{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 3756, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.7699115044247788e-07, "loss": 4.0879, "step": 1 }, { "epoch": 0.0, "learning_rate": 3.5398230088495575e-07, "loss": 4.8096, "step": 2 }, { "epoch": 0.0, "learning_rate": 5.309734513274336e-07, "loss": 4.9408, "step": 3 }, { "epoch": 0.0, "learning_rate": 7.079646017699115e-07, "loss": 5.0644, "step": 4 }, { "epoch": 0.0, "learning_rate": 8.849557522123895e-07, "loss": 5.0214, "step": 5 }, { "epoch": 0.0, "learning_rate": 1.0619469026548673e-06, "loss": 5.0348, "step": 6 }, { "epoch": 0.0, "learning_rate": 1.2389380530973452e-06, "loss": 5.0261, "step": 7 }, { "epoch": 0.0, "learning_rate": 1.415929203539823e-06, "loss": 4.8111, "step": 8 }, { "epoch": 0.0, "learning_rate": 1.592920353982301e-06, "loss": 4.8649, "step": 9 }, { "epoch": 0.0, "learning_rate": 1.769911504424779e-06, "loss": 4.8561, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.9469026548672567e-06, "loss": 4.688, "step": 11 }, { "epoch": 0.0, "learning_rate": 2.1238938053097345e-06, "loss": 4.6344, "step": 12 }, { "epoch": 0.0, "learning_rate": 2.3008849557522127e-06, "loss": 4.6299, "step": 13 }, { "epoch": 0.0, "learning_rate": 2.4778761061946905e-06, "loss": 4.6214, "step": 14 }, { "epoch": 0.0, "learning_rate": 2.6548672566371687e-06, "loss": 4.4317, "step": 15 }, { "epoch": 0.0, "learning_rate": 2.831858407079646e-06, "loss": 4.2308, "step": 16 }, { "epoch": 0.0, "learning_rate": 3.0088495575221242e-06, "loss": 4.0902, "step": 17 }, { "epoch": 0.0, "learning_rate": 3.185840707964602e-06, "loss": 3.9157, "step": 18 }, { "epoch": 0.01, "learning_rate": 3.36283185840708e-06, "loss": 3.9216, "step": 19 }, { "epoch": 0.01, "learning_rate": 3.539823008849558e-06, "loss": 3.7463, "step": 20 }, { "epoch": 0.01, "learning_rate": 3.7168141592920357e-06, "loss": 3.6445, "step": 21 }, { "epoch": 0.01, "learning_rate": 3.8938053097345135e-06, "loss": 3.5936, "step": 22 }, { "epoch": 0.01, "learning_rate": 4.070796460176992e-06, "loss": 3.5806, "step": 23 }, { "epoch": 0.01, "learning_rate": 4.247787610619469e-06, "loss": 3.6178, "step": 24 }, { "epoch": 0.01, "learning_rate": 4.424778761061948e-06, "loss": 3.5333, "step": 25 }, { "epoch": 0.01, "learning_rate": 4.6017699115044254e-06, "loss": 3.6075, "step": 26 }, { "epoch": 0.01, "learning_rate": 4.778761061946903e-06, "loss": 3.5291, "step": 27 }, { "epoch": 0.01, "learning_rate": 4.955752212389381e-06, "loss": 3.4853, "step": 28 }, { "epoch": 0.01, "learning_rate": 5.132743362831859e-06, "loss": 3.4344, "step": 29 }, { "epoch": 0.01, "learning_rate": 5.309734513274337e-06, "loss": 3.3966, "step": 30 }, { "epoch": 0.01, "learning_rate": 5.486725663716814e-06, "loss": 3.3734, "step": 31 }, { "epoch": 0.01, "learning_rate": 5.663716814159292e-06, "loss": 3.2909, "step": 32 }, { "epoch": 0.01, "learning_rate": 5.840707964601771e-06, "loss": 3.2473, "step": 33 }, { "epoch": 0.01, "learning_rate": 6.0176991150442484e-06, "loss": 3.1957, "step": 34 }, { "epoch": 0.01, "learning_rate": 6.194690265486726e-06, "loss": 3.2262, "step": 35 }, { "epoch": 0.01, "learning_rate": 6.371681415929204e-06, "loss": 3.1904, "step": 36 }, { "epoch": 0.01, "learning_rate": 6.548672566371682e-06, "loss": 3.1702, "step": 37 }, { "epoch": 0.01, "learning_rate": 6.72566371681416e-06, "loss": 3.1249, "step": 38 }, { "epoch": 0.01, "learning_rate": 6.902654867256637e-06, "loss": 3.1694, "step": 39 }, { "epoch": 0.01, "learning_rate": 7.079646017699116e-06, "loss": 3.1659, "step": 40 }, { "epoch": 0.01, "learning_rate": 7.256637168141594e-06, "loss": 3.0438, "step": 41 }, { "epoch": 0.01, "learning_rate": 7.4336283185840714e-06, "loss": 3.0664, "step": 42 }, { "epoch": 0.01, "learning_rate": 7.610619469026549e-06, "loss": 3.0374, "step": 43 }, { "epoch": 0.01, "learning_rate": 7.787610619469027e-06, "loss": 2.9876, "step": 44 }, { "epoch": 0.01, "learning_rate": 7.964601769911505e-06, "loss": 3.0711, "step": 45 }, { "epoch": 0.01, "learning_rate": 8.141592920353984e-06, "loss": 3.0268, "step": 46 }, { "epoch": 0.01, "learning_rate": 8.31858407079646e-06, "loss": 2.9807, "step": 47 }, { "epoch": 0.01, "learning_rate": 8.495575221238938e-06, "loss": 2.9752, "step": 48 }, { "epoch": 0.01, "learning_rate": 8.672566371681418e-06, "loss": 2.9747, "step": 49 }, { "epoch": 0.01, "learning_rate": 8.849557522123895e-06, "loss": 2.9257, "step": 50 }, { "epoch": 0.01, "learning_rate": 9.026548672566371e-06, "loss": 2.963, "step": 51 }, { "epoch": 0.01, "learning_rate": 9.203539823008851e-06, "loss": 2.9695, "step": 52 }, { "epoch": 0.01, "learning_rate": 9.380530973451329e-06, "loss": 2.9098, "step": 53 }, { "epoch": 0.01, "learning_rate": 9.557522123893806e-06, "loss": 2.9543, "step": 54 }, { "epoch": 0.01, "learning_rate": 9.734513274336284e-06, "loss": 2.8488, "step": 55 }, { "epoch": 0.01, "learning_rate": 9.911504424778762e-06, "loss": 2.8434, "step": 56 }, { "epoch": 0.02, "learning_rate": 1.008849557522124e-05, "loss": 2.8618, "step": 57 }, { "epoch": 0.02, "learning_rate": 1.0265486725663717e-05, "loss": 2.8717, "step": 58 }, { "epoch": 0.02, "learning_rate": 1.0442477876106197e-05, "loss": 2.8625, "step": 59 }, { "epoch": 0.02, "learning_rate": 1.0619469026548675e-05, "loss": 2.8286, "step": 60 }, { "epoch": 0.02, "learning_rate": 1.079646017699115e-05, "loss": 2.7342, "step": 61 }, { "epoch": 0.02, "learning_rate": 1.0973451327433629e-05, "loss": 2.7499, "step": 62 }, { "epoch": 0.02, "learning_rate": 1.1150442477876106e-05, "loss": 2.7514, "step": 63 }, { "epoch": 0.02, "learning_rate": 1.1327433628318584e-05, "loss": 2.7256, "step": 64 }, { "epoch": 0.02, "learning_rate": 1.1504424778761064e-05, "loss": 2.7508, "step": 65 }, { "epoch": 0.02, "learning_rate": 1.1681415929203541e-05, "loss": 2.7303, "step": 66 }, { "epoch": 0.02, "learning_rate": 1.1858407079646019e-05, "loss": 2.6827, "step": 67 }, { "epoch": 0.02, "learning_rate": 1.2035398230088497e-05, "loss": 2.7102, "step": 68 }, { "epoch": 0.02, "learning_rate": 1.2212389380530973e-05, "loss": 2.6663, "step": 69 }, { "epoch": 0.02, "learning_rate": 1.2389380530973452e-05, "loss": 2.6559, "step": 70 }, { "epoch": 0.02, "learning_rate": 1.256637168141593e-05, "loss": 2.6232, "step": 71 }, { "epoch": 0.02, "learning_rate": 1.2743362831858408e-05, "loss": 2.5894, "step": 72 }, { "epoch": 0.02, "learning_rate": 1.2920353982300886e-05, "loss": 2.5931, "step": 73 }, { "epoch": 0.02, "learning_rate": 1.3097345132743363e-05, "loss": 2.5585, "step": 74 }, { "epoch": 0.02, "learning_rate": 1.3274336283185843e-05, "loss": 2.5667, "step": 75 }, { "epoch": 0.02, "learning_rate": 1.345132743362832e-05, "loss": 2.5338, "step": 76 }, { "epoch": 0.02, "learning_rate": 1.3628318584070797e-05, "loss": 2.5103, "step": 77 }, { "epoch": 0.02, "learning_rate": 1.3805309734513275e-05, "loss": 2.4776, "step": 78 }, { "epoch": 0.02, "learning_rate": 1.3982300884955752e-05, "loss": 2.4765, "step": 79 }, { "epoch": 0.02, "learning_rate": 1.4159292035398232e-05, "loss": 2.4878, "step": 80 }, { "epoch": 0.02, "learning_rate": 1.433628318584071e-05, "loss": 2.4424, "step": 81 }, { "epoch": 0.02, "learning_rate": 1.4513274336283187e-05, "loss": 2.3933, "step": 82 }, { "epoch": 0.02, "learning_rate": 1.4690265486725665e-05, "loss": 2.4143, "step": 83 }, { "epoch": 0.02, "learning_rate": 1.4867256637168143e-05, "loss": 2.3407, "step": 84 }, { "epoch": 0.02, "learning_rate": 1.5044247787610619e-05, "loss": 2.3932, "step": 85 }, { "epoch": 0.02, "learning_rate": 1.5221238938053098e-05, "loss": 2.3102, "step": 86 }, { "epoch": 0.02, "learning_rate": 1.5398230088495576e-05, "loss": 2.331, "step": 87 }, { "epoch": 0.02, "learning_rate": 1.5575221238938054e-05, "loss": 2.3002, "step": 88 }, { "epoch": 0.02, "learning_rate": 1.5752212389380532e-05, "loss": 2.3373, "step": 89 }, { "epoch": 0.02, "learning_rate": 1.592920353982301e-05, "loss": 2.2986, "step": 90 }, { "epoch": 0.02, "learning_rate": 1.6106194690265487e-05, "loss": 2.2915, "step": 91 }, { "epoch": 0.02, "learning_rate": 1.628318584070797e-05, "loss": 2.2597, "step": 92 }, { "epoch": 0.02, "learning_rate": 1.6460176991150443e-05, "loss": 2.274, "step": 93 }, { "epoch": 0.03, "learning_rate": 1.663716814159292e-05, "loss": 2.2741, "step": 94 }, { "epoch": 0.03, "learning_rate": 1.68141592920354e-05, "loss": 2.2669, "step": 95 }, { "epoch": 0.03, "learning_rate": 1.6991150442477876e-05, "loss": 2.2253, "step": 96 }, { "epoch": 0.03, "learning_rate": 1.7168141592920354e-05, "loss": 2.2241, "step": 97 }, { "epoch": 0.03, "learning_rate": 1.7345132743362835e-05, "loss": 2.1995, "step": 98 }, { "epoch": 0.03, "learning_rate": 1.7522123893805313e-05, "loss": 2.1696, "step": 99 }, { "epoch": 0.03, "learning_rate": 1.769911504424779e-05, "loss": 2.1811, "step": 100 }, { "epoch": 0.03, "learning_rate": 1.7876106194690265e-05, "loss": 2.1415, "step": 101 }, { "epoch": 0.03, "learning_rate": 1.8053097345132743e-05, "loss": 2.0982, "step": 102 }, { "epoch": 0.03, "learning_rate": 1.823008849557522e-05, "loss": 2.1441, "step": 103 }, { "epoch": 0.03, "learning_rate": 1.8407079646017702e-05, "loss": 2.1191, "step": 104 }, { "epoch": 0.03, "learning_rate": 1.858407079646018e-05, "loss": 2.0983, "step": 105 }, { "epoch": 0.03, "learning_rate": 1.8761061946902657e-05, "loss": 2.0711, "step": 106 }, { "epoch": 0.03, "learning_rate": 1.8938053097345135e-05, "loss": 2.1227, "step": 107 }, { "epoch": 0.03, "learning_rate": 1.9115044247787613e-05, "loss": 2.0826, "step": 108 }, { "epoch": 0.03, "learning_rate": 1.929203539823009e-05, "loss": 2.0979, "step": 109 }, { "epoch": 0.03, "learning_rate": 1.946902654867257e-05, "loss": 2.0958, "step": 110 }, { "epoch": 0.03, "learning_rate": 1.9646017699115046e-05, "loss": 2.07, "step": 111 }, { "epoch": 0.03, "learning_rate": 1.9823008849557524e-05, "loss": 2.091, "step": 112 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 2.1325, "step": 113 }, { "epoch": 0.03, "learning_rate": 1.9999996281640465e-05, "loss": 2.1855, "step": 114 }, { "epoch": 0.03, "learning_rate": 1.9999985126564622e-05, "loss": 2.1566, "step": 115 }, { "epoch": 0.03, "learning_rate": 1.999996653478077e-05, "loss": 2.1338, "step": 116 }, { "epoch": 0.03, "learning_rate": 1.9999940506302727e-05, "loss": 2.1435, "step": 117 }, { "epoch": 0.03, "learning_rate": 1.999990704114986e-05, "loss": 2.1201, "step": 118 }, { "epoch": 0.03, "learning_rate": 1.9999866139347052e-05, "loss": 2.1777, "step": 119 }, { "epoch": 0.03, "learning_rate": 1.9999817800924717e-05, "loss": 2.1683, "step": 120 }, { "epoch": 0.03, "learning_rate": 1.999976202591881e-05, "loss": 2.1269, "step": 121 }, { "epoch": 0.03, "learning_rate": 1.9999698814370803e-05, "loss": 2.114, "step": 122 }, { "epoch": 0.03, "learning_rate": 1.9999628166327707e-05, "loss": 2.0685, "step": 123 }, { "epoch": 0.03, "learning_rate": 1.9999550081842062e-05, "loss": 2.0872, "step": 124 }, { "epoch": 0.03, "learning_rate": 1.9999464560971935e-05, "loss": 2.0964, "step": 125 }, { "epoch": 0.03, "learning_rate": 1.999937160378093e-05, "loss": 2.0872, "step": 126 }, { "epoch": 0.03, "learning_rate": 1.999927121033817e-05, "loss": 2.0512, "step": 127 }, { "epoch": 0.03, "learning_rate": 1.999916338071832e-05, "loss": 2.1166, "step": 128 }, { "epoch": 0.03, "learning_rate": 1.9999048115001566e-05, "loss": 2.04, "step": 129 }, { "epoch": 0.03, "learning_rate": 1.999892541327363e-05, "loss": 2.0732, "step": 130 }, { "epoch": 0.03, "learning_rate": 1.9998795275625763e-05, "loss": 2.0859, "step": 131 }, { "epoch": 0.04, "learning_rate": 1.9998657702154746e-05, "loss": 2.0228, "step": 132 }, { "epoch": 0.04, "learning_rate": 1.999851269296288e-05, "loss": 1.9826, "step": 133 }, { "epoch": 0.04, "learning_rate": 1.9998360248158013e-05, "loss": 1.9849, "step": 134 }, { "epoch": 0.04, "learning_rate": 1.999820036785351e-05, "loss": 2.0534, "step": 135 }, { "epoch": 0.04, "learning_rate": 1.9998033052168276e-05, "loss": 2.0085, "step": 136 }, { "epoch": 0.04, "learning_rate": 1.9997858301226728e-05, "loss": 2.0729, "step": 137 }, { "epoch": 0.04, "learning_rate": 1.999767611515883e-05, "loss": 2.04, "step": 138 }, { "epoch": 0.04, "learning_rate": 1.999748649410007e-05, "loss": 2.0764, "step": 139 }, { "epoch": 0.04, "learning_rate": 1.9997289438191457e-05, "loss": 1.9994, "step": 140 }, { "epoch": 0.04, "learning_rate": 1.9997084947579544e-05, "loss": 1.9897, "step": 141 }, { "epoch": 0.04, "learning_rate": 1.99968730224164e-05, "loss": 2.0508, "step": 142 }, { "epoch": 0.04, "learning_rate": 1.9996653662859632e-05, "loss": 2.029, "step": 143 }, { "epoch": 0.04, "learning_rate": 1.9996426869072367e-05, "loss": 1.9791, "step": 144 }, { "epoch": 0.04, "learning_rate": 1.9996192641223265e-05, "loss": 2.0017, "step": 145 }, { "epoch": 0.04, "learning_rate": 1.9995950979486518e-05, "loss": 2.0151, "step": 146 }, { "epoch": 0.04, "learning_rate": 1.999570188404184e-05, "loss": 2.0199, "step": 147 }, { "epoch": 0.04, "learning_rate": 1.999544535507448e-05, "loss": 2.0143, "step": 148 }, { "epoch": 0.04, "learning_rate": 1.9995181392775208e-05, "loss": 2.0097, "step": 149 }, { "epoch": 0.04, "learning_rate": 1.9994909997340328e-05, "loss": 2.0113, "step": 150 }, { "epoch": 0.04, "learning_rate": 1.9994631168971667e-05, "loss": 1.9801, "step": 151 }, { "epoch": 0.04, "learning_rate": 1.9994344907876584e-05, "loss": 2.0, "step": 152 }, { "epoch": 0.04, "learning_rate": 1.999405121426796e-05, "loss": 2.035, "step": 153 }, { "epoch": 0.04, "learning_rate": 1.999375008836421e-05, "loss": 1.9994, "step": 154 }, { "epoch": 0.04, "learning_rate": 1.999344153038927e-05, "loss": 2.0372, "step": 155 }, { "epoch": 0.04, "learning_rate": 1.999312554057261e-05, "loss": 2.0257, "step": 156 }, { "epoch": 0.04, "learning_rate": 1.9992802119149216e-05, "loss": 2.0091, "step": 157 }, { "epoch": 0.04, "learning_rate": 1.9992471266359613e-05, "loss": 1.9539, "step": 158 }, { "epoch": 0.04, "learning_rate": 1.9992132982449845e-05, "loss": 1.9965, "step": 159 }, { "epoch": 0.04, "learning_rate": 1.9991787267671486e-05, "loss": 1.9886, "step": 160 }, { "epoch": 0.04, "learning_rate": 1.9991434122281633e-05, "loss": 2.0006, "step": 161 }, { "epoch": 0.04, "learning_rate": 1.9991073546542912e-05, "loss": 1.952, "step": 162 }, { "epoch": 0.04, "learning_rate": 1.999070554072347e-05, "loss": 1.9921, "step": 163 }, { "epoch": 0.04, "learning_rate": 1.999033010509698e-05, "loss": 2.0151, "step": 164 }, { "epoch": 0.04, "learning_rate": 1.9989947239942653e-05, "loss": 1.9579, "step": 165 }, { "epoch": 0.04, "learning_rate": 1.998955694554521e-05, "loss": 1.9629, "step": 166 }, { "epoch": 0.04, "learning_rate": 1.9989159222194895e-05, "loss": 1.9503, "step": 167 }, { "epoch": 0.04, "learning_rate": 1.998875407018749e-05, "loss": 1.965, "step": 168 }, { "epoch": 0.04, "learning_rate": 1.99883414898243e-05, "loss": 1.8939, "step": 169 }, { "epoch": 0.05, "learning_rate": 1.998792148141214e-05, "loss": 1.9407, "step": 170 }, { "epoch": 0.05, "learning_rate": 1.9987494045263365e-05, "loss": 1.9211, "step": 171 }, { "epoch": 0.05, "learning_rate": 1.9987059181695843e-05, "loss": 1.977, "step": 172 }, { "epoch": 0.05, "learning_rate": 1.9986616891032975e-05, "loss": 1.9666, "step": 173 }, { "epoch": 0.05, "learning_rate": 1.9986167173603673e-05, "loss": 2.0063, "step": 174 }, { "epoch": 0.05, "learning_rate": 1.9985710029742384e-05, "loss": 2.0337, "step": 175 }, { "epoch": 0.05, "learning_rate": 1.9985245459789077e-05, "loss": 1.9834, "step": 176 }, { "epoch": 0.05, "learning_rate": 1.9984773464089228e-05, "loss": 1.9336, "step": 177 }, { "epoch": 0.05, "learning_rate": 1.998429404299386e-05, "loss": 1.9009, "step": 178 }, { "epoch": 0.05, "learning_rate": 1.9983807196859492e-05, "loss": 1.9576, "step": 179 }, { "epoch": 0.05, "learning_rate": 1.998331292604819e-05, "loss": 1.9375, "step": 180 }, { "epoch": 0.05, "learning_rate": 1.9982811230927522e-05, "loss": 1.9513, "step": 181 }, { "epoch": 0.05, "learning_rate": 1.9982302111870588e-05, "loss": 1.9222, "step": 182 }, { "epoch": 0.05, "learning_rate": 1.9981785569256e-05, "loss": 1.9298, "step": 183 }, { "epoch": 0.05, "learning_rate": 1.9981261603467903e-05, "loss": 1.9549, "step": 184 }, { "epoch": 0.05, "learning_rate": 1.9980730214895954e-05, "loss": 1.9024, "step": 185 }, { "epoch": 0.05, "learning_rate": 1.9980191403935326e-05, "loss": 1.9106, "step": 186 }, { "epoch": 0.05, "learning_rate": 1.9979645170986727e-05, "loss": 1.9179, "step": 187 }, { "epoch": 0.05, "learning_rate": 1.997909151645637e-05, "loss": 1.9416, "step": 188 }, { "epoch": 0.05, "learning_rate": 1.997853044075599e-05, "loss": 1.9285, "step": 189 }, { "epoch": 0.05, "learning_rate": 1.9977961944302847e-05, "loss": 1.9481, "step": 190 }, { "epoch": 0.05, "learning_rate": 1.9977386027519715e-05, "loss": 1.9732, "step": 191 }, { "epoch": 0.05, "learning_rate": 1.997680269083489e-05, "loss": 1.9502, "step": 192 }, { "epoch": 0.05, "learning_rate": 1.9976211934682172e-05, "loss": 1.9016, "step": 193 }, { "epoch": 0.05, "learning_rate": 1.9975613759500906e-05, "loss": 1.9013, "step": 194 }, { "epoch": 0.05, "learning_rate": 1.9975008165735924e-05, "loss": 1.9454, "step": 195 }, { "epoch": 0.05, "learning_rate": 1.9974395153837598e-05, "loss": 1.9062, "step": 196 }, { "epoch": 0.05, "learning_rate": 1.9973774724261804e-05, "loss": 1.9272, "step": 197 }, { "epoch": 0.05, "learning_rate": 1.9973146877469937e-05, "loss": 1.8834, "step": 198 }, { "epoch": 0.05, "learning_rate": 1.997251161392891e-05, "loss": 1.9093, "step": 199 }, { "epoch": 0.05, "learning_rate": 1.9971868934111153e-05, "loss": 1.872, "step": 200 }, { "epoch": 0.05, "learning_rate": 1.9971218838494607e-05, "loss": 1.9097, "step": 201 }, { "epoch": 0.05, "learning_rate": 1.997056132756273e-05, "loss": 1.9126, "step": 202 }, { "epoch": 0.05, "learning_rate": 1.9969896401804494e-05, "loss": 1.8786, "step": 203 }, { "epoch": 0.05, "learning_rate": 1.9969224061714388e-05, "loss": 1.8951, "step": 204 }, { "epoch": 0.05, "learning_rate": 1.9968544307792407e-05, "loss": 1.9486, "step": 205 }, { "epoch": 0.05, "learning_rate": 1.9967857140544072e-05, "loss": 1.9577, "step": 206 }, { "epoch": 0.06, "learning_rate": 1.9967162560480404e-05, "loss": 1.9117, "step": 207 }, { "epoch": 0.06, "learning_rate": 1.9966460568117947e-05, "loss": 1.9298, "step": 208 }, { "epoch": 0.06, "learning_rate": 1.9965751163978752e-05, "loss": 1.9526, "step": 209 }, { "epoch": 0.06, "learning_rate": 1.9965034348590374e-05, "loss": 1.9334, "step": 210 }, { "epoch": 0.06, "learning_rate": 1.9964310122485906e-05, "loss": 1.9407, "step": 211 }, { "epoch": 0.06, "learning_rate": 1.996357848620392e-05, "loss": 1.9348, "step": 212 }, { "epoch": 0.06, "learning_rate": 1.996283944028852e-05, "loss": 1.9466, "step": 213 }, { "epoch": 0.06, "learning_rate": 1.9962092985289312e-05, "loss": 1.9284, "step": 214 }, { "epoch": 0.06, "learning_rate": 1.9961339121761415e-05, "loss": 1.9198, "step": 215 }, { "epoch": 0.06, "learning_rate": 1.9960577850265452e-05, "loss": 1.9723, "step": 216 }, { "epoch": 0.06, "learning_rate": 1.9959809171367565e-05, "loss": 1.9083, "step": 217 }, { "epoch": 0.06, "learning_rate": 1.995903308563939e-05, "loss": 1.9255, "step": 218 }, { "epoch": 0.06, "learning_rate": 1.9958249593658093e-05, "loss": 1.8936, "step": 219 }, { "epoch": 0.06, "learning_rate": 1.9957458696006324e-05, "loss": 1.8588, "step": 220 }, { "epoch": 0.06, "learning_rate": 1.9956660393272256e-05, "loss": 1.8848, "step": 221 }, { "epoch": 0.06, "learning_rate": 1.9955854686049566e-05, "loss": 1.8799, "step": 222 }, { "epoch": 0.06, "learning_rate": 1.9955041574937433e-05, "loss": 1.8744, "step": 223 }, { "epoch": 0.06, "learning_rate": 1.9954221060540545e-05, "loss": 1.8582, "step": 224 }, { "epoch": 0.06, "learning_rate": 1.9953393143469094e-05, "loss": 1.8595, "step": 225 }, { "epoch": 0.06, "learning_rate": 1.9952557824338785e-05, "loss": 1.9063, "step": 226 }, { "epoch": 0.06, "learning_rate": 1.995171510377081e-05, "loss": 1.9233, "step": 227 }, { "epoch": 0.06, "learning_rate": 1.9950864982391893e-05, "loss": 1.8949, "step": 228 }, { "epoch": 0.06, "learning_rate": 1.9950007460834227e-05, "loss": 1.8587, "step": 229 }, { "epoch": 0.06, "learning_rate": 1.9949142539735543e-05, "loss": 1.8708, "step": 230 }, { "epoch": 0.06, "learning_rate": 1.9948270219739047e-05, "loss": 1.9107, "step": 231 }, { "epoch": 0.06, "learning_rate": 1.9947390501493464e-05, "loss": 1.8866, "step": 232 }, { "epoch": 0.06, "learning_rate": 1.9946503385653014e-05, "loss": 1.8952, "step": 233 }, { "epoch": 0.06, "learning_rate": 1.9945608872877423e-05, "loss": 1.8934, "step": 234 }, { "epoch": 0.06, "learning_rate": 1.994470696383191e-05, "loss": 1.9012, "step": 235 }, { "epoch": 0.06, "learning_rate": 1.9943797659187208e-05, "loss": 1.9032, "step": 236 }, { "epoch": 0.06, "learning_rate": 1.9942880959619534e-05, "loss": 1.8718, "step": 237 }, { "epoch": 0.06, "learning_rate": 1.994195686581061e-05, "loss": 1.8692, "step": 238 }, { "epoch": 0.06, "learning_rate": 1.9941025378447663e-05, "loss": 1.8623, "step": 239 }, { "epoch": 0.06, "learning_rate": 1.9940086498223415e-05, "loss": 1.9061, "step": 240 }, { "epoch": 0.06, "learning_rate": 1.9939140225836083e-05, "loss": 1.9019, "step": 241 }, { "epoch": 0.06, "learning_rate": 1.9938186561989383e-05, "loss": 1.8987, "step": 242 }, { "epoch": 0.06, "learning_rate": 1.9937225507392523e-05, "loss": 1.8767, "step": 243 }, { "epoch": 0.06, "learning_rate": 1.9936257062760222e-05, "loss": 1.9579, "step": 244 }, { "epoch": 0.07, "learning_rate": 1.993528122881268e-05, "loss": 1.926, "step": 245 }, { "epoch": 0.07, "learning_rate": 1.9934298006275597e-05, "loss": 1.8339, "step": 246 }, { "epoch": 0.07, "learning_rate": 1.993330739588017e-05, "loss": 1.8547, "step": 247 }, { "epoch": 0.07, "learning_rate": 1.9932309398363085e-05, "loss": 1.8545, "step": 248 }, { "epoch": 0.07, "learning_rate": 1.9931304014466528e-05, "loss": 1.8995, "step": 249 }, { "epoch": 0.07, "learning_rate": 1.9930291244938174e-05, "loss": 1.8845, "step": 250 }, { "epoch": 0.07, "learning_rate": 1.992927109053119e-05, "loss": 1.8782, "step": 251 }, { "epoch": 0.07, "learning_rate": 1.9928243552004237e-05, "loss": 1.9015, "step": 252 }, { "epoch": 0.07, "learning_rate": 1.9927208630121467e-05, "loss": 1.8725, "step": 253 }, { "epoch": 0.07, "learning_rate": 1.992616632565252e-05, "loss": 1.8647, "step": 254 }, { "epoch": 0.07, "learning_rate": 1.9925116639372528e-05, "loss": 1.8686, "step": 255 }, { "epoch": 0.07, "learning_rate": 1.992405957206212e-05, "loss": 1.8307, "step": 256 }, { "epoch": 0.07, "learning_rate": 1.9922995124507397e-05, "loss": 1.8414, "step": 257 }, { "epoch": 0.07, "learning_rate": 1.992192329749997e-05, "loss": 1.8623, "step": 258 }, { "epoch": 0.07, "learning_rate": 1.992084409183692e-05, "loss": 1.8543, "step": 259 }, { "epoch": 0.07, "learning_rate": 1.991975750832082e-05, "loss": 1.8282, "step": 260 }, { "epoch": 0.07, "learning_rate": 1.991866354775974e-05, "loss": 1.8484, "step": 261 }, { "epoch": 0.07, "learning_rate": 1.9917562210967216e-05, "loss": 1.8852, "step": 262 }, { "epoch": 0.07, "learning_rate": 1.9916453498762293e-05, "loss": 1.8084, "step": 263 }, { "epoch": 0.07, "learning_rate": 1.991533741196948e-05, "loss": 1.8608, "step": 264 }, { "epoch": 0.07, "learning_rate": 1.9914213951418783e-05, "loss": 1.795, "step": 265 }, { "epoch": 0.07, "learning_rate": 1.991308311794569e-05, "loss": 1.8594, "step": 266 }, { "epoch": 0.07, "learning_rate": 1.9911944912391167e-05, "loss": 1.8373, "step": 267 }, { "epoch": 0.07, "learning_rate": 1.991079933560167e-05, "loss": 1.8533, "step": 268 }, { "epoch": 0.07, "learning_rate": 1.9909646388429126e-05, "loss": 1.8841, "step": 269 }, { "epoch": 0.07, "learning_rate": 1.9908486071730954e-05, "loss": 1.8487, "step": 270 }, { "epoch": 0.07, "learning_rate": 1.9907318386370045e-05, "loss": 1.8384, "step": 271 }, { "epoch": 0.07, "learning_rate": 1.9906143333214776e-05, "loss": 1.874, "step": 272 }, { "epoch": 0.07, "learning_rate": 1.9904960913139004e-05, "loss": 1.858, "step": 273 }, { "epoch": 0.07, "learning_rate": 1.9903771127022055e-05, "loss": 1.8386, "step": 274 }, { "epoch": 0.07, "learning_rate": 1.9902573975748746e-05, "loss": 1.7635, "step": 275 }, { "epoch": 0.07, "learning_rate": 1.9901369460209362e-05, "loss": 1.8125, "step": 276 }, { "epoch": 0.07, "learning_rate": 1.9900157581299666e-05, "loss": 1.8203, "step": 277 }, { "epoch": 0.07, "learning_rate": 1.9898938339920902e-05, "loss": 1.7943, "step": 278 }, { "epoch": 0.07, "learning_rate": 1.989771173697978e-05, "loss": 1.814, "step": 279 }, { "epoch": 0.07, "learning_rate": 1.9896477773388496e-05, "loss": 1.8506, "step": 280 }, { "epoch": 0.07, "learning_rate": 1.9895236450064713e-05, "loss": 1.8558, "step": 281 }, { "epoch": 0.08, "learning_rate": 1.9893987767931563e-05, "loss": 1.848, "step": 282 }, { "epoch": 0.08, "learning_rate": 1.9892731727917663e-05, "loss": 1.7731, "step": 283 }, { "epoch": 0.08, "learning_rate": 1.9891468330957087e-05, "loss": 1.7893, "step": 284 }, { "epoch": 0.08, "learning_rate": 1.9890197577989398e-05, "loss": 1.7841, "step": 285 }, { "epoch": 0.08, "learning_rate": 1.9888919469959606e-05, "loss": 1.8079, "step": 286 }, { "epoch": 0.08, "learning_rate": 1.9887634007818218e-05, "loss": 1.793, "step": 287 }, { "epoch": 0.08, "learning_rate": 1.9886341192521182e-05, "loss": 1.8111, "step": 288 }, { "epoch": 0.08, "learning_rate": 1.9885041025029945e-05, "loss": 1.7598, "step": 289 }, { "epoch": 0.08, "learning_rate": 1.988373350631139e-05, "loss": 1.7731, "step": 290 }, { "epoch": 0.08, "learning_rate": 1.988241863733789e-05, "loss": 1.7457, "step": 291 }, { "epoch": 0.08, "learning_rate": 1.9881096419087275e-05, "loss": 1.7545, "step": 292 }, { "epoch": 0.08, "learning_rate": 1.9879766852542838e-05, "loss": 1.776, "step": 293 }, { "epoch": 0.08, "learning_rate": 1.9878429938693342e-05, "loss": 1.7536, "step": 294 }, { "epoch": 0.08, "learning_rate": 1.9877085678533014e-05, "loss": 1.7571, "step": 295 }, { "epoch": 0.08, "learning_rate": 1.9875734073061545e-05, "loss": 1.747, "step": 296 }, { "epoch": 0.08, "learning_rate": 1.987437512328408e-05, "loss": 1.7223, "step": 297 }, { "epoch": 0.08, "learning_rate": 1.9873008830211235e-05, "loss": 1.7077, "step": 298 }, { "epoch": 0.08, "learning_rate": 1.9871635194859083e-05, "loss": 1.7251, "step": 299 }, { "epoch": 0.08, "learning_rate": 1.9870254218249158e-05, "loss": 1.6867, "step": 300 }, { "epoch": 0.08, "learning_rate": 1.9868865901408454e-05, "loss": 1.7148, "step": 301 }, { "epoch": 0.08, "learning_rate": 1.9867470245369423e-05, "loss": 1.6998, "step": 302 }, { "epoch": 0.08, "learning_rate": 1.9866067251169974e-05, "loss": 1.713, "step": 303 }, { "epoch": 0.08, "learning_rate": 1.9864656919853477e-05, "loss": 1.696, "step": 304 }, { "epoch": 0.08, "learning_rate": 1.9863239252468754e-05, "loss": 1.6905, "step": 305 }, { "epoch": 0.08, "learning_rate": 1.9861814250070084e-05, "loss": 1.6908, "step": 306 }, { "epoch": 0.08, "learning_rate": 1.98603819137172e-05, "loss": 1.6485, "step": 307 }, { "epoch": 0.08, "learning_rate": 1.9858942244475297e-05, "loss": 1.6587, "step": 308 }, { "epoch": 0.08, "learning_rate": 1.985749524341501e-05, "loss": 1.5657, "step": 309 }, { "epoch": 0.08, "learning_rate": 1.9856040911612434e-05, "loss": 1.5778, "step": 310 }, { "epoch": 0.08, "learning_rate": 1.9854579250149118e-05, "loss": 1.6071, "step": 311 }, { "epoch": 0.08, "learning_rate": 1.9853110260112053e-05, "loss": 1.5867, "step": 312 }, { "epoch": 0.08, "learning_rate": 1.9851633942593688e-05, "loss": 1.5479, "step": 313 }, { "epoch": 0.08, "learning_rate": 1.9850150298691923e-05, "loss": 1.5655, "step": 314 }, { "epoch": 0.08, "learning_rate": 1.9848659329510097e-05, "loss": 1.5396, "step": 315 }, { "epoch": 0.08, "learning_rate": 1.9847161036157006e-05, "loss": 1.499, "step": 316 }, { "epoch": 0.08, "learning_rate": 1.9845655419746885e-05, "loss": 1.4956, "step": 317 }, { "epoch": 0.08, "learning_rate": 1.984414248139942e-05, "loss": 1.4955, "step": 318 }, { "epoch": 0.08, "learning_rate": 1.984262222223974e-05, "loss": 1.5219, "step": 319 }, { "epoch": 0.09, "learning_rate": 1.984109464339842e-05, "loss": 1.5113, "step": 320 }, { "epoch": 0.09, "learning_rate": 1.9839559746011478e-05, "loss": 1.4916, "step": 321 }, { "epoch": 0.09, "learning_rate": 1.9838017531220374e-05, "loss": 1.5195, "step": 322 }, { "epoch": 0.09, "learning_rate": 1.983646800017201e-05, "loss": 1.5267, "step": 323 }, { "epoch": 0.09, "learning_rate": 1.9834911154018727e-05, "loss": 1.5043, "step": 324 }, { "epoch": 0.09, "learning_rate": 1.983334699391831e-05, "loss": 1.5112, "step": 325 }, { "epoch": 0.09, "learning_rate": 1.9831775521033978e-05, "loss": 1.5234, "step": 326 }, { "epoch": 0.09, "learning_rate": 1.9830196736534393e-05, "loss": 1.5123, "step": 327 }, { "epoch": 0.09, "learning_rate": 1.9828610641593653e-05, "loss": 1.4398, "step": 328 }, { "epoch": 0.09, "learning_rate": 1.9827017237391292e-05, "loss": 1.4767, "step": 329 }, { "epoch": 0.09, "learning_rate": 1.9825416525112277e-05, "loss": 1.4964, "step": 330 }, { "epoch": 0.09, "learning_rate": 1.982380850594702e-05, "loss": 1.4719, "step": 331 }, { "epoch": 0.09, "learning_rate": 1.9822193181091354e-05, "loss": 1.5427, "step": 332 }, { "epoch": 0.09, "learning_rate": 1.982057055174655e-05, "loss": 1.5377, "step": 333 }, { "epoch": 0.09, "learning_rate": 1.9818940619119312e-05, "loss": 1.4708, "step": 334 }, { "epoch": 0.09, "learning_rate": 1.9817303384421782e-05, "loss": 1.4597, "step": 335 }, { "epoch": 0.09, "learning_rate": 1.9815658848871516e-05, "loss": 1.4817, "step": 336 }, { "epoch": 0.09, "learning_rate": 1.9814007013691518e-05, "loss": 1.4863, "step": 337 }, { "epoch": 0.09, "learning_rate": 1.9812347880110202e-05, "loss": 1.4535, "step": 338 }, { "epoch": 0.09, "learning_rate": 1.9810681449361425e-05, "loss": 1.4547, "step": 339 }, { "epoch": 0.09, "learning_rate": 1.9809007722684464e-05, "loss": 1.4215, "step": 340 }, { "epoch": 0.09, "learning_rate": 1.9807326701324023e-05, "loss": 1.4146, "step": 341 }, { "epoch": 0.09, "learning_rate": 1.9805638386530227e-05, "loss": 1.4084, "step": 342 }, { "epoch": 0.09, "learning_rate": 1.980394277955863e-05, "loss": 1.3941, "step": 343 }, { "epoch": 0.09, "learning_rate": 1.980223988167021e-05, "loss": 1.3865, "step": 344 }, { "epoch": 0.09, "learning_rate": 1.9800529694131365e-05, "loss": 1.424, "step": 345 }, { "epoch": 0.09, "learning_rate": 1.9798812218213906e-05, "loss": 1.459, "step": 346 }, { "epoch": 0.09, "learning_rate": 1.9797087455195077e-05, "loss": 1.4091, "step": 347 }, { "epoch": 0.09, "learning_rate": 1.9795355406357535e-05, "loss": 1.4143, "step": 348 }, { "epoch": 0.09, "learning_rate": 1.9793616072989354e-05, "loss": 1.4586, "step": 349 }, { "epoch": 0.09, "learning_rate": 1.9791869456384033e-05, "loss": 1.3983, "step": 350 }, { "epoch": 0.09, "learning_rate": 1.9790115557840476e-05, "loss": 1.4152, "step": 351 }, { "epoch": 0.09, "learning_rate": 1.9788354378663012e-05, "loss": 1.4143, "step": 352 }, { "epoch": 0.09, "learning_rate": 1.9786585920161378e-05, "loss": 1.4565, "step": 353 }, { "epoch": 0.09, "learning_rate": 1.9784810183650727e-05, "loss": 1.4352, "step": 354 }, { "epoch": 0.09, "learning_rate": 1.9783027170451622e-05, "loss": 1.429, "step": 355 }, { "epoch": 0.09, "learning_rate": 1.978123688189005e-05, "loss": 1.4064, "step": 356 }, { "epoch": 0.1, "learning_rate": 1.9779439319297383e-05, "loss": 1.4069, "step": 357 }, { "epoch": 0.1, "learning_rate": 1.977763448401043e-05, "loss": 1.4215, "step": 358 }, { "epoch": 0.1, "learning_rate": 1.977582237737139e-05, "loss": 1.424, "step": 359 }, { "epoch": 0.1, "learning_rate": 1.977400300072788e-05, "loss": 1.4202, "step": 360 }, { "epoch": 0.1, "learning_rate": 1.977217635543292e-05, "loss": 1.4228, "step": 361 }, { "epoch": 0.1, "learning_rate": 1.9770342442844926e-05, "loss": 1.3691, "step": 362 }, { "epoch": 0.1, "learning_rate": 1.976850126432774e-05, "loss": 1.3901, "step": 363 }, { "epoch": 0.1, "learning_rate": 1.9766652821250583e-05, "loss": 1.3605, "step": 364 }, { "epoch": 0.1, "learning_rate": 1.9764797114988097e-05, "loss": 1.3497, "step": 365 }, { "epoch": 0.1, "learning_rate": 1.9762934146920318e-05, "loss": 1.3184, "step": 366 }, { "epoch": 0.1, "learning_rate": 1.976106391843268e-05, "loss": 1.3383, "step": 367 }, { "epoch": 0.1, "learning_rate": 1.975918643091602e-05, "loss": 1.366, "step": 368 }, { "epoch": 0.1, "learning_rate": 1.9757301685766576e-05, "loss": 1.3603, "step": 369 }, { "epoch": 0.1, "learning_rate": 1.9755409684385974e-05, "loss": 1.3436, "step": 370 }, { "epoch": 0.1, "learning_rate": 1.975351042818125e-05, "loss": 1.306, "step": 371 }, { "epoch": 0.1, "learning_rate": 1.9751603918564826e-05, "loss": 1.3329, "step": 372 }, { "epoch": 0.1, "learning_rate": 1.9749690156954515e-05, "loss": 1.3729, "step": 373 }, { "epoch": 0.1, "learning_rate": 1.974776914477353e-05, "loss": 1.356, "step": 374 }, { "epoch": 0.1, "learning_rate": 1.9745840883450474e-05, "loss": 1.4007, "step": 375 }, { "epoch": 0.1, "learning_rate": 1.974390537441934e-05, "loss": 1.3463, "step": 376 }, { "epoch": 0.1, "learning_rate": 1.9741962619119513e-05, "loss": 1.352, "step": 377 }, { "epoch": 0.1, "learning_rate": 1.9740012618995764e-05, "loss": 1.3178, "step": 378 }, { "epoch": 0.1, "learning_rate": 1.9738055375498257e-05, "loss": 1.3788, "step": 379 }, { "epoch": 0.1, "learning_rate": 1.9736090890082534e-05, "loss": 1.2948, "step": 380 }, { "epoch": 0.1, "learning_rate": 1.973411916420953e-05, "loss": 1.3385, "step": 381 }, { "epoch": 0.1, "learning_rate": 1.9732140199345565e-05, "loss": 1.3316, "step": 382 }, { "epoch": 0.1, "learning_rate": 1.9730153996962334e-05, "loss": 1.3608, "step": 383 }, { "epoch": 0.1, "learning_rate": 1.972816055853692e-05, "loss": 1.3213, "step": 384 }, { "epoch": 0.1, "learning_rate": 1.9726159885551793e-05, "loss": 1.3354, "step": 385 }, { "epoch": 0.1, "learning_rate": 1.972415197949479e-05, "loss": 1.3362, "step": 386 }, { "epoch": 0.1, "learning_rate": 1.9722136841859142e-05, "loss": 1.3475, "step": 387 }, { "epoch": 0.1, "learning_rate": 1.972011447414345e-05, "loss": 1.3791, "step": 388 }, { "epoch": 0.1, "learning_rate": 1.9718084877851682e-05, "loss": 1.3412, "step": 389 }, { "epoch": 0.1, "learning_rate": 1.97160480544932e-05, "loss": 1.2926, "step": 390 }, { "epoch": 0.1, "learning_rate": 1.9714004005582735e-05, "loss": 1.3416, "step": 391 }, { "epoch": 0.1, "learning_rate": 1.971195273264038e-05, "loss": 1.3066, "step": 392 }, { "epoch": 0.1, "learning_rate": 1.9709894237191613e-05, "loss": 1.3206, "step": 393 }, { "epoch": 0.1, "learning_rate": 1.9707828520767282e-05, "loss": 1.3125, "step": 394 }, { "epoch": 0.11, "learning_rate": 1.97057555849036e-05, "loss": 1.3061, "step": 395 }, { "epoch": 0.11, "learning_rate": 1.970367543114215e-05, "loss": 1.3014, "step": 396 }, { "epoch": 0.11, "learning_rate": 1.9701588061029884e-05, "loss": 1.2805, "step": 397 }, { "epoch": 0.11, "learning_rate": 1.969949347611912e-05, "loss": 1.2856, "step": 398 }, { "epoch": 0.11, "learning_rate": 1.969739167796755e-05, "loss": 1.3132, "step": 399 }, { "epoch": 0.11, "learning_rate": 1.969528266813821e-05, "loss": 1.276, "step": 400 }, { "epoch": 0.11, "learning_rate": 1.969316644819952e-05, "loss": 1.3028, "step": 401 }, { "epoch": 0.11, "learning_rate": 1.9691043019725252e-05, "loss": 1.2988, "step": 402 }, { "epoch": 0.11, "learning_rate": 1.9688912384294535e-05, "loss": 1.2942, "step": 403 }, { "epoch": 0.11, "learning_rate": 1.9686774543491865e-05, "loss": 1.2771, "step": 404 }, { "epoch": 0.11, "learning_rate": 1.96846294989071e-05, "loss": 1.3104, "step": 405 }, { "epoch": 0.11, "learning_rate": 1.9682477252135444e-05, "loss": 1.2897, "step": 406 }, { "epoch": 0.11, "learning_rate": 1.968031780477746e-05, "loss": 1.316, "step": 407 }, { "epoch": 0.11, "learning_rate": 1.9678151158439075e-05, "loss": 1.3043, "step": 408 }, { "epoch": 0.11, "learning_rate": 1.9675977314731554e-05, "loss": 1.2968, "step": 409 }, { "epoch": 0.11, "learning_rate": 1.9673796275271536e-05, "loss": 1.3308, "step": 410 }, { "epoch": 0.11, "learning_rate": 1.967160804168099e-05, "loss": 1.3371, "step": 411 }, { "epoch": 0.11, "learning_rate": 1.9669412615587244e-05, "loss": 1.2985, "step": 412 }, { "epoch": 0.11, "learning_rate": 1.9667209998622976e-05, "loss": 1.2716, "step": 413 }, { "epoch": 0.11, "learning_rate": 1.9665000192426213e-05, "loss": 1.3384, "step": 414 }, { "epoch": 0.11, "learning_rate": 1.9662783198640323e-05, "loss": 1.2814, "step": 415 }, { "epoch": 0.11, "learning_rate": 1.966055901891402e-05, "loss": 1.2444, "step": 416 }, { "epoch": 0.11, "learning_rate": 1.965832765490137e-05, "loss": 1.2546, "step": 417 }, { "epoch": 0.11, "learning_rate": 1.965608910826177e-05, "loss": 1.2963, "step": 418 }, { "epoch": 0.11, "learning_rate": 1.965384338065997e-05, "loss": 1.2414, "step": 419 }, { "epoch": 0.11, "learning_rate": 1.965159047376605e-05, "loss": 1.2743, "step": 420 }, { "epoch": 0.11, "learning_rate": 1.964933038925543e-05, "loss": 1.2594, "step": 421 }, { "epoch": 0.11, "learning_rate": 1.9647063128808882e-05, "loss": 1.2826, "step": 422 }, { "epoch": 0.11, "learning_rate": 1.9644788694112495e-05, "loss": 1.3069, "step": 423 }, { "epoch": 0.11, "learning_rate": 1.9642507086857706e-05, "loss": 1.2625, "step": 424 }, { "epoch": 0.11, "learning_rate": 1.9640218308741278e-05, "loss": 1.292, "step": 425 }, { "epoch": 0.11, "learning_rate": 1.9637922361465317e-05, "loss": 1.2533, "step": 426 }, { "epoch": 0.11, "learning_rate": 1.9635619246737253e-05, "loss": 1.2235, "step": 427 }, { "epoch": 0.11, "learning_rate": 1.9633308966269845e-05, "loss": 1.2175, "step": 428 }, { "epoch": 0.11, "learning_rate": 1.9630991521781184e-05, "loss": 1.2898, "step": 429 }, { "epoch": 0.11, "learning_rate": 1.962866691499469e-05, "loss": 1.2709, "step": 430 }, { "epoch": 0.11, "learning_rate": 1.962633514763911e-05, "loss": 1.2581, "step": 431 }, { "epoch": 0.12, "learning_rate": 1.962399622144851e-05, "loss": 1.2884, "step": 432 }, { "epoch": 0.12, "learning_rate": 1.9621650138162284e-05, "loss": 1.2284, "step": 433 }, { "epoch": 0.12, "learning_rate": 1.9619296899525147e-05, "loss": 1.2299, "step": 434 }, { "epoch": 0.12, "learning_rate": 1.961693650728714e-05, "loss": 1.2662, "step": 435 }, { "epoch": 0.12, "learning_rate": 1.961456896320362e-05, "loss": 1.2242, "step": 436 }, { "epoch": 0.12, "learning_rate": 1.9612194269035264e-05, "loss": 1.2879, "step": 437 }, { "epoch": 0.12, "learning_rate": 1.960981242654806e-05, "loss": 1.2639, "step": 438 }, { "epoch": 0.12, "learning_rate": 1.9607423437513325e-05, "loss": 1.3079, "step": 439 }, { "epoch": 0.12, "learning_rate": 1.9605027303707675e-05, "loss": 1.3076, "step": 440 }, { "epoch": 0.12, "learning_rate": 1.9602624026913052e-05, "loss": 1.2316, "step": 441 }, { "epoch": 0.12, "learning_rate": 1.9600213608916703e-05, "loss": 1.264, "step": 442 }, { "epoch": 0.12, "learning_rate": 1.9597796051511194e-05, "loss": 1.3153, "step": 443 }, { "epoch": 0.12, "learning_rate": 1.9595371356494385e-05, "loss": 1.2632, "step": 444 }, { "epoch": 0.12, "learning_rate": 1.959293952566946e-05, "loss": 1.2505, "step": 445 }, { "epoch": 0.12, "learning_rate": 1.9590500560844904e-05, "loss": 1.2605, "step": 446 }, { "epoch": 0.12, "learning_rate": 1.9588054463834504e-05, "loss": 1.288, "step": 447 }, { "epoch": 0.12, "learning_rate": 1.9585601236457355e-05, "loss": 1.272, "step": 448 }, { "epoch": 0.12, "learning_rate": 1.9583140880537852e-05, "loss": 1.2597, "step": 449 }, { "epoch": 0.12, "learning_rate": 1.9580673397905694e-05, "loss": 1.2651, "step": 450 }, { "epoch": 0.12, "learning_rate": 1.9578198790395875e-05, "loss": 1.3186, "step": 451 }, { "epoch": 0.12, "learning_rate": 1.9575717059848695e-05, "loss": 1.2669, "step": 452 }, { "epoch": 0.12, "learning_rate": 1.9573228208109748e-05, "loss": 1.259, "step": 453 }, { "epoch": 0.12, "learning_rate": 1.9570732237029922e-05, "loss": 1.263, "step": 454 }, { "epoch": 0.12, "learning_rate": 1.9568229148465393e-05, "loss": 1.2607, "step": 455 }, { "epoch": 0.12, "learning_rate": 1.9565718944277653e-05, "loss": 1.2744, "step": 456 }, { "epoch": 0.12, "learning_rate": 1.9563201626333453e-05, "loss": 1.229, "step": 457 }, { "epoch": 0.12, "learning_rate": 1.956067719650487e-05, "loss": 1.2447, "step": 458 }, { "epoch": 0.12, "learning_rate": 1.9558145656669233e-05, "loss": 1.2609, "step": 459 }, { "epoch": 0.12, "learning_rate": 1.955560700870919e-05, "loss": 1.2309, "step": 460 }, { "epoch": 0.12, "learning_rate": 1.9553061254512663e-05, "loss": 1.3167, "step": 461 }, { "epoch": 0.12, "learning_rate": 1.9550508395972846e-05, "loss": 1.2862, "step": 462 }, { "epoch": 0.12, "learning_rate": 1.9547948434988238e-05, "loss": 1.2457, "step": 463 }, { "epoch": 0.12, "learning_rate": 1.9545381373462606e-05, "loss": 1.2822, "step": 464 }, { "epoch": 0.12, "learning_rate": 1.9542807213305005e-05, "loss": 1.2974, "step": 465 }, { "epoch": 0.12, "learning_rate": 1.954022595642976e-05, "loss": 1.3415, "step": 466 }, { "epoch": 0.12, "learning_rate": 1.9537637604756485e-05, "loss": 1.315, "step": 467 }, { "epoch": 0.12, "learning_rate": 1.953504216021006e-05, "loss": 1.2913, "step": 468 }, { "epoch": 0.12, "learning_rate": 1.953243962472065e-05, "loss": 1.3037, "step": 469 }, { "epoch": 0.13, "learning_rate": 1.952983000022368e-05, "loss": 1.2933, "step": 470 }, { "epoch": 0.13, "learning_rate": 1.952721328865986e-05, "loss": 1.237, "step": 471 }, { "epoch": 0.13, "learning_rate": 1.9524589491975163e-05, "loss": 1.2654, "step": 472 }, { "epoch": 0.13, "learning_rate": 1.9521958612120833e-05, "loss": 1.2838, "step": 473 }, { "epoch": 0.13, "learning_rate": 1.9519320651053385e-05, "loss": 1.2587, "step": 474 }, { "epoch": 0.13, "learning_rate": 1.9516675610734588e-05, "loss": 1.2992, "step": 475 }, { "epoch": 0.13, "learning_rate": 1.9514023493131492e-05, "loss": 1.2926, "step": 476 }, { "epoch": 0.13, "learning_rate": 1.9511364300216397e-05, "loss": 1.2699, "step": 477 }, { "epoch": 0.13, "learning_rate": 1.9508698033966875e-05, "loss": 1.2692, "step": 478 }, { "epoch": 0.13, "learning_rate": 1.9506024696365747e-05, "loss": 1.3608, "step": 479 }, { "epoch": 0.13, "learning_rate": 1.9503344289401106e-05, "loss": 1.2673, "step": 480 }, { "epoch": 0.13, "learning_rate": 1.950065681506629e-05, "loss": 1.3204, "step": 481 }, { "epoch": 0.13, "learning_rate": 1.9497962275359903e-05, "loss": 1.3194, "step": 482 }, { "epoch": 0.13, "learning_rate": 1.9495260672285795e-05, "loss": 1.2927, "step": 483 }, { "epoch": 0.13, "learning_rate": 1.9492552007853075e-05, "loss": 1.2747, "step": 484 }, { "epoch": 0.13, "learning_rate": 1.9489836284076097e-05, "loss": 1.2709, "step": 485 }, { "epoch": 0.13, "learning_rate": 1.948711350297447e-05, "loss": 1.2657, "step": 486 }, { "epoch": 0.13, "learning_rate": 1.948438366657305e-05, "loss": 1.3108, "step": 487 }, { "epoch": 0.13, "learning_rate": 1.9481646776901942e-05, "loss": 1.222, "step": 488 }, { "epoch": 0.13, "learning_rate": 1.947890283599649e-05, "loss": 1.2864, "step": 489 }, { "epoch": 0.13, "learning_rate": 1.9476151845897287e-05, "loss": 1.2947, "step": 490 }, { "epoch": 0.13, "learning_rate": 1.947339380865017e-05, "loss": 1.2564, "step": 491 }, { "epoch": 0.13, "learning_rate": 1.947062872630621e-05, "loss": 1.3159, "step": 492 }, { "epoch": 0.13, "learning_rate": 1.9467856600921724e-05, "loss": 1.3129, "step": 493 }, { "epoch": 0.13, "learning_rate": 1.9465077434558258e-05, "loss": 1.2719, "step": 494 }, { "epoch": 0.13, "learning_rate": 1.946229122928261e-05, "loss": 1.2509, "step": 495 }, { "epoch": 0.13, "learning_rate": 1.9459497987166794e-05, "loss": 1.2487, "step": 496 }, { "epoch": 0.13, "learning_rate": 1.9456697710288068e-05, "loss": 1.282, "step": 497 }, { "epoch": 0.13, "learning_rate": 1.9453890400728922e-05, "loss": 1.239, "step": 498 }, { "epoch": 0.13, "learning_rate": 1.945107606057707e-05, "loss": 1.2053, "step": 499 }, { "epoch": 0.13, "learning_rate": 1.9448254691925455e-05, "loss": 1.2172, "step": 500 }, { "epoch": 0.13, "learning_rate": 1.9445426296872257e-05, "loss": 1.2355, "step": 501 }, { "epoch": 0.13, "learning_rate": 1.944259087752087e-05, "loss": 1.2665, "step": 502 }, { "epoch": 0.13, "learning_rate": 1.9439748435979914e-05, "loss": 1.244, "step": 503 }, { "epoch": 0.13, "learning_rate": 1.9436898974363237e-05, "loss": 1.2393, "step": 504 }, { "epoch": 0.13, "learning_rate": 1.94340424947899e-05, "loss": 1.2491, "step": 505 }, { "epoch": 0.13, "learning_rate": 1.943117899938419e-05, "loss": 1.2702, "step": 506 }, { "epoch": 0.13, "learning_rate": 1.9428308490275606e-05, "loss": 1.2337, "step": 507 }, { "epoch": 0.14, "learning_rate": 1.9425430969598866e-05, "loss": 1.2736, "step": 508 }, { "epoch": 0.14, "learning_rate": 1.94225464394939e-05, "loss": 1.2506, "step": 509 }, { "epoch": 0.14, "learning_rate": 1.9419654902105848e-05, "loss": 1.2779, "step": 510 }, { "epoch": 0.14, "learning_rate": 1.9416756359585073e-05, "loss": 1.1998, "step": 511 }, { "epoch": 0.14, "learning_rate": 1.9413850814087137e-05, "loss": 1.1963, "step": 512 }, { "epoch": 0.14, "learning_rate": 1.941093826777281e-05, "loss": 1.2677, "step": 513 }, { "epoch": 0.14, "learning_rate": 1.9408018722808075e-05, "loss": 1.2444, "step": 514 }, { "epoch": 0.14, "learning_rate": 1.9405092181364113e-05, "loss": 1.2451, "step": 515 }, { "epoch": 0.14, "learning_rate": 1.940215864561731e-05, "loss": 1.2399, "step": 516 }, { "epoch": 0.14, "learning_rate": 1.9399218117749253e-05, "loss": 1.2893, "step": 517 }, { "epoch": 0.14, "learning_rate": 1.9396270599946735e-05, "loss": 1.2291, "step": 518 }, { "epoch": 0.14, "learning_rate": 1.9393316094401737e-05, "loss": 1.2441, "step": 519 }, { "epoch": 0.14, "learning_rate": 1.9390354603311445e-05, "loss": 1.2276, "step": 520 }, { "epoch": 0.14, "learning_rate": 1.9387386128878233e-05, "loss": 1.2159, "step": 521 }, { "epoch": 0.14, "learning_rate": 1.9384410673309676e-05, "loss": 1.2328, "step": 522 }, { "epoch": 0.14, "learning_rate": 1.9381428238818538e-05, "loss": 1.2539, "step": 523 }, { "epoch": 0.14, "learning_rate": 1.937843882762277e-05, "loss": 1.2685, "step": 524 }, { "epoch": 0.14, "learning_rate": 1.9375442441945507e-05, "loss": 1.2493, "step": 525 }, { "epoch": 0.14, "learning_rate": 1.937243908401509e-05, "loss": 1.3042, "step": 526 }, { "epoch": 0.14, "learning_rate": 1.9369428756065015e-05, "loss": 1.2732, "step": 527 }, { "epoch": 0.14, "learning_rate": 1.9366411460333992e-05, "loss": 1.2641, "step": 528 }, { "epoch": 0.14, "learning_rate": 1.936338719906589e-05, "loss": 1.3147, "step": 529 }, { "epoch": 0.14, "learning_rate": 1.9360355974509778e-05, "loss": 1.2317, "step": 530 }, { "epoch": 0.14, "learning_rate": 1.9357317788919885e-05, "loss": 1.2191, "step": 531 }, { "epoch": 0.14, "learning_rate": 1.9354272644555622e-05, "loss": 1.1826, "step": 532 }, { "epoch": 0.14, "learning_rate": 1.9351220543681582e-05, "loss": 1.2314, "step": 533 }, { "epoch": 0.14, "learning_rate": 1.934816148856753e-05, "loss": 1.2341, "step": 534 }, { "epoch": 0.14, "learning_rate": 1.934509548148839e-05, "loss": 1.2433, "step": 535 }, { "epoch": 0.14, "learning_rate": 1.9342022524724272e-05, "loss": 1.2406, "step": 536 }, { "epoch": 0.14, "learning_rate": 1.9338942620560444e-05, "loss": 1.1994, "step": 537 }, { "epoch": 0.14, "learning_rate": 1.9335855771287347e-05, "loss": 1.267, "step": 538 }, { "epoch": 0.14, "learning_rate": 1.933276197920058e-05, "loss": 1.2047, "step": 539 }, { "epoch": 0.14, "learning_rate": 1.9329661246600916e-05, "loss": 1.2197, "step": 540 }, { "epoch": 0.14, "learning_rate": 1.932655357579428e-05, "loss": 1.2988, "step": 541 }, { "epoch": 0.14, "learning_rate": 1.9323438969091753e-05, "loss": 1.2244, "step": 542 }, { "epoch": 0.14, "learning_rate": 1.932031742880959e-05, "loss": 1.2468, "step": 543 }, { "epoch": 0.14, "learning_rate": 1.9317188957269192e-05, "loss": 1.2394, "step": 544 }, { "epoch": 0.15, "learning_rate": 1.9314053556797105e-05, "loss": 1.3103, "step": 545 }, { "epoch": 0.15, "learning_rate": 1.931091122972505e-05, "loss": 1.2302, "step": 546 }, { "epoch": 0.15, "learning_rate": 1.9307761978389883e-05, "loss": 1.264, "step": 547 }, { "epoch": 0.15, "learning_rate": 1.930460580513361e-05, "loss": 1.2955, "step": 548 }, { "epoch": 0.15, "learning_rate": 1.9301442712303395e-05, "loss": 1.2818, "step": 549 }, { "epoch": 0.15, "learning_rate": 1.9298272702251538e-05, "loss": 1.2378, "step": 550 }, { "epoch": 0.15, "learning_rate": 1.9295095777335486e-05, "loss": 1.2645, "step": 551 }, { "epoch": 0.15, "learning_rate": 1.9291911939917832e-05, "loss": 1.1898, "step": 552 }, { "epoch": 0.15, "learning_rate": 1.9288721192366296e-05, "loss": 1.2397, "step": 553 }, { "epoch": 0.15, "learning_rate": 1.9285523537053766e-05, "loss": 1.2539, "step": 554 }, { "epoch": 0.15, "learning_rate": 1.9282318976358228e-05, "loss": 1.283, "step": 555 }, { "epoch": 0.15, "learning_rate": 1.9279107512662838e-05, "loss": 1.223, "step": 556 }, { "epoch": 0.15, "learning_rate": 1.9275889148355865e-05, "loss": 1.2189, "step": 557 }, { "epoch": 0.15, "learning_rate": 1.9272663885830724e-05, "loss": 1.2336, "step": 558 }, { "epoch": 0.15, "learning_rate": 1.926943172748594e-05, "loss": 1.249, "step": 559 }, { "epoch": 0.15, "learning_rate": 1.9266192675725186e-05, "loss": 1.2566, "step": 560 }, { "epoch": 0.15, "learning_rate": 1.926294673295725e-05, "loss": 1.2718, "step": 561 }, { "epoch": 0.15, "learning_rate": 1.9259693901596054e-05, "loss": 1.2926, "step": 562 }, { "epoch": 0.15, "learning_rate": 1.925643418406063e-05, "loss": 1.2265, "step": 563 }, { "epoch": 0.15, "learning_rate": 1.9253167582775145e-05, "loss": 1.2598, "step": 564 }, { "epoch": 0.15, "learning_rate": 1.9249894100168877e-05, "loss": 1.2278, "step": 565 }, { "epoch": 0.15, "learning_rate": 1.924661373867622e-05, "loss": 1.2301, "step": 566 }, { "epoch": 0.15, "learning_rate": 1.9243326500736687e-05, "loss": 1.2274, "step": 567 }, { "epoch": 0.15, "learning_rate": 1.924003238879491e-05, "loss": 1.2401, "step": 568 }, { "epoch": 0.15, "learning_rate": 1.923673140530062e-05, "loss": 1.2388, "step": 569 }, { "epoch": 0.15, "learning_rate": 1.923342355270867e-05, "loss": 1.2453, "step": 570 }, { "epoch": 0.15, "learning_rate": 1.9230108833479017e-05, "loss": 1.2229, "step": 571 }, { "epoch": 0.15, "learning_rate": 1.9226787250076725e-05, "loss": 1.2429, "step": 572 }, { "epoch": 0.15, "learning_rate": 1.9223458804971963e-05, "loss": 1.231, "step": 573 }, { "epoch": 0.15, "learning_rate": 1.9220123500639995e-05, "loss": 1.2367, "step": 574 }, { "epoch": 0.15, "learning_rate": 1.92167813395612e-05, "loss": 1.2202, "step": 575 }, { "epoch": 0.15, "learning_rate": 1.921343232422105e-05, "loss": 1.2164, "step": 576 }, { "epoch": 0.15, "learning_rate": 1.921007645711011e-05, "loss": 1.2972, "step": 577 }, { "epoch": 0.15, "learning_rate": 1.9206713740724047e-05, "loss": 1.2484, "step": 578 }, { "epoch": 0.15, "learning_rate": 1.9203344177563615e-05, "loss": 1.2781, "step": 579 }, { "epoch": 0.15, "learning_rate": 1.9199967770134667e-05, "loss": 1.2443, "step": 580 }, { "epoch": 0.15, "learning_rate": 1.9196584520948137e-05, "loss": 1.276, "step": 581 }, { "epoch": 0.15, "learning_rate": 1.9193194432520063e-05, "loss": 1.2806, "step": 582 }, { "epoch": 0.16, "learning_rate": 1.918979750737155e-05, "loss": 1.2258, "step": 583 }, { "epoch": 0.16, "learning_rate": 1.9186393748028796e-05, "loss": 1.2155, "step": 584 }, { "epoch": 0.16, "learning_rate": 1.9182983157023082e-05, "loss": 1.2317, "step": 585 }, { "epoch": 0.16, "learning_rate": 1.917956573689077e-05, "loss": 1.2609, "step": 586 }, { "epoch": 0.16, "learning_rate": 1.91761414901733e-05, "loss": 1.2373, "step": 587 }, { "epoch": 0.16, "learning_rate": 1.9172710419417187e-05, "loss": 1.2463, "step": 588 }, { "epoch": 0.16, "learning_rate": 1.9169272527174024e-05, "loss": 1.2548, "step": 589 }, { "epoch": 0.16, "learning_rate": 1.916582781600047e-05, "loss": 1.2548, "step": 590 }, { "epoch": 0.16, "learning_rate": 1.9162376288458265e-05, "loss": 1.22, "step": 591 }, { "epoch": 0.16, "learning_rate": 1.9158917947114207e-05, "loss": 1.23, "step": 592 }, { "epoch": 0.16, "learning_rate": 1.9155452794540176e-05, "loss": 1.2402, "step": 593 }, { "epoch": 0.16, "learning_rate": 1.91519808333131e-05, "loss": 1.2546, "step": 594 }, { "epoch": 0.16, "learning_rate": 1.9148502066014983e-05, "loss": 1.2533, "step": 595 }, { "epoch": 0.16, "learning_rate": 1.9145016495232887e-05, "loss": 1.2142, "step": 596 }, { "epoch": 0.16, "learning_rate": 1.9141524123558938e-05, "loss": 1.2709, "step": 597 }, { "epoch": 0.16, "learning_rate": 1.9138024953590303e-05, "loss": 1.1757, "step": 598 }, { "epoch": 0.16, "learning_rate": 1.913451898792922e-05, "loss": 1.2165, "step": 599 }, { "epoch": 0.16, "learning_rate": 1.9131006229182985e-05, "loss": 1.2425, "step": 600 }, { "epoch": 0.16, "learning_rate": 1.9127486679963933e-05, "loss": 1.2424, "step": 601 }, { "epoch": 0.16, "learning_rate": 1.912396034288945e-05, "loss": 1.2552, "step": 602 }, { "epoch": 0.16, "learning_rate": 1.9120427220581977e-05, "loss": 1.239, "step": 603 }, { "epoch": 0.16, "learning_rate": 1.9116887315669004e-05, "loss": 1.2055, "step": 604 }, { "epoch": 0.16, "learning_rate": 1.9113340630783045e-05, "loss": 1.2061, "step": 605 }, { "epoch": 0.16, "learning_rate": 1.9109787168561683e-05, "loss": 1.1645, "step": 606 }, { "epoch": 0.16, "learning_rate": 1.9106226931647525e-05, "loss": 1.1796, "step": 607 }, { "epoch": 0.16, "learning_rate": 1.9102659922688213e-05, "loss": 1.2497, "step": 608 }, { "epoch": 0.16, "learning_rate": 1.9099086144336437e-05, "loss": 1.1734, "step": 609 }, { "epoch": 0.16, "learning_rate": 1.9095505599249912e-05, "loss": 1.244, "step": 610 }, { "epoch": 0.16, "learning_rate": 1.9091918290091395e-05, "loss": 1.2475, "step": 611 }, { "epoch": 0.16, "learning_rate": 1.908832421952866e-05, "loss": 1.2411, "step": 612 }, { "epoch": 0.16, "learning_rate": 1.908472339023452e-05, "loss": 1.2361, "step": 613 }, { "epoch": 0.16, "learning_rate": 1.9081115804886807e-05, "loss": 1.2677, "step": 614 }, { "epoch": 0.16, "learning_rate": 1.907750146616839e-05, "loss": 1.2517, "step": 615 }, { "epoch": 0.16, "learning_rate": 1.907388037676714e-05, "loss": 1.3095, "step": 616 }, { "epoch": 0.16, "learning_rate": 1.9070252539375962e-05, "loss": 1.2766, "step": 617 }, { "epoch": 0.16, "learning_rate": 1.9066617956692785e-05, "loss": 1.2445, "step": 618 }, { "epoch": 0.16, "learning_rate": 1.906297663142054e-05, "loss": 1.2759, "step": 619 }, { "epoch": 0.17, "learning_rate": 1.9059328566267172e-05, "loss": 1.2896, "step": 620 }, { "epoch": 0.17, "learning_rate": 1.9055673763945653e-05, "loss": 1.2508, "step": 621 }, { "epoch": 0.17, "learning_rate": 1.9052012227173958e-05, "loss": 1.211, "step": 622 }, { "epoch": 0.17, "learning_rate": 1.9048343958675067e-05, "loss": 1.2344, "step": 623 }, { "epoch": 0.17, "learning_rate": 1.9044668961176962e-05, "loss": 1.2381, "step": 624 }, { "epoch": 0.17, "learning_rate": 1.9040987237412643e-05, "loss": 1.2505, "step": 625 }, { "epoch": 0.17, "learning_rate": 1.90372987901201e-05, "loss": 1.273, "step": 626 }, { "epoch": 0.17, "learning_rate": 1.9033603622042333e-05, "loss": 1.2591, "step": 627 }, { "epoch": 0.17, "learning_rate": 1.902990173592733e-05, "loss": 1.2367, "step": 628 }, { "epoch": 0.17, "learning_rate": 1.902619313452808e-05, "loss": 1.2161, "step": 629 }, { "epoch": 0.17, "learning_rate": 1.9022477820602568e-05, "loss": 1.2511, "step": 630 }, { "epoch": 0.17, "learning_rate": 1.9018755796913766e-05, "loss": 1.1785, "step": 631 }, { "epoch": 0.17, "learning_rate": 1.9015027066229642e-05, "loss": 1.3031, "step": 632 }, { "epoch": 0.17, "learning_rate": 1.9011291631323143e-05, "loss": 1.2344, "step": 633 }, { "epoch": 0.17, "learning_rate": 1.900754949497221e-05, "loss": 1.2534, "step": 634 }, { "epoch": 0.17, "learning_rate": 1.9003800659959767e-05, "loss": 1.234, "step": 635 }, { "epoch": 0.17, "learning_rate": 1.9000045129073714e-05, "loss": 1.275, "step": 636 }, { "epoch": 0.17, "learning_rate": 1.8996282905106936e-05, "loss": 1.2263, "step": 637 }, { "epoch": 0.17, "learning_rate": 1.8992513990857292e-05, "loss": 1.2775, "step": 638 }, { "epoch": 0.17, "learning_rate": 1.8988738389127614e-05, "loss": 1.2527, "step": 639 }, { "epoch": 0.17, "learning_rate": 1.8984956102725718e-05, "loss": 1.2265, "step": 640 }, { "epoch": 0.17, "learning_rate": 1.898116713446438e-05, "loss": 1.2666, "step": 641 }, { "epoch": 0.17, "learning_rate": 1.897737148716135e-05, "loss": 1.2511, "step": 642 }, { "epoch": 0.17, "learning_rate": 1.8973569163639345e-05, "loss": 1.2319, "step": 643 }, { "epoch": 0.17, "learning_rate": 1.8969760166726045e-05, "loss": 1.2096, "step": 644 }, { "epoch": 0.17, "learning_rate": 1.8965944499254095e-05, "loss": 1.2356, "step": 645 }, { "epoch": 0.17, "learning_rate": 1.8962122164061096e-05, "loss": 1.2376, "step": 646 }, { "epoch": 0.17, "learning_rate": 1.895829316398962e-05, "loss": 1.2522, "step": 647 }, { "epoch": 0.17, "learning_rate": 1.8954457501887176e-05, "loss": 1.1907, "step": 648 }, { "epoch": 0.17, "learning_rate": 1.8950615180606244e-05, "loss": 1.222, "step": 649 }, { "epoch": 0.17, "learning_rate": 1.8946766203004253e-05, "loss": 1.2551, "step": 650 }, { "epoch": 0.17, "learning_rate": 1.8942910571943573e-05, "loss": 1.2448, "step": 651 }, { "epoch": 0.17, "learning_rate": 1.8939048290291534e-05, "loss": 1.2414, "step": 652 }, { "epoch": 0.17, "learning_rate": 1.8935179360920404e-05, "loss": 1.2846, "step": 653 }, { "epoch": 0.17, "learning_rate": 1.8931303786707397e-05, "loss": 1.2081, "step": 654 }, { "epoch": 0.17, "learning_rate": 1.892742157053467e-05, "loss": 1.2611, "step": 655 }, { "epoch": 0.17, "learning_rate": 1.8923532715289315e-05, "loss": 1.2354, "step": 656 }, { "epoch": 0.17, "learning_rate": 1.8919637223863366e-05, "loss": 1.224, "step": 657 }, { "epoch": 0.18, "learning_rate": 1.8915735099153794e-05, "loss": 1.2701, "step": 658 }, { "epoch": 0.18, "learning_rate": 1.8911826344062495e-05, "loss": 1.2722, "step": 659 }, { "epoch": 0.18, "learning_rate": 1.8907910961496304e-05, "loss": 1.2465, "step": 660 }, { "epoch": 0.18, "learning_rate": 1.8903988954366975e-05, "loss": 1.2751, "step": 661 }, { "epoch": 0.18, "learning_rate": 1.89000603255912e-05, "loss": 1.2544, "step": 662 }, { "epoch": 0.18, "learning_rate": 1.889612507809059e-05, "loss": 1.2605, "step": 663 }, { "epoch": 0.18, "learning_rate": 1.8892183214791674e-05, "loss": 1.223, "step": 664 }, { "epoch": 0.18, "learning_rate": 1.8888234738625906e-05, "loss": 1.2387, "step": 665 }, { "epoch": 0.18, "learning_rate": 1.888427965252966e-05, "loss": 1.2126, "step": 666 }, { "epoch": 0.18, "learning_rate": 1.888031795944422e-05, "loss": 1.1988, "step": 667 }, { "epoch": 0.18, "learning_rate": 1.8876349662315788e-05, "loss": 1.2515, "step": 668 }, { "epoch": 0.18, "learning_rate": 1.8872374764095474e-05, "loss": 1.267, "step": 669 }, { "epoch": 0.18, "learning_rate": 1.8868393267739296e-05, "loss": 1.2271, "step": 670 }, { "epoch": 0.18, "learning_rate": 1.886440517620818e-05, "loss": 1.2256, "step": 671 }, { "epoch": 0.18, "learning_rate": 1.886041049246796e-05, "loss": 1.1926, "step": 672 }, { "epoch": 0.18, "learning_rate": 1.8856409219489377e-05, "loss": 1.2331, "step": 673 }, { "epoch": 0.18, "learning_rate": 1.8852401360248054e-05, "loss": 1.1965, "step": 674 }, { "epoch": 0.18, "learning_rate": 1.8848386917724527e-05, "loss": 1.2201, "step": 675 }, { "epoch": 0.18, "learning_rate": 1.8844365894904227e-05, "loss": 1.235, "step": 676 }, { "epoch": 0.18, "learning_rate": 1.8840338294777475e-05, "loss": 1.2085, "step": 677 }, { "epoch": 0.18, "learning_rate": 1.883630412033948e-05, "loss": 1.2357, "step": 678 }, { "epoch": 0.18, "learning_rate": 1.8832263374590348e-05, "loss": 1.265, "step": 679 }, { "epoch": 0.18, "learning_rate": 1.8828216060535067e-05, "loss": 1.2585, "step": 680 }, { "epoch": 0.18, "learning_rate": 1.8824162181183513e-05, "loss": 1.2407, "step": 681 }, { "epoch": 0.18, "learning_rate": 1.8820101739550435e-05, "loss": 1.2348, "step": 682 }, { "epoch": 0.18, "learning_rate": 1.881603473865548e-05, "loss": 1.2667, "step": 683 }, { "epoch": 0.18, "learning_rate": 1.8811961181523154e-05, "loss": 1.2401, "step": 684 }, { "epoch": 0.18, "learning_rate": 1.880788107118285e-05, "loss": 1.3008, "step": 685 }, { "epoch": 0.18, "learning_rate": 1.8803794410668832e-05, "loss": 1.2292, "step": 686 }, { "epoch": 0.18, "learning_rate": 1.8799701203020233e-05, "loss": 1.2211, "step": 687 }, { "epoch": 0.18, "learning_rate": 1.8795601451281058e-05, "loss": 1.2188, "step": 688 }, { "epoch": 0.18, "learning_rate": 1.8791495158500174e-05, "loss": 1.2324, "step": 689 }, { "epoch": 0.18, "learning_rate": 1.878738232773132e-05, "loss": 1.2296, "step": 690 }, { "epoch": 0.18, "learning_rate": 1.8783262962033092e-05, "loss": 1.1839, "step": 691 }, { "epoch": 0.18, "learning_rate": 1.8779137064468947e-05, "loss": 1.2149, "step": 692 }, { "epoch": 0.18, "learning_rate": 1.8775004638107192e-05, "loss": 1.2228, "step": 693 }, { "epoch": 0.18, "learning_rate": 1.8770865686021008e-05, "loss": 1.2323, "step": 694 }, { "epoch": 0.19, "learning_rate": 1.8766720211288407e-05, "loss": 1.1646, "step": 695 }, { "epoch": 0.19, "learning_rate": 1.8762568216992268e-05, "loss": 1.2287, "step": 696 }, { "epoch": 0.19, "learning_rate": 1.875840970622031e-05, "loss": 1.1981, "step": 697 }, { "epoch": 0.19, "learning_rate": 1.8754244682065103e-05, "loss": 1.222, "step": 698 }, { "epoch": 0.19, "learning_rate": 1.8750073147624054e-05, "loss": 1.2383, "step": 699 }, { "epoch": 0.19, "learning_rate": 1.8745895105999423e-05, "loss": 1.2615, "step": 700 }, { "epoch": 0.19, "learning_rate": 1.8741710560298293e-05, "loss": 1.182, "step": 701 }, { "epoch": 0.19, "learning_rate": 1.8737519513632604e-05, "loss": 1.1948, "step": 702 }, { "epoch": 0.19, "learning_rate": 1.873332196911911e-05, "loss": 1.1821, "step": 703 }, { "epoch": 0.19, "learning_rate": 1.8729117929879413e-05, "loss": 1.2049, "step": 704 }, { "epoch": 0.19, "learning_rate": 1.8724907399039935e-05, "loss": 1.2076, "step": 705 }, { "epoch": 0.19, "learning_rate": 1.8720690379731932e-05, "loss": 1.1765, "step": 706 }, { "epoch": 0.19, "learning_rate": 1.8716466875091483e-05, "loss": 1.1941, "step": 707 }, { "epoch": 0.19, "learning_rate": 1.8712236888259486e-05, "loss": 1.187, "step": 708 }, { "epoch": 0.19, "learning_rate": 1.8708000422381667e-05, "loss": 1.186, "step": 709 }, { "epoch": 0.19, "learning_rate": 1.870375748060857e-05, "loss": 1.2054, "step": 710 }, { "epoch": 0.19, "learning_rate": 1.869950806609554e-05, "loss": 1.1731, "step": 711 }, { "epoch": 0.19, "learning_rate": 1.869525218200276e-05, "loss": 1.2097, "step": 712 }, { "epoch": 0.19, "learning_rate": 1.8690989831495206e-05, "loss": 1.2026, "step": 713 }, { "epoch": 0.19, "learning_rate": 1.8686721017742666e-05, "loss": 1.1769, "step": 714 }, { "epoch": 0.19, "learning_rate": 1.868244574391974e-05, "loss": 1.2111, "step": 715 }, { "epoch": 0.19, "learning_rate": 1.867816401320583e-05, "loss": 1.1978, "step": 716 }, { "epoch": 0.19, "learning_rate": 1.8673875828785136e-05, "loss": 1.2161, "step": 717 }, { "epoch": 0.19, "learning_rate": 1.866958119384666e-05, "loss": 1.1864, "step": 718 }, { "epoch": 0.19, "learning_rate": 1.86652801115842e-05, "loss": 1.1924, "step": 719 }, { "epoch": 0.19, "learning_rate": 1.866097258519636e-05, "loss": 1.221, "step": 720 }, { "epoch": 0.19, "learning_rate": 1.8656658617886515e-05, "loss": 1.2148, "step": 721 }, { "epoch": 0.19, "learning_rate": 1.8652338212862847e-05, "loss": 1.1782, "step": 722 }, { "epoch": 0.19, "learning_rate": 1.8648011373338318e-05, "loss": 1.2437, "step": 723 }, { "epoch": 0.19, "learning_rate": 1.8643678102530674e-05, "loss": 1.1929, "step": 724 }, { "epoch": 0.19, "learning_rate": 1.8639338403662455e-05, "loss": 1.253, "step": 725 }, { "epoch": 0.19, "learning_rate": 1.8634992279960966e-05, "loss": 1.2302, "step": 726 }, { "epoch": 0.19, "learning_rate": 1.8630639734658297e-05, "loss": 1.2389, "step": 727 }, { "epoch": 0.19, "learning_rate": 1.862628077099132e-05, "loss": 1.1963, "step": 728 }, { "epoch": 0.19, "learning_rate": 1.862191539220167e-05, "loss": 1.2066, "step": 729 }, { "epoch": 0.19, "learning_rate": 1.8617543601535752e-05, "loss": 1.1784, "step": 730 }, { "epoch": 0.19, "learning_rate": 1.861316540224475e-05, "loss": 1.1709, "step": 731 }, { "epoch": 0.19, "learning_rate": 1.860878079758461e-05, "loss": 1.2621, "step": 732 }, { "epoch": 0.2, "learning_rate": 1.8604389790816032e-05, "loss": 1.2238, "step": 733 }, { "epoch": 0.2, "learning_rate": 1.8599992385204485e-05, "loss": 1.2411, "step": 734 }, { "epoch": 0.2, "learning_rate": 1.8595588584020204e-05, "loss": 1.2267, "step": 735 }, { "epoch": 0.2, "learning_rate": 1.8591178390538163e-05, "loss": 1.2856, "step": 736 }, { "epoch": 0.2, "learning_rate": 1.8586761808038108e-05, "loss": 1.2681, "step": 737 }, { "epoch": 0.2, "learning_rate": 1.8582338839804518e-05, "loss": 1.2765, "step": 738 }, { "epoch": 0.2, "learning_rate": 1.8577909489126634e-05, "loss": 1.2736, "step": 739 }, { "epoch": 0.2, "learning_rate": 1.857347375929844e-05, "loss": 1.2265, "step": 740 }, { "epoch": 0.2, "learning_rate": 1.856903165361867e-05, "loss": 1.2386, "step": 741 }, { "epoch": 0.2, "learning_rate": 1.856458317539078e-05, "loss": 1.2079, "step": 742 }, { "epoch": 0.2, "learning_rate": 1.856012832792299e-05, "loss": 1.2135, "step": 743 }, { "epoch": 0.2, "learning_rate": 1.855566711452824e-05, "loss": 1.2308, "step": 744 }, { "epoch": 0.2, "learning_rate": 1.8551199538524205e-05, "loss": 1.1615, "step": 745 }, { "epoch": 0.2, "learning_rate": 1.8546725603233306e-05, "loss": 1.2618, "step": 746 }, { "epoch": 0.2, "learning_rate": 1.854224531198267e-05, "loss": 1.1738, "step": 747 }, { "epoch": 0.2, "learning_rate": 1.853775866810418e-05, "loss": 1.1912, "step": 748 }, { "epoch": 0.2, "learning_rate": 1.853326567493441e-05, "loss": 1.1687, "step": 749 }, { "epoch": 0.2, "learning_rate": 1.8528766335814684e-05, "loss": 1.1533, "step": 750 }, { "epoch": 0.2, "learning_rate": 1.852426065409103e-05, "loss": 1.2308, "step": 751 }, { "epoch": 0.2, "learning_rate": 1.85197486331142e-05, "loss": 1.2101, "step": 752 }, { "epoch": 0.2, "learning_rate": 1.8515230276239654e-05, "loss": 1.2176, "step": 753 }, { "epoch": 0.2, "learning_rate": 1.8510705586827566e-05, "loss": 1.2372, "step": 754 }, { "epoch": 0.2, "learning_rate": 1.850617456824282e-05, "loss": 1.1994, "step": 755 }, { "epoch": 0.2, "learning_rate": 1.8501637223855016e-05, "loss": 1.2537, "step": 756 }, { "epoch": 0.2, "learning_rate": 1.849709355703844e-05, "loss": 1.2186, "step": 757 }, { "epoch": 0.2, "learning_rate": 1.8492543571172094e-05, "loss": 1.1488, "step": 758 }, { "epoch": 0.2, "learning_rate": 1.8487987269639672e-05, "loss": 1.253, "step": 759 }, { "epoch": 0.2, "learning_rate": 1.848342465582957e-05, "loss": 1.231, "step": 760 }, { "epoch": 0.2, "learning_rate": 1.8478855733134874e-05, "loss": 1.2184, "step": 761 }, { "epoch": 0.2, "learning_rate": 1.847428050495336e-05, "loss": 1.1739, "step": 762 }, { "epoch": 0.2, "learning_rate": 1.8469698974687506e-05, "loss": 1.2569, "step": 763 }, { "epoch": 0.2, "learning_rate": 1.846511114574446e-05, "loss": 1.1986, "step": 764 }, { "epoch": 0.2, "learning_rate": 1.846051702153606e-05, "loss": 1.2808, "step": 765 }, { "epoch": 0.2, "learning_rate": 1.8455916605478833e-05, "loss": 1.2569, "step": 766 }, { "epoch": 0.2, "learning_rate": 1.845130990099398e-05, "loss": 1.233, "step": 767 }, { "epoch": 0.2, "learning_rate": 1.8446696911507374e-05, "loss": 1.262, "step": 768 }, { "epoch": 0.2, "learning_rate": 1.844207764044956e-05, "loss": 1.2251, "step": 769 }, { "epoch": 0.21, "learning_rate": 1.8437452091255772e-05, "loss": 1.2507, "step": 770 }, { "epoch": 0.21, "learning_rate": 1.8432820267365894e-05, "loss": 1.1849, "step": 771 }, { "epoch": 0.21, "learning_rate": 1.8428182172224483e-05, "loss": 1.1569, "step": 772 }, { "epoch": 0.21, "learning_rate": 1.8423537809280762e-05, "loss": 1.1715, "step": 773 }, { "epoch": 0.21, "learning_rate": 1.8418887181988615e-05, "loss": 1.2373, "step": 774 }, { "epoch": 0.21, "learning_rate": 1.8414230293806577e-05, "loss": 1.2143, "step": 775 }, { "epoch": 0.21, "learning_rate": 1.840956714819785e-05, "loss": 1.2226, "step": 776 }, { "epoch": 0.21, "learning_rate": 1.8404897748630283e-05, "loss": 1.2251, "step": 777 }, { "epoch": 0.21, "learning_rate": 1.840022209857638e-05, "loss": 1.266, "step": 778 }, { "epoch": 0.21, "learning_rate": 1.839554020151328e-05, "loss": 1.2404, "step": 779 }, { "epoch": 0.21, "learning_rate": 1.8390852060922796e-05, "loss": 1.2634, "step": 780 }, { "epoch": 0.21, "learning_rate": 1.838615768029135e-05, "loss": 1.2294, "step": 781 }, { "epoch": 0.21, "learning_rate": 1.8381457063110032e-05, "loss": 1.2584, "step": 782 }, { "epoch": 0.21, "learning_rate": 1.837675021287455e-05, "loss": 1.2943, "step": 783 }, { "epoch": 0.21, "learning_rate": 1.8372037133085262e-05, "loss": 1.2392, "step": 784 }, { "epoch": 0.21, "learning_rate": 1.8367317827247155e-05, "loss": 1.2134, "step": 785 }, { "epoch": 0.21, "learning_rate": 1.836259229886984e-05, "loss": 1.204, "step": 786 }, { "epoch": 0.21, "learning_rate": 1.835786055146756e-05, "loss": 1.2323, "step": 787 }, { "epoch": 0.21, "learning_rate": 1.8353122588559182e-05, "loss": 1.2363, "step": 788 }, { "epoch": 0.21, "learning_rate": 1.83483784136682e-05, "loss": 1.2092, "step": 789 }, { "epoch": 0.21, "learning_rate": 1.834362803032272e-05, "loss": 1.2212, "step": 790 }, { "epoch": 0.21, "learning_rate": 1.833887144205547e-05, "loss": 1.2511, "step": 791 }, { "epoch": 0.21, "learning_rate": 1.8334108652403787e-05, "loss": 1.2191, "step": 792 }, { "epoch": 0.21, "learning_rate": 1.8329339664909633e-05, "loss": 1.2311, "step": 793 }, { "epoch": 0.21, "learning_rate": 1.832456448311956e-05, "loss": 1.2413, "step": 794 }, { "epoch": 0.21, "learning_rate": 1.8319783110584743e-05, "loss": 1.1921, "step": 795 }, { "epoch": 0.21, "learning_rate": 1.831499555086095e-05, "loss": 1.2317, "step": 796 }, { "epoch": 0.21, "learning_rate": 1.831020180750856e-05, "loss": 1.2158, "step": 797 }, { "epoch": 0.21, "learning_rate": 1.830540188409254e-05, "loss": 1.264, "step": 798 }, { "epoch": 0.21, "learning_rate": 1.8300595784182464e-05, "loss": 1.2391, "step": 799 }, { "epoch": 0.21, "learning_rate": 1.8295783511352488e-05, "loss": 1.2193, "step": 800 }, { "epoch": 0.21, "learning_rate": 1.8290965069181368e-05, "loss": 1.2093, "step": 801 }, { "epoch": 0.21, "learning_rate": 1.828614046125244e-05, "loss": 1.2225, "step": 802 }, { "epoch": 0.21, "learning_rate": 1.8281309691153632e-05, "loss": 1.2159, "step": 803 }, { "epoch": 0.21, "learning_rate": 1.8276472762477453e-05, "loss": 1.187, "step": 804 }, { "epoch": 0.21, "learning_rate": 1.8271629678820992e-05, "loss": 1.2254, "step": 805 }, { "epoch": 0.21, "learning_rate": 1.8266780443785906e-05, "loss": 1.2361, "step": 806 }, { "epoch": 0.21, "learning_rate": 1.826192506097844e-05, "loss": 1.1828, "step": 807 }, { "epoch": 0.22, "learning_rate": 1.8257063534009415e-05, "loss": 1.2238, "step": 808 }, { "epoch": 0.22, "learning_rate": 1.8252195866494194e-05, "loss": 1.19, "step": 809 }, { "epoch": 0.22, "learning_rate": 1.824732206205274e-05, "loss": 1.2212, "step": 810 }, { "epoch": 0.22, "learning_rate": 1.8242442124309557e-05, "loss": 1.2213, "step": 811 }, { "epoch": 0.22, "learning_rate": 1.8237556056893722e-05, "loss": 1.2313, "step": 812 }, { "epoch": 0.22, "learning_rate": 1.8232663863438862e-05, "loss": 1.2204, "step": 813 }, { "epoch": 0.22, "learning_rate": 1.8227765547583164e-05, "loss": 1.2166, "step": 814 }, { "epoch": 0.22, "learning_rate": 1.8222861112969368e-05, "loss": 1.2168, "step": 815 }, { "epoch": 0.22, "learning_rate": 1.8217950563244767e-05, "loss": 1.1648, "step": 816 }, { "epoch": 0.22, "learning_rate": 1.82130339020612e-05, "loss": 1.2046, "step": 817 }, { "epoch": 0.22, "learning_rate": 1.820811113307504e-05, "loss": 1.1828, "step": 818 }, { "epoch": 0.22, "learning_rate": 1.820318225994722e-05, "loss": 1.1964, "step": 819 }, { "epoch": 0.22, "learning_rate": 1.8198247286343208e-05, "loss": 1.2382, "step": 820 }, { "epoch": 0.22, "learning_rate": 1.8193306215933e-05, "loss": 1.2299, "step": 821 }, { "epoch": 0.22, "learning_rate": 1.818835905239113e-05, "loss": 1.2241, "step": 822 }, { "epoch": 0.22, "learning_rate": 1.8183405799396668e-05, "loss": 1.2208, "step": 823 }, { "epoch": 0.22, "learning_rate": 1.8178446460633208e-05, "loss": 1.1725, "step": 824 }, { "epoch": 0.22, "learning_rate": 1.8173481039788867e-05, "loss": 1.1876, "step": 825 }, { "epoch": 0.22, "learning_rate": 1.8168509540556297e-05, "loss": 1.2068, "step": 826 }, { "epoch": 0.22, "learning_rate": 1.8163531966632652e-05, "loss": 1.2043, "step": 827 }, { "epoch": 0.22, "learning_rate": 1.8158548321719624e-05, "loss": 1.211, "step": 828 }, { "epoch": 0.22, "learning_rate": 1.8153558609523405e-05, "loss": 1.1778, "step": 829 }, { "epoch": 0.22, "learning_rate": 1.8148562833754707e-05, "loss": 1.1521, "step": 830 }, { "epoch": 0.22, "learning_rate": 1.814356099812874e-05, "loss": 1.1569, "step": 831 }, { "epoch": 0.22, "learning_rate": 1.8138553106365235e-05, "loss": 1.16, "step": 832 }, { "epoch": 0.22, "learning_rate": 1.813353916218842e-05, "loss": 1.2086, "step": 833 }, { "epoch": 0.22, "learning_rate": 1.8128519169327026e-05, "loss": 1.1575, "step": 834 }, { "epoch": 0.22, "learning_rate": 1.8123493131514277e-05, "loss": 1.1611, "step": 835 }, { "epoch": 0.22, "learning_rate": 1.8118461052487897e-05, "loss": 1.1967, "step": 836 }, { "epoch": 0.22, "learning_rate": 1.8113422935990104e-05, "loss": 1.1527, "step": 837 }, { "epoch": 0.22, "learning_rate": 1.8108378785767597e-05, "loss": 1.1807, "step": 838 }, { "epoch": 0.22, "learning_rate": 1.810332860557158e-05, "loss": 1.2188, "step": 839 }, { "epoch": 0.22, "learning_rate": 1.8098272399157723e-05, "loss": 1.204, "step": 840 }, { "epoch": 0.22, "learning_rate": 1.8093210170286185e-05, "loss": 1.1769, "step": 841 }, { "epoch": 0.22, "learning_rate": 1.80881419227216e-05, "loss": 1.1494, "step": 842 }, { "epoch": 0.22, "learning_rate": 1.8083067660233088e-05, "loss": 1.2462, "step": 843 }, { "epoch": 0.22, "learning_rate": 1.8077987386594236e-05, "loss": 1.1486, "step": 844 }, { "epoch": 0.22, "learning_rate": 1.8072901105583094e-05, "loss": 1.2559, "step": 845 }, { "epoch": 0.23, "learning_rate": 1.8067808820982193e-05, "loss": 1.1867, "step": 846 }, { "epoch": 0.23, "learning_rate": 1.806271053657852e-05, "loss": 1.2168, "step": 847 }, { "epoch": 0.23, "learning_rate": 1.805760625616352e-05, "loss": 1.1851, "step": 848 }, { "epoch": 0.23, "learning_rate": 1.805249598353311e-05, "loss": 1.2278, "step": 849 }, { "epoch": 0.23, "learning_rate": 1.8047379722487654e-05, "loss": 1.2008, "step": 850 }, { "epoch": 0.23, "learning_rate": 1.804225747683197e-05, "loss": 1.1979, "step": 851 }, { "epoch": 0.23, "learning_rate": 1.8037129250375333e-05, "loss": 1.2266, "step": 852 }, { "epoch": 0.23, "learning_rate": 1.803199504693146e-05, "loss": 1.1756, "step": 853 }, { "epoch": 0.23, "learning_rate": 1.8026854870318506e-05, "loss": 1.1782, "step": 854 }, { "epoch": 0.23, "learning_rate": 1.8021708724359082e-05, "loss": 1.2002, "step": 855 }, { "epoch": 0.23, "learning_rate": 1.8016556612880233e-05, "loss": 1.1756, "step": 856 }, { "epoch": 0.23, "learning_rate": 1.8011398539713442e-05, "loss": 1.2039, "step": 857 }, { "epoch": 0.23, "learning_rate": 1.8006234508694612e-05, "loss": 1.2324, "step": 858 }, { "epoch": 0.23, "learning_rate": 1.8001064523664104e-05, "loss": 1.2422, "step": 859 }, { "epoch": 0.23, "learning_rate": 1.7995888588466676e-05, "loss": 1.2294, "step": 860 }, { "epoch": 0.23, "learning_rate": 1.7990706706951533e-05, "loss": 1.2169, "step": 861 }, { "epoch": 0.23, "learning_rate": 1.798551888297229e-05, "loss": 1.2454, "step": 862 }, { "epoch": 0.23, "learning_rate": 1.7980325120386993e-05, "loss": 1.2311, "step": 863 }, { "epoch": 0.23, "learning_rate": 1.7975125423058092e-05, "loss": 1.2603, "step": 864 }, { "epoch": 0.23, "learning_rate": 1.7969919794852456e-05, "loss": 1.2582, "step": 865 }, { "epoch": 0.23, "learning_rate": 1.7964708239641372e-05, "loss": 1.212, "step": 866 }, { "epoch": 0.23, "learning_rate": 1.7959490761300517e-05, "loss": 1.2061, "step": 867 }, { "epoch": 0.23, "learning_rate": 1.7954267363709986e-05, "loss": 1.2292, "step": 868 }, { "epoch": 0.23, "learning_rate": 1.7949038050754276e-05, "loss": 1.1923, "step": 869 }, { "epoch": 0.23, "learning_rate": 1.7943802826322275e-05, "loss": 1.2494, "step": 870 }, { "epoch": 0.23, "learning_rate": 1.7938561694307277e-05, "loss": 1.1869, "step": 871 }, { "epoch": 0.23, "learning_rate": 1.7933314658606963e-05, "loss": 1.2022, "step": 872 }, { "epoch": 0.23, "learning_rate": 1.7928061723123406e-05, "loss": 1.1736, "step": 873 }, { "epoch": 0.23, "learning_rate": 1.792280289176307e-05, "loss": 1.1962, "step": 874 }, { "epoch": 0.23, "learning_rate": 1.7917538168436794e-05, "loss": 1.1535, "step": 875 }, { "epoch": 0.23, "learning_rate": 1.7912267557059808e-05, "loss": 1.1931, "step": 876 }, { "epoch": 0.23, "learning_rate": 1.7906991061551718e-05, "loss": 1.1583, "step": 877 }, { "epoch": 0.23, "learning_rate": 1.7901708685836503e-05, "loss": 1.2031, "step": 878 }, { "epoch": 0.23, "learning_rate": 1.7896420433842522e-05, "loss": 1.1453, "step": 879 }, { "epoch": 0.23, "learning_rate": 1.7891126309502495e-05, "loss": 1.1358, "step": 880 }, { "epoch": 0.23, "learning_rate": 1.7885826316753516e-05, "loss": 1.1693, "step": 881 }, { "epoch": 0.23, "learning_rate": 1.7880520459537042e-05, "loss": 1.1449, "step": 882 }, { "epoch": 0.24, "learning_rate": 1.7875208741798885e-05, "loss": 1.2067, "step": 883 }, { "epoch": 0.24, "learning_rate": 1.7869891167489224e-05, "loss": 1.166, "step": 884 }, { "epoch": 0.24, "learning_rate": 1.7864567740562588e-05, "loss": 1.1506, "step": 885 }, { "epoch": 0.24, "learning_rate": 1.7859238464977863e-05, "loss": 1.1958, "step": 886 }, { "epoch": 0.24, "learning_rate": 1.7853903344698276e-05, "loss": 1.2008, "step": 887 }, { "epoch": 0.24, "learning_rate": 1.784856238369141e-05, "loss": 1.1663, "step": 888 }, { "epoch": 0.24, "learning_rate": 1.784321558592919e-05, "loss": 1.2281, "step": 889 }, { "epoch": 0.24, "learning_rate": 1.7837862955387875e-05, "loss": 1.2288, "step": 890 }, { "epoch": 0.24, "learning_rate": 1.7832504496048068e-05, "loss": 1.1653, "step": 891 }, { "epoch": 0.24, "learning_rate": 1.78271402118947e-05, "loss": 1.1373, "step": 892 }, { "epoch": 0.24, "learning_rate": 1.7821770106917053e-05, "loss": 1.217, "step": 893 }, { "epoch": 0.24, "learning_rate": 1.7816394185108705e-05, "loss": 1.1769, "step": 894 }, { "epoch": 0.24, "learning_rate": 1.7811012450467587e-05, "loss": 1.1716, "step": 895 }, { "epoch": 0.24, "learning_rate": 1.780562490699594e-05, "loss": 1.1833, "step": 896 }, { "epoch": 0.24, "learning_rate": 1.7800231558700334e-05, "loss": 1.1705, "step": 897 }, { "epoch": 0.24, "learning_rate": 1.7794832409591647e-05, "loss": 1.1932, "step": 898 }, { "epoch": 0.24, "learning_rate": 1.778942746368508e-05, "loss": 1.2083, "step": 899 }, { "epoch": 0.24, "learning_rate": 1.778401672500013e-05, "loss": 1.1891, "step": 900 }, { "epoch": 0.24, "learning_rate": 1.7778600197560612e-05, "loss": 1.1892, "step": 901 }, { "epoch": 0.24, "learning_rate": 1.7773177885394658e-05, "loss": 1.1769, "step": 902 }, { "epoch": 0.24, "learning_rate": 1.7767749792534678e-05, "loss": 1.1754, "step": 903 }, { "epoch": 0.24, "learning_rate": 1.7762315923017395e-05, "loss": 1.1617, "step": 904 }, { "epoch": 0.24, "learning_rate": 1.7756876280883824e-05, "loss": 1.2243, "step": 905 }, { "epoch": 0.24, "learning_rate": 1.7751430870179277e-05, "loss": 1.2258, "step": 906 }, { "epoch": 0.24, "learning_rate": 1.7745979694953347e-05, "loss": 1.176, "step": 907 }, { "epoch": 0.24, "learning_rate": 1.7740522759259928e-05, "loss": 1.1786, "step": 908 }, { "epoch": 0.24, "learning_rate": 1.7735060067157186e-05, "loss": 1.2116, "step": 909 }, { "epoch": 0.24, "learning_rate": 1.772959162270757e-05, "loss": 1.1987, "step": 910 }, { "epoch": 0.24, "learning_rate": 1.772411742997781e-05, "loss": 1.2198, "step": 911 }, { "epoch": 0.24, "learning_rate": 1.771863749303891e-05, "loss": 1.1834, "step": 912 }, { "epoch": 0.24, "learning_rate": 1.771315181596614e-05, "loss": 1.2377, "step": 913 }, { "epoch": 0.24, "learning_rate": 1.7707660402839055e-05, "loss": 1.2027, "step": 914 }, { "epoch": 0.24, "learning_rate": 1.770216325774145e-05, "loss": 1.1991, "step": 915 }, { "epoch": 0.24, "learning_rate": 1.7696660384761413e-05, "loss": 1.2036, "step": 916 }, { "epoch": 0.24, "learning_rate": 1.7691151787991265e-05, "loss": 1.1902, "step": 917 }, { "epoch": 0.24, "learning_rate": 1.76856374715276e-05, "loss": 1.2198, "step": 918 }, { "epoch": 0.24, "learning_rate": 1.7680117439471258e-05, "loss": 1.2526, "step": 919 }, { "epoch": 0.24, "learning_rate": 1.767459169592733e-05, "loss": 1.185, "step": 920 }, { "epoch": 0.25, "learning_rate": 1.7669060245005155e-05, "loss": 1.2256, "step": 921 }, { "epoch": 0.25, "learning_rate": 1.7663523090818328e-05, "loss": 1.2052, "step": 922 }, { "epoch": 0.25, "learning_rate": 1.7657980237484665e-05, "loss": 1.2201, "step": 923 }, { "epoch": 0.25, "learning_rate": 1.7652431689126235e-05, "loss": 1.1766, "step": 924 }, { "epoch": 0.25, "learning_rate": 1.764687744986933e-05, "loss": 1.2078, "step": 925 }, { "epoch": 0.25, "learning_rate": 1.7641317523844494e-05, "loss": 1.2266, "step": 926 }, { "epoch": 0.25, "learning_rate": 1.763575191518648e-05, "loss": 1.2189, "step": 927 }, { "epoch": 0.25, "learning_rate": 1.7630180628034278e-05, "loss": 1.186, "step": 928 }, { "epoch": 0.25, "learning_rate": 1.7624603666531092e-05, "loss": 1.1738, "step": 929 }, { "epoch": 0.25, "learning_rate": 1.7619021034824356e-05, "loss": 1.1913, "step": 930 }, { "epoch": 0.25, "learning_rate": 1.7613432737065716e-05, "loss": 1.1799, "step": 931 }, { "epoch": 0.25, "learning_rate": 1.7607838777411035e-05, "loss": 1.1785, "step": 932 }, { "epoch": 0.25, "learning_rate": 1.760223916002038e-05, "loss": 1.1563, "step": 933 }, { "epoch": 0.25, "learning_rate": 1.7596633889058027e-05, "loss": 1.1631, "step": 934 }, { "epoch": 0.25, "learning_rate": 1.759102296869246e-05, "loss": 1.1406, "step": 935 }, { "epoch": 0.25, "learning_rate": 1.7585406403096365e-05, "loss": 1.1562, "step": 936 }, { "epoch": 0.25, "learning_rate": 1.7579784196446626e-05, "loss": 1.1592, "step": 937 }, { "epoch": 0.25, "learning_rate": 1.7574156352924316e-05, "loss": 1.1302, "step": 938 }, { "epoch": 0.25, "learning_rate": 1.7568522876714706e-05, "loss": 1.1793, "step": 939 }, { "epoch": 0.25, "learning_rate": 1.7562883772007252e-05, "loss": 1.1097, "step": 940 }, { "epoch": 0.25, "learning_rate": 1.75572390429956e-05, "loss": 1.1455, "step": 941 }, { "epoch": 0.25, "learning_rate": 1.7551588693877576e-05, "loss": 1.1491, "step": 942 }, { "epoch": 0.25, "learning_rate": 1.7545932728855183e-05, "loss": 1.1271, "step": 943 }, { "epoch": 0.25, "learning_rate": 1.7540271152134608e-05, "loss": 1.1603, "step": 944 }, { "epoch": 0.25, "learning_rate": 1.7534603967926204e-05, "loss": 1.1436, "step": 945 }, { "epoch": 0.25, "learning_rate": 1.75289311804445e-05, "loss": 1.1841, "step": 946 }, { "epoch": 0.25, "learning_rate": 1.752325279390818e-05, "loss": 1.156, "step": 947 }, { "epoch": 0.25, "learning_rate": 1.7517568812540108e-05, "loss": 1.1682, "step": 948 }, { "epoch": 0.25, "learning_rate": 1.75118792405673e-05, "loss": 1.162, "step": 949 }, { "epoch": 0.25, "learning_rate": 1.750618408222093e-05, "loss": 1.1763, "step": 950 }, { "epoch": 0.25, "learning_rate": 1.7500483341736327e-05, "loss": 1.1834, "step": 951 }, { "epoch": 0.25, "learning_rate": 1.7494777023352974e-05, "loss": 1.1677, "step": 952 }, { "epoch": 0.25, "learning_rate": 1.7489065131314497e-05, "loss": 1.1603, "step": 953 }, { "epoch": 0.25, "learning_rate": 1.7483347669868672e-05, "loss": 1.2207, "step": 954 }, { "epoch": 0.25, "learning_rate": 1.747762464326741e-05, "loss": 1.177, "step": 955 }, { "epoch": 0.25, "learning_rate": 1.747189605576677e-05, "loss": 1.2146, "step": 956 }, { "epoch": 0.25, "learning_rate": 1.746616191162694e-05, "loss": 1.2245, "step": 957 }, { "epoch": 0.26, "learning_rate": 1.746042221511224e-05, "loss": 1.1715, "step": 958 }, { "epoch": 0.26, "learning_rate": 1.7454676970491125e-05, "loss": 1.1927, "step": 959 }, { "epoch": 0.26, "learning_rate": 1.7448926182036165e-05, "loss": 1.1921, "step": 960 }, { "epoch": 0.26, "learning_rate": 1.744316985402407e-05, "loss": 1.1534, "step": 961 }, { "epoch": 0.26, "learning_rate": 1.743740799073565e-05, "loss": 1.1649, "step": 962 }, { "epoch": 0.26, "learning_rate": 1.7431640596455848e-05, "loss": 1.1639, "step": 963 }, { "epoch": 0.26, "learning_rate": 1.7425867675473707e-05, "loss": 1.1543, "step": 964 }, { "epoch": 0.26, "learning_rate": 1.742008923208239e-05, "loss": 1.1142, "step": 965 }, { "epoch": 0.26, "learning_rate": 1.741430527057916e-05, "loss": 1.1417, "step": 966 }, { "epoch": 0.26, "learning_rate": 1.740851579526539e-05, "loss": 1.1686, "step": 967 }, { "epoch": 0.26, "learning_rate": 1.740272081044655e-05, "loss": 1.1349, "step": 968 }, { "epoch": 0.26, "learning_rate": 1.7396920320432206e-05, "loss": 1.2056, "step": 969 }, { "epoch": 0.26, "learning_rate": 1.739111432953602e-05, "loss": 1.1668, "step": 970 }, { "epoch": 0.26, "learning_rate": 1.7385302842075744e-05, "loss": 1.1418, "step": 971 }, { "epoch": 0.26, "learning_rate": 1.7379485862373216e-05, "loss": 1.159, "step": 972 }, { "epoch": 0.26, "learning_rate": 1.7373663394754367e-05, "loss": 1.1662, "step": 973 }, { "epoch": 0.26, "learning_rate": 1.736783544354919e-05, "loss": 1.1555, "step": 974 }, { "epoch": 0.26, "learning_rate": 1.7362002013091785e-05, "loss": 1.1691, "step": 975 }, { "epoch": 0.26, "learning_rate": 1.7356163107720297e-05, "loss": 1.1392, "step": 976 }, { "epoch": 0.26, "learning_rate": 1.7350318731776962e-05, "loss": 1.1673, "step": 977 }, { "epoch": 0.26, "learning_rate": 1.7344468889608078e-05, "loss": 1.1855, "step": 978 }, { "epoch": 0.26, "learning_rate": 1.733861358556401e-05, "loss": 1.1356, "step": 979 }, { "epoch": 0.26, "learning_rate": 1.7332752823999174e-05, "loss": 1.1804, "step": 980 }, { "epoch": 0.26, "learning_rate": 1.7326886609272063e-05, "loss": 1.1681, "step": 981 }, { "epoch": 0.26, "learning_rate": 1.7321014945745212e-05, "loss": 1.1259, "step": 982 }, { "epoch": 0.26, "learning_rate": 1.7315137837785215e-05, "loss": 1.1595, "step": 983 }, { "epoch": 0.26, "learning_rate": 1.730925528976271e-05, "loss": 1.2029, "step": 984 }, { "epoch": 0.26, "learning_rate": 1.7303367306052385e-05, "loss": 1.1539, "step": 985 }, { "epoch": 0.26, "learning_rate": 1.7297473891032966e-05, "loss": 1.1421, "step": 986 }, { "epoch": 0.26, "learning_rate": 1.729157504908722e-05, "loss": 1.1784, "step": 987 }, { "epoch": 0.26, "learning_rate": 1.728567078460195e-05, "loss": 1.1523, "step": 988 }, { "epoch": 0.26, "learning_rate": 1.7279761101967995e-05, "loss": 1.1636, "step": 989 }, { "epoch": 0.26, "learning_rate": 1.7273846005580212e-05, "loss": 1.1721, "step": 990 }, { "epoch": 0.26, "learning_rate": 1.7267925499837503e-05, "loss": 1.1321, "step": 991 }, { "epoch": 0.26, "learning_rate": 1.726199958914277e-05, "loss": 1.1182, "step": 992 }, { "epoch": 0.26, "learning_rate": 1.7256068277902956e-05, "loss": 1.1741, "step": 993 }, { "epoch": 0.26, "learning_rate": 1.7250131570529004e-05, "loss": 1.2052, "step": 994 }, { "epoch": 0.26, "learning_rate": 1.7244189471435885e-05, "loss": 1.1717, "step": 995 }, { "epoch": 0.27, "learning_rate": 1.723824198504256e-05, "loss": 1.2017, "step": 996 }, { "epoch": 0.27, "learning_rate": 1.7232289115772017e-05, "loss": 1.1608, "step": 997 }, { "epoch": 0.27, "learning_rate": 1.722633086805123e-05, "loss": 1.1591, "step": 998 }, { "epoch": 0.27, "learning_rate": 1.7220367246311186e-05, "loss": 1.2065, "step": 999 }, { "epoch": 0.27, "learning_rate": 1.7214398254986858e-05, "loss": 1.1738, "step": 1000 }, { "epoch": 0.27, "learning_rate": 1.7208423898517225e-05, "loss": 1.1958, "step": 1001 }, { "epoch": 0.27, "learning_rate": 1.720244418134524e-05, "loss": 1.1244, "step": 1002 }, { "epoch": 0.27, "learning_rate": 1.7196459107917857e-05, "loss": 1.1554, "step": 1003 }, { "epoch": 0.27, "learning_rate": 1.7190468682686e-05, "loss": 1.1512, "step": 1004 }, { "epoch": 0.27, "learning_rate": 1.7184472910104587e-05, "loss": 1.1499, "step": 1005 }, { "epoch": 0.27, "learning_rate": 1.7178471794632503e-05, "loss": 1.1576, "step": 1006 }, { "epoch": 0.27, "learning_rate": 1.7172465340732606e-05, "loss": 1.1712, "step": 1007 }, { "epoch": 0.27, "learning_rate": 1.716645355287173e-05, "loss": 1.1706, "step": 1008 }, { "epoch": 0.27, "learning_rate": 1.7160436435520676e-05, "loss": 1.1596, "step": 1009 }, { "epoch": 0.27, "learning_rate": 1.7154413993154196e-05, "loss": 1.1492, "step": 1010 }, { "epoch": 0.27, "learning_rate": 1.714838623025102e-05, "loss": 1.194, "step": 1011 }, { "epoch": 0.27, "learning_rate": 1.7142353151293822e-05, "loss": 1.1959, "step": 1012 }, { "epoch": 0.27, "learning_rate": 1.7136314760769234e-05, "loss": 1.1737, "step": 1013 }, { "epoch": 0.27, "learning_rate": 1.713027106316784e-05, "loss": 1.1144, "step": 1014 }, { "epoch": 0.27, "learning_rate": 1.712422206298416e-05, "loss": 1.1564, "step": 1015 }, { "epoch": 0.27, "learning_rate": 1.7118167764716674e-05, "loss": 1.1615, "step": 1016 }, { "epoch": 0.27, "learning_rate": 1.711210817286779e-05, "loss": 1.1726, "step": 1017 }, { "epoch": 0.27, "learning_rate": 1.7106043291943857e-05, "loss": 1.1671, "step": 1018 }, { "epoch": 0.27, "learning_rate": 1.7099973126455153e-05, "loss": 1.1696, "step": 1019 }, { "epoch": 0.27, "learning_rate": 1.70938976809159e-05, "loss": 1.1543, "step": 1020 }, { "epoch": 0.27, "learning_rate": 1.7087816959844224e-05, "loss": 1.2106, "step": 1021 }, { "epoch": 0.27, "learning_rate": 1.708173096776219e-05, "loss": 1.2044, "step": 1022 }, { "epoch": 0.27, "learning_rate": 1.7075639709195784e-05, "loss": 1.147, "step": 1023 }, { "epoch": 0.27, "learning_rate": 1.70695431886749e-05, "loss": 1.1781, "step": 1024 }, { "epoch": 0.27, "learning_rate": 1.706344141073335e-05, "loss": 1.1669, "step": 1025 }, { "epoch": 0.27, "learning_rate": 1.7057334379908852e-05, "loss": 1.1623, "step": 1026 }, { "epoch": 0.27, "learning_rate": 1.7051222100743036e-05, "loss": 1.1464, "step": 1027 }, { "epoch": 0.27, "learning_rate": 1.7045104577781436e-05, "loss": 1.2071, "step": 1028 }, { "epoch": 0.27, "learning_rate": 1.7038981815573478e-05, "loss": 1.1638, "step": 1029 }, { "epoch": 0.27, "learning_rate": 1.7032853818672484e-05, "loss": 1.1718, "step": 1030 }, { "epoch": 0.27, "learning_rate": 1.702672059163568e-05, "loss": 1.1618, "step": 1031 }, { "epoch": 0.27, "learning_rate": 1.702058213902417e-05, "loss": 1.1809, "step": 1032 }, { "epoch": 0.28, "learning_rate": 1.7014438465402952e-05, "loss": 1.1354, "step": 1033 }, { "epoch": 0.28, "learning_rate": 1.7008289575340903e-05, "loss": 1.1609, "step": 1034 }, { "epoch": 0.28, "learning_rate": 1.700213547341078e-05, "loss": 1.1728, "step": 1035 }, { "epoch": 0.28, "learning_rate": 1.699597616418921e-05, "loss": 1.1284, "step": 1036 }, { "epoch": 0.28, "learning_rate": 1.698981165225671e-05, "loss": 1.094, "step": 1037 }, { "epoch": 0.28, "learning_rate": 1.6983641942197644e-05, "loss": 1.12, "step": 1038 }, { "epoch": 0.28, "learning_rate": 1.6977467038600255e-05, "loss": 1.1061, "step": 1039 }, { "epoch": 0.28, "learning_rate": 1.6971286946056648e-05, "loss": 1.1363, "step": 1040 }, { "epoch": 0.28, "learning_rate": 1.696510166916278e-05, "loss": 1.0969, "step": 1041 }, { "epoch": 0.28, "learning_rate": 1.695891121251847e-05, "loss": 1.1184, "step": 1042 }, { "epoch": 0.28, "learning_rate": 1.6952715580727387e-05, "loss": 1.1475, "step": 1043 }, { "epoch": 0.28, "learning_rate": 1.6946514778397047e-05, "loss": 1.1417, "step": 1044 }, { "epoch": 0.28, "learning_rate": 1.6940308810138814e-05, "loss": 1.1579, "step": 1045 }, { "epoch": 0.28, "learning_rate": 1.693409768056789e-05, "loss": 1.1084, "step": 1046 }, { "epoch": 0.28, "learning_rate": 1.692788139430332e-05, "loss": 1.1561, "step": 1047 }, { "epoch": 0.28, "learning_rate": 1.6921659955967983e-05, "loss": 1.1394, "step": 1048 }, { "epoch": 0.28, "learning_rate": 1.691543337018858e-05, "loss": 1.1858, "step": 1049 }, { "epoch": 0.28, "learning_rate": 1.690920164159566e-05, "loss": 1.1713, "step": 1050 }, { "epoch": 0.28, "learning_rate": 1.690296477482357e-05, "loss": 1.1765, "step": 1051 }, { "epoch": 0.28, "learning_rate": 1.6896722774510505e-05, "loss": 1.1623, "step": 1052 }, { "epoch": 0.28, "learning_rate": 1.689047564529846e-05, "loss": 1.1913, "step": 1053 }, { "epoch": 0.28, "learning_rate": 1.6884223391833253e-05, "loss": 1.1594, "step": 1054 }, { "epoch": 0.28, "learning_rate": 1.68779660187645e-05, "loss": 1.1304, "step": 1055 }, { "epoch": 0.28, "learning_rate": 1.6871703530745645e-05, "loss": 1.1706, "step": 1056 }, { "epoch": 0.28, "learning_rate": 1.6865435932433918e-05, "loss": 1.1564, "step": 1057 }, { "epoch": 0.28, "learning_rate": 1.6859163228490358e-05, "loss": 1.1616, "step": 1058 }, { "epoch": 0.28, "learning_rate": 1.68528854235798e-05, "loss": 1.1535, "step": 1059 }, { "epoch": 0.28, "learning_rate": 1.6846602522370864e-05, "loss": 1.1788, "step": 1060 }, { "epoch": 0.28, "learning_rate": 1.6840314529535976e-05, "loss": 1.1771, "step": 1061 }, { "epoch": 0.28, "learning_rate": 1.6834021449751333e-05, "loss": 1.1725, "step": 1062 }, { "epoch": 0.28, "learning_rate": 1.6827723287696927e-05, "loss": 1.1635, "step": 1063 }, { "epoch": 0.28, "learning_rate": 1.682142004805652e-05, "loss": 1.134, "step": 1064 }, { "epoch": 0.28, "learning_rate": 1.6815111735517656e-05, "loss": 1.2084, "step": 1065 }, { "epoch": 0.28, "learning_rate": 1.680879835477165e-05, "loss": 1.1573, "step": 1066 }, { "epoch": 0.28, "learning_rate": 1.6802479910513584e-05, "loss": 1.1606, "step": 1067 }, { "epoch": 0.28, "learning_rate": 1.6796156407442312e-05, "loss": 1.2046, "step": 1068 }, { "epoch": 0.28, "learning_rate": 1.678982785026044e-05, "loss": 1.1915, "step": 1069 }, { "epoch": 0.28, "learning_rate": 1.6783494243674343e-05, "loss": 1.1655, "step": 1070 }, { "epoch": 0.29, "learning_rate": 1.6777155592394145e-05, "loss": 1.2018, "step": 1071 }, { "epoch": 0.29, "learning_rate": 1.677081190113372e-05, "loss": 1.1975, "step": 1072 }, { "epoch": 0.29, "learning_rate": 1.6764463174610696e-05, "loss": 1.1274, "step": 1073 }, { "epoch": 0.29, "learning_rate": 1.675810941754644e-05, "loss": 1.1611, "step": 1074 }, { "epoch": 0.29, "learning_rate": 1.6751750634666067e-05, "loss": 1.1234, "step": 1075 }, { "epoch": 0.29, "learning_rate": 1.6745386830698417e-05, "loss": 1.1486, "step": 1076 }, { "epoch": 0.29, "learning_rate": 1.6739018010376085e-05, "loss": 1.1646, "step": 1077 }, { "epoch": 0.29, "learning_rate": 1.6732644178435375e-05, "loss": 1.1842, "step": 1078 }, { "epoch": 0.29, "learning_rate": 1.6726265339616322e-05, "loss": 1.1832, "step": 1079 }, { "epoch": 0.29, "learning_rate": 1.67198814986627e-05, "loss": 1.2032, "step": 1080 }, { "epoch": 0.29, "learning_rate": 1.671349266032199e-05, "loss": 1.144, "step": 1081 }, { "epoch": 0.29, "learning_rate": 1.6707098829345384e-05, "loss": 1.1573, "step": 1082 }, { "epoch": 0.29, "learning_rate": 1.6700700010487804e-05, "loss": 1.1755, "step": 1083 }, { "epoch": 0.29, "learning_rate": 1.6694296208507863e-05, "loss": 1.1691, "step": 1084 }, { "epoch": 0.29, "learning_rate": 1.66878874281679e-05, "loss": 1.1443, "step": 1085 }, { "epoch": 0.29, "learning_rate": 1.668147367423393e-05, "loss": 1.2002, "step": 1086 }, { "epoch": 0.29, "learning_rate": 1.667505495147569e-05, "loss": 1.148, "step": 1087 }, { "epoch": 0.29, "learning_rate": 1.666863126466661e-05, "loss": 1.1808, "step": 1088 }, { "epoch": 0.29, "learning_rate": 1.6662202618583797e-05, "loss": 1.1725, "step": 1089 }, { "epoch": 0.29, "learning_rate": 1.6655769018008052e-05, "loss": 1.1823, "step": 1090 }, { "epoch": 0.29, "learning_rate": 1.664933046772387e-05, "loss": 1.1514, "step": 1091 }, { "epoch": 0.29, "learning_rate": 1.6642886972519416e-05, "loss": 1.1657, "step": 1092 }, { "epoch": 0.29, "learning_rate": 1.663643853718654e-05, "loss": 1.1932, "step": 1093 }, { "epoch": 0.29, "learning_rate": 1.662998516652076e-05, "loss": 1.1717, "step": 1094 }, { "epoch": 0.29, "learning_rate": 1.6623526865321268e-05, "loss": 1.1625, "step": 1095 }, { "epoch": 0.29, "learning_rate": 1.6617063638390913e-05, "loss": 1.1608, "step": 1096 }, { "epoch": 0.29, "learning_rate": 1.6610595490536227e-05, "loss": 1.1328, "step": 1097 }, { "epoch": 0.29, "learning_rate": 1.6604122426567384e-05, "loss": 1.1157, "step": 1098 }, { "epoch": 0.29, "learning_rate": 1.6597644451298218e-05, "loss": 1.12, "step": 1099 }, { "epoch": 0.29, "learning_rate": 1.6591161569546222e-05, "loss": 1.1499, "step": 1100 }, { "epoch": 0.29, "learning_rate": 1.658467378613253e-05, "loss": 1.1229, "step": 1101 }, { "epoch": 0.29, "learning_rate": 1.6578181105881925e-05, "loss": 1.1667, "step": 1102 }, { "epoch": 0.29, "learning_rate": 1.657168353362283e-05, "loss": 1.1031, "step": 1103 }, { "epoch": 0.29, "learning_rate": 1.656518107418731e-05, "loss": 1.0891, "step": 1104 }, { "epoch": 0.29, "learning_rate": 1.655867373241106e-05, "loss": 1.1291, "step": 1105 }, { "epoch": 0.29, "learning_rate": 1.6552161513133404e-05, "loss": 1.0993, "step": 1106 }, { "epoch": 0.29, "learning_rate": 1.65456444211973e-05, "loss": 1.1176, "step": 1107 }, { "epoch": 0.29, "learning_rate": 1.6539122461449325e-05, "loss": 1.1208, "step": 1108 }, { "epoch": 0.3, "learning_rate": 1.653259563873968e-05, "loss": 1.1251, "step": 1109 }, { "epoch": 0.3, "learning_rate": 1.6526063957922174e-05, "loss": 1.1321, "step": 1110 }, { "epoch": 0.3, "learning_rate": 1.651952742385424e-05, "loss": 1.1282, "step": 1111 }, { "epoch": 0.3, "learning_rate": 1.651298604139691e-05, "loss": 1.1054, "step": 1112 }, { "epoch": 0.3, "learning_rate": 1.650643981541483e-05, "loss": 1.1438, "step": 1113 }, { "epoch": 0.3, "learning_rate": 1.6499888750776243e-05, "loss": 1.1286, "step": 1114 }, { "epoch": 0.3, "learning_rate": 1.6493332852352996e-05, "loss": 1.1425, "step": 1115 }, { "epoch": 0.3, "learning_rate": 1.6486772125020518e-05, "loss": 1.1472, "step": 1116 }, { "epoch": 0.3, "learning_rate": 1.648020657365784e-05, "loss": 1.1461, "step": 1117 }, { "epoch": 0.3, "learning_rate": 1.6473636203147584e-05, "loss": 1.1381, "step": 1118 }, { "epoch": 0.3, "learning_rate": 1.6467061018375946e-05, "loss": 1.1543, "step": 1119 }, { "epoch": 0.3, "learning_rate": 1.6460481024232705e-05, "loss": 1.1167, "step": 1120 }, { "epoch": 0.3, "learning_rate": 1.645389622561122e-05, "loss": 1.154, "step": 1121 }, { "epoch": 0.3, "learning_rate": 1.6447306627408418e-05, "loss": 1.1693, "step": 1122 }, { "epoch": 0.3, "learning_rate": 1.64407122345248e-05, "loss": 1.2104, "step": 1123 }, { "epoch": 0.3, "learning_rate": 1.6434113051864435e-05, "loss": 1.1392, "step": 1124 }, { "epoch": 0.3, "learning_rate": 1.6427509084334943e-05, "loss": 1.1358, "step": 1125 }, { "epoch": 0.3, "learning_rate": 1.642090033684751e-05, "loss": 1.0991, "step": 1126 }, { "epoch": 0.3, "learning_rate": 1.641428681431688e-05, "loss": 1.1138, "step": 1127 }, { "epoch": 0.3, "learning_rate": 1.640766852166134e-05, "loss": 1.0821, "step": 1128 }, { "epoch": 0.3, "learning_rate": 1.6401045463802727e-05, "loss": 1.1131, "step": 1129 }, { "epoch": 0.3, "learning_rate": 1.6394417645666432e-05, "loss": 1.1647, "step": 1130 }, { "epoch": 0.3, "learning_rate": 1.6387785072181364e-05, "loss": 1.1049, "step": 1131 }, { "epoch": 0.3, "learning_rate": 1.638114774827999e-05, "loss": 1.144, "step": 1132 }, { "epoch": 0.3, "learning_rate": 1.6374505678898304e-05, "loss": 1.1585, "step": 1133 }, { "epoch": 0.3, "learning_rate": 1.6367858868975818e-05, "loss": 1.1707, "step": 1134 }, { "epoch": 0.3, "learning_rate": 1.6361207323455585e-05, "loss": 1.1697, "step": 1135 }, { "epoch": 0.3, "learning_rate": 1.635455104728417e-05, "loss": 1.1287, "step": 1136 }, { "epoch": 0.3, "learning_rate": 1.6347890045411655e-05, "loss": 1.1548, "step": 1137 }, { "epoch": 0.3, "learning_rate": 1.6341224322791643e-05, "loss": 1.1526, "step": 1138 }, { "epoch": 0.3, "learning_rate": 1.6334553884381248e-05, "loss": 1.1754, "step": 1139 }, { "epoch": 0.3, "learning_rate": 1.6327878735141084e-05, "loss": 1.2005, "step": 1140 }, { "epoch": 0.3, "learning_rate": 1.632119888003527e-05, "loss": 1.1708, "step": 1141 }, { "epoch": 0.3, "learning_rate": 1.631451432403143e-05, "loss": 1.1979, "step": 1142 }, { "epoch": 0.3, "learning_rate": 1.630782507210068e-05, "loss": 1.1959, "step": 1143 }, { "epoch": 0.3, "learning_rate": 1.6301131129217624e-05, "loss": 1.1347, "step": 1144 }, { "epoch": 0.3, "learning_rate": 1.6294432500360365e-05, "loss": 1.1546, "step": 1145 }, { "epoch": 0.31, "learning_rate": 1.628772919051049e-05, "loss": 1.1683, "step": 1146 }, { "epoch": 0.31, "learning_rate": 1.628102120465305e-05, "loss": 1.1207, "step": 1147 }, { "epoch": 0.31, "learning_rate": 1.627430854777659e-05, "loss": 1.1382, "step": 1148 }, { "epoch": 0.31, "learning_rate": 1.6267591224873128e-05, "loss": 1.0964, "step": 1149 }, { "epoch": 0.31, "learning_rate": 1.626086924093815e-05, "loss": 1.1192, "step": 1150 }, { "epoch": 0.31, "learning_rate": 1.6254142600970597e-05, "loss": 1.1541, "step": 1151 }, { "epoch": 0.31, "learning_rate": 1.624741130997289e-05, "loss": 1.1452, "step": 1152 }, { "epoch": 0.31, "learning_rate": 1.62406753729509e-05, "loss": 1.1622, "step": 1153 }, { "epoch": 0.31, "learning_rate": 1.623393479491395e-05, "loss": 1.1603, "step": 1154 }, { "epoch": 0.31, "learning_rate": 1.6227189580874822e-05, "loss": 1.1533, "step": 1155 }, { "epoch": 0.31, "learning_rate": 1.6220439735849747e-05, "loss": 1.1594, "step": 1156 }, { "epoch": 0.31, "learning_rate": 1.6213685264858386e-05, "loss": 1.1652, "step": 1157 }, { "epoch": 0.31, "learning_rate": 1.6206926172923853e-05, "loss": 1.1905, "step": 1158 }, { "epoch": 0.31, "learning_rate": 1.6200162465072697e-05, "loss": 1.1386, "step": 1159 }, { "epoch": 0.31, "learning_rate": 1.619339414633489e-05, "loss": 1.1455, "step": 1160 }, { "epoch": 0.31, "learning_rate": 1.618662122174385e-05, "loss": 1.1274, "step": 1161 }, { "epoch": 0.31, "learning_rate": 1.617984369633641e-05, "loss": 1.181, "step": 1162 }, { "epoch": 0.31, "learning_rate": 1.6173061575152816e-05, "loss": 1.0885, "step": 1163 }, { "epoch": 0.31, "learning_rate": 1.6166274863236752e-05, "loss": 1.1055, "step": 1164 }, { "epoch": 0.31, "learning_rate": 1.6159483565635297e-05, "loss": 1.0976, "step": 1165 }, { "epoch": 0.31, "learning_rate": 1.6152687687398953e-05, "loss": 1.1211, "step": 1166 }, { "epoch": 0.31, "learning_rate": 1.614588723358162e-05, "loss": 1.1201, "step": 1167 }, { "epoch": 0.31, "learning_rate": 1.6139082209240608e-05, "loss": 1.1473, "step": 1168 }, { "epoch": 0.31, "learning_rate": 1.613227261943662e-05, "loss": 1.1235, "step": 1169 }, { "epoch": 0.31, "learning_rate": 1.6125458469233757e-05, "loss": 1.1163, "step": 1170 }, { "epoch": 0.31, "learning_rate": 1.6118639763699513e-05, "loss": 1.1251, "step": 1171 }, { "epoch": 0.31, "learning_rate": 1.6111816507904764e-05, "loss": 1.1005, "step": 1172 }, { "epoch": 0.31, "learning_rate": 1.6104988706923777e-05, "loss": 1.1236, "step": 1173 }, { "epoch": 0.31, "learning_rate": 1.6098156365834198e-05, "loss": 1.1207, "step": 1174 }, { "epoch": 0.31, "learning_rate": 1.609131948971704e-05, "loss": 1.1339, "step": 1175 }, { "epoch": 0.31, "learning_rate": 1.60844780836567e-05, "loss": 1.1559, "step": 1176 }, { "epoch": 0.31, "learning_rate": 1.607763215274094e-05, "loss": 1.1188, "step": 1177 }, { "epoch": 0.31, "learning_rate": 1.6070781702060883e-05, "loss": 1.1203, "step": 1178 }, { "epoch": 0.31, "learning_rate": 1.6063926736711024e-05, "loss": 1.1691, "step": 1179 }, { "epoch": 0.31, "learning_rate": 1.6057067261789202e-05, "loss": 1.1068, "step": 1180 }, { "epoch": 0.31, "learning_rate": 1.6050203282396616e-05, "loss": 1.2101, "step": 1181 }, { "epoch": 0.31, "learning_rate": 1.6043334803637815e-05, "loss": 1.1752, "step": 1182 }, { "epoch": 0.31, "learning_rate": 1.6036461830620696e-05, "loss": 1.1231, "step": 1183 }, { "epoch": 0.32, "learning_rate": 1.6029584368456496e-05, "loss": 1.1383, "step": 1184 }, { "epoch": 0.32, "learning_rate": 1.602270242225979e-05, "loss": 1.1318, "step": 1185 }, { "epoch": 0.32, "learning_rate": 1.601581599714848e-05, "loss": 1.1342, "step": 1186 }, { "epoch": 0.32, "learning_rate": 1.6008925098243818e-05, "loss": 1.1308, "step": 1187 }, { "epoch": 0.32, "learning_rate": 1.600202973067037e-05, "loss": 1.1431, "step": 1188 }, { "epoch": 0.32, "learning_rate": 1.5995129899556022e-05, "loss": 1.1571, "step": 1189 }, { "epoch": 0.32, "learning_rate": 1.598822561003199e-05, "loss": 1.1571, "step": 1190 }, { "epoch": 0.32, "learning_rate": 1.5981316867232795e-05, "loss": 1.1462, "step": 1191 }, { "epoch": 0.32, "learning_rate": 1.597440367629628e-05, "loss": 1.1503, "step": 1192 }, { "epoch": 0.32, "learning_rate": 1.5967486042363588e-05, "loss": 1.1386, "step": 1193 }, { "epoch": 0.32, "learning_rate": 1.5960563970579167e-05, "loss": 1.1703, "step": 1194 }, { "epoch": 0.32, "learning_rate": 1.5953637466090772e-05, "loss": 1.1216, "step": 1195 }, { "epoch": 0.32, "learning_rate": 1.5946706534049446e-05, "loss": 1.1625, "step": 1196 }, { "epoch": 0.32, "learning_rate": 1.593977117960953e-05, "loss": 1.1458, "step": 1197 }, { "epoch": 0.32, "learning_rate": 1.5932831407928658e-05, "loss": 1.1368, "step": 1198 }, { "epoch": 0.32, "learning_rate": 1.5925887224167734e-05, "loss": 1.1297, "step": 1199 }, { "epoch": 0.32, "learning_rate": 1.5918938633490956e-05, "loss": 1.2047, "step": 1200 }, { "epoch": 0.32, "learning_rate": 1.5911985641065795e-05, "loss": 1.127, "step": 1201 }, { "epoch": 0.32, "learning_rate": 1.5905028252062996e-05, "loss": 1.1673, "step": 1202 }, { "epoch": 0.32, "learning_rate": 1.5898066471656578e-05, "loss": 1.165, "step": 1203 }, { "epoch": 0.32, "learning_rate": 1.5891100305023816e-05, "loss": 1.1677, "step": 1204 }, { "epoch": 0.32, "learning_rate": 1.5884129757345252e-05, "loss": 1.1123, "step": 1205 }, { "epoch": 0.32, "learning_rate": 1.5877154833804693e-05, "loss": 1.0734, "step": 1206 }, { "epoch": 0.32, "learning_rate": 1.5870175539589187e-05, "loss": 1.1043, "step": 1207 }, { "epoch": 0.32, "learning_rate": 1.5863191879889043e-05, "loss": 1.12, "step": 1208 }, { "epoch": 0.32, "learning_rate": 1.5856203859897813e-05, "loss": 1.0884, "step": 1209 }, { "epoch": 0.32, "learning_rate": 1.5849211484812288e-05, "loss": 1.0987, "step": 1210 }, { "epoch": 0.32, "learning_rate": 1.58422147598325e-05, "loss": 1.1124, "step": 1211 }, { "epoch": 0.32, "learning_rate": 1.5835213690161723e-05, "loss": 1.1162, "step": 1212 }, { "epoch": 0.32, "learning_rate": 1.582820828100645e-05, "loss": 1.0641, "step": 1213 }, { "epoch": 0.32, "learning_rate": 1.5821198537576407e-05, "loss": 1.1112, "step": 1214 }, { "epoch": 0.32, "learning_rate": 1.581418446508455e-05, "loss": 1.0821, "step": 1215 }, { "epoch": 0.32, "learning_rate": 1.5807166068747038e-05, "loss": 1.1012, "step": 1216 }, { "epoch": 0.32, "learning_rate": 1.5800143353783262e-05, "loss": 1.0745, "step": 1217 }, { "epoch": 0.32, "learning_rate": 1.5793116325415815e-05, "loss": 1.0913, "step": 1218 }, { "epoch": 0.32, "learning_rate": 1.5786084988870506e-05, "loss": 1.1275, "step": 1219 }, { "epoch": 0.32, "learning_rate": 1.5779049349376334e-05, "loss": 1.0709, "step": 1220 }, { "epoch": 0.33, "learning_rate": 1.577200941216551e-05, "loss": 1.0845, "step": 1221 }, { "epoch": 0.33, "learning_rate": 1.5764965182473438e-05, "loss": 1.0886, "step": 1222 }, { "epoch": 0.33, "learning_rate": 1.5757916665538714e-05, "loss": 1.0439, "step": 1223 }, { "epoch": 0.33, "learning_rate": 1.5750863866603126e-05, "loss": 1.0604, "step": 1224 }, { "epoch": 0.33, "learning_rate": 1.5743806790911634e-05, "loss": 1.1129, "step": 1225 }, { "epoch": 0.33, "learning_rate": 1.573674544371239e-05, "loss": 1.0853, "step": 1226 }, { "epoch": 0.33, "learning_rate": 1.5729679830256723e-05, "loss": 1.0975, "step": 1227 }, { "epoch": 0.33, "learning_rate": 1.572260995579913e-05, "loss": 1.0844, "step": 1228 }, { "epoch": 0.33, "learning_rate": 1.5715535825597276e-05, "loss": 1.0726, "step": 1229 }, { "epoch": 0.33, "learning_rate": 1.570845744491199e-05, "loss": 1.0768, "step": 1230 }, { "epoch": 0.33, "learning_rate": 1.570137481900727e-05, "loss": 1.0794, "step": 1231 }, { "epoch": 0.33, "learning_rate": 1.5694287953150266e-05, "loss": 1.0734, "step": 1232 }, { "epoch": 0.33, "learning_rate": 1.568719685261128e-05, "loss": 1.1038, "step": 1233 }, { "epoch": 0.33, "learning_rate": 1.5680101522663762e-05, "loss": 1.0856, "step": 1234 }, { "epoch": 0.33, "learning_rate": 1.5673001968584313e-05, "loss": 1.0749, "step": 1235 }, { "epoch": 0.33, "learning_rate": 1.566589819565267e-05, "loss": 1.0827, "step": 1236 }, { "epoch": 0.33, "learning_rate": 1.5658790209151712e-05, "loss": 1.119, "step": 1237 }, { "epoch": 0.33, "learning_rate": 1.5651678014367445e-05, "loss": 1.1672, "step": 1238 }, { "epoch": 0.33, "learning_rate": 1.564456161658901e-05, "loss": 1.1076, "step": 1239 }, { "epoch": 0.33, "learning_rate": 1.5637441021108667e-05, "loss": 1.169, "step": 1240 }, { "epoch": 0.33, "learning_rate": 1.5630316233221817e-05, "loss": 1.164, "step": 1241 }, { "epoch": 0.33, "learning_rate": 1.562318725822695e-05, "loss": 1.1126, "step": 1242 }, { "epoch": 0.33, "learning_rate": 1.5616054101425692e-05, "loss": 1.0915, "step": 1243 }, { "epoch": 0.33, "learning_rate": 1.5608916768122768e-05, "loss": 1.071, "step": 1244 }, { "epoch": 0.33, "learning_rate": 1.5601775263626013e-05, "loss": 1.1608, "step": 1245 }, { "epoch": 0.33, "learning_rate": 1.5594629593246366e-05, "loss": 1.1283, "step": 1246 }, { "epoch": 0.33, "learning_rate": 1.558747976229786e-05, "loss": 1.0933, "step": 1247 }, { "epoch": 0.33, "learning_rate": 1.5580325776097616e-05, "loss": 1.1605, "step": 1248 }, { "epoch": 0.33, "learning_rate": 1.557316763996587e-05, "loss": 1.0983, "step": 1249 }, { "epoch": 0.33, "learning_rate": 1.5566005359225907e-05, "loss": 1.1068, "step": 1250 }, { "epoch": 0.33, "learning_rate": 1.5558838939204126e-05, "loss": 1.0671, "step": 1251 }, { "epoch": 0.33, "learning_rate": 1.555166838522999e-05, "loss": 1.0948, "step": 1252 }, { "epoch": 0.33, "learning_rate": 1.5544493702636033e-05, "loss": 1.1043, "step": 1253 }, { "epoch": 0.33, "learning_rate": 1.5537314896757877e-05, "loss": 1.1255, "step": 1254 }, { "epoch": 0.33, "learning_rate": 1.5530131972934187e-05, "loss": 1.073, "step": 1255 }, { "epoch": 0.33, "learning_rate": 1.55229449365067e-05, "loss": 1.0816, "step": 1256 }, { "epoch": 0.33, "learning_rate": 1.5515753792820226e-05, "loss": 1.0969, "step": 1257 }, { "epoch": 0.33, "learning_rate": 1.550855854722261e-05, "loss": 1.1425, "step": 1258 }, { "epoch": 0.34, "learning_rate": 1.5501359205064747e-05, "loss": 1.1145, "step": 1259 }, { "epoch": 0.34, "learning_rate": 1.5494155771700597e-05, "loss": 1.1266, "step": 1260 }, { "epoch": 0.34, "learning_rate": 1.5486948252487146e-05, "loss": 1.1177, "step": 1261 }, { "epoch": 0.34, "learning_rate": 1.547973665278442e-05, "loss": 1.1063, "step": 1262 }, { "epoch": 0.34, "learning_rate": 1.5472520977955485e-05, "loss": 1.1757, "step": 1263 }, { "epoch": 0.34, "learning_rate": 1.5465301233366442e-05, "loss": 1.1255, "step": 1264 }, { "epoch": 0.34, "learning_rate": 1.5458077424386404e-05, "loss": 1.1397, "step": 1265 }, { "epoch": 0.34, "learning_rate": 1.545084955638752e-05, "loss": 1.1611, "step": 1266 }, { "epoch": 0.34, "learning_rate": 1.5443617634744948e-05, "loss": 1.1165, "step": 1267 }, { "epoch": 0.34, "learning_rate": 1.5436381664836867e-05, "loss": 1.177, "step": 1268 }, { "epoch": 0.34, "learning_rate": 1.5429141652044465e-05, "loss": 1.1543, "step": 1269 }, { "epoch": 0.34, "learning_rate": 1.542189760175194e-05, "loss": 1.1291, "step": 1270 }, { "epoch": 0.34, "learning_rate": 1.5414649519346482e-05, "loss": 1.1022, "step": 1271 }, { "epoch": 0.34, "learning_rate": 1.5407397410218286e-05, "loss": 1.0726, "step": 1272 }, { "epoch": 0.34, "learning_rate": 1.540014127976055e-05, "loss": 1.1152, "step": 1273 }, { "epoch": 0.34, "learning_rate": 1.5392881133369445e-05, "loss": 1.1191, "step": 1274 }, { "epoch": 0.34, "learning_rate": 1.5385616976444143e-05, "loss": 1.0871, "step": 1275 }, { "epoch": 0.34, "learning_rate": 1.5378348814386796e-05, "loss": 1.0669, "step": 1276 }, { "epoch": 0.34, "learning_rate": 1.5371076652602525e-05, "loss": 1.1178, "step": 1277 }, { "epoch": 0.34, "learning_rate": 1.5363800496499438e-05, "loss": 1.0949, "step": 1278 }, { "epoch": 0.34, "learning_rate": 1.5356520351488606e-05, "loss": 1.1128, "step": 1279 }, { "epoch": 0.34, "learning_rate": 1.5349236222984067e-05, "loss": 1.0938, "step": 1280 }, { "epoch": 0.34, "learning_rate": 1.5341948116402826e-05, "loss": 1.1417, "step": 1281 }, { "epoch": 0.34, "learning_rate": 1.5334656037164838e-05, "loss": 1.1409, "step": 1282 }, { "epoch": 0.34, "learning_rate": 1.5327359990693018e-05, "loss": 1.095, "step": 1283 }, { "epoch": 0.34, "learning_rate": 1.5320059982413238e-05, "loss": 1.1283, "step": 1284 }, { "epoch": 0.34, "learning_rate": 1.53127560177543e-05, "loss": 1.0951, "step": 1285 }, { "epoch": 0.34, "learning_rate": 1.5305448102147968e-05, "loss": 1.0976, "step": 1286 }, { "epoch": 0.34, "learning_rate": 1.5298136241028924e-05, "loss": 1.0905, "step": 1287 }, { "epoch": 0.34, "learning_rate": 1.5290820439834795e-05, "loss": 1.0867, "step": 1288 }, { "epoch": 0.34, "learning_rate": 1.5283500704006142e-05, "loss": 1.0403, "step": 1289 }, { "epoch": 0.34, "learning_rate": 1.527617703898644e-05, "loss": 1.095, "step": 1290 }, { "epoch": 0.34, "learning_rate": 1.52688494502221e-05, "loss": 1.105, "step": 1291 }, { "epoch": 0.34, "learning_rate": 1.526151794316244e-05, "loss": 1.0759, "step": 1292 }, { "epoch": 0.34, "learning_rate": 1.5254182523259697e-05, "loss": 1.0853, "step": 1293 }, { "epoch": 0.34, "learning_rate": 1.5246843195969013e-05, "loss": 1.0508, "step": 1294 }, { "epoch": 0.34, "learning_rate": 1.5239499966748446e-05, "loss": 1.0962, "step": 1295 }, { "epoch": 0.35, "learning_rate": 1.523215284105894e-05, "loss": 1.0406, "step": 1296 }, { "epoch": 0.35, "learning_rate": 1.5224801824364356e-05, "loss": 1.1071, "step": 1297 }, { "epoch": 0.35, "learning_rate": 1.5217446922131436e-05, "loss": 1.0491, "step": 1298 }, { "epoch": 0.35, "learning_rate": 1.5210088139829809e-05, "loss": 1.0794, "step": 1299 }, { "epoch": 0.35, "learning_rate": 1.5202725482932002e-05, "loss": 1.086, "step": 1300 }, { "epoch": 0.35, "learning_rate": 1.519535895691341e-05, "loss": 1.0452, "step": 1301 }, { "epoch": 0.35, "learning_rate": 1.5187988567252311e-05, "loss": 1.0737, "step": 1302 }, { "epoch": 0.35, "learning_rate": 1.5180614319429867e-05, "loss": 1.0881, "step": 1303 }, { "epoch": 0.35, "learning_rate": 1.5173236218930086e-05, "loss": 1.0856, "step": 1304 }, { "epoch": 0.35, "learning_rate": 1.5165854271239859e-05, "loss": 1.0533, "step": 1305 }, { "epoch": 0.35, "learning_rate": 1.5158468481848937e-05, "loss": 1.0849, "step": 1306 }, { "epoch": 0.35, "learning_rate": 1.515107885624992e-05, "loss": 1.0107, "step": 1307 }, { "epoch": 0.35, "learning_rate": 1.5143685399938264e-05, "loss": 1.0559, "step": 1308 }, { "epoch": 0.35, "learning_rate": 1.5136288118412277e-05, "loss": 1.0697, "step": 1309 }, { "epoch": 0.35, "learning_rate": 1.5128887017173111e-05, "loss": 1.0931, "step": 1310 }, { "epoch": 0.35, "learning_rate": 1.5121482101724752e-05, "loss": 1.0329, "step": 1311 }, { "epoch": 0.35, "learning_rate": 1.5114073377574033e-05, "loss": 1.0901, "step": 1312 }, { "epoch": 0.35, "learning_rate": 1.510666085023061e-05, "loss": 1.0545, "step": 1313 }, { "epoch": 0.35, "learning_rate": 1.5099244525206973e-05, "loss": 1.0723, "step": 1314 }, { "epoch": 0.35, "learning_rate": 1.5091824408018431e-05, "loss": 1.0922, "step": 1315 }, { "epoch": 0.35, "learning_rate": 1.5084400504183125e-05, "loss": 1.0841, "step": 1316 }, { "epoch": 0.35, "learning_rate": 1.5076972819222e-05, "loss": 1.0763, "step": 1317 }, { "epoch": 0.35, "learning_rate": 1.5069541358658813e-05, "loss": 1.084, "step": 1318 }, { "epoch": 0.35, "learning_rate": 1.5062106128020135e-05, "loss": 1.094, "step": 1319 }, { "epoch": 0.35, "learning_rate": 1.505466713283534e-05, "loss": 1.069, "step": 1320 }, { "epoch": 0.35, "learning_rate": 1.5047224378636598e-05, "loss": 1.0875, "step": 1321 }, { "epoch": 0.35, "learning_rate": 1.5039777870958875e-05, "loss": 1.1269, "step": 1322 }, { "epoch": 0.35, "learning_rate": 1.503232761533993e-05, "loss": 1.1072, "step": 1323 }, { "epoch": 0.35, "learning_rate": 1.502487361732031e-05, "loss": 1.1138, "step": 1324 }, { "epoch": 0.35, "learning_rate": 1.5017415882443345e-05, "loss": 1.1142, "step": 1325 }, { "epoch": 0.35, "learning_rate": 1.500995441625514e-05, "loss": 1.0614, "step": 1326 }, { "epoch": 0.35, "learning_rate": 1.500248922430458e-05, "loss": 1.0988, "step": 1327 }, { "epoch": 0.35, "learning_rate": 1.4995020312143318e-05, "loss": 1.1056, "step": 1328 }, { "epoch": 0.35, "learning_rate": 1.4987547685325772e-05, "loss": 1.1094, "step": 1329 }, { "epoch": 0.35, "learning_rate": 1.4980071349409127e-05, "loss": 1.0749, "step": 1330 }, { "epoch": 0.35, "learning_rate": 1.4972591309953323e-05, "loss": 1.1418, "step": 1331 }, { "epoch": 0.35, "learning_rate": 1.4965107572521056e-05, "loss": 1.1592, "step": 1332 }, { "epoch": 0.35, "learning_rate": 1.4957620142677774e-05, "loss": 1.0896, "step": 1333 }, { "epoch": 0.36, "learning_rate": 1.4950129025991662e-05, "loss": 1.0869, "step": 1334 }, { "epoch": 0.36, "learning_rate": 1.4942634228033656e-05, "loss": 1.1238, "step": 1335 }, { "epoch": 0.36, "learning_rate": 1.493513575437743e-05, "loss": 1.1011, "step": 1336 }, { "epoch": 0.36, "learning_rate": 1.4927633610599381e-05, "loss": 1.1016, "step": 1337 }, { "epoch": 0.36, "learning_rate": 1.492012780227865e-05, "loss": 1.128, "step": 1338 }, { "epoch": 0.36, "learning_rate": 1.4912618334997092e-05, "loss": 1.1217, "step": 1339 }, { "epoch": 0.36, "learning_rate": 1.4905105214339286e-05, "loss": 1.1327, "step": 1340 }, { "epoch": 0.36, "learning_rate": 1.489758844589253e-05, "loss": 1.1025, "step": 1341 }, { "epoch": 0.36, "learning_rate": 1.4890068035246836e-05, "loss": 1.1002, "step": 1342 }, { "epoch": 0.36, "learning_rate": 1.4882543987994918e-05, "loss": 1.1006, "step": 1343 }, { "epoch": 0.36, "learning_rate": 1.48750163097322e-05, "loss": 1.078, "step": 1344 }, { "epoch": 0.36, "learning_rate": 1.4867485006056806e-05, "loss": 1.1363, "step": 1345 }, { "epoch": 0.36, "learning_rate": 1.4859950082569554e-05, "loss": 1.1208, "step": 1346 }, { "epoch": 0.36, "learning_rate": 1.4852411544873954e-05, "loss": 1.0882, "step": 1347 }, { "epoch": 0.36, "learning_rate": 1.4844869398576208e-05, "loss": 1.1187, "step": 1348 }, { "epoch": 0.36, "learning_rate": 1.4837323649285195e-05, "loss": 1.1322, "step": 1349 }, { "epoch": 0.36, "learning_rate": 1.4829774302612477e-05, "loss": 1.1162, "step": 1350 }, { "epoch": 0.36, "learning_rate": 1.4822221364172293e-05, "loss": 1.1492, "step": 1351 }, { "epoch": 0.36, "learning_rate": 1.4814664839581551e-05, "loss": 1.1272, "step": 1352 }, { "epoch": 0.36, "learning_rate": 1.4807104734459823e-05, "loss": 1.1102, "step": 1353 }, { "epoch": 0.36, "learning_rate": 1.479954105442935e-05, "loss": 1.1368, "step": 1354 }, { "epoch": 0.36, "learning_rate": 1.4791973805115028e-05, "loss": 1.1638, "step": 1355 }, { "epoch": 0.36, "learning_rate": 1.4784402992144405e-05, "loss": 1.0838, "step": 1356 }, { "epoch": 0.36, "learning_rate": 1.4776828621147685e-05, "loss": 1.1166, "step": 1357 }, { "epoch": 0.36, "learning_rate": 1.4769250697757716e-05, "loss": 1.1192, "step": 1358 }, { "epoch": 0.36, "learning_rate": 1.476166922760998e-05, "loss": 1.1528, "step": 1359 }, { "epoch": 0.36, "learning_rate": 1.475408421634261e-05, "loss": 1.1255, "step": 1360 }, { "epoch": 0.36, "learning_rate": 1.4746495669596365e-05, "loss": 1.1322, "step": 1361 }, { "epoch": 0.36, "learning_rate": 1.4738903593014631e-05, "loss": 1.1282, "step": 1362 }, { "epoch": 0.36, "learning_rate": 1.4731307992243424e-05, "loss": 1.1251, "step": 1363 }, { "epoch": 0.36, "learning_rate": 1.4723708872931382e-05, "loss": 1.118, "step": 1364 }, { "epoch": 0.36, "learning_rate": 1.4716106240729748e-05, "loss": 1.1582, "step": 1365 }, { "epoch": 0.36, "learning_rate": 1.4708500101292394e-05, "loss": 1.1174, "step": 1366 }, { "epoch": 0.36, "learning_rate": 1.470089046027579e-05, "loss": 1.0768, "step": 1367 }, { "epoch": 0.36, "learning_rate": 1.4693277323339011e-05, "loss": 1.1309, "step": 1368 }, { "epoch": 0.36, "learning_rate": 1.4685660696143736e-05, "loss": 1.1093, "step": 1369 }, { "epoch": 0.36, "learning_rate": 1.4678040584354232e-05, "loss": 1.1427, "step": 1370 }, { "epoch": 0.37, "learning_rate": 1.4670416993637364e-05, "loss": 1.1309, "step": 1371 }, { "epoch": 0.37, "learning_rate": 1.4662789929662585e-05, "loss": 1.1231, "step": 1372 }, { "epoch": 0.37, "learning_rate": 1.4655159398101924e-05, "loss": 1.1171, "step": 1373 }, { "epoch": 0.37, "learning_rate": 1.4647525404629997e-05, "loss": 1.105, "step": 1374 }, { "epoch": 0.37, "learning_rate": 1.4639887954923989e-05, "loss": 1.1029, "step": 1375 }, { "epoch": 0.37, "learning_rate": 1.4632247054663653e-05, "loss": 1.103, "step": 1376 }, { "epoch": 0.37, "learning_rate": 1.4624602709531315e-05, "loss": 1.1375, "step": 1377 }, { "epoch": 0.37, "learning_rate": 1.4616954925211864e-05, "loss": 1.1379, "step": 1378 }, { "epoch": 0.37, "learning_rate": 1.4609303707392734e-05, "loss": 1.0883, "step": 1379 }, { "epoch": 0.37, "learning_rate": 1.4601649061763928e-05, "loss": 1.0846, "step": 1380 }, { "epoch": 0.37, "learning_rate": 1.4593990994017986e-05, "loss": 1.1728, "step": 1381 }, { "epoch": 0.37, "learning_rate": 1.458632950985e-05, "loss": 1.1184, "step": 1382 }, { "epoch": 0.37, "learning_rate": 1.45786646149576e-05, "loss": 1.1186, "step": 1383 }, { "epoch": 0.37, "learning_rate": 1.4570996315040954e-05, "loss": 1.155, "step": 1384 }, { "epoch": 0.37, "learning_rate": 1.456332461580276e-05, "loss": 1.1395, "step": 1385 }, { "epoch": 0.37, "learning_rate": 1.4555649522948247e-05, "loss": 1.1804, "step": 1386 }, { "epoch": 0.37, "learning_rate": 1.4547971042185166e-05, "loss": 1.1406, "step": 1387 }, { "epoch": 0.37, "learning_rate": 1.4540289179223781e-05, "loss": 1.1924, "step": 1388 }, { "epoch": 0.37, "learning_rate": 1.4532603939776885e-05, "loss": 1.1348, "step": 1389 }, { "epoch": 0.37, "learning_rate": 1.4524915329559774e-05, "loss": 1.1264, "step": 1390 }, { "epoch": 0.37, "learning_rate": 1.4517223354290248e-05, "loss": 1.1379, "step": 1391 }, { "epoch": 0.37, "learning_rate": 1.4509528019688617e-05, "loss": 1.1796, "step": 1392 }, { "epoch": 0.37, "learning_rate": 1.4501829331477682e-05, "loss": 1.1759, "step": 1393 }, { "epoch": 0.37, "learning_rate": 1.449412729538274e-05, "loss": 1.1412, "step": 1394 }, { "epoch": 0.37, "learning_rate": 1.4486421917131583e-05, "loss": 1.1009, "step": 1395 }, { "epoch": 0.37, "learning_rate": 1.4478713202454483e-05, "loss": 1.1801, "step": 1396 }, { "epoch": 0.37, "learning_rate": 1.4471001157084192e-05, "loss": 1.1183, "step": 1397 }, { "epoch": 0.37, "learning_rate": 1.4463285786755941e-05, "loss": 1.0929, "step": 1398 }, { "epoch": 0.37, "learning_rate": 1.4455567097207441e-05, "loss": 1.1259, "step": 1399 }, { "epoch": 0.37, "learning_rate": 1.4447845094178858e-05, "loss": 1.1309, "step": 1400 }, { "epoch": 0.37, "learning_rate": 1.4440119783412832e-05, "loss": 1.1046, "step": 1401 }, { "epoch": 0.37, "learning_rate": 1.4432391170654457e-05, "loss": 1.1237, "step": 1402 }, { "epoch": 0.37, "learning_rate": 1.4424659261651288e-05, "loss": 1.0906, "step": 1403 }, { "epoch": 0.37, "learning_rate": 1.4416924062153324e-05, "loss": 1.0848, "step": 1404 }, { "epoch": 0.37, "learning_rate": 1.4409185577913022e-05, "loss": 1.124, "step": 1405 }, { "epoch": 0.37, "learning_rate": 1.4401443814685272e-05, "loss": 1.0697, "step": 1406 }, { "epoch": 0.37, "learning_rate": 1.4393698778227402e-05, "loss": 1.1313, "step": 1407 }, { "epoch": 0.37, "learning_rate": 1.4385950474299186e-05, "loss": 1.1143, "step": 1408 }, { "epoch": 0.38, "learning_rate": 1.4378198908662816e-05, "loss": 1.1185, "step": 1409 }, { "epoch": 0.38, "learning_rate": 1.4370444087082911e-05, "loss": 1.0889, "step": 1410 }, { "epoch": 0.38, "learning_rate": 1.4362686015326516e-05, "loss": 1.0554, "step": 1411 }, { "epoch": 0.38, "learning_rate": 1.4354924699163093e-05, "loss": 1.0788, "step": 1412 }, { "epoch": 0.38, "learning_rate": 1.4347160144364512e-05, "loss": 1.0748, "step": 1413 }, { "epoch": 0.38, "learning_rate": 1.4339392356705056e-05, "loss": 1.1112, "step": 1414 }, { "epoch": 0.38, "learning_rate": 1.433162134196141e-05, "loss": 1.0618, "step": 1415 }, { "epoch": 0.38, "learning_rate": 1.4323847105912658e-05, "loss": 1.049, "step": 1416 }, { "epoch": 0.38, "learning_rate": 1.4316069654340286e-05, "loss": 1.1061, "step": 1417 }, { "epoch": 0.38, "learning_rate": 1.4308288993028158e-05, "loss": 1.105, "step": 1418 }, { "epoch": 0.38, "learning_rate": 1.430050512776254e-05, "loss": 1.1144, "step": 1419 }, { "epoch": 0.38, "learning_rate": 1.4292718064332071e-05, "loss": 1.1141, "step": 1420 }, { "epoch": 0.38, "learning_rate": 1.4284927808527771e-05, "loss": 1.126, "step": 1421 }, { "epoch": 0.38, "learning_rate": 1.4277134366143038e-05, "loss": 1.1308, "step": 1422 }, { "epoch": 0.38, "learning_rate": 1.4269337742973632e-05, "loss": 1.0776, "step": 1423 }, { "epoch": 0.38, "learning_rate": 1.4261537944817683e-05, "loss": 1.1473, "step": 1424 }, { "epoch": 0.38, "learning_rate": 1.4253734977475684e-05, "loss": 1.0824, "step": 1425 }, { "epoch": 0.38, "learning_rate": 1.424592884675048e-05, "loss": 1.1336, "step": 1426 }, { "epoch": 0.38, "learning_rate": 1.4238119558447274e-05, "loss": 1.0926, "step": 1427 }, { "epoch": 0.38, "learning_rate": 1.4230307118373614e-05, "loss": 1.0953, "step": 1428 }, { "epoch": 0.38, "learning_rate": 1.4222491532339389e-05, "loss": 1.1018, "step": 1429 }, { "epoch": 0.38, "learning_rate": 1.4214672806156833e-05, "loss": 1.0869, "step": 1430 }, { "epoch": 0.38, "learning_rate": 1.4206850945640516e-05, "loss": 1.1086, "step": 1431 }, { "epoch": 0.38, "learning_rate": 1.419902595660733e-05, "loss": 1.1426, "step": 1432 }, { "epoch": 0.38, "learning_rate": 1.4191197844876504e-05, "loss": 1.0987, "step": 1433 }, { "epoch": 0.38, "learning_rate": 1.418336661626958e-05, "loss": 1.0913, "step": 1434 }, { "epoch": 0.38, "learning_rate": 1.417553227661043e-05, "loss": 1.0862, "step": 1435 }, { "epoch": 0.38, "learning_rate": 1.4167694831725226e-05, "loss": 1.0702, "step": 1436 }, { "epoch": 0.38, "learning_rate": 1.4159854287442456e-05, "loss": 1.0836, "step": 1437 }, { "epoch": 0.38, "learning_rate": 1.4152010649592918e-05, "loss": 1.096, "step": 1438 }, { "epoch": 0.38, "learning_rate": 1.4144163924009697e-05, "loss": 1.0586, "step": 1439 }, { "epoch": 0.38, "learning_rate": 1.4136314116528192e-05, "loss": 1.046, "step": 1440 }, { "epoch": 0.38, "learning_rate": 1.4128461232986077e-05, "loss": 1.1146, "step": 1441 }, { "epoch": 0.38, "learning_rate": 1.412060527922332e-05, "loss": 1.0374, "step": 1442 }, { "epoch": 0.38, "learning_rate": 1.411274626108218e-05, "loss": 1.0637, "step": 1443 }, { "epoch": 0.38, "learning_rate": 1.4104884184407184e-05, "loss": 1.0991, "step": 1444 }, { "epoch": 0.38, "learning_rate": 1.4097019055045133e-05, "loss": 1.125, "step": 1445 }, { "epoch": 0.38, "learning_rate": 1.4089150878845113e-05, "loss": 1.0971, "step": 1446 }, { "epoch": 0.39, "learning_rate": 1.4081279661658458e-05, "loss": 1.0894, "step": 1447 }, { "epoch": 0.39, "learning_rate": 1.4073405409338772e-05, "loss": 1.066, "step": 1448 }, { "epoch": 0.39, "learning_rate": 1.4065528127741916e-05, "loss": 1.0831, "step": 1449 }, { "epoch": 0.39, "learning_rate": 1.4057647822726006e-05, "loss": 1.1224, "step": 1450 }, { "epoch": 0.39, "learning_rate": 1.4049764500151399e-05, "loss": 1.1278, "step": 1451 }, { "epoch": 0.39, "learning_rate": 1.4041878165880705e-05, "loss": 1.0533, "step": 1452 }, { "epoch": 0.39, "learning_rate": 1.4033988825778763e-05, "loss": 1.083, "step": 1453 }, { "epoch": 0.39, "learning_rate": 1.4026096485712656e-05, "loss": 1.1218, "step": 1454 }, { "epoch": 0.39, "learning_rate": 1.4018201151551703e-05, "loss": 1.066, "step": 1455 }, { "epoch": 0.39, "learning_rate": 1.4010302829167432e-05, "loss": 1.1173, "step": 1456 }, { "epoch": 0.39, "learning_rate": 1.4002401524433606e-05, "loss": 1.0629, "step": 1457 }, { "epoch": 0.39, "learning_rate": 1.3994497243226208e-05, "loss": 1.0743, "step": 1458 }, { "epoch": 0.39, "learning_rate": 1.3986589991423427e-05, "loss": 1.1159, "step": 1459 }, { "epoch": 0.39, "learning_rate": 1.397867977490566e-05, "loss": 1.1183, "step": 1460 }, { "epoch": 0.39, "learning_rate": 1.3970766599555518e-05, "loss": 1.0679, "step": 1461 }, { "epoch": 0.39, "learning_rate": 1.396285047125781e-05, "loss": 1.0927, "step": 1462 }, { "epoch": 0.39, "learning_rate": 1.3954931395899529e-05, "loss": 1.0735, "step": 1463 }, { "epoch": 0.39, "learning_rate": 1.3947009379369875e-05, "loss": 1.0698, "step": 1464 }, { "epoch": 0.39, "learning_rate": 1.393908442756023e-05, "loss": 1.116, "step": 1465 }, { "epoch": 0.39, "learning_rate": 1.3931156546364153e-05, "loss": 1.0721, "step": 1466 }, { "epoch": 0.39, "learning_rate": 1.3923225741677393e-05, "loss": 1.1, "step": 1467 }, { "epoch": 0.39, "learning_rate": 1.3915292019397861e-05, "loss": 1.0767, "step": 1468 }, { "epoch": 0.39, "learning_rate": 1.3907355385425645e-05, "loss": 1.1157, "step": 1469 }, { "epoch": 0.39, "learning_rate": 1.3899415845662997e-05, "loss": 1.0758, "step": 1470 }, { "epoch": 0.39, "learning_rate": 1.3891473406014334e-05, "loss": 1.0789, "step": 1471 }, { "epoch": 0.39, "learning_rate": 1.3883528072386216e-05, "loss": 1.1088, "step": 1472 }, { "epoch": 0.39, "learning_rate": 1.3875579850687373e-05, "loss": 1.1478, "step": 1473 }, { "epoch": 0.39, "learning_rate": 1.3867628746828673e-05, "loss": 1.0773, "step": 1474 }, { "epoch": 0.39, "learning_rate": 1.3859674766723119e-05, "loss": 1.0741, "step": 1475 }, { "epoch": 0.39, "learning_rate": 1.3851717916285876e-05, "loss": 1.1436, "step": 1476 }, { "epoch": 0.39, "learning_rate": 1.3843758201434221e-05, "loss": 1.1205, "step": 1477 }, { "epoch": 0.39, "learning_rate": 1.3835795628087577e-05, "loss": 1.1124, "step": 1478 }, { "epoch": 0.39, "learning_rate": 1.382783020216748e-05, "loss": 1.0749, "step": 1479 }, { "epoch": 0.39, "learning_rate": 1.3819861929597595e-05, "loss": 1.1235, "step": 1480 }, { "epoch": 0.39, "learning_rate": 1.3811890816303703e-05, "loss": 1.0883, "step": 1481 }, { "epoch": 0.39, "learning_rate": 1.3803916868213701e-05, "loss": 1.1316, "step": 1482 }, { "epoch": 0.39, "learning_rate": 1.3795940091257585e-05, "loss": 1.0973, "step": 1483 }, { "epoch": 0.4, "learning_rate": 1.3787960491367463e-05, "loss": 1.1179, "step": 1484 }, { "epoch": 0.4, "learning_rate": 1.3779978074477535e-05, "loss": 1.1416, "step": 1485 }, { "epoch": 0.4, "learning_rate": 1.3771992846524107e-05, "loss": 1.1022, "step": 1486 }, { "epoch": 0.4, "learning_rate": 1.3764004813445562e-05, "loss": 1.0899, "step": 1487 }, { "epoch": 0.4, "learning_rate": 1.3756013981182379e-05, "loss": 1.103, "step": 1488 }, { "epoch": 0.4, "learning_rate": 1.3748020355677116e-05, "loss": 1.0866, "step": 1489 }, { "epoch": 0.4, "learning_rate": 1.3740023942874405e-05, "loss": 1.1004, "step": 1490 }, { "epoch": 0.4, "learning_rate": 1.3732024748720956e-05, "loss": 1.0683, "step": 1491 }, { "epoch": 0.4, "learning_rate": 1.3724022779165545e-05, "loss": 1.1027, "step": 1492 }, { "epoch": 0.4, "learning_rate": 1.371601804015901e-05, "loss": 1.0801, "step": 1493 }, { "epoch": 0.4, "learning_rate": 1.3708010537654252e-05, "loss": 1.0584, "step": 1494 }, { "epoch": 0.4, "learning_rate": 1.3700000277606228e-05, "loss": 1.0562, "step": 1495 }, { "epoch": 0.4, "learning_rate": 1.3691987265971938e-05, "loss": 1.0513, "step": 1496 }, { "epoch": 0.4, "learning_rate": 1.3683971508710438e-05, "loss": 1.0929, "step": 1497 }, { "epoch": 0.4, "learning_rate": 1.367595301178282e-05, "loss": 1.1038, "step": 1498 }, { "epoch": 0.4, "learning_rate": 1.3667931781152214e-05, "loss": 1.0801, "step": 1499 }, { "epoch": 0.4, "learning_rate": 1.3659907822783787e-05, "loss": 1.0564, "step": 1500 }, { "epoch": 0.4, "learning_rate": 1.3651881142644728e-05, "loss": 1.0836, "step": 1501 }, { "epoch": 0.4, "learning_rate": 1.364385174670425e-05, "loss": 1.0988, "step": 1502 }, { "epoch": 0.4, "learning_rate": 1.3635819640933598e-05, "loss": 1.1283, "step": 1503 }, { "epoch": 0.4, "learning_rate": 1.3627784831306022e-05, "loss": 1.1157, "step": 1504 }, { "epoch": 0.4, "learning_rate": 1.3619747323796777e-05, "loss": 1.0946, "step": 1505 }, { "epoch": 0.4, "learning_rate": 1.3611707124383137e-05, "loss": 1.1168, "step": 1506 }, { "epoch": 0.4, "learning_rate": 1.3603664239044375e-05, "loss": 1.0756, "step": 1507 }, { "epoch": 0.4, "learning_rate": 1.359561867376175e-05, "loss": 1.1111, "step": 1508 }, { "epoch": 0.4, "learning_rate": 1.3587570434518532e-05, "loss": 1.1082, "step": 1509 }, { "epoch": 0.4, "learning_rate": 1.3579519527299966e-05, "loss": 1.0808, "step": 1510 }, { "epoch": 0.4, "learning_rate": 1.3571465958093285e-05, "loss": 1.1124, "step": 1511 }, { "epoch": 0.4, "learning_rate": 1.3563409732887705e-05, "loss": 1.1655, "step": 1512 }, { "epoch": 0.4, "learning_rate": 1.3555350857674409e-05, "loss": 1.0983, "step": 1513 }, { "epoch": 0.4, "learning_rate": 1.3547289338446562e-05, "loss": 1.1043, "step": 1514 }, { "epoch": 0.4, "learning_rate": 1.3539225181199291e-05, "loss": 1.0531, "step": 1515 }, { "epoch": 0.4, "learning_rate": 1.3531158391929675e-05, "loss": 1.0523, "step": 1516 }, { "epoch": 0.4, "learning_rate": 1.3523088976636764e-05, "loss": 1.1089, "step": 1517 }, { "epoch": 0.4, "learning_rate": 1.3515016941321555e-05, "loss": 1.0819, "step": 1518 }, { "epoch": 0.4, "learning_rate": 1.3506942291986991e-05, "loss": 1.068, "step": 1519 }, { "epoch": 0.4, "learning_rate": 1.3498865034637967e-05, "loss": 1.075, "step": 1520 }, { "epoch": 0.4, "learning_rate": 1.349078517528131e-05, "loss": 1.1119, "step": 1521 }, { "epoch": 0.41, "learning_rate": 1.3482702719925782e-05, "loss": 1.0514, "step": 1522 }, { "epoch": 0.41, "learning_rate": 1.347461767458208e-05, "loss": 1.0631, "step": 1523 }, { "epoch": 0.41, "learning_rate": 1.3466530045262826e-05, "loss": 1.0845, "step": 1524 }, { "epoch": 0.41, "learning_rate": 1.3458439837982563e-05, "loss": 1.0566, "step": 1525 }, { "epoch": 0.41, "learning_rate": 1.3450347058757746e-05, "loss": 1.071, "step": 1526 }, { "epoch": 0.41, "learning_rate": 1.3442251713606755e-05, "loss": 1.0761, "step": 1527 }, { "epoch": 0.41, "learning_rate": 1.3434153808549864e-05, "loss": 1.0544, "step": 1528 }, { "epoch": 0.41, "learning_rate": 1.3426053349609262e-05, "loss": 0.9846, "step": 1529 }, { "epoch": 0.41, "learning_rate": 1.3417950342809034e-05, "loss": 1.1068, "step": 1530 }, { "epoch": 0.41, "learning_rate": 1.340984479417515e-05, "loss": 1.0267, "step": 1531 }, { "epoch": 0.41, "learning_rate": 1.3401736709735489e-05, "loss": 1.1188, "step": 1532 }, { "epoch": 0.41, "learning_rate": 1.33936260955198e-05, "loss": 1.0727, "step": 1533 }, { "epoch": 0.41, "learning_rate": 1.3385512957559719e-05, "loss": 1.062, "step": 1534 }, { "epoch": 0.41, "learning_rate": 1.3377397301888761e-05, "loss": 1.032, "step": 1535 }, { "epoch": 0.41, "learning_rate": 1.336927913454231e-05, "loss": 1.0789, "step": 1536 }, { "epoch": 0.41, "learning_rate": 1.3361158461557618e-05, "loss": 1.041, "step": 1537 }, { "epoch": 0.41, "learning_rate": 1.3353035288973804e-05, "loss": 1.0836, "step": 1538 }, { "epoch": 0.41, "learning_rate": 1.3344909622831842e-05, "loss": 1.0696, "step": 1539 }, { "epoch": 0.41, "learning_rate": 1.3336781469174559e-05, "loss": 0.9871, "step": 1540 }, { "epoch": 0.41, "learning_rate": 1.3328650834046638e-05, "loss": 1.0792, "step": 1541 }, { "epoch": 0.41, "learning_rate": 1.3320517723494604e-05, "loss": 1.0079, "step": 1542 }, { "epoch": 0.41, "learning_rate": 1.3312382143566818e-05, "loss": 1.0692, "step": 1543 }, { "epoch": 0.41, "learning_rate": 1.3304244100313491e-05, "loss": 1.0453, "step": 1544 }, { "epoch": 0.41, "learning_rate": 1.329610359978665e-05, "loss": 1.0527, "step": 1545 }, { "epoch": 0.41, "learning_rate": 1.3287960648040155e-05, "loss": 1.0243, "step": 1546 }, { "epoch": 0.41, "learning_rate": 1.3279815251129698e-05, "loss": 1.063, "step": 1547 }, { "epoch": 0.41, "learning_rate": 1.3271667415112777e-05, "loss": 1.1255, "step": 1548 }, { "epoch": 0.41, "learning_rate": 1.3263517146048708e-05, "loss": 1.063, "step": 1549 }, { "epoch": 0.41, "learning_rate": 1.325536444999862e-05, "loss": 1.0588, "step": 1550 }, { "epoch": 0.41, "learning_rate": 1.3247209333025444e-05, "loss": 1.1248, "step": 1551 }, { "epoch": 0.41, "learning_rate": 1.3239051801193907e-05, "loss": 1.0956, "step": 1552 }, { "epoch": 0.41, "learning_rate": 1.3230891860570542e-05, "loss": 1.0898, "step": 1553 }, { "epoch": 0.41, "learning_rate": 1.3222729517223667e-05, "loss": 1.0834, "step": 1554 }, { "epoch": 0.41, "learning_rate": 1.321456477722338e-05, "loss": 1.0419, "step": 1555 }, { "epoch": 0.41, "learning_rate": 1.3206397646641577e-05, "loss": 1.0673, "step": 1556 }, { "epoch": 0.41, "learning_rate": 1.3198228131551923e-05, "loss": 1.1119, "step": 1557 }, { "epoch": 0.41, "learning_rate": 1.3190056238029852e-05, "loss": 1.0765, "step": 1558 }, { "epoch": 0.42, "learning_rate": 1.3181881972152575e-05, "loss": 1.133, "step": 1559 }, { "epoch": 0.42, "learning_rate": 1.3173705339999063e-05, "loss": 1.0735, "step": 1560 }, { "epoch": 0.42, "learning_rate": 1.3165526347650049e-05, "loss": 1.0587, "step": 1561 }, { "epoch": 0.42, "learning_rate": 1.3157345001188016e-05, "loss": 1.0547, "step": 1562 }, { "epoch": 0.42, "learning_rate": 1.3149161306697204e-05, "loss": 1.0405, "step": 1563 }, { "epoch": 0.42, "learning_rate": 1.3140975270263599e-05, "loss": 1.0653, "step": 1564 }, { "epoch": 0.42, "learning_rate": 1.3132786897974922e-05, "loss": 1.0794, "step": 1565 }, { "epoch": 0.42, "learning_rate": 1.3124596195920642e-05, "loss": 1.0819, "step": 1566 }, { "epoch": 0.42, "learning_rate": 1.3116403170191947e-05, "loss": 1.0596, "step": 1567 }, { "epoch": 0.42, "learning_rate": 1.3108207826881762e-05, "loss": 1.1033, "step": 1568 }, { "epoch": 0.42, "learning_rate": 1.3100010172084735e-05, "loss": 1.106, "step": 1569 }, { "epoch": 0.42, "learning_rate": 1.3091810211897231e-05, "loss": 1.0772, "step": 1570 }, { "epoch": 0.42, "learning_rate": 1.3083607952417333e-05, "loss": 1.0979, "step": 1571 }, { "epoch": 0.42, "learning_rate": 1.3075403399744824e-05, "loss": 1.0893, "step": 1572 }, { "epoch": 0.42, "learning_rate": 1.3067196559981203e-05, "loss": 1.0426, "step": 1573 }, { "epoch": 0.42, "learning_rate": 1.305898743922967e-05, "loss": 1.1177, "step": 1574 }, { "epoch": 0.42, "learning_rate": 1.305077604359511e-05, "loss": 1.0946, "step": 1575 }, { "epoch": 0.42, "learning_rate": 1.3042562379184111e-05, "loss": 1.0627, "step": 1576 }, { "epoch": 0.42, "learning_rate": 1.303434645210495e-05, "loss": 1.0753, "step": 1577 }, { "epoch": 0.42, "learning_rate": 1.3026128268467573e-05, "loss": 1.052, "step": 1578 }, { "epoch": 0.42, "learning_rate": 1.3017907834383612e-05, "loss": 1.0658, "step": 1579 }, { "epoch": 0.42, "learning_rate": 1.3009685155966378e-05, "loss": 1.0617, "step": 1580 }, { "epoch": 0.42, "learning_rate": 1.3001460239330845e-05, "loss": 1.0411, "step": 1581 }, { "epoch": 0.42, "learning_rate": 1.2993233090593648e-05, "loss": 1.0771, "step": 1582 }, { "epoch": 0.42, "learning_rate": 1.2985003715873094e-05, "loss": 1.0952, "step": 1583 }, { "epoch": 0.42, "learning_rate": 1.2976772121289134e-05, "loss": 1.0861, "step": 1584 }, { "epoch": 0.42, "learning_rate": 1.2968538312963368e-05, "loss": 1.0704, "step": 1585 }, { "epoch": 0.42, "learning_rate": 1.2960302297019056e-05, "loss": 1.0888, "step": 1586 }, { "epoch": 0.42, "learning_rate": 1.295206407958109e-05, "loss": 1.0857, "step": 1587 }, { "epoch": 0.42, "learning_rate": 1.2943823666775999e-05, "loss": 1.0671, "step": 1588 }, { "epoch": 0.42, "learning_rate": 1.2935581064731948e-05, "loss": 1.0773, "step": 1589 }, { "epoch": 0.42, "learning_rate": 1.2927336279578726e-05, "loss": 1.1296, "step": 1590 }, { "epoch": 0.42, "learning_rate": 1.291908931744775e-05, "loss": 1.1087, "step": 1591 }, { "epoch": 0.42, "learning_rate": 1.2910840184472056e-05, "loss": 1.0774, "step": 1592 }, { "epoch": 0.42, "learning_rate": 1.2902588886786288e-05, "loss": 1.1318, "step": 1593 }, { "epoch": 0.42, "learning_rate": 1.2894335430526708e-05, "loss": 1.0817, "step": 1594 }, { "epoch": 0.42, "learning_rate": 1.2886079821831177e-05, "loss": 1.1375, "step": 1595 }, { "epoch": 0.42, "learning_rate": 1.2877822066839162e-05, "loss": 1.0742, "step": 1596 }, { "epoch": 0.43, "learning_rate": 1.286956217169172e-05, "loss": 1.0824, "step": 1597 }, { "epoch": 0.43, "learning_rate": 1.2861300142531508e-05, "loss": 1.0743, "step": 1598 }, { "epoch": 0.43, "learning_rate": 1.2853035985502759e-05, "loss": 1.0958, "step": 1599 }, { "epoch": 0.43, "learning_rate": 1.2844769706751297e-05, "loss": 1.0998, "step": 1600 }, { "epoch": 0.43, "learning_rate": 1.2836501312424522e-05, "loss": 1.0497, "step": 1601 }, { "epoch": 0.43, "learning_rate": 1.2828230808671407e-05, "loss": 1.0962, "step": 1602 }, { "epoch": 0.43, "learning_rate": 1.2819958201642493e-05, "loss": 1.1429, "step": 1603 }, { "epoch": 0.43, "learning_rate": 1.2811683497489887e-05, "loss": 1.0807, "step": 1604 }, { "epoch": 0.43, "learning_rate": 1.2803406702367246e-05, "loss": 1.1334, "step": 1605 }, { "epoch": 0.43, "learning_rate": 1.27951278224298e-05, "loss": 1.0336, "step": 1606 }, { "epoch": 0.43, "learning_rate": 1.2786846863834317e-05, "loss": 1.0962, "step": 1607 }, { "epoch": 0.43, "learning_rate": 1.2778563832739106e-05, "loss": 1.0938, "step": 1608 }, { "epoch": 0.43, "learning_rate": 1.2770278735304033e-05, "loss": 1.0757, "step": 1609 }, { "epoch": 0.43, "learning_rate": 1.2761991577690491e-05, "loss": 1.1005, "step": 1610 }, { "epoch": 0.43, "learning_rate": 1.2753702366061402e-05, "loss": 1.1061, "step": 1611 }, { "epoch": 0.43, "learning_rate": 1.2745411106581224e-05, "loss": 1.0514, "step": 1612 }, { "epoch": 0.43, "learning_rate": 1.2737117805415934e-05, "loss": 1.1107, "step": 1613 }, { "epoch": 0.43, "learning_rate": 1.2728822468733021e-05, "loss": 1.1209, "step": 1614 }, { "epoch": 0.43, "learning_rate": 1.2720525102701503e-05, "loss": 1.1337, "step": 1615 }, { "epoch": 0.43, "learning_rate": 1.2712225713491893e-05, "loss": 1.1219, "step": 1616 }, { "epoch": 0.43, "learning_rate": 1.270392430727621e-05, "loss": 1.0937, "step": 1617 }, { "epoch": 0.43, "learning_rate": 1.2695620890227983e-05, "loss": 1.1448, "step": 1618 }, { "epoch": 0.43, "learning_rate": 1.2687315468522225e-05, "loss": 1.0776, "step": 1619 }, { "epoch": 0.43, "learning_rate": 1.2679008048335448e-05, "loss": 1.1389, "step": 1620 }, { "epoch": 0.43, "learning_rate": 1.2670698635845648e-05, "loss": 1.0936, "step": 1621 }, { "epoch": 0.43, "learning_rate": 1.2662387237232299e-05, "loss": 1.1249, "step": 1622 }, { "epoch": 0.43, "learning_rate": 1.2654073858676354e-05, "loss": 1.1092, "step": 1623 }, { "epoch": 0.43, "learning_rate": 1.2645758506360243e-05, "loss": 1.127, "step": 1624 }, { "epoch": 0.43, "learning_rate": 1.2637441186467858e-05, "loss": 1.1176, "step": 1625 }, { "epoch": 0.43, "learning_rate": 1.2629121905184554e-05, "loss": 1.0922, "step": 1626 }, { "epoch": 0.43, "learning_rate": 1.2620800668697148e-05, "loss": 1.0937, "step": 1627 }, { "epoch": 0.43, "learning_rate": 1.261247748319391e-05, "loss": 1.0906, "step": 1628 }, { "epoch": 0.43, "learning_rate": 1.260415235486456e-05, "loss": 1.0855, "step": 1629 }, { "epoch": 0.43, "learning_rate": 1.2595825289900263e-05, "loss": 1.0673, "step": 1630 }, { "epoch": 0.43, "learning_rate": 1.2587496294493622e-05, "loss": 1.1541, "step": 1631 }, { "epoch": 0.43, "learning_rate": 1.2579165374838673e-05, "loss": 1.1089, "step": 1632 }, { "epoch": 0.43, "learning_rate": 1.2570832537130894e-05, "loss": 1.114, "step": 1633 }, { "epoch": 0.44, "learning_rate": 1.2562497787567178e-05, "loss": 1.1358, "step": 1634 }, { "epoch": 0.44, "learning_rate": 1.2554161132345846e-05, "loss": 1.1128, "step": 1635 }, { "epoch": 0.44, "learning_rate": 1.2545822577666634e-05, "loss": 1.1283, "step": 1636 }, { "epoch": 0.44, "learning_rate": 1.253748212973069e-05, "loss": 1.0899, "step": 1637 }, { "epoch": 0.44, "learning_rate": 1.2529139794740565e-05, "loss": 1.1327, "step": 1638 }, { "epoch": 0.44, "learning_rate": 1.2520795578900233e-05, "loss": 1.1163, "step": 1639 }, { "epoch": 0.44, "learning_rate": 1.2512449488415042e-05, "loss": 1.1135, "step": 1640 }, { "epoch": 0.44, "learning_rate": 1.2504101529491748e-05, "loss": 1.0841, "step": 1641 }, { "epoch": 0.44, "learning_rate": 1.2495751708338494e-05, "loss": 1.1127, "step": 1642 }, { "epoch": 0.44, "learning_rate": 1.2487400031164807e-05, "loss": 1.1038, "step": 1643 }, { "epoch": 0.44, "learning_rate": 1.2479046504181594e-05, "loss": 1.072, "step": 1644 }, { "epoch": 0.44, "learning_rate": 1.2470691133601139e-05, "loss": 1.106, "step": 1645 }, { "epoch": 0.44, "learning_rate": 1.2462333925637098e-05, "loss": 1.0483, "step": 1646 }, { "epoch": 0.44, "learning_rate": 1.2453974886504488e-05, "loss": 1.1122, "step": 1647 }, { "epoch": 0.44, "learning_rate": 1.2445614022419697e-05, "loss": 1.1046, "step": 1648 }, { "epoch": 0.44, "learning_rate": 1.2437251339600462e-05, "loss": 1.0975, "step": 1649 }, { "epoch": 0.44, "learning_rate": 1.242888684426587e-05, "loss": 1.0274, "step": 1650 }, { "epoch": 0.44, "learning_rate": 1.2420520542636373e-05, "loss": 1.0799, "step": 1651 }, { "epoch": 0.44, "learning_rate": 1.2412152440933743e-05, "loss": 1.0992, "step": 1652 }, { "epoch": 0.44, "learning_rate": 1.2403782545381107e-05, "loss": 1.0626, "step": 1653 }, { "epoch": 0.44, "learning_rate": 1.2395410862202922e-05, "loss": 1.0426, "step": 1654 }, { "epoch": 0.44, "learning_rate": 1.2387037397624973e-05, "loss": 1.0825, "step": 1655 }, { "epoch": 0.44, "learning_rate": 1.2378662157874366e-05, "loss": 1.09, "step": 1656 }, { "epoch": 0.44, "learning_rate": 1.2370285149179539e-05, "loss": 1.0693, "step": 1657 }, { "epoch": 0.44, "learning_rate": 1.2361906377770236e-05, "loss": 1.0825, "step": 1658 }, { "epoch": 0.44, "learning_rate": 1.2353525849877507e-05, "loss": 1.1225, "step": 1659 }, { "epoch": 0.44, "learning_rate": 1.2345143571733727e-05, "loss": 1.0685, "step": 1660 }, { "epoch": 0.44, "learning_rate": 1.233675954957255e-05, "loss": 1.1176, "step": 1661 }, { "epoch": 0.44, "learning_rate": 1.2328373789628942e-05, "loss": 1.1147, "step": 1662 }, { "epoch": 0.44, "learning_rate": 1.2319986298139159e-05, "loss": 1.0987, "step": 1663 }, { "epoch": 0.44, "learning_rate": 1.2311597081340741e-05, "loss": 1.1329, "step": 1664 }, { "epoch": 0.44, "learning_rate": 1.230320614547251e-05, "loss": 1.1315, "step": 1665 }, { "epoch": 0.44, "learning_rate": 1.2294813496774576e-05, "loss": 1.0587, "step": 1666 }, { "epoch": 0.44, "learning_rate": 1.228641914148831e-05, "loss": 1.124, "step": 1667 }, { "epoch": 0.44, "learning_rate": 1.2278023085856359e-05, "loss": 1.085, "step": 1668 }, { "epoch": 0.44, "learning_rate": 1.2269625336122637e-05, "loss": 1.1236, "step": 1669 }, { "epoch": 0.44, "learning_rate": 1.2261225898532314e-05, "loss": 1.1231, "step": 1670 }, { "epoch": 0.44, "learning_rate": 1.225282477933181e-05, "loss": 1.0932, "step": 1671 }, { "epoch": 0.45, "learning_rate": 1.224442198476881e-05, "loss": 1.0946, "step": 1672 }, { "epoch": 0.45, "learning_rate": 1.2236017521092229e-05, "loss": 1.1278, "step": 1673 }, { "epoch": 0.45, "learning_rate": 1.2227611394552234e-05, "loss": 1.1301, "step": 1674 }, { "epoch": 0.45, "learning_rate": 1.2219203611400228e-05, "loss": 1.0831, "step": 1675 }, { "epoch": 0.45, "learning_rate": 1.2210794177888837e-05, "loss": 1.0409, "step": 1676 }, { "epoch": 0.45, "learning_rate": 1.2202383100271922e-05, "loss": 1.108, "step": 1677 }, { "epoch": 0.45, "learning_rate": 1.2193970384804568e-05, "loss": 1.0857, "step": 1678 }, { "epoch": 0.45, "learning_rate": 1.2185556037743074e-05, "loss": 1.0162, "step": 1679 }, { "epoch": 0.45, "learning_rate": 1.2177140065344953e-05, "loss": 1.0561, "step": 1680 }, { "epoch": 0.45, "learning_rate": 1.2168722473868926e-05, "loss": 1.0494, "step": 1681 }, { "epoch": 0.45, "learning_rate": 1.2160303269574923e-05, "loss": 1.0649, "step": 1682 }, { "epoch": 0.45, "learning_rate": 1.2151882458724063e-05, "loss": 1.0265, "step": 1683 }, { "epoch": 0.45, "learning_rate": 1.2143460047578675e-05, "loss": 1.0211, "step": 1684 }, { "epoch": 0.45, "learning_rate": 1.2135036042402264e-05, "loss": 1.0331, "step": 1685 }, { "epoch": 0.45, "learning_rate": 1.2126610449459527e-05, "loss": 1.0317, "step": 1686 }, { "epoch": 0.45, "learning_rate": 1.211818327501634e-05, "loss": 1.0073, "step": 1687 }, { "epoch": 0.45, "learning_rate": 1.210975452533976e-05, "loss": 1.0477, "step": 1688 }, { "epoch": 0.45, "learning_rate": 1.2101324206698006e-05, "loss": 1.0659, "step": 1689 }, { "epoch": 0.45, "learning_rate": 1.2092892325360472e-05, "loss": 1.0175, "step": 1690 }, { "epoch": 0.45, "learning_rate": 1.2084458887597712e-05, "loss": 1.0518, "step": 1691 }, { "epoch": 0.45, "learning_rate": 1.2076023899681433e-05, "loss": 1.0484, "step": 1692 }, { "epoch": 0.45, "learning_rate": 1.2067587367884504e-05, "loss": 1.0489, "step": 1693 }, { "epoch": 0.45, "learning_rate": 1.205914929848093e-05, "loss": 1.0551, "step": 1694 }, { "epoch": 0.45, "learning_rate": 1.2050709697745875e-05, "loss": 1.0508, "step": 1695 }, { "epoch": 0.45, "learning_rate": 1.2042268571955624e-05, "loss": 1.0407, "step": 1696 }, { "epoch": 0.45, "learning_rate": 1.203382592738761e-05, "loss": 1.0742, "step": 1697 }, { "epoch": 0.45, "learning_rate": 1.202538177032039e-05, "loss": 1.101, "step": 1698 }, { "epoch": 0.45, "learning_rate": 1.2016936107033648e-05, "loss": 1.0659, "step": 1699 }, { "epoch": 0.45, "learning_rate": 1.200848894380818e-05, "loss": 1.0341, "step": 1700 }, { "epoch": 0.45, "learning_rate": 1.2000040286925912e-05, "loss": 1.0888, "step": 1701 }, { "epoch": 0.45, "learning_rate": 1.1991590142669869e-05, "loss": 1.0724, "step": 1702 }, { "epoch": 0.45, "learning_rate": 1.1983138517324184e-05, "loss": 1.078, "step": 1703 }, { "epoch": 0.45, "learning_rate": 1.1974685417174097e-05, "loss": 1.02, "step": 1704 }, { "epoch": 0.45, "learning_rate": 1.196623084850594e-05, "loss": 1.0705, "step": 1705 }, { "epoch": 0.45, "learning_rate": 1.1957774817607138e-05, "loss": 1.0823, "step": 1706 }, { "epoch": 0.45, "learning_rate": 1.1949317330766204e-05, "loss": 1.053, "step": 1707 }, { "epoch": 0.45, "learning_rate": 1.1940858394272732e-05, "loss": 1.0805, "step": 1708 }, { "epoch": 0.46, "learning_rate": 1.1932398014417394e-05, "loss": 1.1236, "step": 1709 }, { "epoch": 0.46, "learning_rate": 1.1923936197491942e-05, "loss": 1.0898, "step": 1710 }, { "epoch": 0.46, "learning_rate": 1.1915472949789189e-05, "loss": 1.1121, "step": 1711 }, { "epoch": 0.46, "learning_rate": 1.1907008277603012e-05, "loss": 1.1367, "step": 1712 }, { "epoch": 0.46, "learning_rate": 1.1898542187228352e-05, "loss": 1.0627, "step": 1713 }, { "epoch": 0.46, "learning_rate": 1.1890074684961206e-05, "loss": 1.1114, "step": 1714 }, { "epoch": 0.46, "learning_rate": 1.188160577709861e-05, "loss": 1.0957, "step": 1715 }, { "epoch": 0.46, "learning_rate": 1.1873135469938659e-05, "loss": 1.0739, "step": 1716 }, { "epoch": 0.46, "learning_rate": 1.1864663769780482e-05, "loss": 1.0691, "step": 1717 }, { "epoch": 0.46, "learning_rate": 1.185619068292424e-05, "loss": 1.0737, "step": 1718 }, { "epoch": 0.46, "learning_rate": 1.1847716215671137e-05, "loss": 1.0679, "step": 1719 }, { "epoch": 0.46, "learning_rate": 1.1839240374323388e-05, "loss": 1.1032, "step": 1720 }, { "epoch": 0.46, "learning_rate": 1.1830763165184243e-05, "loss": 1.095, "step": 1721 }, { "epoch": 0.46, "learning_rate": 1.1822284594557965e-05, "loss": 1.0424, "step": 1722 }, { "epoch": 0.46, "learning_rate": 1.1813804668749826e-05, "loss": 1.101, "step": 1723 }, { "epoch": 0.46, "learning_rate": 1.1805323394066109e-05, "loss": 1.1153, "step": 1724 }, { "epoch": 0.46, "learning_rate": 1.17968407768141e-05, "loss": 1.1016, "step": 1725 }, { "epoch": 0.46, "learning_rate": 1.1788356823302084e-05, "loss": 1.1122, "step": 1726 }, { "epoch": 0.46, "learning_rate": 1.1779871539839338e-05, "loss": 1.0715, "step": 1727 }, { "epoch": 0.46, "learning_rate": 1.1771384932736129e-05, "loss": 1.0668, "step": 1728 }, { "epoch": 0.46, "learning_rate": 1.176289700830371e-05, "loss": 1.0792, "step": 1729 }, { "epoch": 0.46, "learning_rate": 1.1754407772854308e-05, "loss": 1.0853, "step": 1730 }, { "epoch": 0.46, "learning_rate": 1.1745917232701137e-05, "loss": 1.0996, "step": 1731 }, { "epoch": 0.46, "learning_rate": 1.1737425394158363e-05, "loss": 1.1001, "step": 1732 }, { "epoch": 0.46, "learning_rate": 1.1728932263541133e-05, "loss": 1.1326, "step": 1733 }, { "epoch": 0.46, "learning_rate": 1.172043784716555e-05, "loss": 1.0586, "step": 1734 }, { "epoch": 0.46, "learning_rate": 1.1711942151348674e-05, "loss": 1.0559, "step": 1735 }, { "epoch": 0.46, "learning_rate": 1.1703445182408508e-05, "loss": 1.0716, "step": 1736 }, { "epoch": 0.46, "learning_rate": 1.1694946946664018e-05, "loss": 1.11, "step": 1737 }, { "epoch": 0.46, "learning_rate": 1.1686447450435102e-05, "loss": 1.0963, "step": 1738 }, { "epoch": 0.46, "learning_rate": 1.1677946700042588e-05, "loss": 1.0487, "step": 1739 }, { "epoch": 0.46, "learning_rate": 1.1669444701808256e-05, "loss": 1.0437, "step": 1740 }, { "epoch": 0.46, "learning_rate": 1.1660941462054798e-05, "loss": 1.1086, "step": 1741 }, { "epoch": 0.46, "learning_rate": 1.1652436987105833e-05, "loss": 1.1051, "step": 1742 }, { "epoch": 0.46, "learning_rate": 1.1643931283285905e-05, "loss": 1.0522, "step": 1743 }, { "epoch": 0.46, "learning_rate": 1.1635424356920463e-05, "loss": 1.0849, "step": 1744 }, { "epoch": 0.46, "learning_rate": 1.1626916214335869e-05, "loss": 1.1127, "step": 1745 }, { "epoch": 0.46, "learning_rate": 1.1618406861859392e-05, "loss": 1.0605, "step": 1746 }, { "epoch": 0.47, "learning_rate": 1.1609896305819199e-05, "loss": 1.098, "step": 1747 }, { "epoch": 0.47, "learning_rate": 1.1601384552544345e-05, "loss": 1.1019, "step": 1748 }, { "epoch": 0.47, "learning_rate": 1.1592871608364793e-05, "loss": 1.1169, "step": 1749 }, { "epoch": 0.47, "learning_rate": 1.158435747961137e-05, "loss": 1.1095, "step": 1750 }, { "epoch": 0.47, "learning_rate": 1.15758421726158e-05, "loss": 1.063, "step": 1751 }, { "epoch": 0.47, "learning_rate": 1.1567325693710673e-05, "loss": 1.0766, "step": 1752 }, { "epoch": 0.47, "learning_rate": 1.1558808049229462e-05, "loss": 1.0722, "step": 1753 }, { "epoch": 0.47, "learning_rate": 1.1550289245506494e-05, "loss": 1.1274, "step": 1754 }, { "epoch": 0.47, "learning_rate": 1.1541769288876967e-05, "loss": 1.0678, "step": 1755 }, { "epoch": 0.47, "learning_rate": 1.1533248185676933e-05, "loss": 1.0502, "step": 1756 }, { "epoch": 0.47, "learning_rate": 1.1524725942243293e-05, "loss": 1.0872, "step": 1757 }, { "epoch": 0.47, "learning_rate": 1.1516202564913808e-05, "loss": 1.0619, "step": 1758 }, { "epoch": 0.47, "learning_rate": 1.1507678060027068e-05, "loss": 1.0508, "step": 1759 }, { "epoch": 0.47, "learning_rate": 1.149915243392251e-05, "loss": 1.0507, "step": 1760 }, { "epoch": 0.47, "learning_rate": 1.14906256929404e-05, "loss": 1.0293, "step": 1761 }, { "epoch": 0.47, "learning_rate": 1.1482097843421842e-05, "loss": 1.0975, "step": 1762 }, { "epoch": 0.47, "learning_rate": 1.147356889170875e-05, "loss": 1.118, "step": 1763 }, { "epoch": 0.47, "learning_rate": 1.1465038844143872e-05, "loss": 1.0504, "step": 1764 }, { "epoch": 0.47, "learning_rate": 1.1456507707070763e-05, "loss": 1.0522, "step": 1765 }, { "epoch": 0.47, "learning_rate": 1.1447975486833788e-05, "loss": 1.0342, "step": 1766 }, { "epoch": 0.47, "learning_rate": 1.1439442189778122e-05, "loss": 1.0244, "step": 1767 }, { "epoch": 0.47, "learning_rate": 1.1430907822249739e-05, "loss": 1.0812, "step": 1768 }, { "epoch": 0.47, "learning_rate": 1.1422372390595404e-05, "loss": 1.0531, "step": 1769 }, { "epoch": 0.47, "learning_rate": 1.1413835901162682e-05, "loss": 1.0662, "step": 1770 }, { "epoch": 0.47, "learning_rate": 1.140529836029992e-05, "loss": 1.0234, "step": 1771 }, { "epoch": 0.47, "learning_rate": 1.1396759774356243e-05, "loss": 1.0424, "step": 1772 }, { "epoch": 0.47, "learning_rate": 1.1388220149681564e-05, "loss": 1.0561, "step": 1773 }, { "epoch": 0.47, "learning_rate": 1.1379679492626556e-05, "loss": 1.0825, "step": 1774 }, { "epoch": 0.47, "learning_rate": 1.1371137809542669e-05, "loss": 1.0614, "step": 1775 }, { "epoch": 0.47, "learning_rate": 1.1362595106782113e-05, "loss": 1.0537, "step": 1776 }, { "epoch": 0.47, "learning_rate": 1.1354051390697856e-05, "loss": 1.052, "step": 1777 }, { "epoch": 0.47, "learning_rate": 1.1345506667643615e-05, "loss": 1.1247, "step": 1778 }, { "epoch": 0.47, "learning_rate": 1.1336960943973866e-05, "loss": 1.0321, "step": 1779 }, { "epoch": 0.47, "learning_rate": 1.1328414226043825e-05, "loss": 1.125, "step": 1780 }, { "epoch": 0.47, "learning_rate": 1.1319866520209439e-05, "loss": 1.0957, "step": 1781 }, { "epoch": 0.47, "learning_rate": 1.1311317832827405e-05, "loss": 1.0649, "step": 1782 }, { "epoch": 0.47, "learning_rate": 1.1302768170255132e-05, "loss": 1.1237, "step": 1783 }, { "epoch": 0.47, "learning_rate": 1.1294217538850771e-05, "loss": 1.1014, "step": 1784 }, { "epoch": 0.48, "learning_rate": 1.1285665944973187e-05, "loss": 1.1024, "step": 1785 }, { "epoch": 0.48, "learning_rate": 1.1277113394981953e-05, "loss": 1.0245, "step": 1786 }, { "epoch": 0.48, "learning_rate": 1.1268559895237369e-05, "loss": 1.0305, "step": 1787 }, { "epoch": 0.48, "learning_rate": 1.1260005452100428e-05, "loss": 1.0846, "step": 1788 }, { "epoch": 0.48, "learning_rate": 1.1251450071932829e-05, "loss": 1.0938, "step": 1789 }, { "epoch": 0.48, "learning_rate": 1.1242893761096967e-05, "loss": 1.0896, "step": 1790 }, { "epoch": 0.48, "learning_rate": 1.123433652595593e-05, "loss": 1.0543, "step": 1791 }, { "epoch": 0.48, "learning_rate": 1.1225778372873497e-05, "loss": 1.0585, "step": 1792 }, { "epoch": 0.48, "learning_rate": 1.1217219308214122e-05, "loss": 1.0902, "step": 1793 }, { "epoch": 0.48, "learning_rate": 1.1208659338342946e-05, "loss": 1.0709, "step": 1794 }, { "epoch": 0.48, "learning_rate": 1.120009846962577e-05, "loss": 1.0653, "step": 1795 }, { "epoch": 0.48, "learning_rate": 1.1191536708429078e-05, "loss": 1.0663, "step": 1796 }, { "epoch": 0.48, "learning_rate": 1.118297406112001e-05, "loss": 1.1207, "step": 1797 }, { "epoch": 0.48, "learning_rate": 1.1174410534066367e-05, "loss": 1.1123, "step": 1798 }, { "epoch": 0.48, "learning_rate": 1.11658461336366e-05, "loss": 1.1347, "step": 1799 }, { "epoch": 0.48, "learning_rate": 1.1157280866199818e-05, "loss": 1.1302, "step": 1800 }, { "epoch": 0.48, "learning_rate": 1.1148714738125763e-05, "loss": 1.1127, "step": 1801 }, { "epoch": 0.48, "learning_rate": 1.1140147755784831e-05, "loss": 1.055, "step": 1802 }, { "epoch": 0.48, "learning_rate": 1.1131579925548044e-05, "loss": 1.0898, "step": 1803 }, { "epoch": 0.48, "learning_rate": 1.1123011253787051e-05, "loss": 1.0768, "step": 1804 }, { "epoch": 0.48, "learning_rate": 1.111444174687414e-05, "loss": 1.1167, "step": 1805 }, { "epoch": 0.48, "learning_rate": 1.1105871411182209e-05, "loss": 1.0655, "step": 1806 }, { "epoch": 0.48, "learning_rate": 1.1097300253084777e-05, "loss": 1.0877, "step": 1807 }, { "epoch": 0.48, "learning_rate": 1.1088728278955974e-05, "loss": 1.0755, "step": 1808 }, { "epoch": 0.48, "learning_rate": 1.1080155495170534e-05, "loss": 1.1365, "step": 1809 }, { "epoch": 0.48, "learning_rate": 1.1071581908103796e-05, "loss": 1.0502, "step": 1810 }, { "epoch": 0.48, "learning_rate": 1.1063007524131697e-05, "loss": 1.0902, "step": 1811 }, { "epoch": 0.48, "learning_rate": 1.1054432349630767e-05, "loss": 1.0507, "step": 1812 }, { "epoch": 0.48, "learning_rate": 1.1045856390978118e-05, "loss": 1.0629, "step": 1813 }, { "epoch": 0.48, "learning_rate": 1.1037279654551452e-05, "loss": 1.0354, "step": 1814 }, { "epoch": 0.48, "learning_rate": 1.102870214672905e-05, "loss": 1.0539, "step": 1815 }, { "epoch": 0.48, "learning_rate": 1.102012387388976e-05, "loss": 1.0981, "step": 1816 }, { "epoch": 0.48, "learning_rate": 1.1011544842413e-05, "loss": 1.0502, "step": 1817 }, { "epoch": 0.48, "learning_rate": 1.100296505867876e-05, "loss": 1.0196, "step": 1818 }, { "epoch": 0.48, "learning_rate": 1.099438452906758e-05, "loss": 1.134, "step": 1819 }, { "epoch": 0.48, "learning_rate": 1.0985803259960563e-05, "loss": 1.0423, "step": 1820 }, { "epoch": 0.48, "learning_rate": 1.0977221257739357e-05, "loss": 1.0709, "step": 1821 }, { "epoch": 0.49, "learning_rate": 1.0968638528786151e-05, "loss": 1.0871, "step": 1822 }, { "epoch": 0.49, "learning_rate": 1.0960055079483685e-05, "loss": 1.0865, "step": 1823 }, { "epoch": 0.49, "learning_rate": 1.0951470916215225e-05, "loss": 1.0886, "step": 1824 }, { "epoch": 0.49, "learning_rate": 1.0942886045364576e-05, "loss": 1.0773, "step": 1825 }, { "epoch": 0.49, "learning_rate": 1.0934300473316061e-05, "loss": 1.0031, "step": 1826 }, { "epoch": 0.49, "learning_rate": 1.0925714206454534e-05, "loss": 1.061, "step": 1827 }, { "epoch": 0.49, "learning_rate": 1.0917127251165355e-05, "loss": 1.0497, "step": 1828 }, { "epoch": 0.49, "learning_rate": 1.0908539613834408e-05, "loss": 1.0735, "step": 1829 }, { "epoch": 0.49, "learning_rate": 1.0899951300848071e-05, "loss": 1.0489, "step": 1830 }, { "epoch": 0.49, "learning_rate": 1.0891362318593233e-05, "loss": 1.0556, "step": 1831 }, { "epoch": 0.49, "learning_rate": 1.088277267345728e-05, "loss": 1.0528, "step": 1832 }, { "epoch": 0.49, "learning_rate": 1.087418237182809e-05, "loss": 1.0742, "step": 1833 }, { "epoch": 0.49, "learning_rate": 1.0865591420094025e-05, "loss": 1.0682, "step": 1834 }, { "epoch": 0.49, "learning_rate": 1.085699982464394e-05, "loss": 1.0309, "step": 1835 }, { "epoch": 0.49, "learning_rate": 1.0848407591867166e-05, "loss": 1.0911, "step": 1836 }, { "epoch": 0.49, "learning_rate": 1.0839814728153493e-05, "loss": 1.1129, "step": 1837 }, { "epoch": 0.49, "learning_rate": 1.0831221239893202e-05, "loss": 1.0534, "step": 1838 }, { "epoch": 0.49, "learning_rate": 1.0822627133477027e-05, "loss": 1.0788, "step": 1839 }, { "epoch": 0.49, "learning_rate": 1.081403241529616e-05, "loss": 1.0208, "step": 1840 }, { "epoch": 0.49, "learning_rate": 1.0805437091742257e-05, "loss": 1.0927, "step": 1841 }, { "epoch": 0.49, "learning_rate": 1.0796841169207415e-05, "loss": 1.0493, "step": 1842 }, { "epoch": 0.49, "learning_rate": 1.078824465408418e-05, "loss": 1.0922, "step": 1843 }, { "epoch": 0.49, "learning_rate": 1.077964755276554e-05, "loss": 1.0957, "step": 1844 }, { "epoch": 0.49, "learning_rate": 1.0771049871644919e-05, "loss": 1.0688, "step": 1845 }, { "epoch": 0.49, "learning_rate": 1.0762451617116167e-05, "loss": 1.0736, "step": 1846 }, { "epoch": 0.49, "learning_rate": 1.0753852795573569e-05, "loss": 1.0869, "step": 1847 }, { "epoch": 0.49, "learning_rate": 1.0745253413411825e-05, "loss": 1.0599, "step": 1848 }, { "epoch": 0.49, "learning_rate": 1.0736653477026049e-05, "loss": 1.0435, "step": 1849 }, { "epoch": 0.49, "learning_rate": 1.0728052992811782e-05, "loss": 1.1396, "step": 1850 }, { "epoch": 0.49, "learning_rate": 1.0719451967164956e-05, "loss": 1.0694, "step": 1851 }, { "epoch": 0.49, "learning_rate": 1.0710850406481912e-05, "loss": 1.0499, "step": 1852 }, { "epoch": 0.49, "learning_rate": 1.0702248317159392e-05, "loss": 1.0949, "step": 1853 }, { "epoch": 0.49, "learning_rate": 1.0693645705594527e-05, "loss": 1.0734, "step": 1854 }, { "epoch": 0.49, "learning_rate": 1.0685042578184834e-05, "loss": 1.1053, "step": 1855 }, { "epoch": 0.49, "learning_rate": 1.0676438941328225e-05, "loss": 1.1048, "step": 1856 }, { "epoch": 0.49, "learning_rate": 1.0667834801422975e-05, "loss": 1.0456, "step": 1857 }, { "epoch": 0.49, "learning_rate": 1.0659230164867745e-05, "loss": 1.0911, "step": 1858 }, { "epoch": 0.49, "learning_rate": 1.0650625038061562e-05, "loss": 1.0891, "step": 1859 }, { "epoch": 0.5, "learning_rate": 1.0642019427403816e-05, "loss": 1.1335, "step": 1860 }, { "epoch": 0.5, "learning_rate": 1.0633413339294256e-05, "loss": 1.0638, "step": 1861 }, { "epoch": 0.5, "learning_rate": 1.0624806780132993e-05, "loss": 1.0477, "step": 1862 }, { "epoch": 0.5, "learning_rate": 1.0616199756320478e-05, "loss": 1.0894, "step": 1863 }, { "epoch": 0.5, "learning_rate": 1.0607592274257515e-05, "loss": 1.0944, "step": 1864 }, { "epoch": 0.5, "learning_rate": 1.0598984340345246e-05, "loss": 1.1447, "step": 1865 }, { "epoch": 0.5, "learning_rate": 1.0590375960985153e-05, "loss": 1.0514, "step": 1866 }, { "epoch": 0.5, "learning_rate": 1.058176714257904e-05, "loss": 1.0652, "step": 1867 }, { "epoch": 0.5, "learning_rate": 1.057315789152905e-05, "loss": 1.0651, "step": 1868 }, { "epoch": 0.5, "learning_rate": 1.0564548214237636e-05, "loss": 1.1083, "step": 1869 }, { "epoch": 0.5, "learning_rate": 1.055593811710757e-05, "loss": 1.0767, "step": 1870 }, { "epoch": 0.5, "learning_rate": 1.054732760654195e-05, "loss": 1.0807, "step": 1871 }, { "epoch": 0.5, "learning_rate": 1.0538716688944165e-05, "loss": 1.0766, "step": 1872 }, { "epoch": 0.5, "learning_rate": 1.053010537071791e-05, "loss": 1.0761, "step": 1873 }, { "epoch": 0.5, "learning_rate": 1.0521493658267186e-05, "loss": 1.0873, "step": 1874 }, { "epoch": 0.5, "learning_rate": 1.0512881557996273e-05, "loss": 1.0964, "step": 1875 }, { "epoch": 0.5, "learning_rate": 1.050426907630976e-05, "loss": 1.0949, "step": 1876 }, { "epoch": 0.5, "learning_rate": 1.0495656219612497e-05, "loss": 1.0763, "step": 1877 }, { "epoch": 0.5, "learning_rate": 1.0487042994309628e-05, "loss": 1.0668, "step": 1878 }, { "epoch": 0.5, "learning_rate": 1.0478429406806569e-05, "loss": 1.071, "step": 1879 }, { "epoch": 0.5, "learning_rate": 1.0469815463508999e-05, "loss": 1.0543, "step": 1880 }, { "epoch": 0.5, "learning_rate": 1.0461201170822868e-05, "loss": 1.0632, "step": 1881 }, { "epoch": 0.5, "learning_rate": 1.0452586535154382e-05, "loss": 1.0781, "step": 1882 }, { "epoch": 0.5, "learning_rate": 1.0443971562910005e-05, "loss": 1.0797, "step": 1883 }, { "epoch": 0.5, "learning_rate": 1.0435356260496448e-05, "loss": 1.0732, "step": 1884 }, { "epoch": 0.5, "learning_rate": 1.0426740634320671e-05, "loss": 1.0922, "step": 1885 }, { "epoch": 0.5, "learning_rate": 1.0418124690789873e-05, "loss": 1.073, "step": 1886 }, { "epoch": 0.5, "learning_rate": 1.0409508436311487e-05, "loss": 1.06, "step": 1887 }, { "epoch": 0.5, "learning_rate": 1.0400891877293182e-05, "loss": 1.0633, "step": 1888 }, { "epoch": 0.5, "learning_rate": 1.0392275020142852e-05, "loss": 1.0814, "step": 1889 }, { "epoch": 0.5, "learning_rate": 1.0383657871268607e-05, "loss": 1.061, "step": 1890 }, { "epoch": 0.5, "learning_rate": 1.0375040437078783e-05, "loss": 1.0313, "step": 1891 }, { "epoch": 0.5, "learning_rate": 1.0366422723981921e-05, "loss": 1.0968, "step": 1892 }, { "epoch": 0.5, "learning_rate": 1.0357804738386772e-05, "loss": 1.1357, "step": 1893 }, { "epoch": 0.5, "learning_rate": 1.034918648670229e-05, "loss": 1.0572, "step": 1894 }, { "epoch": 0.5, "learning_rate": 1.034056797533763e-05, "loss": 1.0874, "step": 1895 }, { "epoch": 0.5, "learning_rate": 1.0331949210702135e-05, "loss": 1.1177, "step": 1896 }, { "epoch": 0.51, "learning_rate": 1.0323330199205333e-05, "loss": 1.1072, "step": 1897 }, { "epoch": 0.51, "learning_rate": 1.0314710947256949e-05, "loss": 1.1037, "step": 1898 }, { "epoch": 0.51, "learning_rate": 1.0306091461266874e-05, "loss": 1.1058, "step": 1899 }, { "epoch": 0.51, "learning_rate": 1.0297471747645176e-05, "loss": 1.0681, "step": 1900 }, { "epoch": 0.51, "learning_rate": 1.0288851812802098e-05, "loss": 1.1173, "step": 1901 }, { "epoch": 0.51, "learning_rate": 1.0280231663148044e-05, "loss": 1.1872, "step": 1902 }, { "epoch": 0.51, "learning_rate": 1.027161130509357e-05, "loss": 1.073, "step": 1903 }, { "epoch": 0.51, "learning_rate": 1.0262990745049402e-05, "loss": 1.1469, "step": 1904 }, { "epoch": 0.51, "learning_rate": 1.0254369989426406e-05, "loss": 1.139, "step": 1905 }, { "epoch": 0.51, "learning_rate": 1.0245749044635589e-05, "loss": 1.1144, "step": 1906 }, { "epoch": 0.51, "learning_rate": 1.0237127917088114e-05, "loss": 1.0813, "step": 1907 }, { "epoch": 0.51, "learning_rate": 1.022850661319527e-05, "loss": 1.0947, "step": 1908 }, { "epoch": 0.51, "learning_rate": 1.0219885139368478e-05, "loss": 1.0773, "step": 1909 }, { "epoch": 0.51, "learning_rate": 1.0211263502019284e-05, "loss": 1.0847, "step": 1910 }, { "epoch": 0.51, "learning_rate": 1.0202641707559355e-05, "loss": 1.1096, "step": 1911 }, { "epoch": 0.51, "learning_rate": 1.0194019762400486e-05, "loss": 1.0547, "step": 1912 }, { "epoch": 0.51, "learning_rate": 1.0185397672954569e-05, "loss": 1.047, "step": 1913 }, { "epoch": 0.51, "learning_rate": 1.0176775445633609e-05, "loss": 1.0403, "step": 1914 }, { "epoch": 0.51, "learning_rate": 1.0168153086849717e-05, "loss": 1.0622, "step": 1915 }, { "epoch": 0.51, "learning_rate": 1.01595306030151e-05, "loss": 1.0793, "step": 1916 }, { "epoch": 0.51, "learning_rate": 1.0150908000542052e-05, "loss": 1.0624, "step": 1917 }, { "epoch": 0.51, "learning_rate": 1.0142285285842967e-05, "loss": 1.0677, "step": 1918 }, { "epoch": 0.51, "learning_rate": 1.0133662465330311e-05, "loss": 1.1156, "step": 1919 }, { "epoch": 0.51, "learning_rate": 1.0125039545416633e-05, "loss": 1.1136, "step": 1920 }, { "epoch": 0.51, "learning_rate": 1.011641653251456e-05, "loss": 1.0812, "step": 1921 }, { "epoch": 0.51, "learning_rate": 1.0107793433036781e-05, "loss": 1.1084, "step": 1922 }, { "epoch": 0.51, "learning_rate": 1.0099170253396052e-05, "loss": 1.0958, "step": 1923 }, { "epoch": 0.51, "learning_rate": 1.0090547000005194e-05, "loss": 1.0782, "step": 1924 }, { "epoch": 0.51, "learning_rate": 1.0081923679277076e-05, "loss": 1.0812, "step": 1925 }, { "epoch": 0.51, "learning_rate": 1.0073300297624617e-05, "loss": 1.0455, "step": 1926 }, { "epoch": 0.51, "learning_rate": 1.0064676861460788e-05, "loss": 1.0794, "step": 1927 }, { "epoch": 0.51, "learning_rate": 1.0056053377198593e-05, "loss": 1.1186, "step": 1928 }, { "epoch": 0.51, "learning_rate": 1.0047429851251076e-05, "loss": 1.0465, "step": 1929 }, { "epoch": 0.51, "learning_rate": 1.0038806290031314e-05, "loss": 1.1195, "step": 1930 }, { "epoch": 0.51, "learning_rate": 1.0030182699952401e-05, "loss": 1.0836, "step": 1931 }, { "epoch": 0.51, "learning_rate": 1.0021559087427461e-05, "loss": 1.1074, "step": 1932 }, { "epoch": 0.51, "learning_rate": 1.0012935458869638e-05, "loss": 1.0677, "step": 1933 }, { "epoch": 0.51, "learning_rate": 1.0004311820692074e-05, "loss": 1.0688, "step": 1934 }, { "epoch": 0.52, "learning_rate": 9.995688179307929e-06, "loss": 1.0705, "step": 1935 }, { "epoch": 0.52, "learning_rate": 9.987064541130367e-06, "loss": 1.1032, "step": 1936 }, { "epoch": 0.52, "learning_rate": 9.978440912572542e-06, "loss": 1.0972, "step": 1937 }, { "epoch": 0.52, "learning_rate": 9.969817300047602e-06, "loss": 1.1161, "step": 1938 }, { "epoch": 0.52, "learning_rate": 9.961193709968691e-06, "loss": 1.0847, "step": 1939 }, { "epoch": 0.52, "learning_rate": 9.952570148748926e-06, "loss": 1.127, "step": 1940 }, { "epoch": 0.52, "learning_rate": 9.943946622801409e-06, "loss": 1.1149, "step": 1941 }, { "epoch": 0.52, "learning_rate": 9.935323138539214e-06, "loss": 1.0588, "step": 1942 }, { "epoch": 0.52, "learning_rate": 9.926699702375387e-06, "loss": 1.0607, "step": 1943 }, { "epoch": 0.52, "learning_rate": 9.918076320722927e-06, "loss": 1.0974, "step": 1944 }, { "epoch": 0.52, "learning_rate": 9.909452999994808e-06, "loss": 1.1091, "step": 1945 }, { "epoch": 0.52, "learning_rate": 9.900829746603952e-06, "loss": 1.0964, "step": 1946 }, { "epoch": 0.52, "learning_rate": 9.892206566963222e-06, "loss": 1.0765, "step": 1947 }, { "epoch": 0.52, "learning_rate": 9.883583467485446e-06, "loss": 1.0804, "step": 1948 }, { "epoch": 0.52, "learning_rate": 9.87496045458337e-06, "loss": 1.0971, "step": 1949 }, { "epoch": 0.52, "learning_rate": 9.866337534669692e-06, "loss": 1.0248, "step": 1950 }, { "epoch": 0.52, "learning_rate": 9.857714714157036e-06, "loss": 1.0787, "step": 1951 }, { "epoch": 0.52, "learning_rate": 9.84909199945795e-06, "loss": 1.1175, "step": 1952 }, { "epoch": 0.52, "learning_rate": 9.840469396984903e-06, "loss": 1.061, "step": 1953 }, { "epoch": 0.52, "learning_rate": 9.831846913150285e-06, "loss": 1.0316, "step": 1954 }, { "epoch": 0.52, "learning_rate": 9.823224554366394e-06, "loss": 1.0629, "step": 1955 }, { "epoch": 0.52, "learning_rate": 9.814602327045435e-06, "loss": 1.0906, "step": 1956 }, { "epoch": 0.52, "learning_rate": 9.805980237599517e-06, "loss": 1.093, "step": 1957 }, { "epoch": 0.52, "learning_rate": 9.797358292440648e-06, "loss": 1.0803, "step": 1958 }, { "epoch": 0.52, "learning_rate": 9.788736497980719e-06, "loss": 1.0128, "step": 1959 }, { "epoch": 0.52, "learning_rate": 9.780114860631527e-06, "loss": 1.1114, "step": 1960 }, { "epoch": 0.52, "learning_rate": 9.771493386804733e-06, "loss": 1.0896, "step": 1961 }, { "epoch": 0.52, "learning_rate": 9.762872082911887e-06, "loss": 1.0763, "step": 1962 }, { "epoch": 0.52, "learning_rate": 9.754250955364413e-06, "loss": 1.0455, "step": 1963 }, { "epoch": 0.52, "learning_rate": 9.7456300105736e-06, "loss": 1.1063, "step": 1964 }, { "epoch": 0.52, "learning_rate": 9.737009254950601e-06, "loss": 1.0905, "step": 1965 }, { "epoch": 0.52, "learning_rate": 9.728388694906433e-06, "loss": 1.0615, "step": 1966 }, { "epoch": 0.52, "learning_rate": 9.71976833685196e-06, "loss": 1.0667, "step": 1967 }, { "epoch": 0.52, "learning_rate": 9.711148187197904e-06, "loss": 1.0869, "step": 1968 }, { "epoch": 0.52, "learning_rate": 9.702528252354826e-06, "loss": 1.1011, "step": 1969 }, { "epoch": 0.52, "learning_rate": 9.693908538733128e-06, "loss": 1.0642, "step": 1970 }, { "epoch": 0.52, "learning_rate": 9.685289052743055e-06, "loss": 1.1197, "step": 1971 }, { "epoch": 0.53, "learning_rate": 9.67666980079467e-06, "loss": 1.131, "step": 1972 }, { "epoch": 0.53, "learning_rate": 9.66805078929787e-06, "loss": 1.0779, "step": 1973 }, { "epoch": 0.53, "learning_rate": 9.659432024662372e-06, "loss": 1.0747, "step": 1974 }, { "epoch": 0.53, "learning_rate": 9.650813513297714e-06, "loss": 1.0401, "step": 1975 }, { "epoch": 0.53, "learning_rate": 9.642195261613233e-06, "loss": 1.0685, "step": 1976 }, { "epoch": 0.53, "learning_rate": 9.633577276018084e-06, "loss": 1.0539, "step": 1977 }, { "epoch": 0.53, "learning_rate": 9.624959562921224e-06, "loss": 1.0868, "step": 1978 }, { "epoch": 0.53, "learning_rate": 9.616342128731398e-06, "loss": 1.1219, "step": 1979 }, { "epoch": 0.53, "learning_rate": 9.607724979857153e-06, "loss": 1.1016, "step": 1980 }, { "epoch": 0.53, "learning_rate": 9.599108122706823e-06, "loss": 1.0669, "step": 1981 }, { "epoch": 0.53, "learning_rate": 9.590491563688517e-06, "loss": 1.0794, "step": 1982 }, { "epoch": 0.53, "learning_rate": 9.581875309210132e-06, "loss": 1.0752, "step": 1983 }, { "epoch": 0.53, "learning_rate": 9.573259365679334e-06, "loss": 1.1067, "step": 1984 }, { "epoch": 0.53, "learning_rate": 9.564643739503557e-06, "loss": 1.0505, "step": 1985 }, { "epoch": 0.53, "learning_rate": 9.55602843709e-06, "loss": 1.0805, "step": 1986 }, { "epoch": 0.53, "learning_rate": 9.547413464845623e-06, "loss": 1.1495, "step": 1987 }, { "epoch": 0.53, "learning_rate": 9.538798829177136e-06, "loss": 1.071, "step": 1988 }, { "epoch": 0.53, "learning_rate": 9.530184536491004e-06, "loss": 1.0926, "step": 1989 }, { "epoch": 0.53, "learning_rate": 9.521570593193436e-06, "loss": 1.115, "step": 1990 }, { "epoch": 0.53, "learning_rate": 9.512957005690373e-06, "loss": 1.0951, "step": 1991 }, { "epoch": 0.53, "learning_rate": 9.504343780387506e-06, "loss": 1.0882, "step": 1992 }, { "epoch": 0.53, "learning_rate": 9.495730923690245e-06, "loss": 1.0432, "step": 1993 }, { "epoch": 0.53, "learning_rate": 9.487118442003729e-06, "loss": 1.0881, "step": 1994 }, { "epoch": 0.53, "learning_rate": 9.47850634173282e-06, "loss": 1.1016, "step": 1995 }, { "epoch": 0.53, "learning_rate": 9.469894629282095e-06, "loss": 1.0796, "step": 1996 }, { "epoch": 0.53, "learning_rate": 9.46128331105584e-06, "loss": 1.0496, "step": 1997 }, { "epoch": 0.53, "learning_rate": 9.452672393458055e-06, "loss": 1.0725, "step": 1998 }, { "epoch": 0.53, "learning_rate": 9.444061882892434e-06, "loss": 1.0659, "step": 1999 }, { "epoch": 0.53, "learning_rate": 9.43545178576237e-06, "loss": 1.0857, "step": 2000 }, { "epoch": 0.53, "learning_rate": 9.426842108470957e-06, "loss": 1.1067, "step": 2001 }, { "epoch": 0.53, "learning_rate": 9.418232857420966e-06, "loss": 1.0727, "step": 2002 }, { "epoch": 0.53, "learning_rate": 9.409624039014852e-06, "loss": 1.0802, "step": 2003 }, { "epoch": 0.53, "learning_rate": 9.401015659654758e-06, "loss": 1.1034, "step": 2004 }, { "epoch": 0.53, "learning_rate": 9.392407725742492e-06, "loss": 1.0834, "step": 2005 }, { "epoch": 0.53, "learning_rate": 9.383800243679527e-06, "loss": 1.0625, "step": 2006 }, { "epoch": 0.53, "learning_rate": 9.375193219867014e-06, "loss": 1.0907, "step": 2007 }, { "epoch": 0.53, "learning_rate": 9.36658666070575e-06, "loss": 1.1555, "step": 2008 }, { "epoch": 0.53, "learning_rate": 9.35798057259619e-06, "loss": 1.1002, "step": 2009 }, { "epoch": 0.54, "learning_rate": 9.349374961938445e-06, "loss": 1.0993, "step": 2010 }, { "epoch": 0.54, "learning_rate": 9.34076983513226e-06, "loss": 1.0412, "step": 2011 }, { "epoch": 0.54, "learning_rate": 9.33216519857703e-06, "loss": 1.1255, "step": 2012 }, { "epoch": 0.54, "learning_rate": 9.323561058671782e-06, "loss": 1.0668, "step": 2013 }, { "epoch": 0.54, "learning_rate": 9.314957421815171e-06, "loss": 1.0873, "step": 2014 }, { "epoch": 0.54, "learning_rate": 9.306354294405478e-06, "loss": 1.0831, "step": 2015 }, { "epoch": 0.54, "learning_rate": 9.297751682840613e-06, "loss": 1.0632, "step": 2016 }, { "epoch": 0.54, "learning_rate": 9.289149593518091e-06, "loss": 1.0609, "step": 2017 }, { "epoch": 0.54, "learning_rate": 9.280548032835048e-06, "loss": 1.0695, "step": 2018 }, { "epoch": 0.54, "learning_rate": 9.271947007188218e-06, "loss": 1.0621, "step": 2019 }, { "epoch": 0.54, "learning_rate": 9.26334652297395e-06, "loss": 1.0481, "step": 2020 }, { "epoch": 0.54, "learning_rate": 9.254746586588179e-06, "loss": 1.1021, "step": 2021 }, { "epoch": 0.54, "learning_rate": 9.246147204426433e-06, "loss": 1.0884, "step": 2022 }, { "epoch": 0.54, "learning_rate": 9.237548382883834e-06, "loss": 1.1115, "step": 2023 }, { "epoch": 0.54, "learning_rate": 9.228950128355084e-06, "loss": 1.101, "step": 2024 }, { "epoch": 0.54, "learning_rate": 9.220352447234461e-06, "loss": 1.0834, "step": 2025 }, { "epoch": 0.54, "learning_rate": 9.211755345915822e-06, "loss": 1.0359, "step": 2026 }, { "epoch": 0.54, "learning_rate": 9.203158830792588e-06, "loss": 1.0409, "step": 2027 }, { "epoch": 0.54, "learning_rate": 9.194562908257745e-06, "loss": 1.0707, "step": 2028 }, { "epoch": 0.54, "learning_rate": 9.185967584703841e-06, "loss": 1.0163, "step": 2029 }, { "epoch": 0.54, "learning_rate": 9.177372866522977e-06, "loss": 1.077, "step": 2030 }, { "epoch": 0.54, "learning_rate": 9.168778760106798e-06, "loss": 1.0366, "step": 2031 }, { "epoch": 0.54, "learning_rate": 9.160185271846508e-06, "loss": 1.1087, "step": 2032 }, { "epoch": 0.54, "learning_rate": 9.151592408132836e-06, "loss": 1.0725, "step": 2033 }, { "epoch": 0.54, "learning_rate": 9.143000175356058e-06, "loss": 1.0718, "step": 2034 }, { "epoch": 0.54, "learning_rate": 9.134408579905975e-06, "loss": 1.0701, "step": 2035 }, { "epoch": 0.54, "learning_rate": 9.125817628171912e-06, "loss": 1.0287, "step": 2036 }, { "epoch": 0.54, "learning_rate": 9.117227326542721e-06, "loss": 1.089, "step": 2037 }, { "epoch": 0.54, "learning_rate": 9.108637681406768e-06, "loss": 1.1023, "step": 2038 }, { "epoch": 0.54, "learning_rate": 9.10004869915193e-06, "loss": 1.0617, "step": 2039 }, { "epoch": 0.54, "learning_rate": 9.091460386165593e-06, "loss": 1.0765, "step": 2040 }, { "epoch": 0.54, "learning_rate": 9.082872748834645e-06, "loss": 1.0805, "step": 2041 }, { "epoch": 0.54, "learning_rate": 9.074285793545466e-06, "loss": 1.0703, "step": 2042 }, { "epoch": 0.54, "learning_rate": 9.065699526683937e-06, "loss": 1.0806, "step": 2043 }, { "epoch": 0.54, "learning_rate": 9.057113954635426e-06, "loss": 1.0512, "step": 2044 }, { "epoch": 0.54, "learning_rate": 9.048529083784775e-06, "loss": 1.0684, "step": 2045 }, { "epoch": 0.54, "learning_rate": 9.039944920516316e-06, "loss": 1.1106, "step": 2046 }, { "epoch": 0.54, "learning_rate": 9.03136147121385e-06, "loss": 1.0348, "step": 2047 }, { "epoch": 0.55, "learning_rate": 9.022778742260644e-06, "loss": 1.0597, "step": 2048 }, { "epoch": 0.55, "learning_rate": 9.014196740039437e-06, "loss": 1.0909, "step": 2049 }, { "epoch": 0.55, "learning_rate": 9.005615470932421e-06, "loss": 1.0751, "step": 2050 }, { "epoch": 0.55, "learning_rate": 8.997034941321241e-06, "loss": 1.0892, "step": 2051 }, { "epoch": 0.55, "learning_rate": 8.988455157587001e-06, "loss": 1.0855, "step": 2052 }, { "epoch": 0.55, "learning_rate": 8.979876126110244e-06, "loss": 1.0583, "step": 2053 }, { "epoch": 0.55, "learning_rate": 8.971297853270952e-06, "loss": 1.1212, "step": 2054 }, { "epoch": 0.55, "learning_rate": 8.962720345448548e-06, "loss": 1.1021, "step": 2055 }, { "epoch": 0.55, "learning_rate": 8.954143609021885e-06, "loss": 1.0657, "step": 2056 }, { "epoch": 0.55, "learning_rate": 8.945567650369235e-06, "loss": 1.0675, "step": 2057 }, { "epoch": 0.55, "learning_rate": 8.936992475868305e-06, "loss": 1.0479, "step": 2058 }, { "epoch": 0.55, "learning_rate": 8.928418091896207e-06, "loss": 1.0462, "step": 2059 }, { "epoch": 0.55, "learning_rate": 8.919844504829467e-06, "loss": 1.0718, "step": 2060 }, { "epoch": 0.55, "learning_rate": 8.91127172104403e-06, "loss": 1.0646, "step": 2061 }, { "epoch": 0.55, "learning_rate": 8.902699746915226e-06, "loss": 1.0452, "step": 2062 }, { "epoch": 0.55, "learning_rate": 8.894128588817793e-06, "loss": 1.076, "step": 2063 }, { "epoch": 0.55, "learning_rate": 8.885558253125862e-06, "loss": 1.0962, "step": 2064 }, { "epoch": 0.55, "learning_rate": 8.87698874621295e-06, "loss": 1.0901, "step": 2065 }, { "epoch": 0.55, "learning_rate": 8.86842007445196e-06, "loss": 1.0617, "step": 2066 }, { "epoch": 0.55, "learning_rate": 8.85985224421517e-06, "loss": 1.0675, "step": 2067 }, { "epoch": 0.55, "learning_rate": 8.851285261874239e-06, "loss": 1.0431, "step": 2068 }, { "epoch": 0.55, "learning_rate": 8.842719133800185e-06, "loss": 1.0845, "step": 2069 }, { "epoch": 0.55, "learning_rate": 8.834153866363402e-06, "loss": 1.0979, "step": 2070 }, { "epoch": 0.55, "learning_rate": 8.825589465933638e-06, "loss": 1.0756, "step": 2071 }, { "epoch": 0.55, "learning_rate": 8.817025938879991e-06, "loss": 1.0737, "step": 2072 }, { "epoch": 0.55, "learning_rate": 8.808463291570923e-06, "loss": 1.0958, "step": 2073 }, { "epoch": 0.55, "learning_rate": 8.799901530374233e-06, "loss": 1.0476, "step": 2074 }, { "epoch": 0.55, "learning_rate": 8.791340661657057e-06, "loss": 1.0801, "step": 2075 }, { "epoch": 0.55, "learning_rate": 8.78278069178588e-06, "loss": 1.0977, "step": 2076 }, { "epoch": 0.55, "learning_rate": 8.774221627126506e-06, "loss": 1.0704, "step": 2077 }, { "epoch": 0.55, "learning_rate": 8.765663474044071e-06, "loss": 1.0831, "step": 2078 }, { "epoch": 0.55, "learning_rate": 8.757106238903037e-06, "loss": 1.0601, "step": 2079 }, { "epoch": 0.55, "learning_rate": 8.748549928067177e-06, "loss": 1.1061, "step": 2080 }, { "epoch": 0.55, "learning_rate": 8.739994547899575e-06, "loss": 1.0969, "step": 2081 }, { "epoch": 0.55, "learning_rate": 8.731440104762634e-06, "loss": 1.1074, "step": 2082 }, { "epoch": 0.55, "learning_rate": 8.722886605018049e-06, "loss": 1.0841, "step": 2083 }, { "epoch": 0.55, "learning_rate": 8.714334055026816e-06, "loss": 1.1098, "step": 2084 }, { "epoch": 0.56, "learning_rate": 8.70578246114923e-06, "loss": 1.0618, "step": 2085 }, { "epoch": 0.56, "learning_rate": 8.697231829744873e-06, "loss": 1.0741, "step": 2086 }, { "epoch": 0.56, "learning_rate": 8.688682167172599e-06, "loss": 1.0521, "step": 2087 }, { "epoch": 0.56, "learning_rate": 8.680133479790563e-06, "loss": 1.1073, "step": 2088 }, { "epoch": 0.56, "learning_rate": 8.671585773956178e-06, "loss": 1.0588, "step": 2089 }, { "epoch": 0.56, "learning_rate": 8.663039056026135e-06, "loss": 1.1211, "step": 2090 }, { "epoch": 0.56, "learning_rate": 8.654493332356388e-06, "loss": 1.1059, "step": 2091 }, { "epoch": 0.56, "learning_rate": 8.645948609302149e-06, "loss": 1.0753, "step": 2092 }, { "epoch": 0.56, "learning_rate": 8.63740489321789e-06, "loss": 1.109, "step": 2093 }, { "epoch": 0.56, "learning_rate": 8.628862190457334e-06, "loss": 1.1048, "step": 2094 }, { "epoch": 0.56, "learning_rate": 8.620320507373446e-06, "loss": 1.0824, "step": 2095 }, { "epoch": 0.56, "learning_rate": 8.611779850318441e-06, "loss": 1.1018, "step": 2096 }, { "epoch": 0.56, "learning_rate": 8.60324022564376e-06, "loss": 1.1395, "step": 2097 }, { "epoch": 0.56, "learning_rate": 8.594701639700084e-06, "loss": 1.0924, "step": 2098 }, { "epoch": 0.56, "learning_rate": 8.586164098837321e-06, "loss": 1.0618, "step": 2099 }, { "epoch": 0.56, "learning_rate": 8.5776276094046e-06, "loss": 1.0488, "step": 2100 }, { "epoch": 0.56, "learning_rate": 8.569092177750264e-06, "loss": 1.0413, "step": 2101 }, { "epoch": 0.56, "learning_rate": 8.56055781022188e-06, "loss": 1.11, "step": 2102 }, { "epoch": 0.56, "learning_rate": 8.552024513166215e-06, "loss": 1.0741, "step": 2103 }, { "epoch": 0.56, "learning_rate": 8.54349229292924e-06, "loss": 1.0793, "step": 2104 }, { "epoch": 0.56, "learning_rate": 8.534961155856131e-06, "loss": 1.0963, "step": 2105 }, { "epoch": 0.56, "learning_rate": 8.526431108291253e-06, "loss": 1.073, "step": 2106 }, { "epoch": 0.56, "learning_rate": 8.517902156578161e-06, "loss": 1.0578, "step": 2107 }, { "epoch": 0.56, "learning_rate": 8.509374307059602e-06, "loss": 1.0176, "step": 2108 }, { "epoch": 0.56, "learning_rate": 8.500847566077494e-06, "loss": 1.0546, "step": 2109 }, { "epoch": 0.56, "learning_rate": 8.492321939972933e-06, "loss": 1.0516, "step": 2110 }, { "epoch": 0.56, "learning_rate": 8.483797435086195e-06, "loss": 1.0609, "step": 2111 }, { "epoch": 0.56, "learning_rate": 8.475274057756709e-06, "loss": 1.0496, "step": 2112 }, { "epoch": 0.56, "learning_rate": 8.46675181432307e-06, "loss": 1.0413, "step": 2113 }, { "epoch": 0.56, "learning_rate": 8.458230711123035e-06, "loss": 1.0041, "step": 2114 }, { "epoch": 0.56, "learning_rate": 8.449710754493509e-06, "loss": 1.0413, "step": 2115 }, { "epoch": 0.56, "learning_rate": 8.44119195077054e-06, "loss": 1.1032, "step": 2116 }, { "epoch": 0.56, "learning_rate": 8.432674306289328e-06, "loss": 1.0336, "step": 2117 }, { "epoch": 0.56, "learning_rate": 8.424157827384204e-06, "loss": 1.0655, "step": 2118 }, { "epoch": 0.56, "learning_rate": 8.415642520388631e-06, "loss": 1.0408, "step": 2119 }, { "epoch": 0.56, "learning_rate": 8.407128391635212e-06, "loss": 1.0548, "step": 2120 }, { "epoch": 0.56, "learning_rate": 8.398615447455656e-06, "loss": 1.0586, "step": 2121 }, { "epoch": 0.56, "learning_rate": 8.390103694180805e-06, "loss": 1.0543, "step": 2122 }, { "epoch": 0.57, "learning_rate": 8.381593138140612e-06, "loss": 1.0429, "step": 2123 }, { "epoch": 0.57, "learning_rate": 8.373083785664134e-06, "loss": 1.0593, "step": 2124 }, { "epoch": 0.57, "learning_rate": 8.364575643079539e-06, "loss": 1.0174, "step": 2125 }, { "epoch": 0.57, "learning_rate": 8.356068716714098e-06, "loss": 1.0757, "step": 2126 }, { "epoch": 0.57, "learning_rate": 8.347563012894169e-06, "loss": 1.0795, "step": 2127 }, { "epoch": 0.57, "learning_rate": 8.339058537945205e-06, "loss": 1.0432, "step": 2128 }, { "epoch": 0.57, "learning_rate": 8.330555298191747e-06, "loss": 1.0446, "step": 2129 }, { "epoch": 0.57, "learning_rate": 8.322053299957414e-06, "loss": 1.0527, "step": 2130 }, { "epoch": 0.57, "learning_rate": 8.313552549564902e-06, "loss": 1.0429, "step": 2131 }, { "epoch": 0.57, "learning_rate": 8.305053053335983e-06, "loss": 1.04, "step": 2132 }, { "epoch": 0.57, "learning_rate": 8.296554817591493e-06, "loss": 1.0368, "step": 2133 }, { "epoch": 0.57, "learning_rate": 8.28805784865133e-06, "loss": 1.1047, "step": 2134 }, { "epoch": 0.57, "learning_rate": 8.279562152834453e-06, "loss": 1.0191, "step": 2135 }, { "epoch": 0.57, "learning_rate": 8.27106773645887e-06, "loss": 1.0629, "step": 2136 }, { "epoch": 0.57, "learning_rate": 8.26257460584164e-06, "loss": 1.0736, "step": 2137 }, { "epoch": 0.57, "learning_rate": 8.254082767298868e-06, "loss": 1.0998, "step": 2138 }, { "epoch": 0.57, "learning_rate": 8.245592227145694e-06, "loss": 1.0186, "step": 2139 }, { "epoch": 0.57, "learning_rate": 8.237102991696293e-06, "loss": 1.0921, "step": 2140 }, { "epoch": 0.57, "learning_rate": 8.228615067263874e-06, "loss": 1.0355, "step": 2141 }, { "epoch": 0.57, "learning_rate": 8.220128460160667e-06, "loss": 1.0812, "step": 2142 }, { "epoch": 0.57, "learning_rate": 8.21164317669792e-06, "loss": 1.1064, "step": 2143 }, { "epoch": 0.57, "learning_rate": 8.203159223185905e-06, "loss": 1.0429, "step": 2144 }, { "epoch": 0.57, "learning_rate": 8.194676605933896e-06, "loss": 1.0619, "step": 2145 }, { "epoch": 0.57, "learning_rate": 8.18619533125018e-06, "loss": 1.0557, "step": 2146 }, { "epoch": 0.57, "learning_rate": 8.17771540544204e-06, "loss": 1.08, "step": 2147 }, { "epoch": 0.57, "learning_rate": 8.16923683481576e-06, "loss": 1.0637, "step": 2148 }, { "epoch": 0.57, "learning_rate": 8.160759625676615e-06, "loss": 1.1005, "step": 2149 }, { "epoch": 0.57, "learning_rate": 8.15228378432887e-06, "loss": 1.0615, "step": 2150 }, { "epoch": 0.57, "learning_rate": 8.143809317075762e-06, "loss": 1.1178, "step": 2151 }, { "epoch": 0.57, "learning_rate": 8.135336230219523e-06, "loss": 1.0774, "step": 2152 }, { "epoch": 0.57, "learning_rate": 8.126864530061345e-06, "loss": 1.0625, "step": 2153 }, { "epoch": 0.57, "learning_rate": 8.118394222901393e-06, "loss": 1.0285, "step": 2154 }, { "epoch": 0.57, "learning_rate": 8.109925315038799e-06, "loss": 1.052, "step": 2155 }, { "epoch": 0.57, "learning_rate": 8.101457812771651e-06, "loss": 1.0679, "step": 2156 }, { "epoch": 0.57, "learning_rate": 8.092991722396993e-06, "loss": 1.0374, "step": 2157 }, { "epoch": 0.57, "learning_rate": 8.084527050210816e-06, "loss": 1.0207, "step": 2158 }, { "epoch": 0.57, "learning_rate": 8.076063802508063e-06, "loss": 1.0772, "step": 2159 }, { "epoch": 0.58, "learning_rate": 8.067601985582609e-06, "loss": 1.0464, "step": 2160 }, { "epoch": 0.58, "learning_rate": 8.059141605727275e-06, "loss": 1.0735, "step": 2161 }, { "epoch": 0.58, "learning_rate": 8.050682669233803e-06, "loss": 1.0971, "step": 2162 }, { "epoch": 0.58, "learning_rate": 8.042225182392867e-06, "loss": 1.0591, "step": 2163 }, { "epoch": 0.58, "learning_rate": 8.033769151494065e-06, "loss": 1.034, "step": 2164 }, { "epoch": 0.58, "learning_rate": 8.025314582825908e-06, "loss": 1.0369, "step": 2165 }, { "epoch": 0.58, "learning_rate": 8.016861482675819e-06, "loss": 1.0743, "step": 2166 }, { "epoch": 0.58, "learning_rate": 8.008409857330136e-06, "loss": 1.0748, "step": 2167 }, { "epoch": 0.58, "learning_rate": 7.999959713074093e-06, "loss": 1.1161, "step": 2168 }, { "epoch": 0.58, "learning_rate": 7.991511056191823e-06, "loss": 1.0326, "step": 2169 }, { "epoch": 0.58, "learning_rate": 7.983063892966358e-06, "loss": 1.0646, "step": 2170 }, { "epoch": 0.58, "learning_rate": 7.974618229679615e-06, "loss": 1.0907, "step": 2171 }, { "epoch": 0.58, "learning_rate": 7.966174072612393e-06, "loss": 1.0755, "step": 2172 }, { "epoch": 0.58, "learning_rate": 7.957731428044381e-06, "loss": 1.1361, "step": 2173 }, { "epoch": 0.58, "learning_rate": 7.949290302254132e-06, "loss": 1.0891, "step": 2174 }, { "epoch": 0.58, "learning_rate": 7.940850701519072e-06, "loss": 1.0641, "step": 2175 }, { "epoch": 0.58, "learning_rate": 7.932412632115501e-06, "loss": 1.0808, "step": 2176 }, { "epoch": 0.58, "learning_rate": 7.92397610031857e-06, "loss": 1.0803, "step": 2177 }, { "epoch": 0.58, "learning_rate": 7.915541112402293e-06, "loss": 1.0356, "step": 2178 }, { "epoch": 0.58, "learning_rate": 7.907107674639533e-06, "loss": 1.037, "step": 2179 }, { "epoch": 0.58, "learning_rate": 7.898675793302e-06, "loss": 1.0653, "step": 2180 }, { "epoch": 0.58, "learning_rate": 7.890245474660246e-06, "loss": 1.0417, "step": 2181 }, { "epoch": 0.58, "learning_rate": 7.881816724983663e-06, "loss": 1.0474, "step": 2182 }, { "epoch": 0.58, "learning_rate": 7.873389550540475e-06, "loss": 1.0402, "step": 2183 }, { "epoch": 0.58, "learning_rate": 7.864963957597739e-06, "loss": 1.0922, "step": 2184 }, { "epoch": 0.58, "learning_rate": 7.856539952421325e-06, "loss": 1.0343, "step": 2185 }, { "epoch": 0.58, "learning_rate": 7.848117541275937e-06, "loss": 1.1003, "step": 2186 }, { "epoch": 0.58, "learning_rate": 7.83969673042508e-06, "loss": 1.0451, "step": 2187 }, { "epoch": 0.58, "learning_rate": 7.831277526131074e-06, "loss": 1.0432, "step": 2188 }, { "epoch": 0.58, "learning_rate": 7.822859934655049e-06, "loss": 1.0994, "step": 2189 }, { "epoch": 0.58, "learning_rate": 7.814443962256928e-06, "loss": 1.075, "step": 2190 }, { "epoch": 0.58, "learning_rate": 7.806029615195432e-06, "loss": 1.0755, "step": 2191 }, { "epoch": 0.58, "learning_rate": 7.797616899728078e-06, "loss": 1.0261, "step": 2192 }, { "epoch": 0.58, "learning_rate": 7.789205822111165e-06, "loss": 1.0623, "step": 2193 }, { "epoch": 0.58, "learning_rate": 7.780796388599774e-06, "loss": 1.0744, "step": 2194 }, { "epoch": 0.58, "learning_rate": 7.772388605447766e-06, "loss": 1.1166, "step": 2195 }, { "epoch": 0.58, "learning_rate": 7.763982478907773e-06, "loss": 1.0788, "step": 2196 }, { "epoch": 0.58, "learning_rate": 7.755578015231191e-06, "loss": 1.0443, "step": 2197 }, { "epoch": 0.59, "learning_rate": 7.747175220668191e-06, "loss": 1.1109, "step": 2198 }, { "epoch": 0.59, "learning_rate": 7.73877410146769e-06, "loss": 1.0245, "step": 2199 }, { "epoch": 0.59, "learning_rate": 7.730374663877362e-06, "loss": 1.1183, "step": 2200 }, { "epoch": 0.59, "learning_rate": 7.721976914143641e-06, "loss": 1.0838, "step": 2201 }, { "epoch": 0.59, "learning_rate": 7.713580858511693e-06, "loss": 1.0772, "step": 2202 }, { "epoch": 0.59, "learning_rate": 7.705186503225425e-06, "loss": 1.0411, "step": 2203 }, { "epoch": 0.59, "learning_rate": 7.69679385452749e-06, "loss": 1.0749, "step": 2204 }, { "epoch": 0.59, "learning_rate": 7.68840291865926e-06, "loss": 1.0761, "step": 2205 }, { "epoch": 0.59, "learning_rate": 7.680013701860841e-06, "loss": 1.08, "step": 2206 }, { "epoch": 0.59, "learning_rate": 7.671626210371058e-06, "loss": 1.0396, "step": 2207 }, { "epoch": 0.59, "learning_rate": 7.663240450427452e-06, "loss": 1.0747, "step": 2208 }, { "epoch": 0.59, "learning_rate": 7.654856428266275e-06, "loss": 1.0297, "step": 2209 }, { "epoch": 0.59, "learning_rate": 7.646474150122493e-06, "loss": 1.0186, "step": 2210 }, { "epoch": 0.59, "learning_rate": 7.638093622229768e-06, "loss": 1.0454, "step": 2211 }, { "epoch": 0.59, "learning_rate": 7.629714850820461e-06, "loss": 1.0417, "step": 2212 }, { "epoch": 0.59, "learning_rate": 7.621337842125634e-06, "loss": 1.038, "step": 2213 }, { "epoch": 0.59, "learning_rate": 7.612962602375029e-06, "loss": 1.0411, "step": 2214 }, { "epoch": 0.59, "learning_rate": 7.604589137797079e-06, "loss": 1.0444, "step": 2215 }, { "epoch": 0.59, "learning_rate": 7.596217454618895e-06, "loss": 1.0592, "step": 2216 }, { "epoch": 0.59, "learning_rate": 7.587847559066258e-06, "loss": 1.0675, "step": 2217 }, { "epoch": 0.59, "learning_rate": 7.57947945736363e-06, "loss": 1.0836, "step": 2218 }, { "epoch": 0.59, "learning_rate": 7.57111315573413e-06, "loss": 1.0231, "step": 2219 }, { "epoch": 0.59, "learning_rate": 7.562748660399539e-06, "loss": 1.0769, "step": 2220 }, { "epoch": 0.59, "learning_rate": 7.5543859775803034e-06, "loss": 1.028, "step": 2221 }, { "epoch": 0.59, "learning_rate": 7.546025113495512e-06, "loss": 1.0901, "step": 2222 }, { "epoch": 0.59, "learning_rate": 7.537666074362903e-06, "loss": 1.066, "step": 2223 }, { "epoch": 0.59, "learning_rate": 7.529308866398861e-06, "loss": 1.0707, "step": 2224 }, { "epoch": 0.59, "learning_rate": 7.520953495818408e-06, "loss": 1.077, "step": 2225 }, { "epoch": 0.59, "learning_rate": 7.5125999688351946e-06, "loss": 1.0676, "step": 2226 }, { "epoch": 0.59, "learning_rate": 7.504248291661509e-06, "loss": 1.1037, "step": 2227 }, { "epoch": 0.59, "learning_rate": 7.4958984705082565e-06, "loss": 1.0721, "step": 2228 }, { "epoch": 0.59, "learning_rate": 7.48755051158496e-06, "loss": 1.0681, "step": 2229 }, { "epoch": 0.59, "learning_rate": 7.4792044210997705e-06, "loss": 1.0616, "step": 2230 }, { "epoch": 0.59, "learning_rate": 7.4708602052594355e-06, "loss": 1.0636, "step": 2231 }, { "epoch": 0.59, "learning_rate": 7.462517870269315e-06, "loss": 1.0704, "step": 2232 }, { "epoch": 0.59, "learning_rate": 7.45417742233337e-06, "loss": 1.0824, "step": 2233 }, { "epoch": 0.59, "learning_rate": 7.445838867654158e-06, "loss": 1.0527, "step": 2234 }, { "epoch": 0.6, "learning_rate": 7.437502212432823e-06, "loss": 1.0866, "step": 2235 }, { "epoch": 0.6, "learning_rate": 7.429167462869108e-06, "loss": 1.0274, "step": 2236 }, { "epoch": 0.6, "learning_rate": 7.420834625161328e-06, "loss": 1.0644, "step": 2237 }, { "epoch": 0.6, "learning_rate": 7.412503705506382e-06, "loss": 1.099, "step": 2238 }, { "epoch": 0.6, "learning_rate": 7.40417471009974e-06, "loss": 1.0563, "step": 2239 }, { "epoch": 0.6, "learning_rate": 7.395847645135442e-06, "loss": 1.033, "step": 2240 }, { "epoch": 0.6, "learning_rate": 7.3875225168060915e-06, "loss": 1.0558, "step": 2241 }, { "epoch": 0.6, "learning_rate": 7.379199331302856e-06, "loss": 1.0715, "step": 2242 }, { "epoch": 0.6, "learning_rate": 7.37087809481545e-06, "loss": 1.0787, "step": 2243 }, { "epoch": 0.6, "learning_rate": 7.362558813532146e-06, "loss": 1.0292, "step": 2244 }, { "epoch": 0.6, "learning_rate": 7.35424149363976e-06, "loss": 1.026, "step": 2245 }, { "epoch": 0.6, "learning_rate": 7.345926141323648e-06, "loss": 1.0794, "step": 2246 }, { "epoch": 0.6, "learning_rate": 7.3376127627677026e-06, "loss": 1.0899, "step": 2247 }, { "epoch": 0.6, "learning_rate": 7.329301364154355e-06, "loss": 1.0808, "step": 2248 }, { "epoch": 0.6, "learning_rate": 7.320991951664554e-06, "loss": 1.0931, "step": 2249 }, { "epoch": 0.6, "learning_rate": 7.312684531477778e-06, "loss": 1.0932, "step": 2250 }, { "epoch": 0.6, "learning_rate": 7.304379109772022e-06, "loss": 1.0483, "step": 2251 }, { "epoch": 0.6, "learning_rate": 7.296075692723794e-06, "loss": 1.1124, "step": 2252 }, { "epoch": 0.6, "learning_rate": 7.287774286508111e-06, "loss": 1.0774, "step": 2253 }, { "epoch": 0.6, "learning_rate": 7.2794748972985e-06, "loss": 1.1465, "step": 2254 }, { "epoch": 0.6, "learning_rate": 7.27117753126698e-06, "loss": 1.1062, "step": 2255 }, { "epoch": 0.6, "learning_rate": 7.262882194584069e-06, "loss": 1.0428, "step": 2256 }, { "epoch": 0.6, "learning_rate": 7.254588893418777e-06, "loss": 1.0422, "step": 2257 }, { "epoch": 0.6, "learning_rate": 7.2462976339386e-06, "loss": 1.0695, "step": 2258 }, { "epoch": 0.6, "learning_rate": 7.23800842230951e-06, "loss": 1.0654, "step": 2259 }, { "epoch": 0.6, "learning_rate": 7.229721264695968e-06, "loss": 1.0881, "step": 2260 }, { "epoch": 0.6, "learning_rate": 7.221436167260897e-06, "loss": 1.1004, "step": 2261 }, { "epoch": 0.6, "learning_rate": 7.213153136165687e-06, "loss": 1.0763, "step": 2262 }, { "epoch": 0.6, "learning_rate": 7.204872177570203e-06, "loss": 1.098, "step": 2263 }, { "epoch": 0.6, "learning_rate": 7.196593297632757e-06, "loss": 1.0738, "step": 2264 }, { "epoch": 0.6, "learning_rate": 7.188316502510117e-06, "loss": 1.0966, "step": 2265 }, { "epoch": 0.6, "learning_rate": 7.180041798357509e-06, "loss": 1.1105, "step": 2266 }, { "epoch": 0.6, "learning_rate": 7.171769191328595e-06, "loss": 1.0688, "step": 2267 }, { "epoch": 0.6, "learning_rate": 7.1634986875754785e-06, "loss": 1.0766, "step": 2268 }, { "epoch": 0.6, "learning_rate": 7.155230293248705e-06, "loss": 1.1057, "step": 2269 }, { "epoch": 0.6, "learning_rate": 7.146964014497244e-06, "loss": 1.0881, "step": 2270 }, { "epoch": 0.6, "learning_rate": 7.138699857468496e-06, "loss": 1.0266, "step": 2271 }, { "epoch": 0.6, "learning_rate": 7.1304378283082835e-06, "loss": 1.0835, "step": 2272 }, { "epoch": 0.61, "learning_rate": 7.122177933160841e-06, "loss": 1.1142, "step": 2273 }, { "epoch": 0.61, "learning_rate": 7.113920178168825e-06, "loss": 1.0781, "step": 2274 }, { "epoch": 0.61, "learning_rate": 7.105664569473296e-06, "loss": 1.0848, "step": 2275 }, { "epoch": 0.61, "learning_rate": 7.097411113213715e-06, "loss": 1.1063, "step": 2276 }, { "epoch": 0.61, "learning_rate": 7.089159815527948e-06, "loss": 1.0825, "step": 2277 }, { "epoch": 0.61, "learning_rate": 7.0809106825522535e-06, "loss": 1.1028, "step": 2278 }, { "epoch": 0.61, "learning_rate": 7.072663720421276e-06, "loss": 1.1095, "step": 2279 }, { "epoch": 0.61, "learning_rate": 7.064418935268056e-06, "loss": 1.0578, "step": 2280 }, { "epoch": 0.61, "learning_rate": 7.056176333224005e-06, "loss": 1.0763, "step": 2281 }, { "epoch": 0.61, "learning_rate": 7.047935920418911e-06, "loss": 1.0462, "step": 2282 }, { "epoch": 0.61, "learning_rate": 7.0396977029809455e-06, "loss": 1.0718, "step": 2283 }, { "epoch": 0.61, "learning_rate": 7.031461687036633e-06, "loss": 1.0804, "step": 2284 }, { "epoch": 0.61, "learning_rate": 7.0232278787108695e-06, "loss": 1.0897, "step": 2285 }, { "epoch": 0.61, "learning_rate": 7.0149962841269095e-06, "loss": 1.0574, "step": 2286 }, { "epoch": 0.61, "learning_rate": 7.006766909406353e-06, "loss": 1.0596, "step": 2287 }, { "epoch": 0.61, "learning_rate": 6.998539760669157e-06, "loss": 1.1148, "step": 2288 }, { "epoch": 0.61, "learning_rate": 6.9903148440336245e-06, "loss": 1.1159, "step": 2289 }, { "epoch": 0.61, "learning_rate": 6.982092165616392e-06, "loss": 1.0823, "step": 2290 }, { "epoch": 0.61, "learning_rate": 6.973871731532431e-06, "loss": 1.0539, "step": 2291 }, { "epoch": 0.61, "learning_rate": 6.965653547895055e-06, "loss": 1.0862, "step": 2292 }, { "epoch": 0.61, "learning_rate": 6.95743762081589e-06, "loss": 1.056, "step": 2293 }, { "epoch": 0.61, "learning_rate": 6.9492239564048915e-06, "loss": 1.0807, "step": 2294 }, { "epoch": 0.61, "learning_rate": 6.941012560770335e-06, "loss": 1.0643, "step": 2295 }, { "epoch": 0.61, "learning_rate": 6.9328034400188004e-06, "loss": 1.0533, "step": 2296 }, { "epoch": 0.61, "learning_rate": 6.924596600255178e-06, "loss": 1.0828, "step": 2297 }, { "epoch": 0.61, "learning_rate": 6.9163920475826714e-06, "loss": 1.1016, "step": 2298 }, { "epoch": 0.61, "learning_rate": 6.908189788102771e-06, "loss": 1.0559, "step": 2299 }, { "epoch": 0.61, "learning_rate": 6.899989827915265e-06, "loss": 1.0865, "step": 2300 }, { "epoch": 0.61, "learning_rate": 6.89179217311824e-06, "loss": 1.0445, "step": 2301 }, { "epoch": 0.61, "learning_rate": 6.883596829808057e-06, "loss": 1.0888, "step": 2302 }, { "epoch": 0.61, "learning_rate": 6.875403804079361e-06, "loss": 1.0805, "step": 2303 }, { "epoch": 0.61, "learning_rate": 6.8672131020250785e-06, "loss": 1.061, "step": 2304 }, { "epoch": 0.61, "learning_rate": 6.8590247297364055e-06, "loss": 1.0731, "step": 2305 }, { "epoch": 0.61, "learning_rate": 6.850838693302798e-06, "loss": 1.0552, "step": 2306 }, { "epoch": 0.61, "learning_rate": 6.842654998811988e-06, "loss": 1.0623, "step": 2307 }, { "epoch": 0.61, "learning_rate": 6.834473652349958e-06, "loss": 1.1154, "step": 2308 }, { "epoch": 0.61, "learning_rate": 6.826294660000942e-06, "loss": 1.0809, "step": 2309 }, { "epoch": 0.62, "learning_rate": 6.8181180278474305e-06, "loss": 1.0834, "step": 2310 }, { "epoch": 0.62, "learning_rate": 6.809943761970154e-06, "loss": 1.1071, "step": 2311 }, { "epoch": 0.62, "learning_rate": 6.801771868448082e-06, "loss": 1.0937, "step": 2312 }, { "epoch": 0.62, "learning_rate": 6.793602353358426e-06, "loss": 1.0914, "step": 2313 }, { "epoch": 0.62, "learning_rate": 6.785435222776624e-06, "loss": 1.0212, "step": 2314 }, { "epoch": 0.62, "learning_rate": 6.777270482776339e-06, "loss": 1.0853, "step": 2315 }, { "epoch": 0.62, "learning_rate": 6.769108139429461e-06, "loss": 1.0409, "step": 2316 }, { "epoch": 0.62, "learning_rate": 6.760948198806096e-06, "loss": 1.0194, "step": 2317 }, { "epoch": 0.62, "learning_rate": 6.75279066697456e-06, "loss": 1.0809, "step": 2318 }, { "epoch": 0.62, "learning_rate": 6.744635550001383e-06, "loss": 0.9894, "step": 2319 }, { "epoch": 0.62, "learning_rate": 6.736482853951296e-06, "loss": 1.0907, "step": 2320 }, { "epoch": 0.62, "learning_rate": 6.728332584887227e-06, "loss": 1.0726, "step": 2321 }, { "epoch": 0.62, "learning_rate": 6.720184748870307e-06, "loss": 1.1126, "step": 2322 }, { "epoch": 0.62, "learning_rate": 6.712039351959849e-06, "loss": 1.0629, "step": 2323 }, { "epoch": 0.62, "learning_rate": 6.7038964002133565e-06, "loss": 1.0633, "step": 2324 }, { "epoch": 0.62, "learning_rate": 6.695755899686516e-06, "loss": 1.0801, "step": 2325 }, { "epoch": 0.62, "learning_rate": 6.687617856433184e-06, "loss": 1.0729, "step": 2326 }, { "epoch": 0.62, "learning_rate": 6.6794822765054015e-06, "loss": 1.0796, "step": 2327 }, { "epoch": 0.62, "learning_rate": 6.671349165953367e-06, "loss": 1.0732, "step": 2328 }, { "epoch": 0.62, "learning_rate": 6.663218530825444e-06, "loss": 1.0346, "step": 2329 }, { "epoch": 0.62, "learning_rate": 6.655090377168163e-06, "loss": 1.1146, "step": 2330 }, { "epoch": 0.62, "learning_rate": 6.6469647110262e-06, "loss": 1.0774, "step": 2331 }, { "epoch": 0.62, "learning_rate": 6.6388415384423845e-06, "loss": 1.1039, "step": 2332 }, { "epoch": 0.62, "learning_rate": 6.630720865457694e-06, "loss": 1.0201, "step": 2333 }, { "epoch": 0.62, "learning_rate": 6.622602698111244e-06, "loss": 1.0981, "step": 2334 }, { "epoch": 0.62, "learning_rate": 6.614487042440285e-06, "loss": 1.1098, "step": 2335 }, { "epoch": 0.62, "learning_rate": 6.606373904480205e-06, "loss": 1.0751, "step": 2336 }, { "epoch": 0.62, "learning_rate": 6.598263290264517e-06, "loss": 1.056, "step": 2337 }, { "epoch": 0.62, "learning_rate": 6.590155205824853e-06, "loss": 1.1249, "step": 2338 }, { "epoch": 0.62, "learning_rate": 6.582049657190972e-06, "loss": 1.128, "step": 2339 }, { "epoch": 0.62, "learning_rate": 6.5739466503907425e-06, "loss": 1.0652, "step": 2340 }, { "epoch": 0.62, "learning_rate": 6.565846191450138e-06, "loss": 1.0575, "step": 2341 }, { "epoch": 0.62, "learning_rate": 6.557748286393249e-06, "loss": 1.1079, "step": 2342 }, { "epoch": 0.62, "learning_rate": 6.549652941242257e-06, "loss": 1.0796, "step": 2343 }, { "epoch": 0.62, "learning_rate": 6.5415601620174416e-06, "loss": 1.0782, "step": 2344 }, { "epoch": 0.62, "learning_rate": 6.533469954737177e-06, "loss": 1.0857, "step": 2345 }, { "epoch": 0.62, "learning_rate": 6.525382325417924e-06, "loss": 1.086, "step": 2346 }, { "epoch": 0.62, "learning_rate": 6.517297280074222e-06, "loss": 1.0701, "step": 2347 }, { "epoch": 0.63, "learning_rate": 6.509214824718695e-06, "loss": 1.0242, "step": 2348 }, { "epoch": 0.63, "learning_rate": 6.501134965362033e-06, "loss": 1.0939, "step": 2349 }, { "epoch": 0.63, "learning_rate": 6.493057708013009e-06, "loss": 1.1009, "step": 2350 }, { "epoch": 0.63, "learning_rate": 6.484983058678445e-06, "loss": 1.0525, "step": 2351 }, { "epoch": 0.63, "learning_rate": 6.476911023363237e-06, "loss": 1.0992, "step": 2352 }, { "epoch": 0.63, "learning_rate": 6.4688416080703266e-06, "loss": 1.1088, "step": 2353 }, { "epoch": 0.63, "learning_rate": 6.46077481880071e-06, "loss": 1.0964, "step": 2354 }, { "epoch": 0.63, "learning_rate": 6.452710661553437e-06, "loss": 1.0669, "step": 2355 }, { "epoch": 0.63, "learning_rate": 6.444649142325592e-06, "loss": 1.0722, "step": 2356 }, { "epoch": 0.63, "learning_rate": 6.436590267112297e-06, "loss": 1.0379, "step": 2357 }, { "epoch": 0.63, "learning_rate": 6.428534041906715e-06, "loss": 1.0781, "step": 2358 }, { "epoch": 0.63, "learning_rate": 6.420480472700037e-06, "loss": 1.0281, "step": 2359 }, { "epoch": 0.63, "learning_rate": 6.412429565481468e-06, "loss": 1.0561, "step": 2360 }, { "epoch": 0.63, "learning_rate": 6.40438132623825e-06, "loss": 1.0616, "step": 2361 }, { "epoch": 0.63, "learning_rate": 6.396335760955629e-06, "loss": 1.0822, "step": 2362 }, { "epoch": 0.63, "learning_rate": 6.388292875616862e-06, "loss": 1.0818, "step": 2363 }, { "epoch": 0.63, "learning_rate": 6.380252676203224e-06, "loss": 1.1037, "step": 2364 }, { "epoch": 0.63, "learning_rate": 6.3722151686939806e-06, "loss": 1.0626, "step": 2365 }, { "epoch": 0.63, "learning_rate": 6.364180359066401e-06, "loss": 1.148, "step": 2366 }, { "epoch": 0.63, "learning_rate": 6.35614825329575e-06, "loss": 1.0591, "step": 2367 }, { "epoch": 0.63, "learning_rate": 6.3481188573552765e-06, "loss": 1.104, "step": 2368 }, { "epoch": 0.63, "learning_rate": 6.340092177216215e-06, "loss": 1.0838, "step": 2369 }, { "epoch": 0.63, "learning_rate": 6.332068218847786e-06, "loss": 1.1171, "step": 2370 }, { "epoch": 0.63, "learning_rate": 6.3240469882171815e-06, "loss": 1.0818, "step": 2371 }, { "epoch": 0.63, "learning_rate": 6.316028491289562e-06, "loss": 1.0671, "step": 2372 }, { "epoch": 0.63, "learning_rate": 6.308012734028062e-06, "loss": 1.047, "step": 2373 }, { "epoch": 0.63, "learning_rate": 6.299999722393775e-06, "loss": 1.0724, "step": 2374 }, { "epoch": 0.63, "learning_rate": 6.291989462345746e-06, "loss": 1.0712, "step": 2375 }, { "epoch": 0.63, "learning_rate": 6.28398195984099e-06, "loss": 1.0383, "step": 2376 }, { "epoch": 0.63, "learning_rate": 6.2759772208344576e-06, "loss": 1.1041, "step": 2377 }, { "epoch": 0.63, "learning_rate": 6.267975251279044e-06, "loss": 1.0692, "step": 2378 }, { "epoch": 0.63, "learning_rate": 6.259976057125596e-06, "loss": 1.0504, "step": 2379 }, { "epoch": 0.63, "learning_rate": 6.251979644322887e-06, "loss": 1.0553, "step": 2380 }, { "epoch": 0.63, "learning_rate": 6.243986018817623e-06, "loss": 1.0693, "step": 2381 }, { "epoch": 0.63, "learning_rate": 6.23599518655444e-06, "loss": 1.043, "step": 2382 }, { "epoch": 0.63, "learning_rate": 6.2280071534758965e-06, "loss": 1.0901, "step": 2383 }, { "epoch": 0.63, "learning_rate": 6.220021925522465e-06, "loss": 1.0665, "step": 2384 }, { "epoch": 0.63, "learning_rate": 6.212039508632539e-06, "loss": 1.0517, "step": 2385 }, { "epoch": 0.64, "learning_rate": 6.204059908742417e-06, "loss": 1.0621, "step": 2386 }, { "epoch": 0.64, "learning_rate": 6.196083131786299e-06, "loss": 1.0481, "step": 2387 }, { "epoch": 0.64, "learning_rate": 6.188109183696297e-06, "loss": 1.0887, "step": 2388 }, { "epoch": 0.64, "learning_rate": 6.180138070402407e-06, "loss": 1.1495, "step": 2389 }, { "epoch": 0.64, "learning_rate": 6.172169797832522e-06, "loss": 1.0936, "step": 2390 }, { "epoch": 0.64, "learning_rate": 6.1642043719124265e-06, "loss": 1.0877, "step": 2391 }, { "epoch": 0.64, "learning_rate": 6.1562417985657806e-06, "loss": 1.096, "step": 2392 }, { "epoch": 0.64, "learning_rate": 6.148282083714126e-06, "loss": 1.0645, "step": 2393 }, { "epoch": 0.64, "learning_rate": 6.140325233276882e-06, "loss": 1.0687, "step": 2394 }, { "epoch": 0.64, "learning_rate": 6.132371253171332e-06, "loss": 1.0476, "step": 2395 }, { "epoch": 0.64, "learning_rate": 6.124420149312628e-06, "loss": 1.0644, "step": 2396 }, { "epoch": 0.64, "learning_rate": 6.1164719276137865e-06, "loss": 1.1004, "step": 2397 }, { "epoch": 0.64, "learning_rate": 6.108526593985669e-06, "loss": 1.0756, "step": 2398 }, { "epoch": 0.64, "learning_rate": 6.1005841543370035e-06, "loss": 1.0594, "step": 2399 }, { "epoch": 0.64, "learning_rate": 6.092644614574359e-06, "loss": 1.0674, "step": 2400 }, { "epoch": 0.64, "learning_rate": 6.084707980602142e-06, "loss": 1.0723, "step": 2401 }, { "epoch": 0.64, "learning_rate": 6.076774258322611e-06, "loss": 1.074, "step": 2402 }, { "epoch": 0.64, "learning_rate": 6.06884345363585e-06, "loss": 1.0379, "step": 2403 }, { "epoch": 0.64, "learning_rate": 6.060915572439773e-06, "loss": 1.0724, "step": 2404 }, { "epoch": 0.64, "learning_rate": 6.052990620630127e-06, "loss": 1.0784, "step": 2405 }, { "epoch": 0.64, "learning_rate": 6.045068604100475e-06, "loss": 1.0589, "step": 2406 }, { "epoch": 0.64, "learning_rate": 6.037149528742193e-06, "loss": 1.0657, "step": 2407 }, { "epoch": 0.64, "learning_rate": 6.029233400444483e-06, "loss": 1.1098, "step": 2408 }, { "epoch": 0.64, "learning_rate": 6.021320225094343e-06, "loss": 1.0762, "step": 2409 }, { "epoch": 0.64, "learning_rate": 6.013410008576577e-06, "loss": 1.0648, "step": 2410 }, { "epoch": 0.64, "learning_rate": 6.005502756773795e-06, "loss": 1.0602, "step": 2411 }, { "epoch": 0.64, "learning_rate": 5.997598475566395e-06, "loss": 1.0297, "step": 2412 }, { "epoch": 0.64, "learning_rate": 5.989697170832572e-06, "loss": 1.0973, "step": 2413 }, { "epoch": 0.64, "learning_rate": 5.981798848448302e-06, "loss": 1.0446, "step": 2414 }, { "epoch": 0.64, "learning_rate": 5.973903514287344e-06, "loss": 1.0806, "step": 2415 }, { "epoch": 0.64, "learning_rate": 5.966011174221239e-06, "loss": 1.0313, "step": 2416 }, { "epoch": 0.64, "learning_rate": 5.9581218341192995e-06, "loss": 1.0644, "step": 2417 }, { "epoch": 0.64, "learning_rate": 5.950235499848604e-06, "loss": 1.0379, "step": 2418 }, { "epoch": 0.64, "learning_rate": 5.942352177273995e-06, "loss": 1.0586, "step": 2419 }, { "epoch": 0.64, "learning_rate": 5.934471872258085e-06, "loss": 1.0925, "step": 2420 }, { "epoch": 0.64, "learning_rate": 5.926594590661231e-06, "loss": 1.0814, "step": 2421 }, { "epoch": 0.64, "learning_rate": 5.918720338341544e-06, "loss": 1.0921, "step": 2422 }, { "epoch": 0.65, "learning_rate": 5.910849121154889e-06, "loss": 1.025, "step": 2423 }, { "epoch": 0.65, "learning_rate": 5.902980944954868e-06, "loss": 1.0741, "step": 2424 }, { "epoch": 0.65, "learning_rate": 5.895115815592819e-06, "loss": 1.0954, "step": 2425 }, { "epoch": 0.65, "learning_rate": 5.887253738917822e-06, "loss": 1.0306, "step": 2426 }, { "epoch": 0.65, "learning_rate": 5.879394720776682e-06, "loss": 1.0533, "step": 2427 }, { "epoch": 0.65, "learning_rate": 5.871538767013927e-06, "loss": 1.0227, "step": 2428 }, { "epoch": 0.65, "learning_rate": 5.863685883471812e-06, "loss": 1.096, "step": 2429 }, { "epoch": 0.65, "learning_rate": 5.855836075990305e-06, "loss": 1.0598, "step": 2430 }, { "epoch": 0.65, "learning_rate": 5.847989350407084e-06, "loss": 1.0688, "step": 2431 }, { "epoch": 0.65, "learning_rate": 5.840145712557546e-06, "loss": 1.074, "step": 2432 }, { "epoch": 0.65, "learning_rate": 5.832305168274777e-06, "loss": 1.0352, "step": 2433 }, { "epoch": 0.65, "learning_rate": 5.824467723389575e-06, "loss": 1.0754, "step": 2434 }, { "epoch": 0.65, "learning_rate": 5.816633383730421e-06, "loss": 1.0555, "step": 2435 }, { "epoch": 0.65, "learning_rate": 5.808802155123499e-06, "loss": 1.0516, "step": 2436 }, { "epoch": 0.65, "learning_rate": 5.800974043392674e-06, "loss": 1.0854, "step": 2437 }, { "epoch": 0.65, "learning_rate": 5.793149054359488e-06, "loss": 1.0679, "step": 2438 }, { "epoch": 0.65, "learning_rate": 5.7853271938431664e-06, "loss": 1.0867, "step": 2439 }, { "epoch": 0.65, "learning_rate": 5.777508467660615e-06, "loss": 1.0846, "step": 2440 }, { "epoch": 0.65, "learning_rate": 5.76969288162639e-06, "loss": 1.0654, "step": 2441 }, { "epoch": 0.65, "learning_rate": 5.7618804415527275e-06, "loss": 1.0393, "step": 2442 }, { "epoch": 0.65, "learning_rate": 5.754071153249523e-06, "loss": 1.0511, "step": 2443 }, { "epoch": 0.65, "learning_rate": 5.74626502252432e-06, "loss": 1.0552, "step": 2444 }, { "epoch": 0.65, "learning_rate": 5.738462055182319e-06, "loss": 1.0627, "step": 2445 }, { "epoch": 0.65, "learning_rate": 5.730662257026374e-06, "loss": 1.0946, "step": 2446 }, { "epoch": 0.65, "learning_rate": 5.722865633856967e-06, "loss": 1.0618, "step": 2447 }, { "epoch": 0.65, "learning_rate": 5.71507219147223e-06, "loss": 1.0958, "step": 2448 }, { "epoch": 0.65, "learning_rate": 5.707281935667934e-06, "loss": 1.123, "step": 2449 }, { "epoch": 0.65, "learning_rate": 5.699494872237463e-06, "loss": 1.0589, "step": 2450 }, { "epoch": 0.65, "learning_rate": 5.691711006971843e-06, "loss": 1.0304, "step": 2451 }, { "epoch": 0.65, "learning_rate": 5.68393034565972e-06, "loss": 1.0493, "step": 2452 }, { "epoch": 0.65, "learning_rate": 5.676152894087343e-06, "loss": 1.0718, "step": 2453 }, { "epoch": 0.65, "learning_rate": 5.668378658038591e-06, "loss": 1.0399, "step": 2454 }, { "epoch": 0.65, "learning_rate": 5.660607643294948e-06, "loss": 1.0539, "step": 2455 }, { "epoch": 0.65, "learning_rate": 5.652839855635491e-06, "loss": 1.0842, "step": 2456 }, { "epoch": 0.65, "learning_rate": 5.6450753008369085e-06, "loss": 1.0736, "step": 2457 }, { "epoch": 0.65, "learning_rate": 5.637313984673488e-06, "loss": 1.0802, "step": 2458 }, { "epoch": 0.65, "learning_rate": 5.629555912917093e-06, "loss": 1.0314, "step": 2459 }, { "epoch": 0.65, "learning_rate": 5.621801091337187e-06, "loss": 1.05, "step": 2460 }, { "epoch": 0.66, "learning_rate": 5.614049525700818e-06, "loss": 1.0605, "step": 2461 }, { "epoch": 0.66, "learning_rate": 5.606301221772598e-06, "loss": 1.0489, "step": 2462 }, { "epoch": 0.66, "learning_rate": 5.598556185314731e-06, "loss": 1.0587, "step": 2463 }, { "epoch": 0.66, "learning_rate": 5.590814422086983e-06, "loss": 1.044, "step": 2464 }, { "epoch": 0.66, "learning_rate": 5.583075937846678e-06, "loss": 1.0498, "step": 2465 }, { "epoch": 0.66, "learning_rate": 5.575340738348715e-06, "loss": 1.0615, "step": 2466 }, { "epoch": 0.66, "learning_rate": 5.5676088293455475e-06, "loss": 1.061, "step": 2467 }, { "epoch": 0.66, "learning_rate": 5.559880216587173e-06, "loss": 1.104, "step": 2468 }, { "epoch": 0.66, "learning_rate": 5.5521549058211436e-06, "loss": 1.0446, "step": 2469 }, { "epoch": 0.66, "learning_rate": 5.5444329027925644e-06, "loss": 1.0723, "step": 2470 }, { "epoch": 0.66, "learning_rate": 5.536714213244061e-06, "loss": 1.0739, "step": 2471 }, { "epoch": 0.66, "learning_rate": 5.528998842915812e-06, "loss": 1.0911, "step": 2472 }, { "epoch": 0.66, "learning_rate": 5.521286797545524e-06, "loss": 1.058, "step": 2473 }, { "epoch": 0.66, "learning_rate": 5.513578082868421e-06, "loss": 1.0435, "step": 2474 }, { "epoch": 0.66, "learning_rate": 5.505872704617264e-06, "loss": 1.0953, "step": 2475 }, { "epoch": 0.66, "learning_rate": 5.498170668522325e-06, "loss": 1.0726, "step": 2476 }, { "epoch": 0.66, "learning_rate": 5.4904719803113886e-06, "loss": 1.1326, "step": 2477 }, { "epoch": 0.66, "learning_rate": 5.4827766457097545e-06, "loss": 1.068, "step": 2478 }, { "epoch": 0.66, "learning_rate": 5.4750846704402276e-06, "loss": 1.0543, "step": 2479 }, { "epoch": 0.66, "learning_rate": 5.467396060223118e-06, "loss": 1.106, "step": 2480 }, { "epoch": 0.66, "learning_rate": 5.459710820776222e-06, "loss": 1.0479, "step": 2481 }, { "epoch": 0.66, "learning_rate": 5.452028957814838e-06, "loss": 1.0561, "step": 2482 }, { "epoch": 0.66, "learning_rate": 5.444350477051758e-06, "loss": 1.0481, "step": 2483 }, { "epoch": 0.66, "learning_rate": 5.436675384197242e-06, "loss": 1.0277, "step": 2484 }, { "epoch": 0.66, "learning_rate": 5.4290036849590465e-06, "loss": 1.075, "step": 2485 }, { "epoch": 0.66, "learning_rate": 5.421335385042403e-06, "loss": 1.0507, "step": 2486 }, { "epoch": 0.66, "learning_rate": 5.413670490150003e-06, "loss": 1.045, "step": 2487 }, { "epoch": 0.66, "learning_rate": 5.406009005982016e-06, "loss": 1.0371, "step": 2488 }, { "epoch": 0.66, "learning_rate": 5.398350938236078e-06, "loss": 1.074, "step": 2489 }, { "epoch": 0.66, "learning_rate": 5.390696292607269e-06, "loss": 1.0796, "step": 2490 }, { "epoch": 0.66, "learning_rate": 5.38304507478814e-06, "loss": 1.0487, "step": 2491 }, { "epoch": 0.66, "learning_rate": 5.375397290468689e-06, "loss": 1.0983, "step": 2492 }, { "epoch": 0.66, "learning_rate": 5.367752945336351e-06, "loss": 1.062, "step": 2493 }, { "epoch": 0.66, "learning_rate": 5.360112045076015e-06, "loss": 1.0938, "step": 2494 }, { "epoch": 0.66, "learning_rate": 5.35247459537001e-06, "loss": 1.0742, "step": 2495 }, { "epoch": 0.66, "learning_rate": 5.3448406018980805e-06, "loss": 1.0953, "step": 2496 }, { "epoch": 0.66, "learning_rate": 5.337210070337419e-06, "loss": 1.0787, "step": 2497 }, { "epoch": 0.67, "learning_rate": 5.329583006362641e-06, "loss": 1.051, "step": 2498 }, { "epoch": 0.67, "learning_rate": 5.321959415645772e-06, "loss": 1.0393, "step": 2499 }, { "epoch": 0.67, "learning_rate": 5.314339303856268e-06, "loss": 1.0282, "step": 2500 }, { "epoch": 0.67, "learning_rate": 5.306722676660993e-06, "loss": 1.1033, "step": 2501 }, { "epoch": 0.67, "learning_rate": 5.299109539724215e-06, "loss": 1.0303, "step": 2502 }, { "epoch": 0.67, "learning_rate": 5.291499898707608e-06, "loss": 1.0625, "step": 2503 }, { "epoch": 0.67, "learning_rate": 5.283893759270256e-06, "loss": 1.0678, "step": 2504 }, { "epoch": 0.67, "learning_rate": 5.276291127068624e-06, "loss": 1.0582, "step": 2505 }, { "epoch": 0.67, "learning_rate": 5.268692007756577e-06, "loss": 1.0809, "step": 2506 }, { "epoch": 0.67, "learning_rate": 5.261096406985374e-06, "loss": 1.0722, "step": 2507 }, { "epoch": 0.67, "learning_rate": 5.25350433040364e-06, "loss": 1.126, "step": 2508 }, { "epoch": 0.67, "learning_rate": 5.245915783657393e-06, "loss": 1.0931, "step": 2509 }, { "epoch": 0.67, "learning_rate": 5.238330772390025e-06, "loss": 1.0923, "step": 2510 }, { "epoch": 0.67, "learning_rate": 5.2307493022422905e-06, "loss": 1.1078, "step": 2511 }, { "epoch": 0.67, "learning_rate": 5.223171378852319e-06, "loss": 1.0527, "step": 2512 }, { "epoch": 0.67, "learning_rate": 5.215597007855601e-06, "loss": 1.0822, "step": 2513 }, { "epoch": 0.67, "learning_rate": 5.208026194884977e-06, "loss": 1.0505, "step": 2514 }, { "epoch": 0.67, "learning_rate": 5.200458945570653e-06, "loss": 1.0644, "step": 2515 }, { "epoch": 0.67, "learning_rate": 5.192895265540179e-06, "loss": 1.044, "step": 2516 }, { "epoch": 0.67, "learning_rate": 5.18533516041845e-06, "loss": 1.0787, "step": 2517 }, { "epoch": 0.67, "learning_rate": 5.177778635827711e-06, "loss": 1.0414, "step": 2518 }, { "epoch": 0.67, "learning_rate": 5.170225697387524e-06, "loss": 1.0824, "step": 2519 }, { "epoch": 0.67, "learning_rate": 5.162676350714806e-06, "loss": 1.0454, "step": 2520 }, { "epoch": 0.67, "learning_rate": 5.155130601423795e-06, "loss": 1.0895, "step": 2521 }, { "epoch": 0.67, "learning_rate": 5.147588455126047e-06, "loss": 1.1093, "step": 2522 }, { "epoch": 0.67, "learning_rate": 5.1400499174304465e-06, "loss": 1.0499, "step": 2523 }, { "epoch": 0.67, "learning_rate": 5.132514993943196e-06, "loss": 1.0664, "step": 2524 }, { "epoch": 0.67, "learning_rate": 5.124983690267802e-06, "loss": 1.0647, "step": 2525 }, { "epoch": 0.67, "learning_rate": 5.117456012005082e-06, "loss": 1.0374, "step": 2526 }, { "epoch": 0.67, "learning_rate": 5.109931964753168e-06, "loss": 1.0845, "step": 2527 }, { "epoch": 0.67, "learning_rate": 5.102411554107471e-06, "loss": 1.089, "step": 2528 }, { "epoch": 0.67, "learning_rate": 5.094894785660715e-06, "loss": 1.1, "step": 2529 }, { "epoch": 0.67, "learning_rate": 5.087381665002912e-06, "loss": 1.0519, "step": 2530 }, { "epoch": 0.67, "learning_rate": 5.079872197721352e-06, "loss": 1.0713, "step": 2531 }, { "epoch": 0.67, "learning_rate": 5.0723663894006185e-06, "loss": 1.0697, "step": 2532 }, { "epoch": 0.67, "learning_rate": 5.064864245622574e-06, "loss": 1.0476, "step": 2533 }, { "epoch": 0.67, "learning_rate": 5.057365771966345e-06, "loss": 1.0493, "step": 2534 }, { "epoch": 0.67, "learning_rate": 5.049870974008339e-06, "loss": 1.0602, "step": 2535 }, { "epoch": 0.68, "learning_rate": 5.04237985732223e-06, "loss": 1.0368, "step": 2536 }, { "epoch": 0.68, "learning_rate": 5.034892427478943e-06, "loss": 1.1157, "step": 2537 }, { "epoch": 0.68, "learning_rate": 5.027408690046675e-06, "loss": 1.0724, "step": 2538 }, { "epoch": 0.68, "learning_rate": 5.019928650590876e-06, "loss": 1.0239, "step": 2539 }, { "epoch": 0.68, "learning_rate": 5.01245231467423e-06, "loss": 1.0809, "step": 2540 }, { "epoch": 0.68, "learning_rate": 5.004979687856683e-06, "loss": 1.0948, "step": 2541 }, { "epoch": 0.68, "learning_rate": 4.997510775695423e-06, "loss": 1.0473, "step": 2542 }, { "epoch": 0.68, "learning_rate": 4.990045583744862e-06, "loss": 1.0372, "step": 2543 }, { "epoch": 0.68, "learning_rate": 4.982584117556656e-06, "loss": 1.0671, "step": 2544 }, { "epoch": 0.68, "learning_rate": 4.9751263826796925e-06, "loss": 1.039, "step": 2545 }, { "epoch": 0.68, "learning_rate": 4.967672384660071e-06, "loss": 1.0664, "step": 2546 }, { "epoch": 0.68, "learning_rate": 4.960222129041126e-06, "loss": 1.0897, "step": 2547 }, { "epoch": 0.68, "learning_rate": 4.9527756213634015e-06, "loss": 1.1104, "step": 2548 }, { "epoch": 0.68, "learning_rate": 4.945332867164661e-06, "loss": 1.0794, "step": 2549 }, { "epoch": 0.68, "learning_rate": 4.937893871979864e-06, "loss": 1.0645, "step": 2550 }, { "epoch": 0.68, "learning_rate": 4.9304586413411856e-06, "loss": 1.0736, "step": 2551 }, { "epoch": 0.68, "learning_rate": 4.923027180778003e-06, "loss": 1.0546, "step": 2552 }, { "epoch": 0.68, "learning_rate": 4.915599495816874e-06, "loss": 1.0945, "step": 2553 }, { "epoch": 0.68, "learning_rate": 4.908175591981566e-06, "loss": 1.1029, "step": 2554 }, { "epoch": 0.68, "learning_rate": 4.900755474793031e-06, "loss": 1.1039, "step": 2555 }, { "epoch": 0.68, "learning_rate": 4.893339149769393e-06, "loss": 1.0683, "step": 2556 }, { "epoch": 0.68, "learning_rate": 4.88592662242597e-06, "loss": 1.1067, "step": 2557 }, { "epoch": 0.68, "learning_rate": 4.878517898275251e-06, "loss": 1.0423, "step": 2558 }, { "epoch": 0.68, "learning_rate": 4.871112982826892e-06, "loss": 1.0864, "step": 2559 }, { "epoch": 0.68, "learning_rate": 4.863711881587724e-06, "loss": 1.0429, "step": 2560 }, { "epoch": 0.68, "learning_rate": 4.85631460006174e-06, "loss": 1.0656, "step": 2561 }, { "epoch": 0.68, "learning_rate": 4.848921143750083e-06, "loss": 1.0678, "step": 2562 }, { "epoch": 0.68, "learning_rate": 4.841531518151064e-06, "loss": 1.0422, "step": 2563 }, { "epoch": 0.68, "learning_rate": 4.834145728760143e-06, "loss": 1.0304, "step": 2564 }, { "epoch": 0.68, "learning_rate": 4.8267637810699165e-06, "loss": 1.0602, "step": 2565 }, { "epoch": 0.68, "learning_rate": 4.819385680570136e-06, "loss": 1.0561, "step": 2566 }, { "epoch": 0.68, "learning_rate": 4.812011432747689e-06, "loss": 1.0644, "step": 2567 }, { "epoch": 0.68, "learning_rate": 4.804641043086592e-06, "loss": 1.0189, "step": 2568 }, { "epoch": 0.68, "learning_rate": 4.797274517067999e-06, "loss": 1.0005, "step": 2569 }, { "epoch": 0.68, "learning_rate": 4.789911860170193e-06, "loss": 1.0526, "step": 2570 }, { "epoch": 0.68, "learning_rate": 4.782553077868567e-06, "loss": 1.1084, "step": 2571 }, { "epoch": 0.68, "learning_rate": 4.7751981756356435e-06, "loss": 1.0692, "step": 2572 }, { "epoch": 0.69, "learning_rate": 4.767847158941062e-06, "loss": 1.1071, "step": 2573 }, { "epoch": 0.69, "learning_rate": 4.760500033251559e-06, "loss": 1.0226, "step": 2574 }, { "epoch": 0.69, "learning_rate": 4.753156804030988e-06, "loss": 1.0811, "step": 2575 }, { "epoch": 0.69, "learning_rate": 4.745817476740308e-06, "loss": 1.0417, "step": 2576 }, { "epoch": 0.69, "learning_rate": 4.738482056837562e-06, "loss": 1.0555, "step": 2577 }, { "epoch": 0.69, "learning_rate": 4.731150549777901e-06, "loss": 1.0634, "step": 2578 }, { "epoch": 0.69, "learning_rate": 4.723822961013561e-06, "loss": 1.0461, "step": 2579 }, { "epoch": 0.69, "learning_rate": 4.71649929599386e-06, "loss": 1.0492, "step": 2580 }, { "epoch": 0.69, "learning_rate": 4.709179560165205e-06, "loss": 1.0266, "step": 2581 }, { "epoch": 0.69, "learning_rate": 4.70186375897108e-06, "loss": 1.0849, "step": 2582 }, { "epoch": 0.69, "learning_rate": 4.694551897852034e-06, "loss": 1.0285, "step": 2583 }, { "epoch": 0.69, "learning_rate": 4.687243982245698e-06, "loss": 1.0944, "step": 2584 }, { "epoch": 0.69, "learning_rate": 4.679940017586766e-06, "loss": 1.088, "step": 2585 }, { "epoch": 0.69, "learning_rate": 4.672640009306983e-06, "loss": 1.0765, "step": 2586 }, { "epoch": 0.69, "learning_rate": 4.665343962835165e-06, "loss": 1.0441, "step": 2587 }, { "epoch": 0.69, "learning_rate": 4.6580518835971815e-06, "loss": 1.0662, "step": 2588 }, { "epoch": 0.69, "learning_rate": 4.650763777015937e-06, "loss": 1.0628, "step": 2589 }, { "epoch": 0.69, "learning_rate": 4.6434796485113965e-06, "loss": 1.0368, "step": 2590 }, { "epoch": 0.69, "learning_rate": 4.636199503500566e-06, "loss": 1.0839, "step": 2591 }, { "epoch": 0.69, "learning_rate": 4.628923347397476e-06, "loss": 1.0479, "step": 2592 }, { "epoch": 0.69, "learning_rate": 4.621651185613205e-06, "loss": 1.0596, "step": 2593 }, { "epoch": 0.69, "learning_rate": 4.614383023555859e-06, "loss": 1.0417, "step": 2594 }, { "epoch": 0.69, "learning_rate": 4.607118866630557e-06, "loss": 1.0501, "step": 2595 }, { "epoch": 0.69, "learning_rate": 4.599858720239453e-06, "loss": 1.0514, "step": 2596 }, { "epoch": 0.69, "learning_rate": 4.592602589781717e-06, "loss": 1.0446, "step": 2597 }, { "epoch": 0.69, "learning_rate": 4.5853504806535225e-06, "loss": 1.0923, "step": 2598 }, { "epoch": 0.69, "learning_rate": 4.578102398248064e-06, "loss": 1.0826, "step": 2599 }, { "epoch": 0.69, "learning_rate": 4.570858347955538e-06, "loss": 1.0796, "step": 2600 }, { "epoch": 0.69, "learning_rate": 4.563618335163136e-06, "loss": 1.0256, "step": 2601 }, { "epoch": 0.69, "learning_rate": 4.556382365255056e-06, "loss": 1.0341, "step": 2602 }, { "epoch": 0.69, "learning_rate": 4.5491504436124865e-06, "loss": 1.0751, "step": 2603 }, { "epoch": 0.69, "learning_rate": 4.5419225756136e-06, "loss": 1.0363, "step": 2604 }, { "epoch": 0.69, "learning_rate": 4.534698766633562e-06, "loss": 1.0736, "step": 2605 }, { "epoch": 0.69, "learning_rate": 4.527479022044517e-06, "loss": 1.068, "step": 2606 }, { "epoch": 0.69, "learning_rate": 4.520263347215584e-06, "loss": 1.0564, "step": 2607 }, { "epoch": 0.69, "learning_rate": 4.513051747512858e-06, "loss": 1.0613, "step": 2608 }, { "epoch": 0.69, "learning_rate": 4.505844228299407e-06, "loss": 1.0815, "step": 2609 }, { "epoch": 0.69, "learning_rate": 4.498640794935254e-06, "loss": 1.0718, "step": 2610 }, { "epoch": 0.7, "learning_rate": 4.491441452777393e-06, "loss": 1.0665, "step": 2611 }, { "epoch": 0.7, "learning_rate": 4.484246207179777e-06, "loss": 1.0459, "step": 2612 }, { "epoch": 0.7, "learning_rate": 4.477055063493299e-06, "loss": 1.0507, "step": 2613 }, { "epoch": 0.7, "learning_rate": 4.4698680270658165e-06, "loss": 1.025, "step": 2614 }, { "epoch": 0.7, "learning_rate": 4.462685103242129e-06, "loss": 1.0861, "step": 2615 }, { "epoch": 0.7, "learning_rate": 4.455506297363969e-06, "loss": 1.0637, "step": 2616 }, { "epoch": 0.7, "learning_rate": 4.448331614770014e-06, "loss": 1.0702, "step": 2617 }, { "epoch": 0.7, "learning_rate": 4.441161060795879e-06, "loss": 1.0746, "step": 2618 }, { "epoch": 0.7, "learning_rate": 4.433994640774096e-06, "loss": 1.072, "step": 2619 }, { "epoch": 0.7, "learning_rate": 4.426832360034136e-06, "loss": 1.0332, "step": 2620 }, { "epoch": 0.7, "learning_rate": 4.419674223902386e-06, "loss": 1.0465, "step": 2621 }, { "epoch": 0.7, "learning_rate": 4.412520237702145e-06, "loss": 1.0653, "step": 2622 }, { "epoch": 0.7, "learning_rate": 4.405370406753636e-06, "loss": 1.0907, "step": 2623 }, { "epoch": 0.7, "learning_rate": 4.398224736373991e-06, "loss": 0.9964, "step": 2624 }, { "epoch": 0.7, "learning_rate": 4.391083231877236e-06, "loss": 1.0493, "step": 2625 }, { "epoch": 0.7, "learning_rate": 4.383945898574311e-06, "loss": 1.1131, "step": 2626 }, { "epoch": 0.7, "learning_rate": 4.3768127417730545e-06, "loss": 1.0945, "step": 2627 }, { "epoch": 0.7, "learning_rate": 4.3696837667781874e-06, "loss": 1.0923, "step": 2628 }, { "epoch": 0.7, "learning_rate": 4.362558978891332e-06, "loss": 1.0785, "step": 2629 }, { "epoch": 0.7, "learning_rate": 4.3554383834109955e-06, "loss": 1.0124, "step": 2630 }, { "epoch": 0.7, "learning_rate": 4.34832198563256e-06, "loss": 1.0491, "step": 2631 }, { "epoch": 0.7, "learning_rate": 4.341209790848291e-06, "loss": 1.099, "step": 2632 }, { "epoch": 0.7, "learning_rate": 4.334101804347333e-06, "loss": 1.0683, "step": 2633 }, { "epoch": 0.7, "learning_rate": 4.326998031415689e-06, "loss": 1.0497, "step": 2634 }, { "epoch": 0.7, "learning_rate": 4.319898477336238e-06, "loss": 1.0527, "step": 2635 }, { "epoch": 0.7, "learning_rate": 4.312803147388725e-06, "loss": 1.0561, "step": 2636 }, { "epoch": 0.7, "learning_rate": 4.3057120468497375e-06, "loss": 1.0928, "step": 2637 }, { "epoch": 0.7, "learning_rate": 4.298625180992731e-06, "loss": 1.0458, "step": 2638 }, { "epoch": 0.7, "learning_rate": 4.291542555088017e-06, "loss": 1.0309, "step": 2639 }, { "epoch": 0.7, "learning_rate": 4.284464174402731e-06, "loss": 1.0526, "step": 2640 }, { "epoch": 0.7, "learning_rate": 4.277390044200875e-06, "loss": 1.0311, "step": 2641 }, { "epoch": 0.7, "learning_rate": 4.270320169743283e-06, "loss": 1.0498, "step": 2642 }, { "epoch": 0.7, "learning_rate": 4.263254556287614e-06, "loss": 1.0861, "step": 2643 }, { "epoch": 0.7, "learning_rate": 4.25619320908837e-06, "loss": 1.0805, "step": 2644 }, { "epoch": 0.7, "learning_rate": 4.249136133396882e-06, "loss": 1.0395, "step": 2645 }, { "epoch": 0.7, "learning_rate": 4.2420833344612885e-06, "loss": 1.0961, "step": 2646 }, { "epoch": 0.7, "learning_rate": 4.2350348175265645e-06, "loss": 1.0647, "step": 2647 }, { "epoch": 0.71, "learning_rate": 4.227990587834495e-06, "loss": 1.0673, "step": 2648 }, { "epoch": 0.71, "learning_rate": 4.220950650623672e-06, "loss": 1.0793, "step": 2649 }, { "epoch": 0.71, "learning_rate": 4.213915011129499e-06, "loss": 1.0423, "step": 2650 }, { "epoch": 0.71, "learning_rate": 4.206883674584189e-06, "loss": 1.0441, "step": 2651 }, { "epoch": 0.71, "learning_rate": 4.199856646216742e-06, "loss": 1.0661, "step": 2652 }, { "epoch": 0.71, "learning_rate": 4.192833931252965e-06, "loss": 1.0795, "step": 2653 }, { "epoch": 0.71, "learning_rate": 4.185815534915458e-06, "loss": 1.0391, "step": 2654 }, { "epoch": 0.71, "learning_rate": 4.178801462423597e-06, "loss": 1.0252, "step": 2655 }, { "epoch": 0.71, "learning_rate": 4.171791718993554e-06, "loss": 1.0448, "step": 2656 }, { "epoch": 0.71, "learning_rate": 4.16478630983828e-06, "loss": 1.0639, "step": 2657 }, { "epoch": 0.71, "learning_rate": 4.157785240167504e-06, "loss": 1.0671, "step": 2658 }, { "epoch": 0.71, "learning_rate": 4.150788515187717e-06, "loss": 1.0835, "step": 2659 }, { "epoch": 0.71, "learning_rate": 4.1437961401021895e-06, "loss": 1.0341, "step": 2660 }, { "epoch": 0.71, "learning_rate": 4.136808120110961e-06, "loss": 1.0924, "step": 2661 }, { "epoch": 0.71, "learning_rate": 4.129824460410815e-06, "loss": 1.0396, "step": 2662 }, { "epoch": 0.71, "learning_rate": 4.122845166195309e-06, "loss": 1.0235, "step": 2663 }, { "epoch": 0.71, "learning_rate": 4.1158702426547515e-06, "loss": 1.0814, "step": 2664 }, { "epoch": 0.71, "learning_rate": 4.108899694976188e-06, "loss": 1.0898, "step": 2665 }, { "epoch": 0.71, "learning_rate": 4.101933528343426e-06, "loss": 1.0844, "step": 2666 }, { "epoch": 0.71, "learning_rate": 4.094971747937008e-06, "loss": 1.0505, "step": 2667 }, { "epoch": 0.71, "learning_rate": 4.088014358934209e-06, "loss": 1.085, "step": 2668 }, { "epoch": 0.71, "learning_rate": 4.081061366509047e-06, "loss": 1.062, "step": 2669 }, { "epoch": 0.71, "learning_rate": 4.074112775832271e-06, "loss": 1.0075, "step": 2670 }, { "epoch": 0.71, "learning_rate": 4.067168592071346e-06, "loss": 1.0572, "step": 2671 }, { "epoch": 0.71, "learning_rate": 4.060228820390468e-06, "loss": 1.0446, "step": 2672 }, { "epoch": 0.71, "learning_rate": 4.053293465950557e-06, "loss": 1.0774, "step": 2673 }, { "epoch": 0.71, "learning_rate": 4.046362533909232e-06, "loss": 1.0841, "step": 2674 }, { "epoch": 0.71, "learning_rate": 4.039436029420835e-06, "loss": 1.0729, "step": 2675 }, { "epoch": 0.71, "learning_rate": 4.032513957636418e-06, "loss": 1.0734, "step": 2676 }, { "epoch": 0.71, "learning_rate": 4.025596323703725e-06, "loss": 1.0248, "step": 2677 }, { "epoch": 0.71, "learning_rate": 4.018683132767207e-06, "loss": 1.0198, "step": 2678 }, { "epoch": 0.71, "learning_rate": 4.011774389968016e-06, "loss": 1.0328, "step": 2679 }, { "epoch": 0.71, "learning_rate": 4.004870100443978e-06, "loss": 1.0012, "step": 2680 }, { "epoch": 0.71, "learning_rate": 3.997970269329633e-06, "loss": 1.071, "step": 2681 }, { "epoch": 0.71, "learning_rate": 3.991074901756182e-06, "loss": 1.0647, "step": 2682 }, { "epoch": 0.71, "learning_rate": 3.984184002851519e-06, "loss": 1.091, "step": 2683 }, { "epoch": 0.71, "learning_rate": 3.977297577740215e-06, "loss": 1.0773, "step": 2684 }, { "epoch": 0.71, "learning_rate": 3.970415631543506e-06, "loss": 1.1237, "step": 2685 }, { "epoch": 0.72, "learning_rate": 3.963538169379303e-06, "loss": 1.0965, "step": 2686 }, { "epoch": 0.72, "learning_rate": 3.956665196362187e-06, "loss": 1.0952, "step": 2687 }, { "epoch": 0.72, "learning_rate": 3.949796717603386e-06, "loss": 1.0588, "step": 2688 }, { "epoch": 0.72, "learning_rate": 3.9429327382107985e-06, "loss": 1.0491, "step": 2689 }, { "epoch": 0.72, "learning_rate": 3.936073263288978e-06, "loss": 1.0824, "step": 2690 }, { "epoch": 0.72, "learning_rate": 3.929218297939116e-06, "loss": 1.0399, "step": 2691 }, { "epoch": 0.72, "learning_rate": 3.92236784725906e-06, "loss": 1.1202, "step": 2692 }, { "epoch": 0.72, "learning_rate": 3.915521916343302e-06, "loss": 1.0861, "step": 2693 }, { "epoch": 0.72, "learning_rate": 3.908680510282962e-06, "loss": 1.0864, "step": 2694 }, { "epoch": 0.72, "learning_rate": 3.901843634165804e-06, "loss": 1.0563, "step": 2695 }, { "epoch": 0.72, "learning_rate": 3.895011293076224e-06, "loss": 1.0568, "step": 2696 }, { "epoch": 0.72, "learning_rate": 3.888183492095236e-06, "loss": 1.0411, "step": 2697 }, { "epoch": 0.72, "learning_rate": 3.881360236300489e-06, "loss": 1.0234, "step": 2698 }, { "epoch": 0.72, "learning_rate": 3.874541530766246e-06, "loss": 1.0518, "step": 2699 }, { "epoch": 0.72, "learning_rate": 3.867727380563382e-06, "loss": 1.0276, "step": 2700 }, { "epoch": 0.72, "learning_rate": 3.860917790759393e-06, "loss": 1.0872, "step": 2701 }, { "epoch": 0.72, "learning_rate": 3.8541127664183835e-06, "loss": 1.0394, "step": 2702 }, { "epoch": 0.72, "learning_rate": 3.8473123126010505e-06, "loss": 1.0425, "step": 2703 }, { "epoch": 0.72, "learning_rate": 3.840516434364704e-06, "loss": 1.0892, "step": 2704 }, { "epoch": 0.72, "learning_rate": 3.833725136763252e-06, "loss": 1.0757, "step": 2705 }, { "epoch": 0.72, "learning_rate": 3.826938424847184e-06, "loss": 1.0606, "step": 2706 }, { "epoch": 0.72, "learning_rate": 3.820156303663592e-06, "loss": 1.0785, "step": 2707 }, { "epoch": 0.72, "learning_rate": 3.8133787782561505e-06, "loss": 1.0419, "step": 2708 }, { "epoch": 0.72, "learning_rate": 3.8066058536651094e-06, "loss": 1.0681, "step": 2709 }, { "epoch": 0.72, "learning_rate": 3.7998375349273054e-06, "loss": 1.0426, "step": 2710 }, { "epoch": 0.72, "learning_rate": 3.79307382707615e-06, "loss": 1.0516, "step": 2711 }, { "epoch": 0.72, "learning_rate": 3.786314735141616e-06, "loss": 1.0206, "step": 2712 }, { "epoch": 0.72, "learning_rate": 3.779560264150255e-06, "loss": 1.106, "step": 2713 }, { "epoch": 0.72, "learning_rate": 3.772810419125178e-06, "loss": 1.0327, "step": 2714 }, { "epoch": 0.72, "learning_rate": 3.7660652050860515e-06, "loss": 1.0615, "step": 2715 }, { "epoch": 0.72, "learning_rate": 3.7593246270491015e-06, "loss": 1.0938, "step": 2716 }, { "epoch": 0.72, "learning_rate": 3.7525886900271137e-06, "loss": 1.0042, "step": 2717 }, { "epoch": 0.72, "learning_rate": 3.7458573990294057e-06, "loss": 1.0758, "step": 2718 }, { "epoch": 0.72, "learning_rate": 3.739130759061853e-06, "loss": 1.1006, "step": 2719 }, { "epoch": 0.72, "learning_rate": 3.7324087751268743e-06, "loss": 1.0993, "step": 2720 }, { "epoch": 0.72, "learning_rate": 3.725691452223411e-06, "loss": 1.0836, "step": 2721 }, { "epoch": 0.72, "learning_rate": 3.7189787953469525e-06, "loss": 1.071, "step": 2722 }, { "epoch": 0.72, "learning_rate": 3.7122708094895168e-06, "loss": 1.0415, "step": 2723 }, { "epoch": 0.73, "learning_rate": 3.7055674996396353e-06, "loss": 1.0828, "step": 2724 }, { "epoch": 0.73, "learning_rate": 3.698868870782377e-06, "loss": 1.0451, "step": 2725 }, { "epoch": 0.73, "learning_rate": 3.692174927899327e-06, "loss": 1.0562, "step": 2726 }, { "epoch": 0.73, "learning_rate": 3.6854856759685753e-06, "loss": 1.0368, "step": 2727 }, { "epoch": 0.73, "learning_rate": 3.6788011199647334e-06, "loss": 1.0338, "step": 2728 }, { "epoch": 0.73, "learning_rate": 3.672121264858919e-06, "loss": 1.0366, "step": 2729 }, { "epoch": 0.73, "learning_rate": 3.665446115618756e-06, "loss": 1.0822, "step": 2730 }, { "epoch": 0.73, "learning_rate": 3.6587756772083584e-06, "loss": 1.0477, "step": 2731 }, { "epoch": 0.73, "learning_rate": 3.6521099545883453e-06, "loss": 1.0539, "step": 2732 }, { "epoch": 0.73, "learning_rate": 3.6454489527158343e-06, "loss": 1.0542, "step": 2733 }, { "epoch": 0.73, "learning_rate": 3.638792676544416e-06, "loss": 1.0447, "step": 2734 }, { "epoch": 0.73, "learning_rate": 3.6321411310241804e-06, "loss": 1.0325, "step": 2735 }, { "epoch": 0.73, "learning_rate": 3.625494321101699e-06, "loss": 1.0344, "step": 2736 }, { "epoch": 0.73, "learning_rate": 3.618852251720009e-06, "loss": 1.0313, "step": 2737 }, { "epoch": 0.73, "learning_rate": 3.6122149278186357e-06, "loss": 1.0688, "step": 2738 }, { "epoch": 0.73, "learning_rate": 3.6055823543335734e-06, "loss": 1.0941, "step": 2739 }, { "epoch": 0.73, "learning_rate": 3.598954536197273e-06, "loss": 1.0361, "step": 2740 }, { "epoch": 0.73, "learning_rate": 3.592331478338662e-06, "loss": 1.0685, "step": 2741 }, { "epoch": 0.73, "learning_rate": 3.585713185683124e-06, "loss": 1.0255, "step": 2742 }, { "epoch": 0.73, "learning_rate": 3.579099663152491e-06, "loss": 1.0657, "step": 2743 }, { "epoch": 0.73, "learning_rate": 3.572490915665059e-06, "loss": 1.0703, "step": 2744 }, { "epoch": 0.73, "learning_rate": 3.5658869481355686e-06, "loss": 1.0611, "step": 2745 }, { "epoch": 0.73, "learning_rate": 3.559287765475199e-06, "loss": 1.0627, "step": 2746 }, { "epoch": 0.73, "learning_rate": 3.5526933725915813e-06, "loss": 1.0837, "step": 2747 }, { "epoch": 0.73, "learning_rate": 3.546103774388784e-06, "loss": 1.0483, "step": 2748 }, { "epoch": 0.73, "learning_rate": 3.5395189757672976e-06, "loss": 1.0625, "step": 2749 }, { "epoch": 0.73, "learning_rate": 3.5329389816240557e-06, "loss": 1.074, "step": 2750 }, { "epoch": 0.73, "learning_rate": 3.526363796852419e-06, "loss": 1.0117, "step": 2751 }, { "epoch": 0.73, "learning_rate": 3.5197934263421617e-06, "loss": 1.0696, "step": 2752 }, { "epoch": 0.73, "learning_rate": 3.513227874979486e-06, "loss": 1.0431, "step": 2753 }, { "epoch": 0.73, "learning_rate": 3.5066671476470105e-06, "loss": 1.0564, "step": 2754 }, { "epoch": 0.73, "learning_rate": 3.5001112492237588e-06, "loss": 1.0817, "step": 2755 }, { "epoch": 0.73, "learning_rate": 3.493560184585171e-06, "loss": 1.0238, "step": 2756 }, { "epoch": 0.73, "learning_rate": 3.4870139586030937e-06, "loss": 1.0698, "step": 2757 }, { "epoch": 0.73, "learning_rate": 3.480472576145764e-06, "loss": 1.0248, "step": 2758 }, { "epoch": 0.73, "learning_rate": 3.473936042077828e-06, "loss": 1.0714, "step": 2759 }, { "epoch": 0.73, "learning_rate": 3.4674043612603246e-06, "loss": 1.0311, "step": 2760 }, { "epoch": 0.74, "learning_rate": 3.4608775385506767e-06, "loss": 1.1161, "step": 2761 }, { "epoch": 0.74, "learning_rate": 3.4543555788027004e-06, "loss": 1.0714, "step": 2762 }, { "epoch": 0.74, "learning_rate": 3.4478384868665993e-06, "loss": 1.0486, "step": 2763 }, { "epoch": 0.74, "learning_rate": 3.4413262675889438e-06, "loss": 1.0805, "step": 2764 }, { "epoch": 0.74, "learning_rate": 3.434818925812691e-06, "loss": 1.0289, "step": 2765 }, { "epoch": 0.74, "learning_rate": 3.4283164663771716e-06, "loss": 1.0004, "step": 2766 }, { "epoch": 0.74, "learning_rate": 3.4218188941180774e-06, "loss": 1.0589, "step": 2767 }, { "epoch": 0.74, "learning_rate": 3.4153262138674713e-06, "loss": 1.0766, "step": 2768 }, { "epoch": 0.74, "learning_rate": 3.4088384304537815e-06, "loss": 1.0265, "step": 2769 }, { "epoch": 0.74, "learning_rate": 3.4023555487017845e-06, "loss": 1.0726, "step": 2770 }, { "epoch": 0.74, "learning_rate": 3.395877573432619e-06, "loss": 1.0285, "step": 2771 }, { "epoch": 0.74, "learning_rate": 3.3894045094637775e-06, "loss": 1.0595, "step": 2772 }, { "epoch": 0.74, "learning_rate": 3.3829363616090894e-06, "loss": 1.0741, "step": 2773 }, { "epoch": 0.74, "learning_rate": 3.376473134678736e-06, "loss": 1.0584, "step": 2774 }, { "epoch": 0.74, "learning_rate": 3.370014833479244e-06, "loss": 1.0538, "step": 2775 }, { "epoch": 0.74, "learning_rate": 3.3635614628134617e-06, "loss": 1.0004, "step": 2776 }, { "epoch": 0.74, "learning_rate": 3.357113027480584e-06, "loss": 1.0668, "step": 2777 }, { "epoch": 0.74, "learning_rate": 3.3506695322761327e-06, "loss": 1.0635, "step": 2778 }, { "epoch": 0.74, "learning_rate": 3.3442309819919507e-06, "loss": 1.0492, "step": 2779 }, { "epoch": 0.74, "learning_rate": 3.3377973814162068e-06, "loss": 1.0534, "step": 2780 }, { "epoch": 0.74, "learning_rate": 3.3313687353333945e-06, "loss": 1.0512, "step": 2781 }, { "epoch": 0.74, "learning_rate": 3.3249450485243096e-06, "loss": 1.052, "step": 2782 }, { "epoch": 0.74, "learning_rate": 3.3185263257660716e-06, "loss": 1.078, "step": 2783 }, { "epoch": 0.74, "learning_rate": 3.312112571832108e-06, "loss": 1.0585, "step": 2784 }, { "epoch": 0.74, "learning_rate": 3.3057037914921387e-06, "loss": 1.0447, "step": 2785 }, { "epoch": 0.74, "learning_rate": 3.299299989512199e-06, "loss": 1.0461, "step": 2786 }, { "epoch": 0.74, "learning_rate": 3.2929011706546197e-06, "loss": 1.0442, "step": 2787 }, { "epoch": 0.74, "learning_rate": 3.2865073396780146e-06, "loss": 1.0699, "step": 2788 }, { "epoch": 0.74, "learning_rate": 3.2801185013373006e-06, "loss": 1.0744, "step": 2789 }, { "epoch": 0.74, "learning_rate": 3.2737346603836806e-06, "loss": 1.0358, "step": 2790 }, { "epoch": 0.74, "learning_rate": 3.2673558215646305e-06, "loss": 1.1005, "step": 2791 }, { "epoch": 0.74, "learning_rate": 3.260981989623917e-06, "loss": 1.0623, "step": 2792 }, { "epoch": 0.74, "learning_rate": 3.2546131693015835e-06, "loss": 1.0714, "step": 2793 }, { "epoch": 0.74, "learning_rate": 3.2482493653339363e-06, "loss": 1.015, "step": 2794 }, { "epoch": 0.74, "learning_rate": 3.2418905824535618e-06, "loss": 1.0365, "step": 2795 }, { "epoch": 0.74, "learning_rate": 3.235536825389308e-06, "loss": 1.0655, "step": 2796 }, { "epoch": 0.74, "learning_rate": 3.2291880988662836e-06, "loss": 1.0364, "step": 2797 }, { "epoch": 0.74, "learning_rate": 3.2228444076058574e-06, "loss": 1.0612, "step": 2798 }, { "epoch": 0.75, "learning_rate": 3.2165057563256606e-06, "loss": 1.0402, "step": 2799 }, { "epoch": 0.75, "learning_rate": 3.210172149739562e-06, "loss": 1.0373, "step": 2800 }, { "epoch": 0.75, "learning_rate": 3.2038435925576906e-06, "loss": 1.02, "step": 2801 }, { "epoch": 0.75, "learning_rate": 3.197520089486419e-06, "loss": 1.0031, "step": 2802 }, { "epoch": 0.75, "learning_rate": 3.1912016452283533e-06, "loss": 1.035, "step": 2803 }, { "epoch": 0.75, "learning_rate": 3.1848882644823463e-06, "loss": 1.0568, "step": 2804 }, { "epoch": 0.75, "learning_rate": 3.1785799519434856e-06, "loss": 1.0371, "step": 2805 }, { "epoch": 0.75, "learning_rate": 3.1722767123030785e-06, "loss": 1.0642, "step": 2806 }, { "epoch": 0.75, "learning_rate": 3.1659785502486696e-06, "loss": 1.047, "step": 2807 }, { "epoch": 0.75, "learning_rate": 3.15968547046403e-06, "loss": 1.0395, "step": 2808 }, { "epoch": 0.75, "learning_rate": 3.15339747762914e-06, "loss": 1.0656, "step": 2809 }, { "epoch": 0.75, "learning_rate": 3.147114576420205e-06, "loss": 1.0246, "step": 2810 }, { "epoch": 0.75, "learning_rate": 3.1408367715096467e-06, "loss": 1.0384, "step": 2811 }, { "epoch": 0.75, "learning_rate": 3.134564067566085e-06, "loss": 1.0518, "step": 2812 }, { "epoch": 0.75, "learning_rate": 3.128296469254357e-06, "loss": 1.1042, "step": 2813 }, { "epoch": 0.75, "learning_rate": 3.1220339812355025e-06, "loss": 1.102, "step": 2814 }, { "epoch": 0.75, "learning_rate": 3.115776608166753e-06, "loss": 1.027, "step": 2815 }, { "epoch": 0.75, "learning_rate": 3.109524354701542e-06, "loss": 1.0333, "step": 2816 }, { "epoch": 0.75, "learning_rate": 3.1032772254894994e-06, "loss": 1.0725, "step": 2817 }, { "epoch": 0.75, "learning_rate": 3.0970352251764322e-06, "loss": 1.0229, "step": 2818 }, { "epoch": 0.75, "learning_rate": 3.0907983584043455e-06, "loss": 1.0466, "step": 2819 }, { "epoch": 0.75, "learning_rate": 3.084566629811424e-06, "loss": 1.0259, "step": 2820 }, { "epoch": 0.75, "learning_rate": 3.078340044032023e-06, "loss": 1.0736, "step": 2821 }, { "epoch": 0.75, "learning_rate": 3.0721186056966823e-06, "loss": 1.0412, "step": 2822 }, { "epoch": 0.75, "learning_rate": 3.0659023194321145e-06, "loss": 1.0545, "step": 2823 }, { "epoch": 0.75, "learning_rate": 3.0596911898611905e-06, "loss": 1.0391, "step": 2824 }, { "epoch": 0.75, "learning_rate": 3.0534852216029555e-06, "loss": 1.0734, "step": 2825 }, { "epoch": 0.75, "learning_rate": 3.047284419272618e-06, "loss": 1.0604, "step": 2826 }, { "epoch": 0.75, "learning_rate": 3.041088787481534e-06, "loss": 1.0355, "step": 2827 }, { "epoch": 0.75, "learning_rate": 3.0348983308372226e-06, "loss": 1.0364, "step": 2828 }, { "epoch": 0.75, "learning_rate": 3.0287130539433572e-06, "loss": 1.0616, "step": 2829 }, { "epoch": 0.75, "learning_rate": 3.0225329613997478e-06, "loss": 1.0531, "step": 2830 }, { "epoch": 0.75, "learning_rate": 3.016358057802359e-06, "loss": 1.076, "step": 2831 }, { "epoch": 0.75, "learning_rate": 3.0101883477432957e-06, "loss": 1.0657, "step": 2832 }, { "epoch": 0.75, "learning_rate": 3.004023835810791e-06, "loss": 1.0438, "step": 2833 }, { "epoch": 0.75, "learning_rate": 2.9978645265892236e-06, "loss": 1.0849, "step": 2834 }, { "epoch": 0.75, "learning_rate": 2.9917104246591024e-06, "loss": 1.0814, "step": 2835 }, { "epoch": 0.76, "learning_rate": 2.9855615345970514e-06, "loss": 1.0631, "step": 2836 }, { "epoch": 0.76, "learning_rate": 2.979417860975833e-06, "loss": 1.0079, "step": 2837 }, { "epoch": 0.76, "learning_rate": 2.9732794083643235e-06, "loss": 1.0879, "step": 2838 }, { "epoch": 0.76, "learning_rate": 2.9671461813275206e-06, "loss": 1.0426, "step": 2839 }, { "epoch": 0.76, "learning_rate": 2.9610181844265273e-06, "loss": 1.027, "step": 2840 }, { "epoch": 0.76, "learning_rate": 2.9548954222185667e-06, "loss": 1.0888, "step": 2841 }, { "epoch": 0.76, "learning_rate": 2.948777899256966e-06, "loss": 1.0833, "step": 2842 }, { "epoch": 0.76, "learning_rate": 2.942665620091151e-06, "loss": 1.0711, "step": 2843 }, { "epoch": 0.76, "learning_rate": 2.9365585892666536e-06, "loss": 1.0324, "step": 2844 }, { "epoch": 0.76, "learning_rate": 2.930456811325105e-06, "loss": 1.0696, "step": 2845 }, { "epoch": 0.76, "learning_rate": 2.9243602908042168e-06, "loss": 1.0263, "step": 2846 }, { "epoch": 0.76, "learning_rate": 2.9182690322378117e-06, "loss": 1.0029, "step": 2847 }, { "epoch": 0.76, "learning_rate": 2.9121830401557784e-06, "loss": 1.058, "step": 2848 }, { "epoch": 0.76, "learning_rate": 2.906102319084102e-06, "loss": 1.0705, "step": 2849 }, { "epoch": 0.76, "learning_rate": 2.900026873544847e-06, "loss": 1.0376, "step": 2850 }, { "epoch": 0.76, "learning_rate": 2.893956708056145e-06, "loss": 1.0689, "step": 2851 }, { "epoch": 0.76, "learning_rate": 2.887891827132211e-06, "loss": 1.0303, "step": 2852 }, { "epoch": 0.76, "learning_rate": 2.8818322352833285e-06, "loss": 1.056, "step": 2853 }, { "epoch": 0.76, "learning_rate": 2.875777937015841e-06, "loss": 1.0447, "step": 2854 }, { "epoch": 0.76, "learning_rate": 2.869728936832162e-06, "loss": 1.0893, "step": 2855 }, { "epoch": 0.76, "learning_rate": 2.863685239230768e-06, "loss": 1.033, "step": 2856 }, { "epoch": 0.76, "learning_rate": 2.8576468487061782e-06, "loss": 1.0342, "step": 2857 }, { "epoch": 0.76, "learning_rate": 2.85161376974898e-06, "loss": 1.0367, "step": 2858 }, { "epoch": 0.76, "learning_rate": 2.8455860068458054e-06, "loss": 1.0658, "step": 2859 }, { "epoch": 0.76, "learning_rate": 2.839563564479326e-06, "loss": 1.02, "step": 2860 }, { "epoch": 0.76, "learning_rate": 2.833546447128268e-06, "loss": 1.014, "step": 2861 }, { "epoch": 0.76, "learning_rate": 2.8275346592673957e-06, "loss": 0.995, "step": 2862 }, { "epoch": 0.76, "learning_rate": 2.8215282053674985e-06, "loss": 1.0449, "step": 2863 }, { "epoch": 0.76, "learning_rate": 2.8155270898954124e-06, "loss": 1.0314, "step": 2864 }, { "epoch": 0.76, "learning_rate": 2.8095313173140006e-06, "loss": 1.0242, "step": 2865 }, { "epoch": 0.76, "learning_rate": 2.803540892082145e-06, "loss": 1.0399, "step": 2866 }, { "epoch": 0.76, "learning_rate": 2.797555818654759e-06, "loss": 0.9984, "step": 2867 }, { "epoch": 0.76, "learning_rate": 2.791576101482778e-06, "loss": 1.0642, "step": 2868 }, { "epoch": 0.76, "learning_rate": 2.7856017450131425e-06, "loss": 1.021, "step": 2869 }, { "epoch": 0.76, "learning_rate": 2.779632753688816e-06, "loss": 1.066, "step": 2870 }, { "epoch": 0.76, "learning_rate": 2.7736691319487728e-06, "loss": 1.0563, "step": 2871 }, { "epoch": 0.76, "learning_rate": 2.7677108842279877e-06, "loss": 1.0479, "step": 2872 }, { "epoch": 0.76, "learning_rate": 2.7617580149574406e-06, "loss": 1.062, "step": 2873 }, { "epoch": 0.77, "learning_rate": 2.75581052856412e-06, "loss": 1.0316, "step": 2874 }, { "epoch": 0.77, "learning_rate": 2.7498684294709963e-06, "loss": 0.9987, "step": 2875 }, { "epoch": 0.77, "learning_rate": 2.743931722097045e-06, "loss": 1.0388, "step": 2876 }, { "epoch": 0.77, "learning_rate": 2.7380004108572324e-06, "loss": 1.0324, "step": 2877 }, { "epoch": 0.77, "learning_rate": 2.7320745001625015e-06, "loss": 1.0733, "step": 2878 }, { "epoch": 0.77, "learning_rate": 2.726153994419788e-06, "loss": 1.0518, "step": 2879 }, { "epoch": 0.77, "learning_rate": 2.7202388980320103e-06, "loss": 1.0702, "step": 2880 }, { "epoch": 0.77, "learning_rate": 2.7143292153980515e-06, "loss": 1.0901, "step": 2881 }, { "epoch": 0.77, "learning_rate": 2.708424950912781e-06, "loss": 1.0541, "step": 2882 }, { "epoch": 0.77, "learning_rate": 2.7025261089670374e-06, "loss": 1.0716, "step": 2883 }, { "epoch": 0.77, "learning_rate": 2.696632693947616e-06, "loss": 1.0405, "step": 2884 }, { "epoch": 0.77, "learning_rate": 2.6907447102372888e-06, "loss": 1.0471, "step": 2885 }, { "epoch": 0.77, "learning_rate": 2.6848621622147865e-06, "loss": 1.0147, "step": 2886 }, { "epoch": 0.77, "learning_rate": 2.678985054254789e-06, "loss": 1.0539, "step": 2887 }, { "epoch": 0.77, "learning_rate": 2.6731133907279384e-06, "loss": 1.061, "step": 2888 }, { "epoch": 0.77, "learning_rate": 2.667247176000829e-06, "loss": 1.0363, "step": 2889 }, { "epoch": 0.77, "learning_rate": 2.661386414435996e-06, "loss": 1.0105, "step": 2890 }, { "epoch": 0.77, "learning_rate": 2.6555311103919236e-06, "loss": 1.0354, "step": 2891 }, { "epoch": 0.77, "learning_rate": 2.64968126822304e-06, "loss": 1.0487, "step": 2892 }, { "epoch": 0.77, "learning_rate": 2.6438368922797042e-06, "loss": 1.0563, "step": 2893 }, { "epoch": 0.77, "learning_rate": 2.637997986908217e-06, "loss": 1.059, "step": 2894 }, { "epoch": 0.77, "learning_rate": 2.63216455645081e-06, "loss": 0.9959, "step": 2895 }, { "epoch": 0.77, "learning_rate": 2.6263366052456373e-06, "loss": 1.0625, "step": 2896 }, { "epoch": 0.77, "learning_rate": 2.620514137626784e-06, "loss": 1.072, "step": 2897 }, { "epoch": 0.77, "learning_rate": 2.61469715792426e-06, "loss": 1.0505, "step": 2898 }, { "epoch": 0.77, "learning_rate": 2.608885670463982e-06, "loss": 1.0552, "step": 2899 }, { "epoch": 0.77, "learning_rate": 2.6030796795677948e-06, "loss": 1.0836, "step": 2900 }, { "epoch": 0.77, "learning_rate": 2.597279189553452e-06, "loss": 1.0183, "step": 2901 }, { "epoch": 0.77, "learning_rate": 2.5914842047346113e-06, "loss": 1.0751, "step": 2902 }, { "epoch": 0.77, "learning_rate": 2.58569472942084e-06, "loss": 1.0605, "step": 2903 }, { "epoch": 0.77, "learning_rate": 2.5799107679176148e-06, "loss": 1.0405, "step": 2904 }, { "epoch": 0.77, "learning_rate": 2.5741323245262963e-06, "loss": 1.071, "step": 2905 }, { "epoch": 0.77, "learning_rate": 2.568359403544155e-06, "loss": 1.0438, "step": 2906 }, { "epoch": 0.77, "learning_rate": 2.562592009264353e-06, "loss": 1.0301, "step": 2907 }, { "epoch": 0.77, "learning_rate": 2.556830145975934e-06, "loss": 1.0337, "step": 2908 }, { "epoch": 0.77, "learning_rate": 2.551073817963835e-06, "loss": 1.0321, "step": 2909 }, { "epoch": 0.77, "learning_rate": 2.545323029508877e-06, "loss": 1.0411, "step": 2910 }, { "epoch": 0.78, "learning_rate": 2.539577784887762e-06, "loss": 1.0168, "step": 2911 }, { "epoch": 0.78, "learning_rate": 2.533838088373062e-06, "loss": 1.023, "step": 2912 }, { "epoch": 0.78, "learning_rate": 2.52810394423323e-06, "loss": 1.0382, "step": 2913 }, { "epoch": 0.78, "learning_rate": 2.5223753567325926e-06, "loss": 1.0506, "step": 2914 }, { "epoch": 0.78, "learning_rate": 2.5166523301313304e-06, "loss": 1.0013, "step": 2915 }, { "epoch": 0.78, "learning_rate": 2.5109348686855027e-06, "loss": 1.0431, "step": 2916 }, { "epoch": 0.78, "learning_rate": 2.5052229766470282e-06, "loss": 1.0748, "step": 2917 }, { "epoch": 0.78, "learning_rate": 2.499516658263674e-06, "loss": 1.0343, "step": 2918 }, { "epoch": 0.78, "learning_rate": 2.493815917779071e-06, "loss": 1.0357, "step": 2919 }, { "epoch": 0.78, "learning_rate": 2.4881207594327038e-06, "loss": 1.0483, "step": 2920 }, { "epoch": 0.78, "learning_rate": 2.482431187459895e-06, "loss": 1.065, "step": 2921 }, { "epoch": 0.78, "learning_rate": 2.476747206091823e-06, "loss": 1.0161, "step": 2922 }, { "epoch": 0.78, "learning_rate": 2.4710688195555067e-06, "loss": 1.0234, "step": 2923 }, { "epoch": 0.78, "learning_rate": 2.465396032073799e-06, "loss": 1.0435, "step": 2924 }, { "epoch": 0.78, "learning_rate": 2.4597288478653937e-06, "loss": 1.0639, "step": 2925 }, { "epoch": 0.78, "learning_rate": 2.45406727114482e-06, "loss": 1.0193, "step": 2926 }, { "epoch": 0.78, "learning_rate": 2.448411306122428e-06, "loss": 1.0685, "step": 2927 }, { "epoch": 0.78, "learning_rate": 2.442760957004402e-06, "loss": 1.0327, "step": 2928 }, { "epoch": 0.78, "learning_rate": 2.437116227992752e-06, "loss": 1.0459, "step": 2929 }, { "epoch": 0.78, "learning_rate": 2.4314771232852973e-06, "loss": 1.0246, "step": 2930 }, { "epoch": 0.78, "learning_rate": 2.425843647075685e-06, "loss": 1.0231, "step": 2931 }, { "epoch": 0.78, "learning_rate": 2.420215803553377e-06, "loss": 1.0634, "step": 2932 }, { "epoch": 0.78, "learning_rate": 2.414593596903636e-06, "loss": 1.0053, "step": 2933 }, { "epoch": 0.78, "learning_rate": 2.408977031307541e-06, "loss": 1.0524, "step": 2934 }, { "epoch": 0.78, "learning_rate": 2.403366110941978e-06, "loss": 1.1044, "step": 2935 }, { "epoch": 0.78, "learning_rate": 2.397760839979625e-06, "loss": 1.0579, "step": 2936 }, { "epoch": 0.78, "learning_rate": 2.3921612225889677e-06, "loss": 1.0616, "step": 2937 }, { "epoch": 0.78, "learning_rate": 2.386567262934286e-06, "loss": 1.0608, "step": 2938 }, { "epoch": 0.78, "learning_rate": 2.3809789651756454e-06, "loss": 1.047, "step": 2939 }, { "epoch": 0.78, "learning_rate": 2.37539633346891e-06, "loss": 1.0443, "step": 2940 }, { "epoch": 0.78, "learning_rate": 2.369819371965727e-06, "loss": 1.0167, "step": 2941 }, { "epoch": 0.78, "learning_rate": 2.364248084813522e-06, "loss": 1.0691, "step": 2942 }, { "epoch": 0.78, "learning_rate": 2.358682476155506e-06, "loss": 1.0302, "step": 2943 }, { "epoch": 0.78, "learning_rate": 2.35312255013067e-06, "loss": 1.0568, "step": 2944 }, { "epoch": 0.78, "learning_rate": 2.347568310873769e-06, "loss": 1.0336, "step": 2945 }, { "epoch": 0.78, "learning_rate": 2.342019762515336e-06, "loss": 1.0273, "step": 2946 }, { "epoch": 0.78, "learning_rate": 2.3364769091816754e-06, "loss": 1.0467, "step": 2947 }, { "epoch": 0.78, "learning_rate": 2.3309397549948453e-06, "loss": 1.1072, "step": 2948 }, { "epoch": 0.79, "learning_rate": 2.3254083040726727e-06, "loss": 1.0808, "step": 2949 }, { "epoch": 0.79, "learning_rate": 2.319882560528748e-06, "loss": 1.057, "step": 2950 }, { "epoch": 0.79, "learning_rate": 2.3143625284724037e-06, "loss": 1.0562, "step": 2951 }, { "epoch": 0.79, "learning_rate": 2.3088482120087365e-06, "loss": 1.0217, "step": 2952 }, { "epoch": 0.79, "learning_rate": 2.3033396152385912e-06, "loss": 1.0504, "step": 2953 }, { "epoch": 0.79, "learning_rate": 2.29783674225855e-06, "loss": 1.0509, "step": 2954 }, { "epoch": 0.79, "learning_rate": 2.2923395971609486e-06, "loss": 1.0358, "step": 2955 }, { "epoch": 0.79, "learning_rate": 2.286848184033863e-06, "loss": 1.0527, "step": 2956 }, { "epoch": 0.79, "learning_rate": 2.2813625069610944e-06, "loss": 1.0213, "step": 2957 }, { "epoch": 0.79, "learning_rate": 2.2758825700221933e-06, "loss": 1.0645, "step": 2958 }, { "epoch": 0.79, "learning_rate": 2.2704083772924344e-06, "loss": 1.0541, "step": 2959 }, { "epoch": 0.79, "learning_rate": 2.264939932842818e-06, "loss": 1.0895, "step": 2960 }, { "epoch": 0.79, "learning_rate": 2.259477240740073e-06, "loss": 1.0365, "step": 2961 }, { "epoch": 0.79, "learning_rate": 2.254020305046655e-06, "loss": 1.0372, "step": 2962 }, { "epoch": 0.79, "learning_rate": 2.2485691298207267e-06, "loss": 1.0543, "step": 2963 }, { "epoch": 0.79, "learning_rate": 2.243123719116178e-06, "loss": 1.0229, "step": 2964 }, { "epoch": 0.79, "learning_rate": 2.237684076982609e-06, "loss": 1.0058, "step": 2965 }, { "epoch": 0.79, "learning_rate": 2.2322502074653253e-06, "loss": 1.0602, "step": 2966 }, { "epoch": 0.79, "learning_rate": 2.2268221146053435e-06, "loss": 1.0555, "step": 2967 }, { "epoch": 0.79, "learning_rate": 2.221399802439388e-06, "loss": 1.0162, "step": 2968 }, { "epoch": 0.79, "learning_rate": 2.2159832749998743e-06, "loss": 1.0709, "step": 2969 }, { "epoch": 0.79, "learning_rate": 2.2105725363149255e-06, "loss": 1.0727, "step": 2970 }, { "epoch": 0.79, "learning_rate": 2.205167590408356e-06, "loss": 1.0419, "step": 2971 }, { "epoch": 0.79, "learning_rate": 2.199768441299669e-06, "loss": 1.0695, "step": 2972 }, { "epoch": 0.79, "learning_rate": 2.194375093004062e-06, "loss": 1.0565, "step": 2973 }, { "epoch": 0.79, "learning_rate": 2.188987549532419e-06, "loss": 1.0309, "step": 2974 }, { "epoch": 0.79, "learning_rate": 2.1836058148912996e-06, "loss": 1.0461, "step": 2975 }, { "epoch": 0.79, "learning_rate": 2.1782298930829517e-06, "loss": 1.0583, "step": 2976 }, { "epoch": 0.79, "learning_rate": 2.172859788105299e-06, "loss": 1.0739, "step": 2977 }, { "epoch": 0.79, "learning_rate": 2.1674955039519353e-06, "loss": 1.0031, "step": 2978 }, { "epoch": 0.79, "learning_rate": 2.162137044612126e-06, "loss": 1.0502, "step": 2979 }, { "epoch": 0.79, "learning_rate": 2.156784414070814e-06, "loss": 1.0769, "step": 2980 }, { "epoch": 0.79, "learning_rate": 2.1514376163085914e-06, "loss": 1.0376, "step": 2981 }, { "epoch": 0.79, "learning_rate": 2.1460966553017258e-06, "loss": 1.0462, "step": 2982 }, { "epoch": 0.79, "learning_rate": 2.1407615350221434e-06, "loss": 1.0462, "step": 2983 }, { "epoch": 0.79, "learning_rate": 2.1354322594374156e-06, "loss": 1.0144, "step": 2984 }, { "epoch": 0.79, "learning_rate": 2.13010883251078e-06, "loss": 1.0065, "step": 2985 }, { "epoch": 0.79, "learning_rate": 2.124791258201121e-06, "loss": 1.046, "step": 2986 }, { "epoch": 0.8, "learning_rate": 2.1194795404629633e-06, "loss": 1.0464, "step": 2987 }, { "epoch": 0.8, "learning_rate": 2.1141736832464865e-06, "loss": 1.0249, "step": 2988 }, { "epoch": 0.8, "learning_rate": 2.1088736904975094e-06, "loss": 1.032, "step": 2989 }, { "epoch": 0.8, "learning_rate": 2.103579566157482e-06, "loss": 1.007, "step": 2990 }, { "epoch": 0.8, "learning_rate": 2.0982913141634985e-06, "loss": 1.011, "step": 2991 }, { "epoch": 0.8, "learning_rate": 2.0930089384482864e-06, "loss": 1.0666, "step": 2992 }, { "epoch": 0.8, "learning_rate": 2.0877324429401957e-06, "loss": 1.0208, "step": 2993 }, { "epoch": 0.8, "learning_rate": 2.0824618315632097e-06, "loss": 1.0614, "step": 2994 }, { "epoch": 0.8, "learning_rate": 2.0771971082369357e-06, "loss": 1.0112, "step": 2995 }, { "epoch": 0.8, "learning_rate": 2.071938276876596e-06, "loss": 1.0627, "step": 2996 }, { "epoch": 0.8, "learning_rate": 2.0666853413930387e-06, "loss": 1.0789, "step": 2997 }, { "epoch": 0.8, "learning_rate": 2.061438305692728e-06, "loss": 1.0611, "step": 2998 }, { "epoch": 0.8, "learning_rate": 2.05619717367773e-06, "loss": 1.0542, "step": 2999 }, { "epoch": 0.8, "learning_rate": 2.0509619492457288e-06, "loss": 1.0622, "step": 3000 }, { "epoch": 0.8, "learning_rate": 2.04573263629002e-06, "loss": 1.0146, "step": 3001 }, { "epoch": 0.8, "learning_rate": 2.0405092386994883e-06, "loss": 1.049, "step": 3002 }, { "epoch": 0.8, "learning_rate": 2.035291760358632e-06, "loss": 1.0334, "step": 3003 }, { "epoch": 0.8, "learning_rate": 2.0300802051475454e-06, "loss": 1.0413, "step": 3004 }, { "epoch": 0.8, "learning_rate": 2.0248745769419108e-06, "loss": 1.018, "step": 3005 }, { "epoch": 0.8, "learning_rate": 2.0196748796130095e-06, "loss": 1.0595, "step": 3006 }, { "epoch": 0.8, "learning_rate": 2.0144811170277136e-06, "loss": 1.0301, "step": 3007 }, { "epoch": 0.8, "learning_rate": 2.009293293048472e-06, "loss": 1.0382, "step": 3008 }, { "epoch": 0.8, "learning_rate": 2.0041114115333284e-06, "loss": 1.0485, "step": 3009 }, { "epoch": 0.8, "learning_rate": 1.9989354763359036e-06, "loss": 1.0491, "step": 3010 }, { "epoch": 0.8, "learning_rate": 1.9937654913053873e-06, "loss": 1.0354, "step": 3011 }, { "epoch": 0.8, "learning_rate": 1.98860146028656e-06, "loss": 1.0575, "step": 3012 }, { "epoch": 0.8, "learning_rate": 1.983443387119768e-06, "loss": 1.0606, "step": 3013 }, { "epoch": 0.8, "learning_rate": 1.978291275640919e-06, "loss": 1.0417, "step": 3014 }, { "epoch": 0.8, "learning_rate": 1.973145129681495e-06, "loss": 1.047, "step": 3015 }, { "epoch": 0.8, "learning_rate": 1.9680049530685454e-06, "loss": 1.0725, "step": 3016 }, { "epoch": 0.8, "learning_rate": 1.962870749624668e-06, "loss": 1.0732, "step": 3017 }, { "epoch": 0.8, "learning_rate": 1.957742523168028e-06, "loss": 1.0321, "step": 3018 }, { "epoch": 0.8, "learning_rate": 1.952620277512348e-06, "loss": 1.0524, "step": 3019 }, { "epoch": 0.8, "learning_rate": 1.947504016466891e-06, "loss": 1.0419, "step": 3020 }, { "epoch": 0.8, "learning_rate": 1.9423937438364803e-06, "loss": 1.0758, "step": 3021 }, { "epoch": 0.8, "learning_rate": 1.9372894634214845e-06, "loss": 1.0321, "step": 3022 }, { "epoch": 0.8, "learning_rate": 1.9321911790178082e-06, "loss": 1.0782, "step": 3023 }, { "epoch": 0.81, "learning_rate": 1.9270988944169055e-06, "loss": 1.0468, "step": 3024 }, { "epoch": 0.81, "learning_rate": 1.922012613405766e-06, "loss": 1.0718, "step": 3025 }, { "epoch": 0.81, "learning_rate": 1.9169323397669116e-06, "loss": 1.0154, "step": 3026 }, { "epoch": 0.81, "learning_rate": 1.911858077278399e-06, "loss": 1.041, "step": 3027 }, { "epoch": 0.81, "learning_rate": 1.9067898297138198e-06, "loss": 1.0316, "step": 3028 }, { "epoch": 0.81, "learning_rate": 1.9017276008422804e-06, "loss": 1.0483, "step": 3029 }, { "epoch": 0.81, "learning_rate": 1.896671394428421e-06, "loss": 1.0339, "step": 3030 }, { "epoch": 0.81, "learning_rate": 1.891621214232403e-06, "loss": 1.087, "step": 3031 }, { "epoch": 0.81, "learning_rate": 1.886577064009899e-06, "loss": 1.031, "step": 3032 }, { "epoch": 0.81, "learning_rate": 1.8815389475121037e-06, "loss": 1.0538, "step": 3033 }, { "epoch": 0.81, "learning_rate": 1.8765068684857256e-06, "loss": 1.0538, "step": 3034 }, { "epoch": 0.81, "learning_rate": 1.8714808306729758e-06, "loss": 1.0747, "step": 3035 }, { "epoch": 0.81, "learning_rate": 1.8664608378115789e-06, "loss": 1.0634, "step": 3036 }, { "epoch": 0.81, "learning_rate": 1.8614468936347663e-06, "loss": 1.0631, "step": 3037 }, { "epoch": 0.81, "learning_rate": 1.856439001871262e-06, "loss": 0.9801, "step": 3038 }, { "epoch": 0.81, "learning_rate": 1.8514371662452958e-06, "loss": 1.0463, "step": 3039 }, { "epoch": 0.81, "learning_rate": 1.8464413904765965e-06, "loss": 1.0607, "step": 3040 }, { "epoch": 0.81, "learning_rate": 1.8414516782803759e-06, "loss": 1.0959, "step": 3041 }, { "epoch": 0.81, "learning_rate": 1.8364680333673469e-06, "loss": 1.055, "step": 3042 }, { "epoch": 0.81, "learning_rate": 1.8314904594437065e-06, "loss": 1.0264, "step": 3043 }, { "epoch": 0.81, "learning_rate": 1.8265189602111344e-06, "loss": 0.985, "step": 3044 }, { "epoch": 0.81, "learning_rate": 1.8215535393667949e-06, "loss": 1.0118, "step": 3045 }, { "epoch": 0.81, "learning_rate": 1.8165942006033355e-06, "loss": 1.1048, "step": 3046 }, { "epoch": 0.81, "learning_rate": 1.8116409476088714e-06, "loss": 1.0326, "step": 3047 }, { "epoch": 0.81, "learning_rate": 1.8066937840670006e-06, "loss": 1.0559, "step": 3048 }, { "epoch": 0.81, "learning_rate": 1.8017527136567935e-06, "loss": 1.0254, "step": 3049 }, { "epoch": 0.81, "learning_rate": 1.7968177400527787e-06, "loss": 1.0439, "step": 3050 }, { "epoch": 0.81, "learning_rate": 1.7918888669249613e-06, "loss": 1.0951, "step": 3051 }, { "epoch": 0.81, "learning_rate": 1.786966097938806e-06, "loss": 1.0195, "step": 3052 }, { "epoch": 0.81, "learning_rate": 1.7820494367552353e-06, "loss": 0.9891, "step": 3053 }, { "epoch": 0.81, "learning_rate": 1.7771388870306328e-06, "loss": 1.0095, "step": 3054 }, { "epoch": 0.81, "learning_rate": 1.77223445241684e-06, "loss": 1.0461, "step": 3055 }, { "epoch": 0.81, "learning_rate": 1.7673361365611407e-06, "loss": 1.0541, "step": 3056 }, { "epoch": 0.81, "learning_rate": 1.762443943106279e-06, "loss": 1.0469, "step": 3057 }, { "epoch": 0.81, "learning_rate": 1.7575578756904431e-06, "loss": 1.0197, "step": 3058 }, { "epoch": 0.81, "learning_rate": 1.7526779379472603e-06, "loss": 1.0101, "step": 3059 }, { "epoch": 0.81, "learning_rate": 1.747804133505805e-06, "loss": 1.0481, "step": 3060 }, { "epoch": 0.81, "learning_rate": 1.7429364659905901e-06, "loss": 1.052, "step": 3061 }, { "epoch": 0.82, "learning_rate": 1.738074939021559e-06, "loss": 1.067, "step": 3062 }, { "epoch": 0.82, "learning_rate": 1.7332195562140962e-06, "loss": 1.0521, "step": 3063 }, { "epoch": 0.82, "learning_rate": 1.7283703211790137e-06, "loss": 1.0463, "step": 3064 }, { "epoch": 0.82, "learning_rate": 1.7235272375225498e-06, "loss": 1.0355, "step": 3065 }, { "epoch": 0.82, "learning_rate": 1.718690308846369e-06, "loss": 1.0816, "step": 3066 }, { "epoch": 0.82, "learning_rate": 1.713859538747563e-06, "loss": 1.0634, "step": 3067 }, { "epoch": 0.82, "learning_rate": 1.7090349308186348e-06, "loss": 1.0208, "step": 3068 }, { "epoch": 0.82, "learning_rate": 1.704216488647512e-06, "loss": 1.0271, "step": 3069 }, { "epoch": 0.82, "learning_rate": 1.6994042158175383e-06, "loss": 1.1009, "step": 3070 }, { "epoch": 0.82, "learning_rate": 1.6945981159074597e-06, "loss": 1.0785, "step": 3071 }, { "epoch": 0.82, "learning_rate": 1.6897981924914408e-06, "loss": 1.027, "step": 3072 }, { "epoch": 0.82, "learning_rate": 1.6850044491390516e-06, "loss": 1.0612, "step": 3073 }, { "epoch": 0.82, "learning_rate": 1.6802168894152604e-06, "loss": 1.0354, "step": 3074 }, { "epoch": 0.82, "learning_rate": 1.6754355168804426e-06, "loss": 0.9969, "step": 3075 }, { "epoch": 0.82, "learning_rate": 1.6706603350903726e-06, "loss": 1.0042, "step": 3076 }, { "epoch": 0.82, "learning_rate": 1.6658913475962147e-06, "loss": 1.0544, "step": 3077 }, { "epoch": 0.82, "learning_rate": 1.6611285579445334e-06, "loss": 1.0886, "step": 3078 }, { "epoch": 0.82, "learning_rate": 1.656371969677284e-06, "loss": 1.0471, "step": 3079 }, { "epoch": 0.82, "learning_rate": 1.651621586331803e-06, "loss": 1.044, "step": 3080 }, { "epoch": 0.82, "learning_rate": 1.6468774114408192e-06, "loss": 1.0349, "step": 3081 }, { "epoch": 0.82, "learning_rate": 1.6421394485324437e-06, "loss": 1.0509, "step": 3082 }, { "epoch": 0.82, "learning_rate": 1.6374077011301636e-06, "loss": 1.0338, "step": 3083 }, { "epoch": 0.82, "learning_rate": 1.6326821727528465e-06, "loss": 1.044, "step": 3084 }, { "epoch": 0.82, "learning_rate": 1.6279628669147396e-06, "loss": 1.0595, "step": 3085 }, { "epoch": 0.82, "learning_rate": 1.6232497871254516e-06, "loss": 1.0114, "step": 3086 }, { "epoch": 0.82, "learning_rate": 1.618542936889972e-06, "loss": 1.0359, "step": 3087 }, { "epoch": 0.82, "learning_rate": 1.6138423197086527e-06, "loss": 1.0235, "step": 3088 }, { "epoch": 0.82, "learning_rate": 1.609147939077208e-06, "loss": 1.073, "step": 3089 }, { "epoch": 0.82, "learning_rate": 1.6044597984867194e-06, "loss": 1.0254, "step": 3090 }, { "epoch": 0.82, "learning_rate": 1.5997779014236237e-06, "loss": 1.067, "step": 3091 }, { "epoch": 0.82, "learning_rate": 1.5951022513697201e-06, "loss": 1.0695, "step": 3092 }, { "epoch": 0.82, "learning_rate": 1.5904328518021517e-06, "loss": 1.0391, "step": 3093 }, { "epoch": 0.82, "learning_rate": 1.5857697061934252e-06, "loss": 1.0112, "step": 3094 }, { "epoch": 0.82, "learning_rate": 1.58111281801139e-06, "loss": 1.0337, "step": 3095 }, { "epoch": 0.82, "learning_rate": 1.5764621907192402e-06, "loss": 1.0487, "step": 3096 }, { "epoch": 0.82, "learning_rate": 1.5718178277755191e-06, "loss": 1.0861, "step": 3097 }, { "epoch": 0.82, "learning_rate": 1.5671797326341098e-06, "loss": 1.0452, "step": 3098 }, { "epoch": 0.83, "learning_rate": 1.562547908744231e-06, "loss": 1.038, "step": 3099 }, { "epoch": 0.83, "learning_rate": 1.55792235955044e-06, "loss": 1.0382, "step": 3100 }, { "epoch": 0.83, "learning_rate": 1.5533030884926314e-06, "loss": 1.044, "step": 3101 }, { "epoch": 0.83, "learning_rate": 1.5486900990060216e-06, "loss": 1.0215, "step": 3102 }, { "epoch": 0.83, "learning_rate": 1.5440833945211653e-06, "loss": 1.0326, "step": 3103 }, { "epoch": 0.83, "learning_rate": 1.5394829784639408e-06, "loss": 1.0719, "step": 3104 }, { "epoch": 0.83, "learning_rate": 1.534888854255543e-06, "loss": 1.0146, "step": 3105 }, { "epoch": 0.83, "learning_rate": 1.530301025312496e-06, "loss": 1.0597, "step": 3106 }, { "epoch": 0.83, "learning_rate": 1.5257194950466414e-06, "loss": 1.0371, "step": 3107 }, { "epoch": 0.83, "learning_rate": 1.52114426686513e-06, "loss": 0.9664, "step": 3108 }, { "epoch": 0.83, "learning_rate": 1.5165753441704323e-06, "loss": 1.0424, "step": 3109 }, { "epoch": 0.83, "learning_rate": 1.5120127303603305e-06, "loss": 1.0394, "step": 3110 }, { "epoch": 0.83, "learning_rate": 1.507456428827908e-06, "loss": 1.0044, "step": 3111 }, { "epoch": 0.83, "learning_rate": 1.5029064429615604e-06, "loss": 1.0272, "step": 3112 }, { "epoch": 0.83, "learning_rate": 1.498362776144987e-06, "loss": 1.057, "step": 3113 }, { "epoch": 0.83, "learning_rate": 1.49382543175718e-06, "loss": 0.9864, "step": 3114 }, { "epoch": 0.83, "learning_rate": 1.4892944131724363e-06, "loss": 1.0147, "step": 3115 }, { "epoch": 0.83, "learning_rate": 1.484769723760351e-06, "loss": 1.008, "step": 3116 }, { "epoch": 0.83, "learning_rate": 1.4802513668858031e-06, "loss": 1.0363, "step": 3117 }, { "epoch": 0.83, "learning_rate": 1.4757393459089708e-06, "loss": 0.9981, "step": 3118 }, { "epoch": 0.83, "learning_rate": 1.4712336641853187e-06, "loss": 1.0756, "step": 3119 }, { "epoch": 0.83, "learning_rate": 1.4667343250655919e-06, "loss": 1.0783, "step": 3120 }, { "epoch": 0.83, "learning_rate": 1.462241331895824e-06, "loss": 1.0224, "step": 3121 }, { "epoch": 0.83, "learning_rate": 1.457754688017331e-06, "loss": 1.0479, "step": 3122 }, { "epoch": 0.83, "learning_rate": 1.4532743967666984e-06, "loss": 1.0674, "step": 3123 }, { "epoch": 0.83, "learning_rate": 1.4488004614757956e-06, "loss": 1.01, "step": 3124 }, { "epoch": 0.83, "learning_rate": 1.4443328854717652e-06, "loss": 1.0524, "step": 3125 }, { "epoch": 0.83, "learning_rate": 1.4398716720770134e-06, "loss": 1.116, "step": 3126 }, { "epoch": 0.83, "learning_rate": 1.4354168246092203e-06, "loss": 1.0549, "step": 3127 }, { "epoch": 0.83, "learning_rate": 1.4309683463813352e-06, "loss": 1.0905, "step": 3128 }, { "epoch": 0.83, "learning_rate": 1.42652624070156e-06, "loss": 1.0529, "step": 3129 }, { "epoch": 0.83, "learning_rate": 1.4220905108733674e-06, "loss": 1.0929, "step": 3130 }, { "epoch": 0.83, "learning_rate": 1.417661160195486e-06, "loss": 1.1028, "step": 3131 }, { "epoch": 0.83, "learning_rate": 1.4132381919618966e-06, "loss": 1.0724, "step": 3132 }, { "epoch": 0.83, "learning_rate": 1.4088216094618378e-06, "loss": 1.0269, "step": 3133 }, { "epoch": 0.83, "learning_rate": 1.4044114159798006e-06, "loss": 1.0198, "step": 3134 }, { "epoch": 0.83, "learning_rate": 1.4000076147955166e-06, "loss": 1.0702, "step": 3135 }, { "epoch": 0.83, "learning_rate": 1.3956102091839719e-06, "loss": 1.0802, "step": 3136 }, { "epoch": 0.84, "learning_rate": 1.3912192024153948e-06, "loss": 1.0561, "step": 3137 }, { "epoch": 0.84, "learning_rate": 1.386834597755251e-06, "loss": 1.0035, "step": 3138 }, { "epoch": 0.84, "learning_rate": 1.3824563984642491e-06, "loss": 1.0899, "step": 3139 }, { "epoch": 0.84, "learning_rate": 1.3780846077983344e-06, "loss": 1.0169, "step": 3140 }, { "epoch": 0.84, "learning_rate": 1.3737192290086821e-06, "loss": 1.0353, "step": 3141 }, { "epoch": 0.84, "learning_rate": 1.3693602653417026e-06, "loss": 1.0459, "step": 3142 }, { "epoch": 0.84, "learning_rate": 1.3650077200390377e-06, "loss": 1.0121, "step": 3143 }, { "epoch": 0.84, "learning_rate": 1.3606615963375481e-06, "loss": 1.0269, "step": 3144 }, { "epoch": 0.84, "learning_rate": 1.3563218974693259e-06, "loss": 1.07, "step": 3145 }, { "epoch": 0.84, "learning_rate": 1.3519886266616866e-06, "loss": 1.0453, "step": 3146 }, { "epoch": 0.84, "learning_rate": 1.3476617871371556e-06, "loss": 1.079, "step": 3147 }, { "epoch": 0.84, "learning_rate": 1.3433413821134866e-06, "loss": 1.0624, "step": 3148 }, { "epoch": 0.84, "learning_rate": 1.339027414803643e-06, "loss": 1.0384, "step": 3149 }, { "epoch": 0.84, "learning_rate": 1.3347198884157996e-06, "loss": 1.0292, "step": 3150 }, { "epoch": 0.84, "learning_rate": 1.3304188061533418e-06, "loss": 1.0866, "step": 3151 }, { "epoch": 0.84, "learning_rate": 1.3261241712148688e-06, "loss": 1.0055, "step": 3152 }, { "epoch": 0.84, "learning_rate": 1.3218359867941733e-06, "loss": 1.045, "step": 3153 }, { "epoch": 0.84, "learning_rate": 1.3175542560802612e-06, "loss": 1.0676, "step": 3154 }, { "epoch": 0.84, "learning_rate": 1.3132789822573366e-06, "loss": 1.0268, "step": 3155 }, { "epoch": 0.84, "learning_rate": 1.3090101685047972e-06, "loss": 1.0107, "step": 3156 }, { "epoch": 0.84, "learning_rate": 1.3047478179972416e-06, "loss": 1.0355, "step": 3157 }, { "epoch": 0.84, "learning_rate": 1.3004919339044609e-06, "loss": 1.0457, "step": 3158 }, { "epoch": 0.84, "learning_rate": 1.2962425193914353e-06, "loss": 1.0278, "step": 3159 }, { "epoch": 0.84, "learning_rate": 1.2919995776183336e-06, "loss": 1.0705, "step": 3160 }, { "epoch": 0.84, "learning_rate": 1.2877631117405165e-06, "loss": 1.0371, "step": 3161 }, { "epoch": 0.84, "learning_rate": 1.2835331249085214e-06, "loss": 1.0643, "step": 3162 }, { "epoch": 0.84, "learning_rate": 1.27930962026807e-06, "loss": 1.0611, "step": 3163 }, { "epoch": 0.84, "learning_rate": 1.2750926009600684e-06, "loss": 1.0715, "step": 3164 }, { "epoch": 0.84, "learning_rate": 1.2708820701205903e-06, "loss": 1.0759, "step": 3165 }, { "epoch": 0.84, "learning_rate": 1.2666780308808925e-06, "loss": 1.0646, "step": 3166 }, { "epoch": 0.84, "learning_rate": 1.262480486367401e-06, "loss": 1.0146, "step": 3167 }, { "epoch": 0.84, "learning_rate": 1.2582894397017088e-06, "loss": 1.0092, "step": 3168 }, { "epoch": 0.84, "learning_rate": 1.2541048940005806e-06, "loss": 1.0643, "step": 3169 }, { "epoch": 0.84, "learning_rate": 1.2499268523759489e-06, "loss": 1.0993, "step": 3170 }, { "epoch": 0.84, "learning_rate": 1.2457553179349014e-06, "loss": 1.0305, "step": 3171 }, { "epoch": 0.84, "learning_rate": 1.2415902937796931e-06, "loss": 1.0331, "step": 3172 }, { "epoch": 0.84, "learning_rate": 1.2374317830077366e-06, "loss": 1.013, "step": 3173 }, { "epoch": 0.85, "learning_rate": 1.2332797887115967e-06, "loss": 1.0358, "step": 3174 }, { "epoch": 0.85, "learning_rate": 1.2291343139789957e-06, "loss": 1.0488, "step": 3175 }, { "epoch": 0.85, "learning_rate": 1.2249953618928101e-06, "loss": 1.0356, "step": 3176 }, { "epoch": 0.85, "learning_rate": 1.2208629355310553e-06, "loss": 1.0109, "step": 3177 }, { "epoch": 0.85, "learning_rate": 1.2167370379669075e-06, "loss": 0.9891, "step": 3178 }, { "epoch": 0.85, "learning_rate": 1.21261767226868e-06, "loss": 1.024, "step": 3179 }, { "epoch": 0.85, "learning_rate": 1.2085048414998257e-06, "loss": 1.0515, "step": 3180 }, { "epoch": 0.85, "learning_rate": 1.2043985487189424e-06, "loss": 1.0451, "step": 3181 }, { "epoch": 0.85, "learning_rate": 1.2002987969797686e-06, "loss": 1.0851, "step": 3182 }, { "epoch": 0.85, "learning_rate": 1.1962055893311697e-06, "loss": 1.0461, "step": 3183 }, { "epoch": 0.85, "learning_rate": 1.1921189288171497e-06, "loss": 1.0264, "step": 3184 }, { "epoch": 0.85, "learning_rate": 1.1880388184768477e-06, "loss": 1.0207, "step": 3185 }, { "epoch": 0.85, "learning_rate": 1.1839652613445219e-06, "loss": 1.0707, "step": 3186 }, { "epoch": 0.85, "learning_rate": 1.1798982604495635e-06, "loss": 1.062, "step": 3187 }, { "epoch": 0.85, "learning_rate": 1.1758378188164898e-06, "loss": 1.0302, "step": 3188 }, { "epoch": 0.85, "learning_rate": 1.1717839394649332e-06, "loss": 1.0095, "step": 3189 }, { "epoch": 0.85, "learning_rate": 1.1677366254096523e-06, "loss": 1.0423, "step": 3190 }, { "epoch": 0.85, "learning_rate": 1.1636958796605213e-06, "loss": 1.0191, "step": 3191 }, { "epoch": 0.85, "learning_rate": 1.1596617052225268e-06, "loss": 1.0737, "step": 3192 }, { "epoch": 0.85, "learning_rate": 1.1556341050957731e-06, "loss": 1.0786, "step": 3193 }, { "epoch": 0.85, "learning_rate": 1.1516130822754735e-06, "loss": 1.0876, "step": 3194 }, { "epoch": 0.85, "learning_rate": 1.1475986397519479e-06, "loss": 1.057, "step": 3195 }, { "epoch": 0.85, "learning_rate": 1.143590780510625e-06, "loss": 1.0403, "step": 3196 }, { "epoch": 0.85, "learning_rate": 1.1395895075320383e-06, "loss": 1.0308, "step": 3197 }, { "epoch": 0.85, "learning_rate": 1.1355948237918214e-06, "loss": 1.053, "step": 3198 }, { "epoch": 0.85, "learning_rate": 1.1316067322607072e-06, "loss": 1.0489, "step": 3199 }, { "epoch": 0.85, "learning_rate": 1.1276252359045304e-06, "loss": 1.0349, "step": 3200 }, { "epoch": 0.85, "learning_rate": 1.1236503376842134e-06, "loss": 1.0303, "step": 3201 }, { "epoch": 0.85, "learning_rate": 1.1196820405557795e-06, "loss": 1.0266, "step": 3202 }, { "epoch": 0.85, "learning_rate": 1.1157203474703415e-06, "loss": 1.0341, "step": 3203 }, { "epoch": 0.85, "learning_rate": 1.1117652613740959e-06, "loss": 1.0398, "step": 3204 }, { "epoch": 0.85, "learning_rate": 1.1078167852083288e-06, "loss": 1.0579, "step": 3205 }, { "epoch": 0.85, "learning_rate": 1.1038749219094147e-06, "loss": 1.0246, "step": 3206 }, { "epoch": 0.85, "learning_rate": 1.099939674408802e-06, "loss": 1.0519, "step": 3207 }, { "epoch": 0.85, "learning_rate": 1.0960110456330274e-06, "loss": 1.0566, "step": 3208 }, { "epoch": 0.85, "learning_rate": 1.0920890385037008e-06, "loss": 1.0804, "step": 3209 }, { "epoch": 0.85, "learning_rate": 1.0881736559375067e-06, "loss": 1.0319, "step": 3210 }, { "epoch": 0.85, "learning_rate": 1.084264900846208e-06, "loss": 1.0846, "step": 3211 }, { "epoch": 0.86, "learning_rate": 1.0803627761366352e-06, "loss": 1.0624, "step": 3212 }, { "epoch": 0.86, "learning_rate": 1.0764672847106872e-06, "loss": 1.0228, "step": 3213 }, { "epoch": 0.86, "learning_rate": 1.072578429465333e-06, "loss": 1.0629, "step": 3214 }, { "epoch": 0.86, "learning_rate": 1.068696213292606e-06, "loss": 0.9903, "step": 3215 }, { "epoch": 0.86, "learning_rate": 1.0648206390795978e-06, "loss": 1.022, "step": 3216 }, { "epoch": 0.86, "learning_rate": 1.0609517097084676e-06, "loss": 1.0424, "step": 3217 }, { "epoch": 0.86, "learning_rate": 1.0570894280564293e-06, "loss": 1.043, "step": 3218 }, { "epoch": 0.86, "learning_rate": 1.0532337969957496e-06, "loss": 1.0146, "step": 3219 }, { "epoch": 0.86, "learning_rate": 1.0493848193937572e-06, "loss": 1.0183, "step": 3220 }, { "epoch": 0.86, "learning_rate": 1.045542498112827e-06, "loss": 1.0324, "step": 3221 }, { "epoch": 0.86, "learning_rate": 1.0417068360103854e-06, "loss": 1.041, "step": 3222 }, { "epoch": 0.86, "learning_rate": 1.0378778359389053e-06, "loss": 1.0355, "step": 3223 }, { "epoch": 0.86, "learning_rate": 1.0340555007459096e-06, "loss": 1.004, "step": 3224 }, { "epoch": 0.86, "learning_rate": 1.0302398332739583e-06, "loss": 1.0256, "step": 3225 }, { "epoch": 0.86, "learning_rate": 1.0264308363606578e-06, "loss": 1.0226, "step": 3226 }, { "epoch": 0.86, "learning_rate": 1.0226285128386527e-06, "loss": 1.0559, "step": 3227 }, { "epoch": 0.86, "learning_rate": 1.0188328655356217e-06, "loss": 1.0168, "step": 3228 }, { "epoch": 0.86, "learning_rate": 1.0150438972742838e-06, "loss": 1.0419, "step": 3229 }, { "epoch": 0.86, "learning_rate": 1.0112616108723883e-06, "loss": 1.0397, "step": 3230 }, { "epoch": 0.86, "learning_rate": 1.007486009142712e-06, "loss": 1.0225, "step": 3231 }, { "epoch": 0.86, "learning_rate": 1.0037170948930664e-06, "loss": 1.0053, "step": 3232 }, { "epoch": 0.86, "learning_rate": 9.999548709262874e-07, "loss": 1.0238, "step": 3233 }, { "epoch": 0.86, "learning_rate": 9.961993400402347e-07, "loss": 1.05, "step": 3234 }, { "epoch": 0.86, "learning_rate": 9.9245050502779e-07, "loss": 1.0402, "step": 3235 }, { "epoch": 0.86, "learning_rate": 9.887083686768594e-07, "loss": 1.0235, "step": 3236 }, { "epoch": 0.86, "learning_rate": 9.849729337703606e-07, "loss": 1.0451, "step": 3237 }, { "epoch": 0.86, "learning_rate": 9.812442030862345e-07, "loss": 1.0402, "step": 3238 }, { "epoch": 0.86, "learning_rate": 9.775221793974343e-07, "loss": 1.0307, "step": 3239 }, { "epoch": 0.86, "learning_rate": 9.738068654719214e-07, "loss": 1.0648, "step": 3240 }, { "epoch": 0.86, "learning_rate": 9.70098264072672e-07, "loss": 1.0501, "step": 3241 }, { "epoch": 0.86, "learning_rate": 9.663963779576701e-07, "loss": 1.0403, "step": 3242 }, { "epoch": 0.86, "learning_rate": 9.627012098799005e-07, "loss": 1.0665, "step": 3243 }, { "epoch": 0.86, "learning_rate": 9.59012762587359e-07, "loss": 1.014, "step": 3244 }, { "epoch": 0.86, "learning_rate": 9.553310388230409e-07, "loss": 1.034, "step": 3245 }, { "epoch": 0.86, "learning_rate": 9.516560413249376e-07, "loss": 1.0234, "step": 3246 }, { "epoch": 0.86, "learning_rate": 9.479877728260433e-07, "loss": 1.0365, "step": 3247 }, { "epoch": 0.86, "learning_rate": 9.443262360543482e-07, "loss": 1.0665, "step": 3248 }, { "epoch": 0.87, "learning_rate": 9.406714337328304e-07, "loss": 1.087, "step": 3249 }, { "epoch": 0.87, "learning_rate": 9.370233685794649e-07, "loss": 1.0529, "step": 3250 }, { "epoch": 0.87, "learning_rate": 9.333820433072194e-07, "loss": 1.0799, "step": 3251 }, { "epoch": 0.87, "learning_rate": 9.297474606240387e-07, "loss": 1.051, "step": 3252 }, { "epoch": 0.87, "learning_rate": 9.261196232328628e-07, "loss": 1.0954, "step": 3253 }, { "epoch": 0.87, "learning_rate": 9.224985338316151e-07, "loss": 1.0295, "step": 3254 }, { "epoch": 0.87, "learning_rate": 9.188841951131944e-07, "loss": 1.0413, "step": 3255 }, { "epoch": 0.87, "learning_rate": 9.152766097654831e-07, "loss": 1.0442, "step": 3256 }, { "epoch": 0.87, "learning_rate": 9.116757804713439e-07, "loss": 1.0424, "step": 3257 }, { "epoch": 0.87, "learning_rate": 9.080817099086081e-07, "loss": 1.0334, "step": 3258 }, { "epoch": 0.87, "learning_rate": 9.044944007500889e-07, "loss": 1.0562, "step": 3259 }, { "epoch": 0.87, "learning_rate": 9.009138556635666e-07, "loss": 1.0048, "step": 3260 }, { "epoch": 0.87, "learning_rate": 8.973400773117891e-07, "loss": 1.0278, "step": 3261 }, { "epoch": 0.87, "learning_rate": 8.937730683524781e-07, "loss": 1.0294, "step": 3262 }, { "epoch": 0.87, "learning_rate": 8.902128314383185e-07, "loss": 1.0629, "step": 3263 }, { "epoch": 0.87, "learning_rate": 8.866593692169545e-07, "loss": 1.0318, "step": 3264 }, { "epoch": 0.87, "learning_rate": 8.831126843310001e-07, "loss": 1.0379, "step": 3265 }, { "epoch": 0.87, "learning_rate": 8.795727794180243e-07, "loss": 1.0726, "step": 3266 }, { "epoch": 0.87, "learning_rate": 8.760396571105522e-07, "loss": 1.0729, "step": 3267 }, { "epoch": 0.87, "learning_rate": 8.725133200360702e-07, "loss": 1.0676, "step": 3268 }, { "epoch": 0.87, "learning_rate": 8.689937708170171e-07, "loss": 1.0744, "step": 3269 }, { "epoch": 0.87, "learning_rate": 8.654810120707801e-07, "loss": 1.0876, "step": 3270 }, { "epoch": 0.87, "learning_rate": 8.619750464096999e-07, "loss": 1.0431, "step": 3271 }, { "epoch": 0.87, "learning_rate": 8.584758764410661e-07, "loss": 1.058, "step": 3272 }, { "epoch": 0.87, "learning_rate": 8.549835047671129e-07, "loss": 1.0364, "step": 3273 }, { "epoch": 0.87, "learning_rate": 8.514979339850171e-07, "loss": 1.0685, "step": 3274 }, { "epoch": 0.87, "learning_rate": 8.480191666869009e-07, "loss": 1.0166, "step": 3275 }, { "epoch": 0.87, "learning_rate": 8.445472054598281e-07, "loss": 1.0641, "step": 3276 }, { "epoch": 0.87, "learning_rate": 8.410820528857933e-07, "loss": 1.0493, "step": 3277 }, { "epoch": 0.87, "learning_rate": 8.376237115417374e-07, "loss": 1.0476, "step": 3278 }, { "epoch": 0.87, "learning_rate": 8.34172183999532e-07, "loss": 1.0156, "step": 3279 }, { "epoch": 0.87, "learning_rate": 8.307274728259784e-07, "loss": 1.0703, "step": 3280 }, { "epoch": 0.87, "learning_rate": 8.27289580582813e-07, "loss": 1.0719, "step": 3281 }, { "epoch": 0.87, "learning_rate": 8.238585098267015e-07, "loss": 1.0591, "step": 3282 }, { "epoch": 0.87, "learning_rate": 8.20434263109231e-07, "loss": 1.0538, "step": 3283 }, { "epoch": 0.87, "learning_rate": 8.170168429769188e-07, "loss": 1.0001, "step": 3284 }, { "epoch": 0.87, "learning_rate": 8.136062519712074e-07, "loss": 1.02, "step": 3285 }, { "epoch": 0.87, "learning_rate": 8.102024926284535e-07, "loss": 1.0069, "step": 3286 }, { "epoch": 0.88, "learning_rate": 8.068055674799391e-07, "loss": 1.0987, "step": 3287 }, { "epoch": 0.88, "learning_rate": 8.034154790518633e-07, "loss": 1.0477, "step": 3288 }, { "epoch": 0.88, "learning_rate": 8.000322298653363e-07, "loss": 1.0986, "step": 3289 }, { "epoch": 0.88, "learning_rate": 7.966558224363863e-07, "loss": 1.023, "step": 3290 }, { "epoch": 0.88, "learning_rate": 7.932862592759571e-07, "loss": 1.047, "step": 3291 }, { "epoch": 0.88, "learning_rate": 7.899235428898922e-07, "loss": 1.0336, "step": 3292 }, { "epoch": 0.88, "learning_rate": 7.86567675778952e-07, "loss": 1.0585, "step": 3293 }, { "epoch": 0.88, "learning_rate": 7.832186604388015e-07, "loss": 1.0262, "step": 3294 }, { "epoch": 0.88, "learning_rate": 7.798764993600072e-07, "loss": 0.9935, "step": 3295 }, { "epoch": 0.88, "learning_rate": 7.765411950280411e-07, "loss": 1.0679, "step": 3296 }, { "epoch": 0.88, "learning_rate": 7.732127499232778e-07, "loss": 1.0958, "step": 3297 }, { "epoch": 0.88, "learning_rate": 7.698911665209841e-07, "loss": 1.0291, "step": 3298 }, { "epoch": 0.88, "learning_rate": 7.665764472913306e-07, "loss": 1.0247, "step": 3299 }, { "epoch": 0.88, "learning_rate": 7.632685946993834e-07, "loss": 1.0557, "step": 3300 }, { "epoch": 0.88, "learning_rate": 7.599676112050935e-07, "loss": 1.0325, "step": 3301 }, { "epoch": 0.88, "learning_rate": 7.566734992633141e-07, "loss": 1.0377, "step": 3302 }, { "epoch": 0.88, "learning_rate": 7.533862613237842e-07, "loss": 1.0428, "step": 3303 }, { "epoch": 0.88, "learning_rate": 7.501058998311273e-07, "loss": 1.0946, "step": 3304 }, { "epoch": 0.88, "learning_rate": 7.468324172248565e-07, "loss": 1.0461, "step": 3305 }, { "epoch": 0.88, "learning_rate": 7.435658159393722e-07, "loss": 1.0489, "step": 3306 }, { "epoch": 0.88, "learning_rate": 7.4030609840395e-07, "loss": 1.0542, "step": 3307 }, { "epoch": 0.88, "learning_rate": 7.37053267042751e-07, "loss": 1.0869, "step": 3308 }, { "epoch": 0.88, "learning_rate": 7.338073242748178e-07, "loss": 1.0545, "step": 3309 }, { "epoch": 0.88, "learning_rate": 7.305682725140629e-07, "loss": 1.0512, "step": 3310 }, { "epoch": 0.88, "learning_rate": 7.273361141692803e-07, "loss": 1.0206, "step": 3311 }, { "epoch": 0.88, "learning_rate": 7.241108516441353e-07, "loss": 1.0594, "step": 3312 }, { "epoch": 0.88, "learning_rate": 7.208924873371625e-07, "loss": 1.0443, "step": 3313 }, { "epoch": 0.88, "learning_rate": 7.17681023641772e-07, "loss": 1.039, "step": 3314 }, { "epoch": 0.88, "learning_rate": 7.144764629462397e-07, "loss": 1.0459, "step": 3315 }, { "epoch": 0.88, "learning_rate": 7.112788076337029e-07, "loss": 1.033, "step": 3316 }, { "epoch": 0.88, "learning_rate": 7.080880600821727e-07, "loss": 1.0844, "step": 3317 }, { "epoch": 0.88, "learning_rate": 7.049042226645164e-07, "loss": 1.0922, "step": 3318 }, { "epoch": 0.88, "learning_rate": 7.017272977484646e-07, "loss": 1.0484, "step": 3319 }, { "epoch": 0.88, "learning_rate": 6.985572876966063e-07, "loss": 1.062, "step": 3320 }, { "epoch": 0.88, "learning_rate": 6.953941948663912e-07, "loss": 1.0649, "step": 3321 }, { "epoch": 0.88, "learning_rate": 6.922380216101199e-07, "loss": 1.0773, "step": 3322 }, { "epoch": 0.88, "learning_rate": 6.890887702749516e-07, "loss": 1.0387, "step": 3323 }, { "epoch": 0.88, "learning_rate": 6.859464432028961e-07, "loss": 1.0607, "step": 3324 }, { "epoch": 0.89, "learning_rate": 6.828110427308121e-07, "loss": 0.9808, "step": 3325 }, { "epoch": 0.89, "learning_rate": 6.796825711904098e-07, "loss": 1.0619, "step": 3326 }, { "epoch": 0.89, "learning_rate": 6.765610309082471e-07, "loss": 1.0725, "step": 3327 }, { "epoch": 0.89, "learning_rate": 6.734464242057237e-07, "loss": 1.041, "step": 3328 }, { "epoch": 0.89, "learning_rate": 6.703387533990846e-07, "loss": 1.0186, "step": 3329 }, { "epoch": 0.89, "learning_rate": 6.672380207994211e-07, "loss": 1.0397, "step": 3330 }, { "epoch": 0.89, "learning_rate": 6.641442287126565e-07, "loss": 1.1046, "step": 3331 }, { "epoch": 0.89, "learning_rate": 6.610573794395581e-07, "loss": 1.0474, "step": 3332 }, { "epoch": 0.89, "learning_rate": 6.579774752757329e-07, "loss": 1.0268, "step": 3333 }, { "epoch": 0.89, "learning_rate": 6.549045185116131e-07, "loss": 1.0843, "step": 3334 }, { "epoch": 0.89, "learning_rate": 6.518385114324732e-07, "loss": 1.0118, "step": 3335 }, { "epoch": 0.89, "learning_rate": 6.487794563184191e-07, "loss": 0.9777, "step": 3336 }, { "epoch": 0.89, "learning_rate": 6.457273554443788e-07, "loss": 1.0521, "step": 3337 }, { "epoch": 0.89, "learning_rate": 6.426822110801178e-07, "loss": 1.05, "step": 3338 }, { "epoch": 0.89, "learning_rate": 6.396440254902248e-07, "loss": 1.0692, "step": 3339 }, { "epoch": 0.89, "learning_rate": 6.366128009341088e-07, "loss": 1.029, "step": 3340 }, { "epoch": 0.89, "learning_rate": 6.335885396660102e-07, "loss": 1.062, "step": 3341 }, { "epoch": 0.89, "learning_rate": 6.305712439349876e-07, "loss": 1.0155, "step": 3342 }, { "epoch": 0.89, "learning_rate": 6.27560915984915e-07, "loss": 1.0796, "step": 3343 }, { "epoch": 0.89, "learning_rate": 6.245575580544927e-07, "loss": 1.0604, "step": 3344 }, { "epoch": 0.89, "learning_rate": 6.21561172377233e-07, "loss": 1.088, "step": 3345 }, { "epoch": 0.89, "learning_rate": 6.185717611814624e-07, "loss": 1.0336, "step": 3346 }, { "epoch": 0.89, "learning_rate": 6.155893266903223e-07, "loss": 1.0522, "step": 3347 }, { "epoch": 0.89, "learning_rate": 6.126138711217678e-07, "loss": 1.0216, "step": 3348 }, { "epoch": 0.89, "learning_rate": 6.096453966885574e-07, "loss": 1.0561, "step": 3349 }, { "epoch": 0.89, "learning_rate": 6.066839055982642e-07, "loss": 1.0678, "step": 3350 }, { "epoch": 0.89, "learning_rate": 6.037294000532679e-07, "loss": 1.0617, "step": 3351 }, { "epoch": 0.89, "learning_rate": 6.007818822507483e-07, "loss": 1.0375, "step": 3352 }, { "epoch": 0.89, "learning_rate": 5.978413543826922e-07, "loss": 1.0682, "step": 3353 }, { "epoch": 0.89, "learning_rate": 5.949078186358903e-07, "loss": 1.0507, "step": 3354 }, { "epoch": 0.89, "learning_rate": 5.919812771919264e-07, "loss": 1.0525, "step": 3355 }, { "epoch": 0.89, "learning_rate": 5.890617322271897e-07, "loss": 1.0493, "step": 3356 }, { "epoch": 0.89, "learning_rate": 5.861491859128642e-07, "loss": 1.0438, "step": 3357 }, { "epoch": 0.89, "learning_rate": 5.832436404149266e-07, "loss": 1.0624, "step": 3358 }, { "epoch": 0.89, "learning_rate": 5.803450978941516e-07, "loss": 1.0127, "step": 3359 }, { "epoch": 0.89, "learning_rate": 5.774535605061037e-07, "loss": 1.065, "step": 3360 }, { "epoch": 0.89, "learning_rate": 5.745690304011353e-07, "loss": 1.0039, "step": 3361 }, { "epoch": 0.9, "learning_rate": 5.716915097243936e-07, "loss": 1.0374, "step": 3362 }, { "epoch": 0.9, "learning_rate": 5.688210006158101e-07, "loss": 1.0454, "step": 3363 }, { "epoch": 0.9, "learning_rate": 5.659575052100986e-07, "loss": 1.0453, "step": 3364 }, { "epoch": 0.9, "learning_rate": 5.63101025636762e-07, "loss": 1.0405, "step": 3365 }, { "epoch": 0.9, "learning_rate": 5.602515640200856e-07, "loss": 1.0439, "step": 3366 }, { "epoch": 0.9, "learning_rate": 5.574091224791312e-07, "loss": 1.0176, "step": 3367 }, { "epoch": 0.9, "learning_rate": 5.545737031277432e-07, "loss": 1.015, "step": 3368 }, { "epoch": 0.9, "learning_rate": 5.517453080745449e-07, "loss": 1.0347, "step": 3369 }, { "epoch": 0.9, "learning_rate": 5.489239394229328e-07, "loss": 1.0651, "step": 3370 }, { "epoch": 0.9, "learning_rate": 5.461095992710797e-07, "loss": 1.052, "step": 3371 }, { "epoch": 0.9, "learning_rate": 5.433022897119333e-07, "loss": 1.0718, "step": 3372 }, { "epoch": 0.9, "learning_rate": 5.405020128332072e-07, "loss": 1.0445, "step": 3373 }, { "epoch": 0.9, "learning_rate": 5.377087707173912e-07, "loss": 1.0716, "step": 3374 }, { "epoch": 0.9, "learning_rate": 5.349225654417412e-07, "loss": 1.024, "step": 3375 }, { "epoch": 0.9, "learning_rate": 5.321433990782776e-07, "loss": 1.0643, "step": 3376 }, { "epoch": 0.9, "learning_rate": 5.293712736937895e-07, "loss": 1.0768, "step": 3377 }, { "epoch": 0.9, "learning_rate": 5.26606191349831e-07, "loss": 1.0379, "step": 3378 }, { "epoch": 0.9, "learning_rate": 5.23848154102713e-07, "loss": 1.0293, "step": 3379 }, { "epoch": 0.9, "learning_rate": 5.210971640035101e-07, "loss": 1.011, "step": 3380 }, { "epoch": 0.9, "learning_rate": 5.183532230980604e-07, "loss": 1.0597, "step": 3381 }, { "epoch": 0.9, "learning_rate": 5.156163334269504e-07, "loss": 1.0565, "step": 3382 }, { "epoch": 0.9, "learning_rate": 5.12886497025531e-07, "loss": 1.059, "step": 3383 }, { "epoch": 0.9, "learning_rate": 5.101637159239048e-07, "loss": 1.0756, "step": 3384 }, { "epoch": 0.9, "learning_rate": 5.074479921469277e-07, "loss": 1.0014, "step": 3385 }, { "epoch": 0.9, "learning_rate": 5.047393277142054e-07, "loss": 1.0552, "step": 3386 }, { "epoch": 0.9, "learning_rate": 5.020377246400987e-07, "loss": 1.0262, "step": 3387 }, { "epoch": 0.9, "learning_rate": 4.993431849337104e-07, "loss": 1.0031, "step": 3388 }, { "epoch": 0.9, "learning_rate": 4.966557105988956e-07, "loss": 0.9906, "step": 3389 }, { "epoch": 0.9, "learning_rate": 4.939753036342553e-07, "loss": 1.0464, "step": 3390 }, { "epoch": 0.9, "learning_rate": 4.913019660331286e-07, "loss": 1.0394, "step": 3391 }, { "epoch": 0.9, "learning_rate": 4.886356997836051e-07, "loss": 1.0905, "step": 3392 }, { "epoch": 0.9, "learning_rate": 4.859765068685118e-07, "loss": 1.1047, "step": 3393 }, { "epoch": 0.9, "learning_rate": 4.833243892654138e-07, "loss": 0.9997, "step": 3394 }, { "epoch": 0.9, "learning_rate": 4.806793489466189e-07, "loss": 1.0303, "step": 3395 }, { "epoch": 0.9, "learning_rate": 4.780413878791679e-07, "loss": 1.0342, "step": 3396 }, { "epoch": 0.9, "learning_rate": 4.754105080248372e-07, "loss": 1.0552, "step": 3397 }, { "epoch": 0.9, "learning_rate": 4.7278671134013965e-07, "loss": 1.005, "step": 3398 }, { "epoch": 0.9, "learning_rate": 4.701699997763209e-07, "loss": 1.0962, "step": 3399 }, { "epoch": 0.91, "learning_rate": 4.675603752793523e-07, "loss": 1.0388, "step": 3400 }, { "epoch": 0.91, "learning_rate": 4.649578397899401e-07, "loss": 1.0467, "step": 3401 }, { "epoch": 0.91, "learning_rate": 4.6236239524351744e-07, "loss": 1.0819, "step": 3402 }, { "epoch": 0.91, "learning_rate": 4.5977404357024137e-07, "loss": 1.0316, "step": 3403 }, { "epoch": 0.91, "learning_rate": 4.5719278669499765e-07, "loss": 1.0471, "step": 3404 }, { "epoch": 0.91, "learning_rate": 4.546186265373964e-07, "loss": 1.0261, "step": 3405 }, { "epoch": 0.91, "learning_rate": 4.5205156501176517e-07, "loss": 1.0211, "step": 3406 }, { "epoch": 0.91, "learning_rate": 4.4949160402715575e-07, "loss": 1.0517, "step": 3407 }, { "epoch": 0.91, "learning_rate": 4.4693874548734195e-07, "loss": 1.0179, "step": 3408 }, { "epoch": 0.91, "learning_rate": 4.443929912908096e-07, "loss": 1.012, "step": 3409 }, { "epoch": 0.91, "learning_rate": 4.418543433307665e-07, "loss": 1.0284, "step": 3410 }, { "epoch": 0.91, "learning_rate": 4.393228034951347e-07, "loss": 1.0756, "step": 3411 }, { "epoch": 0.91, "learning_rate": 4.3679837366654597e-07, "loss": 1.0589, "step": 3412 }, { "epoch": 0.91, "learning_rate": 4.342810557223498e-07, "loss": 1.0092, "step": 3413 }, { "epoch": 0.91, "learning_rate": 4.317708515346075e-07, "loss": 1.0912, "step": 3414 }, { "epoch": 0.91, "learning_rate": 4.2926776297008255e-07, "loss": 1.0598, "step": 3415 }, { "epoch": 0.91, "learning_rate": 4.2677179189025364e-07, "loss": 1.0546, "step": 3416 }, { "epoch": 0.91, "learning_rate": 4.24282940151306e-07, "loss": 1.0644, "step": 3417 }, { "epoch": 0.91, "learning_rate": 4.2180120960412573e-07, "loss": 1.0519, "step": 3418 }, { "epoch": 0.91, "learning_rate": 4.1932660209430875e-07, "loss": 1.0321, "step": 3419 }, { "epoch": 0.91, "learning_rate": 4.1685911946215073e-07, "loss": 1.0364, "step": 3420 }, { "epoch": 0.91, "learning_rate": 4.143987635426472e-07, "loss": 1.0773, "step": 3421 }, { "epoch": 0.91, "learning_rate": 4.119455361654967e-07, "loss": 1.0044, "step": 3422 }, { "epoch": 0.91, "learning_rate": 4.094994391550977e-07, "loss": 1.0325, "step": 3423 }, { "epoch": 0.91, "learning_rate": 4.0706047433054065e-07, "loss": 1.0246, "step": 3424 }, { "epoch": 0.91, "learning_rate": 4.046286435056157e-07, "loss": 1.0599, "step": 3425 }, { "epoch": 0.91, "learning_rate": 4.022039484888107e-07, "loss": 1.0701, "step": 3426 }, { "epoch": 0.91, "learning_rate": 3.9978639108329774e-07, "loss": 1.0564, "step": 3427 }, { "epoch": 0.91, "learning_rate": 3.9737597308694976e-07, "loss": 1.0257, "step": 3428 }, { "epoch": 0.91, "learning_rate": 3.9497269629232725e-07, "loss": 1.0338, "step": 3429 }, { "epoch": 0.91, "learning_rate": 3.925765624866784e-07, "loss": 1.0251, "step": 3430 }, { "epoch": 0.91, "learning_rate": 3.9018757345194005e-07, "loss": 1.0446, "step": 3431 }, { "epoch": 0.91, "learning_rate": 3.878057309647376e-07, "loss": 1.0216, "step": 3432 }, { "epoch": 0.91, "learning_rate": 3.854310367963798e-07, "loss": 1.0091, "step": 3433 }, { "epoch": 0.91, "learning_rate": 3.830634927128596e-07, "loss": 1.0684, "step": 3434 }, { "epoch": 0.91, "learning_rate": 3.80703100474854e-07, "loss": 1.0661, "step": 3435 }, { "epoch": 0.91, "learning_rate": 3.7834986183771905e-07, "loss": 1.0538, "step": 3436 }, { "epoch": 0.92, "learning_rate": 3.760037785514925e-07, "loss": 1.0563, "step": 3437 }, { "epoch": 0.92, "learning_rate": 3.736648523608932e-07, "loss": 1.0725, "step": 3438 }, { "epoch": 0.92, "learning_rate": 3.7133308500531095e-07, "loss": 1.0589, "step": 3439 }, { "epoch": 0.92, "learning_rate": 3.690084782188175e-07, "loss": 1.0278, "step": 3440 }, { "epoch": 0.92, "learning_rate": 3.666910337301588e-07, "loss": 1.0704, "step": 3441 }, { "epoch": 0.92, "learning_rate": 3.643807532627508e-07, "loss": 1.0384, "step": 3442 }, { "epoch": 0.92, "learning_rate": 3.620776385346847e-07, "loss": 1.0776, "step": 3443 }, { "epoch": 0.92, "learning_rate": 3.5978169125872484e-07, "loss": 1.0551, "step": 3444 }, { "epoch": 0.92, "learning_rate": 3.574929131422988e-07, "loss": 1.0392, "step": 3445 }, { "epoch": 0.92, "learning_rate": 3.552113058875084e-07, "loss": 1.0109, "step": 3446 }, { "epoch": 0.92, "learning_rate": 3.5293687119112294e-07, "loss": 1.0284, "step": 3447 }, { "epoch": 0.92, "learning_rate": 3.5066961074457173e-07, "loss": 1.0561, "step": 3448 }, { "epoch": 0.92, "learning_rate": 3.4840952623395485e-07, "loss": 1.0976, "step": 3449 }, { "epoch": 0.92, "learning_rate": 3.461566193400323e-07, "loss": 1.0633, "step": 3450 }, { "epoch": 0.92, "learning_rate": 3.439108917382306e-07, "loss": 1.0341, "step": 3451 }, { "epoch": 0.92, "learning_rate": 3.416723450986315e-07, "loss": 1.0662, "step": 3452 }, { "epoch": 0.92, "learning_rate": 3.3944098108598e-07, "loss": 1.0565, "step": 3453 }, { "epoch": 0.92, "learning_rate": 3.3721680135967993e-07, "loss": 1.0557, "step": 3454 }, { "epoch": 0.92, "learning_rate": 3.3499980757378923e-07, "loss": 1.0209, "step": 3455 }, { "epoch": 0.92, "learning_rate": 3.3279000137702464e-07, "loss": 1.0505, "step": 3456 }, { "epoch": 0.92, "learning_rate": 3.305873844127594e-07, "loss": 1.0604, "step": 3457 }, { "epoch": 0.92, "learning_rate": 3.2839195831901315e-07, "loss": 1.0292, "step": 3458 }, { "epoch": 0.92, "learning_rate": 3.262037247284666e-07, "loss": 1.0441, "step": 3459 }, { "epoch": 0.92, "learning_rate": 3.240226852684458e-07, "loss": 1.0269, "step": 3460 }, { "epoch": 0.92, "learning_rate": 3.218488415609289e-07, "loss": 1.0176, "step": 3461 }, { "epoch": 0.92, "learning_rate": 3.1968219522254153e-07, "loss": 1.0433, "step": 3462 }, { "epoch": 0.92, "learning_rate": 3.175227478645604e-07, "loss": 1.0477, "step": 3463 }, { "epoch": 0.92, "learning_rate": 3.153705010929031e-07, "loss": 1.0409, "step": 3464 }, { "epoch": 0.92, "learning_rate": 3.1322545650813495e-07, "loss": 1.0252, "step": 3465 }, { "epoch": 0.92, "learning_rate": 3.1108761570546875e-07, "loss": 1.0923, "step": 3466 }, { "epoch": 0.92, "learning_rate": 3.0895698027475164e-07, "loss": 1.0631, "step": 3467 }, { "epoch": 0.92, "learning_rate": 3.068335518004817e-07, "loss": 1.0623, "step": 3468 }, { "epoch": 0.92, "learning_rate": 3.0471733186179133e-07, "loss": 1.042, "step": 3469 }, { "epoch": 0.92, "learning_rate": 3.026083220324527e-07, "loss": 1.0917, "step": 3470 }, { "epoch": 0.92, "learning_rate": 3.005065238808791e-07, "loss": 1.0663, "step": 3471 }, { "epoch": 0.92, "learning_rate": 2.98411938970119e-07, "loss": 1.0552, "step": 3472 }, { "epoch": 0.92, "learning_rate": 2.963245688578531e-07, "loss": 1.0737, "step": 3473 }, { "epoch": 0.92, "learning_rate": 2.94244415096403e-07, "loss": 1.0612, "step": 3474 }, { "epoch": 0.93, "learning_rate": 2.9217147923272126e-07, "loss": 1.0124, "step": 3475 }, { "epoch": 0.93, "learning_rate": 2.901057628083892e-07, "loss": 1.051, "step": 3476 }, { "epoch": 0.93, "learning_rate": 2.880472673596224e-07, "loss": 1.0609, "step": 3477 }, { "epoch": 0.93, "learning_rate": 2.859959944172697e-07, "loss": 1.0338, "step": 3478 }, { "epoch": 0.93, "learning_rate": 2.8395194550680074e-07, "loss": 1.0164, "step": 3479 }, { "epoch": 0.93, "learning_rate": 2.819151221483196e-07, "loss": 1.0697, "step": 3480 }, { "epoch": 0.93, "learning_rate": 2.7988552585655473e-07, "loss": 1.0784, "step": 3481 }, { "epoch": 0.93, "learning_rate": 2.778631581408575e-07, "loss": 1.0125, "step": 3482 }, { "epoch": 0.93, "learning_rate": 2.758480205052094e-07, "loss": 1.0489, "step": 3483 }, { "epoch": 0.93, "learning_rate": 2.7384011444820945e-07, "loss": 1.0483, "step": 3484 }, { "epoch": 0.93, "learning_rate": 2.7183944146308203e-07, "loss": 1.0823, "step": 3485 }, { "epoch": 0.93, "learning_rate": 2.6984600303767036e-07, "loss": 1.0355, "step": 3486 }, { "epoch": 0.93, "learning_rate": 2.678598006544397e-07, "loss": 1.0509, "step": 3487 }, { "epoch": 0.93, "learning_rate": 2.6588083579047185e-07, "loss": 1.0497, "step": 3488 }, { "epoch": 0.93, "learning_rate": 2.6390910991746843e-07, "loss": 1.0436, "step": 3489 }, { "epoch": 0.93, "learning_rate": 2.619446245017465e-07, "loss": 1.0233, "step": 3490 }, { "epoch": 0.93, "learning_rate": 2.5998738100423746e-07, "loss": 1.0711, "step": 3491 }, { "epoch": 0.93, "learning_rate": 2.5803738088048924e-07, "loss": 1.0645, "step": 3492 }, { "epoch": 0.93, "learning_rate": 2.560946255806629e-07, "loss": 1.0557, "step": 3493 }, { "epoch": 0.93, "learning_rate": 2.541591165495294e-07, "loss": 1.0295, "step": 3494 }, { "epoch": 0.93, "learning_rate": 2.5223085522647295e-07, "loss": 1.0421, "step": 3495 }, { "epoch": 0.93, "learning_rate": 2.5030984304548867e-07, "loss": 1.0681, "step": 3496 }, { "epoch": 0.93, "learning_rate": 2.4839608143517715e-07, "loss": 1.0008, "step": 3497 }, { "epoch": 0.93, "learning_rate": 2.464895718187499e-07, "loss": 1.0189, "step": 3498 }, { "epoch": 0.93, "learning_rate": 2.445903156140261e-07, "loss": 1.0155, "step": 3499 }, { "epoch": 0.93, "learning_rate": 2.4269831423342713e-07, "loss": 1.0682, "step": 3500 }, { "epoch": 0.93, "learning_rate": 2.40813569083983e-07, "loss": 1.0531, "step": 3501 }, { "epoch": 0.93, "learning_rate": 2.3893608156732474e-07, "loss": 1.0682, "step": 3502 }, { "epoch": 0.93, "learning_rate": 2.3706585307968654e-07, "loss": 1.0663, "step": 3503 }, { "epoch": 0.93, "learning_rate": 2.3520288501190592e-07, "loss": 1.0472, "step": 3504 }, { "epoch": 0.93, "learning_rate": 2.3334717874942014e-07, "loss": 1.0059, "step": 3505 }, { "epoch": 0.93, "learning_rate": 2.314987356722642e-07, "loss": 1.0988, "step": 3506 }, { "epoch": 0.93, "learning_rate": 2.296575571550741e-07, "loss": 1.0456, "step": 3507 }, { "epoch": 0.93, "learning_rate": 2.2782364456708228e-07, "loss": 1.0513, "step": 3508 }, { "epoch": 0.93, "learning_rate": 2.2599699927211892e-07, "loss": 1.0452, "step": 3509 }, { "epoch": 0.93, "learning_rate": 2.2417762262860964e-07, "loss": 1.0362, "step": 3510 }, { "epoch": 0.93, "learning_rate": 2.2236551598957102e-07, "loss": 1.0501, "step": 3511 }, { "epoch": 0.94, "learning_rate": 2.2056068070261726e-07, "loss": 1.0874, "step": 3512 }, { "epoch": 0.94, "learning_rate": 2.1876311810995477e-07, "loss": 1.0415, "step": 3513 }, { "epoch": 0.94, "learning_rate": 2.1697282954837861e-07, "loss": 1.0356, "step": 3514 }, { "epoch": 0.94, "learning_rate": 2.1518981634927605e-07, "loss": 1.0191, "step": 3515 }, { "epoch": 0.94, "learning_rate": 2.1341407983862527e-07, "loss": 1.0763, "step": 3516 }, { "epoch": 0.94, "learning_rate": 2.1164562133699107e-07, "loss": 1.0272, "step": 3517 }, { "epoch": 0.94, "learning_rate": 2.0988444215952475e-07, "loss": 1.0517, "step": 3518 }, { "epoch": 0.94, "learning_rate": 2.0813054361596973e-07, "loss": 1.0617, "step": 3519 }, { "epoch": 0.94, "learning_rate": 2.0638392701064713e-07, "loss": 1.0724, "step": 3520 }, { "epoch": 0.94, "learning_rate": 2.0464459364246792e-07, "loss": 1.0851, "step": 3521 }, { "epoch": 0.94, "learning_rate": 2.0291254480492516e-07, "loss": 1.1186, "step": 3522 }, { "epoch": 0.94, "learning_rate": 2.0118778178609744e-07, "loss": 1.0252, "step": 3523 }, { "epoch": 0.94, "learning_rate": 1.9947030586863869e-07, "loss": 1.0144, "step": 3524 }, { "epoch": 0.94, "learning_rate": 1.9776011832979058e-07, "loss": 1.0507, "step": 3525 }, { "epoch": 0.94, "learning_rate": 1.960572204413702e-07, "loss": 1.0649, "step": 3526 }, { "epoch": 0.94, "learning_rate": 1.9436161346977456e-07, "loss": 1.0791, "step": 3527 }, { "epoch": 0.94, "learning_rate": 1.926732986759794e-07, "loss": 1.0845, "step": 3528 }, { "epoch": 0.94, "learning_rate": 1.9099227731553705e-07, "loss": 1.0376, "step": 3529 }, { "epoch": 0.94, "learning_rate": 1.893185506385764e-07, "loss": 1.0476, "step": 3530 }, { "epoch": 0.94, "learning_rate": 1.876521198897996e-07, "loss": 1.0538, "step": 3531 }, { "epoch": 0.94, "learning_rate": 1.859929863084853e-07, "loss": 1.0467, "step": 3532 }, { "epoch": 0.94, "learning_rate": 1.8434115112848427e-07, "loss": 1.0787, "step": 3533 }, { "epoch": 0.94, "learning_rate": 1.826966155782195e-07, "loss": 1.0656, "step": 3534 }, { "epoch": 0.94, "learning_rate": 1.810593808806882e-07, "loss": 1.0706, "step": 3535 }, { "epoch": 0.94, "learning_rate": 1.79429448253452e-07, "loss": 1.0366, "step": 3536 }, { "epoch": 0.94, "learning_rate": 1.7780681890864904e-07, "loss": 1.0475, "step": 3537 }, { "epoch": 0.94, "learning_rate": 1.7619149405298298e-07, "loss": 1.1094, "step": 3538 }, { "epoch": 0.94, "learning_rate": 1.7458347488772287e-07, "loss": 1.0694, "step": 3539 }, { "epoch": 0.94, "learning_rate": 1.7298276260870995e-07, "loss": 1.1082, "step": 3540 }, { "epoch": 0.94, "learning_rate": 1.7138935840634974e-07, "loss": 1.0617, "step": 3541 }, { "epoch": 0.94, "learning_rate": 1.6980326346560883e-07, "loss": 1.0744, "step": 3542 }, { "epoch": 0.94, "learning_rate": 1.6822447896602368e-07, "loss": 1.0629, "step": 3543 }, { "epoch": 0.94, "learning_rate": 1.6665300608169289e-07, "loss": 1.0723, "step": 3544 }, { "epoch": 0.94, "learning_rate": 1.6508884598127385e-07, "loss": 1.0594, "step": 3545 }, { "epoch": 0.94, "learning_rate": 1.6353199982799163e-07, "loss": 1.071, "step": 3546 }, { "epoch": 0.94, "learning_rate": 1.6198246877962676e-07, "loss": 1.0067, "step": 3547 }, { "epoch": 0.94, "learning_rate": 1.6044025398852193e-07, "loss": 1.016, "step": 3548 }, { "epoch": 0.94, "learning_rate": 1.589053566015797e-07, "loss": 1.0211, "step": 3549 }, { "epoch": 0.95, "learning_rate": 1.5737777776026142e-07, "loss": 1.0892, "step": 3550 }, { "epoch": 0.95, "learning_rate": 1.5585751860058285e-07, "loss": 1.0855, "step": 3551 }, { "epoch": 0.95, "learning_rate": 1.5434458025311738e-07, "loss": 1.0317, "step": 3552 }, { "epoch": 0.95, "learning_rate": 1.5283896384299613e-07, "loss": 1.085, "step": 3553 }, { "epoch": 0.95, "learning_rate": 1.5134067048990342e-07, "loss": 1.103, "step": 3554 }, { "epoch": 0.95, "learning_rate": 1.4984970130807796e-07, "loss": 1.0728, "step": 3555 }, { "epoch": 0.95, "learning_rate": 1.4836605740631283e-07, "loss": 1.0129, "step": 3556 }, { "epoch": 0.95, "learning_rate": 1.4688973988794873e-07, "loss": 1.0753, "step": 3557 }, { "epoch": 0.95, "learning_rate": 1.4542074985088528e-07, "loss": 1.0438, "step": 3558 }, { "epoch": 0.95, "learning_rate": 1.4395908838756744e-07, "loss": 1.0544, "step": 3559 }, { "epoch": 0.95, "learning_rate": 1.425047565849924e-07, "loss": 1.0232, "step": 3560 }, { "epoch": 0.95, "learning_rate": 1.4105775552470502e-07, "loss": 1.0727, "step": 3561 }, { "epoch": 0.95, "learning_rate": 1.3961808628280005e-07, "loss": 1.0919, "step": 3562 }, { "epoch": 0.95, "learning_rate": 1.3818574992991772e-07, "loss": 1.0275, "step": 3563 }, { "epoch": 0.95, "learning_rate": 1.3676074753124824e-07, "loss": 1.0589, "step": 3564 }, { "epoch": 0.95, "learning_rate": 1.3534308014652608e-07, "loss": 1.0253, "step": 3565 }, { "epoch": 0.95, "learning_rate": 1.3393274883002794e-07, "loss": 1.0305, "step": 3566 }, { "epoch": 0.95, "learning_rate": 1.3252975463057925e-07, "loss": 1.0567, "step": 3567 }, { "epoch": 0.95, "learning_rate": 1.311340985915488e-07, "loss": 1.0785, "step": 3568 }, { "epoch": 0.95, "learning_rate": 1.2974578175084407e-07, "loss": 1.0466, "step": 3569 }, { "epoch": 0.95, "learning_rate": 1.283648051409192e-07, "loss": 1.0146, "step": 3570 }, { "epoch": 0.95, "learning_rate": 1.2699116978876823e-07, "loss": 1.0617, "step": 3571 }, { "epoch": 0.95, "learning_rate": 1.2562487671592182e-07, "loss": 1.0426, "step": 3572 }, { "epoch": 0.95, "learning_rate": 1.242659269384572e-07, "loss": 1.0416, "step": 3573 }, { "epoch": 0.95, "learning_rate": 1.229143214669859e-07, "loss": 1.0693, "step": 3574 }, { "epoch": 0.95, "learning_rate": 1.2157006130665838e-07, "loss": 1.024, "step": 3575 }, { "epoch": 0.95, "learning_rate": 1.2023314745716496e-07, "loss": 1.0255, "step": 3576 }, { "epoch": 0.95, "learning_rate": 1.1890358091272925e-07, "loss": 1.0502, "step": 3577 }, { "epoch": 0.95, "learning_rate": 1.1758136266211251e-07, "loss": 1.0673, "step": 3578 }, { "epoch": 0.95, "learning_rate": 1.1626649368861153e-07, "loss": 1.0433, "step": 3579 }, { "epoch": 0.95, "learning_rate": 1.149589749700586e-07, "loss": 1.0604, "step": 3580 }, { "epoch": 0.95, "learning_rate": 1.1365880747881697e-07, "loss": 1.0752, "step": 3581 }, { "epoch": 0.95, "learning_rate": 1.1236599218178546e-07, "loss": 1.0216, "step": 3582 }, { "epoch": 0.95, "learning_rate": 1.1108053004039499e-07, "loss": 1.0495, "step": 3583 }, { "epoch": 0.95, "learning_rate": 1.0980242201060643e-07, "loss": 1.0764, "step": 3584 }, { "epoch": 0.95, "learning_rate": 1.085316690429139e-07, "loss": 1.0546, "step": 3585 }, { "epoch": 0.95, "learning_rate": 1.0726827208234036e-07, "loss": 1.0301, "step": 3586 }, { "epoch": 0.96, "learning_rate": 1.0601223206843869e-07, "loss": 1.0766, "step": 3587 }, { "epoch": 0.96, "learning_rate": 1.0476354993528948e-07, "loss": 1.079, "step": 3588 }, { "epoch": 0.96, "learning_rate": 1.0352222661150435e-07, "loss": 1.069, "step": 3589 }, { "epoch": 0.96, "learning_rate": 1.0228826302021932e-07, "loss": 1.0559, "step": 3590 }, { "epoch": 0.96, "learning_rate": 1.0106166007909924e-07, "loss": 1.075, "step": 3591 }, { "epoch": 0.96, "learning_rate": 9.98424187003344e-08, "loss": 1.0289, "step": 3592 }, { "epoch": 0.96, "learning_rate": 9.863053979063952e-08, "loss": 1.0611, "step": 3593 }, { "epoch": 0.96, "learning_rate": 9.74260242512548e-08, "loss": 1.0361, "step": 3594 }, { "epoch": 0.96, "learning_rate": 9.622887297794592e-08, "loss": 1.0219, "step": 3595 }, { "epoch": 0.96, "learning_rate": 9.50390868609985e-08, "loss": 1.0618, "step": 3596 }, { "epoch": 0.96, "learning_rate": 9.385666678522477e-08, "loss": 1.0322, "step": 3597 }, { "epoch": 0.96, "learning_rate": 9.268161362995687e-08, "loss": 1.0404, "step": 3598 }, { "epoch": 0.96, "learning_rate": 9.151392826904915e-08, "loss": 1.0567, "step": 3599 }, { "epoch": 0.96, "learning_rate": 9.035361157087585e-08, "loss": 1.0671, "step": 3600 }, { "epoch": 0.96, "learning_rate": 8.920066439833231e-08, "loss": 1.093, "step": 3601 }, { "epoch": 0.96, "learning_rate": 8.805508760883263e-08, "loss": 1.0381, "step": 3602 }, { "epoch": 0.96, "learning_rate": 8.691688205431092e-08, "loss": 1.046, "step": 3603 }, { "epoch": 0.96, "learning_rate": 8.578604858121675e-08, "loss": 1.058, "step": 3604 }, { "epoch": 0.96, "learning_rate": 8.466258803052185e-08, "loss": 1.0029, "step": 3605 }, { "epoch": 0.96, "learning_rate": 8.354650123770903e-08, "loss": 1.0683, "step": 3606 }, { "epoch": 0.96, "learning_rate": 8.243778903278432e-08, "loss": 1.0357, "step": 3607 }, { "epoch": 0.96, "learning_rate": 8.133645224026376e-08, "loss": 1.0909, "step": 3608 }, { "epoch": 0.96, "learning_rate": 8.024249167917997e-08, "loss": 1.0556, "step": 3609 }, { "epoch": 0.96, "learning_rate": 7.915590816308216e-08, "loss": 1.095, "step": 3610 }, { "epoch": 0.96, "learning_rate": 7.807670250003064e-08, "loss": 1.0923, "step": 3611 }, { "epoch": 0.96, "learning_rate": 7.700487549260227e-08, "loss": 1.0819, "step": 3612 }, { "epoch": 0.96, "learning_rate": 7.594042793788281e-08, "loss": 1.071, "step": 3613 }, { "epoch": 0.96, "learning_rate": 7.488336062747236e-08, "loss": 1.0711, "step": 3614 }, { "epoch": 0.96, "learning_rate": 7.383367434748323e-08, "loss": 0.9789, "step": 3615 }, { "epoch": 0.96, "learning_rate": 7.279136987853763e-08, "loss": 1.02, "step": 3616 }, { "epoch": 0.96, "learning_rate": 7.175644799576553e-08, "loss": 1.042, "step": 3617 }, { "epoch": 0.96, "learning_rate": 7.072890946881239e-08, "loss": 1.0375, "step": 3618 }, { "epoch": 0.96, "learning_rate": 6.970875506182806e-08, "loss": 1.0377, "step": 3619 }, { "epoch": 0.96, "learning_rate": 6.869598553347345e-08, "loss": 1.0879, "step": 3620 }, { "epoch": 0.96, "learning_rate": 6.76906016369161e-08, "loss": 1.0494, "step": 3621 }, { "epoch": 0.96, "learning_rate": 6.669260411983236e-08, "loss": 1.0594, "step": 3622 }, { "epoch": 0.96, "learning_rate": 6.570199372440522e-08, "loss": 1.0624, "step": 3623 }, { "epoch": 0.96, "learning_rate": 6.471877118732207e-08, "loss": 1.0729, "step": 3624 }, { "epoch": 0.97, "learning_rate": 6.37429372397802e-08, "loss": 1.0826, "step": 3625 }, { "epoch": 0.97, "learning_rate": 6.277449260747803e-08, "loss": 1.0391, "step": 3626 }, { "epoch": 0.97, "learning_rate": 6.181343801062279e-08, "loss": 1.0309, "step": 3627 }, { "epoch": 0.97, "learning_rate": 6.085977416392163e-08, "loss": 1.0564, "step": 3628 }, { "epoch": 0.97, "learning_rate": 5.991350177658839e-08, "loss": 1.0304, "step": 3629 }, { "epoch": 0.97, "learning_rate": 5.897462155233902e-08, "loss": 1.006, "step": 3630 }, { "epoch": 0.97, "learning_rate": 5.804313418939278e-08, "loss": 1.0506, "step": 3631 }, { "epoch": 0.97, "learning_rate": 5.7119040380470004e-08, "loss": 1.049, "step": 3632 }, { "epoch": 0.97, "learning_rate": 5.62023408127943e-08, "loss": 1.0946, "step": 3633 }, { "epoch": 0.97, "learning_rate": 5.529303616808812e-08, "loss": 1.0413, "step": 3634 }, { "epoch": 0.97, "learning_rate": 5.4391127122578325e-08, "loss": 1.0349, "step": 3635 }, { "epoch": 0.97, "learning_rate": 5.349661434698616e-08, "loss": 1.0849, "step": 3636 }, { "epoch": 0.97, "learning_rate": 5.2609498506537294e-08, "loss": 1.0688, "step": 3637 }, { "epoch": 0.97, "learning_rate": 5.1729780260953986e-08, "loss": 1.0176, "step": 3638 }, { "epoch": 0.97, "learning_rate": 5.085746026445959e-08, "loss": 1.0577, "step": 3639 }, { "epoch": 0.97, "learning_rate": 4.999253916577185e-08, "loss": 1.0368, "step": 3640 }, { "epoch": 0.97, "learning_rate": 4.913501760811068e-08, "loss": 1.0666, "step": 3641 }, { "epoch": 0.97, "learning_rate": 4.8284896229188195e-08, "loss": 1.0473, "step": 3642 }, { "epoch": 0.97, "learning_rate": 4.744217566121756e-08, "loss": 1.0367, "step": 3643 }, { "epoch": 0.97, "learning_rate": 4.660685653090635e-08, "loss": 1.0477, "step": 3644 }, { "epoch": 0.97, "learning_rate": 4.577893945945655e-08, "loss": 1.0135, "step": 3645 }, { "epoch": 0.97, "learning_rate": 4.495842506256898e-08, "loss": 1.0831, "step": 3646 }, { "epoch": 0.97, "learning_rate": 4.414531395043553e-08, "loss": 1.0256, "step": 3647 }, { "epoch": 0.97, "learning_rate": 4.333960672774362e-08, "loss": 1.0398, "step": 3648 }, { "epoch": 0.97, "learning_rate": 4.254130399367617e-08, "loss": 1.0673, "step": 3649 }, { "epoch": 0.97, "learning_rate": 4.175040634190941e-08, "loss": 1.0515, "step": 3650 }, { "epoch": 0.97, "learning_rate": 4.096691436060951e-08, "loss": 1.0194, "step": 3651 }, { "epoch": 0.97, "learning_rate": 4.019082863243817e-08, "loss": 1.0347, "step": 3652 }, { "epoch": 0.97, "learning_rate": 3.9422149734550385e-08, "loss": 1.0652, "step": 3653 }, { "epoch": 0.97, "learning_rate": 3.8660878238588884e-08, "loss": 1.0705, "step": 3654 }, { "epoch": 0.97, "learning_rate": 3.7907014710689694e-08, "loss": 1.0655, "step": 3655 }, { "epoch": 0.97, "learning_rate": 3.716055971148214e-08, "loss": 1.0442, "step": 3656 }, { "epoch": 0.97, "learning_rate": 3.642151379608105e-08, "loss": 1.0283, "step": 3657 }, { "epoch": 0.97, "learning_rate": 3.56898775140968e-08, "loss": 1.0325, "step": 3658 }, { "epoch": 0.97, "learning_rate": 3.496565140962527e-08, "loss": 1.0651, "step": 3659 }, { "epoch": 0.97, "learning_rate": 3.42488360212534e-08, "loss": 1.053, "step": 3660 }, { "epoch": 0.97, "learning_rate": 3.353943188205588e-08, "loss": 1.0705, "step": 3661 }, { "epoch": 0.97, "learning_rate": 3.283743951959739e-08, "loss": 1.0715, "step": 3662 }, { "epoch": 0.98, "learning_rate": 3.214285945593032e-08, "loss": 1.0667, "step": 3663 }, { "epoch": 0.98, "learning_rate": 3.14556922075937e-08, "loss": 1.0255, "step": 3664 }, { "epoch": 0.98, "learning_rate": 3.077593828561431e-08, "loss": 1.0631, "step": 3665 }, { "epoch": 0.98, "learning_rate": 3.010359819550668e-08, "loss": 1.0392, "step": 3666 }, { "epoch": 0.98, "learning_rate": 2.9438672437271942e-08, "loss": 1.0268, "step": 3667 }, { "epoch": 0.98, "learning_rate": 2.8781161505394563e-08, "loss": 1.0453, "step": 3668 }, { "epoch": 0.98, "learning_rate": 2.8131065888848952e-08, "loss": 1.0391, "step": 3669 }, { "epoch": 0.98, "learning_rate": 2.7488386071091722e-08, "loss": 1.0411, "step": 3670 }, { "epoch": 0.98, "learning_rate": 2.6853122530066113e-08, "loss": 1.0162, "step": 3671 }, { "epoch": 0.98, "learning_rate": 2.6225275738199775e-08, "loss": 1.0485, "step": 3672 }, { "epoch": 0.98, "learning_rate": 2.5604846162404774e-08, "loss": 1.0429, "step": 3673 }, { "epoch": 0.98, "learning_rate": 2.4991834264077586e-08, "loss": 1.0992, "step": 3674 }, { "epoch": 0.98, "learning_rate": 2.438624049909799e-08, "loss": 1.0501, "step": 3675 }, { "epoch": 0.98, "learning_rate": 2.3788065317827956e-08, "loss": 1.042, "step": 3676 }, { "epoch": 0.98, "learning_rate": 2.3197309165113868e-08, "loss": 1.0676, "step": 3677 }, { "epoch": 0.98, "learning_rate": 2.2613972480285407e-08, "loss": 1.0262, "step": 3678 }, { "epoch": 0.98, "learning_rate": 2.203805569715334e-08, "loss": 1.0906, "step": 3679 }, { "epoch": 0.98, "learning_rate": 2.146955924401062e-08, "loss": 1.0636, "step": 3680 }, { "epoch": 0.98, "learning_rate": 2.0908483543632396e-08, "loss": 1.0467, "step": 3681 }, { "epoch": 0.98, "learning_rate": 2.035482901327379e-08, "loss": 1.0441, "step": 3682 }, { "epoch": 0.98, "learning_rate": 1.9808596064673225e-08, "loss": 1.0517, "step": 3683 }, { "epoch": 0.98, "learning_rate": 1.9269785104047976e-08, "loss": 1.0497, "step": 3684 }, { "epoch": 0.98, "learning_rate": 1.8738396532097524e-08, "loss": 1.0345, "step": 3685 }, { "epoch": 0.98, "learning_rate": 1.82144307440002e-08, "loss": 1.0363, "step": 3686 }, { "epoch": 0.98, "learning_rate": 1.769788812941542e-08, "loss": 1.0339, "step": 3687 }, { "epoch": 0.98, "learning_rate": 1.7188769072479238e-08, "loss": 1.0845, "step": 3688 }, { "epoch": 0.98, "learning_rate": 1.6687073951812126e-08, "loss": 1.0655, "step": 3689 }, { "epoch": 0.98, "learning_rate": 1.6192803140507862e-08, "loss": 1.1102, "step": 3690 }, { "epoch": 0.98, "learning_rate": 1.5705957006143525e-08, "loss": 1.0632, "step": 3691 }, { "epoch": 0.98, "learning_rate": 1.5226535910772834e-08, "loss": 1.0433, "step": 3692 }, { "epoch": 0.98, "learning_rate": 1.4754540210926149e-08, "loss": 1.0744, "step": 3693 }, { "epoch": 0.98, "learning_rate": 1.4289970257616026e-08, "loss": 1.0354, "step": 3694 }, { "epoch": 0.98, "learning_rate": 1.3832826396327215e-08, "loss": 1.0679, "step": 3695 }, { "epoch": 0.98, "learning_rate": 1.338310896702777e-08, "loss": 1.097, "step": 3696 }, { "epoch": 0.98, "learning_rate": 1.2940818304157942e-08, "loss": 1.0752, "step": 3697 }, { "epoch": 0.98, "learning_rate": 1.250595473663685e-08, "loss": 1.0666, "step": 3698 }, { "epoch": 0.98, "learning_rate": 1.2078518587861355e-08, "loss": 1.082, "step": 3699 }, { "epoch": 0.99, "learning_rate": 1.1658510175702742e-08, "loss": 1.0283, "step": 3700 }, { "epoch": 0.99, "learning_rate": 1.1245929812510047e-08, "loss": 1.0761, "step": 3701 }, { "epoch": 0.99, "learning_rate": 1.0840777805106728e-08, "loss": 1.0538, "step": 3702 }, { "epoch": 0.99, "learning_rate": 1.0443054454793988e-08, "loss": 1.0843, "step": 3703 }, { "epoch": 0.99, "learning_rate": 1.0052760057348565e-08, "loss": 1.0784, "step": 3704 }, { "epoch": 0.99, "learning_rate": 9.669894903018284e-09, "loss": 1.0613, "step": 3705 }, { "epoch": 0.99, "learning_rate": 9.294459276532053e-09, "loss": 1.0671, "step": 3706 }, { "epoch": 0.99, "learning_rate": 8.926453457089867e-09, "loss": 1.0314, "step": 3707 }, { "epoch": 0.99, "learning_rate": 8.565877718367255e-09, "loss": 1.0646, "step": 3708 }, { "epoch": 0.99, "learning_rate": 8.212732328514161e-09, "loss": 1.0698, "step": 3709 }, { "epoch": 0.99, "learning_rate": 7.867017550154954e-09, "loss": 1.0529, "step": 3710 }, { "epoch": 0.99, "learning_rate": 7.528733640388419e-09, "loss": 1.0688, "step": 3711 }, { "epoch": 0.99, "learning_rate": 7.197880850785544e-09, "loss": 1.0842, "step": 3712 }, { "epoch": 0.99, "learning_rate": 6.874459427393954e-09, "loss": 1.0669, "step": 3713 }, { "epoch": 0.99, "learning_rate": 6.5584696107312555e-09, "loss": 1.0933, "step": 3714 }, { "epoch": 0.99, "learning_rate": 6.2499116357916945e-09, "loss": 1.0181, "step": 3715 }, { "epoch": 0.99, "learning_rate": 5.948785732041717e-09, "loss": 1.0499, "step": 3716 }, { "epoch": 0.99, "learning_rate": 5.655092123417749e-09, "loss": 1.0289, "step": 3717 }, { "epoch": 0.99, "learning_rate": 5.368831028332855e-09, "loss": 1.0625, "step": 3718 }, { "epoch": 0.99, "learning_rate": 5.0900026596722994e-09, "loss": 1.0947, "step": 3719 }, { "epoch": 0.99, "learning_rate": 4.818607224791327e-09, "loss": 1.0711, "step": 3720 }, { "epoch": 0.99, "learning_rate": 4.554644925519602e-09, "loss": 1.0406, "step": 3721 }, { "epoch": 0.99, "learning_rate": 4.298115958158988e-09, "loss": 1.0708, "step": 3722 }, { "epoch": 0.99, "learning_rate": 4.049020513482438e-09, "loss": 1.0836, "step": 3723 }, { "epoch": 0.99, "learning_rate": 3.8073587767362145e-09, "loss": 1.0907, "step": 3724 }, { "epoch": 0.99, "learning_rate": 3.5731309276354486e-09, "loss": 1.0636, "step": 3725 }, { "epoch": 0.99, "learning_rate": 3.346337140369693e-09, "loss": 1.031, "step": 3726 }, { "epoch": 0.99, "learning_rate": 3.1269775835995884e-09, "loss": 1.0771, "step": 3727 }, { "epoch": 0.99, "learning_rate": 2.9150524204568653e-09, "loss": 1.0645, "step": 3728 }, { "epoch": 0.99, "learning_rate": 2.7105618085421225e-09, "loss": 1.0219, "step": 3729 }, { "epoch": 0.99, "learning_rate": 2.5135058999326e-09, "loss": 1.0077, "step": 3730 }, { "epoch": 0.99, "learning_rate": 2.3238848411710757e-09, "loss": 1.0755, "step": 3731 }, { "epoch": 0.99, "learning_rate": 2.141698773273637e-09, "loss": 1.0426, "step": 3732 }, { "epoch": 0.99, "learning_rate": 1.9669478317274614e-09, "loss": 1.036, "step": 3733 }, { "epoch": 0.99, "learning_rate": 1.7996321464897048e-09, "loss": 1.0822, "step": 3734 }, { "epoch": 0.99, "learning_rate": 1.6397518419886127e-09, "loss": 1.0475, "step": 3735 }, { "epoch": 0.99, "learning_rate": 1.4873070371212993e-09, "loss": 1.0653, "step": 3736 }, { "epoch": 0.99, "learning_rate": 1.3422978452581891e-09, "loss": 1.0531, "step": 3737 }, { "epoch": 1.0, "learning_rate": 1.204724374238575e-09, "loss": 1.0559, "step": 3738 }, { "epoch": 1.0, "learning_rate": 1.074586726371729e-09, "loss": 1.0545, "step": 3739 }, { "epoch": 1.0, "learning_rate": 9.518849984369028e-10, "loss": 1.1161, "step": 3740 }, { "epoch": 1.0, "learning_rate": 8.366192816833262e-10, "loss": 1.0238, "step": 3741 }, { "epoch": 1.0, "learning_rate": 7.287896618324297e-10, "loss": 1.0365, "step": 3742 }, { "epoch": 1.0, "learning_rate": 6.283962190734016e-10, "loss": 1.056, "step": 3743 }, { "epoch": 1.0, "learning_rate": 5.3543902806652e-10, "loss": 1.032, "step": 3744 }, { "epoch": 1.0, "learning_rate": 4.4991815793982154e-10, "loss": 1.0807, "step": 3745 }, { "epoch": 1.0, "learning_rate": 3.718336722946525e-10, "loss": 1.0314, "step": 3746 }, { "epoch": 1.0, "learning_rate": 3.0118562919900787e-10, "loss": 1.065, "step": 3747 }, { "epoch": 1.0, "learning_rate": 2.3797408119197174e-10, "loss": 1.0512, "step": 3748 }, { "epoch": 1.0, "learning_rate": 1.8219907528371772e-10, "loss": 1.041, "step": 3749 }, { "epoch": 1.0, "learning_rate": 1.3386065294995754e-10, "loss": 1.0707, "step": 3750 }, { "epoch": 1.0, "learning_rate": 9.295885014082295e-11, "loss": 1.0836, "step": 3751 }, { "epoch": 1.0, "learning_rate": 5.949369727309417e-11, "loss": 1.0624, "step": 3752 }, { "epoch": 1.0, "learning_rate": 3.3465219233530524e-11, "loss": 1.0888, "step": 3753 }, { "epoch": 1.0, "learning_rate": 1.4873435378870426e-11, "loss": 1.0682, "step": 3754 }, { "epoch": 1.0, "learning_rate": 3.71835953583144e-12, "loss": 1.0739, "step": 3755 }, { "epoch": 1.0, "learning_rate": 0.0, "loss": 1.0259, "step": 3756 }, { "epoch": 1.0, "step": 3756, "total_flos": 5.0002717575098316e+23, "train_loss": 1.2089478572209675, "train_runtime": 66670.3775, "train_samples_per_second": 57.689, "train_steps_per_second": 0.056 } ], "logging_steps": 1.0, "max_steps": 3756, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "save_steps": 188, "total_flos": 5.0002717575098316e+23, "train_batch_size": 1, "trial_name": null, "trial_params": null }