 2.7916610984899103e-08, "loss": 0.0513, "step": 5396 }, { "epoch": 0.72, "learning_rate": 2.790324736068422e-08, "loss": 0.077, "step": 5397 }, { "epoch": 0.72, "learning_rate": 2.7889883736469326e-08, "loss": 0.0955, "step": 5398 }, { "epoch": 0.72, "learning_rate": 2.7876520112254442e-08, "loss": 0.0887, "step": 5399 }, { "epoch": 0.72, "learning_rate": 2.786315648803956e-08, "loss": 0.0668, "step": 5400 }, { "epoch": 0.72, "learning_rate": 2.784979286382467e-08, "loss": 0.0573, "step": 5401 }, { "epoch": 0.72, "learning_rate": 2.783642923960978e-08, "loss": 0.0551, "step": 5402 }, { "epoch": 0.72, "learning_rate": 2.782306561539489e-08, "loss": 0.0592, "step": 5403 }, { "epoch": 0.72, "learning_rate": 2.7809701991180008e-08, "loss": 0.0954, "step": 5404 }, { "epoch": 0.72, "learning_rate": 2.779633836696512e-08, "loss": 0.0499, "step": 5405 }, { "epoch": 0.72, "learning_rate": 2.778297474275023e-08, "loss": 0.0502, "step": 5406 }, { "epoch": 0.72, "learning_rate": 2.7769611118535347e-08, "loss": 0.0795, "step": 5407 }, { "epoch": 0.72, "learning_rate": 2.7756247494320457e-08, "loss": 0.0641, "step": 5408 }, { "epoch": 0.72, "learning_rate": 2.774288387010557e-08, "loss": 0.0775, "step": 5409 }, { "epoch": 0.72, "learning_rate": 2.7729520245890687e-08, "loss": 0.0778, "step": 5410 }, { "epoch": 0.72, "learning_rate": 2.7716156621675797e-08, "loss": 0.061, "step": 5411 }, { "epoch": 0.72, "learning_rate": 2.770279299746091e-08, "loss": 0.0606, "step": 5412 }, { "epoch": 0.72, "learning_rate": 2.768942937324602e-08, "loss": 0.0551, "step": 5413 }, { "epoch": 0.72, "learning_rate": 2.7676065749031136e-08, "loss": 0.0666, "step": 5414 }, { "epoch": 0.72, "learning_rate": 2.766270212481625e-08, "loss": 0.1111, "step": 5415 }, { "epoch": 0.72, "learning_rate": 2.764933850060136e-08, "loss": 0.07, "step": 5416 }, { "epoch": 0.72, "learning_rate": 2.7635974876386475e-08, "loss": 0.0695, "step": 5417 }, { "epoch": 0.72, "learning_rate": 2.7622611252171592e-08, "loss": 0.0612, "step": 5418 }, { "epoch": 0.72, "learning_rate": 2.76092476279567e-08, "loss": 0.0705, "step": 5419 }, { "epoch": 0.72, "learning_rate": 2.7595884003741815e-08, "loss": 0.06, "step": 5420 }, { "epoch": 0.72, "learning_rate": 2.7582520379526925e-08, "loss": 0.0578, "step": 5421 }, { "epoch": 0.72, "learning_rate": 2.756915675531204e-08, "loss": 0.0464, "step": 5422 }, { "epoch": 0.72, "learning_rate": 2.7555793131097154e-08, "loss": 0.0604, "step": 5423 }, { "epoch": 0.72, "learning_rate": 2.7542429506882264e-08, "loss": 0.0587, "step": 5424 }, { "epoch": 0.72, "learning_rate": 2.752906588266738e-08, "loss": 0.0808, "step": 5425 }, { "epoch": 0.73, "learning_rate": 2.751570225845249e-08, "loss": 0.0654, "step": 5426 }, { "epoch": 0.73, "learning_rate": 2.7502338634237604e-08, "loss": 0.0548, "step": 5427 }, { "epoch": 0.73, "learning_rate": 2.748897501002272e-08, "loss": 0.0533, "step": 5428 }, { "epoch": 0.73, "learning_rate": 2.747561138580783e-08, "loss": 0.0764, "step": 5429 }, { "epoch": 0.73, "learning_rate": 2.7462247761592943e-08, "loss": 0.0653, "step": 5430 }, { "epoch": 0.73, "learning_rate": 2.7448884137378053e-08, "loss": 0.0752, "step": 5431 }, { "epoch": 0.73, "learning_rate": 2.743552051316317e-08, "loss": 0.094, "step": 5432 }, { "epoch": 0.73, "learning_rate": 2.7422156888948282e-08, "loss": 0.051, "step": 5433 }, { "epoch": 0.73, "learning_rate": 2.7408793264733392e-08, "loss": 0.0665, "step": 5434 }, { "epoch": 0.73, "learning_rate": 2.739542964051851e-08, "loss": 0.0629, "step": 5435 }, { "epoch": 0.73, "learning_rate": 2.738206601630362e-08, "loss": 0.0835, "step": 5436 }, { "epoch": 0.73, "learning_rate": 2.7368702392088732e-08, "loss": 0.0731, "step": 5437 }, { "epoch": 0.73, "learning_rate": 2.7355338767873848e-08, "loss": 0.077, "step": 5438 }, { "epoch": 0.73, "learning_rate": 2.7341975143658958e-08, "loss": 0.086, "step": 5439 }, { "epoch": 0.73, "learning_rate": 2.732861151944407e-08, "loss": 0.0442, "step": 5440 }, { "epoch": 0.73, "learning_rate": 2.7315247895229188e-08, "loss": 0.0704, "step": 5441 }, { "epoch": 0.73, "learning_rate": 2.7301884271014297e-08, "loss": 0.0791, "step": 5442 }, { "epoch": 0.73, "learning_rate": 2.7288520646799414e-08, "loss": 0.0598, "step": 5443 }, { "epoch": 0.73, "learning_rate": 2.727515702258452e-08, "loss": 0.0847, "step": 5444 }, { "epoch": 0.73, "learning_rate": 2.7261793398369637e-08, "loss": 0.0571, "step": 5445 }, { "epoch": 0.73, "learning_rate": 2.7248429774154753e-08, "loss": 0.0692, "step": 5446 }, { "epoch": 0.73, "learning_rate": 2.7235066149939863e-08, "loss": 0.0613, "step": 5447 }, { "epoch": 0.73, "learning_rate": 2.7221702525724976e-08, "loss": 0.0968, "step": 5448 }, { "epoch": 0.73, "learning_rate": 2.7208338901510086e-08, "loss": 0.0519, "step": 5449 }, { "epoch": 0.73, "learning_rate": 2.7194975277295203e-08, "loss": 0.0825, "step": 5450 }, { "epoch": 0.73, "learning_rate": 2.7181611653080316e-08, "loss": 0.0868, "step": 5451 }, { "epoch": 0.73, "learning_rate": 2.7168248028865425e-08, "loss": 0.0496, "step": 5452 }, { "epoch": 0.73, "learning_rate": 2.7154884404650542e-08, "loss": 0.0648, "step": 5453 }, { "epoch": 0.73, "learning_rate": 2.7141520780435652e-08, "loss": 0.0585, "step": 5454 }, { "epoch": 0.73, "learning_rate": 2.7128157156220765e-08, "loss": 0.0546, "step": 5455 }, { "epoch": 0.73, "learning_rate": 2.711479353200588e-08, "loss": 0.0619, "step": 5456 }, { "epoch": 0.73, "learning_rate": 2.710142990779099e-08, "loss": 0.0816, "step": 5457 }, { "epoch": 0.73, "learning_rate": 2.7088066283576104e-08, "loss": 0.0804, "step": 5458 }, { "epoch": 0.73, "learning_rate": 2.7074702659361214e-08, "loss": 0.0457, "step": 5459 }, { "epoch": 0.73, "learning_rate": 2.706133903514633e-08, "loss": 0.1284, "step": 5460 }, { "epoch": 0.73, "learning_rate": 2.7047975410931444e-08, "loss": 0.0579, "step": 5461 }, { "epoch": 0.73, "learning_rate": 2.7034611786716554e-08, "loss": 0.0559, "step": 5462 }, { "epoch": 0.73, "learning_rate": 2.702124816250167e-08, "loss": 0.0543, "step": 5463 }, { "epoch": 0.73, "learning_rate": 2.7007884538286783e-08, "loss": 0.0742, "step": 5464 }, { "epoch": 0.73, "learning_rate": 2.6994520914071893e-08, "loss": 0.0651, "step": 5465 }, { "epoch": 0.73, "learning_rate": 2.698115728985701e-08, "loss": 0.0967, "step": 5466 }, { "epoch": 0.73, "learning_rate": 2.696779366564212e-08, "loss": 0.0717, "step": 5467 }, { "epoch": 0.73, "learning_rate": 2.6954430041427232e-08, "loss": 0.0714, "step": 5468 }, { "epoch": 0.73, "learning_rate": 2.694106641721235e-08, "loss": 0.0451, "step": 5469 }, { "epoch": 0.73, "learning_rate": 2.692770279299746e-08, "loss": 0.0671, "step": 5470 }, { "epoch": 0.73, "learning_rate": 2.6914339168782575e-08, "loss": 0.0622, "step": 5471 }, { "epoch": 0.73, "learning_rate": 2.6900975544567685e-08, "loss": 0.0476, "step": 5472 }, { "epoch": 0.73, "learning_rate": 2.6887611920352798e-08, "loss": 0.0458, "step": 5473 }, { "epoch": 0.73, "learning_rate": 2.6874248296137915e-08, "loss": 0.0673, "step": 5474 }, { "epoch": 0.73, "learning_rate": 2.6860884671923024e-08, "loss": 0.0584, "step": 5475 }, { "epoch": 0.73, "learning_rate": 2.6847521047708138e-08, "loss": 0.0714, "step": 5476 }, { "epoch": 0.73, "learning_rate": 2.6834157423493247e-08, "loss": 0.0465, "step": 5477 }, { "epoch": 0.73, "learning_rate": 2.6820793799278364e-08, "loss": 0.0501, "step": 5478 }, { "epoch": 0.73, "learning_rate": 2.6807430175063477e-08, "loss": 0.0732, "step": 5479 }, { "epoch": 0.73, "learning_rate": 2.6794066550848587e-08, "loss": 0.0608, "step": 5480 }, { "epoch": 0.73, "learning_rate": 2.6780702926633703e-08, "loss": 0.0697, "step": 5481 }, { "epoch": 0.73, "learning_rate": 2.6767339302418813e-08, "loss": 0.0588, "step": 5482 }, { "epoch": 0.73, "learning_rate": 2.6753975678203926e-08, "loss": 0.0851, "step": 5483 }, { "epoch": 0.73, "learning_rate": 2.6740612053989043e-08, "loss": 0.0814, "step": 5484 }, { "epoch": 0.73, "learning_rate": 2.6727248429774153e-08, "loss": 0.0375, "step": 5485 }, { "epoch": 0.73, "learning_rate": 2.6713884805559266e-08, "loss": 0.0506, "step": 5486 }, { "epoch": 0.73, "learning_rate": 2.6700521181344382e-08, "loss": 0.0573, "step": 5487 }, { "epoch": 0.73, "learning_rate": 2.6687157557129492e-08, "loss": 0.0701, "step": 5488 }, { "epoch": 0.73, "learning_rate": 2.6673793932914605e-08, "loss": 0.0483, "step": 5489 }, { "epoch": 0.73, "learning_rate": 2.6660430308699715e-08, "loss": 0.053, "step": 5490 }, { "epoch": 0.73, "learning_rate": 2.664706668448483e-08, "loss": 0.0676, "step": 5491 }, { "epoch": 0.73, "learning_rate": 2.6633703060269948e-08, "loss": 0.0648, "step": 5492 }, { "epoch": 0.73, "learning_rate": 2.6620339436055054e-08, "loss": 0.0502, "step": 5493 }, { "epoch": 0.73, "learning_rate": 2.660697581184017e-08, "loss": 0.0633, "step": 5494 }, { "epoch": 0.73, "learning_rate": 2.659361218762528e-08, "loss": 0.0651, "step": 5495 }, { "epoch": 0.73, "learning_rate": 2.6580248563410397e-08, "loss": 0.0753, "step": 5496 }, { "epoch": 0.73, "learning_rate": 2.656688493919551e-08, "loss": 0.0837, "step": 5497 }, { "epoch": 0.73, "learning_rate": 2.655352131498062e-08, "loss": 0.0576, "step": 5498 }, { "epoch": 0.73, "learning_rate": 2.6540157690765737e-08, "loss": 0.0777, "step": 5499 }, { "epoch": 0.73, "learning_rate": 2.6526794066550846e-08, "loss": 0.0662, "step": 5500 }, { "epoch": 0.74, "learning_rate": 2.651343044233596e-08, "loss": 0.076, "step": 5501 }, { "epoch": 0.74, "learning_rate": 2.6500066818121076e-08, "loss": 0.0607, "step": 5502 }, { "epoch": 0.74, "learning_rate": 2.6486703193906186e-08, "loss": 0.042, "step": 5503 }, { "epoch": 0.74, "learning_rate": 2.64733395696913e-08, "loss": 0.0821, "step": 5504 }, { "epoch": 0.74, "learning_rate": 2.645997594547641e-08, "loss": 0.0387, "step": 5505 }, { "epoch": 0.74, "learning_rate": 2.6446612321261525e-08, "loss": 0.0753, "step": 5506 }, { "epoch": 0.74, "learning_rate": 2.643324869704664e-08, "loss": 0.0991, "step": 5507 }, { "epoch": 0.74, "learning_rate": 2.6419885072831748e-08, "loss": 0.0411, "step": 5508 }, { "epoch": 0.74, "learning_rate": 2.6406521448616865e-08, "loss": 0.0633, "step": 5509 }, { "epoch": 0.74, "learning_rate": 2.6393157824401978e-08, "loss": 0.0529, "step": 5510 }, { "epoch": 0.74, "learning_rate": 2.6379794200187088e-08, "loss": 0.0811, "step": 5511 }, { "epoch": 0.74, "learning_rate": 2.6366430575972204e-08, "loss": 0.0538, "step": 5512 }, { "epoch": 0.74, "learning_rate": 2.6353066951757314e-08, "loss": 0.0925, "step": 5513 }, { "epoch": 0.74, "learning_rate": 2.6339703327542427e-08, "loss": 0.0701, "step": 5514 }, { "epoch": 0.74, "learning_rate": 2.6326339703327544e-08, "loss": 0.0822, "step": 5515 }, { "epoch": 0.74, "learning_rate": 2.6312976079112653e-08, "loss": 0.0846, "step": 5516 }, { "epoch": 0.74, "learning_rate": 2.629961245489777e-08, "loss": 0.0478, "step": 5517 }, { "epoch": 0.74, "learning_rate": 2.6286248830682876e-08, "loss": 0.0504, "step": 5518 }, { "epoch": 0.74, "learning_rate": 2.6272885206467993e-08, "loss": 0.0467, "step": 5519 }, { "epoch": 0.74, "learning_rate": 2.625952158225311e-08, "loss": 0.0584, "step": 5520 }, { "epoch": 0.74, "learning_rate": 2.624615795803822e-08, "loss": 0.0623, "step": 5521 }, { "epoch": 0.74, "learning_rate": 2.6232794333823332e-08, "loss": 0.0603, "step": 5522 }, { "epoch": 0.74, "learning_rate": 2.6219430709608442e-08, "loss": 0.0666, "step": 5523 }, { "epoch": 0.74, "learning_rate": 2.620606708539356e-08, "loss": 0.0524, "step": 5524 }, { "epoch": 0.74, "learning_rate": 2.619270346117867e-08, "loss": 0.0663, "step": 5525 }, { "epoch": 0.74, "learning_rate": 2.617933983696378e-08, "loss": 0.0622, "step": 5526 }, { "epoch": 0.74, "learning_rate": 2.6165976212748898e-08, "loss": 0.0556, "step": 5527 }, { "epoch": 0.74, "learning_rate": 2.6152612588534008e-08, "loss": 0.0712, "step": 5528 }, { "epoch": 0.74, "learning_rate": 2.613924896431912e-08, "loss": 0.0896, "step": 5529 }, { "epoch": 0.74, "learning_rate": 2.6125885340104237e-08, "loss": 0.0568, "step": 5530 }, { "epoch": 0.74, "learning_rate": 2.6112521715889347e-08, "loss": 0.0882, "step": 5531 }, { "epoch": 0.74, "learning_rate": 2.609915809167446e-08, "loss": 0.042, "step": 5532 }, { "epoch": 0.74, "learning_rate": 2.6085794467459577e-08, "loss": 0.1014, "step": 5533 }, { "epoch": 0.74, "learning_rate": 2.6072430843244687e-08, "loss": 0.095, "step": 5534 }, { "epoch": 0.74, "learning_rate": 2.60590672190298e-08, "loss": 0.0716, "step": 5535 }, { "epoch": 0.74, "learning_rate": 2.604570359481491e-08, "loss": 0.0682, "step": 5536 }, { "epoch": 0.74, "learning_rate": 2.6032339970600026e-08, "loss": 0.0943, "step": 5537 }, { "epoch": 0.74, "learning_rate": 2.6018976346385142e-08, "loss": 0.0439, "step": 5538 }, { "epoch": 0.74, "learning_rate": 2.600561272217025e-08, "loss": 0.0685, "step": 5539 }, { "epoch": 0.74, "learning_rate": 2.5992249097955365e-08, "loss": 0.0426, "step": 5540 }, { "epoch": 0.74, "learning_rate": 2.5978885473740475e-08, "loss": 0.0748, "step": 5541 }, { "epoch": 0.74, "learning_rate": 2.5965521849525592e-08, "loss": 0.06, "step": 5542 }, { "epoch": 0.74, "learning_rate": 2.5952158225310705e-08, "loss": 0.075, "step": 5543 }, { "epoch": 0.74, "learning_rate": 2.5938794601095815e-08, "loss": 0.0869, "step": 5544 }, { "epoch": 0.74, "learning_rate": 2.592543097688093e-08, "loss": 0.0642, "step": 5545 }, { "epoch": 0.74, "learning_rate": 2.591206735266604e-08, "loss": 0.0857, "step": 5546 }, { "epoch": 0.74, "learning_rate": 2.5898703728451154e-08, "loss": 0.1051, "step": 5547 }, { "epoch": 0.74, "learning_rate": 2.588534010423627e-08, "loss": 0.0409, "step": 5548 }, { "epoch": 0.74, "learning_rate": 2.587197648002138e-08, "loss": 0.0613, "step": 5549 }, { "epoch": 0.74, "learning_rate": 2.5858612855806494e-08, "loss": 0.0586, "step": 5550 }, { "epoch": 0.74, "learning_rate": 2.5845249231591603e-08, "loss": 0.076, "step": 5551 }, { "epoch": 0.74, "learning_rate": 2.583188560737672e-08, "loss": 0.0675, "step": 5552 }, { "epoch": 0.74, "learning_rate": 2.5818521983161833e-08, "loss": 0.0843, "step": 5553 }, { "epoch": 0.74, "learning_rate": 2.5805158358946943e-08, "loss": 0.0939, "step": 5554 }, { "epoch": 0.74, "learning_rate": 2.579179473473206e-08, "loss": 0.0567, "step": 5555 }, { "epoch": 0.74, "learning_rate": 2.5778431110517172e-08, "loss": 0.0588, "step": 5556 }, { "epoch": 0.74, "learning_rate": 2.5765067486302282e-08, "loss": 0.1259, "step": 5557 }, { "epoch": 0.74, "learning_rate": 2.57517038620874e-08, "loss": 0.0452, "step": 5558 }, { "epoch": 0.74, "learning_rate": 2.573834023787251e-08, "loss": 0.0611, "step": 5559 }, { "epoch": 0.74, "learning_rate": 2.5724976613657622e-08, "loss": 0.0855, "step": 5560 }, { "epoch": 0.74, "learning_rate": 2.5711612989442738e-08, "loss": 0.0514, "step": 5561 }, { "epoch": 0.74, "learning_rate": 2.5698249365227848e-08, "loss": 0.0832, "step": 5562 }, { "epoch": 0.74, "learning_rate": 2.568488574101296e-08, "loss": 0.0724, "step": 5563 }, { "epoch": 0.74, "learning_rate": 2.567152211679807e-08, "loss": 0.0659, "step": 5564 }, { "epoch": 0.74, "learning_rate": 2.5658158492583187e-08, "loss": 0.072, "step": 5565 }, { "epoch": 0.74, "learning_rate": 2.5644794868368304e-08, "loss": 0.1058, "step": 5566 }, { "epoch": 0.74, "learning_rate": 2.563143124415341e-08, "loss": 0.0832, "step": 5567 }, { "epoch": 0.74, "learning_rate": 2.5618067619938527e-08, "loss": 0.0623, "step": 5568 }, { "epoch": 0.74, "learning_rate": 2.5604703995723637e-08, "loss": 0.0677, "step": 5569 }, { "epoch": 0.74, "learning_rate": 2.5591340371508753e-08, "loss": 0.0811, "step": 5570 }, { "epoch": 0.74, "learning_rate": 2.5577976747293866e-08, "loss": 0.0519, "step": 5571 }, { "epoch": 0.74, "learning_rate": 2.5564613123078976e-08, "loss": 0.0564, "step": 5572 }, { "epoch": 0.74, "learning_rate": 2.5551249498864093e-08, "loss": 0.0638, "step": 5573 }, { "epoch": 0.74, "learning_rate": 2.5537885874649202e-08, "loss": 0.0567, "step": 5574 }, { "epoch": 0.75, "learning_rate": 2.5524522250434315e-08, "loss": 0.0391, "step": 5575 }, { "epoch": 0.75, "learning_rate": 2.5511158626219432e-08, "loss": 0.0527, "step": 5576 }, { "epoch": 0.75, "learning_rate": 2.5497795002004542e-08, "loss": 0.0792, "step": 5577 }, { "epoch": 0.75, "learning_rate": 2.5484431377789655e-08, "loss": 0.0636, "step": 5578 }, { "epoch": 0.75, "learning_rate": 2.547106775357477e-08, "loss": 0.0663, "step": 5579 }, { "epoch": 0.75, "learning_rate": 2.545770412935988e-08, "loss": 0.0449, "step": 5580 }, { "epoch": 0.75, "learning_rate": 2.5444340505144994e-08, "loss": 0.0729, "step": 5581 }, { "epoch": 0.75, "learning_rate": 2.5430976880930104e-08, "loss": 0.0643, "step": 5582 }, { "epoch": 0.75, "learning_rate": 2.541761325671522e-08, "loss": 0.0606, "step": 5583 }, { "epoch": 0.75, "learning_rate": 2.5404249632500334e-08, "loss": 0.0525, "step": 5584 }, { "epoch": 0.75, "learning_rate": 2.5390886008285444e-08, "loss": 0.0548, "step": 5585 }, { "epoch": 0.75, "learning_rate": 2.537752238407056e-08, "loss": 0.0439, "step": 5586 }, { "epoch": 0.75, "learning_rate": 2.536415875985567e-08, "loss": 0.0503, "step": 5587 }, { "epoch": 0.75, "learning_rate": 2.5350795135640783e-08, "loss": 0.0472, "step": 5588 }, { "epoch": 0.75, "learning_rate": 2.53374315114259e-08, "loss": 0.0698, "step": 5589 }, { "epoch": 0.75, "learning_rate": 2.532406788721101e-08, "loss": 0.0709, "step": 5590 }, { "epoch": 0.75, "learning_rate": 2.5310704262996126e-08, "loss": 0.0709, "step": 5591 }, { "epoch": 0.75, "learning_rate": 2.5297340638781232e-08, "loss": 0.0633, "step": 5592 }, { "epoch": 0.75, "learning_rate": 2.528397701456635e-08, "loss": 0.048, "step": 5593 }, { "epoch": 0.75, "learning_rate": 2.5270613390351465e-08, "loss": 0.063, "step": 5594 }, { "epoch": 0.75, "learning_rate": 2.5257249766136575e-08, "loss": 0.0547, "step": 5595 }, { "epoch": 0.75, "learning_rate": 2.5243886141921688e-08, "loss": 0.0644, "step": 5596 }, { "epoch": 0.75, "learning_rate": 2.5230522517706798e-08, "loss": 0.0352, "step": 5597 }, { "epoch": 0.75, "learning_rate": 2.5217158893491914e-08, "loss": 0.0506, "step": 5598 }, { "epoch": 0.75, "learning_rate": 2.5203795269277028e-08, "loss": 0.0589, "step": 5599 }, { "epoch": 0.75, "learning_rate": 2.5190431645062137e-08, "loss": 0.0441, "step": 5600 }, { "epoch": 0.75, "learning_rate": 2.5177068020847254e-08, "loss": 0.0899, "step": 5601 }, { "epoch": 0.75, "learning_rate": 2.5163704396632367e-08, "loss": 0.0399, "step": 5602 }, { "epoch": 0.75, "learning_rate": 2.5150340772417477e-08, "loss": 0.0802, "step": 5603 }, { "epoch": 0.75, "learning_rate": 2.5136977148202593e-08, "loss": 0.0745, "step": 5604 }, { "epoch": 0.75, "learning_rate": 2.5123613523987703e-08, "loss": 0.0473, "step": 5605 }, { "epoch": 0.75, "learning_rate": 2.5110249899772816e-08, "loss": 0.0634, "step": 5606 }, { "epoch": 0.75, "learning_rate": 2.5096886275557933e-08, "loss": 0.088, "step": 5607 }, { "epoch": 0.75, "learning_rate": 2.5083522651343043e-08, "loss": 0.047, "step": 5608 }, { "epoch": 0.75, "learning_rate": 2.5070159027128156e-08, "loss": 0.0608, "step": 5609 }, { "epoch": 0.75, "learning_rate": 2.5056795402913266e-08, "loss": 0.045, "step": 5610 }, { "epoch": 0.75, "learning_rate": 2.5043431778698382e-08, "loss": 0.0622, "step": 5611 }, { "epoch": 0.75, "learning_rate": 2.50300681544835e-08, "loss": 0.0691, "step": 5612 }, { "epoch": 0.75, "learning_rate": 2.5016704530268605e-08, "loss": 0.0642, "step": 5613 }, { "epoch": 0.75, "learning_rate": 2.500334090605372e-08, "loss": 0.0658, "step": 5614 }, { "epoch": 0.75, "learning_rate": 2.4989977281838835e-08, "loss": 0.0827, "step": 5615 }, { "epoch": 0.75, "learning_rate": 2.4976613657623948e-08, "loss": 0.0694, "step": 5616 }, { "epoch": 0.75, "learning_rate": 2.4963250033409058e-08, "loss": 0.0727, "step": 5617 }, { "epoch": 0.75, "learning_rate": 2.4949886409194174e-08, "loss": 0.0451, "step": 5618 }, { "epoch": 0.75, "learning_rate": 2.4936522784979287e-08, "loss": 0.0591, "step": 5619 }, { "epoch": 0.75, "learning_rate": 2.4923159160764397e-08, "loss": 0.0637, "step": 5620 }, { "epoch": 0.75, "learning_rate": 2.490979553654951e-08, "loss": 0.0698, "step": 5621 }, { "epoch": 0.75, "learning_rate": 2.4896431912334623e-08, "loss": 0.0655, "step": 5622 }, { "epoch": 0.75, "learning_rate": 2.4883068288119736e-08, "loss": 0.1167, "step": 5623 }, { "epoch": 0.75, "learning_rate": 2.486970466390485e-08, "loss": 0.1066, "step": 5624 }, { "epoch": 0.75, "learning_rate": 2.4856341039689963e-08, "loss": 0.0591, "step": 5625 }, { "epoch": 0.75, "learning_rate": 2.4842977415475076e-08, "loss": 0.0684, "step": 5626 }, { "epoch": 0.75, "learning_rate": 2.482961379126019e-08, "loss": 0.0605, "step": 5627 }, { "epoch": 0.75, "learning_rate": 2.4816250167045302e-08, "loss": 0.0776, "step": 5628 }, { "epoch": 0.75, "learning_rate": 2.4802886542830415e-08, "loss": 0.0539, "step": 5629 }, { "epoch": 0.75, "learning_rate": 2.478952291861553e-08, "loss": 0.0603, "step": 5630 }, { "epoch": 0.75, "learning_rate": 2.4776159294400638e-08, "loss": 0.0334, "step": 5631 }, { "epoch": 0.75, "learning_rate": 2.4762795670185755e-08, "loss": 0.0585, "step": 5632 }, { "epoch": 0.75, "learning_rate": 2.4749432045970868e-08, "loss": 0.0568, "step": 5633 }, { "epoch": 0.75, "learning_rate": 2.4736068421755978e-08, "loss": 0.079, "step": 5634 }, { "epoch": 0.75, "learning_rate": 2.472270479754109e-08, "loss": 0.0493, "step": 5635 }, { "epoch": 0.75, "learning_rate": 2.4709341173326204e-08, "loss": 0.076, "step": 5636 }, { "epoch": 0.75, "learning_rate": 2.4695977549111317e-08, "loss": 0.075, "step": 5637 }, { "epoch": 0.75, "learning_rate": 2.468261392489643e-08, "loss": 0.0562, "step": 5638 }, { "epoch": 0.75, "learning_rate": 2.4669250300681543e-08, "loss": 0.0738, "step": 5639 }, { "epoch": 0.75, "learning_rate": 2.4655886676466656e-08, "loss": 0.0623, "step": 5640 }, { "epoch": 0.75, "learning_rate": 2.464252305225177e-08, "loss": 0.0369, "step": 5641 }, { "epoch": 0.75, "learning_rate": 2.4629159428036883e-08, "loss": 0.0946, "step": 5642 }, { "epoch": 0.75, "learning_rate": 2.4615795803821996e-08, "loss": 0.0998, "step": 5643 }, { "epoch": 0.75, "learning_rate": 2.460243217960711e-08, "loss": 0.0814, "step": 5644 }, { "epoch": 0.75, "learning_rate": 2.458906855539222e-08, "loss": 0.0746, "step": 5645 }, { "epoch": 0.75, "learning_rate": 2.4575704931177335e-08, "loss": 0.0625, "step": 5646 }, { "epoch": 0.75, "learning_rate": 2.456234130696245e-08, "loss": 0.0491, "step": 5647 }, { "epoch": 0.75, "learning_rate": 2.4548977682747558e-08, "loss": 0.0669, "step": 5648 }, { "epoch": 0.75, "learning_rate": 2.453561405853267e-08, "loss": 0.0773, "step": 5649 }, { "epoch": 0.76, "learning_rate": 2.4522250434317788e-08, "loss": 0.061, "step": 5650 }, { "epoch": 0.76, "learning_rate": 2.45088868101029e-08, "loss": 0.0734, "step": 5651 }, { "epoch": 0.76, "learning_rate": 2.449552318588801e-08, "loss": 0.0441, "step": 5652 }, { "epoch": 0.76, "learning_rate": 2.4482159561673124e-08, "loss": 0.0648, "step": 5653 }, { "epoch": 0.76, "learning_rate": 2.4468795937458237e-08, "loss": 0.1072, "step": 5654 }, { "epoch": 0.76, "learning_rate": 2.445543231324335e-08, "loss": 0.0631, "step": 5655 }, { "epoch": 0.76, "learning_rate": 2.4442068689028463e-08, "loss": 0.0989, "step": 5656 }, { "epoch": 0.76, "learning_rate": 2.4428705064813577e-08, "loss": 0.0901, "step": 5657 }, { "epoch": 0.76, "learning_rate": 2.441534144059869e-08, "loss": 0.0794, "step": 5658 }, { "epoch": 0.76, "learning_rate": 2.44019778163838e-08, "loss": 0.0793, "step": 5659 }, { "epoch": 0.76, "learning_rate": 2.4388614192168916e-08, "loss": 0.0437, "step": 5660 }, { "epoch": 0.76, "learning_rate": 2.437525056795403e-08, "loss": 0.0683, "step": 5661 }, { "epoch": 0.76, "learning_rate": 2.436188694373914e-08, "loss": 0.0595, "step": 5662 }, { "epoch": 0.76, "learning_rate": 2.4348523319524252e-08, "loss": 0.0572, "step": 5663 }, { "epoch": 0.76, "learning_rate": 2.433515969530937e-08, "loss": 0.0535, "step": 5664 }, { "epoch": 0.76, "learning_rate": 2.4321796071094482e-08, "loss": 0.1018, "step": 5665 }, { "epoch": 0.76, "learning_rate": 2.430843244687959e-08, "loss": 0.0854, "step": 5666 }, { "epoch": 0.76, "learning_rate": 2.4295068822664705e-08, "loss": 0.0633, "step": 5667 }, { "epoch": 0.76, "learning_rate": 2.4281705198449818e-08, "loss": 0.055, "step": 5668 }, { "epoch": 0.76, "learning_rate": 2.426834157423493e-08, "loss": 0.124, "step": 5669 }, { "epoch": 0.76, "learning_rate": 2.4254977950020044e-08, "loss": 0.0454, "step": 5670 }, { "epoch": 0.76, "learning_rate": 2.4241614325805157e-08, "loss": 0.0828, "step": 5671 }, { "epoch": 0.76, "learning_rate": 2.422825070159027e-08, "loss": 0.0801, "step": 5672 }, { "epoch": 0.76, "learning_rate": 2.4214887077375384e-08, "loss": 0.0647, "step": 5673 }, { "epoch": 0.76, "learning_rate": 2.4201523453160497e-08, "loss": 0.0345, "step": 5674 }, { "epoch": 0.76, "learning_rate": 2.418815982894561e-08, "loss": 0.0604, "step": 5675 }, { "epoch": 0.76, "learning_rate": 2.4174796204730723e-08, "loss": 0.0496, "step": 5676 }, { "epoch": 0.76, "learning_rate": 2.4161432580515833e-08, "loss": 0.0592, "step": 5677 }, { "epoch": 0.76, "learning_rate": 2.414806895630095e-08, "loss": 0.0774, "step": 5678 }, { "epoch": 0.76, "learning_rate": 2.4134705332086062e-08, "loss": 0.0588, "step": 5679 }, { "epoch": 0.76, "learning_rate": 2.4121341707871172e-08, "loss": 0.0677, "step": 5680 }, { "epoch": 0.76, "learning_rate": 2.4107978083656285e-08, "loss": 0.0599, "step": 5681 }, { "epoch": 0.76, "learning_rate": 2.40946144594414e-08, "loss": 0.0583, "step": 5682 }, { "epoch": 0.76, "learning_rate": 2.4081250835226512e-08, "loss": 0.041, "step": 5683 }, { "epoch": 0.76, "learning_rate": 2.4067887211011625e-08, "loss": 0.0548, "step": 5684 }, { "epoch": 0.76, "learning_rate": 2.4054523586796738e-08, "loss": 0.0567, "step": 5685 }, { "epoch": 0.76, "learning_rate": 2.404115996258185e-08, "loss": 0.064, "step": 5686 }, { "epoch": 0.76, "learning_rate": 2.4027796338366964e-08, "loss": 0.0545, "step": 5687 }, { "epoch": 0.76, "learning_rate": 2.4014432714152077e-08, "loss": 0.0839, "step": 5688 }, { "epoch": 0.76, "learning_rate": 2.400106908993719e-08, "loss": 0.0476, "step": 5689 }, { "epoch": 0.76, "learning_rate": 2.3987705465722304e-08, "loss": 0.0552, "step": 5690 }, { "epoch": 0.76, "learning_rate": 2.3974341841507413e-08, "loss": 0.0891, "step": 5691 }, { "epoch": 0.76, "learning_rate": 2.396097821729253e-08, "loss": 0.0289, "step": 5692 }, { "epoch": 0.76, "learning_rate": 2.3947614593077643e-08, "loss": 0.0714, "step": 5693 }, { "epoch": 0.76, "learning_rate": 2.3934250968862753e-08, "loss": 0.0538, "step": 5694 }, { "epoch": 0.76, "learning_rate": 2.3920887344647866e-08, "loss": 0.0774, "step": 5695 }, { "epoch": 0.76, "learning_rate": 2.3907523720432983e-08, "loss": 0.0495, "step": 5696 }, { "epoch": 0.76, "learning_rate": 2.3894160096218092e-08, "loss": 0.0732, "step": 5697 }, { "epoch": 0.76, "learning_rate": 2.3880796472003205e-08, "loss": 0.0845, "step": 5698 }, { "epoch": 0.76, "learning_rate": 2.386743284778832e-08, "loss": 0.0809, "step": 5699 }, { "epoch": 0.76, "learning_rate": 2.3854069223573432e-08, "loss": 0.0698, "step": 5700 }, { "epoch": 0.76, "learning_rate": 2.3840705599358545e-08, "loss": 0.0657, "step": 5701 }, { "epoch": 0.76, "learning_rate": 2.3827341975143658e-08, "loss": 0.0687, "step": 5702 }, { "epoch": 0.76, "learning_rate": 2.381397835092877e-08, "loss": 0.0571, "step": 5703 }, { "epoch": 0.76, "learning_rate": 2.3800614726713884e-08, "loss": 0.0863, "step": 5704 }, { "epoch": 0.76, "learning_rate": 2.3787251102498997e-08, "loss": 0.0976, "step": 5705 }, { "epoch": 0.76, "learning_rate": 2.377388747828411e-08, "loss": 0.0411, "step": 5706 }, { "epoch": 0.76, "learning_rate": 2.3760523854069224e-08, "loss": 0.0865, "step": 5707 }, { "epoch": 0.76, "learning_rate": 2.3747160229854334e-08, "loss": 0.0423, "step": 5708 }, { "epoch": 0.76, "learning_rate": 2.3733796605639447e-08, "loss": 0.0598, "step": 5709 }, { "epoch": 0.76, "learning_rate": 2.3720432981424563e-08, "loss": 0.0461, "step": 5710 }, { "epoch": 0.76, "learning_rate": 2.3707069357209676e-08, "loss": 0.0429, "step": 5711 }, { "epoch": 0.76, "learning_rate": 2.3693705732994786e-08, "loss": 0.0542, "step": 5712 }, { "epoch": 0.76, "learning_rate": 2.36803421087799e-08, "loss": 0.0499, "step": 5713 }, { "epoch": 0.76, "learning_rate": 2.3666978484565012e-08, "loss": 0.0786, "step": 5714 }, { "epoch": 0.76, "learning_rate": 2.3653614860350126e-08, "loss": 0.0661, "step": 5715 }, { "epoch": 0.76, "learning_rate": 2.364025123613524e-08, "loss": 0.0575, "step": 5716 }, { "epoch": 0.76, "learning_rate": 2.3626887611920352e-08, "loss": 0.0656, "step": 5717 }, { "epoch": 0.76, "learning_rate": 2.3613523987705465e-08, "loss": 0.0587, "step": 5718 }, { "epoch": 0.76, "learning_rate": 2.3600160363490578e-08, "loss": 0.0837, "step": 5719 }, { "epoch": 0.76, "learning_rate": 2.358679673927569e-08, "loss": 0.0712, "step": 5720 }, { "epoch": 0.76, "learning_rate": 2.3573433115060804e-08, "loss": 0.0807, "step": 5721 }, { "epoch": 0.76, "learning_rate": 2.3560069490845914e-08, "loss": 0.0549, "step": 5722 }, { "epoch": 0.76, "learning_rate": 2.3546705866631027e-08, "loss": 0.0663, "step": 5723 }, { "epoch": 0.76, "learning_rate": 2.3533342242416144e-08, "loss": 0.0644, "step": 5724 }, { "epoch": 0.77, "learning_rate": 2.3519978618201257e-08, "loss": 0.0684, "step": 5725 }, { "epoch": 0.77, "learning_rate": 2.3506614993986367e-08, "loss": 0.0369, "step": 5726 }, { "epoch": 0.77, "learning_rate": 2.349325136977148e-08, "loss": 0.0452, "step": 5727 }, { "epoch": 0.77, "learning_rate": 2.3479887745556596e-08, "loss": 0.0823, "step": 5728 }, { "epoch": 0.77, "learning_rate": 2.3466524121341706e-08, "loss": 0.0756, "step": 5729 }, { "epoch": 0.77, "learning_rate": 2.345316049712682e-08, "loss": 0.0692, "step": 5730 }, { "epoch": 0.77, "learning_rate": 2.3439796872911933e-08, "loss": 0.0528, "step": 5731 }, { "epoch": 0.77, "learning_rate": 2.3426433248697046e-08, "loss": 0.0536, "step": 5732 }, { "epoch": 0.77, "learning_rate": 2.341306962448216e-08, "loss": 0.0508, "step": 5733 }, { "epoch": 0.77, "learning_rate": 2.3399706000267272e-08, "loss": 0.0585, "step": 5734 }, { "epoch": 0.77, "learning_rate": 2.3386342376052385e-08, "loss": 0.0588, "step": 5735 }, { "epoch": 0.77, "learning_rate": 2.3372978751837495e-08, "loss": 0.0521, "step": 5736 }, { "epoch": 0.77, "learning_rate": 2.3359615127622608e-08, "loss": 0.041, "step": 5737 }, { "epoch": 0.77, "learning_rate": 2.3346251503407725e-08, "loss": 0.0853, "step": 5738 }, { "epoch": 0.77, "learning_rate": 2.3332887879192838e-08, "loss": 0.0669, "step": 5739 }, { "epoch": 0.77, "learning_rate": 2.3319524254977948e-08, "loss": 0.0716, "step": 5740 }, { "epoch": 0.77, "learning_rate": 2.330616063076306e-08, "loss": 0.0544, "step": 5741 }, { "epoch": 0.77, "learning_rate": 2.3292797006548177e-08, "loss": 0.0703, "step": 5742 }, { "epoch": 0.77, "learning_rate": 2.3279433382333287e-08, "loss": 0.0844, "step": 5743 }, { "epoch": 0.77, "learning_rate": 2.32660697581184e-08, "loss": 0.0675, "step": 5744 }, { "epoch": 0.77, "learning_rate": 2.3252706133903513e-08, "loss": 0.0332, "step": 5745 }, { "epoch": 0.77, "learning_rate": 2.3239342509688626e-08, "loss": 0.0772, "step": 5746 }, { "epoch": 0.77, "learning_rate": 2.322597888547374e-08, "loss": 0.1069, "step": 5747 }, { "epoch": 0.77, "learning_rate": 2.3212615261258853e-08, "loss": 0.0651, "step": 5748 }, { "epoch": 0.77, "learning_rate": 2.3199251637043966e-08, "loss": 0.0495, "step": 5749 }, { "epoch": 0.77, "learning_rate": 2.318588801282908e-08, "loss": 0.0939, "step": 5750 }, { "epoch": 0.77, "learning_rate": 2.3172524388614192e-08, "loss": 0.0624, "step": 5751 }, { "epoch": 0.77, "learning_rate": 2.3159160764399305e-08, "loss": 0.1065, "step": 5752 }, { "epoch": 0.77, "learning_rate": 2.314579714018442e-08, "loss": 0.0914, "step": 5753 }, { "epoch": 0.77, "learning_rate": 2.3132433515969528e-08, "loss": 0.0515, "step": 5754 }, { "epoch": 0.77, "learning_rate": 2.311906989175464e-08, "loss": 0.0451, "step": 5755 }, { "epoch": 0.77, "learning_rate": 2.3105706267539758e-08, "loss": 0.0679, "step": 5756 }, { "epoch": 0.77, "learning_rate": 2.3092342643324868e-08, "loss": 0.0712, "step": 5757 }, { "epoch": 0.77, "learning_rate": 2.307897901910998e-08, "loss": 0.0455, "step": 5758 }, { "epoch": 0.77, "learning_rate": 2.3065615394895094e-08, "loss": 0.0596, "step": 5759 }, { "epoch": 0.77, "learning_rate": 2.3052251770680207e-08, "loss": 0.0604, "step": 5760 }, { "epoch": 0.77, "learning_rate": 2.303888814646532e-08, "loss": 0.0468, "step": 5761 }, { "epoch": 0.77, "learning_rate": 2.3025524522250433e-08, "loss": 0.0729, "step": 5762 }, { "epoch": 0.77, "learning_rate": 2.3012160898035546e-08, "loss": 0.0597, "step": 5763 }, { "epoch": 0.77, "learning_rate": 2.299879727382066e-08, "loss": 0.0628, "step": 5764 }, { "epoch": 0.77, "learning_rate": 2.2985433649605773e-08, "loss": 0.0376, "step": 5765 }, { "epoch": 0.77, "learning_rate": 2.2972070025390886e-08, "loss": 0.0485, "step": 5766 }, { "epoch": 0.77, "learning_rate": 2.2958706401176e-08, "loss": 0.0524, "step": 5767 }, { "epoch": 0.77, "learning_rate": 2.294534277696111e-08, "loss": 0.0443, "step": 5768 }, { "epoch": 0.77, "learning_rate": 2.2931979152746222e-08, "loss": 0.0741, "step": 5769 }, { "epoch": 0.77, "learning_rate": 2.291861552853134e-08, "loss": 0.0549, "step": 5770 }, { "epoch": 0.77, "learning_rate": 2.2905251904316448e-08, "loss": 0.0433, "step": 5771 }, { "epoch": 0.77, "learning_rate": 2.289188828010156e-08, "loss": 0.1014, "step": 5772 }, { "epoch": 0.77, "learning_rate": 2.2878524655886675e-08, "loss": 0.0619, "step": 5773 }, { "epoch": 0.77, "learning_rate": 2.286516103167179e-08, "loss": 0.0538, "step": 5774 }, { "epoch": 0.77, "learning_rate": 2.28517974074569e-08, "loss": 0.0671, "step": 5775 }, { "epoch": 0.77, "learning_rate": 2.2838433783242014e-08, "loss": 0.0942, "step": 5776 }, { "epoch": 0.77, "learning_rate": 2.2825070159027127e-08, "loss": 0.0544, "step": 5777 }, { "epoch": 0.77, "learning_rate": 2.281170653481224e-08, "loss": 0.0728, "step": 5778 }, { "epoch": 0.77, "learning_rate": 2.2798342910597353e-08, "loss": 0.0904, "step": 5779 }, { "epoch": 0.77, "learning_rate": 2.2784979286382467e-08, "loss": 0.0653, "step": 5780 }, { "epoch": 0.77, "learning_rate": 2.277161566216758e-08, "loss": 0.0476, "step": 5781 }, { "epoch": 0.77, "learning_rate": 2.275825203795269e-08, "loss": 0.0595, "step": 5782 }, { "epoch": 0.77, "learning_rate": 2.2744888413737803e-08, "loss": 0.0754, "step": 5783 }, { "epoch": 0.77, "learning_rate": 2.273152478952292e-08, "loss": 0.0625, "step": 5784 }, { "epoch": 0.77, "learning_rate": 2.2718161165308032e-08, "loss": 0.0749, "step": 5785 }, { "epoch": 0.77, "learning_rate": 2.2704797541093142e-08, "loss": 0.0519, "step": 5786 }, { "epoch": 0.77, "learning_rate": 2.2691433916878255e-08, "loss": 0.0582, "step": 5787 }, { "epoch": 0.77, "learning_rate": 2.2678070292663372e-08, "loss": 0.0812, "step": 5788 }, { "epoch": 0.77, "learning_rate": 2.266470666844848e-08, "loss": 0.0774, "step": 5789 }, { "epoch": 0.77, "learning_rate": 2.2651343044233595e-08, "loss": 0.0655, "step": 5790 }, { "epoch": 0.77, "learning_rate": 2.2637979420018708e-08, "loss": 0.0666, "step": 5791 }, { "epoch": 0.77, "learning_rate": 2.262461579580382e-08, "loss": 0.0922, "step": 5792 }, { "epoch": 0.77, "learning_rate": 2.2611252171588934e-08, "loss": 0.0409, "step": 5793 }, { "epoch": 0.77, "learning_rate": 2.2597888547374047e-08, "loss": 0.0743, "step": 5794 }, { "epoch": 0.77, "learning_rate": 2.258452492315916e-08, "loss": 0.0411, "step": 5795 }, { "epoch": 0.77, "learning_rate": 2.257116129894427e-08, "loss": 0.0547, "step": 5796 }, { "epoch": 0.77, "learning_rate": 2.2557797674729387e-08, "loss": 0.0539, "step": 5797 }, { "epoch": 0.77, "learning_rate": 2.25444340505145e-08, "loss": 0.0529, "step": 5798 }, { "epoch": 0.77, "learning_rate": 2.2531070426299613e-08, "loss": 0.0612, "step": 5799 }, { "epoch": 0.78, "learning_rate": 2.2517706802084723e-08, "loss": 0.0528, "step": 5800 }, { "epoch": 0.78, "learning_rate": 2.2504343177869836e-08, "loss": 0.0459, "step": 5801 }, { "epoch": 0.78, "learning_rate": 2.2490979553654952e-08, "loss": 0.051, "step": 5802 }, { "epoch": 0.78, "learning_rate": 2.2477615929440062e-08, "loss": 0.0533, "step": 5803 }, { "epoch": 0.78, "learning_rate": 2.2464252305225175e-08, "loss": 0.0514, "step": 5804 }, { "epoch": 0.78, "learning_rate": 2.245088868101029e-08, "loss": 0.0684, "step": 5805 }, { "epoch": 0.78, "learning_rate": 2.2437525056795402e-08, "loss": 0.106, "step": 5806 }, { "epoch": 0.78, "learning_rate": 2.2424161432580515e-08, "loss": 0.0639, "step": 5807 }, { "epoch": 0.78, "learning_rate": 2.2410797808365628e-08, "loss": 0.0608, "step": 5808 }, { "epoch": 0.78, "learning_rate": 2.239743418415074e-08, "loss": 0.0524, "step": 5809 }, { "epoch": 0.78, "learning_rate": 2.238407055993585e-08, "loss": 0.0673, "step": 5810 }, { "epoch": 0.78, "learning_rate": 2.2370706935720967e-08, "loss": 0.0743, "step": 5811 }, { "epoch": 0.78, "learning_rate": 2.235734331150608e-08, "loss": 0.0489, "step": 5812 }, { "epoch": 0.78, "learning_rate": 2.2343979687291194e-08, "loss": 0.0732, "step": 5813 }, { "epoch": 0.78, "learning_rate": 2.2330616063076304e-08, "loss": 0.0503, "step": 5814 }, { "epoch": 0.78, "learning_rate": 2.2317252438861417e-08, "loss": 0.0658, "step": 5815 }, { "epoch": 0.78, "learning_rate": 2.2303888814646533e-08, "loss": 0.055, "step": 5816 }, { "epoch": 0.78, "learning_rate": 2.2290525190431643e-08, "loss": 0.0419, "step": 5817 }, { "epoch": 0.78, "learning_rate": 2.2277161566216756e-08, "loss": 0.0387, "step": 5818 }, { "epoch": 0.78, "learning_rate": 2.226379794200187e-08, "loss": 0.0566, "step": 5819 }, { "epoch": 0.78, "learning_rate": 2.2250434317786986e-08, "loss": 0.0694, "step": 5820 }, { "epoch": 0.78, "learning_rate": 2.2237070693572096e-08, "loss": 0.0615, "step": 5821 }, { "epoch": 0.78, "learning_rate": 2.222370706935721e-08, "loss": 0.0675, "step": 5822 }, { "epoch": 0.78, "learning_rate": 2.2210343445142322e-08, "loss": 0.0596, "step": 5823 }, { "epoch": 0.78, "learning_rate": 2.2196979820927435e-08, "loss": 0.061, "step": 5824 }, { "epoch": 0.78, "learning_rate": 2.2183616196712548e-08, "loss": 0.0612, "step": 5825 }, { "epoch": 0.78, "learning_rate": 2.217025257249766e-08, "loss": 0.0576, "step": 5826 }, { "epoch": 0.78, "learning_rate": 2.2156888948282774e-08, "loss": 0.0958, "step": 5827 }, { "epoch": 0.78, "learning_rate": 2.2143525324067884e-08, "loss": 0.0728, "step": 5828 }, { "epoch": 0.78, "learning_rate": 2.2130161699852997e-08, "loss": 0.0556, "step": 5829 }, { "epoch": 0.78, "learning_rate": 2.2116798075638114e-08, "loss": 0.1109, "step": 5830 }, { "epoch": 0.78, "learning_rate": 2.2103434451423224e-08, "loss": 0.0591, "step": 5831 }, { "epoch": 0.78, "learning_rate": 2.2090070827208337e-08, "loss": 0.0702, "step": 5832 }, { "epoch": 0.78, "learning_rate": 2.207670720299345e-08, "loss": 0.0574, "step": 5833 }, { "epoch": 0.78, "learning_rate": 2.2063343578778566e-08, "loss": 0.0633, "step": 5834 }, { "epoch": 0.78, "learning_rate": 2.2049979954563676e-08, "loss": 0.0806, "step": 5835 }, { "epoch": 0.78, "learning_rate": 2.203661633034879e-08, "loss": 0.0704, "step": 5836 }, { "epoch": 0.78, "learning_rate": 2.2023252706133902e-08, "loss": 0.0725, "step": 5837 }, { "epoch": 0.78, "learning_rate": 2.2009889081919016e-08, "loss": 0.0408, "step": 5838 }, { "epoch": 0.78, "learning_rate": 2.199652545770413e-08, "loss": 0.0647, "step": 5839 }, { "epoch": 0.78, "learning_rate": 2.1983161833489242e-08, "loss": 0.0814, "step": 5840 }, { "epoch": 0.78, "learning_rate": 2.1969798209274355e-08, "loss": 0.0498, "step": 5841 }, { "epoch": 0.78, "learning_rate": 2.1956434585059465e-08, "loss": 0.0515, "step": 5842 }, { "epoch": 0.78, "learning_rate": 2.194307096084458e-08, "loss": 0.0653, "step": 5843 }, { "epoch": 0.78, "learning_rate": 2.1929707336629694e-08, "loss": 0.0583, "step": 5844 }, { "epoch": 0.78, "learning_rate": 2.1916343712414808e-08, "loss": 0.0633, "step": 5845 }, { "epoch": 0.78, "learning_rate": 2.1902980088199917e-08, "loss": 0.0735, "step": 5846 }, { "epoch": 0.78, "learning_rate": 2.188961646398503e-08, "loss": 0.0744, "step": 5847 }, { "epoch": 0.78, "learning_rate": 2.1876252839770147e-08, "loss": 0.0701, "step": 5848 }, { "epoch": 0.78, "learning_rate": 2.1862889215555257e-08, "loss": 0.0738, "step": 5849 }, { "epoch": 0.78, "learning_rate": 2.184952559134037e-08, "loss": 0.0721, "step": 5850 }, { "epoch": 0.78, "learning_rate": 2.1836161967125483e-08, "loss": 0.0407, "step": 5851 }, { "epoch": 0.78, "learning_rate": 2.1822798342910596e-08, "loss": 0.063, "step": 5852 }, { "epoch": 0.78, "learning_rate": 2.180943471869571e-08, "loss": 0.0577, "step": 5853 }, { "epoch": 0.78, "learning_rate": 2.1796071094480823e-08, "loss": 0.0445, "step": 5854 }, { "epoch": 0.78, "learning_rate": 2.1782707470265936e-08, "loss": 0.0505, "step": 5855 }, { "epoch": 0.78, "learning_rate": 2.1769343846051046e-08, "loss": 0.065, "step": 5856 }, { "epoch": 0.78, "learning_rate": 2.1755980221836162e-08, "loss": 0.0567, "step": 5857 }, { "epoch": 0.78, "learning_rate": 2.1742616597621275e-08, "loss": 0.0662, "step": 5858 }, { "epoch": 0.78, "learning_rate": 2.1729252973406388e-08, "loss": 0.0711, "step": 5859 }, { "epoch": 0.78, "learning_rate": 2.1715889349191498e-08, "loss": 0.0371, "step": 5860 }, { "epoch": 0.78, "learning_rate": 2.170252572497661e-08, "loss": 0.0742, "step": 5861 }, { "epoch": 0.78, "learning_rate": 2.1689162100761728e-08, "loss": 0.0398, "step": 5862 }, { "epoch": 0.78, "learning_rate": 2.1675798476546838e-08, "loss": 0.0615, "step": 5863 }, { "epoch": 0.78, "learning_rate": 2.166243485233195e-08, "loss": 0.0632, "step": 5864 }, { "epoch": 0.78, "learning_rate": 2.1649071228117064e-08, "loss": 0.1038, "step": 5865 }, { "epoch": 0.78, "learning_rate": 2.1635707603902177e-08, "loss": 0.0614, "step": 5866 }, { "epoch": 0.78, "learning_rate": 2.162234397968729e-08, "loss": 0.0834, "step": 5867 }, { "epoch": 0.78, "learning_rate": 2.1608980355472403e-08, "loss": 0.0539, "step": 5868 }, { "epoch": 0.78, "learning_rate": 2.1595616731257516e-08, "loss": 0.0772, "step": 5869 }, { "epoch": 0.78, "learning_rate": 2.1582253107042626e-08, "loss": 0.078, "step": 5870 }, { "epoch": 0.78, "learning_rate": 2.1568889482827743e-08, "loss": 0.0625, "step": 5871 }, { "epoch": 0.78, "learning_rate": 2.1555525858612856e-08, "loss": 0.0657, "step": 5872 }, { "epoch": 0.78, "learning_rate": 2.154216223439797e-08, "loss": 0.0581, "step": 5873 }, { "epoch": 0.78, "learning_rate": 2.152879861018308e-08, "loss": 0.0395, "step": 5874 }, { "epoch": 0.79, "learning_rate": 2.1515434985968195e-08, "loss": 0.0391, "step": 5875 }, { "epoch": 0.79, "learning_rate": 2.150207136175331e-08, "loss": 0.0926, "step": 5876 }, { "epoch": 0.79, "learning_rate": 2.1488707737538418e-08, "loss": 0.0572, "step": 5877 }, { "epoch": 0.79, "learning_rate": 2.147534411332353e-08, "loss": 0.0695, "step": 5878 }, { "epoch": 0.79, "learning_rate": 2.1461980489108645e-08, "loss": 0.0733, "step": 5879 }, { "epoch": 0.79, "learning_rate": 2.144861686489376e-08, "loss": 0.0604, "step": 5880 }, { "epoch": 0.79, "learning_rate": 2.143525324067887e-08, "loss": 0.0758, "step": 5881 }, { "epoch": 0.79, "learning_rate": 2.1421889616463984e-08, "loss": 0.0696, "step": 5882 }, { "epoch": 0.79, "learning_rate": 2.1408525992249097e-08, "loss": 0.0619, "step": 5883 }, { "epoch": 0.79, "learning_rate": 2.139516236803421e-08, "loss": 0.0643, "step": 5884 }, { "epoch": 0.79, "learning_rate": 2.1381798743819323e-08, "loss": 0.0479, "step": 5885 }, { "epoch": 0.79, "learning_rate": 2.1368435119604437e-08, "loss": 0.0724, "step": 5886 }, { "epoch": 0.79, "learning_rate": 2.135507149538955e-08, "loss": 0.0469, "step": 5887 }, { "epoch": 0.79, "learning_rate": 2.134170787117466e-08, "loss": 0.0685, "step": 5888 }, { "epoch": 0.79, "learning_rate": 2.1328344246959776e-08, "loss": 0.0737, "step": 5889 }, { "epoch": 0.79, "learning_rate": 2.131498062274489e-08, "loss": 0.0737, "step": 5890 }, { "epoch": 0.79, "learning_rate": 2.130161699853e-08, "loss": 0.0649, "step": 5891 }, { "epoch": 0.79, "learning_rate": 2.1288253374315112e-08, "loss": 0.1527, "step": 5892 }, { "epoch": 0.79, "learning_rate": 2.1274889750100225e-08, "loss": 0.0767, "step": 5893 }, { "epoch": 0.79, "learning_rate": 2.126152612588534e-08, "loss": 0.0809, "step": 5894 }, { "epoch": 0.79, "learning_rate": 2.124816250167045e-08, "loss": 0.0979, "step": 5895 }, { "epoch": 0.79, "learning_rate": 2.1234798877455565e-08, "loss": 0.0656, "step": 5896 }, { "epoch": 0.79, "learning_rate": 2.1221435253240678e-08, "loss": 0.074, "step": 5897 }, { "epoch": 0.79, "learning_rate": 2.120807162902579e-08, "loss": 0.086, "step": 5898 }, { "epoch": 0.79, "learning_rate": 2.1194708004810904e-08, "loss": 0.0705, "step": 5899 }, { "epoch": 0.79, "learning_rate": 2.1181344380596017e-08, "loss": 0.0673, "step": 5900 }, { "epoch": 0.79, "learning_rate": 2.116798075638113e-08, "loss": 0.0592, "step": 5901 }, { "epoch": 0.79, "learning_rate": 2.115461713216624e-08, "loss": 0.0535, "step": 5902 }, { "epoch": 0.79, "learning_rate": 2.1141253507951357e-08, "loss": 0.0571, "step": 5903 }, { "epoch": 0.79, "learning_rate": 2.112788988373647e-08, "loss": 0.0326, "step": 5904 }, { "epoch": 0.79, "learning_rate": 2.111452625952158e-08, "loss": 0.0571, "step": 5905 }, { "epoch": 0.79, "learning_rate": 2.1101162635306693e-08, "loss": 0.0402, "step": 5906 }, { "epoch": 0.79, "learning_rate": 2.1087799011091806e-08, "loss": 0.0786, "step": 5907 }, { "epoch": 0.79, "learning_rate": 2.1074435386876922e-08, "loss": 0.08, "step": 5908 }, { "epoch": 0.79, "learning_rate": 2.1061071762662032e-08, "loss": 0.0507, "step": 5909 }, { "epoch": 0.79, "learning_rate": 2.1047708138447145e-08, "loss": 0.0645, "step": 5910 }, { "epoch": 0.79, "learning_rate": 2.103434451423226e-08, "loss": 0.0434, "step": 5911 }, { "epoch": 0.79, "learning_rate": 2.102098089001737e-08, "loss": 0.0609, "step": 5912 }, { "epoch": 0.79, "learning_rate": 2.1007617265802485e-08, "loss": 0.0789, "step": 5913 }, { "epoch": 0.79, "learning_rate": 2.0994253641587598e-08, "loss": 0.0683, "step": 5914 }, { "epoch": 0.79, "learning_rate": 2.098089001737271e-08, "loss": 0.0494, "step": 5915 }, { "epoch": 0.79, "learning_rate": 2.096752639315782e-08, "loss": 0.0733, "step": 5916 }, { "epoch": 0.79, "learning_rate": 2.0954162768942937e-08, "loss": 0.0462, "step": 5917 }, { "epoch": 0.79, "learning_rate": 2.094079914472805e-08, "loss": 0.0613, "step": 5918 }, { "epoch": 0.79, "learning_rate": 2.0927435520513164e-08, "loss": 0.0573, "step": 5919 }, { "epoch": 0.79, "learning_rate": 2.0914071896298273e-08, "loss": 0.0633, "step": 5920 }, { "epoch": 0.79, "learning_rate": 2.090070827208339e-08, "loss": 0.0634, "step": 5921 }, { "epoch": 0.79, "learning_rate": 2.0887344647868503e-08, "loss": 0.045, "step": 5922 }, { "epoch": 0.79, "learning_rate": 2.0873981023653613e-08, "loss": 0.0754, "step": 5923 }, { "epoch": 0.79, "learning_rate": 2.0860617399438726e-08, "loss": 0.0781, "step": 5924 }, { "epoch": 0.79, "learning_rate": 2.084725377522384e-08, "loss": 0.0476, "step": 5925 }, { "epoch": 0.79, "learning_rate": 2.0833890151008952e-08, "loss": 0.0664, "step": 5926 }, { "epoch": 0.79, "learning_rate": 2.0820526526794065e-08, "loss": 0.0551, "step": 5927 }, { "epoch": 0.79, "learning_rate": 2.080716290257918e-08, "loss": 0.0633, "step": 5928 }, { "epoch": 0.79, "learning_rate": 2.0793799278364292e-08, "loss": 0.0491, "step": 5929 }, { "epoch": 0.79, "learning_rate": 2.07804356541494e-08, "loss": 0.0713, "step": 5930 }, { "epoch": 0.79, "learning_rate": 2.0767072029934518e-08, "loss": 0.0522, "step": 5931 }, { "epoch": 0.79, "learning_rate": 2.075370840571963e-08, "loss": 0.0719, "step": 5932 }, { "epoch": 0.79, "learning_rate": 2.0740344781504744e-08, "loss": 0.0645, "step": 5933 }, { "epoch": 0.79, "learning_rate": 2.0726981157289854e-08, "loss": 0.0651, "step": 5934 }, { "epoch": 0.79, "learning_rate": 2.071361753307497e-08, "loss": 0.0819, "step": 5935 }, { "epoch": 0.79, "learning_rate": 2.0700253908860084e-08, "loss": 0.0953, "step": 5936 }, { "epoch": 0.79, "learning_rate": 2.0686890284645194e-08, "loss": 0.057, "step": 5937 }, { "epoch": 0.79, "learning_rate": 2.0673526660430307e-08, "loss": 0.0475, "step": 5938 }, { "epoch": 0.79, "learning_rate": 2.066016303621542e-08, "loss": 0.0837, "step": 5939 }, { "epoch": 0.79, "learning_rate": 2.0646799412000533e-08, "loss": 0.0653, "step": 5940 }, { "epoch": 0.79, "learning_rate": 2.0633435787785646e-08, "loss": 0.0652, "step": 5941 }, { "epoch": 0.79, "learning_rate": 2.062007216357076e-08, "loss": 0.0813, "step": 5942 }, { "epoch": 0.79, "learning_rate": 2.0606708539355872e-08, "loss": 0.0435, "step": 5943 }, { "epoch": 0.79, "learning_rate": 2.0593344915140986e-08, "loss": 0.0707, "step": 5944 }, { "epoch": 0.79, "learning_rate": 2.05799812909261e-08, "loss": 0.0644, "step": 5945 }, { "epoch": 0.79, "learning_rate": 2.0566617666711212e-08, "loss": 0.0444, "step": 5946 }, { "epoch": 0.79, "learning_rate": 2.0553254042496325e-08, "loss": 0.063, "step": 5947 }, { "epoch": 0.79, "learning_rate": 2.0539890418281435e-08, "loss": 0.1253, "step": 5948 }, { "epoch": 0.8, "learning_rate": 2.052652679406655e-08, "loss": 0.0701, "step": 5949 }, { "epoch": 0.8, "learning_rate": 2.0513163169851664e-08, "loss": 0.063, "step": 5950 }, { "epoch": 0.8, "learning_rate": 2.0499799545636774e-08, "loss": 0.0747, "step": 5951 }, { "epoch": 0.8, "learning_rate": 2.0486435921421887e-08, "loss": 0.0469, "step": 5952 }, { "epoch": 0.8, "learning_rate": 2.0473072297207e-08, "loss": 0.0653, "step": 5953 }, { "epoch": 0.8, "learning_rate": 2.0459708672992117e-08, "loss": 0.0742, "step": 5954 }, { "epoch": 0.8, "learning_rate": 2.0446345048777227e-08, "loss": 0.0426, "step": 5955 }, { "epoch": 0.8, "learning_rate": 2.043298142456234e-08, "loss": 0.0666, "step": 5956 }, { "epoch": 0.8, "learning_rate": 2.0419617800347453e-08, "loss": 0.0537, "step": 5957 }, { "epoch": 0.8, "learning_rate": 2.0406254176132566e-08, "loss": 0.0792, "step": 5958 }, { "epoch": 0.8, "learning_rate": 2.039289055191768e-08, "loss": 0.0868, "step": 5959 }, { "epoch": 0.8, "learning_rate": 2.0379526927702792e-08, "loss": 0.0651, "step": 5960 }, { "epoch": 0.8, "learning_rate": 2.0366163303487906e-08, "loss": 0.0604, "step": 5961 }, { "epoch": 0.8, "learning_rate": 2.0352799679273015e-08, "loss": 0.0565, "step": 5962 }, { "epoch": 0.8, "learning_rate": 2.0339436055058132e-08, "loss": 0.062, "step": 5963 }, { "epoch": 0.8, "learning_rate": 2.0326072430843245e-08, "loss": 0.0642, "step": 5964 }, { "epoch": 0.8, "learning_rate": 2.0312708806628355e-08, "loss": 0.0648, "step": 5965 }, { "epoch": 0.8, "learning_rate": 2.0299345182413468e-08, "loss": 0.0369, "step": 5966 }, { "epoch": 0.8, "learning_rate": 2.0285981558198584e-08, "loss": 0.0721, "step": 5967 }, { "epoch": 0.8, "learning_rate": 2.0272617933983698e-08, "loss": 0.0955, "step": 5968 }, { "epoch": 0.8, "learning_rate": 2.0259254309768807e-08, "loss": 0.0696, "step": 5969 }, { "epoch": 0.8, "learning_rate": 2.024589068555392e-08, "loss": 0.0332, "step": 5970 }, { "epoch": 0.8, "learning_rate": 2.0232527061339034e-08, "loss": 0.0781, "step": 5971 }, { "epoch": 0.8, "learning_rate": 2.0219163437124147e-08, "loss": 0.0728, "step": 5972 }, { "epoch": 0.8, "learning_rate": 2.020579981290926e-08, "loss": 0.0631, "step": 5973 }, { "epoch": 0.8, "learning_rate": 2.0192436188694373e-08, "loss": 0.07, "step": 5974 }, { "epoch": 0.8, "learning_rate": 2.0179072564479486e-08, "loss": 0.0595, "step": 5975 }, { "epoch": 0.8, "learning_rate": 2.01657089402646e-08, "loss": 0.0788, "step": 5976 }, { "epoch": 0.8, "learning_rate": 2.0152345316049713e-08, "loss": 0.058, "step": 5977 }, { "epoch": 0.8, "learning_rate": 2.0138981691834826e-08, "loss": 0.0468, "step": 5978 }, { "epoch": 0.8, "learning_rate": 2.0125618067619936e-08, "loss": 0.0673, "step": 5979 }, { "epoch": 0.8, "learning_rate": 2.011225444340505e-08, "loss": 0.0567, "step": 5980 }, { "epoch": 0.8, "learning_rate": 2.0098890819190165e-08, "loss": 0.0504, "step": 5981 }, { "epoch": 0.8, "learning_rate": 2.0085527194975278e-08, "loss": 0.0512, "step": 5982 }, { "epoch": 0.8, "learning_rate": 2.0072163570760388e-08, "loss": 0.1252, "step": 5983 }, { "epoch": 0.8, "learning_rate": 2.00587999465455e-08, "loss": 0.0573, "step": 5984 }, { "epoch": 0.8, "learning_rate": 2.0045436322330614e-08, "loss": 0.049, "step": 5985 }, { "epoch": 0.8, "learning_rate": 2.0032072698115728e-08, "loss": 0.0698, "step": 5986 }, { "epoch": 0.8, "learning_rate": 2.001870907390084e-08, "loss": 0.0726, "step": 5987 }, { "epoch": 0.8, "learning_rate": 2.0005345449685954e-08, "loss": 0.0559, "step": 5988 }, { "epoch": 0.8, "learning_rate": 1.9991981825471067e-08, "loss": 0.0377, "step": 5989 }, { "epoch": 0.8, "learning_rate": 1.997861820125618e-08, "loss": 0.0496, "step": 5990 }, { "epoch": 0.8, "learning_rate": 1.9965254577041293e-08, "loss": 0.0618, "step": 5991 }, { "epoch": 0.8, "learning_rate": 1.9951890952826406e-08, "loss": 0.0551, "step": 5992 }, { "epoch": 0.8, "learning_rate": 1.993852732861152e-08, "loss": 0.0484, "step": 5993 }, { "epoch": 0.8, "learning_rate": 1.992516370439663e-08, "loss": 0.0581, "step": 5994 }, { "epoch": 0.8, "learning_rate": 1.9911800080181746e-08, "loss": 0.0525, "step": 5995 }, { "epoch": 0.8, "learning_rate": 1.989843645596686e-08, "loss": 0.0635, "step": 5996 }, { "epoch": 0.8, "learning_rate": 1.988507283175197e-08, "loss": 0.053, "step": 5997 }, { "epoch": 0.8, "learning_rate": 1.9871709207537082e-08, "loss": 0.0787, "step": 5998 }, { "epoch": 0.8, "learning_rate": 1.98583455833222e-08, "loss": 0.0887, "step": 5999 }, { "epoch": 0.8, "learning_rate": 1.9844981959107308e-08, "loss": 0.0524, "step": 6000 }, { "epoch": 0.8, "learning_rate": 1.983161833489242e-08, "loss": 0.0709, "step": 6001 }, { "epoch": 0.8, "learning_rate": 1.9818254710677535e-08, "loss": 0.0703, "step": 6002 }, { "epoch": 0.8, "learning_rate": 1.9804891086462648e-08, "loss": 0.0648, "step": 6003 }, { "epoch": 0.8, "learning_rate": 1.979152746224776e-08, "loss": 0.0528, "step": 6004 }, { "epoch": 0.8, "learning_rate": 1.9778163838032874e-08, "loss": 0.0566, "step": 6005 }, { "epoch": 0.8, "learning_rate": 1.9764800213817987e-08, "loss": 0.0399, "step": 6006 }, { "epoch": 0.8, "learning_rate": 1.97514365896031e-08, "loss": 0.0506, "step": 6007 }, { "epoch": 0.8, "learning_rate": 1.973807296538821e-08, "loss": 0.0775, "step": 6008 }, { "epoch": 0.8, "learning_rate": 1.9724709341173327e-08, "loss": 0.0499, "step": 6009 }, { "epoch": 0.8, "learning_rate": 1.971134571695844e-08, "loss": 0.0852, "step": 6010 }, { "epoch": 0.8, "learning_rate": 1.969798209274355e-08, "loss": 0.0452, "step": 6011 }, { "epoch": 0.8, "learning_rate": 1.9684618468528663e-08, "loss": 0.0538, "step": 6012 }, { "epoch": 0.8, "learning_rate": 1.967125484431378e-08, "loss": 0.1007, "step": 6013 }, { "epoch": 0.8, "learning_rate": 1.965789122009889e-08, "loss": 0.0623, "step": 6014 }, { "epoch": 0.8, "learning_rate": 1.9644527595884002e-08, "loss": 0.0698, "step": 6015 }, { "epoch": 0.8, "learning_rate": 1.9631163971669115e-08, "loss": 0.1014, "step": 6016 }, { "epoch": 0.8, "learning_rate": 1.9617800347454228e-08, "loss": 0.0636, "step": 6017 }, { "epoch": 0.8, "learning_rate": 1.960443672323934e-08, "loss": 0.0565, "step": 6018 }, { "epoch": 0.8, "learning_rate": 1.9591073099024455e-08, "loss": 0.0667, "step": 6019 }, { "epoch": 0.8, "learning_rate": 1.9577709474809568e-08, "loss": 0.0378, "step": 6020 }, { "epoch": 0.8, "learning_rate": 1.956434585059468e-08, "loss": 0.0657, "step": 6021 }, { "epoch": 0.8, "learning_rate": 1.9550982226379794e-08, "loss": 0.0634, "step": 6022 }, { "epoch": 0.8, "learning_rate": 1.9537618602164907e-08, "loss": 0.0581, "step": 6023 }, { "epoch": 0.81, "learning_rate": 1.952425497795002e-08, "loss": 0.0719, "step": 6024 }, { "epoch": 0.81, "learning_rate": 1.951089135373513e-08, "loss": 0.0337, "step": 6025 }, { "epoch": 0.81, "learning_rate": 1.9497527729520243e-08, "loss": 0.0481, "step": 6026 }, { "epoch": 0.81, "learning_rate": 1.948416410530536e-08, "loss": 0.0641, "step": 6027 }, { "epoch": 0.81, "learning_rate": 1.9470800481090473e-08, "loss": 0.0539, "step": 6028 }, { "epoch": 0.81, "learning_rate": 1.9457436856875583e-08, "loss": 0.0464, "step": 6029 }, { "epoch": 0.81, "learning_rate": 1.9444073232660696e-08, "loss": 0.0568, "step": 6030 }, { "epoch": 0.81, "learning_rate": 1.943070960844581e-08, "loss": 0.0712, "step": 6031 }, { "epoch": 0.81, "learning_rate": 1.9417345984230922e-08, "loss": 0.0587, "step": 6032 }, { "epoch": 0.81, "learning_rate": 1.9403982360016035e-08, "loss": 0.1167, "step": 6033 }, { "epoch": 0.81, "learning_rate": 1.939061873580115e-08, "loss": 0.0526, "step": 6034 }, { "epoch": 0.81, "learning_rate": 1.937725511158626e-08, "loss": 0.0569, "step": 6035 }, { "epoch": 0.81, "learning_rate": 1.9363891487371375e-08, "loss": 0.0912, "step": 6036 }, { "epoch": 0.81, "learning_rate": 1.9350527863156488e-08, "loss": 0.0493, "step": 6037 }, { "epoch": 0.81, "learning_rate": 1.93371642389416e-08, "loss": 0.0745, "step": 6038 }, { "epoch": 0.81, "learning_rate": 1.932380061472671e-08, "loss": 0.0619, "step": 6039 }, { "epoch": 0.81, "learning_rate": 1.9310436990511824e-08, "loss": 0.0555, "step": 6040 }, { "epoch": 0.81, "learning_rate": 1.929707336629694e-08, "loss": 0.0552, "step": 6041 }, { "epoch": 0.81, "learning_rate": 1.9283709742082054e-08, "loss": 0.0735, "step": 6042 }, { "epoch": 0.81, "learning_rate": 1.9270346117867163e-08, "loss": 0.0463, "step": 6043 }, { "epoch": 0.81, "learning_rate": 1.9256982493652277e-08, "loss": 0.0443, "step": 6044 }, { "epoch": 0.81, "learning_rate": 1.9243618869437393e-08, "loss": 0.0938, "step": 6045 }, { "epoch": 0.81, "learning_rate": 1.9230255245222503e-08, "loss": 0.0627, "step": 6046 }, { "epoch": 0.81, "learning_rate": 1.9216891621007616e-08, "loss": 0.0467, "step": 6047 }, { "epoch": 0.81, "learning_rate": 1.920352799679273e-08, "loss": 0.0632, "step": 6048 }, { "epoch": 0.81, "learning_rate": 1.9190164372577842e-08, "loss": 0.0786, "step": 6049 }, { "epoch": 0.81, "learning_rate": 1.9176800748362955e-08, "loss": 0.0487, "step": 6050 }, { "epoch": 0.81, "learning_rate": 1.916343712414807e-08, "loss": 0.0585, "step": 6051 }, { "epoch": 0.81, "learning_rate": 1.9150073499933182e-08, "loss": 0.057, "step": 6052 }, { "epoch": 0.81, "learning_rate": 1.9136709875718295e-08, "loss": 0.0435, "step": 6053 }, { "epoch": 0.81, "learning_rate": 1.9123346251503405e-08, "loss": 0.0811, "step": 6054 }, { "epoch": 0.81, "learning_rate": 1.910998262728852e-08, "loss": 0.0944, "step": 6055 }, { "epoch": 0.81, "learning_rate": 1.9096619003073634e-08, "loss": 0.0606, "step": 6056 }, { "epoch": 0.81, "learning_rate": 1.9083255378858744e-08, "loss": 0.0521, "step": 6057 }, { "epoch": 0.81, "learning_rate": 1.9069891754643857e-08, "loss": 0.0757, "step": 6058 }, { "epoch": 0.81, "learning_rate": 1.9056528130428974e-08, "loss": 0.0671, "step": 6059 }, { "epoch": 0.81, "learning_rate": 1.9043164506214084e-08, "loss": 0.0429, "step": 6060 }, { "epoch": 0.81, "learning_rate": 1.9029800881999197e-08, "loss": 0.0782, "step": 6061 }, { "epoch": 0.81, "learning_rate": 1.901643725778431e-08, "loss": 0.0476, "step": 6062 }, { "epoch": 0.81, "learning_rate": 1.9003073633569423e-08, "loss": 0.0706, "step": 6063 }, { "epoch": 0.81, "learning_rate": 1.8989710009354536e-08, "loss": 0.0415, "step": 6064 }, { "epoch": 0.81, "learning_rate": 1.897634638513965e-08, "loss": 0.0576, "step": 6065 }, { "epoch": 0.81, "learning_rate": 1.8962982760924762e-08, "loss": 0.1119, "step": 6066 }, { "epoch": 0.81, "learning_rate": 1.8949619136709876e-08, "loss": 0.0592, "step": 6067 }, { "epoch": 0.81, "learning_rate": 1.893625551249499e-08, "loss": 0.0752, "step": 6068 }, { "epoch": 0.81, "learning_rate": 1.8922891888280102e-08, "loss": 0.0889, "step": 6069 }, { "epoch": 0.81, "learning_rate": 1.8909528264065215e-08, "loss": 0.0597, "step": 6070 }, { "epoch": 0.81, "learning_rate": 1.8896164639850325e-08, "loss": 0.0494, "step": 6071 }, { "epoch": 0.81, "learning_rate": 1.8882801015635438e-08, "loss": 0.0686, "step": 6072 }, { "epoch": 0.81, "learning_rate": 1.8869437391420554e-08, "loss": 0.0968, "step": 6073 }, { "epoch": 0.81, "learning_rate": 1.8856073767205664e-08, "loss": 0.0734, "step": 6074 }, { "epoch": 0.81, "learning_rate": 1.8842710142990777e-08, "loss": 0.0624, "step": 6075 }, { "epoch": 0.81, "learning_rate": 1.882934651877589e-08, "loss": 0.0527, "step": 6076 }, { "epoch": 0.81, "learning_rate": 1.8815982894561004e-08, "loss": 0.075, "step": 6077 }, { "epoch": 0.81, "learning_rate": 1.8802619270346117e-08, "loss": 0.0625, "step": 6078 }, { "epoch": 0.81, "learning_rate": 1.878925564613123e-08, "loss": 0.0569, "step": 6079 }, { "epoch": 0.81, "learning_rate": 1.8775892021916343e-08, "loss": 0.0632, "step": 6080 }, { "epoch": 0.81, "learning_rate": 1.8762528397701456e-08, "loss": 0.0561, "step": 6081 }, { "epoch": 0.81, "learning_rate": 1.874916477348657e-08, "loss": 0.0686, "step": 6082 }, { "epoch": 0.81, "learning_rate": 1.8735801149271682e-08, "loss": 0.0703, "step": 6083 }, { "epoch": 0.81, "learning_rate": 1.8722437525056796e-08, "loss": 0.0646, "step": 6084 }, { "epoch": 0.81, "learning_rate": 1.8709073900841905e-08, "loss": 0.0639, "step": 6085 }, { "epoch": 0.81, "learning_rate": 1.869571027662702e-08, "loss": 0.0613, "step": 6086 }, { "epoch": 0.81, "learning_rate": 1.8682346652412135e-08, "loss": 0.0577, "step": 6087 }, { "epoch": 0.81, "learning_rate": 1.8668983028197248e-08, "loss": 0.0466, "step": 6088 }, { "epoch": 0.81, "learning_rate": 1.8655619403982358e-08, "loss": 0.0621, "step": 6089 }, { "epoch": 0.81, "learning_rate": 1.864225577976747e-08, "loss": 0.0546, "step": 6090 }, { "epoch": 0.81, "learning_rate": 1.8628892155552588e-08, "loss": 0.0795, "step": 6091 }, { "epoch": 0.81, "learning_rate": 1.8615528531337697e-08, "loss": 0.0668, "step": 6092 }, { "epoch": 0.81, "learning_rate": 1.860216490712281e-08, "loss": 0.0513, "step": 6093 }, { "epoch": 0.81, "learning_rate": 1.8588801282907924e-08, "loss": 0.0456, "step": 6094 }, { "epoch": 0.81, "learning_rate": 1.8575437658693037e-08, "loss": 0.0719, "step": 6095 }, { "epoch": 0.81, "learning_rate": 1.856207403447815e-08, "loss": 0.0909, "step": 6096 }, { "epoch": 0.81, "learning_rate": 1.8548710410263263e-08, "loss": 0.0713, "step": 6097 }, { "epoch": 0.81, "learning_rate": 1.8535346786048376e-08, "loss": 0.0502, "step": 6098 }, { "epoch": 0.82, "learning_rate": 1.8521983161833486e-08, "loss": 0.072, "step": 6099 }, { "epoch": 0.82, "learning_rate": 1.85086195376186e-08, "loss": 0.0652, "step": 6100 }, { "epoch": 0.82, "learning_rate": 1.8495255913403716e-08, "loss": 0.0747, "step": 6101 }, { "epoch": 0.82, "learning_rate": 1.848189228918883e-08, "loss": 0.0739, "step": 6102 }, { "epoch": 0.82, "learning_rate": 1.846852866497394e-08, "loss": 0.0647, "step": 6103 }, { "epoch": 0.82, "learning_rate": 1.8455165040759052e-08, "loss": 0.0636, "step": 6104 }, { "epoch": 0.82, "learning_rate": 1.8441801416544168e-08, "loss": 0.063, "step": 6105 }, { "epoch": 0.82, "learning_rate": 1.8428437792329278e-08, "loss": 0.0627, "step": 6106 }, { "epoch": 0.82, "learning_rate": 1.841507416811439e-08, "loss": 0.0741, "step": 6107 }, { "epoch": 0.82, "learning_rate": 1.8401710543899504e-08, "loss": 0.0706, "step": 6108 }, { "epoch": 0.82, "learning_rate": 1.8388346919684618e-08, "loss": 0.0608, "step": 6109 }, { "epoch": 0.82, "learning_rate": 1.837498329546973e-08, "loss": 0.0427, "step": 6110 }, { "epoch": 0.82, "learning_rate": 1.8361619671254844e-08, "loss": 0.0736, "step": 6111 }, { "epoch": 0.82, "learning_rate": 1.8348256047039957e-08, "loss": 0.0494, "step": 6112 }, { "epoch": 0.82, "learning_rate": 1.8334892422825067e-08, "loss": 0.0573, "step": 6113 }, { "epoch": 0.82, "learning_rate": 1.8321528798610183e-08, "loss": 0.0952, "step": 6114 }, { "epoch": 0.82, "learning_rate": 1.8308165174395296e-08, "loss": 0.0517, "step": 6115 }, { "epoch": 0.82, "learning_rate": 1.829480155018041e-08, "loss": 0.0627, "step": 6116 }, { "epoch": 0.82, "learning_rate": 1.828143792596552e-08, "loss": 0.0656, "step": 6117 }, { "epoch": 0.82, "learning_rate": 1.8268074301750633e-08, "loss": 0.0503, "step": 6118 }, { "epoch": 0.82, "learning_rate": 1.825471067753575e-08, "loss": 0.0625, "step": 6119 }, { "epoch": 0.82, "learning_rate": 1.824134705332086e-08, "loss": 0.054, "step": 6120 }, { "epoch": 0.82, "learning_rate": 1.8227983429105972e-08, "loss": 0.0568, "step": 6121 }, { "epoch": 0.82, "learning_rate": 1.8214619804891085e-08, "loss": 0.061, "step": 6122 }, { "epoch": 0.82, "learning_rate": 1.82012561806762e-08, "loss": 0.0538, "step": 6123 }, { "epoch": 0.82, "learning_rate": 1.818789255646131e-08, "loss": 0.0746, "step": 6124 }, { "epoch": 0.82, "learning_rate": 1.8174528932246425e-08, "loss": 0.091, "step": 6125 }, { "epoch": 0.82, "learning_rate": 1.8161165308031538e-08, "loss": 0.0547, "step": 6126 }, { "epoch": 0.82, "learning_rate": 1.814780168381665e-08, "loss": 0.0548, "step": 6127 }, { "epoch": 0.82, "learning_rate": 1.8134438059601764e-08, "loss": 0.0751, "step": 6128 }, { "epoch": 0.82, "learning_rate": 1.8121074435386877e-08, "loss": 0.0639, "step": 6129 }, { "epoch": 0.82, "learning_rate": 1.810771081117199e-08, "loss": 0.062, "step": 6130 }, { "epoch": 0.82, "learning_rate": 1.80943471869571e-08, "loss": 0.0661, "step": 6131 }, { "epoch": 0.82, "learning_rate": 1.8080983562742213e-08, "loss": 0.0828, "step": 6132 }, { "epoch": 0.82, "learning_rate": 1.806761993852733e-08, "loss": 0.0825, "step": 6133 }, { "epoch": 0.82, "learning_rate": 1.805425631431244e-08, "loss": 0.0641, "step": 6134 }, { "epoch": 0.82, "learning_rate": 1.8040892690097553e-08, "loss": 0.0713, "step": 6135 }, { "epoch": 0.82, "learning_rate": 1.8027529065882666e-08, "loss": 0.0789, "step": 6136 }, { "epoch": 0.82, "learning_rate": 1.8014165441667782e-08, "loss": 0.0605, "step": 6137 }, { "epoch": 0.82, "learning_rate": 1.8000801817452892e-08, "loss": 0.0542, "step": 6138 }, { "epoch": 0.82, "learning_rate": 1.7987438193238005e-08, "loss": 0.0935, "step": 6139 }, { "epoch": 0.82, "learning_rate": 1.797407456902312e-08, "loss": 0.0812, "step": 6140 }, { "epoch": 0.82, "learning_rate": 1.796071094480823e-08, "loss": 0.6617, "step": 6141 }, { "epoch": 0.82, "learning_rate": 1.7947347320593345e-08, "loss": 0.0617, "step": 6142 }, { "epoch": 0.82, "learning_rate": 1.7933983696378458e-08, "loss": 0.0525, "step": 6143 }, { "epoch": 0.82, "learning_rate": 1.792062007216357e-08, "loss": 0.0622, "step": 6144 }, { "epoch": 0.82, "learning_rate": 1.790725644794868e-08, "loss": 0.0577, "step": 6145 }, { "epoch": 0.82, "learning_rate": 1.7893892823733797e-08, "loss": 0.0523, "step": 6146 }, { "epoch": 0.82, "learning_rate": 1.788052919951891e-08, "loss": 0.0655, "step": 6147 }, { "epoch": 0.82, "learning_rate": 1.786716557530402e-08, "loss": 0.0864, "step": 6148 }, { "epoch": 0.82, "learning_rate": 1.7853801951089133e-08, "loss": 0.0991, "step": 6149 }, { "epoch": 0.82, "learning_rate": 1.7840438326874246e-08, "loss": 0.0711, "step": 6150 }, { "epoch": 0.82, "learning_rate": 1.7827074702659363e-08, "loss": 0.0578, "step": 6151 }, { "epoch": 0.82, "learning_rate": 1.7813711078444473e-08, "loss": 0.0637, "step": 6152 }, { "epoch": 0.82, "learning_rate": 1.7800347454229586e-08, "loss": 0.0736, "step": 6153 }, { "epoch": 0.82, "learning_rate": 1.77869838300147e-08, "loss": 0.0556, "step": 6154 }, { "epoch": 0.82, "learning_rate": 1.7773620205799812e-08, "loss": 0.0691, "step": 6155 }, { "epoch": 0.82, "learning_rate": 1.7760256581584925e-08, "loss": 0.0958, "step": 6156 }, { "epoch": 0.82, "learning_rate": 1.774689295737004e-08, "loss": 0.048, "step": 6157 }, { "epoch": 0.82, "learning_rate": 1.773352933315515e-08, "loss": 0.062, "step": 6158 }, { "epoch": 0.82, "learning_rate": 1.772016570894026e-08, "loss": 0.056, "step": 6159 }, { "epoch": 0.82, "learning_rate": 1.7706802084725378e-08, "loss": 0.0618, "step": 6160 }, { "epoch": 0.82, "learning_rate": 1.769343846051049e-08, "loss": 0.0692, "step": 6161 }, { "epoch": 0.82, "learning_rate": 1.7680074836295604e-08, "loss": 0.0708, "step": 6162 }, { "epoch": 0.82, "learning_rate": 1.7666711212080714e-08, "loss": 0.0572, "step": 6163 }, { "epoch": 0.82, "learning_rate": 1.7653347587865827e-08, "loss": 0.0564, "step": 6164 }, { "epoch": 0.82, "learning_rate": 1.7639983963650944e-08, "loss": 0.0579, "step": 6165 }, { "epoch": 0.82, "learning_rate": 1.7626620339436053e-08, "loss": 0.0442, "step": 6166 }, { "epoch": 0.82, "learning_rate": 1.7613256715221167e-08, "loss": 0.0528, "step": 6167 }, { "epoch": 0.82, "learning_rate": 1.759989309100628e-08, "loss": 0.0459, "step": 6168 }, { "epoch": 0.82, "learning_rate": 1.7586529466791393e-08, "loss": 0.0748, "step": 6169 }, { "epoch": 0.82, "learning_rate": 1.7573165842576506e-08, "loss": 0.0531, "step": 6170 }, { "epoch": 0.82, "learning_rate": 1.755980221836162e-08, "loss": 0.0531, "step": 6171 }, { "epoch": 0.82, "learning_rate": 1.7546438594146732e-08, "loss": 0.0741, "step": 6172 }, { "epoch": 0.82, "learning_rate": 1.7533074969931842e-08, "loss": 0.057, "step": 6173 }, { "epoch": 0.83, "learning_rate": 1.751971134571696e-08, "loss": 0.0736, "step": 6174 }, { "epoch": 0.83, "learning_rate": 1.7506347721502072e-08, "loss": 0.076, "step": 6175 }, { "epoch": 0.83, "learning_rate": 1.7492984097287185e-08, "loss": 0.0588, "step": 6176 }, { "epoch": 0.83, "learning_rate": 1.7479620473072295e-08, "loss": 0.0862, "step": 6177 }, { "epoch": 0.83, "learning_rate": 1.7466256848857408e-08, "loss": 0.0425, "step": 6178 }, { "epoch": 0.83, "learning_rate": 1.7452893224642524e-08, "loss": 0.0471, "step": 6179 }, { "epoch": 0.83, "learning_rate": 1.7439529600427634e-08, "loss": 0.0602, "step": 6180 }, { "epoch": 0.83, "learning_rate": 1.7426165976212747e-08, "loss": 0.0374, "step": 6181 }, { "epoch": 0.83, "learning_rate": 1.741280235199786e-08, "loss": 0.0662, "step": 6182 }, { "epoch": 0.83, "learning_rate": 1.7399438727782974e-08, "loss": 0.0669, "step": 6183 }, { "epoch": 0.83, "learning_rate": 1.7386075103568087e-08, "loss": 0.0627, "step": 6184 }, { "epoch": 0.83, "learning_rate": 1.73727114793532e-08, "loss": 0.053, "step": 6185 }, { "epoch": 0.83, "learning_rate": 1.7359347855138313e-08, "loss": 0.0469, "step": 6186 }, { "epoch": 0.83, "learning_rate": 1.7345984230923423e-08, "loss": 0.0701, "step": 6187 }, { "epoch": 0.83, "learning_rate": 1.733262060670854e-08, "loss": 0.0746, "step": 6188 }, { "epoch": 0.83, "learning_rate": 1.7319256982493652e-08, "loss": 0.064, "step": 6189 }, { "epoch": 0.83, "learning_rate": 1.7305893358278766e-08, "loss": 0.0667, "step": 6190 }, { "epoch": 0.83, "learning_rate": 1.7292529734063875e-08, "loss": 0.0639, "step": 6191 }, { "epoch": 0.83, "learning_rate": 1.7279166109848992e-08, "loss": 0.0581, "step": 6192 }, { "epoch": 0.83, "learning_rate": 1.7265802485634105e-08, "loss": 0.0681, "step": 6193 }, { "epoch": 0.83, "learning_rate": 1.7252438861419215e-08, "loss": 0.0982, "step": 6194 }, { "epoch": 0.83, "learning_rate": 1.7239075237204328e-08, "loss": 0.0578, "step": 6195 }, { "epoch": 0.83, "learning_rate": 1.722571161298944e-08, "loss": 0.048, "step": 6196 }, { "epoch": 0.83, "learning_rate": 1.7212347988774558e-08, "loss": 0.0554, "step": 6197 }, { "epoch": 0.83, "learning_rate": 1.7198984364559667e-08, "loss": 0.053, "step": 6198 }, { "epoch": 0.83, "learning_rate": 1.718562074034478e-08, "loss": 0.0708, "step": 6199 }, { "epoch": 0.83, "learning_rate": 1.7172257116129894e-08, "loss": 0.0513, "step": 6200 }, { "epoch": 0.83, "learning_rate": 1.7158893491915007e-08, "loss": 0.0618, "step": 6201 }, { "epoch": 0.83, "learning_rate": 1.714552986770012e-08, "loss": 0.0567, "step": 6202 }, { "epoch": 0.83, "learning_rate": 1.7132166243485233e-08, "loss": 0.081, "step": 6203 }, { "epoch": 0.83, "learning_rate": 1.7118802619270346e-08, "loss": 0.0858, "step": 6204 }, { "epoch": 0.83, "learning_rate": 1.7105438995055456e-08, "loss": 0.0739, "step": 6205 }, { "epoch": 0.83, "learning_rate": 1.7092075370840572e-08, "loss": 0.0752, "step": 6206 }, { "epoch": 0.83, "learning_rate": 1.7078711746625686e-08, "loss": 0.0736, "step": 6207 }, { "epoch": 0.83, "learning_rate": 1.7065348122410795e-08, "loss": 0.0419, "step": 6208 }, { "epoch": 0.83, "learning_rate": 1.705198449819591e-08, "loss": 0.0748, "step": 6209 }, { "epoch": 0.83, "learning_rate": 1.7038620873981022e-08, "loss": 0.0529, "step": 6210 }, { "epoch": 0.83, "learning_rate": 1.7025257249766138e-08, "loss": 0.0921, "step": 6211 }, { "epoch": 0.83, "learning_rate": 1.7011893625551248e-08, "loss": 0.0711, "step": 6212 }, { "epoch": 0.83, "learning_rate": 1.699853000133636e-08, "loss": 0.089, "step": 6213 }, { "epoch": 0.83, "learning_rate": 1.6985166377121474e-08, "loss": 0.0629, "step": 6214 }, { "epoch": 0.83, "learning_rate": 1.6971802752906587e-08, "loss": 0.0569, "step": 6215 }, { "epoch": 0.83, "learning_rate": 1.69584391286917e-08, "loss": 0.0659, "step": 6216 }, { "epoch": 0.83, "learning_rate": 1.6945075504476814e-08, "loss": 0.0751, "step": 6217 }, { "epoch": 0.83, "learning_rate": 1.6931711880261927e-08, "loss": 0.1202, "step": 6218 }, { "epoch": 0.83, "learning_rate": 1.6918348256047037e-08, "loss": 0.0636, "step": 6219 }, { "epoch": 0.83, "learning_rate": 1.6904984631832153e-08, "loss": 0.1036, "step": 6220 }, { "epoch": 0.83, "learning_rate": 1.6891621007617266e-08, "loss": 0.0779, "step": 6221 }, { "epoch": 0.83, "learning_rate": 1.687825738340238e-08, "loss": 0.0503, "step": 6222 }, { "epoch": 0.83, "learning_rate": 1.686489375918749e-08, "loss": 0.0691, "step": 6223 }, { "epoch": 0.83, "learning_rate": 1.6851530134972602e-08, "loss": 0.0707, "step": 6224 }, { "epoch": 0.83, "learning_rate": 1.683816651075772e-08, "loss": 0.0735, "step": 6225 }, { "epoch": 0.83, "learning_rate": 1.682480288654283e-08, "loss": 0.0497, "step": 6226 }, { "epoch": 0.83, "learning_rate": 1.6811439262327942e-08, "loss": 0.0556, "step": 6227 }, { "epoch": 0.83, "learning_rate": 1.6798075638113055e-08, "loss": 0.0748, "step": 6228 }, { "epoch": 0.83, "learning_rate": 1.6784712013898168e-08, "loss": 0.0554, "step": 6229 }, { "epoch": 0.83, "learning_rate": 1.677134838968328e-08, "loss": 0.0571, "step": 6230 }, { "epoch": 0.83, "learning_rate": 1.6757984765468394e-08, "loss": 0.0969, "step": 6231 }, { "epoch": 0.83, "learning_rate": 1.6744621141253508e-08, "loss": 0.0597, "step": 6232 }, { "epoch": 0.83, "learning_rate": 1.6731257517038617e-08, "loss": 0.0625, "step": 6233 }, { "epoch": 0.83, "learning_rate": 1.6717893892823734e-08, "loss": 0.0543, "step": 6234 }, { "epoch": 0.83, "learning_rate": 1.6704530268608847e-08, "loss": 0.0836, "step": 6235 }, { "epoch": 0.83, "learning_rate": 1.669116664439396e-08, "loss": 0.0868, "step": 6236 }, { "epoch": 0.83, "learning_rate": 1.667780302017907e-08, "loss": 0.0689, "step": 6237 }, { "epoch": 0.83, "learning_rate": 1.6664439395964186e-08, "loss": 0.0736, "step": 6238 }, { "epoch": 0.83, "learning_rate": 1.66510757717493e-08, "loss": 0.0911, "step": 6239 }, { "epoch": 0.83, "learning_rate": 1.663771214753441e-08, "loss": 0.0616, "step": 6240 }, { "epoch": 0.83, "learning_rate": 1.6624348523319523e-08, "loss": 0.0823, "step": 6241 }, { "epoch": 0.83, "learning_rate": 1.6610984899104636e-08, "loss": 0.0863, "step": 6242 }, { "epoch": 0.83, "learning_rate": 1.659762127488975e-08, "loss": 0.0881, "step": 6243 }, { "epoch": 0.83, "learning_rate": 1.6584257650674862e-08, "loss": 0.0412, "step": 6244 }, { "epoch": 0.83, "learning_rate": 1.6570894026459975e-08, "loss": 0.07, "step": 6245 }, { "epoch": 0.83, "learning_rate": 1.6557530402245088e-08, "loss": 0.0611, "step": 6246 }, { "epoch": 0.83, "learning_rate": 1.6544166778030198e-08, "loss": 0.0536, "step": 6247 }, { "epoch": 0.83, "learning_rate": 1.6530803153815315e-08, "loss": 0.0518, "step": 6248 }, { "epoch": 0.84, "learning_rate": 1.6517439529600428e-08, "loss": 0.0845, "step": 6249 }, { "epoch": 0.84, "learning_rate": 1.650407590538554e-08, "loss": 0.0508, "step": 6250 }, { "epoch": 0.84, "learning_rate": 1.649071228117065e-08, "loss": 0.0773, "step": 6251 }, { "epoch": 0.84, "learning_rate": 1.6477348656955767e-08, "loss": 0.0595, "step": 6252 }, { "epoch": 0.84, "learning_rate": 1.646398503274088e-08, "loss": 0.0866, "step": 6253 }, { "epoch": 0.84, "learning_rate": 1.645062140852599e-08, "loss": 0.048, "step": 6254 }, { "epoch": 0.84, "learning_rate": 1.6437257784311103e-08, "loss": 0.0646, "step": 6255 }, { "epoch": 0.84, "learning_rate": 1.6423894160096216e-08, "loss": 0.0793, "step": 6256 }, { "epoch": 0.84, "learning_rate": 1.6410530535881333e-08, "loss": 0.0472, "step": 6257 }, { "epoch": 0.84, "learning_rate": 1.6397166911666443e-08, "loss": 0.079, "step": 6258 }, { "epoch": 0.84, "learning_rate": 1.6383803287451556e-08, "loss": 0.0461, "step": 6259 }, { "epoch": 0.84, "learning_rate": 1.637043966323667e-08, "loss": 0.0563, "step": 6260 }, { "epoch": 0.84, "learning_rate": 1.6357076039021782e-08, "loss": 0.0591, "step": 6261 }, { "epoch": 0.84, "learning_rate": 1.6343712414806895e-08, "loss": 0.0633, "step": 6262 }, { "epoch": 0.84, "learning_rate": 1.633034879059201e-08, "loss": 0.0652, "step": 6263 }, { "epoch": 0.84, "learning_rate": 1.631698516637712e-08, "loss": 0.0595, "step": 6264 }, { "epoch": 0.84, "learning_rate": 1.630362154216223e-08, "loss": 0.0838, "step": 6265 }, { "epoch": 0.84, "learning_rate": 1.6290257917947348e-08, "loss": 0.062, "step": 6266 }, { "epoch": 0.84, "learning_rate": 1.627689429373246e-08, "loss": 0.0648, "step": 6267 }, { "epoch": 0.84, "learning_rate": 1.626353066951757e-08, "loss": 0.0456, "step": 6268 }, { "epoch": 0.84, "learning_rate": 1.6250167045302684e-08, "loss": 0.054, "step": 6269 }, { "epoch": 0.84, "learning_rate": 1.62368034210878e-08, "loss": 0.0793, "step": 6270 }, { "epoch": 0.84, "learning_rate": 1.6223439796872913e-08, "loss": 0.0742, "step": 6271 }, { "epoch": 0.84, "learning_rate": 1.6210076172658023e-08, "loss": 0.0497, "step": 6272 }, { "epoch": 0.84, "learning_rate": 1.6196712548443136e-08, "loss": 0.0623, "step": 6273 }, { "epoch": 0.84, "learning_rate": 1.618334892422825e-08, "loss": 0.0764, "step": 6274 }, { "epoch": 0.84, "learning_rate": 1.6169985300013363e-08, "loss": 0.0714, "step": 6275 }, { "epoch": 0.84, "learning_rate": 1.6156621675798476e-08, "loss": 0.0692, "step": 6276 }, { "epoch": 0.84, "learning_rate": 1.614325805158359e-08, "loss": 0.0878, "step": 6277 }, { "epoch": 0.84, "learning_rate": 1.6129894427368702e-08, "loss": 0.0946, "step": 6278 }, { "epoch": 0.84, "learning_rate": 1.6116530803153812e-08, "loss": 0.0431, "step": 6279 }, { "epoch": 0.84, "learning_rate": 1.610316717893893e-08, "loss": 0.0525, "step": 6280 }, { "epoch": 0.84, "learning_rate": 1.608980355472404e-08, "loss": 0.1124, "step": 6281 }, { "epoch": 0.84, "learning_rate": 1.607643993050915e-08, "loss": 0.0593, "step": 6282 }, { "epoch": 0.84, "learning_rate": 1.6063076306294265e-08, "loss": 0.0666, "step": 6283 }, { "epoch": 0.84, "learning_rate": 1.604971268207938e-08, "loss": 0.055, "step": 6284 }, { "epoch": 0.84, "learning_rate": 1.6036349057864494e-08, "loss": 0.0725, "step": 6285 }, { "epoch": 0.84, "learning_rate": 1.6022985433649604e-08, "loss": 0.047, "step": 6286 }, { "epoch": 0.84, "learning_rate": 1.6009621809434717e-08, "loss": 0.0967, "step": 6287 }, { "epoch": 0.84, "learning_rate": 1.599625818521983e-08, "loss": 0.0823, "step": 6288 }, { "epoch": 0.84, "learning_rate": 1.5982894561004943e-08, "loss": 0.0763, "step": 6289 }, { "epoch": 0.84, "learning_rate": 1.5969530936790057e-08, "loss": 0.0519, "step": 6290 }, { "epoch": 0.84, "learning_rate": 1.595616731257517e-08, "loss": 0.0671, "step": 6291 }, { "epoch": 0.84, "learning_rate": 1.5942803688360283e-08, "loss": 0.0529, "step": 6292 }, { "epoch": 0.84, "learning_rate": 1.5929440064145396e-08, "loss": 0.076, "step": 6293 }, { "epoch": 0.84, "learning_rate": 1.591607643993051e-08, "loss": 0.0676, "step": 6294 }, { "epoch": 0.84, "learning_rate": 1.5902712815715622e-08, "loss": 0.0548, "step": 6295 }, { "epoch": 0.84, "learning_rate": 1.5889349191500735e-08, "loss": 0.0824, "step": 6296 }, { "epoch": 0.84, "learning_rate": 1.5875985567285845e-08, "loss": 0.0702, "step": 6297 }, { "epoch": 0.84, "learning_rate": 1.5862621943070962e-08, "loss": 0.0699, "step": 6298 }, { "epoch": 0.84, "learning_rate": 1.5849258318856075e-08, "loss": 0.0675, "step": 6299 }, { "epoch": 0.84, "learning_rate": 1.5835894694641185e-08, "loss": 0.0531, "step": 6300 }, { "epoch": 0.84, "learning_rate": 1.5822531070426298e-08, "loss": 0.0639, "step": 6301 }, { "epoch": 0.84, "learning_rate": 1.580916744621141e-08, "loss": 0.0628, "step": 6302 }, { "epoch": 0.84, "learning_rate": 1.5795803821996524e-08, "loss": 0.0783, "step": 6303 }, { "epoch": 0.84, "learning_rate": 1.5782440197781637e-08, "loss": 0.1141, "step": 6304 }, { "epoch": 0.84, "learning_rate": 1.576907657356675e-08, "loss": 0.0586, "step": 6305 }, { "epoch": 0.84, "learning_rate": 1.5755712949351864e-08, "loss": 0.0881, "step": 6306 }, { "epoch": 0.84, "learning_rate": 1.5742349325136977e-08, "loss": 0.0759, "step": 6307 }, { "epoch": 0.84, "learning_rate": 1.572898570092209e-08, "loss": 0.0623, "step": 6308 }, { "epoch": 0.84, "learning_rate": 1.5715622076707203e-08, "loss": 0.0532, "step": 6309 }, { "epoch": 0.84, "learning_rate": 1.5702258452492316e-08, "loss": 0.0682, "step": 6310 }, { "epoch": 0.84, "learning_rate": 1.5688894828277426e-08, "loss": 0.0596, "step": 6311 }, { "epoch": 0.84, "learning_rate": 1.5675531204062542e-08, "loss": 0.0669, "step": 6312 }, { "epoch": 0.84, "learning_rate": 1.5662167579847656e-08, "loss": 0.0454, "step": 6313 }, { "epoch": 0.84, "learning_rate": 1.5648803955632765e-08, "loss": 0.0486, "step": 6314 }, { "epoch": 0.84, "learning_rate": 1.563544033141788e-08, "loss": 0.0796, "step": 6315 }, { "epoch": 0.84, "learning_rate": 1.5622076707202995e-08, "loss": 0.0578, "step": 6316 }, { "epoch": 0.84, "learning_rate": 1.5608713082988105e-08, "loss": 0.0682, "step": 6317 }, { "epoch": 0.84, "learning_rate": 1.5595349458773218e-08, "loss": 0.0765, "step": 6318 }, { "epoch": 0.84, "learning_rate": 1.558198583455833e-08, "loss": 0.0538, "step": 6319 }, { "epoch": 0.84, "learning_rate": 1.5568622210343444e-08, "loss": 0.0855, "step": 6320 }, { "epoch": 0.84, "learning_rate": 1.5555258586128557e-08, "loss": 0.0665, "step": 6321 }, { "epoch": 0.84, "learning_rate": 1.554189496191367e-08, "loss": 0.0602, "step": 6322 }, { "epoch": 0.84, "learning_rate": 1.5528531337698784e-08, "loss": 0.0479, "step": 6323 }, { "epoch": 0.85, "learning_rate": 1.5515167713483897e-08, "loss": 0.0599, "step": 6324 }, { "epoch": 0.85, "learning_rate": 1.5501804089269007e-08, "loss": 0.0564, "step": 6325 }, { "epoch": 0.85, "learning_rate": 1.5488440465054123e-08, "loss": 0.1108, "step": 6326 }, { "epoch": 0.85, "learning_rate": 1.5475076840839236e-08, "loss": 0.0455, "step": 6327 }, { "epoch": 0.85, "learning_rate": 1.5461713216624346e-08, "loss": 0.0612, "step": 6328 }, { "epoch": 0.85, "learning_rate": 1.544834959240946e-08, "loss": 0.0942, "step": 6329 }, { "epoch": 0.85, "learning_rate": 1.5434985968194576e-08, "loss": 0.0802, "step": 6330 }, { "epoch": 0.85, "learning_rate": 1.542162234397969e-08, "loss": 0.0935, "step": 6331 }, { "epoch": 0.85, "learning_rate": 1.54082587197648e-08, "loss": 0.0501, "step": 6332 }, { "epoch": 0.85, "learning_rate": 1.5394895095549912e-08, "loss": 0.0648, "step": 6333 }, { "epoch": 0.85, "learning_rate": 1.5381531471335025e-08, "loss": 0.0774, "step": 6334 }, { "epoch": 0.85, "learning_rate": 1.5368167847120138e-08, "loss": 0.0603, "step": 6335 }, { "epoch": 0.85, "learning_rate": 1.535480422290525e-08, "loss": 0.0601, "step": 6336 }, { "epoch": 0.85, "learning_rate": 1.5341440598690364e-08, "loss": 0.0605, "step": 6337 }, { "epoch": 0.85, "learning_rate": 1.5328076974475477e-08, "loss": 0.1002, "step": 6338 }, { "epoch": 0.85, "learning_rate": 1.531471335026059e-08, "loss": 0.0749, "step": 6339 }, { "epoch": 0.85, "learning_rate": 1.5301349726045704e-08, "loss": 0.068, "step": 6340 }, { "epoch": 0.85, "learning_rate": 1.5287986101830817e-08, "loss": 0.0708, "step": 6341 }, { "epoch": 0.85, "learning_rate": 1.5274622477615927e-08, "loss": 0.0574, "step": 6342 }, { "epoch": 0.85, "learning_rate": 1.526125885340104e-08, "loss": 0.0497, "step": 6343 }, { "epoch": 0.85, "learning_rate": 1.5247895229186156e-08, "loss": 0.0883, "step": 6344 }, { "epoch": 0.85, "learning_rate": 1.523453160497127e-08, "loss": 0.0382, "step": 6345 }, { "epoch": 0.85, "learning_rate": 1.522116798075638e-08, "loss": 0.0404, "step": 6346 }, { "epoch": 0.85, "learning_rate": 1.5207804356541492e-08, "loss": 0.077, "step": 6347 }, { "epoch": 0.85, "learning_rate": 1.5194440732326606e-08, "loss": 0.033, "step": 6348 }, { "epoch": 0.85, "learning_rate": 1.518107710811172e-08, "loss": 0.0606, "step": 6349 }, { "epoch": 0.85, "learning_rate": 1.5167713483896832e-08, "loss": 0.0483, "step": 6350 }, { "epoch": 0.85, "learning_rate": 1.5154349859681945e-08, "loss": 0.0771, "step": 6351 }, { "epoch": 0.85, "learning_rate": 1.5140986235467058e-08, "loss": 0.0693, "step": 6352 }, { "epoch": 0.85, "learning_rate": 1.512762261125217e-08, "loss": 0.0698, "step": 6353 }, { "epoch": 0.85, "learning_rate": 1.5114258987037284e-08, "loss": 0.0461, "step": 6354 }, { "epoch": 0.85, "learning_rate": 1.5100895362822398e-08, "loss": 0.1232, "step": 6355 }, { "epoch": 0.85, "learning_rate": 1.5087531738607507e-08, "loss": 0.0524, "step": 6356 }, { "epoch": 0.85, "learning_rate": 1.507416811439262e-08, "loss": 0.0714, "step": 6357 }, { "epoch": 0.85, "learning_rate": 1.5060804490177737e-08, "loss": 0.0582, "step": 6358 }, { "epoch": 0.85, "learning_rate": 1.504744086596285e-08, "loss": 0.0709, "step": 6359 }, { "epoch": 0.85, "learning_rate": 1.503407724174796e-08, "loss": 0.07, "step": 6360 }, { "epoch": 0.85, "learning_rate": 1.5020713617533073e-08, "loss": 0.0505, "step": 6361 }, { "epoch": 0.85, "learning_rate": 1.500734999331819e-08, "loss": 0.0673, "step": 6362 }, { "epoch": 0.85, "learning_rate": 1.49939863691033e-08, "loss": 0.0657, "step": 6363 }, { "epoch": 0.85, "learning_rate": 1.4980622744888413e-08, "loss": 0.0679, "step": 6364 }, { "epoch": 0.85, "learning_rate": 1.4967259120673526e-08, "loss": 0.0612, "step": 6365 }, { "epoch": 0.85, "learning_rate": 1.495389549645864e-08, "loss": 0.0768, "step": 6366 }, { "epoch": 0.85, "learning_rate": 1.4940531872243752e-08, "loss": 0.0699, "step": 6367 }, { "epoch": 0.85, "learning_rate": 1.4927168248028865e-08, "loss": 0.0777, "step": 6368 }, { "epoch": 0.85, "learning_rate": 1.4913804623813978e-08, "loss": 0.0736, "step": 6369 }, { "epoch": 0.85, "learning_rate": 1.490044099959909e-08, "loss": 0.0682, "step": 6370 }, { "epoch": 0.85, "learning_rate": 1.4887077375384203e-08, "loss": 0.0524, "step": 6371 }, { "epoch": 0.85, "learning_rate": 1.4873713751169318e-08, "loss": 0.0671, "step": 6372 }, { "epoch": 0.85, "learning_rate": 1.486035012695443e-08, "loss": 0.0645, "step": 6373 }, { "epoch": 0.85, "learning_rate": 1.4846986502739542e-08, "loss": 0.0706, "step": 6374 }, { "epoch": 0.85, "learning_rate": 1.4833622878524654e-08, "loss": 0.0441, "step": 6375 }, { "epoch": 0.85, "learning_rate": 1.4820259254309769e-08, "loss": 0.0744, "step": 6376 }, { "epoch": 0.85, "learning_rate": 1.4806895630094882e-08, "loss": 0.0723, "step": 6377 }, { "epoch": 0.85, "learning_rate": 1.4793532005879993e-08, "loss": 0.0776, "step": 6378 }, { "epoch": 0.85, "learning_rate": 1.4780168381665106e-08, "loss": 0.0457, "step": 6379 }, { "epoch": 0.85, "learning_rate": 1.4766804757450218e-08, "loss": 0.1013, "step": 6380 }, { "epoch": 0.85, "learning_rate": 1.4753441133235334e-08, "loss": 0.0771, "step": 6381 }, { "epoch": 0.85, "learning_rate": 1.4740077509020446e-08, "loss": 0.0428, "step": 6382 }, { "epoch": 0.85, "learning_rate": 1.4726713884805559e-08, "loss": 0.0657, "step": 6383 }, { "epoch": 0.85, "learning_rate": 1.471335026059067e-08, "loss": 0.0597, "step": 6384 }, { "epoch": 0.85, "learning_rate": 1.4699986636375785e-08, "loss": 0.0426, "step": 6385 }, { "epoch": 0.85, "learning_rate": 1.4686623012160898e-08, "loss": 0.0542, "step": 6386 }, { "epoch": 0.85, "learning_rate": 1.467325938794601e-08, "loss": 0.0762, "step": 6387 }, { "epoch": 0.85, "learning_rate": 1.4659895763731123e-08, "loss": 0.0619, "step": 6388 }, { "epoch": 0.85, "learning_rate": 1.4646532139516234e-08, "loss": 0.0599, "step": 6389 }, { "epoch": 0.85, "learning_rate": 1.463316851530135e-08, "loss": 0.0508, "step": 6390 }, { "epoch": 0.85, "learning_rate": 1.4619804891086462e-08, "loss": 0.1016, "step": 6391 }, { "epoch": 0.85, "learning_rate": 1.4606441266871576e-08, "loss": 0.0617, "step": 6392 }, { "epoch": 0.85, "learning_rate": 1.4593077642656687e-08, "loss": 0.052, "step": 6393 }, { "epoch": 0.85, "learning_rate": 1.45797140184418e-08, "loss": 0.052, "step": 6394 }, { "epoch": 0.85, "learning_rate": 1.4566350394226915e-08, "loss": 0.0818, "step": 6395 }, { "epoch": 0.85, "learning_rate": 1.4552986770012026e-08, "loss": 0.0574, "step": 6396 }, { "epoch": 0.85, "learning_rate": 1.453962314579714e-08, "loss": 0.0438, "step": 6397 }, { "epoch": 0.86, "learning_rate": 1.4526259521582251e-08, "loss": 0.0678, "step": 6398 }, { "epoch": 0.86, "learning_rate": 1.4512895897367366e-08, "loss": 0.0464, "step": 6399 }, { "epoch": 0.86, "learning_rate": 1.4499532273152479e-08, "loss": 0.0877, "step": 6400 }, { "epoch": 0.86, "learning_rate": 1.448616864893759e-08, "loss": 0.0687, "step": 6401 }, { "epoch": 0.86, "learning_rate": 1.4472805024722704e-08, "loss": 0.0623, "step": 6402 }, { "epoch": 0.86, "learning_rate": 1.4459441400507815e-08, "loss": 0.0539, "step": 6403 }, { "epoch": 0.86, "learning_rate": 1.4446077776292932e-08, "loss": 0.0882, "step": 6404 }, { "epoch": 0.86, "learning_rate": 1.4432714152078043e-08, "loss": 0.0705, "step": 6405 }, { "epoch": 0.86, "learning_rate": 1.4419350527863156e-08, "loss": 0.0787, "step": 6406 }, { "epoch": 0.86, "learning_rate": 1.4405986903648268e-08, "loss": 0.0359, "step": 6407 }, { "epoch": 0.86, "learning_rate": 1.4392623279433383e-08, "loss": 0.0682, "step": 6408 }, { "epoch": 0.86, "learning_rate": 1.4379259655218496e-08, "loss": 0.0648, "step": 6409 }, { "epoch": 0.86, "learning_rate": 1.4365896031003607e-08, "loss": 0.0683, "step": 6410 }, { "epoch": 0.86, "learning_rate": 1.435253240678872e-08, "loss": 0.076, "step": 6411 }, { "epoch": 0.86, "learning_rate": 1.4339168782573832e-08, "loss": 0.0714, "step": 6412 }, { "epoch": 0.86, "learning_rate": 1.4325805158358947e-08, "loss": 0.0719, "step": 6413 }, { "epoch": 0.86, "learning_rate": 1.431244153414406e-08, "loss": 0.0674, "step": 6414 }, { "epoch": 0.86, "learning_rate": 1.4299077909929171e-08, "loss": 0.0339, "step": 6415 }, { "epoch": 0.86, "learning_rate": 1.4285714285714284e-08, "loss": 0.0856, "step": 6416 }, { "epoch": 0.86, "learning_rate": 1.4272350661499399e-08, "loss": 0.0658, "step": 6417 }, { "epoch": 0.86, "learning_rate": 1.4258987037284512e-08, "loss": 0.0416, "step": 6418 }, { "epoch": 0.86, "learning_rate": 1.4245623413069624e-08, "loss": 0.038, "step": 6419 }, { "epoch": 0.86, "learning_rate": 1.4232259788854737e-08, "loss": 0.0551, "step": 6420 }, { "epoch": 0.86, "learning_rate": 1.4218896164639848e-08, "loss": 0.0663, "step": 6421 }, { "epoch": 0.86, "learning_rate": 1.4205532540424963e-08, "loss": 0.0754, "step": 6422 }, { "epoch": 0.86, "learning_rate": 1.4192168916210076e-08, "loss": 0.0633, "step": 6423 }, { "epoch": 0.86, "learning_rate": 1.4178805291995188e-08, "loss": 0.0633, "step": 6424 }, { "epoch": 0.86, "learning_rate": 1.4165441667780301e-08, "loss": 0.0835, "step": 6425 }, { "epoch": 0.86, "learning_rate": 1.4152078043565412e-08, "loss": 0.0583, "step": 6426 }, { "epoch": 0.86, "learning_rate": 1.4138714419350529e-08, "loss": 0.0754, "step": 6427 }, { "epoch": 0.86, "learning_rate": 1.412535079513564e-08, "loss": 0.0548, "step": 6428 }, { "epoch": 0.86, "learning_rate": 1.4111987170920754e-08, "loss": 0.0525, "step": 6429 }, { "epoch": 0.86, "learning_rate": 1.4098623546705865e-08, "loss": 0.0443, "step": 6430 }, { "epoch": 0.86, "learning_rate": 1.408525992249098e-08, "loss": 0.0653, "step": 6431 }, { "epoch": 0.86, "learning_rate": 1.4071896298276093e-08, "loss": 0.0831, "step": 6432 }, { "epoch": 0.86, "learning_rate": 1.4058532674061204e-08, "loss": 0.0657, "step": 6433 }, { "epoch": 0.86, "learning_rate": 1.4045169049846318e-08, "loss": 0.0561, "step": 6434 }, { "epoch": 0.86, "learning_rate": 1.4031805425631429e-08, "loss": 0.0521, "step": 6435 }, { "epoch": 0.86, "learning_rate": 1.4018441801416544e-08, "loss": 0.0611, "step": 6436 }, { "epoch": 0.86, "learning_rate": 1.4005078177201657e-08, "loss": 0.0586, "step": 6437 }, { "epoch": 0.86, "learning_rate": 1.3991714552986769e-08, "loss": 0.0861, "step": 6438 }, { "epoch": 0.86, "learning_rate": 1.3978350928771882e-08, "loss": 0.0405, "step": 6439 }, { "epoch": 0.86, "learning_rate": 1.3964987304556996e-08, "loss": 0.0652, "step": 6440 }, { "epoch": 0.86, "learning_rate": 1.395162368034211e-08, "loss": 0.0728, "step": 6441 }, { "epoch": 0.86, "learning_rate": 1.3938260056127221e-08, "loss": 0.0559, "step": 6442 }, { "epoch": 0.86, "learning_rate": 1.3924896431912334e-08, "loss": 0.042, "step": 6443 }, { "epoch": 0.86, "learning_rate": 1.3911532807697446e-08, "loss": 0.096, "step": 6444 }, { "epoch": 0.86, "learning_rate": 1.389816918348256e-08, "loss": 0.0367, "step": 6445 }, { "epoch": 0.86, "learning_rate": 1.3884805559267674e-08, "loss": 0.0762, "step": 6446 }, { "epoch": 0.86, "learning_rate": 1.3871441935052785e-08, "loss": 0.0597, "step": 6447 }, { "epoch": 0.86, "learning_rate": 1.3858078310837898e-08, "loss": 0.059, "step": 6448 }, { "epoch": 0.86, "learning_rate": 1.384471468662301e-08, "loss": 0.0627, "step": 6449 }, { "epoch": 0.86, "learning_rate": 1.3831351062408125e-08, "loss": 0.0689, "step": 6450 }, { "epoch": 0.86, "learning_rate": 1.3817987438193238e-08, "loss": 0.063, "step": 6451 }, { "epoch": 0.86, "learning_rate": 1.380462381397835e-08, "loss": 0.0931, "step": 6452 }, { "epoch": 0.86, "learning_rate": 1.3791260189763462e-08, "loss": 0.0376, "step": 6453 }, { "epoch": 0.86, "learning_rate": 1.3777896565548577e-08, "loss": 0.0914, "step": 6454 }, { "epoch": 0.86, "learning_rate": 1.376453294133369e-08, "loss": 0.051, "step": 6455 }, { "epoch": 0.86, "learning_rate": 1.3751169317118802e-08, "loss": 0.0842, "step": 6456 }, { "epoch": 0.86, "learning_rate": 1.3737805692903915e-08, "loss": 0.0588, "step": 6457 }, { "epoch": 0.86, "learning_rate": 1.3724442068689026e-08, "loss": 0.0553, "step": 6458 }, { "epoch": 0.86, "learning_rate": 1.3711078444474141e-08, "loss": 0.0404, "step": 6459 }, { "epoch": 0.86, "learning_rate": 1.3697714820259254e-08, "loss": 0.0746, "step": 6460 }, { "epoch": 0.86, "learning_rate": 1.3684351196044366e-08, "loss": 0.087, "step": 6461 }, { "epoch": 0.86, "learning_rate": 1.3670987571829479e-08, "loss": 0.0585, "step": 6462 }, { "epoch": 0.86, "learning_rate": 1.3657623947614594e-08, "loss": 0.0487, "step": 6463 }, { "epoch": 0.86, "learning_rate": 1.3644260323399707e-08, "loss": 0.064, "step": 6464 }, { "epoch": 0.86, "learning_rate": 1.3630896699184818e-08, "loss": 0.0728, "step": 6465 }, { "epoch": 0.86, "learning_rate": 1.3617533074969932e-08, "loss": 0.0601, "step": 6466 }, { "epoch": 0.86, "learning_rate": 1.3604169450755043e-08, "loss": 0.0482, "step": 6467 }, { "epoch": 0.86, "learning_rate": 1.3590805826540158e-08, "loss": 0.0531, "step": 6468 }, { "epoch": 0.86, "learning_rate": 1.3577442202325271e-08, "loss": 0.0574, "step": 6469 }, { "epoch": 0.86, "learning_rate": 1.3564078578110382e-08, "loss": 0.0578, "step": 6470 }, { "epoch": 0.86, "learning_rate": 1.3550714953895496e-08, "loss": 0.0752, "step": 6471 }, { "epoch": 0.86, "learning_rate": 1.3537351329680607e-08, "loss": 0.0581, "step": 6472 }, { "epoch": 0.87, "learning_rate": 1.3523987705465722e-08, "loss": 0.0529, "step": 6473 }, { "epoch": 0.87, "learning_rate": 1.3510624081250835e-08, "loss": 0.0468, "step": 6474 }, { "epoch": 0.87, "learning_rate": 1.3497260457035947e-08, "loss": 0.0556, "step": 6475 }, { "epoch": 0.87, "learning_rate": 1.348389683282106e-08, "loss": 0.0388, "step": 6476 }, { "epoch": 0.87, "learning_rate": 1.3470533208606174e-08, "loss": 0.0813, "step": 6477 }, { "epoch": 0.87, "learning_rate": 1.3457169584391288e-08, "loss": 0.0729, "step": 6478 }, { "epoch": 0.87, "learning_rate": 1.3443805960176399e-08, "loss": 0.0725, "step": 6479 }, { "epoch": 0.87, "learning_rate": 1.3430442335961512e-08, "loss": 0.0438, "step": 6480 }, { "epoch": 0.87, "learning_rate": 1.3417078711746624e-08, "loss": 0.0715, "step": 6481 }, { "epoch": 0.87, "learning_rate": 1.3403715087531739e-08, "loss": 0.0862, "step": 6482 }, { "epoch": 0.87, "learning_rate": 1.3390351463316852e-08, "loss": 0.0759, "step": 6483 }, { "epoch": 0.87, "learning_rate": 1.3376987839101963e-08, "loss": 0.0625, "step": 6484 }, { "epoch": 0.87, "learning_rate": 1.3363624214887076e-08, "loss": 0.0485, "step": 6485 }, { "epoch": 0.87, "learning_rate": 1.3350260590672191e-08, "loss": 0.054, "step": 6486 }, { "epoch": 0.87, "learning_rate": 1.3336896966457303e-08, "loss": 0.057, "step": 6487 }, { "epoch": 0.87, "learning_rate": 1.3323533342242416e-08, "loss": 0.0672, "step": 6488 }, { "epoch": 0.87, "learning_rate": 1.3310169718027527e-08, "loss": 0.0709, "step": 6489 }, { "epoch": 0.87, "learning_rate": 1.329680609381264e-08, "loss": 0.0619, "step": 6490 }, { "epoch": 0.87, "learning_rate": 1.3283442469597755e-08, "loss": 0.0433, "step": 6491 }, { "epoch": 0.87, "learning_rate": 1.3270078845382868e-08, "loss": 0.0757, "step": 6492 }, { "epoch": 0.87, "learning_rate": 1.325671522116798e-08, "loss": 0.0503, "step": 6493 }, { "epoch": 0.87, "learning_rate": 1.3243351596953093e-08, "loss": 0.0869, "step": 6494 }, { "epoch": 0.87, "learning_rate": 1.3229987972738204e-08, "loss": 0.0948, "step": 6495 }, { "epoch": 0.87, "learning_rate": 1.321662434852332e-08, "loss": 0.0534, "step": 6496 }, { "epoch": 0.87, "learning_rate": 1.3203260724308432e-08, "loss": 0.0611, "step": 6497 }, { "epoch": 0.87, "learning_rate": 1.3189897100093544e-08, "loss": 0.0733, "step": 6498 }, { "epoch": 0.87, "learning_rate": 1.3176533475878657e-08, "loss": 0.0532, "step": 6499 }, { "epoch": 0.87, "learning_rate": 1.3163169851663772e-08, "loss": 0.0772, "step": 6500 }, { "epoch": 0.87, "learning_rate": 1.3149806227448885e-08, "loss": 0.0521, "step": 6501 }, { "epoch": 0.87, "learning_rate": 1.3136442603233996e-08, "loss": 0.1587, "step": 6502 }, { "epoch": 0.87, "learning_rate": 1.312307897901911e-08, "loss": 0.0552, "step": 6503 }, { "epoch": 0.87, "learning_rate": 1.3109715354804221e-08, "loss": 0.0701, "step": 6504 }, { "epoch": 0.87, "learning_rate": 1.3096351730589336e-08, "loss": 0.07, "step": 6505 }, { "epoch": 0.87, "learning_rate": 1.3082988106374449e-08, "loss": 0.0649, "step": 6506 }, { "epoch": 0.87, "learning_rate": 1.306962448215956e-08, "loss": 0.0807, "step": 6507 }, { "epoch": 0.87, "learning_rate": 1.3056260857944674e-08, "loss": 0.1043, "step": 6508 }, { "epoch": 0.87, "learning_rate": 1.3042897233729788e-08, "loss": 0.0902, "step": 6509 }, { "epoch": 0.87, "learning_rate": 1.30295336095149e-08, "loss": 0.0712, "step": 6510 }, { "epoch": 0.87, "learning_rate": 1.3016169985300013e-08, "loss": 0.0543, "step": 6511 }, { "epoch": 0.87, "learning_rate": 1.3002806361085124e-08, "loss": 0.0473, "step": 6512 }, { "epoch": 0.87, "learning_rate": 1.2989442736870238e-08, "loss": 0.0399, "step": 6513 }, { "epoch": 0.87, "learning_rate": 1.2976079112655352e-08, "loss": 0.0452, "step": 6514 }, { "epoch": 0.87, "learning_rate": 1.2962715488440466e-08, "loss": 0.1004, "step": 6515 }, { "epoch": 0.87, "learning_rate": 1.2949351864225577e-08, "loss": 0.1081, "step": 6516 }, { "epoch": 0.87, "learning_rate": 1.293598824001069e-08, "loss": 0.0615, "step": 6517 }, { "epoch": 0.87, "learning_rate": 1.2922624615795802e-08, "loss": 0.0628, "step": 6518 }, { "epoch": 0.87, "learning_rate": 1.2909260991580916e-08, "loss": 0.0843, "step": 6519 }, { "epoch": 0.87, "learning_rate": 1.289589736736603e-08, "loss": 0.0541, "step": 6520 }, { "epoch": 0.87, "learning_rate": 1.2882533743151141e-08, "loss": 0.0661, "step": 6521 }, { "epoch": 0.87, "learning_rate": 1.2869170118936254e-08, "loss": 0.0763, "step": 6522 }, { "epoch": 0.87, "learning_rate": 1.2855806494721369e-08, "loss": 0.0564, "step": 6523 }, { "epoch": 0.87, "learning_rate": 1.284244287050648e-08, "loss": 0.0577, "step": 6524 }, { "epoch": 0.87, "learning_rate": 1.2829079246291594e-08, "loss": 0.0421, "step": 6525 }, { "epoch": 0.87, "learning_rate": 1.2815715622076705e-08, "loss": 0.0549, "step": 6526 }, { "epoch": 0.87, "learning_rate": 1.2802351997861818e-08, "loss": 0.0615, "step": 6527 }, { "epoch": 0.87, "learning_rate": 1.2788988373646933e-08, "loss": 0.0891, "step": 6528 }, { "epoch": 0.87, "learning_rate": 1.2775624749432046e-08, "loss": 0.0541, "step": 6529 }, { "epoch": 0.87, "learning_rate": 1.2762261125217158e-08, "loss": 0.0512, "step": 6530 }, { "epoch": 0.87, "learning_rate": 1.2748897501002271e-08, "loss": 0.0613, "step": 6531 }, { "epoch": 0.87, "learning_rate": 1.2735533876787386e-08, "loss": 0.0467, "step": 6532 }, { "epoch": 0.87, "learning_rate": 1.2722170252572497e-08, "loss": 0.1059, "step": 6533 }, { "epoch": 0.87, "learning_rate": 1.270880662835761e-08, "loss": 0.0549, "step": 6534 }, { "epoch": 0.87, "learning_rate": 1.2695443004142722e-08, "loss": 0.0455, "step": 6535 }, { "epoch": 0.87, "learning_rate": 1.2682079379927835e-08, "loss": 0.0554, "step": 6536 }, { "epoch": 0.87, "learning_rate": 1.266871575571295e-08, "loss": 0.0584, "step": 6537 }, { "epoch": 0.87, "learning_rate": 1.2655352131498063e-08, "loss": 0.0504, "step": 6538 }, { "epoch": 0.87, "learning_rate": 1.2641988507283174e-08, "loss": 0.0792, "step": 6539 }, { "epoch": 0.87, "learning_rate": 1.2628624883068288e-08, "loss": 0.0503, "step": 6540 }, { "epoch": 0.87, "learning_rate": 1.2615261258853399e-08, "loss": 0.0758, "step": 6541 }, { "epoch": 0.87, "learning_rate": 1.2601897634638514e-08, "loss": 0.0794, "step": 6542 }, { "epoch": 0.87, "learning_rate": 1.2588534010423627e-08, "loss": 0.0599, "step": 6543 }, { "epoch": 0.87, "learning_rate": 1.2575170386208738e-08, "loss": 0.0564, "step": 6544 }, { "epoch": 0.87, "learning_rate": 1.2561806761993852e-08, "loss": 0.0387, "step": 6545 }, { "epoch": 0.87, "learning_rate": 1.2548443137778966e-08, "loss": 0.0731, "step": 6546 }, { "epoch": 0.87, "learning_rate": 1.2535079513564078e-08, "loss": 0.0758, "step": 6547 }, { "epoch": 0.88, "learning_rate": 1.2521715889349191e-08, "loss": 0.0629, "step": 6548 }, { "epoch": 0.88, "learning_rate": 1.2508352265134302e-08, "loss": 0.0842, "step": 6549 }, { "epoch": 0.88, "learning_rate": 1.2494988640919417e-08, "loss": 0.0623, "step": 6550 }, { "epoch": 0.88, "learning_rate": 1.2481625016704529e-08, "loss": 0.0829, "step": 6551 }, { "epoch": 0.88, "learning_rate": 1.2468261392489644e-08, "loss": 0.0625, "step": 6552 }, { "epoch": 0.88, "learning_rate": 1.2454897768274755e-08, "loss": 0.046, "step": 6553 }, { "epoch": 0.88, "learning_rate": 1.2441534144059868e-08, "loss": 0.0503, "step": 6554 }, { "epoch": 0.88, "learning_rate": 1.2428170519844981e-08, "loss": 0.0641, "step": 6555 }, { "epoch": 0.88, "learning_rate": 1.2414806895630094e-08, "loss": 0.067, "step": 6556 }, { "epoch": 0.88, "learning_rate": 1.2401443271415208e-08, "loss": 0.0529, "step": 6557 }, { "epoch": 0.88, "learning_rate": 1.2388079647200319e-08, "loss": 0.0468, "step": 6558 }, { "epoch": 0.88, "learning_rate": 1.2374716022985434e-08, "loss": 0.0725, "step": 6559 }, { "epoch": 0.88, "learning_rate": 1.2361352398770545e-08, "loss": 0.071, "step": 6560 }, { "epoch": 0.88, "learning_rate": 1.2347988774555659e-08, "loss": 0.0481, "step": 6561 }, { "epoch": 0.88, "learning_rate": 1.2334625150340772e-08, "loss": 0.0394, "step": 6562 }, { "epoch": 0.88, "learning_rate": 1.2321261526125885e-08, "loss": 0.0729, "step": 6563 }, { "epoch": 0.88, "learning_rate": 1.2307897901910998e-08, "loss": 0.0531, "step": 6564 }, { "epoch": 0.88, "learning_rate": 1.229453427769611e-08, "loss": 0.0536, "step": 6565 }, { "epoch": 0.88, "learning_rate": 1.2281170653481224e-08, "loss": 0.0739, "step": 6566 }, { "epoch": 0.88, "learning_rate": 1.2267807029266336e-08, "loss": 0.0494, "step": 6567 }, { "epoch": 0.88, "learning_rate": 1.225444340505145e-08, "loss": 0.0536, "step": 6568 }, { "epoch": 0.88, "learning_rate": 1.2241079780836562e-08, "loss": 0.0349, "step": 6569 }, { "epoch": 0.88, "learning_rate": 1.2227716156621675e-08, "loss": 0.0618, "step": 6570 }, { "epoch": 0.88, "learning_rate": 1.2214352532406788e-08, "loss": 0.0619, "step": 6571 }, { "epoch": 0.88, "learning_rate": 1.22009889081919e-08, "loss": 0.0666, "step": 6572 }, { "epoch": 0.88, "learning_rate": 1.2187625283977015e-08, "loss": 0.0709, "step": 6573 }, { "epoch": 0.88, "learning_rate": 1.2174261659762126e-08, "loss": 0.0678, "step": 6574 }, { "epoch": 0.88, "learning_rate": 1.2160898035547241e-08, "loss": 0.0837, "step": 6575 }, { "epoch": 0.88, "learning_rate": 1.2147534411332352e-08, "loss": 0.0695, "step": 6576 }, { "epoch": 0.88, "learning_rate": 1.2134170787117465e-08, "loss": 0.0449, "step": 6577 }, { "epoch": 0.88, "learning_rate": 1.2120807162902579e-08, "loss": 0.0619, "step": 6578 }, { "epoch": 0.88, "learning_rate": 1.2107443538687692e-08, "loss": 0.0857, "step": 6579 }, { "epoch": 0.88, "learning_rate": 1.2094079914472805e-08, "loss": 0.0607, "step": 6580 }, { "epoch": 0.88, "learning_rate": 1.2080716290257916e-08, "loss": 0.067, "step": 6581 }, { "epoch": 0.88, "learning_rate": 1.2067352666043031e-08, "loss": 0.0424, "step": 6582 }, { "epoch": 0.88, "learning_rate": 1.2053989041828143e-08, "loss": 0.1059, "step": 6583 }, { "epoch": 0.88, "learning_rate": 1.2040625417613256e-08, "loss": 0.0838, "step": 6584 }, { "epoch": 0.88, "learning_rate": 1.2027261793398369e-08, "loss": 0.0811, "step": 6585 }, { "epoch": 0.88, "learning_rate": 1.2013898169183482e-08, "loss": 0.0465, "step": 6586 }, { "epoch": 0.88, "learning_rate": 1.2000534544968595e-08, "loss": 0.0481, "step": 6587 }, { "epoch": 0.88, "learning_rate": 1.1987170920753707e-08, "loss": 0.0845, "step": 6588 }, { "epoch": 0.88, "learning_rate": 1.1973807296538822e-08, "loss": 0.0641, "step": 6589 }, { "epoch": 0.88, "learning_rate": 1.1960443672323933e-08, "loss": 0.0891, "step": 6590 }, { "epoch": 0.88, "learning_rate": 1.1947080048109046e-08, "loss": 0.0589, "step": 6591 }, { "epoch": 0.88, "learning_rate": 1.193371642389416e-08, "loss": 0.1137, "step": 6592 }, { "epoch": 0.88, "learning_rate": 1.1920352799679272e-08, "loss": 0.0726, "step": 6593 }, { "epoch": 0.88, "learning_rate": 1.1906989175464386e-08, "loss": 0.0841, "step": 6594 }, { "epoch": 0.88, "learning_rate": 1.1893625551249499e-08, "loss": 0.082, "step": 6595 }, { "epoch": 0.88, "learning_rate": 1.1880261927034612e-08, "loss": 0.0569, "step": 6596 }, { "epoch": 0.88, "learning_rate": 1.1866898302819723e-08, "loss": 0.072, "step": 6597 }, { "epoch": 0.88, "learning_rate": 1.1853534678604838e-08, "loss": 0.0794, "step": 6598 }, { "epoch": 0.88, "learning_rate": 1.184017105438995e-08, "loss": 0.1855, "step": 6599 }, { "epoch": 0.88, "learning_rate": 1.1826807430175063e-08, "loss": 0.0754, "step": 6600 }, { "epoch": 0.88, "learning_rate": 1.1813443805960176e-08, "loss": 0.0749, "step": 6601 }, { "epoch": 0.88, "learning_rate": 1.1800080181745289e-08, "loss": 0.0941, "step": 6602 }, { "epoch": 0.88, "learning_rate": 1.1786716557530402e-08, "loss": 0.0613, "step": 6603 }, { "epoch": 0.88, "learning_rate": 1.1773352933315514e-08, "loss": 0.0597, "step": 6604 }, { "epoch": 0.88, "learning_rate": 1.1759989309100629e-08, "loss": 0.0656, "step": 6605 }, { "epoch": 0.88, "learning_rate": 1.174662568488574e-08, "loss": 0.0559, "step": 6606 }, { "epoch": 0.88, "learning_rate": 1.1733262060670853e-08, "loss": 0.0619, "step": 6607 }, { "epoch": 0.88, "learning_rate": 1.1719898436455966e-08, "loss": 0.0565, "step": 6608 }, { "epoch": 0.88, "learning_rate": 1.170653481224108e-08, "loss": 0.1158, "step": 6609 }, { "epoch": 0.88, "learning_rate": 1.1693171188026193e-08, "loss": 0.0487, "step": 6610 }, { "epoch": 0.88, "learning_rate": 1.1679807563811304e-08, "loss": 0.0611, "step": 6611 }, { "epoch": 0.88, "learning_rate": 1.1666443939596419e-08, "loss": 0.0771, "step": 6612 }, { "epoch": 0.88, "learning_rate": 1.165308031538153e-08, "loss": 0.0448, "step": 6613 }, { "epoch": 0.88, "learning_rate": 1.1639716691166643e-08, "loss": 0.0671, "step": 6614 }, { "epoch": 0.88, "learning_rate": 1.1626353066951757e-08, "loss": 0.069, "step": 6615 }, { "epoch": 0.88, "learning_rate": 1.161298944273687e-08, "loss": 0.0423, "step": 6616 }, { "epoch": 0.88, "learning_rate": 1.1599625818521983e-08, "loss": 0.0866, "step": 6617 }, { "epoch": 0.88, "learning_rate": 1.1586262194307096e-08, "loss": 0.0564, "step": 6618 }, { "epoch": 0.88, "learning_rate": 1.157289857009221e-08, "loss": 0.0625, "step": 6619 }, { "epoch": 0.88, "learning_rate": 1.155953494587732e-08, "loss": 0.049, "step": 6620 }, { "epoch": 0.88, "learning_rate": 1.1546171321662434e-08, "loss": 0.0649, "step": 6621 }, { "epoch": 0.88, "learning_rate": 1.1532807697447547e-08, "loss": 0.0567, "step": 6622 }, { "epoch": 0.89, "learning_rate": 1.151944407323266e-08, "loss": 0.0681, "step": 6623 }, { "epoch": 0.89, "learning_rate": 1.1506080449017773e-08, "loss": 0.0913, "step": 6624 }, { "epoch": 0.89, "learning_rate": 1.1492716824802886e-08, "loss": 0.0567, "step": 6625 }, { "epoch": 0.89, "learning_rate": 1.1479353200588e-08, "loss": 0.0569, "step": 6626 }, { "epoch": 0.89, "learning_rate": 1.1465989576373111e-08, "loss": 0.0389, "step": 6627 }, { "epoch": 0.89, "learning_rate": 1.1452625952158224e-08, "loss": 0.0641, "step": 6628 }, { "epoch": 0.89, "learning_rate": 1.1439262327943337e-08, "loss": 0.0515, "step": 6629 }, { "epoch": 0.89, "learning_rate": 1.142589870372845e-08, "loss": 0.0569, "step": 6630 }, { "epoch": 0.89, "learning_rate": 1.1412535079513564e-08, "loss": 0.0585, "step": 6631 }, { "epoch": 0.89, "learning_rate": 1.1399171455298677e-08, "loss": 0.0814, "step": 6632 }, { "epoch": 0.89, "learning_rate": 1.138580783108379e-08, "loss": 0.082, "step": 6633 }, { "epoch": 0.89, "learning_rate": 1.1372444206868901e-08, "loss": 0.0711, "step": 6634 }, { "epoch": 0.89, "learning_rate": 1.1359080582654016e-08, "loss": 0.0676, "step": 6635 }, { "epoch": 0.89, "learning_rate": 1.1345716958439128e-08, "loss": 0.0951, "step": 6636 }, { "epoch": 0.89, "learning_rate": 1.133235333422424e-08, "loss": 0.0592, "step": 6637 }, { "epoch": 0.89, "learning_rate": 1.1318989710009354e-08, "loss": 0.0406, "step": 6638 }, { "epoch": 0.89, "learning_rate": 1.1305626085794467e-08, "loss": 0.0774, "step": 6639 }, { "epoch": 0.89, "learning_rate": 1.129226246157958e-08, "loss": 0.0419, "step": 6640 }, { "epoch": 0.89, "learning_rate": 1.1278898837364693e-08, "loss": 0.0646, "step": 6641 }, { "epoch": 0.89, "learning_rate": 1.1265535213149806e-08, "loss": 0.0652, "step": 6642 }, { "epoch": 0.89, "learning_rate": 1.1252171588934918e-08, "loss": 0.0947, "step": 6643 }, { "epoch": 0.89, "learning_rate": 1.1238807964720031e-08, "loss": 0.0627, "step": 6644 }, { "epoch": 0.89, "learning_rate": 1.1225444340505144e-08, "loss": 0.0706, "step": 6645 }, { "epoch": 0.89, "learning_rate": 1.1212080716290257e-08, "loss": 0.0375, "step": 6646 }, { "epoch": 0.89, "learning_rate": 1.119871709207537e-08, "loss": 0.0581, "step": 6647 }, { "epoch": 0.89, "learning_rate": 1.1185353467860484e-08, "loss": 0.0707, "step": 6648 }, { "epoch": 0.89, "learning_rate": 1.1171989843645597e-08, "loss": 0.0882, "step": 6649 }, { "epoch": 0.89, "learning_rate": 1.1158626219430708e-08, "loss": 0.0456, "step": 6650 }, { "epoch": 0.89, "learning_rate": 1.1145262595215821e-08, "loss": 0.0696, "step": 6651 }, { "epoch": 0.89, "learning_rate": 1.1131898971000935e-08, "loss": 0.046, "step": 6652 }, { "epoch": 0.89, "learning_rate": 1.1118535346786048e-08, "loss": 0.0672, "step": 6653 }, { "epoch": 0.89, "learning_rate": 1.1105171722571161e-08, "loss": 0.0693, "step": 6654 }, { "epoch": 0.89, "learning_rate": 1.1091808098356274e-08, "loss": 0.0459, "step": 6655 }, { "epoch": 0.89, "learning_rate": 1.1078444474141387e-08, "loss": 0.0691, "step": 6656 }, { "epoch": 0.89, "learning_rate": 1.1065080849926499e-08, "loss": 0.0673, "step": 6657 }, { "epoch": 0.89, "learning_rate": 1.1051717225711612e-08, "loss": 0.0496, "step": 6658 }, { "epoch": 0.89, "learning_rate": 1.1038353601496725e-08, "loss": 0.0603, "step": 6659 }, { "epoch": 0.89, "learning_rate": 1.1024989977281838e-08, "loss": 0.0385, "step": 6660 }, { "epoch": 0.89, "learning_rate": 1.1011626353066951e-08, "loss": 0.0473, "step": 6661 }, { "epoch": 0.89, "learning_rate": 1.0998262728852064e-08, "loss": 0.052, "step": 6662 }, { "epoch": 0.89, "learning_rate": 1.0984899104637178e-08, "loss": 0.0447, "step": 6663 }, { "epoch": 0.89, "learning_rate": 1.097153548042229e-08, "loss": 0.0774, "step": 6664 }, { "epoch": 0.89, "learning_rate": 1.0958171856207404e-08, "loss": 0.0821, "step": 6665 }, { "epoch": 0.89, "learning_rate": 1.0944808231992515e-08, "loss": 0.0554, "step": 6666 }, { "epoch": 0.89, "learning_rate": 1.0931444607777628e-08, "loss": 0.0628, "step": 6667 }, { "epoch": 0.89, "learning_rate": 1.0918080983562742e-08, "loss": 0.0594, "step": 6668 }, { "epoch": 0.89, "learning_rate": 1.0904717359347855e-08, "loss": 0.0526, "step": 6669 }, { "epoch": 0.89, "learning_rate": 1.0891353735132968e-08, "loss": 0.05, "step": 6670 }, { "epoch": 0.89, "learning_rate": 1.0877990110918081e-08, "loss": 0.0719, "step": 6671 }, { "epoch": 0.89, "learning_rate": 1.0864626486703194e-08, "loss": 0.0774, "step": 6672 }, { "epoch": 0.89, "learning_rate": 1.0851262862488306e-08, "loss": 0.0711, "step": 6673 }, { "epoch": 0.89, "learning_rate": 1.0837899238273419e-08, "loss": 0.0688, "step": 6674 }, { "epoch": 0.89, "learning_rate": 1.0824535614058532e-08, "loss": 0.0964, "step": 6675 }, { "epoch": 0.89, "learning_rate": 1.0811171989843645e-08, "loss": 0.049, "step": 6676 }, { "epoch": 0.89, "learning_rate": 1.0797808365628758e-08, "loss": 0.0406, "step": 6677 }, { "epoch": 0.89, "learning_rate": 1.0784444741413871e-08, "loss": 0.0457, "step": 6678 }, { "epoch": 0.89, "learning_rate": 1.0771081117198984e-08, "loss": 0.0612, "step": 6679 }, { "epoch": 0.89, "learning_rate": 1.0757717492984098e-08, "loss": 0.0583, "step": 6680 }, { "epoch": 0.89, "learning_rate": 1.0744353868769209e-08, "loss": 0.0521, "step": 6681 }, { "epoch": 0.89, "learning_rate": 1.0730990244554322e-08, "loss": 0.0392, "step": 6682 }, { "epoch": 0.89, "learning_rate": 1.0717626620339435e-08, "loss": 0.0702, "step": 6683 }, { "epoch": 0.89, "learning_rate": 1.0704262996124549e-08, "loss": 0.0627, "step": 6684 }, { "epoch": 0.89, "learning_rate": 1.0690899371909662e-08, "loss": 0.059, "step": 6685 }, { "epoch": 0.89, "learning_rate": 1.0677535747694775e-08, "loss": 0.0445, "step": 6686 }, { "epoch": 0.89, "learning_rate": 1.0664172123479888e-08, "loss": 0.0744, "step": 6687 }, { "epoch": 0.89, "learning_rate": 1.0650808499265e-08, "loss": 0.0853, "step": 6688 }, { "epoch": 0.89, "learning_rate": 1.0650808499265e-08, "loss": 0.0682, "step": 6689 }, { "epoch": 0.89, "learning_rate": 1.0637444875050113e-08, "loss": 0.0554, "step": 6690 }, { "epoch": 0.89, "learning_rate": 1.0624081250835226e-08, "loss": 0.0506, "step": 6691 }, { "epoch": 0.89, "learning_rate": 1.0610717626620339e-08, "loss": 0.0984, "step": 6692 }, { "epoch": 0.89, "learning_rate": 1.0597354002405452e-08, "loss": 0.0739, "step": 6693 }, { "epoch": 0.89, "learning_rate": 1.0583990378190565e-08, "loss": 0.0802, "step": 6694 }, { "epoch": 0.89, "learning_rate": 1.0570626753975678e-08, "loss": 0.071, "step": 6695 }, { "epoch": 0.89, "learning_rate": 1.055726312976079e-08, "loss": 0.0847, "step": 6696 }, { "epoch": 0.89, "learning_rate": 1.0543899505545903e-08, "loss": 0.0637, "step": 6697 }, { "epoch": 0.9, "learning_rate": 1.0530535881331016e-08, "loss": 0.0327, "step": 6698 }, { "epoch": 0.9, "learning_rate": 1.051717225711613e-08, "loss": 0.0438, "step": 6699 }, { "epoch": 0.9, "learning_rate": 1.0503808632901242e-08, "loss": 0.0476, "step": 6700 }, { "epoch": 0.9, "learning_rate": 1.0490445008686356e-08, "loss": 0.0705, "step": 6701 }, { "epoch": 0.9, "learning_rate": 1.0477081384471469e-08, "loss": 0.0622, "step": 6702 }, { "epoch": 0.9, "learning_rate": 1.0463717760256582e-08, "loss": 0.0706, "step": 6703 }, { "epoch": 0.9, "learning_rate": 1.0450354136041695e-08, "loss": 0.056, "step": 6704 }, { "epoch": 0.9, "learning_rate": 1.0436990511826806e-08, "loss": 0.0493, "step": 6705 }, { "epoch": 0.9, "learning_rate": 1.042362688761192e-08, "loss": 0.0373, "step": 6706 }, { "epoch": 0.9, "learning_rate": 1.0410263263397033e-08, "loss": 0.0677, "step": 6707 }, { "epoch": 0.9, "learning_rate": 1.0396899639182146e-08, "loss": 0.0671, "step": 6708 }, { "epoch": 0.9, "learning_rate": 1.0383536014967259e-08, "loss": 0.0684, "step": 6709 }, { "epoch": 0.9, "learning_rate": 1.0370172390752372e-08, "loss": 0.067, "step": 6710 }, { "epoch": 0.9, "learning_rate": 1.0356808766537485e-08, "loss": 0.0574, "step": 6711 }, { "epoch": 0.9, "learning_rate": 1.0343445142322597e-08, "loss": 0.0623, "step": 6712 }, { "epoch": 0.9, "learning_rate": 1.033008151810771e-08, "loss": 0.0684, "step": 6713 }, { "epoch": 0.9, "learning_rate": 1.0316717893892823e-08, "loss": 0.0768, "step": 6714 }, { "epoch": 0.9, "learning_rate": 1.0303354269677936e-08, "loss": 0.0644, "step": 6715 }, { "epoch": 0.9, "learning_rate": 1.028999064546305e-08, "loss": 0.1076, "step": 6716 }, { "epoch": 0.9, "learning_rate": 1.0276627021248162e-08, "loss": 0.036, "step": 6717 }, { "epoch": 0.9, "learning_rate": 1.0263263397033276e-08, "loss": 0.072, "step": 6718 }, { "epoch": 0.9, "learning_rate": 1.0249899772818387e-08, "loss": 0.0587, "step": 6719 }, { "epoch": 0.9, "learning_rate": 1.02365361486035e-08, "loss": 0.0575, "step": 6720 }, { "epoch": 0.9, "learning_rate": 1.0223172524388613e-08, "loss": 0.0554, "step": 6721 }, { "epoch": 0.9, "learning_rate": 1.0209808900173727e-08, "loss": 0.0581, "step": 6722 }, { "epoch": 0.9, "learning_rate": 1.019644527595884e-08, "loss": 0.0622, "step": 6723 }, { "epoch": 0.9, "learning_rate": 1.0183081651743953e-08, "loss": 0.0982, "step": 6724 }, { "epoch": 0.9, "learning_rate": 1.0169718027529066e-08, "loss": 0.0586, "step": 6725 }, { "epoch": 0.9, "learning_rate": 1.0156354403314177e-08, "loss": 0.0557, "step": 6726 }, { "epoch": 0.9, "learning_rate": 1.0142990779099292e-08, "loss": 0.1132, "step": 6727 }, { "epoch": 0.9, "learning_rate": 1.0129627154884404e-08, "loss": 0.0532, "step": 6728 }, { "epoch": 0.9, "learning_rate": 1.0116263530669517e-08, "loss": 0.0558, "step": 6729 }, { "epoch": 0.9, "learning_rate": 1.010289990645463e-08, "loss": 0.0724, "step": 6730 }, { "epoch": 0.9, "learning_rate": 1.0089536282239743e-08, "loss": 0.045, "step": 6731 }, { "epoch": 0.9, "learning_rate": 1.0076172658024856e-08, "loss": 0.0769, "step": 6732 }, { "epoch": 0.9, "learning_rate": 1.0062809033809968e-08, "loss": 0.0673, "step": 6733 }, { "epoch": 0.9, "learning_rate": 1.0049445409595083e-08, "loss": 0.0665, "step": 6734 }, { "epoch": 0.9, "learning_rate": 1.0036081785380194e-08, "loss": 0.094, "step": 6735 }, { "epoch": 0.9, "learning_rate": 1.0022718161165307e-08, "loss": 0.0599, "step": 6736 }, { "epoch": 0.9, "learning_rate": 1.000935453695042e-08, "loss": 0.0479, "step": 6737 }, { "epoch": 0.9, "learning_rate": 9.995990912735533e-09, "loss": 0.041, "step": 6738 }, { "epoch": 0.9, "learning_rate": 9.982627288520647e-09, "loss": 0.057, "step": 6739 }, { "epoch": 0.9, "learning_rate": 9.96926366430576e-09, "loss": 0.0477, "step": 6740 }, { "epoch": 0.9, "learning_rate": 9.955900040090873e-09, "loss": 0.0497, "step": 6741 }, { "epoch": 0.9, "learning_rate": 9.942536415875984e-09, "loss": 0.0825, "step": 6742 }, { "epoch": 0.9, "learning_rate": 9.9291727916611e-09, "loss": 0.0513, "step": 6743 }, { "epoch": 0.9, "learning_rate": 9.91580916744621e-09, "loss": 0.048, "step": 6744 }, { "epoch": 0.9, "learning_rate": 9.902445543231324e-09, "loss": 0.0753, "step": 6745 }, { "epoch": 0.9, "learning_rate": 9.889081919016437e-09, "loss": 0.0857, "step": 6746 }, { "epoch": 0.9, "learning_rate": 9.87571829480155e-09, "loss": 0.0554, "step": 6747 }, { "epoch": 0.9, "learning_rate": 9.862354670586663e-09, "loss": 0.0737, "step": 6748 }, { "epoch": 0.9, "learning_rate": 9.848991046371775e-09, "loss": 0.0722, "step": 6749 }, { "epoch": 0.9, "learning_rate": 9.83562742215689e-09, "loss": 0.047, "step": 6750 }, { "epoch": 0.9, "learning_rate": 9.822263797942001e-09, "loss": 0.049, "step": 6751 }, { "epoch": 0.9, "learning_rate": 9.808900173727114e-09, "loss": 0.0795, "step": 6752 }, { "epoch": 0.9, "learning_rate": 9.795536549512227e-09, "loss": 0.1313, "step": 6753 }, { "epoch": 0.9, "learning_rate": 9.78217292529734e-09, "loss": 0.068, "step": 6754 }, { "epoch": 0.9, "learning_rate": 9.768809301082454e-09, "loss": 0.0624, "step": 6755 }, { "epoch": 0.9, "learning_rate": 9.755445676867565e-09, "loss": 0.0673, "step": 6756 }, { "epoch": 0.9, "learning_rate": 9.74208205265268e-09, "loss": 0.0655, "step": 6757 }, { "epoch": 0.9, "learning_rate": 9.728718428437791e-09, "loss": 0.0662, "step": 6758 }, { "epoch": 0.9, "learning_rate": 9.715354804222905e-09, "loss": 0.037, "step": 6759 }, { "epoch": 0.9, "learning_rate": 9.701991180008018e-09, "loss": 0.0465, "step": 6760 }, { "epoch": 0.9, "learning_rate": 9.68862755579313e-09, "loss": 0.0536, "step": 6761 }, { "epoch": 0.9, "learning_rate": 9.675263931578244e-09, "loss": 0.073, "step": 6762 }, { "epoch": 0.9, "learning_rate": 9.661900307363355e-09, "loss": 0.075, "step": 6763 }, { "epoch": 0.9, "learning_rate": 9.64853668314847e-09, "loss": 0.0562, "step": 6764 }, { "epoch": 0.9, "learning_rate": 9.635173058933582e-09, "loss": 0.0803, "step": 6765 }, { "epoch": 0.9, "learning_rate": 9.621809434718697e-09, "loss": 0.027, "step": 6766 }, { "epoch": 0.9, "learning_rate": 9.608445810503808e-09, "loss": 0.0501, "step": 6767 }, { "epoch": 0.9, "learning_rate": 9.595082186288921e-09, "loss": 0.0497, "step": 6768 }, { "epoch": 0.9, "learning_rate": 9.581718562074034e-09, "loss": 0.0645, "step": 6769 }, { "epoch": 0.9, "learning_rate": 9.568354937859147e-09, "loss": 0.067, "step": 6770 }, { "epoch": 0.9, "learning_rate": 9.55499131364426e-09, "loss": 0.0974, "step": 6771 }, { "epoch": 0.9, "learning_rate": 9.541627689429372e-09, "loss": 0.0762, "step": 6772 }, { "epoch": 0.91, "learning_rate": 9.528264065214487e-09, "loss": 0.0542, "step": 6773 }, { "epoch": 0.91, "learning_rate": 9.514900440999598e-09, "loss": 0.0484, "step": 6774 }, { "epoch": 0.91, "learning_rate": 9.501536816784711e-09, "loss": 0.0539, "step": 6775 }, { "epoch": 0.91, "learning_rate": 9.488173192569825e-09, "loss": 0.0761, "step": 6776 }, { "epoch": 0.91, "learning_rate": 9.474809568354938e-09, "loss": 0.0644, "step": 6777 }, { "epoch": 0.91, "learning_rate": 9.461445944140051e-09, "loss": 0.0702, "step": 6778 }, { "epoch": 0.91, "learning_rate": 9.448082319925162e-09, "loss": 0.0421, "step": 6779 }, { "epoch": 0.91, "learning_rate": 9.434718695710277e-09, "loss": 0.0615, "step": 6780 }, { "epoch": 0.91, "learning_rate": 9.421355071495389e-09, "loss": 0.0528, "step": 6781 }, { "epoch": 0.91, "learning_rate": 9.407991447280502e-09, "loss": 0.0549, "step": 6782 }, { "epoch": 0.91, "learning_rate": 9.394627823065615e-09, "loss": 0.0949, "step": 6783 }, { "epoch": 0.91, "learning_rate": 9.381264198850728e-09, "loss": 0.0701, "step": 6784 }, { "epoch": 0.91, "learning_rate": 9.367900574635841e-09, "loss": 0.0878, "step": 6785 }, { "epoch": 0.91, "learning_rate": 9.354536950420953e-09, "loss": 0.0526, "step": 6786 }, { "epoch": 0.91, "learning_rate": 9.341173326206068e-09, "loss": 0.0407, "step": 6787 }, { "epoch": 0.91, "learning_rate": 9.327809701991179e-09, "loss": 0.0555, "step": 6788 }, { "epoch": 0.91, "learning_rate": 9.314446077776294e-09, "loss": 0.0859, "step": 6789 }, { "epoch": 0.91, "learning_rate": 9.301082453561405e-09, "loss": 0.0656, "step": 6790 }, { "epoch": 0.91, "learning_rate": 9.287718829346518e-09, "loss": 0.039, "step": 6791 }, { "epoch": 0.91, "learning_rate": 9.274355205131632e-09, "loss": 0.0374, "step": 6792 }, { "epoch": 0.91, "learning_rate": 9.260991580916743e-09, "loss": 0.0743, "step": 6793 }, { "epoch": 0.91, "learning_rate": 9.247627956701858e-09, "loss": 0.089, "step": 6794 }, { "epoch": 0.91, "learning_rate": 9.23426433248697e-09, "loss": 0.0585, "step": 6795 }, { "epoch": 0.91, "learning_rate": 9.220900708272084e-09, "loss": 0.0706, "step": 6796 }, { "epoch": 0.91, "learning_rate": 9.207537084057196e-09, "loss": 0.0495, "step": 6797 }, { "epoch": 0.91, "learning_rate": 9.194173459842309e-09, "loss": 0.0524, "step": 6798 }, { "epoch": 0.91, "learning_rate": 9.180809835627422e-09, "loss": 0.0495, "step": 6799 }, { "epoch": 0.91, "learning_rate": 9.167446211412533e-09, "loss": 0.0643, "step": 6800 }, { "epoch": 0.91, "learning_rate": 9.154082587197648e-09, "loss": 0.0391, "step": 6801 }, { "epoch": 0.91, "learning_rate": 9.14071896298276e-09, "loss": 0.0756, "step": 6802 }, { "epoch": 0.91, "learning_rate": 9.127355338767874e-09, "loss": 0.0458, "step": 6803 }, { "epoch": 0.91, "learning_rate": 9.113991714552986e-09, "loss": 0.0468, "step": 6804 }, { "epoch": 0.91, "learning_rate": 9.1006280903381e-09, "loss": 0.0407, "step": 6805 }, { "epoch": 0.91, "learning_rate": 9.087264466123212e-09, "loss": 0.0575, "step": 6806 }, { "epoch": 0.91, "learning_rate": 9.073900841908325e-09, "loss": 0.044, "step": 6807 }, { "epoch": 0.91, "learning_rate": 9.060537217693439e-09, "loss": 0.0405, "step": 6808 }, { "epoch": 0.91, "learning_rate": 9.04717359347855e-09, "loss": 0.072, "step": 6809 }, { "epoch": 0.91, "learning_rate": 9.033809969263665e-09, "loss": 0.0653, "step": 6810 }, { "epoch": 0.91, "learning_rate": 9.020446345048776e-09, "loss": 0.0626, "step": 6811 }, { "epoch": 0.91, "learning_rate": 9.007082720833891e-09, "loss": 0.0611, "step": 6812 }, { "epoch": 0.91, "learning_rate": 8.993719096619003e-09, "loss": 0.0565, "step": 6813 }, { "epoch": 0.91, "learning_rate": 8.980355472404116e-09, "loss": 0.049, "step": 6814 }, { "epoch": 0.91, "learning_rate": 8.966991848189229e-09, "loss": 0.0739, "step": 6815 }, { "epoch": 0.91, "learning_rate": 8.95362822397434e-09, "loss": 0.0565, "step": 6816 }, { "epoch": 0.91, "learning_rate": 8.940264599759455e-09, "loss": 0.0688, "step": 6817 }, { "epoch": 0.91, "learning_rate": 8.926900975544567e-09, "loss": 0.0726, "step": 6818 }, { "epoch": 0.91, "learning_rate": 8.913537351329681e-09, "loss": 0.0637, "step": 6819 }, { "epoch": 0.91, "learning_rate": 8.900173727114793e-09, "loss": 0.0493, "step": 6820 }, { "epoch": 0.91, "learning_rate": 8.886810102899906e-09, "loss": 0.04, "step": 6821 }, { "epoch": 0.91, "learning_rate": 8.87344647868502e-09, "loss": 0.0782, "step": 6822 }, { "epoch": 0.91, "learning_rate": 8.86008285447013e-09, "loss": 0.0865, "step": 6823 }, { "epoch": 0.91, "learning_rate": 8.846719230255246e-09, "loss": 0.05, "step": 6824 }, { "epoch": 0.91, "learning_rate": 8.833355606040357e-09, "loss": 0.0642, "step": 6825 }, { "epoch": 0.91, "learning_rate": 8.819991981825472e-09, "loss": 0.0798, "step": 6826 }, { "epoch": 0.91, "learning_rate": 8.806628357610583e-09, "loss": 0.0692, "step": 6827 }, { "epoch": 0.91, "learning_rate": 8.793264733395696e-09, "loss": 0.0421, "step": 6828 }, { "epoch": 0.91, "learning_rate": 8.77990110918081e-09, "loss": 0.0684, "step": 6829 }, { "epoch": 0.91, "learning_rate": 8.766537484965921e-09, "loss": 0.0821, "step": 6830 }, { "epoch": 0.91, "learning_rate": 8.753173860751036e-09, "loss": 0.0369, "step": 6831 }, { "epoch": 0.91, "learning_rate": 8.739810236536147e-09, "loss": 0.0371, "step": 6832 }, { "epoch": 0.91, "learning_rate": 8.726446612321262e-09, "loss": 0.0628, "step": 6833 }, { "epoch": 0.91, "learning_rate": 8.713082988106374e-09, "loss": 0.0687, "step": 6834 }, { "epoch": 0.91, "learning_rate": 8.699719363891487e-09, "loss": 0.0453, "step": 6835 }, { "epoch": 0.91, "learning_rate": 8.6863557396766e-09, "loss": 0.0595, "step": 6836 }, { "epoch": 0.91, "learning_rate": 8.672992115461711e-09, "loss": 0.0616, "step": 6837 }, { "epoch": 0.91, "learning_rate": 8.659628491246826e-09, "loss": 0.0708, "step": 6838 }, { "epoch": 0.91, "learning_rate": 8.646264867031938e-09, "loss": 0.0586, "step": 6839 }, { "epoch": 0.91, "learning_rate": 8.632901242817052e-09, "loss": 0.0558, "step": 6840 }, { "epoch": 0.91, "learning_rate": 8.619537618602164e-09, "loss": 0.0555, "step": 6841 }, { "epoch": 0.91, "learning_rate": 8.606173994387279e-09, "loss": 0.0679, "step": 6842 }, { "epoch": 0.91, "learning_rate": 8.59281037017239e-09, "loss": 0.0702, "step": 6843 }, { "epoch": 0.91, "learning_rate": 8.579446745957503e-09, "loss": 0.0616, "step": 6844 }, { "epoch": 0.91, "learning_rate": 8.566083121742617e-09, "loss": 0.0468, "step": 6845 }, { "epoch": 0.91, "learning_rate": 8.552719497527728e-09, "loss": 0.0785, "step": 6846 }, { "epoch": 0.92, "learning_rate": 8.539355873312843e-09, "loss": 0.0534, "step": 6847 }, { "epoch": 0.92, "learning_rate": 8.525992249097954e-09, "loss": 0.0556, "step": 6848 }, { "epoch": 0.92, "learning_rate": 8.512628624883069e-09, "loss": 0.0673, "step": 6849 }, { "epoch": 0.92, "learning_rate": 8.49926500066818e-09, "loss": 0.0658, "step": 6850 }, { "epoch": 0.92, "learning_rate": 8.485901376453294e-09, "loss": 0.05, "step": 6851 }, { "epoch": 0.92, "learning_rate": 8.472537752238407e-09, "loss": 0.0862, "step": 6852 }, { "epoch": 0.92, "learning_rate": 8.459174128023518e-09, "loss": 0.1125, "step": 6853 }, { "epoch": 0.92, "learning_rate": 8.445810503808633e-09, "loss": 0.0794, "step": 6854 }, { "epoch": 0.92, "learning_rate": 8.432446879593745e-09, "loss": 0.0731, "step": 6855 }, { "epoch": 0.92, "learning_rate": 8.41908325537886e-09, "loss": 0.0886, "step": 6856 }, { "epoch": 0.92, "learning_rate": 8.405719631163971e-09, "loss": 0.0736, "step": 6857 }, { "epoch": 0.92, "learning_rate": 8.392356006949084e-09, "loss": 0.0605, "step": 6858 }, { "epoch": 0.92, "learning_rate": 8.378992382734197e-09, "loss": 0.0711, "step": 6859 }, { "epoch": 0.92, "learning_rate": 8.365628758519309e-09, "loss": 0.0405, "step": 6860 }, { "epoch": 0.92, "learning_rate": 8.352265134304423e-09, "loss": 0.1056, "step": 6861 }, { "epoch": 0.92, "learning_rate": 8.338901510089535e-09, "loss": 0.0663, "step": 6862 }, { "epoch": 0.92, "learning_rate": 8.32553788587465e-09, "loss": 0.0487, "step": 6863 }, { "epoch": 0.92, "learning_rate": 8.312174261659761e-09, "loss": 0.0637, "step": 6864 }, { "epoch": 0.92, "learning_rate": 8.298810637444874e-09, "loss": 0.0568, "step": 6865 }, { "epoch": 0.92, "learning_rate": 8.285447013229988e-09, "loss": 0.0601, "step": 6866 }, { "epoch": 0.92, "learning_rate": 8.272083389015099e-09, "loss": 0.0722, "step": 6867 }, { "epoch": 0.92, "learning_rate": 8.258719764800214e-09, "loss": 0.0509, "step": 6868 }, { "epoch": 0.92, "learning_rate": 8.245356140585325e-09, "loss": 0.0555, "step": 6869 }, { "epoch": 0.92, "learning_rate": 8.23199251637044e-09, "loss": 0.06, "step": 6870 }, { "epoch": 0.92, "learning_rate": 8.218628892155552e-09, "loss": 0.062, "step": 6871 }, { "epoch": 0.92, "learning_rate": 8.205265267940666e-09, "loss": 0.0595, "step": 6872 }, { "epoch": 0.92, "learning_rate": 8.191901643725778e-09, "loss": 0.0683, "step": 6873 }, { "epoch": 0.92, "learning_rate": 8.178538019510891e-09, "loss": 0.0674, "step": 6874 }, { "epoch": 0.92, "learning_rate": 8.165174395296004e-09, "loss": 0.0604, "step": 6875 }, { "epoch": 0.92, "learning_rate": 8.151810771081116e-09, "loss": 0.1088, "step": 6876 }, { "epoch": 0.92, "learning_rate": 8.13844714686623e-09, "loss": 0.0769, "step": 6877 }, { "epoch": 0.92, "learning_rate": 8.125083522651342e-09, "loss": 0.0874, "step": 6878 }, { "epoch": 0.92, "learning_rate": 8.111719898436457e-09, "loss": 0.0517, "step": 6879 }, { "epoch": 0.92, "learning_rate": 8.098356274221568e-09, "loss": 0.0572, "step": 6880 }, { "epoch": 0.92, "learning_rate": 8.084992650006681e-09, "loss": 0.045, "step": 6881 }, { "epoch": 0.92, "learning_rate": 8.071629025791795e-09, "loss": 0.0505, "step": 6882 }, { "epoch": 0.92, "learning_rate": 8.058265401576906e-09, "loss": 0.0787, "step": 6883 }, { "epoch": 0.92, "learning_rate": 8.04490177736202e-09, "loss": 0.0611, "step": 6884 }, { "epoch": 0.92, "learning_rate": 8.031538153147132e-09, "loss": 0.0655, "step": 6885 }, { "epoch": 0.92, "learning_rate": 8.018174528932247e-09, "loss": 0.0432, "step": 6886 }, { "epoch": 0.92, "learning_rate": 8.004810904717359e-09, "loss": 0.0606, "step": 6887 }, { "epoch": 0.92, "learning_rate": 7.991447280502472e-09, "loss": 0.0925, "step": 6888 }, { "epoch": 0.92, "learning_rate": 7.978083656287585e-09, "loss": 0.0562, "step": 6889 }, { "epoch": 0.92, "learning_rate": 7.964720032072698e-09, "loss": 0.0671, "step": 6890 }, { "epoch": 0.92, "learning_rate": 7.951356407857811e-09, "loss": 0.0535, "step": 6891 }, { "epoch": 0.92, "learning_rate": 7.937992783642923e-09, "loss": 0.0875, "step": 6892 }, { "epoch": 0.92, "learning_rate": 7.924629159428037e-09, "loss": 0.0523, "step": 6893 }, { "epoch": 0.92, "learning_rate": 7.911265535213149e-09, "loss": 0.069, "step": 6894 }, { "epoch": 0.92, "learning_rate": 7.897901910998262e-09, "loss": 0.0539, "step": 6895 }, { "epoch": 0.92, "learning_rate": 7.884538286783375e-09, "loss": 0.0555, "step": 6896 }, { "epoch": 0.92, "learning_rate": 7.871174662568488e-09, "loss": 0.0688, "step": 6897 }, { "epoch": 0.92, "learning_rate": 7.857811038353601e-09, "loss": 0.036, "step": 6898 }, { "epoch": 0.92, "learning_rate": 7.844447414138713e-09, "loss": 0.0764, "step": 6899 }, { "epoch": 0.92, "learning_rate": 7.831083789923828e-09, "loss": 0.0738, "step": 6900 }, { "epoch": 0.92, "learning_rate": 7.81772016570894e-09, "loss": 0.0602, "step": 6901 }, { "epoch": 0.92, "learning_rate": 7.804356541494052e-09, "loss": 0.0595, "step": 6902 }, { "epoch": 0.92, "learning_rate": 7.790992917279166e-09, "loss": 0.039, "step": 6903 }, { "epoch": 0.92, "learning_rate": 7.777629293064279e-09, "loss": 0.0635, "step": 6904 }, { "epoch": 0.92, "learning_rate": 7.764265668849392e-09, "loss": 0.0445, "step": 6905 }, { "epoch": 0.92, "learning_rate": 7.750902044634503e-09, "loss": 0.057, "step": 6906 }, { "epoch": 0.92, "learning_rate": 7.737538420419618e-09, "loss": 0.0597, "step": 6907 }, { "epoch": 0.92, "learning_rate": 7.72417479620473e-09, "loss": 0.0665, "step": 6908 }, { "epoch": 0.92, "learning_rate": 7.710811171989844e-09, "loss": 0.0611, "step": 6909 }, { "epoch": 0.92, "learning_rate": 7.697447547774956e-09, "loss": 0.0634, "step": 6910 }, { "epoch": 0.92, "learning_rate": 7.684083923560069e-09, "loss": 0.0793, "step": 6911 }, { "epoch": 0.92, "learning_rate": 7.670720299345182e-09, "loss": 0.1875, "step": 6912 }, { "epoch": 0.92, "learning_rate": 7.657356675130295e-09, "loss": 0.1011, "step": 6913 }, { "epoch": 0.92, "learning_rate": 7.643993050915408e-09, "loss": 0.0473, "step": 6914 }, { "epoch": 0.92, "learning_rate": 7.63062942670052e-09, "loss": 0.0973, "step": 6915 }, { "epoch": 0.92, "learning_rate": 7.617265802485635e-09, "loss": 0.0947, "step": 6916 }, { "epoch": 0.92, "learning_rate": 7.603902178270746e-09, "loss": 0.0672, "step": 6917 }, { "epoch": 0.92, "learning_rate": 7.59053855405586e-09, "loss": 0.0712, "step": 6918 }, { "epoch": 0.92, "learning_rate": 7.577174929840973e-09, "loss": 0.0623, "step": 6919 }, { "epoch": 0.92, "learning_rate": 7.563811305626086e-09, "loss": 0.062, "step": 6920 }, { "epoch": 0.92, "learning_rate": 7.550447681411199e-09, "loss": 0.0569, "step": 6921 }, { "epoch": 0.93, "learning_rate": 7.53708405719631e-09, "loss": 0.0445, "step": 6922 }, { "epoch": 0.93, "learning_rate": 7.523720432981425e-09, "loss": 0.0591, "step": 6923 }, { "epoch": 0.93, "learning_rate": 7.510356808766537e-09, "loss": 0.0786, "step": 6924 }, { "epoch": 0.93, "learning_rate": 7.49699318455165e-09, "loss": 0.0514, "step": 6925 }, { "epoch": 0.93, "learning_rate": 7.483629560336763e-09, "loss": 0.0427, "step": 6926 }, { "epoch": 0.93, "learning_rate": 7.470265936121876e-09, "loss": 0.0576, "step": 6927 }, { "epoch": 0.93, "learning_rate": 7.456902311906989e-09, "loss": 0.0554, "step": 6928 }, { "epoch": 0.93, "learning_rate": 7.4435386876921014e-09, "loss": 0.0425, "step": 6929 }, { "epoch": 0.93, "learning_rate": 7.430175063477215e-09, "loss": 0.0614, "step": 6930 }, { "epoch": 0.93, "learning_rate": 7.416811439262327e-09, "loss": 0.0685, "step": 6931 }, { "epoch": 0.93, "learning_rate": 7.403447815047441e-09, "loss": 0.0682, "step": 6932 }, { "epoch": 0.93, "learning_rate": 7.390084190832553e-09, "loss": 0.0456, "step": 6933 }, { "epoch": 0.93, "learning_rate": 7.376720566617667e-09, "loss": 0.0607, "step": 6934 }, { "epoch": 0.93, "learning_rate": 7.3633569424027795e-09, "loss": 0.0461, "step": 6935 }, { "epoch": 0.93, "learning_rate": 7.349993318187893e-09, "loss": 0.0819, "step": 6936 }, { "epoch": 0.93, "learning_rate": 7.336629693973005e-09, "loss": 0.0431, "step": 6937 }, { "epoch": 0.93, "learning_rate": 7.323266069758117e-09, "loss": 0.0554, "step": 6938 }, { "epoch": 0.93, "learning_rate": 7.309902445543231e-09, "loss": 0.067, "step": 6939 }, { "epoch": 0.93, "learning_rate": 7.2965388213283435e-09, "loss": 0.0562, "step": 6940 }, { "epoch": 0.93, "learning_rate": 7.2831751971134575e-09, "loss": 0.0715, "step": 6941 }, { "epoch": 0.93, "learning_rate": 7.26981157289857e-09, "loss": 0.0714, "step": 6942 }, { "epoch": 0.93, "learning_rate": 7.256447948683683e-09, "loss": 0.0598, "step": 6943 }, { "epoch": 0.93, "learning_rate": 7.243084324468795e-09, "loss": 0.073, "step": 6944 }, { "epoch": 0.93, "learning_rate": 7.2297207002539076e-09, "loss": 0.0498, "step": 6945 }, { "epoch": 0.93, "learning_rate": 7.2163570760390216e-09, "loss": 0.08, "step": 6946 }, { "epoch": 0.93, "learning_rate": 7.202993451824134e-09, "loss": 0.0794, "step": 6947 }, { "epoch": 0.93, "learning_rate": 7.189629827609248e-09, "loss": 0.0695, "step": 6948 }, { "epoch": 0.93, "learning_rate": 7.17626620339436e-09, "loss": 0.0696, "step": 6949 }, { "epoch": 0.93, "learning_rate": 7.162902579179473e-09, "loss": 0.0959, "step": 6950 }, { "epoch": 0.93, "learning_rate": 7.149538954964586e-09, "loss": 0.0633, "step": 6951 }, { "epoch": 0.93, "learning_rate": 7.1361753307496996e-09, "loss": 0.0784, "step": 6952 }, { "epoch": 0.93, "learning_rate": 7.122811706534812e-09, "loss": 0.0468, "step": 6953 }, { "epoch": 0.93, "learning_rate": 7.109448082319924e-09, "loss": 0.0664, "step": 6954 }, { "epoch": 0.93, "learning_rate": 7.096084458105038e-09, "loss": 0.0425, "step": 6955 }, { "epoch": 0.93, "learning_rate": 7.0827208338901505e-09, "loss": 0.0524, "step": 6956 }, { "epoch": 0.93, "learning_rate": 7.0693572096752645e-09, "loss": 0.0623, "step": 6957 }, { "epoch": 0.93, "learning_rate": 7.055993585460377e-09, "loss": 0.0498, "step": 6958 }, { "epoch": 0.93, "learning_rate": 7.04262996124549e-09, "loss": 0.0631, "step": 6959 }, { "epoch": 0.93, "learning_rate": 7.029266337030602e-09, "loss": 0.0793, "step": 6960 }, { "epoch": 0.93, "learning_rate": 7.0159027128157145e-09, "loss": 0.0894, "step": 6961 }, { "epoch": 0.93, "learning_rate": 7.0025390886008285e-09, "loss": 0.0704, "step": 6962 }, { "epoch": 0.93, "learning_rate": 6.989175464385941e-09, "loss": 0.0548, "step": 6963 }, { "epoch": 0.93, "learning_rate": 6.975811840171055e-09, "loss": 0.0754, "step": 6964 }, { "epoch": 0.93, "learning_rate": 6.962448215956167e-09, "loss": 0.0564, "step": 6965 }, { "epoch": 0.93, "learning_rate": 6.94908459174128e-09, "loss": 0.0916, "step": 6966 }, { "epoch": 0.93, "learning_rate": 6.9357209675263926e-09, "loss": 0.0754, "step": 6967 }, { "epoch": 0.93, "learning_rate": 6.922357343311505e-09, "loss": 0.0629, "step": 6968 }, { "epoch": 0.93, "learning_rate": 6.908993719096619e-09, "loss": 0.0658, "step": 6969 }, { "epoch": 0.93, "learning_rate": 6.895630094881731e-09, "loss": 0.0655, "step": 6970 }, { "epoch": 0.93, "learning_rate": 6.882266470666845e-09, "loss": 0.0566, "step": 6971 }, { "epoch": 0.93, "learning_rate": 6.8689028464519575e-09, "loss": 0.0754, "step": 6972 }, { "epoch": 0.93, "learning_rate": 6.855539222237071e-09, "loss": 0.0453, "step": 6973 }, { "epoch": 0.93, "learning_rate": 6.842175598022183e-09, "loss": 0.0863, "step": 6974 }, { "epoch": 0.93, "learning_rate": 6.828811973807297e-09, "loss": 0.0509, "step": 6975 }, { "epoch": 0.93, "learning_rate": 6.815448349592409e-09, "loss": 0.0594, "step": 6976 }, { "epoch": 0.93, "learning_rate": 6.8020847253775215e-09, "loss": 0.0469, "step": 6977 }, { "epoch": 0.93, "learning_rate": 6.7887211011626355e-09, "loss": 0.0702, "step": 6978 }, { "epoch": 0.93, "learning_rate": 6.775357476947748e-09, "loss": 0.0546, "step": 6979 }, { "epoch": 0.93, "learning_rate": 6.761993852732861e-09, "loss": 0.0452, "step": 6980 }, { "epoch": 0.93, "learning_rate": 6.748630228517973e-09, "loss": 0.0472, "step": 6981 }, { "epoch": 0.93, "learning_rate": 6.735266604303087e-09, "loss": 0.0841, "step": 6982 }, { "epoch": 0.93, "learning_rate": 6.7219029800881995e-09, "loss": 0.052, "step": 6983 }, { "epoch": 0.93, "learning_rate": 6.708539355873312e-09, "loss": 0.0705, "step": 6984 }, { "epoch": 0.93, "learning_rate": 6.695175731658426e-09, "loss": 0.0654, "step": 6985 }, { "epoch": 0.93, "learning_rate": 6.681812107443538e-09, "loss": 0.078, "step": 6986 }, { "epoch": 0.93, "learning_rate": 6.668448483228651e-09, "loss": 0.069, "step": 6987 }, { "epoch": 0.93, "learning_rate": 6.655084859013764e-09, "loss": 0.0525, "step": 6988 }, { "epoch": 0.93, "learning_rate": 6.6417212347988776e-09, "loss": 0.0914, "step": 6989 }, { "epoch": 0.93, "learning_rate": 6.62835761058399e-09, "loss": 0.0935, "step": 6990 }, { "epoch": 0.93, "learning_rate": 6.614993986369102e-09, "loss": 0.0806, "step": 6991 }, { "epoch": 0.93, "learning_rate": 6.601630362154216e-09, "loss": 0.069, "step": 6992 }, { "epoch": 0.93, "learning_rate": 6.5882667379393285e-09, "loss": 0.0817, "step": 6993 }, { "epoch": 0.93, "learning_rate": 6.5749031137244424e-09, "loss": 0.0422, "step": 6994 }, { "epoch": 0.93, "learning_rate": 6.561539489509555e-09, "loss": 0.0877, "step": 6995 }, { "epoch": 0.93, "learning_rate": 6.548175865294668e-09, "loss": 0.0674, "step": 6996 }, { "epoch": 0.94, "learning_rate": 6.53481224107978e-09, "loss": 0.0394, "step": 6997 }, { "epoch": 0.94, "learning_rate": 6.521448616864894e-09, "loss": 0.0669, "step": 6998 }, { "epoch": 0.94, "learning_rate": 6.5080849926500065e-09, "loss": 0.0735, "step": 6999 }, { "epoch": 0.94, "learning_rate": 6.494721368435119e-09, "loss": 0.0521, "step": 7000 }, { "epoch": 0.94, "learning_rate": 6.481357744220233e-09, "loss": 0.0605, "step": 7001 }, { "epoch": 0.94, "learning_rate": 6.467994120005345e-09, "loss": 0.1427, "step": 7002 }, { "epoch": 0.94, "learning_rate": 6.454630495790458e-09, "loss": 0.0437, "step": 7003 }, { "epoch": 0.94, "learning_rate": 6.4412668715755706e-09, "loss": 0.0456, "step": 7004 }, { "epoch": 0.94, "learning_rate": 6.4279032473606845e-09, "loss": 0.047, "step": 7005 }, { "epoch": 0.94, "learning_rate": 6.414539623145797e-09, "loss": 0.0505, "step": 7006 }, { "epoch": 0.94, "learning_rate": 6.401175998930909e-09, "loss": 0.0766, "step": 7007 }, { "epoch": 0.94, "learning_rate": 6.387812374716023e-09, "loss": 0.0768, "step": 7008 }, { "epoch": 0.94, "learning_rate": 6.3744487505011354e-09, "loss": 0.0656, "step": 7009 }, { "epoch": 0.94, "learning_rate": 6.3610851262862486e-09, "loss": 0.0449, "step": 7010 }, { "epoch": 0.94, "learning_rate": 6.347721502071361e-09, "loss": 0.0446, "step": 7011 }, { "epoch": 0.94, "learning_rate": 6.334357877856475e-09, "loss": 0.0672, "step": 7012 }, { "epoch": 0.94, "learning_rate": 6.320994253641587e-09, "loss": 0.0698, "step": 7013 }, { "epoch": 0.94, "learning_rate": 6.3076306294266995e-09, "loss": 0.0428, "step": 7014 }, { "epoch": 0.94, "learning_rate": 6.2942670052118135e-09, "loss": 0.0644, "step": 7015 }, { "epoch": 0.94, "learning_rate": 6.280903380996926e-09, "loss": 0.059, "step": 7016 }, { "epoch": 0.94, "learning_rate": 6.267539756782039e-09, "loss": 0.0707, "step": 7017 }, { "epoch": 0.94, "learning_rate": 6.254176132567151e-09, "loss": 0.0668, "step": 7018 }, { "epoch": 0.94, "learning_rate": 6.240812508352264e-09, "loss": 0.0535, "step": 7019 }, { "epoch": 0.94, "learning_rate": 6.2274488841373775e-09, "loss": 0.0722, "step": 7020 }, { "epoch": 0.94, "learning_rate": 6.214085259922491e-09, "loss": 0.0902, "step": 7021 }, { "epoch": 0.94, "learning_rate": 6.200721635707604e-09, "loss": 0.0855, "step": 7022 }, { "epoch": 0.94, "learning_rate": 6.187358011492717e-09, "loss": 0.0552, "step": 7023 }, { "epoch": 0.94, "learning_rate": 6.173994387277829e-09, "loss": 0.0592, "step": 7024 }, { "epoch": 0.94, "learning_rate": 6.160630763062942e-09, "loss": 0.0705, "step": 7025 }, { "epoch": 0.94, "learning_rate": 6.147267138848055e-09, "loss": 0.0416, "step": 7026 }, { "epoch": 0.94, "learning_rate": 6.133903514633168e-09, "loss": 0.1984, "step": 7027 }, { "epoch": 0.94, "learning_rate": 6.120539890418281e-09, "loss": 0.068, "step": 7028 }, { "epoch": 0.94, "learning_rate": 6.107176266203394e-09, "loss": 0.094, "step": 7029 }, { "epoch": 0.94, "learning_rate": 6.093812641988507e-09, "loss": 0.0748, "step": 7030 }, { "epoch": 0.94, "learning_rate": 6.0804490177736204e-09, "loss": 0.0737, "step": 7031 }, { "epoch": 0.94, "learning_rate": 6.067085393558733e-09, "loss": 0.0483, "step": 7032 }, { "epoch": 0.94, "learning_rate": 6.053721769343846e-09, "loss": 0.0881, "step": 7033 }, { "epoch": 0.94, "learning_rate": 6.040358145128958e-09, "loss": 0.0571, "step": 7034 }, { "epoch": 0.94, "learning_rate": 6.026994520914071e-09, "loss": 0.0481, "step": 7035 }, { "epoch": 0.94, "learning_rate": 6.0136308966991845e-09, "loss": 0.0956, "step": 7036 }, { "epoch": 0.94, "learning_rate": 6.000267272484298e-09, "loss": 0.0569, "step": 7037 }, { "epoch": 0.94, "learning_rate": 5.986903648269411e-09, "loss": 0.0445, "step": 7038 }, { "epoch": 0.94, "learning_rate": 5.973540024054523e-09, "loss": 0.0496, "step": 7039 }, { "epoch": 0.94, "learning_rate": 5.960176399839636e-09, "loss": 0.0777, "step": 7040 }, { "epoch": 0.94, "learning_rate": 5.946812775624749e-09, "loss": 0.0695, "step": 7041 }, { "epoch": 0.94, "learning_rate": 5.933449151409862e-09, "loss": 0.0593, "step": 7042 }, { "epoch": 0.94, "learning_rate": 5.920085527194975e-09, "loss": 0.0586, "step": 7043 }, { "epoch": 0.94, "learning_rate": 5.906721902980088e-09, "loss": 0.0488, "step": 7044 }, { "epoch": 0.94, "learning_rate": 5.893358278765201e-09, "loss": 0.071, "step": 7045 }, { "epoch": 0.94, "learning_rate": 5.879994654550314e-09, "loss": 0.0631, "step": 7046 }, { "epoch": 0.94, "learning_rate": 5.8666310303354266e-09, "loss": 0.0782, "step": 7047 }, { "epoch": 0.94, "learning_rate": 5.85326740612054e-09, "loss": 0.0698, "step": 7048 }, { "epoch": 0.94, "learning_rate": 5.85326740612054e-09, "loss": 0.0668, "step": 7049 }, { "epoch": 0.94, "learning_rate": 5.839903781905652e-09, "loss": 0.0532, "step": 7050 }, { "epoch": 0.94, "learning_rate": 5.826540157690765e-09, "loss": 0.0432, "step": 7051 }, { "epoch": 0.94, "learning_rate": 5.813176533475878e-09, "loss": 0.051, "step": 7052 }, { "epoch": 0.94, "learning_rate": 5.7998129092609915e-09, "loss": 0.0484, "step": 7053 }, { "epoch": 0.94, "learning_rate": 5.786449285046105e-09, "loss": 0.0618, "step": 7054 }, { "epoch": 0.94, "learning_rate": 5.773085660831217e-09, "loss": 0.0676, "step": 7055 }, { "epoch": 0.94, "learning_rate": 5.75972203661633e-09, "loss": 0.0522, "step": 7056 }, { "epoch": 0.94, "learning_rate": 5.746358412401443e-09, "loss": 0.0513, "step": 7057 }, { "epoch": 0.94, "learning_rate": 5.7329947881865555e-09, "loss": 0.0851, "step": 7058 }, { "epoch": 0.94, "learning_rate": 5.719631163971669e-09, "loss": 0.077, "step": 7059 }, { "epoch": 0.94, "learning_rate": 5.706267539756782e-09, "loss": 0.0701, "step": 7060 }, { "epoch": 0.94, "learning_rate": 5.692903915541895e-09, "loss": 0.0565, "step": 7061 }, { "epoch": 0.94, "learning_rate": 5.679540291327008e-09, "loss": 0.0567, "step": 7062 }, { "epoch": 0.94, "learning_rate": 5.66617666711212e-09, "loss": 0.0618, "step": 7063 }, { "epoch": 0.94, "learning_rate": 5.6528130428972335e-09, "loss": 0.049, "step": 7064 }, { "epoch": 0.94, "learning_rate": 5.639449418682347e-09, "loss": 0.0535, "step": 7065 }, { "epoch": 0.94, "learning_rate": 5.626085794467459e-09, "loss": 0.0557, "step": 7066 }, { "epoch": 0.94, "learning_rate": 5.612722170252572e-09, "loss": 0.0444, "step": 7067 }, { "epoch": 0.94, "learning_rate": 5.599358546037685e-09, "loss": 0.0362, "step": 7068 }, { "epoch": 0.94, "learning_rate": 5.5859949218227984e-09, "loss": 0.0403, "step": 7069 }, { "epoch": 0.94, "learning_rate": 5.572631297607911e-09, "loss": 0.0882, "step": 7070 }, { "epoch": 0.94, "learning_rate": 5.559267673393024e-09, "loss": 0.0758, "step": 7071 }, { "epoch": 0.95, "learning_rate": 5.545904049178137e-09, "loss": 0.0603, "step": 7072 }, { "epoch": 0.95, "learning_rate": 5.532540424963249e-09, "loss": 0.0525, "step": 7073 }, { "epoch": 0.95, "learning_rate": 5.5191768007483625e-09, "loss": 0.0298, "step": 7074 }, { "epoch": 0.95, "learning_rate": 5.505813176533476e-09, "loss": 0.0602, "step": 7075 }, { "epoch": 0.95, "learning_rate": 5.492449552318589e-09, "loss": 0.0533, "step": 7076 }, { "epoch": 0.95, "learning_rate": 5.479085928103702e-09, "loss": 0.0603, "step": 7077 }, { "epoch": 0.95, "learning_rate": 5.465722303888814e-09, "loss": 0.045, "step": 7078 }, { "epoch": 0.95, "learning_rate": 5.452358679673927e-09, "loss": 0.0774, "step": 7079 }, { "epoch": 0.95, "learning_rate": 5.4389950554590405e-09, "loss": 0.084, "step": 7080 }, { "epoch": 0.95, "learning_rate": 5.425631431244153e-09, "loss": 0.0889, "step": 7081 }, { "epoch": 0.95, "learning_rate": 5.412267807029266e-09, "loss": 0.0666, "step": 7082 }, { "epoch": 0.95, "learning_rate": 5.398904182814379e-09, "loss": 0.0555, "step": 7083 }, { "epoch": 0.95, "learning_rate": 5.385540558599492e-09, "loss": 0.0353, "step": 7084 }, { "epoch": 0.95, "learning_rate": 5.3721769343846046e-09, "loss": 0.0545, "step": 7085 }, { "epoch": 0.95, "learning_rate": 5.358813310169718e-09, "loss": 0.0719, "step": 7086 }, { "epoch": 0.95, "learning_rate": 5.345449685954831e-09, "loss": 0.0938, "step": 7087 }, { "epoch": 0.95, "learning_rate": 5.332086061739944e-09, "loss": 0.0405, "step": 7088 }, { "epoch": 0.95, "learning_rate": 5.318722437525056e-09, "loss": 0.076, "step": 7089 }, { "epoch": 0.95, "learning_rate": 5.3053588133101694e-09, "loss": 0.0472, "step": 7090 }, { "epoch": 0.95, "learning_rate": 5.2919951890952826e-09, "loss": 0.0765, "step": 7091 }, { "epoch": 0.95, "learning_rate": 5.278631564880395e-09, "loss": 0.0852, "step": 7092 }, { "epoch": 0.95, "learning_rate": 5.265267940665508e-09, "loss": 0.1493, "step": 7093 }, { "epoch": 0.95, "learning_rate": 5.251904316450621e-09, "loss": 0.0855, "step": 7094 }, { "epoch": 0.95, "learning_rate": 5.238540692235734e-09, "loss": 0.0647, "step": 7095 }, { "epoch": 0.95, "learning_rate": 5.2251770680208475e-09, "loss": 0.0602, "step": 7096 }, { "epoch": 0.95, "learning_rate": 5.21181344380596e-09, "loss": 0.0643, "step": 7097 }, { "epoch": 0.95, "learning_rate": 5.198449819591073e-09, "loss": 0.0735, "step": 7098 }, { "epoch": 0.95, "learning_rate": 5.185086195376186e-09, "loss": 0.0727, "step": 7099 }, { "epoch": 0.95, "learning_rate": 5.171722571161298e-09, "loss": 0.04, "step": 7100 }, { "epoch": 0.95, "learning_rate": 5.1583589469464115e-09, "loss": 0.0873, "step": 7101 }, { "epoch": 0.95, "learning_rate": 5.144995322731525e-09, "loss": 0.0491, "step": 7102 }, { "epoch": 0.95, "learning_rate": 5.131631698516638e-09, "loss": 0.0495, "step": 7103 }, { "epoch": 0.95, "learning_rate": 5.11826807430175e-09, "loss": 0.0739, "step": 7104 }, { "epoch": 0.95, "learning_rate": 5.104904450086863e-09, "loss": 0.0676, "step": 7105 }, { "epoch": 0.95, "learning_rate": 5.091540825871976e-09, "loss": 0.0847, "step": 7106 }, { "epoch": 0.95, "learning_rate": 5.078177201657089e-09, "loss": 0.0832, "step": 7107 }, { "epoch": 0.95, "learning_rate": 5.064813577442202e-09, "loss": 0.0808, "step": 7108 }, { "epoch": 0.95, "learning_rate": 5.051449953227315e-09, "loss": 0.0591, "step": 7109 }, { "epoch": 0.95, "learning_rate": 5.038086329012428e-09, "loss": 0.0696, "step": 7110 }, { "epoch": 0.95, "learning_rate": 5.024722704797541e-09, "loss": 0.0597, "step": 7111 }, { "epoch": 0.95, "learning_rate": 5.011359080582654e-09, "loss": 0.0501, "step": 7112 }, { "epoch": 0.95, "learning_rate": 4.997995456367767e-09, "loss": 0.0436, "step": 7113 }, { "epoch": 0.95, "learning_rate": 4.98463183215288e-09, "loss": 0.0648, "step": 7114 }, { "epoch": 0.95, "learning_rate": 4.971268207937992e-09, "loss": 0.07, "step": 7115 }, { "epoch": 0.95, "learning_rate": 4.957904583723105e-09, "loss": 0.0602, "step": 7116 }, { "epoch": 0.95, "learning_rate": 4.9445409595082185e-09, "loss": 0.04, "step": 7117 }, { "epoch": 0.95, "learning_rate": 4.931177335293332e-09, "loss": 0.0581, "step": 7118 }, { "epoch": 0.95, "learning_rate": 4.917813711078445e-09, "loss": 0.0946, "step": 7119 }, { "epoch": 0.95, "learning_rate": 4.904450086863557e-09, "loss": 0.0702, "step": 7120 }, { "epoch": 0.95, "learning_rate": 4.89108646264867e-09, "loss": 0.0601, "step": 7121 }, { "epoch": 0.95, "learning_rate": 4.8777228384337825e-09, "loss": 0.0628, "step": 7122 }, { "epoch": 0.95, "learning_rate": 4.864359214218896e-09, "loss": 0.0741, "step": 7123 }, { "epoch": 0.95, "learning_rate": 4.850995590004009e-09, "loss": 0.0788, "step": 7124 }, { "epoch": 0.95, "learning_rate": 4.837631965789122e-09, "loss": 0.0451, "step": 7125 }, { "epoch": 0.95, "learning_rate": 4.824268341574235e-09, "loss": 0.0312, "step": 7126 }, { "epoch": 0.95, "learning_rate": 4.810904717359348e-09, "loss": 0.0452, "step": 7127 }, { "epoch": 0.95, "learning_rate": 4.7975410931444606e-09, "loss": 0.0673, "step": 7128 }, { "epoch": 0.95, "learning_rate": 4.784177468929574e-09, "loss": 0.0566, "step": 7129 }, { "epoch": 0.95, "learning_rate": 4.770813844714686e-09, "loss": 0.072, "step": 7130 }, { "epoch": 0.95, "learning_rate": 4.757450220499799e-09, "loss": 0.0634, "step": 7131 }, { "epoch": 0.95, "learning_rate": 4.744086596284912e-09, "loss": 0.0677, "step": 7132 }, { "epoch": 0.95, "learning_rate": 4.7307229720700255e-09, "loss": 0.1053, "step": 7133 }, { "epoch": 0.95, "learning_rate": 4.717359347855139e-09, "loss": 0.0505, "step": 7134 }, { "epoch": 0.95, "learning_rate": 4.703995723640251e-09, "loss": 0.0552, "step": 7135 }, { "epoch": 0.95, "learning_rate": 4.690632099425364e-09, "loss": 0.3053, "step": 7136 }, { "epoch": 0.95, "learning_rate": 4.677268475210476e-09, "loss": 0.0631, "step": 7137 }, { "epoch": 0.95, "learning_rate": 4.6639048509955895e-09, "loss": 0.0549, "step": 7138 }, { "epoch": 0.95, "learning_rate": 4.650541226780703e-09, "loss": 0.0526, "step": 7139 }, { "epoch": 0.95, "learning_rate": 4.637177602565816e-09, "loss": 0.0674, "step": 7140 }, { "epoch": 0.95, "learning_rate": 4.623813978350929e-09, "loss": 0.0467, "step": 7141 }, { "epoch": 0.95, "learning_rate": 4.610450354136042e-09, "loss": 0.0687, "step": 7142 }, { "epoch": 0.95, "learning_rate": 4.597086729921154e-09, "loss": 0.0725, "step": 7143 }, { "epoch": 0.95, "learning_rate": 4.583723105706267e-09, "loss": 0.0805, "step": 7144 }, { "epoch": 0.95, "learning_rate": 4.57035948149138e-09, "loss": 0.038, "step": 7145 }, { "epoch": 0.95, "learning_rate": 4.556995857276493e-09, "loss": 0.0523, "step": 7146 }, { "epoch": 0.96, "learning_rate": 4.543632233061606e-09, "loss": 0.0629, "step": 7147 }, { "epoch": 0.96, "learning_rate": 4.530268608846719e-09, "loss": 0.073, "step": 7148 }, { "epoch": 0.96, "learning_rate": 4.5169049846318324e-09, "loss": 0.0673, "step": 7149 }, { "epoch": 0.96, "learning_rate": 4.5035413604169456e-09, "loss": 0.0817, "step": 7150 }, { "epoch": 0.96, "learning_rate": 4.490177736202058e-09, "loss": 0.0453, "step": 7151 }, { "epoch": 0.96, "learning_rate": 4.47681411198717e-09, "loss": 0.0579, "step": 7152 }, { "epoch": 0.96, "learning_rate": 4.463450487772283e-09, "loss": 0.06, "step": 7153 }, { "epoch": 0.96, "learning_rate": 4.4500868635573965e-09, "loss": 0.0664, "step": 7154 }, { "epoch": 0.96, "learning_rate": 4.43672323934251e-09, "loss": 0.0592, "step": 7155 }, { "epoch": 0.96, "learning_rate": 4.423359615127623e-09, "loss": 0.0692, "step": 7156 }, { "epoch": 0.96, "learning_rate": 4.409995990912736e-09, "loss": 0.0405, "step": 7157 }, { "epoch": 0.96, "learning_rate": 4.396632366697848e-09, "loss": 0.0894, "step": 7158 }, { "epoch": 0.96, "learning_rate": 4.3832687424829605e-09, "loss": 0.06, "step": 7159 }, { "epoch": 0.96, "learning_rate": 4.369905118268074e-09, "loss": 0.0768, "step": 7160 }, { "epoch": 0.96, "learning_rate": 4.356541494053187e-09, "loss": 0.0449, "step": 7161 }, { "epoch": 0.96, "learning_rate": 4.3431778698383e-09, "loss": 0.0564, "step": 7162 }, { "epoch": 0.96, "learning_rate": 4.329814245623413e-09, "loss": 0.0827, "step": 7163 }, { "epoch": 0.96, "learning_rate": 4.316450621408526e-09, "loss": 0.0803, "step": 7164 }, { "epoch": 0.96, "learning_rate": 4.303086997193639e-09, "loss": 0.1102, "step": 7165 }, { "epoch": 0.96, "learning_rate": 4.289723372978752e-09, "loss": 0.0736, "step": 7166 }, { "epoch": 0.96, "learning_rate": 4.276359748763864e-09, "loss": 0.0599, "step": 7167 }, { "epoch": 0.96, "learning_rate": 4.262996124548977e-09, "loss": 0.0483, "step": 7168 }, { "epoch": 0.96, "learning_rate": 4.24963250033409e-09, "loss": 0.0765, "step": 7169 }, { "epoch": 0.96, "learning_rate": 4.2362688761192034e-09, "loss": 0.0833, "step": 7170 }, { "epoch": 0.96, "learning_rate": 4.2229052519043166e-09, "loss": 0.0576, "step": 7171 }, { "epoch": 0.96, "learning_rate": 4.20954162768943e-09, "loss": 0.0683, "step": 7172 }, { "epoch": 0.96, "learning_rate": 4.196178003474542e-09, "loss": 0.0623, "step": 7173 }, { "epoch": 0.96, "learning_rate": 4.182814379259654e-09, "loss": 0.0949, "step": 7174 }, { "epoch": 0.96, "learning_rate": 4.1694507550447675e-09, "loss": 0.119, "step": 7175 }, { "epoch": 0.96, "learning_rate": 4.156087130829881e-09, "loss": 0.0586, "step": 7176 }, { "epoch": 0.96, "learning_rate": 4.142723506614994e-09, "loss": 0.0748, "step": 7177 }, { "epoch": 0.96, "learning_rate": 4.129359882400107e-09, "loss": 0.0421, "step": 7178 }, { "epoch": 0.96, "learning_rate": 4.11599625818522e-09, "loss": 0.0692, "step": 7179 }, { "epoch": 0.96, "learning_rate": 4.102632633970333e-09, "loss": 0.0621, "step": 7180 }, { "epoch": 0.96, "learning_rate": 4.0892690097554455e-09, "loss": 0.0578, "step": 7181 }, { "epoch": 0.96, "learning_rate": 4.075905385540558e-09, "loss": 0.0511, "step": 7182 }, { "epoch": 0.96, "learning_rate": 4.062541761325671e-09, "loss": 0.0654, "step": 7183 }, { "epoch": 0.96, "learning_rate": 4.049178137110784e-09, "loss": 0.0527, "step": 7184 }, { "epoch": 0.96, "learning_rate": 4.035814512895897e-09, "loss": 0.0545, "step": 7185 }, { "epoch": 0.96, "learning_rate": 4.02245088868101e-09, "loss": 0.0611, "step": 7186 }, { "epoch": 0.96, "learning_rate": 4.0090872644661235e-09, "loss": 0.0901, "step": 7187 }, { "epoch": 0.96, "learning_rate": 3.995723640251236e-09, "loss": 0.0698, "step": 7188 }, { "epoch": 0.96, "learning_rate": 3.982360016036349e-09, "loss": 0.0711, "step": 7189 }, { "epoch": 0.96, "learning_rate": 3.968996391821461e-09, "loss": 0.0495, "step": 7190 }, { "epoch": 0.96, "learning_rate": 3.9556327676065745e-09, "loss": 0.0586, "step": 7191 }, { "epoch": 0.96, "learning_rate": 3.942269143391688e-09, "loss": 0.0522, "step": 7192 }, { "epoch": 0.96, "learning_rate": 3.928905519176801e-09, "loss": 0.0465, "step": 7193 }, { "epoch": 0.96, "learning_rate": 3.915541894961914e-09, "loss": 0.0581, "step": 7194 }, { "epoch": 0.96, "learning_rate": 3.902178270747026e-09, "loss": 0.0537, "step": 7195 }, { "epoch": 0.96, "learning_rate": 3.888814646532139e-09, "loss": 0.065, "step": 7196 }, { "epoch": 0.96, "learning_rate": 3.875451022317252e-09, "loss": 0.0683, "step": 7197 }, { "epoch": 0.96, "learning_rate": 3.862087398102365e-09, "loss": 0.0737, "step": 7198 }, { "epoch": 0.96, "learning_rate": 3.848723773887478e-09, "loss": 0.0564, "step": 7199 }, { "epoch": 0.96, "learning_rate": 3.835360149672591e-09, "loss": 0.0805, "step": 7200 }, { "epoch": 0.96, "learning_rate": 3.821996525457704e-09, "loss": 0.0507, "step": 7201 }, { "epoch": 0.96, "learning_rate": 3.808632901242817e-09, "loss": 0.0562, "step": 7202 }, { "epoch": 0.96, "learning_rate": 3.79526927702793e-09, "loss": 0.0577, "step": 7203 }, { "epoch": 0.96, "learning_rate": 3.781905652813043e-09, "loss": 0.06, "step": 7204 }, { "epoch": 0.96, "learning_rate": 3.768542028598155e-09, "loss": 0.0639, "step": 7205 }, { "epoch": 0.96, "learning_rate": 3.755178404383268e-09, "loss": 0.0575, "step": 7206 }, { "epoch": 0.96, "learning_rate": 3.7418147801683814e-09, "loss": 0.1113, "step": 7207 }, { "epoch": 0.96, "learning_rate": 3.7284511559534946e-09, "loss": 0.0738, "step": 7208 }, { "epoch": 0.96, "learning_rate": 3.7150875317386073e-09, "loss": 0.0715, "step": 7209 }, { "epoch": 0.96, "learning_rate": 3.7017239075237204e-09, "loss": 0.1089, "step": 7210 }, { "epoch": 0.96, "learning_rate": 3.6883602833088336e-09, "loss": 0.0651, "step": 7211 }, { "epoch": 0.96, "learning_rate": 3.6749966590939463e-09, "loss": 0.0518, "step": 7212 }, { "epoch": 0.96, "learning_rate": 3.6616330348790586e-09, "loss": 0.1401, "step": 7213 }, { "epoch": 0.96, "learning_rate": 3.6482694106641718e-09, "loss": 0.0552, "step": 7214 }, { "epoch": 0.96, "learning_rate": 3.634905786449285e-09, "loss": 0.0581, "step": 7215 }, { "epoch": 0.96, "learning_rate": 3.6215421622343976e-09, "loss": 0.059, "step": 7216 }, { "epoch": 0.96, "learning_rate": 3.6081785380195108e-09, "loss": 0.0588, "step": 7217 }, { "epoch": 0.96, "learning_rate": 3.594814913804624e-09, "loss": 0.0613, "step": 7218 }, { "epoch": 0.96, "learning_rate": 3.5814512895897366e-09, "loss": 0.0563, "step": 7219 }, { "epoch": 0.96, "learning_rate": 3.5680876653748498e-09, "loss": 0.0907, "step": 7220 }, { "epoch": 0.96, "learning_rate": 3.554724041159962e-09, "loss": 0.0513, "step": 7221 }, { "epoch": 0.97, "learning_rate": 3.5413604169450752e-09, "loss": 0.0416, "step": 7222 }, { "epoch": 0.97, "learning_rate": 3.5279967927301884e-09, "loss": 0.0735, "step": 7223 }, { "epoch": 0.97, "learning_rate": 3.514633168515301e-09, "loss": 0.0892, "step": 7224 }, { "epoch": 0.97, "learning_rate": 3.5012695443004143e-09, "loss": 0.035, "step": 7225 }, { "epoch": 0.97, "learning_rate": 3.4879059200855274e-09, "loss": 0.0757, "step": 7226 }, { "epoch": 0.97, "learning_rate": 3.47454229587064e-09, "loss": 0.0579, "step": 7227 }, { "epoch": 0.97, "learning_rate": 3.4611786716557524e-09, "loss": 0.0503, "step": 7228 }, { "epoch": 0.97, "learning_rate": 3.4478150474408656e-09, "loss": 0.0822, "step": 7229 }, { "epoch": 0.97, "learning_rate": 3.4344514232259787e-09, "loss": 0.0613, "step": 7230 }, { "epoch": 0.97, "learning_rate": 3.4210877990110915e-09, "loss": 0.053, "step": 7231 }, { "epoch": 0.97, "learning_rate": 3.4077241747962046e-09, "loss": 0.0741, "step": 7232 }, { "epoch": 0.97, "learning_rate": 3.3943605505813177e-09, "loss": 0.0694, "step": 7233 }, { "epoch": 0.97, "learning_rate": 3.3809969263664305e-09, "loss": 0.0553, "step": 7234 }, { "epoch": 0.97, "learning_rate": 3.3676333021515436e-09, "loss": 0.0571, "step": 7235 }, { "epoch": 0.97, "learning_rate": 3.354269677936656e-09, "loss": 0.0499, "step": 7236 }, { "epoch": 0.97, "learning_rate": 3.340906053721769e-09, "loss": 0.0475, "step": 7237 }, { "epoch": 0.97, "learning_rate": 3.327542429506882e-09, "loss": 0.0822, "step": 7238 }, { "epoch": 0.97, "learning_rate": 3.314178805291995e-09, "loss": 0.0716, "step": 7239 }, { "epoch": 0.97, "learning_rate": 3.300815181077108e-09, "loss": 0.0553, "step": 7240 }, { "epoch": 0.97, "learning_rate": 3.2874515568622212e-09, "loss": 0.0566, "step": 7241 }, { "epoch": 0.97, "learning_rate": 3.274087932647334e-09, "loss": 0.0824, "step": 7242 }, { "epoch": 0.97, "learning_rate": 3.260724308432447e-09, "loss": 0.0508, "step": 7243 }, { "epoch": 0.97, "learning_rate": 3.2473606842175594e-09, "loss": 0.0671, "step": 7244 }, { "epoch": 0.97, "learning_rate": 3.2339970600026726e-09, "loss": 0.059, "step": 7245 }, { "epoch": 0.97, "learning_rate": 3.2206334357877853e-09, "loss": 0.0497, "step": 7246 }, { "epoch": 0.97, "learning_rate": 3.2072698115728984e-09, "loss": 0.0498, "step": 7247 }, { "epoch": 0.97, "learning_rate": 3.1939061873580116e-09, "loss": 0.0773, "step": 7248 }, { "epoch": 0.97, "learning_rate": 3.1805425631431243e-09, "loss": 0.0471, "step": 7249 }, { "epoch": 0.97, "learning_rate": 3.1671789389282374e-09, "loss": 0.0635, "step": 7250 }, { "epoch": 0.97, "learning_rate": 3.1538153147133497e-09, "loss": 0.0637, "step": 7251 }, { "epoch": 0.97, "learning_rate": 3.140451690498463e-09, "loss": 0.0478, "step": 7252 }, { "epoch": 0.97, "learning_rate": 3.1270880662835756e-09, "loss": 0.0702, "step": 7253 }, { "epoch": 0.97, "learning_rate": 3.1137244420686888e-09, "loss": 0.037, "step": 7254 }, { "epoch": 0.97, "learning_rate": 3.100360817853802e-09, "loss": 0.066, "step": 7255 }, { "epoch": 0.97, "learning_rate": 3.0869971936389146e-09, "loss": 0.0719, "step": 7256 }, { "epoch": 0.97, "learning_rate": 3.0736335694240274e-09, "loss": 0.0755, "step": 7257 }, { "epoch": 0.97, "learning_rate": 3.0602699452091405e-09, "loss": 0.0526, "step": 7258 }, { "epoch": 0.97, "learning_rate": 3.0469063209942536e-09, "loss": 0.0626, "step": 7259 }, { "epoch": 0.97, "learning_rate": 3.0335426967793664e-09, "loss": 0.0639, "step": 7260 }, { "epoch": 0.97, "learning_rate": 3.020179072564479e-09, "loss": 0.0959, "step": 7261 }, { "epoch": 0.97, "learning_rate": 3.0068154483495922e-09, "loss": 0.0528, "step": 7262 }, { "epoch": 0.97, "learning_rate": 2.9934518241347054e-09, "loss": 0.0595, "step": 7263 }, { "epoch": 0.97, "learning_rate": 2.980088199919818e-09, "loss": 0.0598, "step": 7264 }, { "epoch": 0.97, "learning_rate": 2.966724575704931e-09, "loss": 0.0685, "step": 7265 }, { "epoch": 0.97, "learning_rate": 2.953360951490044e-09, "loss": 0.072, "step": 7266 }, { "epoch": 0.97, "learning_rate": 2.939997327275157e-09, "loss": 0.0851, "step": 7267 }, { "epoch": 0.97, "learning_rate": 2.92663370306027e-09, "loss": 0.0604, "step": 7268 }, { "epoch": 0.97, "learning_rate": 2.9132700788453826e-09, "loss": 0.0339, "step": 7269 }, { "epoch": 0.97, "learning_rate": 2.8999064546304957e-09, "loss": 0.0779, "step": 7270 }, { "epoch": 0.97, "learning_rate": 2.8865428304156085e-09, "loss": 0.0727, "step": 7271 }, { "epoch": 0.97, "learning_rate": 2.8731792062007216e-09, "loss": 0.0722, "step": 7272 }, { "epoch": 0.97, "learning_rate": 2.8598155819858343e-09, "loss": 0.0598, "step": 7273 }, { "epoch": 0.97, "learning_rate": 2.8464519577709475e-09, "loss": 0.0523, "step": 7274 }, { "epoch": 0.97, "learning_rate": 2.83308833355606e-09, "loss": 0.0729, "step": 7275 }, { "epoch": 0.97, "learning_rate": 2.8197247093411733e-09, "loss": 0.076, "step": 7276 }, { "epoch": 0.97, "learning_rate": 2.806361085126286e-09, "loss": 0.0677, "step": 7277 }, { "epoch": 0.97, "learning_rate": 2.7929974609113992e-09, "loss": 0.064, "step": 7278 }, { "epoch": 0.97, "learning_rate": 2.779633836696512e-09, "loss": 0.0514, "step": 7279 }, { "epoch": 0.97, "learning_rate": 2.7662702124816247e-09, "loss": 0.0691, "step": 7280 }, { "epoch": 0.97, "learning_rate": 2.752906588266738e-09, "loss": 0.0507, "step": 7281 }, { "epoch": 0.97, "learning_rate": 2.739542964051851e-09, "loss": 0.0946, "step": 7282 }, { "epoch": 0.97, "learning_rate": 2.7261793398369637e-09, "loss": 0.0525, "step": 7283 }, { "epoch": 0.97, "learning_rate": 2.7128157156220764e-09, "loss": 0.0871, "step": 7284 }, { "epoch": 0.97, "learning_rate": 2.6994520914071895e-09, "loss": 0.0533, "step": 7285 }, { "epoch": 0.97, "learning_rate": 2.6860884671923023e-09, "loss": 0.0998, "step": 7286 }, { "epoch": 0.97, "learning_rate": 2.6727248429774154e-09, "loss": 0.054, "step": 7287 }, { "epoch": 0.97, "learning_rate": 2.659361218762528e-09, "loss": 0.0609, "step": 7288 }, { "epoch": 0.97, "learning_rate": 2.6459975945476413e-09, "loss": 0.0796, "step": 7289 }, { "epoch": 0.97, "learning_rate": 2.632633970332754e-09, "loss": 0.0497, "step": 7290 }, { "epoch": 0.97, "learning_rate": 2.619270346117867e-09, "loss": 0.0634, "step": 7291 }, { "epoch": 0.97, "learning_rate": 2.60590672190298e-09, "loss": 0.0471, "step": 7292 }, { "epoch": 0.97, "learning_rate": 2.592543097688093e-09, "loss": 0.0523, "step": 7293 }, { "epoch": 0.97, "learning_rate": 2.5791794734732058e-09, "loss": 0.0432, "step": 7294 }, { "epoch": 0.97, "learning_rate": 2.565815849258319e-09, "loss": 0.0496, "step": 7295 }, { "epoch": 0.98, "learning_rate": 2.5524522250434316e-09, "loss": 0.0833, "step": 7296 }, { "epoch": 0.98, "learning_rate": 2.5390886008285444e-09, "loss": 0.0412, "step": 7297 }, { "epoch": 0.98, "learning_rate": 2.5257249766136575e-09, "loss": 0.0903, "step": 7298 }, { "epoch": 0.98, "learning_rate": 2.5123613523987706e-09, "loss": 0.0592, "step": 7299 }, { "epoch": 0.98, "learning_rate": 2.4989977281838834e-09, "loss": 0.0559, "step": 7300 }, { "epoch": 0.98, "learning_rate": 2.485634103968996e-09, "loss": 0.0886, "step": 7301 }, { "epoch": 0.98, "learning_rate": 2.4722704797541092e-09, "loss": 0.0644, "step": 7302 }, { "epoch": 0.98, "learning_rate": 2.4589068555392224e-09, "loss": 0.0541, "step": 7303 }, { "epoch": 0.98, "learning_rate": 2.445543231324335e-09, "loss": 0.0722, "step": 7304 }, { "epoch": 0.98, "learning_rate": 2.432179607109448e-09, "loss": 0.0471, "step": 7305 }, { "epoch": 0.98, "learning_rate": 2.418815982894561e-09, "loss": 0.0704, "step": 7306 }, { "epoch": 0.98, "learning_rate": 2.405452358679674e-09, "loss": 0.046, "step": 7307 }, { "epoch": 0.98, "learning_rate": 2.392088734464787e-09, "loss": 0.0552, "step": 7308 }, { "epoch": 0.98, "learning_rate": 2.3787251102498996e-09, "loss": 0.0698, "step": 7309 }, { "epoch": 0.98, "learning_rate": 2.3653614860350127e-09, "loss": 0.0604, "step": 7310 }, { "epoch": 0.98, "learning_rate": 2.3519978618201255e-09, "loss": 0.0743, "step": 7311 }, { "epoch": 0.98, "learning_rate": 2.338634237605238e-09, "loss": 0.0648, "step": 7312 }, { "epoch": 0.98, "learning_rate": 2.3252706133903513e-09, "loss": 0.0362, "step": 7313 }, { "epoch": 0.98, "learning_rate": 2.3119069891754645e-09, "loss": 0.0584, "step": 7314 }, { "epoch": 0.98, "learning_rate": 2.298543364960577e-09, "loss": 0.0644, "step": 7315 }, { "epoch": 0.98, "learning_rate": 2.28517974074569e-09, "loss": 0.0711, "step": 7316 }, { "epoch": 0.98, "learning_rate": 2.271816116530803e-09, "loss": 0.0666, "step": 7317 }, { "epoch": 0.98, "learning_rate": 2.2584524923159162e-09, "loss": 0.0622, "step": 7318 }, { "epoch": 0.98, "learning_rate": 2.245088868101029e-09, "loss": 0.075, "step": 7319 }, { "epoch": 0.98, "learning_rate": 2.2317252438861417e-09, "loss": 0.0731, "step": 7320 }, { "epoch": 0.98, "learning_rate": 2.218361619671255e-09, "loss": 0.0372, "step": 7321 }, { "epoch": 0.98, "learning_rate": 2.204997995456368e-09, "loss": 0.0835, "step": 7322 }, { "epoch": 0.98, "learning_rate": 2.1916343712414803e-09, "loss": 0.0666, "step": 7323 }, { "epoch": 0.98, "learning_rate": 2.1782707470265934e-09, "loss": 0.0657, "step": 7324 }, { "epoch": 0.98, "learning_rate": 2.1649071228117065e-09, "loss": 0.0863, "step": 7325 }, { "epoch": 0.98, "learning_rate": 2.1515434985968197e-09, "loss": 0.0508, "step": 7326 }, { "epoch": 0.98, "learning_rate": 2.138179874381932e-09, "loss": 0.0624, "step": 7327 }, { "epoch": 0.98, "learning_rate": 2.124816250167045e-09, "loss": 0.0594, "step": 7328 }, { "epoch": 0.98, "learning_rate": 2.1114526259521583e-09, "loss": 0.0538, "step": 7329 }, { "epoch": 0.98, "learning_rate": 2.098089001737271e-09, "loss": 0.0598, "step": 7330 }, { "epoch": 0.98, "learning_rate": 2.0847253775223837e-09, "loss": 0.0843, "step": 7331 }, { "epoch": 0.98, "learning_rate": 2.071361753307497e-09, "loss": 0.0623, "step": 7332 }, { "epoch": 0.98, "learning_rate": 2.05799812909261e-09, "loss": 0.0813, "step": 7333 }, { "epoch": 0.98, "learning_rate": 2.0446345048777228e-09, "loss": 0.061, "step": 7334 }, { "epoch": 0.98, "learning_rate": 2.0312708806628355e-09, "loss": 0.0605, "step": 7335 }, { "epoch": 0.98, "learning_rate": 2.0179072564479486e-09, "loss": 0.0709, "step": 7336 }, { "epoch": 0.98, "learning_rate": 2.0045436322330618e-09, "loss": 0.0469, "step": 7337 }, { "epoch": 0.98, "learning_rate": 1.9911800080181745e-09, "loss": 0.0528, "step": 7338 }, { "epoch": 0.98, "learning_rate": 1.9778163838032872e-09, "loss": 0.0379, "step": 7339 }, { "epoch": 0.98, "learning_rate": 1.9644527595884004e-09, "loss": 0.0696, "step": 7340 }, { "epoch": 0.98, "learning_rate": 1.951089135373513e-09, "loss": 0.037, "step": 7341 }, { "epoch": 0.98, "learning_rate": 1.937725511158626e-09, "loss": 0.067, "step": 7342 }, { "epoch": 0.98, "learning_rate": 1.924361886943739e-09, "loss": 0.0987, "step": 7343 }, { "epoch": 0.98, "learning_rate": 1.910998262728852e-09, "loss": 0.0775, "step": 7344 }, { "epoch": 0.98, "learning_rate": 1.897634638513965e-09, "loss": 0.069, "step": 7345 }, { "epoch": 0.98, "learning_rate": 1.8842710142990776e-09, "loss": 0.0611, "step": 7346 }, { "epoch": 0.98, "learning_rate": 1.8709073900841907e-09, "loss": 0.0609, "step": 7347 }, { "epoch": 0.98, "learning_rate": 1.8575437658693036e-09, "loss": 0.0751, "step": 7348 }, { "epoch": 0.98, "learning_rate": 1.8441801416544168e-09, "loss": 0.0566, "step": 7349 }, { "epoch": 0.98, "learning_rate": 1.8308165174395293e-09, "loss": 0.0413, "step": 7350 }, { "epoch": 0.98, "learning_rate": 1.8174528932246425e-09, "loss": 0.0624, "step": 7351 }, { "epoch": 0.98, "learning_rate": 1.8040892690097554e-09, "loss": 0.0774, "step": 7352 }, { "epoch": 0.98, "learning_rate": 1.7907256447948683e-09, "loss": 0.0586, "step": 7353 }, { "epoch": 0.98, "learning_rate": 1.777362020579981e-09, "loss": 0.0607, "step": 7354 }, { "epoch": 0.98, "learning_rate": 1.7639983963650942e-09, "loss": 0.0712, "step": 7355 }, { "epoch": 0.98, "learning_rate": 1.7506347721502071e-09, "loss": 0.1369, "step": 7356 }, { "epoch": 0.98, "learning_rate": 1.73727114793532e-09, "loss": 0.0537, "step": 7357 }, { "epoch": 0.98, "learning_rate": 1.7239075237204328e-09, "loss": 0.0539, "step": 7358 }, { "epoch": 0.98, "learning_rate": 1.7105438995055457e-09, "loss": 0.0587, "step": 7359 }, { "epoch": 0.98, "learning_rate": 1.6971802752906589e-09, "loss": 0.0574, "step": 7360 }, { "epoch": 0.98, "learning_rate": 1.6838166510757718e-09, "loss": 0.0725, "step": 7361 }, { "epoch": 0.98, "learning_rate": 1.6704530268608845e-09, "loss": 0.0515, "step": 7362 }, { "epoch": 0.98, "learning_rate": 1.6570894026459975e-09, "loss": 0.073, "step": 7363 }, { "epoch": 0.98, "learning_rate": 1.6437257784311106e-09, "loss": 0.0651, "step": 7364 }, { "epoch": 0.98, "learning_rate": 1.6303621542162235e-09, "loss": 0.0472, "step": 7365 }, { "epoch": 0.98, "learning_rate": 1.6169985300013363e-09, "loss": 0.0512, "step": 7366 }, { "epoch": 0.98, "learning_rate": 1.6036349057864492e-09, "loss": 0.0957, "step": 7367 }, { "epoch": 0.98, "learning_rate": 1.5902712815715621e-09, "loss": 0.0799, "step": 7368 }, { "epoch": 0.98, "learning_rate": 1.5769076573566749e-09, "loss": 0.0412, "step": 7369 }, { "epoch": 0.98, "learning_rate": 1.5635440331417878e-09, "loss": 0.0751, "step": 7370 }, { "epoch": 0.99, "learning_rate": 1.550180408926901e-09, "loss": 0.0754, "step": 7371 }, { "epoch": 0.99, "learning_rate": 1.5368167847120137e-09, "loss": 0.0506, "step": 7372 }, { "epoch": 0.99, "learning_rate": 1.5234531604971268e-09, "loss": 0.0655, "step": 7373 }, { "epoch": 0.99, "learning_rate": 1.5100895362822396e-09, "loss": 0.0577, "step": 7374 }, { "epoch": 0.99, "learning_rate": 1.4967259120673527e-09, "loss": 0.0606, "step": 7375 }, { "epoch": 0.99, "learning_rate": 1.4833622878524654e-09, "loss": 0.0648, "step": 7376 }, { "epoch": 0.99, "learning_rate": 1.4699986636375786e-09, "loss": 0.0532, "step": 7377 }, { "epoch": 0.99, "learning_rate": 1.4566350394226913e-09, "loss": 0.0384, "step": 7378 }, { "epoch": 0.99, "learning_rate": 1.4432714152078042e-09, "loss": 0.0699, "step": 7379 }, { "epoch": 0.99, "learning_rate": 1.4299077909929172e-09, "loss": 0.0726, "step": 7380 }, { "epoch": 0.99, "learning_rate": 1.41654416677803e-09, "loss": 0.0733, "step": 7381 }, { "epoch": 0.99, "learning_rate": 1.403180542563143e-09, "loss": 0.0655, "step": 7382 }, { "epoch": 0.99, "learning_rate": 1.389816918348256e-09, "loss": 0.0572, "step": 7383 }, { "epoch": 0.99, "learning_rate": 1.376453294133369e-09, "loss": 0.0656, "step": 7384 }, { "epoch": 0.99, "learning_rate": 1.3630896699184818e-09, "loss": 0.0538, "step": 7385 }, { "epoch": 0.99, "learning_rate": 1.3497260457035948e-09, "loss": 0.0609, "step": 7386 }, { "epoch": 0.99, "learning_rate": 1.3363624214887077e-09, "loss": 0.0717, "step": 7387 }, { "epoch": 0.99, "learning_rate": 1.3229987972738206e-09, "loss": 0.1244, "step": 7388 }, { "epoch": 0.99, "learning_rate": 1.3096351730589336e-09, "loss": 0.0423, "step": 7389 }, { "epoch": 0.99, "learning_rate": 1.2962715488440465e-09, "loss": 0.0459, "step": 7390 }, { "epoch": 0.99, "learning_rate": 1.2829079246291595e-09, "loss": 0.0668, "step": 7391 }, { "epoch": 0.99, "learning_rate": 1.2695443004142722e-09, "loss": 0.0895, "step": 7392 }, { "epoch": 0.99, "learning_rate": 1.2561806761993853e-09, "loss": 0.0766, "step": 7393 }, { "epoch": 0.99, "learning_rate": 1.242817051984498e-09, "loss": 0.0589, "step": 7394 }, { "epoch": 0.99, "learning_rate": 1.2294534277696112e-09, "loss": 0.091, "step": 7395 }, { "epoch": 0.99, "learning_rate": 1.216089803554724e-09, "loss": 0.0744, "step": 7396 }, { "epoch": 0.99, "learning_rate": 1.202726179339837e-09, "loss": 0.0636, "step": 7397 }, { "epoch": 0.99, "learning_rate": 1.1893625551249498e-09, "loss": 0.0558, "step": 7398 }, { "epoch": 0.99, "learning_rate": 1.1759989309100627e-09, "loss": 0.0901, "step": 7399 }, { "epoch": 0.99, "learning_rate": 1.1626353066951757e-09, "loss": 0.0565, "step": 7400 }, { "epoch": 0.99, "learning_rate": 1.1492716824802886e-09, "loss": 0.0464, "step": 7401 }, { "epoch": 0.99, "learning_rate": 1.1359080582654015e-09, "loss": 0.0541, "step": 7402 }, { "epoch": 0.99, "learning_rate": 1.1225444340505145e-09, "loss": 0.0461, "step": 7403 }, { "epoch": 0.99, "learning_rate": 1.1091808098356274e-09, "loss": 0.0644, "step": 7404 }, { "epoch": 0.99, "learning_rate": 1.0958171856207401e-09, "loss": 0.0391, "step": 7405 }, { "epoch": 0.99, "learning_rate": 1.0824535614058533e-09, "loss": 0.0432, "step": 7406 }, { "epoch": 0.99, "learning_rate": 1.069089937190966e-09, "loss": 0.0653, "step": 7407 }, { "epoch": 0.99, "learning_rate": 1.0557263129760791e-09, "loss": 0.0745, "step": 7408 }, { "epoch": 0.99, "learning_rate": 1.0423626887611919e-09, "loss": 0.0397, "step": 7409 }, { "epoch": 0.99, "learning_rate": 1.028999064546305e-09, "loss": 0.0705, "step": 7410 }, { "epoch": 0.99, "learning_rate": 1.0156354403314177e-09, "loss": 0.0607, "step": 7411 }, { "epoch": 0.99, "learning_rate": 1.0022718161165309e-09, "loss": 0.1086, "step": 7412 }, { "epoch": 0.99, "learning_rate": 9.889081919016436e-10, "loss": 0.0413, "step": 7413 }, { "epoch": 0.99, "learning_rate": 9.755445676867565e-10, "loss": 0.0648, "step": 7414 }, { "epoch": 0.99, "learning_rate": 9.621809434718695e-10, "loss": 0.0736, "step": 7415 }, { "epoch": 0.99, "learning_rate": 9.488173192569824e-10, "loss": 0.066, "step": 7416 }, { "epoch": 0.99, "learning_rate": 9.354536950420954e-10, "loss": 0.1111, "step": 7417 }, { "epoch": 0.99, "learning_rate": 9.220900708272084e-10, "loss": 0.0565, "step": 7418 }, { "epoch": 0.99, "learning_rate": 9.087264466123212e-10, "loss": 0.0571, "step": 7419 }, { "epoch": 0.99, "learning_rate": 8.953628223974342e-10, "loss": 0.0484, "step": 7420 }, { "epoch": 0.99, "learning_rate": 8.819991981825471e-10, "loss": 0.0646, "step": 7421 }, { "epoch": 0.99, "learning_rate": 8.6863557396766e-10, "loss": 0.0598, "step": 7422 }, { "epoch": 0.99, "learning_rate": 8.552719497527729e-10, "loss": 0.0429, "step": 7423 }, { "epoch": 0.99, "learning_rate": 8.419083255378859e-10, "loss": 0.0644, "step": 7424 }, { "epoch": 0.99, "learning_rate": 8.285447013229987e-10, "loss": 0.0601, "step": 7425 }, { "epoch": 0.99, "learning_rate": 8.151810771081118e-10, "loss": 0.0639, "step": 7426 }, { "epoch": 0.99, "learning_rate": 8.018174528932246e-10, "loss": 0.0467, "step": 7427 }, { "epoch": 0.99, "learning_rate": 7.884538286783374e-10, "loss": 0.0873, "step": 7428 }, { "epoch": 0.99, "learning_rate": 7.750902044634505e-10, "loss": 0.0516, "step": 7429 }, { "epoch": 0.99, "learning_rate": 7.617265802485634e-10, "loss": 0.0562, "step": 7430 }, { "epoch": 0.99, "learning_rate": 7.483629560336763e-10, "loss": 0.0579, "step": 7431 }, { "epoch": 0.99, "learning_rate": 7.349993318187893e-10, "loss": 0.0531, "step": 7432 }, { "epoch": 0.99, "learning_rate": 7.216357076039021e-10, "loss": 0.044, "step": 7433 }, { "epoch": 0.99, "learning_rate": 7.08272083389015e-10, "loss": 0.0819, "step": 7434 }, { "epoch": 0.99, "learning_rate": 6.94908459174128e-10, "loss": 0.0796, "step": 7435 }, { "epoch": 0.99, "learning_rate": 6.815448349592409e-10, "loss": 0.039, "step": 7436 }, { "epoch": 0.99, "learning_rate": 6.681812107443539e-10, "loss": 0.0627, "step": 7437 }, { "epoch": 0.99, "learning_rate": 6.548175865294668e-10, "loss": 0.0483, "step": 7438 }, { "epoch": 0.99, "learning_rate": 6.414539623145797e-10, "loss": 0.0836, "step": 7439 }, { "epoch": 0.99, "learning_rate": 6.280903380996927e-10, "loss": 0.0542, "step": 7440 }, { "epoch": 0.99, "learning_rate": 6.147267138848056e-10, "loss": 0.0417, "step": 7441 }, { "epoch": 0.99, "learning_rate": 6.013630896699185e-10, "loss": 0.0536, "step": 7442 }, { "epoch": 0.99, "learning_rate": 5.879994654550314e-10, "loss": 0.0528, "step": 7443 }, { "epoch": 0.99, "learning_rate": 5.746358412401443e-10, "loss": 0.0486, "step": 7444 }, { "epoch": 0.99, "learning_rate": 5.612722170252572e-10, "loss": 0.0844, "step": 7445 }, { "epoch": 1.0, "learning_rate": 5.479085928103701e-10, "loss": 0.0423, "step": 7446 }, { "epoch": 1.0, "learning_rate": 5.34544968595483e-10, "loss": 0.0786, "step": 7447 }, { "epoch": 1.0, "learning_rate": 5.211813443805959e-10, "loss": 0.0435, "step": 7448 }, { "epoch": 1.0, "learning_rate": 5.078177201657089e-10, "loss": 0.0481, "step": 7449 }, { "epoch": 1.0, "learning_rate": 4.944540959508218e-10, "loss": 0.066, "step": 7450 }, { "epoch": 1.0, "learning_rate": 4.810904717359347e-10, "loss": 0.0842, "step": 7451 }, { "epoch": 1.0, "learning_rate": 4.677268475210477e-10, "loss": 0.0576, "step": 7452 }, { "epoch": 1.0, "learning_rate": 4.543632233061606e-10, "loss": 0.0555, "step": 7453 }, { "epoch": 1.0, "learning_rate": 4.4099959909127355e-10, "loss": 0.0424, "step": 7454 }, { "epoch": 1.0, "learning_rate": 4.2763597487638643e-10, "loss": 0.0638, "step": 7455 }, { "epoch": 1.0, "learning_rate": 4.1427235066149937e-10, "loss": 0.0468, "step": 7456 }, { "epoch": 1.0, "learning_rate": 4.009087264466123e-10, "loss": 0.097, "step": 7457 }, { "epoch": 1.0, "learning_rate": 3.8754510223172524e-10, "loss": 0.0638, "step": 7458 }, { "epoch": 1.0, "learning_rate": 3.7418147801683817e-10, "loss": 0.0714, "step": 7459 }, { "epoch": 1.0, "learning_rate": 3.6081785380195106e-10, "loss": 0.0449, "step": 7460 }, { "epoch": 1.0, "learning_rate": 3.47454229587064e-10, "loss": 0.041, "step": 7461 }, { "epoch": 1.0, "learning_rate": 3.3409060537217693e-10, "loss": 0.0888, "step": 7462 }, { "epoch": 1.0, "learning_rate": 3.2072698115728986e-10, "loss": 0.108, "step": 7463 }, { "epoch": 1.0, "learning_rate": 3.073633569424028e-10, "loss": 0.0605, "step": 7464 }, { "epoch": 1.0, "learning_rate": 2.939997327275157e-10, "loss": 0.0579, "step": 7465 }, { "epoch": 1.0, "learning_rate": 2.806361085126286e-10, "loss": 0.0506, "step": 7466 }, { "epoch": 1.0, "learning_rate": 2.672724842977415e-10, "loss": 0.0819, "step": 7467 }, { "epoch": 1.0, "learning_rate": 2.5390886008285444e-10, "loss": 0.0577, "step": 7468 }, { "epoch": 1.0, "learning_rate": 2.4054523586796737e-10, "loss": 0.0743, "step": 7469 }, { "epoch": 1.0, "learning_rate": 2.271816116530803e-10, "loss": 0.1112, "step": 7470 }, { "epoch": 1.0, "learning_rate": 2.1381798743819322e-10, "loss": 0.0573, "step": 7471 }, { "epoch": 1.0, "learning_rate": 2.0045436322330615e-10, "loss": 0.0801, "step": 7472 }, { "epoch": 1.0, "learning_rate": 1.8709073900841909e-10, "loss": 0.058, "step": 7473 }, { "epoch": 1.0, "learning_rate": 1.73727114793532e-10, "loss": 0.0483, "step": 7474 }, { "epoch": 1.0, "learning_rate": 1.6036349057864493e-10, "loss": 0.0422, "step": 7475 }, { "epoch": 1.0, "learning_rate": 1.4699986636375784e-10, "loss": 0.0476, "step": 7476 }, { "epoch": 1.0, "learning_rate": 1.3363624214887075e-10, "loss": 0.0614, "step": 7477 }, { "epoch": 1.0, "learning_rate": 1.2027261793398369e-10, "loss": 0.0549, "step": 7478 }, { "epoch": 1.0, "learning_rate": 1.0690899371909661e-10, "loss": 0.086, "step": 7479 }, { "epoch": 1.0, "learning_rate": 9.354536950420954e-11, "loss": 0.0414, "step": 7480 }, { "epoch": 1.0, "learning_rate": 8.018174528932247e-11, "loss": 0.0587, "step": 7481 }, { "epoch": 1.0, "learning_rate": 6.681812107443538e-11, "loss": 0.052, "step": 7482 }, { "epoch": 1.0, "learning_rate": 5.3454496859548304e-11, "loss": 0.0653, "step": 7483 }, { "epoch": 1.0, "eval_loss": 0.12977056205272675, "eval_runtime": 2751.0222, "eval_samples_per_second": 5.838, "eval_steps_per_second": 0.182, "eval_wer": 0.19192736670293797, "step": 7483 }, { "epoch": 1.0, "step": 7483, "total_flos": 3.094611758752735e+20, "train_loss": 0.05640954649210032, "train_runtime": 111640.7688, "train_samples_per_second": 4.29, "train_steps_per_second": 0.067 } ], "max_steps": 7483, "num_train_epochs": 1, "total_flos": 3.094611758752735e+20, "trial_name": null, "trial_params": null }